Vous êtes sur la page 1sur 365

Basics of

Social Research
Qualitat ive and Quanftative Approaches

w. Lawrence Neuman
University of Wisconsin- Whitewater

Boston New York San Francisco

Mexico Cit y Mont real Toront o London Madrid Munich Paris
Hong Kong Singapore Tokyo Cape Town Sydney
Senior Series Editor: Jeff Lasser
Editorial Assistant: ErikkaAdams
Senior Marketing Manager: KellyMay
Production Editor: Roberta Sherman
Composition Buyer: Linda Cox
Manufacturing Buyer: JoAnneSweeney PART ONI
Editorial Production Services and Electronic Composition: Publishers' Design and Production Services, Inc. CHAPTER'
Cover Administrator: KristinaM ose-Libon
For related titles and support materials, visit our online catalog at www.ablongman.com.

Copyright © 2007, 2004 Pearson Education, Inc. CHAPTERj

All rights reserved. No part of the material protected by this copyright notice may be reproduced or utilized in CHAPTER!
any form or by any means, electronic or mechanical, including photocopying, recording, or by any information
storage and retrieval system, without written permission from the copyright owner.

To obtain permission(s) to use material from this work, please submit a written request to Allyn and Bacon,
Permissions Department, 75 Arlington Street, Boston, MA 02116 or fax your request to 617-848-7320. PART TWCl
Between the time Website information is gathered and then published, it is not unusual for some sites to have CHAPTER ~
closed. Also, the transcription ofURLs can result in typographical errors. The publisher would appreciate
notification where these errors occur so that they may be corrected in subsequent editions. CHAPTER I

Cataloging-in-Publication data unavailable at press time.





Printed in the United States of America

10 9 8 7 6 5 4 3 2 1 RRD-VA 10 09 08 07 06
PAR TON E Foundations
CHAPTER 1 Doing Social Research
CHAPTER 2 Theory and Social Research
CHAPTER 3 Ethics in Social Research
CHAPTER 4 Reviewing the Scholarly literature and Planning a Study
CHAPTER 5 Qualitative and Quantitative Measurement
CHAPTER 6 Qualitative and Quantitative Sampling

PART TWO Conducting Quantitative Research

CHAPTER 7 Survey Research
CHAPTER 8 Experimental Research
CHAPTER 9 Nonreactive Research and Secondary Analysis
CHAPTER 10 Analysis of Quantitative Data

PAR T T H R EE Conducting Qualitative Research

CHAPTER 11 Field Research
CHAPTER 12 Historical-Comparative Research
CHAPTER 13 Analysis of Qualitative Data

PART FOUR Writing a Research Report

CHAPTER 14 Writing the Research Report

Doing Social Research

Alternatives to Social Research
Common Sense
Media Myths
Personal Experience
How Science Works
The Scientific Community
The Scientific Method and Attitude
Journal Articles in Science
Steps in the Research Process
Dimensions of Research
Use of Research
Purpose of a Study
Time Dimension in Research
Data Collection Techniques

ceptions. When I asked students in my classes an exciting pml

what they think social research entails, they gave persistence, peIlI
Social research is all around us. Educators, gov­ the following answers: biguity, inter.d
ernment officials, business managers, human ing quality"ui
service providers, and health care professionals • It is based on facts alone; there is no theory Reading till
regularly use social research methods and find­ or personal judgment. into an expert nI
ings. People use social research to raise children, • Only experts with a Ph.D. degree or college be a better coIlSlli
reduce crime, improve public health, sell prod­ professors read it or do it. to understand bli
ucts, or just understand one's life. Reports of re­ • It means going to the library and finding a and prepare ~
search appear on broadcast news programs, in lot of magazine articles or books on a topic. jeets. After studJI
popular magazines, in newspapers, and on the • It is when someone hangs around a group of what researd
Internet. and observes. properly condud
Research findings can affect people's daily • It means conducting a controlled experi­
lives and public policies. For example, I recently ment.
heard a debate regarding a U.S. federal govern­
ment program to offer teenagers sexual absti­
• Social research is drawing a sample of peo­
ple and giving them questionnaires to com­
nence counseling. A high-level government plete.
official argued for such counseling and strongly • It is looking up lots of statistical tables Unless you are ~
opposed offering teens birth control informa­ and information from official government about the social ~
tion. An independent health administrator reports. cial research. Y0IIIIl
noted that there is no scientific evidence show­ • To do it, one must use computers to create you know using ~
ing that abstinence-only counseling works. He statistics, charts, and graphs. It is based on "mj
said that 80 percent of teens are already sexually (e.g., friends, ~
active by the age of 18, therefore it is essential to The first two answers are wrong, and the have knowledge •
provide birth control information. He pointed others describe only part of what constitutes so­ ences, the books j
to many research studies showing that birth cial research. It is unwise to confuse one part and the movies ali
control instruction for teens reduces pregnancy with the whole. You may also usel
rates and the spread of sexually transmitted dis­ People conduct social research to learn More than a4
eases. The government abstinence-only advo­ something new about the social world; or to research is a proce
cate relied on moral persuasion because he had carefully document guesses, hunches, or beliefs is a more strucna
no research evidence. Ideology, faith, and poli­ about it; or to refine their understanding of how process than the OIl
tics shape many government programs rather the social world works. A researcher combines in daily life. ~
than solid research evidence, but good social re­ theories or ideas with facts in a careful, system­ often correct, bUli
search can help all of us make informed deci­ atic way and uses creativity. He or she learns to is more likely to ,
sions. The evidence also explains why many organize and plan carefully and to select the ap­ Althoughr~
programs fail to accomplish much or may do propriate technique to address a specific kind of fect knowledge, all
more harm than good. question. A researcher also must treat the people much less likelr III
This book is about social research. In simple in a study in ethical and moral ways. In addition, alternatives befORl
terms, research is a way of going about finding a researcher must fully and clearly communicate
answers to questions. Professors, professional the results of a study to others.
researchers, practitioners, and students in many Social research is a process in which people
fields conduct research to seek answers to ques­ combine a set of principles, outlooks, and ideas You have acquire
tions about the social world. You probably al­ (i.e., methodology) with a collection of specific teachers, and expeIl
ready have some notion of what social research practices, techniques, and strategies (i.e., a vision, and othes
entails. First, let me end some possible miscon­ method of inquiry) to produce knowledge. It is something as bei:ot

an exciting process of discovery, but it requires position of authority says it is true or because it
persistence, personal integrity, tolerance for am­ is in an authoritative publication, you are relying
biguity, interaction with others, and pride in do­ on authority as a basis for knowledge. Relying
ing quality work. on the wisdom of authorities is a quick, simple,
Reading this book cannot transform you and cheap way to learn something. Authorities
into an expert researcher, but it can teach you to often spend time and effort to learn something,
be a better consumer of research results, help you and you can benefit from their experience and
to understand how the research enterprise works, work.
and prepare you to conduct small research pro­ J There are also limitations to relying on au­

jects. After studying this book, you will be aware thority. First, it is easyto overestimate the exper­
of what research can and cannot do, and why tise of other people. You may assume that they
properly conducted research is important. are right when they are not. History is full of past
experts whom we now see as being misinformed.
For example, some "experts" of the past mea­
sured intelligence by counting bumps on the
skull; other "experts" used bloodletting to try to
cure diseases. Their errors seem obvious now,
Unless you are unusual, most of what you know but can you be certain that today's experts will
about the social world is not based on doing so­ not become tomorrow's fools? Second, authori­
cial research. You probably learned most of what ties may not agree, and all authorities may not be
you know using an alternative to social research. equally dependable. Whom should we believe if
It is based on what your parents and other people authorities disagree? Third, authorities may
(e.g., friends, teachers) have told you. You also speak on fields they know little about or be plain
have knowledge based on your personal experi­ wrong. An expert who is very informed about
ences, the books and magazines you have read, one area may use his or her authority in an un­
and the movies and television you have watched. related area. Also, using the halo effect (dis­
You may also use plain old "common sense." cussed later), expertise in one area may spill over
More than a collection of techniques, social illegitimately to be authority in a totally different
research is a process for producing knowledge. It area. Have you ever seen television commercials
is a more structured, organized, and systematic where a movie star uses his or her fame as au­
process than the alternatives that most of us use thority to convince you to buy a car?We need to
in daily life. Knowledge from the alternatives is ask: Who is or is not an authority?
often correct, but knowledge based on research An additional issue is the misuse of author­
is more likely to be true and have fewer errors. ity. Sometimes organizations or individuals
Although research does not always produce per­ give an appearance of authority so they can con­
fect knowledge, compared to the alternatives it is vince others to agree to something that they
much less likely to be flawed. Let us review the might not otherwise agree to. A related situation
alternatives before examining social research. occurs when a person with little training and ex­
pertise is named as a "senior fellow" or "adjunct
scholar" in a private "think tank" with an im­
pressive name, such as the Center for the Study
You have acquired knowledge from parents, of X or the Institute on Y Research. Some think
teachers, and experts as well as from books, tele­ tanks are legitimate research centers, but many
vision, and other media. When you accept are mere fronts created by wealthy special-inter­
something as being true because someone in a est groups to engage in advocacy politics. Think

tanks can make anyone a "scholar" to facilitate Common Sense

the mass media accepting the person as an au­
You know a lot about the social world from your
thority on an issue. In reality, the person may
everyday reasoning or common sense. You rely
not have any real expertise.! Also, too much re­
on what everyone knows and what "just makes
liance on authorities can be dangerous to a de­
sense." For example, it "just makes sense" that
mocratic society. Experts may promote ideas
murder rates are higher in nations that do not
that strengthen their own power and position.
have a death penalty, because people are less
When we accept the authority of experts, but do
likely to kill if they face execution for doing so.
not know how they arrived at their knowledge,
This and other widely held commonsense be­
we lose the ability to evaluate what the experts
liefs, such as that poor youth are more likely to
say and lose control of our destiny.
commit deviant acts than those from the middle
class or that most Catholics do not use birth
Tradition control, are false.
Common sense is valuable in daily living,
People sometimes rely on tradition for knowl­
but it allows logical fallacies to slip into thinking.
edge. Tradition is a special case of authority­
For example, the so-called gambler's fallacysays:
the authority of the past. Tradition means you
"If! have a long string oflosses playing a lottery,
accept something as being true because "it's the
the next time I play, my chances of winning will
way things have always been." For example,
be better." In terms of probability and the facts,
my father-in-law says that drinking a shot of
this is false. Also, common sense contains con­
whiskey cures a cold. When I asked about his
tradictory ideas that often go unnoticed because
statement, he said that he had learned it from his
people use the ideas at different times, such as
father when he was a child, and it had come
"opposites attract" and "birds of a feather flock
down from past generations. Tradition was the
together." Common sense can originate in tradi­
basis of the knowledge for the cure. Here is an
tion. It is useful and sometimes correct, but it
example from the social world: Many people be­
also contains errors, misinformation, contradic­
lieve that children who are raised at home by
tion, and prejudice.
their mothers grow up to be better adjusted and
have fewer personal problems than those raised
in other settings. People "know" this, but how
Media Myths
did they learn it? Most accept it because they be­
lieve (rightly or wrongly) that it was true in the Television shows, movies, and newspaper and
past or is the way things have always been done. magazine articles are important sources of in­
Some traditional social knowledge begins as formation. For example, most people have no
simple prejudice. You might rely on tradition contact with criminals but learn about crime by
without being fully aware of it with a belief such watching television shows and movies and by
as "People from that side of the tracks will never reading newspapers. However, the television
amount to anything" or "You never can trust portrayals of crime, and of many other things,
that type of person" or "That's the way men (or do not accurately reflect social reality. The writ­
women) are." Even if traditional knowledge was ers who create or "adapt" images from life for
once true, it can become distorted as it is passed television shows and movie scripts distort real­
on, and soon it is no longer true. People may ity either out of ignorance or because they rely
cling to traditional knowledge without real un­ on authority, tradition, and common sense.
derstanding; they assume that because some­ Their primary goal is to entertain, not to repre­
thing may have worked or been true in the past, sent reality accurately. Although many journal­
it will continue to be true. ists try to present a realistic picture ofthe world,

they must write stories in short time periods

with limited information and within editorial
Unfortunately, the media tend to perpetu­
ate the myths of a culture. For example, the me­ Americans hear a lot about roadrage. Newsweek mag­
dia show that most people who receive welfare azine, Time magazine, and newspapers in most major
are Black (actually, most are White), that most cities have carried headlines about it. Leading na­
people who are mentally ill are violent and dan­ tional political officials have held public hearings on
gerous (only a small percentage actually are), it, and the federal government gives millions of dol­
and that most people who are elderly are senile lars in grants to law enforcement and transportation
departments to reduce it. Today, even psychologists
and in nursing homes (a tiny minority are).
specialize in this disorder.
Also, mass media "hype" can create a feeling that
The term road rage first appeared in 1988, and
a major problem exists when it may not (see Box
by 1997, the print media were carrying over 4,000
1.1). People are misled by visual images more
articles per year on it. Despite media attention about
easily than other forms of "lying"; this means "aggressive driving" and "anger behind the wheel:'
that stories or stereotypes that appear on film there is no scientific evidence for road rage. The term
and television can have a powerful effect on peo­ is not precisely defined and can refer to anything
ple. For example, television repeatedly shows from gunshots from cars, use of hand gestures, run­
low-income, inner-city, African American youth ning bicyclists off the road, tailgating, and even anger
using illegal drugs. Eventually, most people over auto repair bills! All the data on crashes and ac­
"know" that urban Blacks use illegal drugs at a cidents show declines during the period when road
higher rate than other groups in the United rage reached an epidemic.
States, even though this notion is false. Perhaps media reports fueled perceptions of road
Competing interests use the media to win rage. After hearing or reading about road rage and
public support.? Public relations campaigns try having a label for the behavior, people began to no­
to alter what the public thinks about scientific tice rude driving behavior and engaged in selective ob­
findings, making it difficult for the public to servation. We will not know for sure until it is properly
judge research findings. For example, a large studied, but the amount of such behavior may be un­
majority of scientific research supports the changed. It may turn out that the national epidemic
global warming thesis (i.e., pollutants from in­ of road rage is a widely held myth stimulated by re­
dustrialization and massive deforestation are ports in the mass media. (For more information, see
raising the earth's temperature and will cause Michael Fumento, "Road Rage versus Reality:'
dramatic climate change and bring about envi­ Atlantic Monthly [August 1998].)
ronmental disasters). The scientific evidence is
growing and gets stronger each year. The media
give equal attention to a few dissenters who
question global warming, creating the impres­
sion in the public mind that "no one really vironmental regulations, not to advance knowl­
knows" or that scientists are undecided about edge.
the issue of global warming. The media sources Newspapers offer horoscopes, and televi­
fail to mention that the dissenters represent less sion programs or movies report on supernatural
than 2 percent of all scientists, or that most dis­ powers, ESP (extrasensory perception), UFOs
senting studies are paid for by heavily polluting (unidentified flying objects), and angels or
industries. The industries also spend millions of ghosts. Although no scientific evidence exists for
dollars to publicize the findings because their such, between 25 and 50 percent ofthe U.S. pub­
goal is to deflect growing criticism and delay en- lic accepts them as true, and the percentage with

such beliefs has been growing over time as the tive to features that confirm what we think, but
entertainment media give the phenomenon ignore features that contradict it. For example, I TABLE 1-1
more prominence. 3 believe tall people are excellent singers. This may
be because of stereotypes, what my mother told
me, or whatever. I observe tall people and, with­
Personal Experience
out awareness, pay particular attention to their
If something happens to you, if you personally singing. I look at a chorus or top vocalist and no­
see it or experience it, you accept it as true. Per­ tice those who are tall. Without realizing it, I no­
sonal experience, or "seeing is believing," has a tice and remember people and situations that
strong impact and is a powerful source of reinforce my preconceived ideas. Psychologists Authority
knowledge. Unfortunately, personal experience found that people tend to "seek out" and distort
can lead you astray. Something similar to an op­ their memories to make them more consistent
tical illusion or mirage can occur. What appears with what they already think."
true may actually be due to a slight error or dis­ A third error is premature closure. It often
tortion in judgment. The power of immediacy operates with and reinforces the first two errors.
and direct personal contact is very strong. Even Premature closure occurs when you feel you
knowing that, people fall for illusions. Many have the answer and do not need to listen, seek
people believe what they see or personally expe­ information, or raise questions any longer. Un­
rience rather than what very carefully designed fortunately, most of us are a little lazy or get a lit­
research has discovered. tie sloppy. We take a few pieces of evidence or
The four errors of personal experience rein­ look at events for a short while and then think
force each other and can occur in other areas, as we have it figured out. We look for evidence to
well. They are a basis for misleading people confirm or reject an idea and stop when a small CommonSemr
through propaganda, cons or fraud, magic, amount of evidence is present. In a word, we
stereotyping, and some advertising. The most jump to conclusions. For example, I want to
frequent problem is overgeneralization; it occurs learn whether people in my town support Mary
when some evidence supports your belief, but Smith or Jon Van Horn for mayor. I ask 20 peo­
you falsely assume that it applies to many other ple; 16 say they favor Mary, 2 are undecided, and Media Myth
situations, too. Limited generalization may be only 2 favor Jon, so I stop there and believe Mary
appropriate; under certain conditions, a small will win.
amount of evidence can explain a larger situa­ Another common error is the haloeffect; it is
tion. The problem is that many people general­ when we overgeneralize from what we accept as Personal
ize far beyond limited evidence. For example, being highly positive or prestigious and let its Experience
over the years, I have known five blind people. strong reputation or prestige "rub off' onto
All of them were very friendly. Can I conclude other areas. Thus, I pick up a report by a person
that all blind people are friendly? Do the five from a prestigious university, say Harvard or
people with whom I happened to have personal Cambridge University. I assume that the author
experience with represent all blind people? is smart and talented and that the report will be
The second error, selective observation, oc­ excellent. I do not make this assumption about a
curs when you take special notice of some people report by someone from Unknown University. I
or events and tend to seek out evidence that con­ form an opinion and prejudge the report and HOW SCI
firms what you already believe and ignore con­ may not approach it by considering its own mer­ Although it
tradictory information. People often focus on or its alone. How the various alternatives to social natWe,,~oi
observe particular cases or situations, especially research might address the issue of laundry is ..rhat sepaGIIII5
when they fit preconceived ideas. Weare sensi­ shown in Table 1.1. umllves tlbi. . . .
~ut the sociII
TABLE 1.1 Alternatives to Social

processes. This suggests that we examine the

meaning of science and how its works.

The term science suggests an image of test tubes,
computers, rocket ships, and people in white lab
coats. These outward trappings are a part of sci­
ence, especially natural science (i.e., astronomy,
Authority Experts say that as children,
biology, chemistry, geology, and physics,), that
females are taught to make,
deals with the physical and material world (e.g.,
select, mend, and clean clothing
plants, chemicals, rocks, stars, and electricity).
as part of a female focus on
physical appearance and on
The social sciences, such as anthropology, psy­
caring for children or others in a chology, political science, and sociology, involve
family. Women do the laundry the study of people-their beliefs, behavior, in­
based on their childhood teraction, institutions, and so forth. Fewer peo­
preparation. ple associate these disciplines with the word
Tradition Women have done the laundry
science. Science is a social institution and a way
for centuries, so it is a
to produce knowledge. Not everyone is well in­
continuation of what has formed about science. For example, a 2001 sur­
happened for a long time. vey found that about only one-third of U.S.
adults could correctly explain the basics of
Common Sense Men just are not as concerned
about clothing as much as
women, so it only makes sense
Scientists gather data using specialized tech­
that women do the laundry
niques and use the data to support or reject the­
more often. ories. Data are the empirical evidence or
information that one gathers carefully accord­
Media Myth Television commercials show
ing to rules or procedures. The data can be
women often doing laundry and
quantitative (i.e., expressed as numbers) or
enjoying it, so they do laundry
because they think it's fun.
qualitative (i.e., expressed as words, visual im­
ages, sounds, or objects). Empirical evidence
Personal My mother and the mothers of
refers to observations that people experience
Experience all my friends did the laundry.
through the senses-touch, sight, hearing, smell,
My female friends did it for their
and taste. This confuses people, because re­
boyfriends, but never the other
searchers cannot use their senses to directly ob­
way around. It just feels natural
for the woman to do it.
serve many aspects of the social world about
which they seek answers (e.g., intelligence, atti­
tudes, opinions, feelings, emotions, power, au­
thority, etc.). Researchers have many specialized
techniques to observe and indirectly measure
such aspects of the social world.
Although it builds on some aspects of the alter­
native ways of developing knowledge, science is
The Scientific Community
what separates social research. Social research
involves thinking scientifically about questions Science comes to life through the operation of
about the social world and following scientific the scientific community, which sustains the as­

sumptions, attitudes, and techniques of science. vate industry in organizations such as the Na­ tJ5CS,. ~ ~11111111
The scientific community is a collection of people tional Opinion Research Center and the Rand vithin~
who practice science and a set of norms, behav­ Corporation. Most, however, work at the ap­
iors, and attitudes that bind them together. It is a proximately 200 research universities and insti­
~ OIl pnIlil:!lliil
professional community-a group of interacting tutes located in a dozen advanced industrialized ~. aalliliil
people who share ethical principles, beliefs and countries. Thus, the scientific community is ~h*-»-
values, techniques and training, and career paths. scattered geographically, but its members tend such as boIlaI~
For the most part, the scientific community in­ to work together in small clusters. search, ~
cludes both the natural and social sciences.? How big is the scientific community? This is one roodw1nt
Many people outside the core scientific
community use scientific research techniques. A
not an easy question to answer. Using the broad­
est definition (including all scientists and those
its of the
teristics of­
range of practitioners and technicians apply re­ in science-related professions, such as engi­
search techniques that scientists developed and neers), it includes about 15 percent of the labor
refined. Many use the research techniques (e.g., force in advanced industrialized countries. A
a survey) without possessing a deep knowledge better way to look at the scientific community is
of scientific research. Yet, anyone who uses the to examine the basic unit of the larger commu­
techniques or results of science can do so better nity: the discipline (e.g., sociology, biology, psy­
if they also understand the principles and chology, etc.). Scientists are most familiar with a
processes of the scientific community. particular discipline because knowledge is spe­
The boundaries of the scientific community cialized. Compared to other fields with ad­
and its membership are defined loosely. There is vanced training, the numbers are very small. For ence or a rrxfti..
no membership card or master roster. Many example, each year, about 500 people receive
people treat a Ph.D. degree in a scientific field as Ph.D.s in sociology, 16,000 receive medical de­
an informal "entry ticket" to membership in the grees, and 38,000 receive law degrees.
scientific community. The Ph.D., which stands A discipline such as sociology may have
for doctorate of philosophy, is an advanced about 8,000 active researchers worldwide. Most
graduate degree beyond the master's that pre­ researchers complete only two or three studies
pares one to conduct independent research. in their careers, whereas a small number of
Some researchers do not have Ph.D.s and not all highly active researchers conduct many dozens
those who receive Ph.D.s enter occupations in ofstudies. In a specialty or topic area (e.g., study
which they conduct research. They enter many of the death penalty, social movements, di­
occupations and may have other responsibilities vorce), only about 100 researchers are very ac­
(e.g., teaching, administration, consulting, clin­ tive and conduct most research studies.
ical practice, advising, etc.). In fact, about one­ Although research results represent what hu­
half of the people who receive scientific Ph.D.s manity knows and it has a major impact on the
do not follow careers as active researchers. lives of many millions of people, only a small
At the core of the scientific community are number of people are actually producing most
researchers who conduct studies on a full-time new scientific knowledge.
or part-tirne basis, usually with the help of assis­
tants. Many research assistants are graduate stu­
The Scientific Method and Attitude
dents, and some are undergraduates. Working
as a research assistant is the way that most scien­ You have probably heard of the scientific
tists gain a real grasp on the details of doing re­ method, and you may be wondering how it fits
search. Colleges and universities employ most into all this. The scientific method is not one sin­
members of the scientific community's core. gle thing; it refers to the ideas, rules, techniques,
Some scientists work for the government or pri­ and approaches that the scientific community

uses. The method arises from a loose consensus regularly reject half of the submissions. Thus,
within the community of scientists. It includes a several experienced researchers screen a journal
way of looking at the world that places a high article based on its merits alone, and publication
value on professionalism, craftsmanship, ethical represents the study's tentative acceptance by the
integrity, creativity, rigorous standards, and dili­ scientific community as a valid contribution to
gence. It also includes strong professional norms knowledge. Unlike the authors of articles for the
such as honesty and uprightness in doing re­ popular magazines found at newsstands, scien­
search, great candor and openness about how tists are not paid for publishing in scholarly jour­
one conducted a study, and a focus on the mer­ nals. In fact, they may have to pay a small fee to
its of the research itself and not on any charac­ help defray costs just to have their papers consid­
teristics of individuals who conducted the study. ered. Researchers are happy to make their re­
search available to their peers (i.e., other
scientists and researchers) through scholarly
Journal Articles in Science
journals. The article communicates the results of
Consider what happens once a researcher fin­ a study that a researcher might have devoted
ishes a study. First, he or she writes a detailed de­ years of his or her life to, and it is the way re­
scription of the study and the results as a searchers gain respect and visibility among their
research report or a paper using a special format. professional peers. Likewise, the reviewers are
Often, he or she also givesan oral presentation of not paid for reviewing papers, but consider it an
the paper before other researchers at a confer­ honor to be asked to conduct the reviews and to
ence or a meeting of a professional association carry out one of the responsibilities of being in
and seeks comments and suggestions. Next, the the scientific community. The scientific commu­
researcher sends several copies to the editor of a nity imparts great respect to researchers who
scholarly journal. Each editor, a respected re­ publish many articles in the foremost scholarly
searcher chosen by other scientists to oversee the journals because these researchers are directly
journal, removes the title page, which is the only advancing the scientific community's primary
place the author's name appears, and sends the goal-the accumulation of carefully developed
article to several reviewers. The reviewers are re­ knowledge. A researcher gains prestige and
spected scientists who have conducted studies in honor and a reputation as an accomplished re­
the same specialty area or topic. The reviewers searcher through such publications.
do not know who did the study, and the author You may never publish an article in a schol­
of the paper does not know who the reviewers arly journal, but you will probably read many
are. This reinforces the scientific principle of such articles. It is important to see how they are
judging a study on its merits alone. Reviewers a vital component in the system of scientific re­
evaluate the research based on its clarity, origi­ search. Researchers actively read what appears in
nality, standards of good research methods, and the journals to learn about new research findings
advancing knowledge. They return their evalua­ and the methods used to conduct a study. Even­
tions to the editor, who decides to reject the pa­ tually, the new knowledge is disseminated in
per, ask the author to revise and resubmit it, or textbooks, new reports, or public talks.
accept it for publication. It is a very careful, cau­
tious method to ensure quality control.
The scholarly journals that are highly re­

spected and regularly read by most researchers in


a field receive far more papers than they can pub­

lish. They accept only 10to 15 percent ofsubmit­ Social research proceeds in a sequence of steps,
ted manuscripts. Even lower-ranked journals although various approaches to research suggest

slightly different steps. Most studies follow the

seven steps discussed here. To begin the process,
you select a topic-a general area of study or is­
sue, such as domestic abuse, homelessness, or
FIG U R E 1 . 1 Steps in the Research


she clew I.A II
SocilII . . .
sizes. Btbr~
powerful corporate elites. A topic is too broad 1. Select Topic make~.
for conducting a study. This makes the next step resean:h~ . .
crucial. You must then narrow down the topic,
or focus the topic into a specific research ques­
7. Inform
Others 2. Focus
need to"'"
vantages oJfall
tion for a study (e.g., "Are people who marry
younger more likely to engage in physical abuse the ~])(5."
of a spouse under conditions of high stress than eachoftOor.
those who marry older?"). As you learn about a 6. Interpret ....c: ---­ Tbe&Jr"
topic and narrow the focus, you should review Data '"Jl.... 3. Design research is ~
past research, or the literature, on a topic or Study
search. The-llll
question. You also want to develop a possible
answer, or hypothesis, and theory can be impor­ next two
tant at this stage.
After specifying a research question, you
have to develop a highly detailed plan on how
you will carry-6·lit the study. Th.iS third step re­
5. Analyze
is incorpoI3lal

The .
mensions an:
quires that you de~gl~Q~14e_many practical de­ of a study and a~
tailsgfsl,ojpgJheJesearch (e.g., whether to use a ing an end. The seven steps are for one research you learn the ~
surveyor qualitative observing in the field, how project; it is one cycleof going through the steps howthep .
many subjects to use, etc.). It is only after com­ in a single study on a specific topic. want to inv ­
pleting the design stage that you are ready to Science is an ongoing enterprise that builds with certain
gather the data or evidence (e.g., ask people the on prior research and builds a larger, collectively lecting data. In
questions, record answers, etc.). Once you have created body of knowledge. Anyone study is a mensions of
very carefully collected the data, your next step is small part of the much larger whole of science. A understand the
to manipulate or analyze the data. This will help single researcher may be working on multiple
you see any patterns in it and help you to give research projects at once, or several researchers
meaning to or interpret the data (e.g., "People may collaborate on one project. Likewise, one Use of R.eselill1l
who marry young and grew up in families with project may result in one scholarly article or sev­ For over a
abuse have higher rates of physical domestic eral, and sometimes several smaller projects are Some reseal'
abuse than those with different family histo­ reported in a single article. entific, and
ries"). Finally, you must inform others by writing more activist. ~
a report that describes the study's background, oriented. This ~
how you conducted it, and what you discovered. searchers in the ~
The seven-step process shown in Figure 1.1 tain friendly rdl
is oversimplified. In practice, you will rarely Three years after they graduated from college. from one wing ~
complete one step totally then leave it behind to Tim and Sharon met for lunch. Tim asked their careers. In_
move to the next step. Rather, the process is in­ Sharon, "So, how is your new job as a researcher concentrate on!
teractive in which the steps blend into each for Social Data, Inc.? What are you doing?" over the long •
other. What you do in a later step may stimulate Sharon answered, "Right now I'm working on studies to SONel
you to reconsider and slightly adjust your think­ an applied research project on day care quality in Those who CODIII
ing in a previous one. The process is not strictly which we're doing a cross-sectional survey to get dam ental natura
linear and may flow back and forth before reach­ descriptive data for an evaluation study." Sharon basic research..

touched on four dimensions of social research as Basic Research. Basic social research advances
she described her research on day care. fundamental knowledge about the social world.
Social research comes in several shapes and Basic researchers focus on refuting or support­
sizes. Before you begin a study, you will need to ing theories that explain how the social world
make several decisions about the specific type of operates, what makes things happen, why social
research you are going to conduct. Researchers relations are a certain way, and why society
need to understand the advantages and disad­ changes. Basic research is the source of most new
vantages of each type, although most end up scientific ideas and ways of thinking about the
specializing in doing one type. We can think of world. Many nonscientists criticize basic re­
the types as fitting into one of the categories in search and ask, "What good is it?" and consider
each of four dimensions of research. it to be a waste oftime and money. Although ba­
The first dimension is a distinction in how sic research often lacks a practical application in
research is used, or between applied and basic re­ the short term, it provides a foundation for
search. The next is the purpose of doing research, knowledge that advances understanding in
or its goal, to explore, describe, or explain. The many policy areas, problems, or areas of study.
next two dimensions are more specific: how time Basic research is the source of most of the tools,
is incorporated into the study design, and the methods, theories, and ideas about underlying
specific data collection technique used. causes of how people act or think used by ap­
The dimensions overlap, in that certain di­ plied researchers. It provides the major break­
mensions are often found together (e.g., the goal throughs that significant advances in knowledge;
of a study and a data collection technique). Once it is the painstaking study of broad questions
you learn the dimensions, you will begin to see that has the potential of shifting how we think
how the particular research questions you might about a wide range of issues. It may have an im­
want to investigate tend to be more compatible pact for the next 50 years or century. Often, the
with certain ways of designing a study and col­ applications ofbasic research appear many years
lecting data. In addition, being aware of the di­ or decades later. Practical applications may be
mensions of research will make it easier to apparent only after many accumulated advances
understand the research reports by others. in basic knowledge build over a long time pe­
riod. For example, in 1984, Alec Jeffreys, a ge­
neticist at the University of Leicester in England,
Use of Research
was engaged in basic research studying the evo­
For over a century, science has had two wings. lution ofgenes. As an indirect accidential side ef­
Some researchers adopt a detached, purely sci­ fect of a new technique he developed, he
entific, and academic orientation; others are discovered a way to produce what is now call hu­
more activist, pragmatic, and interventionist man DNA "fingerprints" or unique markings of
oriented. This is not a rigid separation. Re­ the DNA of individuals. This was not his intent.
searchers in the two wings cooperate and main­ He even said he would have never thought ofthe
tain friendly relations. Some individuals move technique if DNA fingerprints had been his goal.
from one wing to another at different stages in Within 10 years applied uses of the technique
their careers. In simple terms, some researchers were developed. Today, DNA analysis is a widely
concentrate on advancing general knowledge used technique in criminal investigations.
over the long term, whereas others conduct
studies to solve specific, immediate problems. Applied Research. Applied socialresearch is de­
Those who concentrate on examining the fun­ signed to address a specific concern or to offer
damental nature of social reality are engaged in solutions to a problem identified by an em­
basic research. ployer, club, agency, social movement, or orga­


nization. Applied social researchers are rarely wisely.Sometimes despite serious problems with
concerned with building, testing, or connecting a study's methodology and cautions from the re­
to a larger theory, developing a long-term gen­ searchers, politicians use results to justify cutting
eral understanding, or carrying out a large-scale programs they dislike or to advance programs
investigation that might span years. Instead, they they favor. Because applied research often has
usually conduct a quick, small-scale study that immediate implications or involves controver­
provides practical results for use in the short sial issues, it often generates conflict. One
term (i.e., next month or next year). For exam­ famous researcher, William Whyte (1984), en­
ple, the student government of University X countered conflict over findings in his applied
wants to know whether the number of Univer­ research on a factory in Oklahoma and on
sity X students who are arrested for driving while restaurants in Chicago. In the first case, the
intoxicated or involved in auto accidents will de­ management was more interested in defeating a
cline if it sponsors alcohol-free parties next year. union than in learning about employment rela­
Applied research would be most applicable for tions; in the other, restaurant owners really
this situation. sought to make the industry look good and did
People employed in businesses, government not want findings on the nitty-gritty of its oper­
offices, health care facilities, social service agen­ ations made public.
cies, political organizations, and educational in­ Applied and basic researchers adopt differ­
stitutions often conduct applied research and ent orientations toward research methodology
use the results in decision making. Applied re­ (see Table 1.2). Basic researchers emphasize high
search affects decisions such as the following: methodological standards and try to conduct
Should an agency start a new program to reduce near-perfect research. Applied researchers must
the wait time before a client receives benefits? make more tradeoffs. They may compromise
Should a police force adopt a new type of re­ scientific rigor to get quick, usable results, but
sponse to reduce spousal abuse? Should a politi­ compromise is never an excuse for sloppy re­
cal candidate emphasize his or her stand on the search. Applied researchers try to squeeze re­
environment instead of the economy? Should a search into the constraints of an applied setting tals, goveTnllDelnr,J
company market a skin care product to mature and balance rigor against practical needs. Such demonstrate the
adults instead of teenagers? balancing requires an in-depth knowledge of re­ doing. An es
The scientific community is the primary search and an awareness of the consequences of techniques .
consumer of basic research. The consumers of compromising standards. researchers. The
applied research findings are practitioners such decision makers.
as teachers, counselors, and social workers, or Types of Applied Research. There are many themselves, d
decision makers such as managers, agency ad­ specific types of applied research. Here, you will research. Also~·
ministrators, and public officials. Often, some­ learn about three major types: evaluation, ac­ a practical si .
one other than the researcher who conducted tion, and social impact assessment. Evaluation
the study uses the results. elude: Does a ~
Applied research results are less likely to en­ Evaluation Research Study. Evaluation research improve learning ~
ter the public domain in publications and may study is applied research designed to find out forcement pr~
be available only to few decision makers or prac­ whether a program, a new way of doing some­ spouse abuse? ~
titioners. This means that applied research find­ thing, a marketing campaign, a policy, and so employee prod~
ings often are not widely disseminated and that forth, is effective-in other words, "Does it measure the effedll
well-qualified researchers rarely get to judge the work?" The most widely used type of applied re­ or way of doing ~
quality of applied studies. search is evaluation research." This type of re­ research technique
The decision makers who use the results of search is widely used in large bureaucratic can be used, the e:IIj
an applied study mayor may not use them organizations (e.g., businesses, schools, hospi­ ally preferred. Prad

TAB L E 1.2 Basic and Applied Social Research Compared

1. Research is intrinsically satisfying and 1. Research is part of a job and is judged by

judgments are by other sociologists. sponsors who are outside the discipline of
2. Research problems and subjects are selected sociology.
with a great deal of freedom. 2. Research problems are "narrowly constrained"
3. Research is judged by absolute norms of to the demands of employers or sponsors.
scientific rigor, and the highest standards of 3. The rigor and standards of scholarship depend
scholarship are sought. on the uses of results. Research can be "quick
4. The primary concern is with the internal logic and dirty" or may match high scientific
and rigor of research design. standards.

5. The driving goal is to contribute to basic, 4. The primary concern is with the ability to
theoretical knowledge. generalize findings to areas of interest to
6. Success comes when results appear in a
scholarly journal and have an impact on others 5. The driving goal is to have practical payoffs or
in the scientific community. uses for results.
6. Success comes when results are used by
sponsors in decision making.

Source: Adapted from Freeman and Rossi (' 984:572-573).

tals, government, large nonprofit agencies) to icy or program may conduct evaluation research
demonstrate the effectiveness of what they are for their own information or at the request of
doing. An evaluation researcher does not use outside decision makers. The decision makers
techniques different from those of other social may place limits on the research by fixing
researchers. The difference lies in the fact that boundaries on what can be studied and by de­
decision makers, who may not be researchers termining the outcome of interest. This often
themselves, define the scope and purpose of the creates ethical dilemmas for a researcher.
research. Also, their objective is to use results in Ethical and political conflicts often arise in
a practical situation.f evaluation research because people can have op­
Evaluation research questions might in­ posing interests in the findings. The findings of
clude: Does a Socratic teaching technique research can affectwho gets or keeps a job, it can
improve learning over lecturing? Does a law-en­ build political popularity, or it may help pro­
forcement program of mandatory arrest reduce mote an alternative program. People who are
spouse abuse? Does a flextime program increase personally displeased with the findings may at­
employee productivity? Evaluation researchers tack the researcher or his or her methods.
measure the effectiveness of a program, policy, Evaluation research has several limitations:
or way of doing something and often use several The reports of research rarely go through a peer
research techniques (e.g., survey and field). If it review process, raw data are rarely publicly avail­
can be used, the experimental technique is usu­ able, and the focus is narrowed to select inputs
ally preferred. Practitioners involved with a pol­ and outputs more than the full process hywhich

a program affects people's lives. In addition, de­
cision makers may selectivelyuse or ignore eval­
mestic violence that will be discussed shortly as
an explanatory study example (Cherlin et al., ., ,
uation findings. 2004) testified in the United States Senate. The
study findings and the testimony helped to alter
Action Research Study. Action research is ap­ marriage promotion provisions in a 2005 wel­
plied research that treats knowledge as a form of fare reform law.?
power and abolishes the division between creat­
ing knowledge and using knowledge to engage in Social Impact Assessment Research Study. A re­
political action. There are several types of action searcher who conducts social impact assessment
research, but most share five characteristics: (1) (SIA) estimates the likely consequences of a
the people being studied actively participate in planned intervention or intentional change to
the research process; (2) the research incorpo­ occur in the future. It may be part ofa larger en- .
rates ordinary or popular knowledge; (3) the re­ vironmental impact statement required by gov­
search focuses on issues of power; (4) the ernment agencies and used for planning and
research seeks to raise consciousness or increase making choices among alternative policies. He
awareness of issues; and (5) the research is tied or she forecasts how aspects of the social envi­
directly to a plan or program of political action. ronment may change and suggests ways to miti­
Action research tends to be associated with a so­ gate changes likely to be adverse from the point
cial movement, political cause, or advocacy for ofview of an affected population. Impacts are the
an issue. It can be conducted to advance a range difference between a forecast of the future with
of political positions. Some action research has the project or policy and without the project or
an insurgent orientation with goals of empower­ policy. For example, the SIA might estimate the
ing the powerless, fighting oppression and injus­ ability of a local hospital to respond to an earth­
tice, and reducing inequality. Wealthy and quake, determine how housing availability for
powerful groups or organizations also sponsor the elderly will change if a major new highway is
and conduct action research to defend their sta­ built, or assessthe impact on college admissions
tus, position, and privileges in society. if students receive interest-free loans. Re­
Most action researchers are explicitly politi­ searchers who conduct SIAs often examine a
cal, not value neutral. Because the primary goal range of social outcomes and work in an inter­
is to affect sociopolitical conditions, publishing disciplinary research team to estimate the social
results in formal reports, articles, or books is sec­ outcomes. The outcomes include measuring
ondary. Most action researchers also believe that "quality of life" issues, such as access to health •
knowledge develops from direct experience, par­ care, illegal drug and alcohol use, employment
ticularly the experience of engaging in sociopo­ opportunities, schooling quality, teen pregnancy
litical action. rates, commuting time and traffic congestion, • Create a g
For example, most feminist research is ac­ availability of parks and recreation facilities, picture ofe
tion research. It has a dual mission: to create so­ shopping choices, viable cultural institutions, • Formulate and ~
cial change by transforming gender relations and crime rates, interracial tensions, or social isola­ questions few' ~
to contribute to the advancement of knowledge. tion. There is an international professional asso­ • Generate new ~
A feminist researcher who studies sexual harass­ ciation for SIA research that advances SIA conjectures.. Of" ~
ment might recommend policy changes to re­ techniques and promotes SIA by governments, • Determine the til
duce it as well as to inform potential victims so corporations, and other organizations. conducting ~
they can protect themselves and defend their Social impact assessments are rarely re­ • Develop technici
rights. At times, researchers will explain study quired, but a few governments mandate them. measuring and Id
results in a public hearing to try to modify new For example, in New South Wales, Australia, a data.
policies or laws. The authors of a study on do­ registered club or hotel cannot increase the

number of poker machines unless the Liquor lence), economic (e.g., unemployment, bank­
Administration Board in the Department Gam­ ruptcy, tourism expansion), and cultural im­
ing and Racing approves an SIA for the club or pacts (e.g.,time awayfrom other leisure activity)
hotel. The SIA enables the board to assess the listed by their effect on all gamblers, problem
likely local community impact from increasing gamblers, the local community, and the
the number of poker machines. The format in­ region. 10
cludes a matrix that allows the board to identify
the social and economic impacts, positive and
Purpose of a Study
negative, financial or nonfinancial, quantified or
qualitative. In New Zealand, the Gambling Act If you ask someone why he or she is conducting
of 2003 requires an SIA before expanding gam­ a study, you might get a range of responses: "My
bling. In one 2004 study in New Zealand for the boss told me to"; "It was a class assignment"; "I
Auckland City Council, it noted that 90 percent was curious"; "My roommate thought it would
of New Zealand's adults gamble, 10 percent gam­ be a good idea." There are almost as many rea­
ble regularly (once a week or more often), and sons to do research as there are researchers. Yet,
about 1 percent are problem gamblers, although the purposes of social research may be organized
this varies by age, income, and ethnicity. The into three groups based on what the researcher is
SIA recommended limiting the locations of new trying to accomplish-explore a new topic, de­
gambling venues, monitoring their usage, and scribe a social phenomenon, or explain why
tracing the amount of gambling revenues that something occurs. Studies may have multiple
are returned to the community in various ways purposes (e.g., both to explore and to describe),
(e.g., clubs, trusts, etc.). It contained a matrix but one of three major purposes is usually dom­
with social (e.g, arrests, divorce, domestic vio­ inant (see Box 1.2).

Exploratory Descriptive Explanatory

• Become familiar with the • Provide a detailed, highly • Test a theory's predictions or
basic facts, setting, and accurate picture. principle.
concerns. • Locate new data that • Elaborate and enrich a
• Create a general mental contradict past data. theory's explanation.
picture of conditions. • Create a set of categories or • Extend a theory to new issues
• Formulate and focus classify types. or topics.
questions for future research. • Clarify a sequence of steps or • Support or refute an
• Generate new ideas, stages. explanation or prediction.
conjectures, or hypotheses. • Document a causal process • Link issues or topics with a
• Determine the feasibility of or mechanism. general principle.
conducting research. • Report on the background or • Determine which of several
• Develop techniques for context of a situation. explanations is best.
measuring and locating future

Exploration. Perhaps you have explored a new Description. Perhaps you have a more highly .-Idrca ......
topic or issue in order to learn about it. If the is­ developed idea about a social phenomenon and J*x. .~...,
sue was new or no researchers had written about want to describe it. Descriptive research presents pIIICdS;~. . .
it, you began at the beginning. In exploratory re­ a picture of the specific details of a situation, so­ IOpercaaai..
search, a researcher examines a new area to for­ cial setting, or relationship; it focuses on "how?"
.:ooditiom __
mulate precise questions that he or she can and "who?" questions: "How did it happen?"
address in future research. Exploratory research "Who is involved?" A great deal of social re­ pb:oatory~
may be the first stage in a sequence of studies. A search is descriptive. Descriptive researchers use ears are ......
researcher may need to conduct an exploratory most data-gathering techniques-surveys, field Cberiin, . . . .
study in order to know enough to design and ex­ research, content analysis, and historical-com­ ~1~~~
• 'iii
ecute a second, more systematic and extensive parative research. Only experimental research is usinga_~
study. It addresses the "what?" question: "What less often used. Much of the social research phvsical abase.. 1
is this social activity really about?" found in scholarly journals or used for making women ,.ith ~
Many higher-education officials are con­ policy decisions is descriptive. likely IIJ.alIT
cerned about college students' low retention Descriptive and exploratory research often The authors
rates, especially students from minority-disad­ blur together in practice. In descriptive research, abused~
vantaged social backgrounds. For example, of a researcher begins with a well-defined subject to resist 01"
Latinos who enroll in college, 80 percent leave and conducts a study to describe it accurately more likely 10
without receiving a degree. Officials seek ways to and the outcome is a detailed picture of the sub­ guilt, and low"
reduce dropouts and increase the chances that ject. The results may indicate the percentage of rionofh~.
students who begin college will stay until they people who hold a particular view or engage in sive experience l
earn a degree. Garza and Landeck (2004) con­ specific behaviors-for example, that 8 percent distance and a I
ducted an exploratory study of over 500 Latino ofparents physically or sexually abuse their chil­ commitments, ..
students at a college along the Texas-Mexico dren. A descriptive study presents a picture of rive data gatheraj
border who had dropped out. They wanted to types of people or of social activities. inthreeci~
learn the influencing factors and rationales in Stack, Wasserman, and Kern (2004) con­ nio-thev~
student decision making. The authors discovered ducted a descriptive study on pornography use perienced P'& l
that the primary factors and rationales were un­ on the Internet by people in the United States. married,~~
related to teaching quality or university services. They found that the greatest users were those weremost.~
Instead, the students who dropped out had been with weak social bonds. More specifically, the
overwhelmed by personal problems or had seri­ types of people who were adult users of pornog­ found the
ous difficulties with family or job responsibilities. raphy tended to be males with unhappy mar­ an adult were
Such factors were a major reason given by over riages and weak ties to organized religion. women who half
80 percent of the students who dropped out. Pornography users were also more likely to have less likely to lea1lll
Exploratory researchers tend to use qualita­ engaged in nonconventional sexual behavior ries of unstable. •
tive data and not be wedded to a specific theory (i.e., had an extramarital affair or engaged in
or research question. Exploratory research rarely paid sex) but not other forms of deviance, such
yields definitive answers. If you conduct an ex­ as illegal drug use. TimeDim~

ploratory study, you may get frustrated and feel An awareness of!
it is difficult because there are few guidelines to Explanation. When you encounter an issue mension will hdf
follow. Everything is potentially important, the that is well recognized and have a description of This is because diiI
steps are not well defined, and the direction of it, you might begin to wonder why things are the sues incorporate!
inquiry changes frequently. You need to be cre­ way they are. Explanatory research identifies the give a snapshot d
ative, open-minded, and flexible; adopt an in­ sources of social behaviors, beliefs, conditions, allow you to anal)
vestigative stance; and explore all sources of and events; it documents causes, tests theories, Other studies pI1l
information. and provides reasons. It builds on exploratory you follow eYed

and descriptive research. For example, an ex­ over several time points (longitudinal). Quanti­
ploratory study discovers a new type of abuse by tative studies generally Look at many cases, peo­
parents; a descriptive researcher documents that ple, or units, and measure limited features about
10 percent of parents abuse their children in this them in the form of numbers. By contrast, a
new way and describes the kinds of parents and qualitative study usually involves qualitative
conditions for which it is most frequent; the ex­ data and examines many diverse features of a
planatory researcher focuses on why certain par­ small number of cases across either a short or
ents are abusing their children in this manner. long time period (see Figure 1.2).
Cherlin, Burton, Hurt, and Purvin (2004) ex­
plained instability in marriage or cohabitation Cross-Sectional Research. Most social re­
using a woman's past experience with sexual or search studies are cross-sectional; they examine a
physical abuse. They tested the hypothesis that single point in time or take a one-time snapshot
women with a history of abuse would be less approach. Cross-sectional research is usually the
likely marry than those without such histories. simplest and least costly alternative. Its disad­
The authors reasoned that those who were vantage is that it cannot capture social processes
abused have fewer social supports and resources or change. Cross-sectional research can be ex­
to resist or avoid abusive partners, and they are ploratory, descriptive, or explanatory, but it is
more likely to harbor feelings of self-blame, most consistent with a descriptive approach to
guilt, and low self-esteem that inhibit the forma­ research. The descriptive study by Stack,
tion of healthy romantic relationships. An abu­ Wasserman, and Kern (2004) on pornography
sive experience also creates greater emotional use was cross-sectional, based on a national U.S.
distance and a hesitancy to make long-term survey conducted in 2000.
commitments. Using quantitative and qualita­
tive data gathered in Low-income neighborhoods Longitudinal Research. Researchers using
in three cities-Boston, Chicago, and San Anto­ longitudinal research examine features of people
nio-they found that adult women who had ex­ or other units at more than one time. It is usually
perienced past abuse were less likely to be more complex and costly than cross-sectional
married, and those with multiple forms of abuse research, but it is also more powerful and infor­
were most likelyto remain single, It appears that mative. Descriptive and explanatory researchers
women without a past history of abuse who use longitudinal approaches. Let us now look at
found themselves in an abusive relationship as the three main types of longitudinal research:
an adult were likely to withdraw from it, but time series, panel, and cohort.
women who had been abused as children were
less likely to leave and tended to enter into a se­ Time-Series Study. A time-series study is longi­
ries of unstable, transitory relations. tudinal research in which a researcher gathers
the same type of information across two or more
time periods. Researchers can observe stability
Time Dimension in Research
or change in the features of the units or can track
An awareness of how a study uses the time di­ conditions over time. The specific individuals
mension will help you read or conduct research. may change but the overall pattern is clear. For
This is because different research questions or is­ example, there has been a nationwide survey of a
sues incorporate time differently. Some studies large sample of incoming freshman students
give a snapshot of a single, fixed time point and since 1966. Since it began, over 11 million stu­
allowyou to analyzeit in detail (cross-sectional). dents at more than 1,800 colleges participated.
Other studies provide a moving picture that lets The fall 2003 survey of 276,449 students found
you follow events, people, or social relations many facts and trends, such as only 34 percent of

Fie U R E 1.2 The Time Dimension in Social Research
CROSS-SECTIONAL: Observe a collection of people at one time.

February 2007

TIME SERIES: Observe different people at multiple times.

1950 1970 1990 2010

PANEL: Observe the exact same people at two or more times.
~ ....
P «Mill.
~ .-m.:rM:IIIII

1986 1996 2006

COHORT: Observe people who shared an experience at two or more times.

Married in 1967 1987 2007

CASE STUDY: Observe a small set intensely across time .

• ~r


entering freshmen studied six or more hours per term panel studies can clearly show the impact
week. This was the lowest level since the ques­ of a particular life event. For example, Oesterle,
tion was asked in 1987 (when it was 47 percent). Johnson, and Mortimer (2004) examined panel
Yet, alcohol consumption was down. In 2003, data from a longitudinal study that began in
44.8 percent reported drinking beer, which rep­ 1988 with 1,000 ninth-grade students enrolled
resented a steady decline from 73.7 percent in in the St. Paul, Minnesota, public school district
1982. In 2003, freshmen were more interested in and looked at volunteering activities during late
keeping up with politics. The 33.9 percent who adolescence and young adulthood, covering
said it was very important to stay politically in­ nine years from age 18-19 (1992) to age 26-27
formed was up from a low of 28.1 percent in (2000). They found that volunteering at an ear­
2000, and 22.5 percent said they discussed poli­ lier stage strongly affected whether one volun­
tics regularly, up from 19.4 percent in 2002 teered at a later stage. Also, people who devoted
(which had been the highest since a low point in full time to working or parenting at an earlier
1993). These figures are still far lower than the stage (18-19 years old) were less likely to volun­
60.3 percent who expressed an interest in politics teer at a later stage (26-27 years old) than those
in 1966, or the one-third who discussed politics whose major activity was attending school.
regularly in 1968. The importance of family has
steadily increased over the years, with 74.8 per­ Cohort Study. A cohort study is similar to a
cent of students calling it essential or very im­ panel study, but rather than observing the exact
portant. This is up from the low point of 58.8 same people, the study focuses on a category of
percent in 1977 when the question was first people who share a similar life experience in a
asked. However, religious involvement declined. specified time period. Researchers examine the
The percentage of students who attended reli­ category as a whole for important features and
gious services regularly was at its lowest level in focus on the cohort, or category, not on specific
35 years. In addition, the percent claiming individuals. Commonly used cohorts include all
"none" as a religious preference reached a record people born in the same year (called birth co­
high of 17.6 percent, compared to a record low of horts), all people hired at the same time, and all
6.6 percent in 1966. Another trend over the past people who graduate in a given year. Unlike
two decades has been a steady growth in opposi­ panel studies, researchers do not have to find the
tion to the death penalty. Nearly one in three in­ exact same people for cohort studies; rather,
coming students advocated ending capital they need only to identify those who experienced
punishment. This is the highest score since 1980 a common life event. In a study of Generation X
(when it was 33.2 percent), although the percent in the United States, Andolina and Mayer (2003)
withholding an opinion was far higher earlier in focused on the cohort of people born between
time; it exceeded 60 percent in the 1970. 11 1967 and 1974. They compared 10 birth cohorts
at different time periods over several decades,
Panel Study. The panel study is a powerful type tracing questions across 24 years. The authors
of longitudinal research in which the researcher found that White Xers are distinct in their
observes exactly the same people, group, or or­ support for school racial integration and for
ganization across multiple time points. It is government action to enforce such efforts, com­
more difficult to conduct than time-series re­ pared to other birth cohorts, but not in their at­
search. Panel research is formidable to conduct titudes toward employment opportunities or
and very costly. Tracking people over time is of­ affirmative action. Despite greater general sup­
ten difficult because some people die or cannot port than other cohorts for equality through in­
be located. Nevertheless, the results of a well-de­ tegration, it does not extend to issues beyond the
signed panel study are very valuable. Even short- schoolyard.

CaseStudies. In cross-sectional and longitudi­ elude experiments, surveys, content analyses,

nal research, a researcher examines features on and existing statistics.
many people or units, either at one time period
or across time periods, and measures several Experiments. Experimentalresearch closely fol­
common features on them, often using num­ lows the logic and principles found in natural
bers. In case-study research, a researcher exam­ science research; researchers create situations
ines, in depth, many features of a few cases over and examine their effects on participants. A re­
. I a duration of time with very detailed, varied, and searcher conducts experiments in laboratories or
extensive data, often in a qualitative form. The in real life with a relatively small number of peo­
researcher carefully selects a few key cases to il­ ple and a well-focused research question. Exper­
lustrate an issue and study it (or them) in detail iments are most effective for explanatory
and considers the specific context of each case. research. In the typical experiment, the re­
This contrasts with other longitudinal studies in searcher divides the people being studied into
which the researcher gathers data on many units two or more groups. He or she then treats both
or cases, then looks for general patterns in the groups identically, except that one group but not
mass of numbers. the other is given a condition he or she is inter­
For example, Snow and Anderson (1992) ested in: the "treatment." The researcher mea­
conducted a case study on homeless people in sures the reactions of both groups precisely. By
Austin, Texas. It provided a wealth of details controlling the setting for both groups and giv­
about the lives and conditions of homeless peo­ ing only one group the treatment, the researcher
ple, identified several types of homeless people, can conclude that any differences in the reac­
outlined the paths by which they became home­ tions of the groups are due to the treatment
less, and discussed several processes that kept alone.
them homeless. This case study used many types
of detailed qualitative and quantitative data, Surveys. A survey researcher asks people ques­
with exploratory, descriptive, and explanatory tions in a written questionnaire (mailed or
phases to reveal a great amount of unexpected handed to people) or during an interview and
and new information. 12 then records answers. The researcher manipu­
lates no situation or condition; he or she simply
asks many people numerous questions in a short
Data Collection Techniques
time period. Typically, he or she then summa­
Social researchers collect data using one or more rizes answers to questions in percentages, tables,
specific techniques. This section givesyou a brief or graphs. Researchers use survey techniques in
overview of the major techniques. In later chap­ descriptive or explanatory research. Surveys give
ters, you will read about these techniques in de­ the researcher a picture of what many people Qualitative ~
tail and learn how to use them. Some techniques think or report doing. Survey researchers often Techniques for'll
are more effective when addressing specific use a sample or a smaller group of selected peo­ elude field researd
kinds of questions or topics. It takes skill, prac­ ple (e.g., 150 students), but generalize results to research.
tice, and creativity to match a research question a larger group (e.g., 5,000 students) from which
to an appropriate data collection technique. The the smaller group was selected. Survey research FieldResearch. ~
techniques fall into two categories based on is very widely used in many fields. case studies lookD
whether the data being gathered are quantitative over a length of
or qualitative. Content Analyses. A content analysis is a tech­ years). A field re:sl!l
nique for examining information, or content, in formulated idea 01
Quantitative Data Collection Techniques. written or symbolic material (e.g., pictures, or natural setting
Techniques for quantitative data collection in- movies, song lyrics, etc.). In content analysis, a adopts a social r~

researcher first identifies a body of material to in detail. The researcher gets to know personally
analyze (e.g.,books, newspapers, films, etc.) and the people being studied, may conduct open­
then creates a system for recording specific as­ ended and informal interviews, and takes de­
pects of it. The system might include counting tailed notes on a daily basis. After leaving the
how often certain words or themes occur. Fi­ field site, the researcher carefully rereads the
nally, the researcher records what was found in notes and prepares written reports. Field re­
the material. He or she often measures informa­ search is used most often for exploratory and de­
tion in the content as numbers and presents it as scriptive studies; it is rarely used for explanatory
tables or graphs. This technique lets a researcher research.
discover features in the content oflarge amounts
of material that might otherwise go unnoticed. Historical-Comparative Research. Historical­
Researchers can use content analysis for ex­ comparative researchers examine aspects of social
ploratory and explanatory research, but primar­ life in a past historical era or across different cul­
ily it is used for descriptive research. tures. Researchers who use this technique may
focus on one historical period or several, com­
ExistingStatistics. In existingstatistics research, pare one or more cultures, or mix historical pe­
a researcher locates previously collected infor­ riods and cultures. Like field research, a
mation, often in the form of government reports researcher combines theory building/testing
or previously conducted surveys, then reorga­ with data collection and begins with a loosely
nizes or combines the information in new ways formulated question that is refined during the
to address a research question. Locating sources research process. Researchers often gather a
can be time consuming, so the researcher needs wide array of evidence, including existing statis­
to consider carefully the meaning of what he or tics and dowments (e.g., novels, officialreports,
she finds. Frequently, a researcher does not books, newspapers, diaries, photographs, and
know whether the information of interest is maps) for study. In addition, they may make di­
available when he or she begins a study. Some­ rect observations and conduct interviews. His­
times, the existing quantitative information con­ torical-comparative research can be exploratory,
sists of stored surveys or other data that a descriptive, or explanatory and can blend types.
researcher reexamines using various statistical
procedures. Existing statistics research can be
used for exploratory, descriptive, or explanatory
purposes, but it is most frequently used for de­
scriptive research. This chapter gave you an overview of social re­
search. You saw how social research differs from
Qualitative Data Collection Techniques. the ordinary ways of learning-knowing about
Techniques for qualitative data collection in­ the social world, how doing research is based on
clude field research and historical-comparative science and the scientific community, and about
research. several types of social research based on its di­
mensions (e.g., its purpose, the technique used
Field Research. Most field researchers conduct to gather data, etc.). The dimensions of research
case studies looking at a small group of people loosely overlap with each other. The dimensions
over a length of time (e.g., weeks, months, of social research are a kind of "road map" to
years). A field researcher begins with a loosely help you make your way through the terrain of
formulated idea or topic, selects a social group social research. In the next chapter, we turn to
or natural setting for study, gains access and social theory. You read about it a little in this
adopts a social role in the setting, and observes chapter. In the next chapter, you will learn how

theory and research methods work together and social impact assessment study
about several types of theory. social research
survey research
time-series study
Key Terms

action research study Endnotes

applied social research
basic social research 1. See Rampton and Stauber (2001:247-277 and
case study 305-306).
cohort study 2. See Best (2001:15) on advocates and media.
cross-sectional research 3. See National Science Board (2002:735-739).
4. Schacter (2001) provides a summary of memory
descriptive research
5. National Science Board (2002:739).
empirical evidence 6. Discussions of the scientific community can be
evaluation research study found in Cole and Gordon (1995), Crane (1972),
existing statistics research Hagstrom (1965), Merton (1973), Mulkay (1991),
experimental research and Ziman (1999).
explanatory research 7. See Patton (200l) and Weiss (1997) for a more
exploratory research detailed discussion of recent advances in evalua­
field research tion research.
halo effect 8. Beck (1995) provides a useful overview.
historical comparative research 9. See Herring and Ebner (2005) on the use of do­
mestic violence study findings. .
longitudinal research
10. See Adams (2004) for more information on the
Auckland City study.
panel study 11. See the website at www.gseis.ucla.edu/heri/heri.
premature closure html.
qualitative data 12. Also see Snow and Anderson (1991) for a discus­
quantitative data sion of the case-study method in their study of
scientific community homeless people. Also see George and Bennett
scientific method (2005) on the case-study method generally.
selective observation
Theory and Social Research

What Is Theory?
Blame Analysis
The Parts of Theory
The Aspects of Theory
Direction of Theorizing
Range of Theory
Levels of Theory
Forms of Explanation
The Three Major Approaches to Social Science
Positivist Approach
Interpretive Approach
Critical Approach
The Dynamic Duo


Suppose you want to make sense of the hostility
between people of different races. Trying to un­
derstand it, you ask a teacher, who responds:
in mind three things about how social scientific
theories work. First, social theories explain re­
curring patterns, not unique or one-time events.
For example, they are not good for explaining
why terrorists decided to attack New York's
World Trade Center on September 11, 2001, but
~- 1JIr

~*- ...

Most raciallyprejudiced people learn nega­ they can explain patterns, such as the conditions
tive stereotypes about another racial group that generally lead to increased levels of fear and
from their families, friends, and others in feelings of patriotism in a people. Second, social
their immediate surroundings. If they lack theories are explanations for aggregates, not par­
sufficient intimate social contact with mem­ ticular individuals. Aggregates are collections of
bers of the group or intense information that many individuals, cases, or other units (e.g.,
contradicts those stereotypes,they remain businesses, schools, families, clubs, cities, na­
prejudiced. tions, etc.). A social theory rarely can explain
why Josephine decided to major in nursing
This makes sense to you because it is consis­ rather than engineering, but it can explain why
tent with what you know about how the social females more than males in general choose nurs­
world works. This is an example of a small-scale ing over engineering as a major. Third, social
social theory, a type that researchers use when theories state a probability, chance, or tendency
conducting a study. for events to occur, rather than state that one
What do you think of when you hear the event must absolutely follow another. For exam­
word theory? Theory is one of the least well un­ ple, instead of stating that when someone is
derstood terms for students learning social sci­ abused as a child, that person will always later
ence. My students' eyelids droop if! begin a class
by saying, "Today we are going to examine the
theory of ..." The mental picture many students
have of theory is something that floats high
among the clouds. My students have called it "a
abuse his or her own children, a theory might
state that when someone experiences abuse dur­
ing his or her childhood, that person will tend to
or is more likely to become an abusive parent
when an adult. Likewise, it might state that peo­

SF '&$.

tangled maze of jargon" and "abstractions that ple who did not experience childhood abuse tie~

...-- .

are irrelevant to the real world." might become abusive parents, but they are less e
Contrary to these views, theory has an im­ likely to than someone who has experienced
portant role in research and is an essential ally abuse as a child.
for the researcher. Researchers use theory differ­ ...« - .
ently in various types of research, but some type
of theory is present in most social research. It is
less evident in applied or descriptive than in ba­
sic or explanatory research. In simple terms, re­
searchers interweave a story about the operation In Chapter 1, social theory was defined as a sys­
of the social world (the theory) with what they tem of interconnected abstractions or ideas that
observe when they examine it systematically (the condenses and organizes knowledge about the
data). social world. It is a compact way to think of the
People who seek absolute, fixed answers for social world. People are constantly developing
a specific individual or a particular one-time new theories about how the world works.
event may be frustrated with science and social Some people confuse the history of social
theories. To avoid frustration, it is wise to keep thought, or what great thinkers said, with social

theory. The classical social theorists (e.g., Almost all research involves some theory, so
Durkheim, Weber, Marx, and Tennies) played the question is less whether you should use the­
an important role in generating innovative ideas. ory than how you should use it. Being explicit
They developed original theories that laid the about the theory makes it easier to read someone
foundation for subsequent generations of social else's research or to conduct your own. An
thinkers. People study the classical theorists be­ awareness of how theory fits into the research
cause they provided many creative and interre­ process produces better designed, easier to un­
lated ideas at once. They radically changed the derstand, and better conducted studies. Most re­
way people understood and saw the social world. searchers disparage atheoretical or "crude
We study them because geniuses who generate empiricist" research.
many original, insightful ideas and fundamen­
tally shift how people saw the social world are
Blame Analysis
At times people confuse theory with a hunch Blameanalysis is a type of counterfeit argument
or speculative guessing. They may say, "It's only presented as if it were a theoretical explanation.
a theory" or ask, "What's your theory about it?" It substitutes attributing blame for a causal ex­
This lax use of the term theory causes confusion. planation that is backed by supporting empirical
Such guessing differs from a serious social the­ evidence. Blame belongs to the realm of making
ory that has been carefully built and debated moral, legal, or ideological claims. It implies an
over many years by dozens of researchers who intention, negligence, or responsibility for an
found support for the theory's key parts in re­ event or situation (usually an unfavorable one).
peated empirical tests. A related confusion is It shifts the focus from Why did it occur? to
when what people consider to be a "fact" (i.e., Who is responsible? Blame analysis assumes
light a match in a gasoline-filled room and it will there is a party or source to which a fixed
explode) is what scientists call a theory (i.e., a amount of responsibility can be attached. The
theory of how combining certain quantities of goal of inquiry is to identify a responsible party.
particular chemicals with oxygen and a level of Often, some sources are exempted or shielded.
heat is likely to produce the outcome of explo­ This may be the injured party, members of a
sive force). People use simple theories without sympathetic audience, or a sacred value or
making them explicit or labeling them as such. principle.
For example, newspaper articles or television re­ Blame analysis clouds discussion because it
ports on social issues usually have unstated so­ confuses blame with cause; it gives an account
cial theories embedded within them. A news (or story) instead of a logical explanation with
report on the difficultyof implementing a school intervening causal mechanisms; and it fails to
desegregation plan will contain an implicit the­ explore empirical evidence for and against sev­
ory about race relations. Likewise, political lead­ eral alternative causes. Blame analysis first pre­
ers frequently express social theories when they sents an unfavorable event or situation. It could
discuss public issues. Politicians who claim that be a bank is robbed, a group is systematically
inadequate education causes poverty or that a paid less in the labor force, or traffic congestion
decline in traditional moral values causes higher is terrible in an urban area. It next identifies one
crime rates are expressing theories. Compared or more responsible parties, then it provides se­
to the theories of social scientists, such layper­ lective evidence that shields certain parties or
sons' theories are less systematic, less well for­ sources (e.g., employment conditions, the
mulated, and harder to test with empirical choices available to the underpaid group, trans­
evidence. portation policy, and land cost).'

measure height or compare it. A height of zero is

possible, and height can increase or decrease
Concepts over time. As with many words, we use the word
in several ways. Height is used in the expressions
All theories contain concepts, and concepts are the height of the battle, the height of the summer,
the building blocks of theory.f A concept is an and the height offashion.
idea expressed as a symbol or in words. Natural The word height refers to an abstract idea.
science concepts are often expressed in symbolic We associate its sound and its written form with
forms, such as Greek letters (e.g., b) or formulas that idea. There is nothing inherent in the
(e.g., s == d/t; s == speed, d == distance, t == time). sounds that make up the word and the idea it
Most social science concepts are expressed as represents. The connection is arbitrary, but it is
words. The exotic symbols of natural science still useful. People can express the abstract idea
concepts make many people nervous, as the use to one another using the symbol alone.

- ....
of everyday words in specialized social science Concepts have two parts: a symbol (word or
concepts can create confusion. term) and a definition. We learn definitions in
I do not want to exaggerate the distinction many ways. I learned the word height and its de­
between concepts expressed as words and con­ finition from my parents. I learned it as I learned
cepts expressed as symbols. Words, after all, are to speak and was socialized to the culture. My
symbols, too; they are symbols we learn with parents never gave me a dictionary definition. I
language. Height is a concept with which you are learned it through a diffuse, nonverbal, informal
already familiar. For example, I can say the word process. My parents showed me many examples;
height or write it down; the spoken sounds and I observed and listened to others use the word; I
written words are part of the English language. used the word incorrectly and was corrected;
The combination ofletters in the sound symbol­ and I used it correctly and was understood.
izes, or stands for, the idea of a height. Chinese or Eventually, I mastered the concept.
Arabic characters, the French word hauteur, the This example shows how people learn con­
German word hohe, the Spanish word altura­ cepts in everyday language and how we share
all symbolize the same idea. In a sense, a lan­ concepts. Suppose my parents had isolated me
guage is merely an agreement to represent ideas from television and other people, then taught
by sounds or written characters that people me that the word for the idea height was zdged. I
learned at some point in their lives. Learning would have had difficulty communicating with
concepts and theory is like learning a language. 3 others. People must share the terms for concepts e..-.,.u....
Concepts are everywhere, and you use them and their definitions if they are to be ofvalue. ftobtioR.
all the time. Height is a simple concept from Everyday life is filled with concepts, but !:JOUPS. or­
everyday experience. What does it mean? It is many have vague and unclear definitions. Like­ aplSin _
easy to use the concept of height, but describing wise, the values, misconceptions, and experi­ social~_
the concept itself is difficult. It represents an ab­ ences of people in a culture may limit everyday meiatm
stract idea about physical relations. How would concepts. Social scientists borrow concepts from aDy remlloRillll&Jl
you describe it to a very young child or a crea­ everyday culture, but they refine these concepts meaning. f«
ture from a distant planet who was totally unfa­ and add new ones. Many concepts such as cept such as
miliar with it? A new concept from a social sexism, life-style, peer group, urban sprawl, and sociated
theory may seem just as alien when you en­ social class began as precise, technical concepts economic gr;
counter it for the first time. Height is a charac­ in social theory but have diffused into the larger city, revitalJ"tZal-"
teristic of a physical object, the distance from top culture and become less precise. noritiesy. ~
to bottom. All people, buildings, trees, moun­ We create concepts from personal experi­ Someco~
tains, books, and so forth have a height. We can ence, creative thought, or observation. The clas­ quantities. or CIJDII

sical theorists originated many concepts. Exam­ concept are amount ofincome, temperature, den­
ple concepts include family system, gender role, sity ofpopulation,yearsof schooling, and degree of
socialization, self-worth, frustration, and displaced violence. These are called variables, and you will
aggresSlOn. read about them in a later chapter. Other con­
Some concepts, especially simple, concrete cepts express types of nonvariable phenomena
concepts such as book or height, can be defined (e.g., bureaucracy, family, revolution, homeless,
through a simple nonverbal process. Most social and cold). Theories use both kinds of concepts.
science concepts are more complex and abstract.
They are defined by formal, dictionary-type de­ Classification Concepts. Some concepts are
finitions that build on other concepts. It may simple; they have one dimension and vary along
seem odd to use concepts to define other con­ a single continuum. Others are complex; they
cepts, but we do this all the time. For example, I have multiple dimensions or many subparts.
defined heightas a distance between top and bot­ You can break complex concepts into a set of
tom. Top, bottom, and distance are all concepts. simple, or single-dimension, concepts. For ex­
We often combine simple, concrete concepts ample, Rueschemeyer and associates (1992:43­
from ordinary experience to create more ab­ 44) stated that democracy has three dimensions:
stract concepts. Height is more abstract than top (1) regular, free elections with universal suffrage;
or bottom. Abstract concepts refer to aspects of (2) an elected legislative body that controls gov­
the world we do not directly experience. They ernment; and (3) freedom of expression and as­
organize thinking and extend understanding of sociation. The authors recognized that each
reality. dimension varies by degree. They combined the
Researchers define scientific concepts more dimensions to create a set of types of regimes.
precisely than those we use in daily discourse. Regimes very low on all three dimensions are to­
Social theory requires well-defined concepts. talitarian, those high on all three are democra­
The definition helps to link theory with research. cies, and ones with other mixes are either
A valuable goal of exploratory research, and of authoritarian or liberal oligarchies.
most good research, is to clarify and refine con­ Classifications are partway between a single,
cepts. Weak, contradictory, or unclear defini­ simple concept and a theory." They help to orga­
tions of concepts restrict the advance of nize abstract, complex concepts. To create a new
knowledge. classification, a researcher logically specifies and
combines the characteristics of simpler con­
Concept Clusters. Concepts are rarely used in cepts. You can best grasp this idea by looking at
isolation. Rather, they form interconnected some examples.
groups, or concept clusters. This is true for con­ The idealtype is a well-known classification.
cepts in everyday language as well as for those in Ideal types are pure, abstract models that define
social theory. Theories contain collections of as­ the essence of the phenomenon in question.
sociated concepts that are consistent and mutu­ They are mental pictures that define the central
ally reinforcing. Together, they form a web of aspects of a concept. Ideal types are not explana­
meaning. For example, if! want to discuss a con­ tions because they do not tell why or how some­
cept such as urban decay, I will need a set of as­ thing occurs. They are smaller than theories, and
sociated concepts (e.g., urban expansion, researchers use them to build a theory. They are
economicgrowth, urbanization, suburbs, center broader, more abstract concepts that bring to­
city, revitalization, mass transit, and racial mi­ gether several narrower, more concrete con­
norities). cepts. Qualitative researchers often use ideal
Some concepts take on a range of values, types to see how well observable phenomena
quantities, or amounts. Examples of this kind of match up to the ideal model. For example, Max

Weber developed an ideal type of the concept used for a much broader range of specific time
bureaucracy. Many people use Weber's ideal type points and situations. More concrete concepts
(see Box 2.1). It distinguishes a bureaucracy are easy to recognize but apply to fewer situa­
from other organizational forms (e.g., social tions. The concepts skin pigmentation, casting a
movements, kingdoms, etc.). It also clarifies crit­ ballotin an election, and age based on the date on
ical features of a kind of organization that people a birth certificate are less abstract and more con­
once found nebulous and hard to think about. crete than the concepts racial group, democracy,
No real-life organization perfectly matches the and maturity. Theories that use many abstract
ideal type, but the model helps us think about concepts can apply to a wider range of social
and study bureaucracy. phenomena than those with concrete concepts.
An example of a theoretical relationship is: In­
Scope. Concepts vary by scope. Some are creased size creates centralization, which in turn
highly abstract, some are at a middle level of ab­ creates greater formalization. Size, centralization,
straction, and some are at a concrete level (i.e., and formalization are very abstract concepts.
they are easy to directly experience with the They can refer to features of a group, organiza­
senses such as sight or touch). More abstract tion, or society. We can translate this to say that
concepts have wider scope; that is, they can be as an organization or group gets bigger, author­
ity and power relations within it become cen­
tralized and concentrated in a small elite. The
elite will tend to rely more on written policies,
rules, or laws to control and organize others in
the group or organization. When you think ex­
plicitly about the scope of concepts, you make a
theory stronger and will be able to communicate
• It is a continuous organization governed by a sys­ it more clearly to others.
tem of rules.
• Conduct is governed by detached, impersonal Assumptions
• There is division of labor, in which different of­
fices are assigned different spheres of compe­
Concepts contain built-in assumptions, state­
ments about the nature of things that are not ob­
servable or testable. We accept them as a
('• ..:­. . . . . r

• Hierarchical authority relations prevail; that is,

necessary starting point. Concepts and theories
lower offices are under control of higher ones. build on assumptions about the nature of hu­
man beings, social reality, or a particular phe­
• Administrative actions, rules, and so on are in
nomenon. Assumptions often remain hidden or
writing and maintained in files.
unstated. One way for a researcher to deepen his
• Individuals do not own and cannot buy or sell or her understanding of a concept is to identify
their offices.
the assumptions on which it is based.
• Officials receive salaries rather than receiving di­ For example, the concept book assumes a
rect payment from clients in order to ensure loy­ system of writing, people who can read, and the
alty to the organization. existence of paper. Without such assumptions,
• Property of the organization is separate from per­ the idea of a book makes little sense. A social sci­
sonal property of officeholders. ence concept, such as racial prejudice, rests on callv possible ~
several assumptions. These include people who explanation. ~
Source: Adapted from Chafetz (1 978:72). make distinctions among individuals based on able. There are ~
their racial heritage, attach specific motivations contenders.

and characteristics to membership in a racial Direction of Theorizing

group, and make judgments about the goodness
Researchers approach the building and testing of
of specific motivations and characteristics. If
theory from two directions. Some begin with ab­
race became irrelevant, people would cease to
stract thinking. They logicallyconnect the ideas
distinguish among individuals on the basis of
in theory to concrete evidence, then test the
race, to attach specific characteristics to a racial
ideas against the evidence. Others begin with
group, and to make judgments about character­
specific observations of empirical evidence. On
istics. If that occurred, the concept of racial prej­
the basis of the evidence, they generalize and
udice would cease to be useful for research. All
build toward increasingly abstract ideas. In prac­
concepts contain assumptions about social rela­
tice, most researchers are flexible and use both
tions or how people behave.
approaches at various points in a study (see
Figure 2.1).
Deductive. In a deductive approach, you begin
Theories contain concepts, their definitions, and with an abstract, logical relationship among
assumptions. More significantly, theories specify concepts, then move toward concrete empirical
how concepts relate to one another. Theories tell evidence. You may have ideas about how the
us whether concepts are related or not. If they world operates and want to test these ideas
are related, the theory states how they relate to against "hard data."
each other. In addition, theories give reasons for Weitzer and Tuch (2004, 2005) used a de­
why the relationship does or does not exist. It is ductive approach in a study of perceptions of
a relationship, such as:Economic distress among police misconduct. They began with Group
the White population caused an increase in mob Position theory (a middle-range theory dis­
violence against African Americans. When a re­ cussed later) within the conflict theory frame­
searcher empirically tests or evaluates such a work (see Range of Theory later in this chapter).
relationship, it is called a hypothesis. After many Group position theory states that dominant and
careful tests of a hypothesis with data confirm subordinate racial-ethnic groups are in compe­
the hypothesis, it is treated as a proposition. A tition for resources and status in a multiethnic
proposition is a relationship in a theory in which society that has a racial hierachy, and such com­
the scientific community starts to gain greater petition affectsracial beliefs and attitudes. Dom­
confidence and feels it is likely to be truthful. inant groups believe they are entitled to
privileges and a position of superiority, and they
fear losing their privileges. Subordinate groups
believe their position can be enhanced if they
challenge the existing order. The authors de­
Theory can be baffling because it comes in so duced that group competition extends beyond
many forms. To simplify, we can categorize a attitudes to perceptions of social institutions, es­
theory by (1) the direction of its reasoning, (2) pecially institutions of social control such as
the level of social reality that it explains, (3) the policing. They argued that subordinate group
forms of explanation it employs, and (4) the members (i.e., Blacks and Latino/Hispanics)
overall framework of assumptions and concepts would preceive police misconduct (measured as
in which it is embedded. Fortunately, all logi­ unjustified stops of citizens, verbal abuse by
cally possible combinations of direction, level, police, an excessive use of force, and police cor­
explanation, and framework are not equally vi­ ruption) differently than members of the domi­
able. There are only about half a dozen serious nant group (Whites). The authors thought that
contenders. perceptions operated via three mechanisms:

FIGURE 2.1 Deductive and Inductive Theorizing

Middle-Range Middle-Range
Theory Theory

Hypothesis Hypothesis, Grounded Concept Formation,

Testing Empirical Generalization Theorizing Empirical Generalization
... 1

Empirical Social Reality Empirical Social Reality

Range of
personal encounters with the police; reports of Duneier (1999) used an inductive approach Social theories
police encounters by friends, family, or neigh­ in his study oflife on the sidewalk. He noted that source of the
bors; and noticing and interpreting news reports in much of social science, both quantitative sec­ the range at ~
about police activity. In these three areas, they ondary analysis research and qualitative field re­ are highly
predicted that non-Whites would interpret neg­ search, a researcher develops a theoretical cepts of - .
ative events or reports as strong evidence of seri­ understanding only after data have been col­ whole systems
ous and systematic police misconduct. By lected. He stated, "I began to get ideas from the tremelvab
constrast, Whites would tend to ignore or dis­ things I was seeing and hearing on the street" (p. building, verifJi
miss such events or reports or see them as iso­ 341). Many researchers who adopt an inductive connects th~
lated incidents. Data from a national survey of approach use grounded theory. Grounded theory ranges tog~
U.S. metropolitan areas (over 100,000 popula­ is part of an inductive approach in which a re­ boxes that fit ~
tion) supported predictions of the theory. searcher builds ideas and theoretical generaliza­ ian dolls. ~
tions based on closely examining and creatively
Inductive. If you use an inductive approach, thinking about the data (see Box 2.2). A re­ Empirical Gena
eralization is m.j
you begin with detailed observations of the
world and move toward more abstract general­
izations and ideas. When you begin, you may
searcher creates grounded theory out of a
process of trying to explain, interpret, and ren­
der meaning from data. It arises from trying to
ment and has
statement aboa
have only a topic and a few vague concepts. As account for, understand, or "make sense of' the among two or ~
you observe, you refine the concepts, develop evidence. Duneier (1999:342) has suggested that very close to ell
empirical generalizations, and identify prelimi­ the process is similar to seeing many symptoms "More men thai!
nary relationships. You build the theory from and later arriving at a diagnosis (i.e., a story that a college major.t
the ground up. explains the source of the symptoms). tween gender ...

Grounded theory is a widely used approach in qual­ generalizations by making comparisons across so­
itative research. It is not the only approach and it is cial situations.
not used by all qualitative researchers. Grounded the­ Qualitative researchers use alternatives to
oryis "a qualitative research method that uses a sys­ grounded theory. Some qualitative researchers offer
tematic set of procedures to develop an inductively an in-depth depiction that is true to an informant's
derived theory about a phenomenon" (Strauss and worldview. They excavate a single social situation to
Corbin, 1990:24). The purpose of grounded the­ elucidate the micro processes that sustain stable social
ory is to build a theory that is faithful to the evi­ interaction. The goal of other researchers is to provide
dence. It is a method for discovering new theory. In a very exacting depiction of events or a setting. They
it, the researcher compares unlike phenomena with a analyze specific events or settings in order to gain in­
view toward learning similarities. He or she sees mi­ sight into the larger dynamics of a society. Still other
cro-level events as the foundation for a more macro­ researchers apply an existing theory to analyze specific
level explanation. Grounded theory shares several settings that they have placed in a macro-level histor­
goals with more positivist-oriented theory. It seeks ical context. They show connections among micro­
theory that is comparable with the evidence that is level events and between micro-level situations and
precise and rigorous, capable of replication. and larger social forces for the purpose of reconstructing
generalizable. A grounded theory approach pursues the theory and informing social action.

Range of Theory easy to test or observe. It is called a generaliza­

tion because the pattern operates across many
Social theories operate with varying ranges. One
time periods and social contexts. The finding in
source of the confusion about theories involves
the study on Internet pornography discussed in
the range at which a theory operates. At one end
Chapter 1 that unhappily married men are more
are highly specific theories with concrete con­
likely than happily married men to use Internet
cepts of limited scope. At the opposite end are
porn is an empirical generalization.
whole systems with many theories that are ex­
tremely abstract. As part of the task of theory Middle-Range Theory. Middle-range theories
building, verifying, and testing, a researcher are slightly more abstract than empirical gener­
connects theoretical statements of different alizations or a specific hypothesis. A middle­
ranges together, like a series of different-sized range theory focuses on a specific substantive
boxes that fit into one another or a set of Russ­ topic area (e.g., domestic violence, military
ian dolls. coups, student volunteering), includes a multi­
ple empirical generalization, and builds a theo­
Empirical Generalization. An empirical gen­ retical explanation (see Forms of Explanation
eralization is the least abstract theoretical state­ later in this chapter). As Merton (1967:39)
ment and has a very narrow range. It is a simple stated, "Middle-range theory is principally used
statement about a pattern or generalization in sociology to guide empirical inquiry." A mid­
among two or more concrete concepts that are die-range theory used in a study discussed in
very close to empirical reality. For example, Chapter 1 said that girls who suffer physical or
"More men than women choose engineering as sexual abuse experience self-blame and guilt
a college major." This summarizes a pattern be­ feelings that inhibits them from developing a
tween gender and choice of college major. It is healthy socialnetwork or forming stable romantic

relationships, and that these factors lead to them

staying single or experiencing greater marital in­
are used in inductive and deductive approaches
to theorizing. Few researchers make precise dis­
lion -idt .a:..
stability when they become adults. tinctions among the ranges of theorizing. They Sea. ••.,.
rarely use a theoretical framework directly in era! .ouior.­
Theoretical Frameworks. A theoretical frame­ empirical research. A researcher may test parts lol"'Ods .-e . ­
work (also called a paradigm or theoretical sys­ of a theory on a topic and occasionally contrast looking .
tem) is more abstract than a middle-range parts of the theories from different frameworks. tiomoi. . . .
theory. Figure 2.1 shows the levels and how they Box 2.3 illustrates the various degrees of abstrac­ planarioa. n.
rim ....
theories oI~
struetlmIl ' ­
lolidrin the
Theoretical Framework lege. they have opportunities to meet other unmar­
OIl Inilll:lo-llrllil
Kalmijn. Structural functionalism holds that the ried people. In modern society. education has be­
processes of industrialization and urbanization change come a major socialization agent. It affects future
human society from a traditional to a modern form. In earnings, moral beliefs and values, and leisure inter­
this process of modernization. social institutions and ests. Thus. young adults select marriage partners less
practices evolve. This evolution includes those that fill on the basis of shared religious or local ties and more
the social system's basic needs. socialize people to cul­ on the basis of common educational levels.
tural values, and regulate social behavior. Institutions
Weitzer andTuch. Group-position theory uses group
that filled needs and maintained the social system in a
competition over material rewards, power. and status
traditional society (such as religion) are superseded
to explain intergroup attitudes and behaviors. Each
by modern ones (such as formal schooling).
group perceives and experiences real or imagined ~c-.­
Weitzer and Tuch. Conflict theory holds that estab­ threats to its social position differently. Members of a ciIision rlw.r
lished social. political. and legal institutions protect dominant group tend to viewpolice or government ac­
the dominant or privileged groups of a society. Ma­ tions taken to defend its interests as being fair or fa­
pelldtftt . . ­
jor institutions operate in ways that contain or sup­ vorable, whereas members of subodorinate groups
press the activities of nondominant groups in tend to see the same actions negatively.
't)lpe.. . . . .
society, especially if they challenge or threaten the
established social-economic hierarchy. Thus, conflict Empirical Generalization SIXiI!ty--''­
between the dominant and subordinate social groups
is reflected in how major institutions operate. espe­
Kalmijn. Americans once married others with simi­ ..-.a
lar religious beliefs and affiliation. This practice is be­
cially institutions that are charged with maintaining
order and engaged in formal social control, such as
ing replaced by marriage to others with similar levels
of education.
E '

law enforcement.
Weitzer and Tuch. Non-Whites experience more
Middle-Range Substantive Theory negative interpersonal encounters with police and ClyA ;S

Kalmijn. A theory of intermarriage patterns notes tend to interpret media reports about police mis­ ft_FFIEir
that young adults in modern society spend less time conduct as evidence of serious and systematic prob­ -.u5C!91 .

-. ....
in small, local settings, where family. religion. and lems with the police. By contrast. Whites have .,·­ 4
community all have a strong influence. Instead. different police encounters or interpret their en­ ~
young adults spend increasing amounts of time in counters and media reports about police actions .--
school settings. In these settings. especially in col­ more favorably. -~

tion with Kalmijn's study of changing marriage in sociology and briefly describes the key con­
partner selection (see also page 40). cepts and assumptions of each.
Sociologyand other social sciences have sev­
eral major theoretical frameworks.f The frame­
Levels of Theory
works are orientations or sweeping ways of
looking at the social world. They provide collec­ Social theories can be divided into three broad
tions of assumptions, concepts, and forms of ex­ groupings by the level ofsocial reality with which
planation. Frameworks include theories for they deal. Most of us devote the majority of our
many substantive areas (e.g., theories of crime, time to thinking about the micro level of reality,
theories of the family, etc.). Thus, there can be a the individuals we see and interact with on a day­
structural functional theory, an exchange the­ by-day basis. Micro-level theory deals with small
ory, and a conflict theory of the family. Theories slices of time, space, or numbers of people. The
within the same framework share assumptions concepts are usually not very abstract.
and major concepts. Some frameworks are ori­ Brase and Richmond (2004) used a micro­
ented more to the micro level; others focus more level theory about doctor-patient interactions
on macro-level phenomena (see Levels of The­ and perceptions. The theory stated that physican
ory next). Box 2.4 shows four major frameworks attire affects doctor-patient interactions. It sug-

Structural Functionalism Symbolic Interactionism

Major Concepts. System, equilibrium, dysfunction, Major Concepts. Self, reference group, role-playing,
division of labor perception

Key Assumptions. Society is a system of interde­ Key Assumptions. People transmit and receive sym­
pendent parts that is in equilibrium or balance. Over bolic communication when they socially interact.
time, society has evolved from a simple to a complex People create perceptions of each other and social
type, which has highly specialized parts. The parts of settings. People largely act on their perceptions.
society fulfill different needs or functions of the social How people think about themselves and others is
system. A basic consensus on values or a value sys­ based on their interactions.
tem holds society together.
Conflict Theory
Exchange Theory (also Rational Choice) Major Concepts. Power, exploitation, struggle, in­
Major Concepts. Opportunities, rewards, approval, equality, alienation
balance, credit
Key Assumptions. Society is made up of groups that
Key Assumptions. Human interactions are similar to have opposing interests. Coercion and attempts to
economic transactions. People give and receive re­ gain power are ever-present aspects of human rela­
sources (symbolic, social approval, or material) and tions. Those in power attempt to hold on to their
try to maximize their rewards while avoiding pain, ex­ power by spreading myths or by using violence if
pense, and embarrassment. Exchange relations tend necessary.
to be balanced. Ifthey are unbalanced, persons with
credit can dominate others.

gested that a patient makes judgments about a would produce gains because other workers and ilBrp IM­
physican's abilities based on attire and that a pa­ government authorities would support their ac­ - C"
tient's trust-openness toward a physican is also af­
fected. It said that perceptions of physican
authority increased with traditional professional
formal attire over informal attire, but that trust­
Macro-level theoryconcerns the operation of
larger aggregates such as social institutions, en­
tire cultural systems, and whole societies. It uses
iaII . ._

openness was influenced in the opposite direction

as authority. Thirty-eight male and 40 female re­
search participants rated their perceptions of
more concepts that are abstract.
Marx's study (1998) on race in the United
States, South Africa, and Brazil used a macro-level
...... .
tiom . . . .

~OI" . .~. .
same- and opposite-gender models who were theory. He wanted to explain the conditions that tJ3III5il"
identified as being medical doctors, but who were led Black people to engage in protest to gain full good tadIa"­
wearing different attire. Findings showed that a citizenship rights and he examined patterns ofna­ Thetwol!'fll:Mllll
white coat and formal attire are clearly superior to tional racial politics in three counties across two This~
makes iD'. . . .l11li
casual attire in establishing physican authority,
but it did not reduce trust-openness as expected.
Meso-level theorylinks macro and micro lev­
centuries. His theory said that protest resulted in
an interaction between (1) race-based political
mobilization and (2) national government poli­
Pt_ ­

els and operates at an intermediate level. Theo­ cies of racial domination (i.e., apartheid in South will OCOII'. It..
ries of organizations, social movements, and Africa, Jim Crow laws in southern United States, and an e:q:J'-'-'
communities are often at this level. and no legalized race-based domination in prediction
Roscigno and Danaher (2001) used meso­ Brazil). Policies of racial domination developed diet. An e:IIJI...
level theory in a study on the 1930s labor move­ from practices of slavery, exploitation, and dis­ one outCOlDlC,
ment among southern textile workers. The crimination that justified White superiority. The predicted ~
researchers used a theory of movement subcul­ policies reinforced specific racial ideologies that it is less J)OIIICI*I
ture and political opportunity to explain grow­ shaped national development during the twenti­ ple are entnllOlllll
ing labor movement strength and increased eth century. A critical causal factor was how prediction.
strike activity among workers in one industry in
a region of the United States across several years.
national political elites used the legalized domina­
tion of Blacks to reduce divisions among Whites.
A gam"'.
ence betweea
They expected strike activity to grow as the result In nations that had large regional or class divi­ enter a casiDo
of a strong movement subculture that carried a sions among Whites, national elites tried to predict the oar
message of injustice and a "political opportu­ increase White backing for the national govern­ ber on a nHllkll~
nity" or the expectation among people that col­ ment by creating legalized forms of racial domi­ may win a lot
lective action at a particular time would produce nation. Over time, such legalized domination officials realizr
positive results. Their study showed that a tech­ froze racial divisions, which promoted a sense of me. Yet, my
nological innovation (i.e., the spread of new ra­ racial identity and consciousness among Blacks. more inten:sti.
dio stations with songs and discussions of The strong sense of racial identity became a key Telling you.
working conditions and unfair treatment) con­ resource when Blacks mobilized politically to de­ more fascirJatiIDlll
tributed to the growth of a subculture of move­ mand full citizenship rights. Legalizedracial dom­ Here is
ment solidarity among the textile workers and ination also intensified the Blacks' protest and sun "rises" each
fostered self-identity as a worker who had com­ directed it against the national government as the at some time, ell
mon interests with the other textile workers. The societal institution that reinforced their experi­ clouds obscure ~
technological innovation and events in the polit­ ence of racial inequality. this so? One ~
ical environment (i.e., union organizers and carries the sun 4
speeches by the President of the United States) other expl~
Forms of Explanation
also created a political opportunity for the work­ ablaze, which ~
ers. The workers believed that collection action Prediction and Explanation. A theory's pri­ shoots it across ..
(i.e., strike) was necessary to achieve justice and mary purpose is to explain. Many people con­ these ancient exJ

fuse prediction with explanation. There are two probably accept involves a theory about the ro­
meanings or uses of the term explanation. Re­ tation of the earth and the position of the sun,
searchers focus on theoretical explanation, a log­ the star of our solar system. In this explanation,
ical argument that tells why something occurs. It the sun only appears to rise. The sun does not
refers to a general rule or principle. These are a move; its apparent movement depends on the
researcher's theoretical argument or connec­ earth's rotation. Weare on a planet that both
tions among concepts. The second type ofexpla­ spins on its axis and orbits around a star millions
nation, ordinary explanation, makes something of miles away in space. All three explanations
clear or describes something in a way that illus­ make the same prediction: The sun rises each
trates it and makes it intelligible. For example, a morning. As you can see, a weak explanation can
good teacher"explains" in the ordinary sense. produce an accurate prediction. A good expla­
The two types ofexplanation can blend together. nation depends on a well-developed theory and
This occurs when a researcher explains (i.e., is confirmed in research by empirical observa­
makes intelligible) his or her explanation (i.e., a tions.
logical argument involving theory).
Prediction is a statement that something Causal Explanation. Causal explanation, the
will occur. It is easier to predict than to explain, most common type ofexplanation, is used when
and an explanation has more logical power than the relationship is one of cause and effect. We
prediction because good explanations also pre­ use it all the time in everyday language, which
dict. An explanation rarely predicts more than tends to be sloppy and ambiguous. What do we
one outcome, but the same outcome may be mean when we say cause? For example, you may
predicted by opposing explanations. Although say that poverty causes crime or that looseness in
it is less powerful than explanation, many peo­ morals causes an increase in divorce. This does
ple are entranced by the dramatic visibility of a not tell how or why the causal process works.
prediction. Researchers try to be more precise and exact
A gambling example illustrates the differ­ when discussing causal relations.
ence between explanation and prediction. If I Philosophers have long debated the idea of
enter a casino and consistently and accurately cause. Some people argue that causality occurs
predict the next card to appear or the next num­ in the empirical world, but it cannot be proved.
ber on a roulette wheel, it will be sensational. I Causality is "out there" in objective reality, and
may win a lot of money, at least until the casino researchers can only try to find evidence for it.
officials realize I am always winning and expel Others argue that causality is only an idea that
me. Yet, my method of making the predictions is exists in the human mind, a mental construc­
more interesting than the fact that I can do so. tion, not something "real" in the world. This
Telling you what I do to predict the next card is second position holds that causality is only a
more fascinating than being able to predict. convenient way of thinking about the world.
Here is another example. You know that the Without entering into the lengthy philosophical
sun "rises" each morning. You can predict that debate, many researchers pursue causal relation­
at some time, every morning, whether or not ships.
clouds obscure it, the sun will rise. But why is You need three things to establish causality:
this so? One explanation is that the Great Turtle temporal order, association, and the elimination
carries the sun across the sky on its back. An­ of plausible alternatives. An implicit fourth
other explanation is that a god sets his arrow condition is an assumption that a causal rela­
ablaze, which appears to us as the sun, and tionship makes sense or fits with broader as­
shoots it across the sky. Few people today believe sumptions or a theoretical framework. Let us
these ancient explanations. The explanation you examine the three basic conditions.

The temporal order condition means that a the effect. Most studies examine unidirectional
cause must come before an effect. This com­ relations. More complex theories specify recip­
monsense assumption establishes the direction rocal-effect causal relations-that is, a mutual
of causality: from the cause toward the effect. causal relationship or simultaneous causality.
You may ask, How can the cause come after For example, studying a lot causes a student to
what it is to affect? It cannot, but temporal order get good grades, but getting good grades also
is only one of the conditions needed for causal­ motivates the student to continue to study. The­
ity. Temporal order is necessary but not suffi­ ories often have reciprocal or feedback relation­
cient to infer causality. Sometimes people make ships, but these are difficult to test. Some
the mistake of talking about "cause" on the basis researchers call unidirectional relations nonre­
of temporal order alone. For example, a profes­ cursive and reciprocal-effect relations recursive.
sional baseball player pitches no-hit games when A researcher also needs an association for
he kisses his wife just before a game. The kissing causality. Two phenomena are associated if they
occurred before the no-hit games. Does that occur together in a patterned way or appear to
mean the kissing is the cause of the pitching per­ act together. People sometimes confuse correla­
formance? It is very unlikely. As another exam­ tion with association. Correlation has a specific
ple, race riots occurred in four separate cities in technical meaning, whereas association is a more
1968, one day after an intense wave of sunspots. general idea. A correlation coefficient is a statisti­
The temporal ordering does not establish a cal measure that indicates the amount of associ­ !
causal link between sunspots and race riots. M­ ation, but there are many ways to measure If a reseanl
ter all, all prior human history occurred before association. Figure 2.2 shows 38 people from a causal rela~
some specific event. The temporal order condi­ lower-income neighborhood and 35 people from searchers at1emJl
tion simply eliminates from consideration po­ an upper-income neighborhood. Can you see an measures of as5lIIj
tential causes that occurred later in time. association between race and income level? ten find an assed
It is not always easy to establish temporal More people mistake association for causal­ sociation e~
order. With cross-sectional research, temporal ity than confuse it with temporal order. For ex­ associated, burl
order is tricky. For example, a researcher finds ample, when I was in college,I got high grades on cause. It is a neal
that people who have a lot of education are also the exams I took on Fridays but low grades on tion. In other ..
less prejudiced than others. Does more educa­ those I took on Mondays. There was an associa­ but it is not ~
tion cause a reduction in prejudice? Or do highly tion between the day of the week and the exam Anassocialll
prejudiced people avoid education or lack the grade, but it did not mean that the day of the (i.e., everytime~
motivation, self-discipline, and intelligence week caused the exam grade. Instead, the reason also is) to s~~
needed to succeed in school? Here is another ex­ was that I worked 20 hours each weekend and ing exam gra~
ample. The students who get high grades in my was very tired on Mondays. As another example, association if 0lIIIl
class say I am an excellent teacher. Does getting the number of children born in India increased 1 C, whereas ~
high grades make them happy, so they return the until the late 1960s,then slowed in the 1970s. The were 6 Ds, 2 Cs".l
favor by saying that I am an excellent teacher number of U.S.-made cars driven in the United but the days ofd
(i.e., high grades cause a positive evaluation)? Or States increased until the late 1960s,then slowed not perfectly aSlI
am I doing a great job, so students study hard in the 1970s. The number of Indian children level associatioa
and learn a lot, which the grades reflect (i.e., born and the number of U.S. cars driven are as­ imperfect assodI
their learning causes them to get high grades)? It sociated: They vary together or increase and de­ Elimi~
is a chicken-or-egg problem. To resolve it, a re­ crease at the same time. Yet there is no causal searcher interea
searcher needs to bring in other information or connection. By coincidence, the Indian govern­ that the effect ..
design research to test for the temporal order. ment instituted a birth control program that not to som~
Simple causal relations are unidirectional, slowed the number of births at the same time ousness becausei
operating in a single direction from the cause to that Americans were buying more imported cars. that is actuallv •


Fie U R E 2.2 Association of Income and Race

Lower Income Upper Income

•• 0

0t.." o
Q~ o~~~O o~'

0 0 0 •

o'o?~ n .'i ti


Q1f.t. ia 'ti
oO~'i .Q.~~ t

Q~ ~o o~'fat.

O~I. ,. ?tl O~ oQo~o~~ Q~"

i ti"" Q~

0 0


If a researcher cannot find an association, a ognized cause is called a spurious relationship,

causal relationship is unlikely. This is why re­ which is discussed in Chapter 4 (see Box 2.5).
searchers attempt to find correlations and other Researchers can observe temporal order and
measures of association. Yet, a researcher can of­ associations. They cannot observe the elimina­
ten find an association without causality. The as­ tion of alternatives. They can only demonstrate
sociation eliminates potential causes that are not it indirectly. Eliminating alternatives is an ideal
associated, but it cannot definitely identify a because eliminating all possible alternatives is
cause. It is a necessary but not a sufficient condi­ impossible. A researcher tries to eliminate major
tion. In other words, you need it for causality, alternative explanations in two ways: through
but it is not enough alone. built-in design controls and by measuring po­
An association does not have to be perfect tential hidden causes. Experimental researchers
(i.e., every time one variable is present, the other build controls into the study design itself to
also is) to show causality. In the example involv­ eliminate alternative causes. They isolate an ex­
ing exam grades and days ofthe week, there is an perimental situation from the influence of all
association if on 10 Fridays I got 7 As, 2 Bs, and variables except the main causal variable.
1 C, whereas my exam grades on 10 Mondays Researchers also try to eliminate alternatives
were 6 Ds, 2 Cs, and 2 Bs. An association exists, by measuring possible alternative causes. This is
but the days ofthe week and the exam grades are common in survey research and is called
not perfectly associated. The race and income­ controlling for another variable. Researchers use
level association shown in Figure 2.2 is also an statistical techniques to learn whether the causal
imperfect association. variable or something else operates on the effect
Eliminating alternatives means that a re­ variable.
searcher interested in causality needs to show Causal explanations are usually in a linear
that the effect is due to the causal variable and form or state cause and effect in a straight line: A
not to something else. It is also called no spuri­ causes B, B causes C, C causes D.
ousness because an apparent causal relationship The study by Brase and Richmond (2004)
that is actually due to an alternative but unrec­ on doctor-patient interactions discussed earlier

As I was driving home from the universityone day, I bias: the educational experience of students. It
heard a radio news report about gender and racial turns out that girls and boys take different numbers
bias in standardized tests. A person who claimedthat and types of mathematics courses in high school.
bias was a major problem said that the tests should Girls tend to take fewer math courses. Among the
be changed. Since I workin the field of education and girls who complete the same mathematics curricu­
disdain racial or gender bias, the report caught my lum as boys, the gender difference dissolves. Like­
attention. Yet, as a social scientist, I critically evalu­ wise, a large percentage of African Americans
ated the news story. The evidence for a bias charge attend racially segregated, poor-quality schools in
was the consistent pattern of higher scores in math­ inner cities or in impoverished rural areas. For
ematics for male high school seniors versus female African Americans who attend high-quality subur­
high school seniors, and for European-background ban schools and complete the same courses, racial
students versus African American'students. Was the differences in test scores disappear. This evidence
cause of the pattern of different test scores a bias suggests that inequality in education causes test
built into the tests? score differences. Although the tests may have
When questioned by someone who had de­ problems, identifying the real cause implies that
signed the tests, the person charging bias lacked a changing the tests without first improvingor equal­
crucial piece of evidence to support a claim of test izing education could be a mistake.

used a causal explanation; it said physican attire X leads to Y, X produces Y, X influences Y, X is Structural
causes certain types of patient perceptions. The related to Y, the greater X the higher Y. tum is used _ida
studybyWeitzer and Tuch (2004,2005) on po­ Here is a simple causal theory: A rise in un­ sequential. aocl

lice misconduct cited earlier used a causal ex­ employment causes an increase in child abuse. causal effect
planation. The cause was a person's group The subject to be explained is an increase in the balls lined up dJIIll
position and competitive pressure with other occurrence of child abuse. What explains it is a to bounce in ~
groups. These are causally linked to police en­ rise in unemployment. We "explain" the in­
with spokes
counters, either directly or indirectly, and inter­ crease in child abuse by identifying its cause. A in which each
pretions of news reports, which differ by group complete explanation also requires elaborating researcher ­
position. The police encounters and the inter­ the causal mechanism. My theory says that when a set of int
pretations of news reports cause very different people lose their jobs, they feel a loss of self­ and relationshiplll
perceptions of police misconduct. We can re­ worth. Once they lose self-worth, they become he or she uses ~
state the logic in a deductive causal form: If the easily frustrated, upset, and angry. Frustrated lationships "'maIIIl
proposition is true, then we observe certain people often express their anger by directing vi­ lations within 4
. things in the empirical evidence. Good causal olence toward those with whom they have close reinforcing ~~
explanations identify a causal relationship and personal contact (e.g., friends, spouse, children, researcher ~
specify a causal mechanism. A simple causal ex­ etc.). This is especially true if they do not under­ tifies essential JIll
planation is: X causes Yor Yoccurs because of stand the source of the anger or cannot direct it ",-hole.
X, where X and Yare concepts (e.g., early mar­ toward its true cause (e.g., an employer, govern­ Structural ai
riage and divorce). Some researchers state ment policy, or "economic forces"). theory. Sanders.'
causality in a predictive form: If X occurs, then The unemployment and child abuse exam­ plained Asian iml
Yfollows. Causality can be stated in many ways: ple illustrates a chain of causes and a causal work theory. 11

mechanism. Researchers can test different parts immigrants from the Philippines, Korea, Tai­
of the chain. They might test whether unem­ wan, and China in Los Angeles and found that
ployment rates and child abuse occur together, social networks matched and sorted immigrants
or whether frustrated people become violent to­ with jobs. New immigrants with limited lan­
ward the people close to them. A typical research guage and job skills sought employment either
strategy is to divide a larger theory into parts and with a co-ethnic employer or through informal
test various relationships against the data. social ties (i.e., they consulted experienced
Relationships between variables can be pos­ friends, relatives, and acquaintances and asked
itive or negative. Researchers imply a positive re­ them to be intermediaries). Network users ex­
lationship if they say nothing. A positive panded job opportunities beyond employers in
relationship means that a higher value on the their own ethnic group. Thus, ethnic network
causal variable goJS with a higher value on the ties were "bridge ties" (i.e., they helped immi­
effect variable. For example, the more education grants get jobs beyond their ethnic community
a person has, the longer his or her life expectancy by using co-ethnics who already made the tran­
is. A negative relationship means that a higher sition to mainstream employment). Over time,
value on the causal variable goes with a lower as language and job skills improved, these im­
value on the effect variable. For example, the migrants moved on to mainstream jobs. Immi­
more frequently a couple attends religious ser­ grants lacking social ties, in limited networks, or
vices, the lower the chances of their divorcing who worked for co-ethnics found it difficult to
each other. In diagrams, a plus sign (+) signifies get a mainstream job. Thus, a person's network
a positive relationship and a negative sign (-) location, access to a large and diverse network,
signifies a negative relationship. and use of network ties is what facilitated ob­
taining a mainstream job.
Structural Explanation. A structural explana­ Structural explanations are also used in se­
tion is used with three types of theories: network, quence theory. The panel study on volun­
sequential, and functional theories. Unlike a teerism by Oesterle, Johnson, and Mortimer
causal effect chain, which is similar to a string of (2004) discussed in Chapter 1 employs se­
balls lined up that hit one another causing each quence theory. The authors used a "life course"
to bounce in turn, it is more similar to a wheel perspective in which the impact of an event
with spokes from a central idea or a spider web happening at one phase of a person's life differs
in which each strand forms part of the whole. A what it would have been if the same happened at
researcher making a structural explanation uses other phases, and early events generally shape
a set of interconnected assumptions, concepts, events in later phases. The authors noted that
and relationships. Instead of causal statements, the transition to adulthood is a critical stage
he or she uses metaphors or analogies so that re­ when a person learns new social roles and adult
lationships "make sense." The concepts and re­ expectations. They found that the amounts and
lations within a theory form a mutually types of volunteer activity in the last stage they
reinforcing system. In structural explanations, a observed (age 26-27) was strongly influenced
researcher specifiesa sequence of phases or iden­ by such activities at prior stages of a person's life
tifies essential parts that form an interlocked (age 18-19). People who volunteered at an early
whole. stage tended to volunteer at later stages. Those
Structural explanations are used in network who did not volunteer at an early stage or who
theory. Sanders, Nee, and Sernau (2002) ex­ devoted full time to working or parenting at
plained Asian immigrant job seeking with net­ other prior stages (18-19 years old) were less
work theory. They used interview data on likely to volunteer at a later stage (26-27 years

old). Thus, later events flowed from an inter­ In modern society, people spend time away
connected process in which earlier stages set a from small local settings in school settings. In
course or direction that pointed to specific these school settings, especially in college, they
events in a later stage. meet other unmarried people. Education is a
Additionally, structural explanations are major socialization agent in modern society.
used in functional theory.6 Functional theorists Increasingly, it affects a person's future earn­
explain an event by locating it within a larger, ings, moral beliefs and values, and ways of
ongoing, balanced social system. They often use spending leisure time. This explains why there
biological metaphors. These researchers explain has been a trend in the United States for people
something by identifying its function within a to marry less within the same religion and in­
larger system or the need it fulfills for the sys­ creasingly to marry persons with a similar level
tem. Functional explanations are in this form: "L of education. In traditional societies, the family
occurs because it serves needs in the system M." and religious organization served the function
Theorists assume that a system will operate to of socializing people to moral values and link­
stay in equilibrium and to continue over time. A ing them to potential marriage partners who
functional theory of social change says that, over held similar values. In modern society, educa­
time, a social system, or society, moves through tional institutions largely fulfill this function
developmental stages, becoming increasingly for the social system. -.r ......
differentiated and more complex. It evolves a fl.. i
specialized division oflabor and develops greater Interpretive Explanation. The purpose of an -.iIs~
individualism. These developments create interpretive explanation is to foster understand­ . . .GfII:II3IIIiiI
greater efficiency for the system as a whole. Spe­ ing. The interpretive theorist attempts to dis­ .-ar
cialization and individualism create temporary cover the meaning of an event or practice by
disruptions. The traditional ways of doing things placing it within a specific social context. He or
weaken, but new social relations emerge. The she tries to comprehend or mentally grasp the
system generates new ways to fulfill functions or operation of the social world, as well as get a feel
satisfy its needs. for something or to see the world as another per­
Kalmijn (1991) used a functional explana­ son does. Because each person's subjective
tion to explain a shift in how people in the worldview shapes how he or she acts, the re­
United States select marriage partners. He relied searcher attempts to discern others' reasoning
on secularization theory, which holds that on­ and view of things. The process is similar to
going historical processes of industrialization decoding a text or work of literature. Meaning
and urbanization shape the development of so­ comes from the context of a cultural symbol
ciety. During these modernization processes, system.
people rely less on traditional ways of doing Duneier's (1999) study of sidewalk life in
things. Religious beliefs and local community New York City discussed earlier in this chapter
ties weaken, as does the family's control over used an interpretive explanation. An interpretive
young adults. People no longer live their entire
lives in small, homogeneous communities.
Young adults become more independent from
their parents and from the religious organiza­
explanation is also illustrated by Edelman,
Fuller, and Mara-Drita's (2001) study of how
companies adopted policies related to diversity
issues in the early 1990s-that is, affirmative ac­
tions that formerly played a critical role in se­
lecting marriage partners.
Society has a basic need to organize the way
people select marriage partners and find part­
tion and equal opportunity. The authors exam­
ined what managers said, or their rhetoric, about
diversity concerns. Rhetoric included various
statements about diversity made by professional

ners with whom they share fundamental values. managers, business school professors, and con­

sultants in professional workshops, meetings,

specialized magazines, and electronic forums.
Edelman and colleagues (2001) found that
managers took legal ideas, terms, and concepts We began this chapter by looking at small-scale
and converted them into ones that fit into their parts of a theory (i.e., ideas or concepts). We
organizational setting. Professional managers moved toward larger aspects of social theory,
converted vague legal mandates and terms that and arrived at major theoretical frameworks in
were based on ideas about racial discrimination the last section. Now, we move to an even a
and ending injustice. They interjected their broader, more abstract level of the linkage be­
own views, values, training, and interests and tween theory and research-fundamental ap­
produced slightly different ideas and proce­ proaches to social science. It involves issues
dures. Management rhetoric changed legal sometimes called meta-methodological (i.e., be­
ideas from takirig specific actions to end yond or supersized methodological concerns)
racial-ethnic or gender discrimination and and blurs into areas of philosophy that studies
changed them into a "new idea" for effective what science means. We only briefly touch on
corporate management. The "new idea" was the issues here, but we cannot ignore them be­
that corporations benefit from a culturally di­ cause they affect how people do social research
verse workforce. Simply put, diversity is good studies.
for company profits. They consolidated various About 45 years ago, a now famous philoso­
studies and discussions on how to improve cor­ pher of science, Thomas Kuhn, argued that the
porate operations around the new idea-a so­ way science develops in a specific field across
cially heterogeneous workforce is more time is based on researchers sharing a general
creative, productive, and profitable. approach, or paradigm. A paradigm is an inte­
The authors created a theory of "manageri­ grated set of assumptions, beliefs, models of do­
alization of law" from their data. This theory ing good research, and techniques for gathering
states that professional managers operate in a and analyzing data. It organizes core ideas, theo­
corporate environment. They will not simply retical frameworks, and research methods. Kuhn
take ideas and mandates created in a govern­ observed that scientific fields tend to be held to­
ment-legal environment and impose them di­ gether around a paradigm for a long period of
rectlyonto a corporation's internal operations. time. Very few researchers question the para­
In fact, on the issue of affirmative action, many digm, and most focus on operating within its
corporate officials saw the legal ideas and re­ general boundaries to accumulate new knowl­
quirements as hostile or alien. So the managers edge. On rare occasions in history, intellectual
converted, or translated, the legal ideas into an difficulties increase, unexpected issues grow, and
acceptable form-one acceptable from a man­ troubling concerns over proper methods multi­
agerial point of view. They used new forms to ply. Slowly,the members of a scientific field shift
move their corporations in a direction that in how they see things and switch to a new para­
would comply with the legal requirements. This digm. Once the new paradigm becomes fully es­
is an interpretive explanation because the au­ tablished and widely adopted, the process of
thors explained a social event (i.e., corporations accumulating knowledge begins anew.
embracing programs and rhetoric to favor Kuhn's explanation covered how most sci­
cultural diversity) by examining how the man­ ences operate most of the time, but some fields
agers subjectively constructed new ways of look­ operate with multiple or competing paradigms.
ing at, thinking about, and talking about the This is the case in several of the social sciences.
diversity issue (i.e., they constructed a new This greatly bothers some social scientists, and
interpretation). they believe having multiple paradigms hinders

the growth of knowledge. They see multiple par­ Positivism sees social science research as funda­
adigms as a sign of the immaturity or underde­ mentally the same as natural science research; it
velopment ofthe "science" in the social sciences. assumes that social reality is made up of objec­
Some believe all social science researchers tive facts that value-free researchers can precisely
should embrace a single paradigm and stop us­ measure and use statistics to test causal theories.
ing alternatives to it. Large-scale bureaucratic agencies, companies,
Other social scientists accept the coexistence and many people in the general public favor a
of multiple paradigms. They recognize that this positivist approach because it emphasizes get­
can be confusing and often makes communicat­ ting objective measures of "hard facts" in the
ing difficult among those who use a different ap­ form of numbers.
proach. Despite this, they argue that each social Positivists put a great value on the principle
science paradigm provides important kinds of of replication, even if only a few studies are repli­
knowledge and insights, so to drop one would cated. Replication occurs when researchers or
limit what we can learn about the social world. others repeat the basics of a study and get iden­
These social scientists note that no one definitely tical or very similar findings. Positivists em­
can saywhich approach is "best" or even whether phasize replication and the ultimate test of
it is necessary or highly desirable to have only one knowledge. This is because they believe that dif­
paradigm. So instead of closing off an approach ferent observers looking at the same facts will get
that offers innovative ways to study social life and the same results if they carefully specify their
gain insight into human behavior, they argue for ideas, precisely measure the facts, and follow the
keeping a diversity of approaches. standards of objective research. When many
In this section, we will look at three funda­ studies by independent researchers yield similar
mental paradigms or approaches used in social findings, confidence grows that we accurately
science. Each approach has been around for over captured the workings of social reality and there­
150 years and is used by many highly respected fore scientific knowledge increases.
professional researchers. These approaches are If a researcher repeats a study and does not
unequal in terms of the number of followers, get similar findings, one or more offive possibil­
quantity of new studies, and types of issues ad­ ities may be occurring: (1) the initial study was
dressed. Often, people who strongly adhere to an unusual fluke or based on a misguided un­
one approach disagree with researchers who use derstanding of the social world; (2) important
another, or see the other approaches as being less conditions were present in the initial study, but
valuable or less "scientific" than their approach. no one was aware of their significance so they
Although adherents to each approach may use were not specified; (3) the initial study, or the
various research techniques, theories, and theo­ repeat of it, was sloppy-it did not include very The interpretree
retical frameworks, researchers who adopt one careful, precise measures; (4) the initial study, or its sees the idra
approach tend to favor certain research tech­ the repeat of it, was improperly conducted-re­ positivism,
niques, theories, or theoretical frameworks over searchers failed to closely follow the highest pretive re5leB1;h11
others. The three approaches are positivism, in­ standards for procedures and techniques, or qualita~
terpretive, and critical; each has internal divi­
sions, offshoots, and extensions, but these are
failed to be completely objective; or (5) the re­
peated study was an unusual fluke.
by science. This.
not just boJTOW4
the core ideas of the three major approaches. The positivist approach is nomothetic; it the natural sc:ie:IIII
means explanations use law or law-like princi­ necessary to ~
ples. Positivists may use inductive and deductive based on the uniq
.Positivist Approach
inquiry, but the ideal is to develop a general can really captunl
Positivism is the most widely practiced social sci­ causal law or principle then use logical deduc­ Most TeSeUlIIl
ence approach, especially in North America. tion to specify how it operates in concrete situa- approach adopt..

tions. Next, the researcher empirically tests out­ view of social reality. This view holds that hu­
comes predicted by the principle in concrete set­ man social life is based less on objective, hard,
tings using very precise measures. In this way, a factual reality than on the ideas, beliefs,and per­
general law or principle covers many specificsit­ ceptions that people hold about reality. In other
uations. For example, a general principle says words, people socially interact and respond
that when two social groups are unequal and based as much, if not more, on what they believe
compete for scarce resources, in-group feelings to be real than what is objectively real. This
and hostility toward the other groups intensify, means that social scientists will be able to under­
and the competing groups are likely to engage in stand social life only if they study how people go
conflict. The principle applies to sports teams, about constructing social reality. As people grow
countries, ethnic groups, families, and other so­ up, interact, and live their daily lives, they con­
cial groupings. A researcher might deduce that tinuously create ideas, relationships, symbols,
in cities with high levels of interracial inequality, and roles that they consider to be meaningful or
when jobs become more scarce and thereby in­ important. These include things such as intimate
crease economic competition, each group will emotional attachments, religious or moral
express greater hostility about the other racial ideals, beliefsin patriotic values, racial-ethnic or
groups, and intergroup conflict (e.g., riots, gender differences, and artistic expressions.
demonstrations, violent attacks) will increase. Rarely do people relate to the objective facts of
The vast majority of positivist studies are reality directly; instead, they do so through the
quantitative, and positivists generally see the ex- ­ filter of these socially constructed beliefs and
periment as the ideal way to do research. Posi­ perceptions. What positivists and many people
tivist researchers also use other quantitative view to be objective facts (e.g., a person's
research techniques, such as surveys or existing height), interpretive researchers say are only at
statistics, but tend to see them as approxima­ the trivial surface level of social life. Or, the
tions of the experiment for situations where an "facts" are images/categories that humans cre­
experiment is impossible. Positivist researchers ated (i.e., I am two meters tall) and we "forget"
advocate value-free science, seek precise quanti­ that people originated the images/categories but
tative measures, test causal theories with statis­ now treat them as being separate from people
tics, and believe in the importance of replicating and objectively real.
studies. Interpretive researchers are skeptical of the
positivist attempts to produce precise quantita­
tive measures of objective facts. This is because
Interpretive Approach
they view social reality as very fluid. For most
The interpretive approach is also scientific, but humans, social reality is largely the shifting per­
its sees the idea of "scientific" differently from ceptions that they are constantly constructing,
positivism. Unlike the positivist approach, inter­ testing, reinforcing, or changing and that have
pretive researchers say that human social life is become embedded in social traditions or institu­
qualitatively different from other things studied tions. For this reason, interpretive researchers
by science. This means that social scientists can­ tend to trust and favor qualitative data. They be­
not just borrow the principles of science from lieve that qualitative data can more accurately
the natural sciences. Instead, they believe it is capture the fluid processes of social reality. In
necessary to create a special type of science, one addition, they favor interpretive over causal
based on the uniqueness of humans and one that forms of theory (see discussion earlier in this
can really capture human social life. chapter).
Most researchers who use an interpretive Interpretive researchers are not likely to
approach adopt a version of the constructionist adopt a nomothetic approach, but instead favor

an idiographic form of explanation and use in­ because it profoundly shapes much of human
ductive reasoning. Idiographic literally means action. TABLE 2.1
specific description and refers to explaining an The critical approach has an activist orien­
aspect of the social world by offering a highly de­ tation and favors action research. Praxis is the
tailed picture or description of a specific social ultimate test of how good an explanation is in
setting, process, or type of relationship. For ex­ the critical approach. It is a blending of theory
Direction loci
ample, qualitative researchers do not see replica­ and concrete action; theory informs one about
tion as the ultimate test of knowledge. Instead, the specific real-world actions one should take Level Mici
they emphasize verstehen or empathetic under­ to advance social change, and one uses the expe­ Explanation c:.a.
standing. Verstehen is the desire of a researcher riences of engaging in action for social change to

to get inside the worldview of those he or she is reformulate the theory. All the approaches see a
studying and accurately represent how the peo­ mutual relationship between abstract theory and
ple being studied see the world, feel about it, and concrete empirical evidence, but the critical ap­
act. In other words, the best test of good social proach goes further and tries to dissolve the gap
knowledge is not replication but whether the re­ between abstract theory and the empirical expe­
searcher can demonstrate that he or she really riences of using the theory to make changes in Theory has .al
captured the inner world and personal perspec­
tive of the people studied.
the world.
tra1 in
but its promineoq
sic-explanatorv I"CS
descriptive r~
Critical Approach THE DYNAMIC DUO
cepts are often IDIJl
The critical approach shares many features with You have seen that theory and research are in­ to create zeneral I
an interpretive approach, but it blends an objec­ terrelated. Only the naive, new researcher mis­ searchers use thoIl
tive/materialist with a constructionist view ofso­
cial reality. The key feature of the critical
approach is a desire to put knowledge into ac­
takenly believes that theory is irrelevant to
research or that a researcher just collects the
data. Researchers who attempt to proceed with­
on", and inS;
refine concepts, etl

Theon- does
tion and a belief that research is not value free. out theory may waste time collecting useless provisional and


Research is the creation of knowledge, and peo­ data. They easily fall into the trap of hazy and
ple regularly use knowledge to advance political­ vague thinking, faulty logic, and imprecise con­
moral ends. This gives doing social research a cepts. They find it difficult to converge onto a
strong connection to political-moral issues. The crisp research issue or to generate a lucid ac­ rists toil to think.
researcher can decide to ignore and help those count of their study's purpose. They also find dIon has limia
with power and authority in society, or advance themselves adrift as they attempt to design or ica.nt progress is
social justice and empower the powerless. conduct empirical research. findings. I
Critical approach emphasizes the multilay­ The reason is simple. Theory frames how we The scienri6c
ered nature of social reality. On the surface level, look at and think about a topic. It gives us con­ ters theories
there is often illusion, myth, and distorted think­ cepts, provides basic assumptions, directs us to se:arc:hen ,,-ho
ing. The critical approach notes that people are the important questions, and suggests ways for me thoory to
often misled, are subject to manipulated mes­ us to make sense of data. Theory enables us to ~oi
sages, or hold false ideas. Yet, beneath the sur­ connect a single study to the immense base of mod:itY the
face level at a deeper, often hidden level lies knowledge to which other researchers con­ researchers
"real" objective reality. Part of the task of social tribute. To use an analogy, theory helps a re­ tearehin~
research is to strip away the surface layer of illu­ searcher see the forest instead of just a single dimr.:e thai: ~
sion or falsehood. Although a researcher wants tree. Theory increases a researcher's awareness ~modift-~
to see beyond this layer, he or she does not en­ of interconnections and of the broader signifi­ 'a:t them if irQ~
tirely ignore it. Such an outer layer is important cance of data (see Table 2.1). apne~


TABLE 2.1 Major Aspects and Types

and central tenets are more difficult to test and
are refuted less often. In a slow process, re­
of Social Theory searchers may decide to abandon or change a
theory as the evidence against it mounts over
time and cannot be logically reconciled.
Researchers adopting an inductive ap­
Direction Inductive or deductive
proach follow a slightly different process. Induc­
Level Micro, meso, or macro tive theorizing begins with a few assumptions
Explanation Causal, interpretive, or structural and broad orienting concepts. Theory develops
Abstraction Empirical generalization, middle
from the ground up as the researchers gather
range, framework, or paradigm and analyze the data. Theory emerges slowly,
concept by concept and proposition by proposi­
tion in a specific area. The process is similar to a
long pregnancy. Over time, the concepts and
empirical generalizations emerge and mature.
Theory has a place in virtually all research, Soon, relationships become visible, and re­
but its prominence varies. It is generally less cen­ searchers weave together knowledge from differ­
tral in applied-descriptive research than in ba­ ent studies into more abstract theory.
sic-explanatory research. Its role in applied and
descriptive research may be indirect. The con­
cepts are often more concrete, and the goal is not
to create general knowledge. Nevertheless, re­
searchers use theory in descriptive research to In this chapter, you learned about social the­
refine concepts, evaluate assumptions of a the­ ory-its parts, purposes, and types. The di­
ory, and indirectly test hypotheses. chotomy between theory and research is an
Theory does not remain fixed over time; it is artificial one. The value of theory and its neces­
provisional and open to revision. Theories grow sity for conducting good research should be
into more accurate and comprehensive explana­ clear. Researchers who proceed without theory
tions about the make-up and operation of the rarely conduct top-quality research and fre­
social world in two ways. They advance as theo­ quently find themselves in a quandary. Likewise,
rists toil to think clearly and logically, but this theorists who proceed without linking theory to
effort has limits. The way a theory makes signif­ research or anchoring it to empirical reality are
icant progress is by interacting with research in jeopardy of floating off into incomprehensible
fmdings. speculation and conjecture. You are now famil­
The scientific community expands and al­ iar with the scientific community, the dimen­
ters theories based on empirical results. Re­ sions of research, and social theory.
searchers who adopt a more deductive approach
use theory to guide the design of a study and the
interpretation of results. They refute, extend, or Key Terms
modify the theory on the basis of results. As
researchers continue to conduct empirical re­ association
search in testing a theory, they develop confi­ assumption
dence that some parts of it are true. Researchers blame analysis
may modify some propositions of a theory or re­ causal explanation
ject them if several well-conducted studies have classification concept
negative findings. A theory's core propositions concept cluster

deductive approach Endnotes

empirical generalization
functional theory 1. See Felson (1991), Felson and Felson (1993), and
grounded theory Logan (1991) for a discussion of blame analysis.
ideal type 2. For more detailed discussions of concepts, see
Chafetz (1978:45-61), Hage (1972:9-85), Kaplan
(1964:34-80), Mullins (1971:7-18), Reynolds
inductive approach
(1971), and Stinchcombe (1968,1973).
macro-level theory 3. Turner (1980) discussed how sociological expla­
meso-level theory nation and theorizing can be conceptualized as
micro-level theory translation.
negative relationship 4. Classifications are discussed in Chafetz (1978:
nomothetic 63-73) and Hage (1972).
paradigm 5. Introductions to alternative theoretical frame­
positive relationship works and social theories are provided in Craib
praxis (1984), Phillips (1985:44-59), and Skidmore
prediction (1979).
6. An introduction to functional explanation can be
found in Chafetz (1978:22-25).
Ethics in Social Research

Why Be Ethical?
Scientific Misconduct
Unethical but Legal
Power Relations
Ethical Issues Involving Research Participants
Origins of Research Participant Protection
Physical Harm, PsychologicalAbuse, and Legal Jeopardy
Other Harm to Participants
Informed Consent
Special Populations and New Inequalities
Privacy, Anonymity, and Confidentiality
Mandated Protections of Research Participants
Ethics and the Scientific Community
Ethics and the Sponsors of Research
Arriving at Particular Findings
Suppressing Findings
Concealing the True Sponsor
Politics of Research
Value-Free and Objective Research


ness will help you better understand the overall

research process.
Ethics include the concerns, dilemmas, and con­ Ethics begin and end with you, the individ­
flicts that arise over the proper way to conduct ual social researcher. A strong personal moral
research. Ethics help to define what is or is not code by the researcher is the best defense against
legitimate to do, or what "moral" research pro­
cedure involves. This is not as simple as it may
unethical behavior. Before, during, and after
conducting a study, a researcher has opportuni­
appear, because there are few ethical absolutes ties to, and should, reflect on the ethics of re­
and only agreed-upon broad principles. These search actions and consult his or her conscience.
principles require judgment to apply and some Ultimately, ethical research depends on the in­
may conflict with others in practice. Many ethi­ tegrity of an individual researcher.
cal issues ask you to balance two values: the pur­
suit of knowledge and the rights of research
participants or of others in society. Social re­
searchers balance potential benefits-such as

advancing the understanding of social life, im­

proving decision making, or helping research
participants-against potential costs-such as
loss of dignity, self-esteem, privacy, or democra­
Given that most people who conduct social re­
search are genuinely concerned about others,
you might ask, Why would any researcher ever
act in an ethically irresponsible manner? Most

tic freedoms. unethical behavior is due to a lack of awareness

Social researchers confront many ethical and pressures on researchers to take ethical
dilemmas and must decide how to act. They shortcuts. Researchers face pressures to build a
have a moral and professional obligation to be career, publish new findings, advance knowl­
ethical, even if research participants are unaware edge, gain prestige, impress family and friends,
of or unconcerned about ethics. hold on to a job, and so forth. Ethical research
Many areas of professional practice have will take longer to complete, cost more money,
ethical standards (e.g.,journalists, police depart­ be more complicated, and be less likely to pro­
ments, business corporations, etc.), but the eth­ duce unambiguous results. Plus, there are many
ical standards for doing social research are often opportunities in research to act unethically, the
stricter. To do professional social research, you odds of getting caught are small, and written
must both know the proper research techniques ethical standards are in the form of vague, loose
(e.g., sampling) and be sensitive to ethical con­ principles.
cerns. This is not always easy. For centuries, The ethical researcher gets few rewards and
moral, legal, and political philosophers debated wins no praise. The unethical researcher, if
the issues researchers regularly face. caught, faces public humiliation, a ruined career, ....... ...,.a.r

It is difficult to appreciate fully the ethical

dilemmas experienced by researchers until you
and possible legal action. The best preparation
for ethical behavior is to internalize a sensitivity
... ~_t_"
actually begin to do research, but waiting until
the middle of a study is too late. You need to pre­
to ethical concerns, to adopt a serious profes­
sional role, and to interact regularly with other
- .-dthat a 1918
pare yourself ahead of time and consider ethical researchers. Moreover, the scientific community I.adeanfrom
concerns as you design a study so that you can demands ethical behavior without exceptions. awzined~
build sound ethical practices into a study's de­ "asociol~~
sign. In addition, by developing sensitivity to ...... ~thr41
Scientific Misconduct
ethical issues, you will be alert to potential ethi­
cal concerns that can arise as you make decisions The research community and agencies that fund ~~.l~~~~
~S ration ~
while conducting a study. Also, an ethical aware­ research oppose a type of unethical behavior w1Ih the C.S. goYenl

called scientific misconduct; it includes research -----------1..

FIG U RE 3.1 Typology of Legal and
fraud and plagiarism. Scientific misconduct oc­
curs when a researcher falsifies or distorts the Moral Actions in Social
data or the methods of data collection, or plagia­ Research
rizes the work of others. It also includes signifi­
cant, unjustified departures from the generally ETHICAL

accepted scientific practices for doing and re­ Yes No

porting on research. Research fraud occurs when
a researcher fakes or invents data that he or she Yes Moral and Legal Legal but Immoral
did not really collect, or fails to honestly and No Illegalbut Moral Immoral and Illegal
fully report how he or she conducted a study. Al­
though rare, it is considered a very serious viola­
tion. The most famous case of research fraud
was that of Sir Cyril Burt, the father of British clearly unethical according to standards of pro­
educational psychology. Burt died in 1971 as an fessional behavior.i (See Figure 3.1 for relations
esteemed researcher who was famous for his between legal and moral actions.)
studies with twins that showed a genetic basis of
intelligence. In 1976, it was discovered that he
had falsified data and the names of coauthors.
Unfortunately, the scientific community had
been misled for nearly 30 years. More recently, a A professional researcher and the research par­
social psychologist was discovered to have fabri­ ticipants or employee-assistants are in a rela­
cated data for several experiments on sex bias tionship of unequal power and trust. An
conducted at Harvard University in the 1990s. experimenter, survey director, or research inves­
Plagiarism occurs when a researcher "steals" the tigator has power over participants and assis­
ideas or writings of another or uses them with­ tants, and in turn, they trust his or her judgment
out citing the source. Plagiarism also includes and authority. The researcher's credentials,
stealing the work of another researcher, an assis­ training, professional role, and the place of sci­
tant, or a student, and misrepresenting it as ence in modern society legitimate the power and
one's own. These are serious breaches of ethical make it into a form of expert authority. Some
standards. 1 ethical issues involve an abuse of power and
trust. A researcher's authority to conduct social
research and to earn the trust of others is ac­
Unethical but Legal
companied always by an unyielding ethical re­
Behavior may be unethical but legal (i.e., not sponsibility to guide, protect, and oversee the
break any law). A plagiarism case illustrates the interests of the people being studied.
distinction between legal and ethical behavior. When looking for ethical guidance, re­
The American Sociological Association docu­ searchers are not alone. They can turn to a num­
mented that a 1988 book without any footnotes ber of resources: professional colleagues, ethical
by a dean from Eastern New Mexico University advisory committees, institutional review boards
contained large sections of a 1978 dissertation or human subjects committees at a college or in­
that a sociology professor at Tufts University stitution (discussed later), codes of ethics by
wrote. Copying the dissertation was not illegal; it professional associations (discussed later in this
did not violate copyright law because the sociol­ chapter), and writings on ethics in research. The
ogist's dissertation did not have a copyright flied larger research community firmly supports and
with the U.S. government. Nevertheless, it was upholds ethical behavior, even if an individual

researcher is ultimately responsible to do what is others in Nazi Germany, and similar "medical
ethical in specific situations. experiments" to test biological weapons by
Japan in the 1940s. In these experiments, terrible
tortures were committed. For example, people
were placed in freezing water to see how long it
took them to die, people were purposely starved
to death, people were intentionally infected with
Have you ever been a participant in a research horrible diseases, and limbs were severed from
study? If so, how were you treated? More atten­ children and transplanted onto others.f
tion is focused on the possible negative effects of Such human rights violations did not occur
research on those being studied than any other
ethical issue, beginning with concerns about
biomedical research. Acting ethically requires
that a researcher balance the value of advancing
knowledge against the value of noninterference
only long ago. In a famous case of unethical re­
search, the Tuskegee Syphilis Study, also known
as Bad Blood, the President of the United States
admitted wrongdoing and formally apologized
in 1997 to the participant-victims. Until the
.. ......

in the lives of others. Either extreme causes 1970s, when a newspaper report caused a scan­
problems. Giving research participants absolute dal to erupt, the U.S. Public Health Service
rights of noninterference could make empirical sponsored a study in which poor, uneducated
research impossible, but giving researchers ab­ African American men in Alabama suffered and
solute rights of inquiry could nullify partici­ died of untreated syphilis, while researchers
pants' basic human rights. The moral question studied the severe physical disabilities that ap­
becomes: When, if ever, are researchers justified pear in advanced stages of the disease. The un­
in risking physical harm or injury to those being ethical study began in 1929, before penicillin was
studied, causing them great embarrassment or available to treat the disease, but it continued
inconvenience, violating their privacy, or fright­ long after treatment was available. Despite their
ening them? unethical treatment of the people, the re­
The law and codes of ethics recognize some searchers were able to publish their results for 40
clear prohibitions: Never cause unnecessary or years. The study ended in 1972, but a formal
irreversible harm to subjects; secure prior vol­ apology took another 25 years.!
untary consent when possible; and never unnec­ Unfortunately, the Bad Blood scandal is not
essarily humiliate, degrade, or release harmful unique. During the Cold War era, the U.S. gov­
information about specific individuals that was ernment periodically compromised ethical re­
collected for research purposes. In other words, search principles for military and political goals.
you should always show respect for the research In 1995, reports revealed that the government
participant. These are minimal standards and authorized injecting unknowing people with ra­
are subject to interpretation (e.g., What does dioactive material in the late 1940s. In the 1950s,
unnecessary mean in a specific situation?). the government warned Eastman Kodak and
other film manufacturers about nuclear fallout
from atomic tests to prevent fogged film, but it
Origins of Research Participant
did not warn nearby citizens of health hazards.
In the 1960s, the U.S. army gave unsuspecting people in
Concern over the treatment of research partici­ soldiers LSD (a hallucinogenic drug), causing se­ comfort or~
pants arose after the revelation of gross viola­ rious trauma. Today, researchers widely recog­ comfort! 1M 4
tions of basic human rights in the name of nize these to be violations of two fundamental obedience stu4j
science. The most notorious violations were ethical principles: Avoid physical harm and ob­ Some sav ~
"medical experiments" conducted on Jews and tain informed consent. 5 knowledge gaia

Physical Harm, Psychological Abuse, tential psychological harm that research partici­
and Legal Jeopardy pants experienced. Others believe that the ex­
treme stress and the risk of permanent harm
Social research can harm a research participant
were too great. Such an experiment could not be
in several ways: physical, psychological, and legal
conducted today because ofheightened sensitiv­
harm, as well as harm to a person's career, repu­
ity to the ethical issues involved.
tation, or income. Different types of harm are
Social researchers have created high levels of
more likely in other types ofresearch (e.g., in ex­
anxiety or discomfort. They have exposed par­
periments versus field research). It is a re­
ticipants to gruesome photos; falsely told male
searcher's responsibility to be aware of all types
students that they have strong feminine person­
of potential harm and to take specific actions to
ality traits; falsely told students that they have
minimize the risk to participants at all times. failed; created a situation of high fear (e.g.,
Physical Harm. Physical harm is rare. Even in smoke entering a room in which the door is
biomedical research, where the intervention into locked); asked participants to harm others;
a person's life is much greater, 3 to 5 percent of placed people in situations where they face social
studies involved any person who suffered any pressure to deny their convictions; and had par­
harm.? A straightforward ethical principle is that ticipants lie, cheat, or steal." Researchers who
researchers should never cause physical harm. study helping behavior often place participants
An ethical researcher anticipates risks before be­ in emergency situations to see whether they will
ginning a study, including basic safety concerns lend assistance. For example, Piliavin and asso­
(e.g., safe buildings, furniture, and equipment). ciates (1969) studied helping behavior in sub­
This means that he or she screens out high-risk ways by faking someone's collapse onto the
subjects (those with heart conditions, mental floor. In the field experiment, the riders in the
breakdown, seizures, etc.) if great stress is in­ subway car were unaware of the experiment and
volved and anticipates possible sources of injury did not volunteer to participate in it.
or physical attacks on research participants or The only researchers who might even con­
assistants. The researcher accepts moral and le­ sider conducting a study that purposely induces
gal responsibility for injury due to participation great stress or anxiety in research participants
in research and terminates a project immediately are very experienced and take all necessary pre­
if he or she can no longer fully guarantee the cautions before inducing anxiety or discomfort.
physical safety of the people involved (see the The researchers should consult with others who
Zimbardo study in Box 3.1). have conducted similar studies and mental
health professionals as they plan the study. They
Psychological Abuse, Stress, or Loss of Self­ should screen out high-risk populations (e.g.,
Esteem. The risk of physical harm is rare, but those with emotional problems or weak hearts),
social researchers can place people in highly and arrange for emergency interventions or ter­
stressful, embarrassing, anxiety-producing, or mination of the research if dangerous situations
unpleasant situations. Researchers want to learn arise. They must always obtain written informed
about people's responses in real-life, high-anxi­ consent (to be discussed) before the research
ety-producing situations, so they might place and debrief the people immediately afterward
people in realistic situations of psychological dis­ (i.e., explain any deception and what actually
comfort or stress. Is it unethical to cause dis­ happened in the study). Researchers should
comfort? The ethics of the famous Milgram never create unnecessary stress (i.e., beyond the
obedience study are still debated (see Box 3.1). minimal amount needed to create the desired ef­
Some say that the precautions taken and the fect) or stress that lacks a very clear, legitimate
knowledge gained outweighed the stress and po­ research purpose. Knowing what "minimal

Stanley Milgram's obedience study (Milgram, 1963, later, in disguise, Humphreys used a deceptive story
1965, 1974) attempted to discover how the hor­ about a health survey to interview the subjects in
rors of the Holocaust under the Nazis could have oc­ their homes. Humphreys was careful to keep names
curred by examining the strength of social pressure in safety deposit boxes, and identifiers with subject
to obey authority, After signing "informed consent names were burned. He significantly advanced knowl­
forms," subjects were assigned, in rigged random se­ edge of homosexuals who frequent "tearooms" and
lection,to be a "teacher" while a confederate was the overturned previous false beliefsabout them. There
"pupil," The teacher was to test the pupil's memory has been controversy over the study: The subjects
of word lists and increase the electric shock level if neverconsented; deception was used; and the names
the pupil made mistakes, The pupil was located in a could have been used to blackmail subjects, to end
nearby room, so the teacher could hear but not see marriages, or to initiate criminal prosecution.
the pupil. The shock apparatus was clearly labeled In the limbardo prison experiment (Zimbardo,
with increasing voltage. As the pupil made mistakes 1972,1973; Zimbardo et al., 1973, 1974), male
and the teacher turned switches, she or he also made students were divided into two role-playing groups:
noises as ifin severe pain.The researcher was present guards and prisoners. Before the experiment, volun­
and made comments such as "You must go on" to teer students were given personality tests, and only
the teacher. Milgram reported, "Subjects were ob­ those in the "normal" range were chosen. Volunteers
served to sweat, tremble, stutter, bite their lips, signed up for two weeks, and prisoners were told that
groan and dig their fingernails into their flesh. These they would be under surveillance and would have 'It"
were characteristic rather than exceptional re­ some civil rights suspended, but that no physical
sponses to the experiment" (Milgram, 1963:375).
The percentage of subjects who wouldshock to dan­
gerous levels was dramatically higher than expected.
Ethical concerns arose over the use of deception and
the extreme emotional stress experienced by
abuse was allowed. Ina simulated prison in the base­
ment of a Stanford University building, prisoners
were deindividualized (dressed in standard uniforms
and called only by their numbers) and guards were
militarized (with uniforms, nightsticks, and reflective
.......- ..



. .l1li£. . . . . .
subjects. sunglasses). Guards were told to maintain a reason­ filii
In Laud Humphreys's (Humphreys, 1975) tea­ able degree of order and served 8-hour shifts, while
room trade study (a study of male homosexual en­ prisoners were locked up 24 hours per day. Unex­ n

counters in public restroorns), about 100 men were pectedly, the volunteers became too caught up in
observed engaging in sexualacts as Humphreys pre­ their roles. Prisoners became passive and disorga­
tended to be a "watchqueen" (a voyeur and look­ nized, while guards became aggressive, arbitrary, and
out). Subjects were followed to their cars, and their dehumanizing. By the sixth day, Zimbardo called off
license numbers were secretly recorded. Names and the experiment for ethical reasons. The risk of per­
addresses were obtained from police registers when manent psychological harm,and even physical harm,
Humphreys posed as a market researcher. One year was too great.

amount" means comes with experience. It is best sitive researchers reduces the chances of making
to begin with too little stress, risking a finding of an ethical misjudgment.
no effect, than to create too much. It is always Research that induces great stress and anx­
wise to work in collaboration with other re­ iety in participants also carries the danger that
searchers when the risk to participants is high, experimenters will develop a callous or manip­
because the involvement ofseveral ethically sen­ ulative attitude toward others. Researchers

have reported feeling guilt and regret after con­ undermining future social research. At the same
ducting experiments that caused psychological time, a researcher who failsto report illegal behav­
harm to people. Experiments that place sub­ ior is indirectly permitting criminal behavior. He
jects in anxiety-producing situations may pro­ or she could be charged as an accessoryto a crime.
duce significant personal discomfort for the Cooperation with law-enforcement officials raises
ethical researcher. the question, Is the researcher a professional sci­
entist who protects research participants in the
Legal Harm. A researcher is responsible for process of seeking knowledge, or a free-lance un­
protecting research participants from increased dercover informant who is reallyworking for the
risk of arrest. If participation in research in­ police trying to "catch" criminals?
creases the risk of arrest, few individuals will
trust researchers or be willing to participate in Other Harm to Participants
future research. Potential legal harm is one crit­
Research participants may face other types of
icism of Humphreys's 1975 tearoom trade study
harm. For example, a survey interview may cre­
(see Box 3.1).
ate anxiety and discomfort if it asks people to re­
A related ethical issue arises when a re­
call unpleasant or traumatic events. An ethical
searcher learns of illegal activity when collecting
researcher must be sensitive to any harm to par­
data. A researcher must weigh the value of pro­
ticipants, consider precautions, and weigh po­
tecting the researcher-subject relationship and
tential harm against potential benefits.
the benefits to future researchers against poten­
Another type of harm is a negative impact
tial serious harm to innocent people. The re­
on the careers, reputations, or incomes of re­
searcher bears the cost of his or her judgment.
search participants. For example, a researcher
For example, in his field research on police, Van
conducts a survey of employees and concludes
Maanen (1982:114-115) reported seeing police
that the supervisor's performance is poor. As a
beat people and witnessing illegal acts and irreg­
consequence, the supervisor loses her job. Or, a
ular procedures, but said, "On and following
researcher studies homeless people living on the
these troublesome incidents I followed police
street. The fmdings show that many engage in
custom: I kept my mouth shut."
petty illegal acts to get food. As a consequence, a
Field researchers in particular can face
city government"cracks down" on the petty ille­
difficult ethical decisions. For example, when
gal acts and the homeless people can no longer
studying a mental institution, Taylor (1987) dis­
eat. What is the researcher's responsibility? The
covered the mistreatment and abuse of inmates
ethical researcher considers the consequences of
by the staff. He had two choices: Abandon the
research for those being studied. The general
study and call for an immediate investigation, or
goal is not to cause any harm simply because
keep quiet and continue with the study for sev­
someone was a research participant. However,
eral months, publicize the findings afterwards,
there is no set answer to such questions. A re­
and then become an advocate to end the abuse.
searcher must evaluate each case, weigh poten­
After weighing the situation, he followed the lat­
tial harm against potential benefits, and bear the
ter course and is now an activist for the rights of
responsibility for the decision.
mental institution inmates.
In some studies, observing illegal behavior
may be central to the research project. If a re­
searcher covertly observes and records illegal Has anyone ever told you a half-truth or lie to
behavior, then supplies the information to law-en­ get you to do something? How did you feel
forcement authorities, he or she isviolating ethical about it? Social researchers follow the ethical
standards regarding research participants and is principle of voluntary consent: Never force any­

one to participate in research, and do not lie to

anyone unless it is necessary and the only way to
be best. When in doubt, it is best to err in the di­
rection ofdisclosing one's true identity and pur­
...... "
cik.-*:1-. ­. . . . .
accomplish a legitimate research purpose. The pose. Covert research remains controversial, and ~ 1iIr
people who participate in social research should many researchers feel that all covert research is ......cd .

_ .
explicitly agree to participate. A person's right unethical. Even those who accept covert research
not to participate can be a critical issue when­ as ethical in certain situations say that it should "iripmtt""
ever the researcher uses deception, disguises the be used only when overt observation is impossi­
research, or uses covert research methods. ble. Whenever possible, the researcher should
Social researchers sometimes deceive or lie inform participants of the observation immedi­
to participants in field and experimental re­ ately afterwards and give them an opportunity
search. A researcher might misrepresent his or to express concerns.
her actions or true intentions for legitimate Deception and covert research may increase
methodological reasons. For example, if partici­ mistrust and cynicism as well as diminish public
pants knew the true purpose, they would modify respect for social research. Misrepresentation in
their behavior, making it impossible to learn of field research is analogous to being an under­
their real behavior. Another situation occurs cover agent or government informer in nonde­
when access to a research site would be impossi­ mocratic societies. The use of deception has a
ble if the researcher told the truth. Deception is long-term negative effect. It increases distrust
never preferable if the researcher can accomplish among people who are frequently studied and
the same thing without using deception. makes doing social research more difficult in the
Experimental researchers often deceive sub­ long term.
jects to prevent them from learning the hypoth­
esis being tested and to reduce "reactive effects"
Informed Consent
(see Chapter 8). Deception is acceptable only if a
researcher can show that it has a clear, specific A fundamental ethical principle of social re­
methodological purpose, and even then, the re­ search is: Never coerce anyone into participat­
searcher should use it only to the minimal de­ ing; participation must be voluntary at all times.
gree necessary. Researchers who use deception Permission alone is not enough; people need to
should always obtain informed consent, never know what they are being asked to participate in 2. A statement
misrepresent risks, and always explain the actual so that they can make an informed decision. Par­ ated with
conditions to participants afterwards. You might ticipants can become aware of their rights and 3. A guarantee
ask, How can a researcher obtain prior informed what they are getting involved in when they read
consent and still use deception? He or she can and sign a statement giving informed consent­
4. The identi
describe the basic procedures involved and con­ an agreement by participants stating they are where to
ceal only specific information about hypotheses willing to be in a study and they know some­ rights or .
being tested. thing about what the research procedure will in­ 5. AstatemerrtuJ
Sometimes field researchers use covert ob­ volve. untary and GIRl
servation to gain entry to field research settings. Governments vary in the requirement for out penatty 1
6. A statement of~
In studies of cults, small extremist political sects, informed consent. The U.S. federal government
illegal or deviant behavior, or behavior in a large does not require informed consent in all re­ be used ;
public area, it may be impossible to conduct re­ search involving human subjects. Nevertheless,
search if a researcher announces and discloses researchers should get written informed consent
7. A statement aIj
provided to ~
her or his true purpose. If a covert stance is not unless there are good reasons for not obtaining it involved 1
essential, a researcher should not use it. If he or (e.g., covert field research, use of secondary data,
8. An offer to ~
she does not know whether covert access is nec­ etc.) as judged by an institutional review board
essary, then a strategy of gradual disclosure may (IRB) (see the later discussion ofIRBs).

Informed consent statements provide spe­ dition of continued employment. It is unethical

cific information (see Box 3.2). A general state­ even if someone other than the researcher (e.g.,
ment about the kinds of procedures or questions an employer) coerces people (e.g., employees) to
involved and the uses of the data are sufficient participate in research.
for informed consent. Studies suggest that par­ Full disclosure with the researcher's identifi­
ticipants who receive a full informed consent cation helps to protect research participants
statement do not respond differently from those against fraudulent research and to protect legit­
who do not. If anything, people who refused to imate researchers. Informed consent lessens the
sign such a statement were more likely to guess chance that a con artist in the guise of a re­
or answer "no response" to questions. searcher will defraud or abuse people. It also re­
It is unethical to coerce people to partici­ duces the chance that someone will use a bogus
pate, including offering them special benefits researcher identity to market products or obtain
that they cannot otherwise attain. For example, personal information on people for unethical
it is unethical for a commanding officer to order purposes.
a soldier to participate in a study, for a professor Legally, a signed informed consent state­
to require a student to be a research subject in ment is optional for most survey, field, and sec­
order to pass a course, or for an employer to ex­ ondary data research, but it is often mandatory
pect an employee to complete a survey as a con- for experimental research. Informed consent is
impossible to obtain in existing statistics and
documentary research. The general rule is: The
greater the risk of potential harm to research
participants, the greater the need to obtain a
written informed consent statement from them.
In sum, there are many sound reasons to get in­
Informed consent statements contain the following:
formed consent and few reasons not to get it.
1. A brief description of the purpose and proce­
dure of the research, including the expected du­
Special Populations and New
ration of the study
2. A statement of any risks or discomfort associ­
ated with participation Some populations or groups of research partici­
3. A guarantee of anonymity and the confidential­ pants are not capable of giving true voluntary in­
ity of records formed consent. Special populations are people
4. The identification of the researcher and of who lack the necessary cognitive competency to
where to receive information about subjects' give valid informed consent or people in a weak
rights or questions about the study position who might cast aside their freedom to
5. A statement that participation is completely vol­ refuse to participate in a study. Students, prison
untary and can be terminated at any time with­ inmates, employees, military personnel, the
out penalty homeless, welfare recipients, children, and the
developmentally disabled may not be fully capa­
6. A statement of alternative procedures that may
be used
ble of making a decision, or they may agree to
participate only because they see their participa­
7. A statement of any benefits or compensation
tion as a way to obtain a desired good-such as
provided to subjects and the number of subjects
higher grades, early parole, promotions, or addi­
tional services. It is unethical to involve "incom­
8. An offer to provide a summary of findings petent" people (e.g., children, mentally disabled,
etc.) in research unless a researcher meets two

minimal conditions: (1) a legal guardian grants long as it meets three conditions: it is attached to
written permission and (2) the researcher fol­ a clear educational objective, the students have a
lows all standard ethical principles to protect choice of research experience or an alternative
participants from harm. For example, a re­ activity, and all other ethical principles of re­
searcher wants to conduct a survey of high search are followed.
school students to learn about their sexual be­
havior and drug/alcohol use. If the survey is con­ Avoid Creating New Inequalities. Another
ducted on schoolproperty, school officials must type of harm occurs when one group of people is
give officialpermission. For any research partic­ denied a service or benefit as a result of partici­
ipant who is a legal minor (usually under 18 pating in a research project. For example, a re­
years old), written parental permission is searcher might have a new treatment for people
needed. It is best to ask permission from each with a terrible disease, such as acquired immune
student, as well. deficiency syndrome (AIDS). To determine the
The use of coercion to participate can be a effects of the new treatment, half the group is
tricky issue, and it depends on the specifics of a randomly chosen to receive the treatment, while
situation. For example, a convicted criminal others receive nothing. The design may clearly
faces the alternative of imprisonment or partici­ show whether the treatment is effective, but par­
pation in an experimental rehabilitation pro­ ticipants in the group who receive no treatment
gram. The convicted criminal may not believe in may die. Of course, those receiving the treat­
the benefits of the program, but the researcher ment may also die, until more is known about
may believe that it will help the criminal. This is whether it is effective. Is it ethical to deny people
a case of coercion. A researcher must honestly who have been randomly assigned to a study
judge whether the benefits to the criminal and to group the potentially life-saving treatment?
society greatly outweigh the ethical prohibition What if a clear, definitive test of whether a treat­
on coercion. This is risky. History shows many ment is effective requires that one study group
cases in which a researcher believed he or she receive no treatment?
was doing something "for the good of' someone A researcher can reduce creating a new in­
in a powerless position (e.g., prisoners, students, equality among research participants when the
homosexuals), but it turned out that the "good" outcome has a major impact on their survival or
actually was for the researcher or a powerful or­ quality oflife in three ways.First, the people who
ganization in society, and it did more harm than do not receive the "new, improved" treatment
good to the research participant. continue to receive the best previously accept­
You may have been in a social science class able treatment. In other words, instead of deny­
in which a teacher required you to participate as ing all assistance, they get the best treatment
a subject in a research project. This is a special available prior to the new one being tested. This
case of coercion and is usually ethical. Teachers ensures that people will not suffer in absolute
have made three arguments in favor of requiring terms, even if they temporarily fall behind in rel­
student participation: (1) it would be difficult ative terms. Second, researchers can use a
and prohibitively expensive to get participants crossover design, which is when a study group
otherwise, (2) the knowledge created from re­ that gets no treatment in the first phase of the
search with students serving as subjects benefits experiment becomes the group with the treat­
future students and society, and (3) students will ment in the second phase, and viceversa. Finally,
learn more about research by experiencing it di­ the researcher continuously monitors results. If
rectly in a realistic research setting. Of the three it appears early in the study that the new treat­
arguments, only the third justifies limited coer­ ment is highly effective, the researcher should
cion. This limited coercion is acceptable only as offer it to those in the control group. Also, in

high-risk experiments with medical treatments Anonymity. Researchers protect privacy by

or possible physical harm, researchers may use not disclosing a participant's identity after in­
animal or other surrogates for humans. formation is gathered. This takes two forms,
both ofwhich require separating an individual's
Privacy, Anonymity, and identity from his or her responses: anonymity
Confidentiality and confidentiality. Anonymity means that peo­
ple remain anonymous or nameless. For exam­
How would you feelif private details about your
ple, a Heldresearcher provides a social picture of
personal lifewere shared with the public without
a particular individual, but gives a fictitious
your knowledge? Because social researchers name and location, and alters some characteris­
sometimes transgress the privacy of people in tics. The subject's identity is protected, and the
order to study social behavior, they must take individual remains unknown or anonymous.
several precautions to protect research partici­ Survey and experimental researchers discard the
pants' privacy.
names or addresses of subjects as soon as possi­
Privacy. Survey researchers invade a person's ble and refer to participants by a code number
privacy when they probe into beliefs, back­ only to protect anonymity. If a researcher uses a
grounds, and behaviors in a way that reveals in­ mail survey and includes a code on the ques­
timate private details. Experimental researchers tionnaire to determine which respondents failed
sometimes use two-way mirrors or hidden mi­ to respond, he or she is not keeping respondents
crophones to "spy" on subjects. Even if people anonymous during that phase of the study. In
know they are being studied, they are unaware of panel studies, researchers track the same indi­
what the experimenter is looking for. Field re­ viduals over time, so they do not uphold partic­
searchers may observe private aspects ofbehav­ ipant anonymity within the study. Likewise,
ior or eavesdrop on conversations. historical researchers use specific names in his­
In Held research, privacy may be violated torical or documentary research. They may do
without advance warning. When Humphreys so if the original information was from public
(1975) served as a "watchqueen" in a public rest­ sources; if the sources were not publicly avail­
room where homosexual contacts took place, he able, a researcher must obtain written permis­
observed very private behavior without inform­ sion from the owner of the documents to use
ing subjects. When Piliavin and colleagues specific names.
(1969) had people collapse on subways to study It is difficult to protect research participant
helping behavior, those in the subway car had anonymity. rn one study about a fictitious town,
the privacy of their ride violated. People have "Springdale," in Small Town in Mass Society
been studied in public places (e.g., in waiting (Vidich and Bensman, 1968), it was easyto iden­
rooms, walking down the street, in classrooms, tify the town and specificindividuals in it. Town
etc.), but some "public" places are more private residents became upset about how the re­
than others (consider, for example, the use of searchers portrayed them and staged a parade
periscopes to observe people who thought they mocking the researchers. People often recognize
were alone in a public toilet stall). the towns studied in community research. Yet, if
Eavesdropping on conversations and ob­ a researcher protects the identities of individuals
serving people in quasi-private areas raises ethi­ with fictitious information, the gap between
cal concerns. To be ethical, a researcher violates what was studied and what is reported to others
privacy only to the minimum degree necessary raises questions about what was found and what
and only for legitimate research purposes. In ad­ was made up. A researcher may breach a promise
dition' he or she takes steps to protect the infor­ of anonymity unknowingly in small samples. For
mation on participants from public disclosure. example, let us say you conduct a survey of 100

college students and ask many questions on a obligated them to destroy the records rather
questionnaire, including age, sex, religion, and than give them to government officials.
hometown. The sample contains one 22-year-old
Jewishmale born in Stratford, Ontario. With this
information, you could find out who the specific
individual is and how he answered very personal
Confidentiality can sometimes protect re­
search participants from legal or physical harm.
In a study of illegal drug users in rural Ohio,
Draus and associates (2005) took great care to ..
- ­• .-IIiai. .


questions, even though his name was not directly protect the research participants. They con­
recorded on the questionnaire. ducted interviews in large multiuse buildings, -
avoided references to illegal drugs in written
Confidentiality. Even if a researcher cannot documents, did not mention of names of drug . . . .asAr
protect anonymity, he or she alwaysshould pro­ dealers and locations, and did not affiliate with 6r "
tect participant confidentiality. Anonymity drug rehabilitation services, which had ties to iaaJIs; . .
means protecting the identity of specific individ­ law enforcement. They noted, "We intentionally
uals from being known. Confidentiality can avoided contact with local police, prosecutors,
include information with participant names at­ or parole officers" and "surveillance of the pro­
tached, but the researcher holds it in confidence ject by local law enforcement was a source of
or keeps it secret from public disclosure. The re­ concern" (p. 169). In other situations, other
searcher releasesdata in a way that does not per­ principles may take precedence over protecting
mit linking specific individuals to responses and research participant confidentiality. For exam­
presents it publicly only in an aggregate form ple, when studying patients in a mental hospital,
(e.g., as percentages, statistical means, etc.). a researcher discovers that a patient is preparing
A researcher can provide anonymity with­ to kill an attendant. The researcher must weigh
out confidentiality, or vice versa, although they the benefit of confidentiality against the poten­
usually go together. Anonymity without confi­ tial harm to the attendant.
dentiality occurs if all the details about a specific Social researchers can pay high personal
individual are made public, but the individual's costs for being ethical. Although he was never
name is withheld. Confidentiality without accused or convicted of breaking any law and he
anonymity occurs if detailed information is not closely followed the ethical principles of the
made public, but a researcher privately links in­ American Sociological Association, Professor
dividual names to specific responses. Rik Scarce spent 16 weeks in a Spokane jail for
Attempts to protect the identity of subjects contempt of court because he refused to testify
from public disclosure has resulted in elaborate before a grand jury and break the confidentiality
procedures: eliciting anonymous responses, us­ of social research data. Scarce had been studying
ing a third-party custodian who holds the key to radical animal liberation groups and had already
coded lists, or using the random-response tech­ published one book on the subject. He had in­
nique. Past abuses suggest that such measures terviewed a research participant who was sus­
may be necessary. For example, Diener and pected ofleading a group that broke into animal
Crandall (1978:70) reported that during the facilities and caused $150,000 damage. Two
1950s,the U.S. State Department and the FBIre­ judges refused to acknowledge the confidential­
quested research records on individuals who had ity of social research data."
been involved in the famous Kinsey sex study. A special concern with anonymity and con­
The Kinsey Sex Institute refused to comply with fidentiality arises when a researcher studies
the government. The institute threatened to de­ "captive" populations (e.g., students, prisoners,
stroy all records rather than release any. Eventu­ employees, patients, and soldiers). Gatekeepers,
ally,the government agenciesbacked down. The or those in positions of authority, may restrict
moral duty and ethical code of the researchers access unless they receive information on sub­

jects.?For example, a researcher studies drug use pact of research procedures on human partici­
and sexual activity among high school students. pants and applies ethical guidelines by reviewing
School authorities agree to cooperate under two research procedures at a preliminary stage when
conditions: (1) students need parental permis­ first proposed. Some forms of research, educa­
sion to participate and (2) school officials get the tional tests, normal educational practice, most
names of all drug users and sexually active stu­ nonsensitive surveys, most observation ofpublic
dents in order to assist the students with coun­ behavior, and studies of existing data in which
seling and to inform the students' parents. An individuals cannot be identified are exempt
ethical researcher will refuse to continue rather from institutional review boards.
than meet the second condition. Even though
the officials claim to have the participants' best
interests in mind, the privacy of participants will
be violated and they could be in legal harm as a
result of disclosure. If the school officials really
want to assist the students and not use re­ Physicians, attorneys, family counselors, social
searchers as spies, they could develop an out­ workers, and other professionals have a code of
reach program of their own. ethics and peer review boards or licensing regu­
lations. The codes formalize professional stan­
dards and provide guidance when questions
Mandated Protections of Research
arise in practice. Social researchers do not pro­
vide a service for a fee, they receive limited ethi­
Many governments have regulations and laws to cal training, and rarely are they licensed. They
protect research participants and their rights. In incorporate ethical concerns into research be­
the United States, legal restraint is found in rules cause it is morally and socially responsible, and
and regulations issued by the U.S. Department to protect social research from charges of insen­
of Health and Human Services Office for the sitivity or abusing people. Professional social sci­
Protection from Research Risks.Although this is ence associations have codes of ethics that
only one federal agency, most researchers and identify proper and improper behavior. They
other government agencies look to it for guid­ represent a consensus of professionals on ethics.
ance. The National Research Act (1974) estab­ All researchers may not agree on all ethical is­
lished the National Commission for the sues, and ethical rules are subject to interpreta­
Protection of Human Subjects in Biomedical tion, but researchers are expected to uphold
and Behavioral Research, which significantly ex­ ethical standards as part of their membership in
panded regulations and required informed con­ a professional community.
sent in most social research. The responsibility Codes of research ethics can be traced to the
for safeguarding ethical standards was assigned Nuremberg code adopted during the Nurem­
to research institutes and universities. The De­ berg Military Tribunal on Nazi war crimes held
partment of Health and Human Services issued by the Allied Powers immediately after World
regulations in 1981, which are still in force. Fed­ War II. The code, developed as a response to the
eral regulations follow a biomedical model and cruelty ofconcentration camp experiments, out­
protect subjects from physical harm. Other rules lines ethical principles and rights of human sub­
require institutional review boards (IREs) at all jects. These include the following:
research institutes, colleges, and universities to
review all use of human subjects. An IRB is a • The principle of voluntary consent
committee ofresearchers and community mem­ • Avoidance of unnecessary physical and
bers that oversees, monitors, and reviewsthe im­ mental suffering

• Avoidance of any experiment where death

or disabling injury is likely
• Termination of research if its continuation
is likely to cause injury, disability, or death
• The principle that experiments should be
conducted by highly qualified people using
the highest levels of skill and care
• Ethical responsibility rests with the individual re­
• Do not exploit subjects or students for personal ••

• The principle that the results should be for
the good of society and unattainable by any
• Some form of informed consent is highly recom­
_an b..,

other method mended or required.

The principles in the Nuremberg code dealt • Honor all guarantees of privacy, confidentiality,
and anonymity.
with the treatment of human subjects and fo­
cused on medical experimentation, but they be­ • Do not coerce or humiliate subjects. • ...a.........
came the basis for the ethical codes in social • Use deception only if needed, and always accom­ .td~

research. Similar codes of human rights, such as pany it with debriefing. Jdilsrthr=.-_
the 1948 Universal Declaration of Human ...._~~
• Use the research method that is appropriate to a
Rights by the United Nations and the 1964 Dec­ topic.
laration of Helsinki, also have implications for Z
social researchers. Box 3.3 lists some of the basic • Detect and remove undesirable consequences to
research subjects.
principles of ethical social research.
Professional social science associations have • Anticipate repercussions of the research or publi­
committees that review codes of ethics and hear cation of results.
about possible violations, but there is no formal • Identify the sponsor who funded the research.
policing of the codes. The penalty for a minor vi­
• Cooperate with host nations when doing compar­
olation rarely goes beyond a letter of complaint.
ative research.
If laws have not been violated, the most extreme
penalty is the negative publicity surrounding a • Release the details of the study design with the
well-documented and serious ethical violation. results.
The publicity may result in the loss of employ­ • Make interpretations of results consistent with
ment, a refusal to publish the researcher's find­ the data.
ings in scholarlyjournals, and a prohibition from • Use high methodological standards and strive for
receiving funding for research-in other words, accuracy.
banishment from the community of professional
• Do not conduct secret research.
Codes of ethics do more than codify think­
ing and provide individual researchers with
guidance; they also help universities and other
institutions defend ethical research against
abuses. For example, after interviewing 24 staff the university and demanded to know who on
members and conducting observations, a re­ their staff had talked to the researcher, with im­
searcher in 1994 documented that the staff at the plications that there might be reprisals. The uni­
Milwaukee Public Defenders Office were seri­ versity administration defended the researcher
ously overworked and could not effectivelypro­ and refused to release the information, citing
vide legal defense for poor people. Learning of widely accepted codes that protect human re­
the findings, top officials at the office contacted search participants. 10

tions, lowered pay, an undesirable transfer,

abandonment by friends at work, or incurring
legal costs. There is no guarantee that doing the
ethical-moral thing will stop the unethical be­
havior or protect the honest researcher from
You might find a job where you do research for retaliation.
a sponsor-an employer, a government agency, Applied social researchers in sponsored re­
or a private firm that contracts with a researcher search settings need to think seriously about
to conduct research. Special ethical problems their professional roles. They may want to main­
arise when a sponsor pays for research, especially tain some independence from an employer and
applied research. Researchers may be asked to affirm their membership in a community of
compromise ethical or professional research dedicated professionals. Many find a defense
standards as a condition for receiving a contract against sponsor pressures by participating in
or for continued employment. Researchers need professional organizations (e.g., the Evaluation
to set ethical boundaries beyond which they will Research Society), maintaining regular contacts
refuse the sponsor's demands. When confronted with researchers outside the sponsoring organi­
with an illegitimate demand from a sponsor, a zation, and staying current with the best re­
researcher has three basic choices: loyalty to an search practices. The researcher least likely to
organization or larger group, exiting from the uphold ethical standards in a sponsored setting
situation, or voicing opposition. I I These present is someone who is isolated and professionally in­
themselves as caving in to the sponsor, quitting, secure. Whatever the situation, unethical behav­
or becoming a whistle-blower. The researcher ior is never justified by the argument that "If I
must choose his or her own course of action, but didn't do it, someone else would have."
it is best to consider ethical issues early in a rela­
tionship with a sponsor and to express concerns
Arriving at Particular Findings
up front. Whistle-blowinginvolves the researcher
who sees an ethical wrongdoing, and who can­ What should you do if a sponsor tells you, di­
not stop it after informing superiors and ex­ rectly or indirectly, what results you should
hausting internal avenues to resolve the issue. come up with before you do a study? An ethical
He or she then turns to outsiders and informs an researcher will refuse to participate ifhe or she is
external audience, agency, or the media. The told to arrive at specific results as a precondition
whistle-blowing researcher must be convinced for doing research. Legitimate research is con­
that the breach of ethics is serious and approved ducted without restrictions on the possible find­
of in the organization. It is risky. The outsiders ings that a study might yield.
mayor may not be interested in the problem or An example of pressure to arrive at particu­
able to help. Outsiders often have their own pri­ lar findings is in the area of educational testing.
orities (making an organization look bad, sensa­ Standardized tests to measure achievement by
tionalizing the problem, etc.) that differ from the U.S. school children have come under criticism.
researcher's primary concern (ending the uneth­ For example, children in about 90 percent of
ical behavior). Supervisors or managers may try school districts in the United States score "above
to discredit or punish anyone who exposes prob­ average" on such tests. This was called the Lake
lems and acts disloyal. Under the best of condi­ Wobegon effect after the mythical town of Lake
tions, the issue may take a long time to resolve Wobegon, where, according to radio show host
and create great emotional strain. Bydoing what Garrison Keillor, "all the children are above av­
is moral, a whistle-blower needs to be prepared erage." The main reason for this finding was that
to make sacrifices-loss of a job or no promo­ the researchers compared scores of current stu­

dents with those of students many years ago. gives the sponsors whatever they want, even if it
Many teachers, school principals, superinten­ is ethically wrong, or a professional who is oblig­
dents, and school boards pressured for a type of ated to teach, guide, or even oppose sponsors in
result that would allow them to report to par­ the service of higher moral principles.
ents and voters that their school district was A researcher should ask: Why would spon­
"above average."l2 sors want the social research conducted if they
are not interested in using the findings or in the
Limits on How to Conduct Studies. Is it ethi­ truth? The answer is that some sponsors are not
cally acceptable for a sponsor to limit research by interested in the truth and have no respect for
defining what a researcher can study or by limit­ the scientific process. They see social research
ing the techniques used? Sponsors can legiti­ only as "a cover" to legitimate a decision or prac­
mately set some conditions on research tice that they plan to carry out, but use research
techniques used (e.g., survey versus experiment) to justify their action or deflect criticism. They
and limit costs for research. However, the re­ abuse the researcher's professional status and
searcher must follow generally accepted research
methods. Researchers must give a realistic ap­
undermine integrity of science to advance their
own narrow goals. They are being deceitful by

.. .­
praisal of what can be accomplished for a given
level of funding. The issue of limits is common
in contract research, when a firm or government
agency asks for work on a particular research
trying to "cash in" on social research's reputa­
tion for honesty. When such a situation occurs,
an ethical researcher has a moral responsibility
to expose and stop the abuse.
.. .. '
project. There is often a tradeoff between quality
and cost. Plus, once the research begins, a re­
Suppressing Findings
searcher may need to redesign the project, or
costs may be higher. The contract procedure What happens if you conduct a study and the

makes midstream changes difficult. A researcher findings make the sponsor look bad, then the
may find that he or she is forced by the contract sponsor does not want to release the results? i 5 "rl
to use research procedures or methods that are This is a common situation for many applied re­ g "


less than ideal. The researcher then confronts a searchers. For example, a sociologist conducted
dilemma: complete the contract and do low­ a study for a state government lottery commis­ ~to
....' "
01' to
quality research, or fail to fulfill the contract and sion on the effects of state government-spon­
lose money and future jobs. sored gambling. After she completed the report, "

A researcher should refuse to continue a but before releasing it to the public, the commis­ dcKmne.a ..
study if he or she cannot uphold generally ac­ sion asked her to remove sections that outlined potitial appai. .
cepted standards of research. If a sponsor de­ the many negative social effects of gambling and who hadpm.-"
mands a biased sample or leading survey to eliminate her recommendations to create so­ -asdlargOO
questions, the ethical researcher should refuse to cial services to help the anticipated increase of ports to ~
cooperate. If a legitimate study shows a spon­ compulsive gamblers. The researcher found her­ documented
sor's pet idea or project to be disaster, a re­ self in a difficult position and faced two conflict­ mdglobal-'"
searcher may anticipate the end of employment ing values: do what the sponsor requested and Inspol-1IllII
or pressure to violate professional research stan­ paid for, or reveal the truth to the public but ~ CODlibl"
dards. In the long run, the sponsor, the re­ then suffer the consequencesrP bt?ginning the
searcher, the scientific community, and society Government agencies may suppress scien­ effect. It mae
in general are harmed by the violation of sound tific information that contradicts official policy ...i thout such .a
research practice. The researcher has to decide or embarrasses high officials. Retaliation against researchers ...no
whether he or she is a "hired hand" who always social researchers employed by government do so. Altenlali.

agencies who make the information public also sponsor's criticism and hostility and release the
occurs. In 2004, leading scientists, Nobel laure­ findings over the sponsor's objections. Most re­
ates, leading medical experts, former federal searchers prefer the first choice, since the second
agency directors, and university chairs and pres­ one may scare away future sponsors.
idents signed a statement voicing concern over Social researchers sometimes self-censor or
the misuse of science by the George W. Bush ad­ delay the release of findings. They do this to pro­
ministration. Major accusations included su­ tect the identity of informants, to maintain ac­
pressing research findings and stacking scientific cess to a research site, to hold on to their jobs, or
advisory committees with ideologically commit­ to protect the personal safety of themselves or
ted advocates rather than impartial scientists. family members. IS This is a less disturbing type
Other complaints included limiting the public of censorship because it is not imposed by an
release studies on auto-saftey data, negative data outside power. It is done by someone who is
about pharmaceuticals, and studies on pollu­ close to the research and who is knowledgeable
tion. These involved industries that were major about possible consequences. Researchers shoul­
political campaign supporters of the administra­ der the ultimate responsibility for their research.
tion. Additional criticisms appeared over re­ Often, they can draw on many different re­
moving a government fact sheet citing studies sources but they face many competing pressures,
that showed no relationship between abortions as well.
and breast cancer, removing study results about
positive effects of condom use in pregnancy pre­
Concealing the True Sponsor
vention, holding back information on positive
aspects of stem cell research, and requiring re­ Is it ethical to keep the identity of a sponsor se­
searchers to revise their study findings on dan­ cret? For example, an abortion clinic funds a
gers of arctic oil drilling and endangered species study on members of religious groups who op­
so they would conform to the administration's pose abortion, but it tells the researcher not to
political agenda. An independent 2005 survey of reveal to participants who is funding the study.
460 biologists who worked for Fisheries Service The researcher must balance the ethical rule that
found that about one-third said they were di­ it is usually best to reveal a sponsor's identity to
rected to suppress findings for nonscientific rea­ participants against both the sponsor's desire for
sons or to inappropriately exclude or alter confidentiality and reduced cooperation by par­
technical information from an official scientific ticipants in the study. In general, an ethical re­
document. In June 2005, it was discovered that a searcher will tell subjects who is sponsoring a
political appointee without scientific training study unless there is a strong methodological
who had previously been an oil industry lobbyist reason for not doing so. When reporting or pub­
was charged with editing official government re­ lishing results, the ethical mandate is very clear:
ports to play down the research findings that A researcher must alwaysreveal the sponsor who
documented linkages between such emissions provides funds for a study.
and global warming. 14
In sponsored research, a researcher can ne­
gotiate conditions for releasing findings prior to
beginning the study and sign a contract to that
effect. It may be unwise to conduct the study Ethics largely address moral concerns and stan­
without such a guarantee, although competing dards of professional conduct in research that
researchers who have fewer ethical scruples may are under the researcher's control. Political con­
do so. Alternatively, a researcher can accept the cerns also affectsocial research, but many are be­

1~"\.\.~\\\.~ '-~"\.\.\."'t~\~\ "'t~%~'O.."'t,-\\.~"'t~.. \.\\.~ ~~\\.\.\.~ ~\ 'l'l'O.."\.\.\. \.~ ~"'t~\.~,-\. ~"'t 'O..~'l'O.."\.\.'-~ \.~~\."'t ~~\\.\.\.'-'O..\­
research usually involve actions by organized ad­ fmancial position, and lear social researchers
vocacy groups, powerful interests in society, might yield findings showing that their actions
governments, or politicians trying to restrict or are harmful to the public or some sectors of
control the direction of social research. Histori­ society. And third, some people in society do not
cally, the politieal influence over social research respect the ideals of science to pursue truth/
has included preventing researchers from con­ knowledge and instead view scientific research
ducting a study, cutting off or redirecting funds only as cover for advancing private interests (see
for research, harassing individual researchers, Box 3.4).
censoring the release of research findings, and

...• ...-...
using social research as a cover or guise for
covert government intelligence/military actions.

For example, U.S. Congress members targeted _ r5


and eliminated funding for research projects

that independent panels of scientists recom­ You have undoubtedly heard about "value-free"
mended because Congress did not like the topics research and the importance of being "objec­ -~
that would be studied, and politically appointed tive" in research. This is not as simple at it might
officials shifted research funds to support more first appear for several reasons. First, there are
studies on topics consistent with their political different meanings of the terms valuefree and
views while ending support for studies on topics objective. Second, different approaches to social
that might contradict their views. A large com­ science (positivism, interpretative, critical) hold
pany threatened an individual researcher with a different views on the issue. And last, even re­
lawsuit for delivering expert testimony in public searchers who agree that social research should
about research findings that revealed its past bad be value free and objective do not believe that it
conduct. Until about a decade ago, social re­ needs to be totally devoid of all values.
searchers who appeared to be independent were There are two basic ways the term value free
actually conducting covert U.S. government in­ is used: research that is free from any prior as­
telligence activities.l?
Most uses of political or financial influence
to control social research share a desire to limit
knowledge creation or restrict the autonomous
scientific investigation of controversial topics.
sumptions, theoretical stand, or value position,
and research that is conducted free of influence
from an individual researcher's personal preju­
dices/beliefs. Likewise, objective can mean focus­
ing only on what is external or visible, or it can
....... .

~ I I r I .

Attempts at control seem motivated by a fear mean following clear and publicly accepted re­ ......aIIIIII......
that researchers might discover something dam­ search procedures and not haphazard, personal JIIIlIPIr ~G, . . .
aging if they have freedom of inquiry. This ones.
shows that free scientific inquiry is connected to The three approaches to social science that
fundamental political ideals of open public de­ you read about in Chapter 2 hold different posi­
bate, democracy, and freedom of expression. tions on the importance of value-free, objective
The attempts to block and steer social re­ research. Positivism puts a high value on such
search have three main reasons. First, some peo­ research. An interpretive approach seriously
ple defend or advance positions and knowledge questions whether it is possible, since human
that originate in deeply held ideological, politi­ values/beliefs pervade all aspects of human ac­
cal, or religious beliefs, and fear that social re­ tivities, including research. Instead of eliminat­
searchers might produce knowledge that ing values and subjective dimension, it suggestsa
contradicts them. Second, powerful interests relativist stance-no single value position is bet­

Michael Burawoy (2004, 2005) distinguished clients. Both rely on professional social science for
among four ideal types of social research: policy, pro­ theories, bodies of knowledge, and techniques for
fessional, critical, and public. The aim of public soci­ gathering and analyzing data. Critical social science,
ology (or social science, more generally) is to enrich as was discussed in Chapter 2, emphasizes demysti­
public debate over moral and political issues by in­ fying and raising questioning about basic conditions.
fusing such debate with social theory and research. The primary audience for professional and critical
Public sociology frequently overlaps with action-ori­ social science are members of the scientific commu­
ented research. Burawoy argued that the place of so­ nity, whereas the main audience for public and policy
cial research in society centers on how one answers research are nonexperts and practitioners. Both crit­
two questions: Knowledge for whom? and Knowl­ ical and public social science seek to infuse a moral,
edge for what? The first question focuses on the value dimension into social research and they try to
sources of research questions and how results are generate debates over moral-political values. Profes­
used. The second question looks at the source of re­ sional and policy social science are less concerned
search goals. Are they handed down by some exter­ about debates over moral or value issues and may
nal sponsor or agency or are they concerned with avoid them. Instead, their focus is more on being ef­
debates over larger societal political-moral issues? fective in providing advances to basic knowledge or
Public social science tries to generate a conversation specific solutions to practical problems. Both public
or debate between researchers and public. By con­ and policy social science are applied research and
strast, policy social science focuses on finding solu­ have a relevance beyond the community of scientific
tions to specific problems as defined by sponsors or researchers.

ter than any other. A critical approach also ques­ or her own value position explicit, reflect care­
tions value-free research, but sees it often as a fully on reasons for doing a study and the proce­
sham. dures used, and communicate in a candid, clear
Value free means free of everyone's values manner exactlyhow the study was conducted. In
except those of science, and objective means fol­ this way, other researchers see the role of a re­
lowing established rules or procedures that some searcher's values and judge for themselves
people created, without considering who they whether the values unfairly influenced a study's
represent and how they created the rules. In findings.
other words, a critical approach sees all research Even highly positivist researchers who ad­
as containing some values, so those who claim to vocate value-free and objective studies admit a
be value free are just hiding theirs. Those who limited place for some personal, moral values.
follow an interpretive and critical approach and Many hold that a researcher's personal, moral
reject value-free research do not embrace sloppy position can enter when it comes to deciding
and haphazard research, research procedures what topic to study and how to disseminate
that follow a particular researcher's whims, or a findings. Being value free and objective only
study that has a foregone conclusion and auto­ refers to actually conducting the study. This
matically supports a specific value position. means that you can study the issues you believe
They believe that a researcher should make his to be important and after completing a study

you can share the results with specific interest Endnotes

groups in addition to making them available to
the scientific community. 1. For a discussion of research fraud, see Broad and
Wade (1982), Diener and Crandall (1978), and
Weinstein (1979). Hearnshaw (1979) and Wade
(1976) discuss the Cyril Burt case, and see
CONCLUSION Holden (2000) on the social psychologist case.
Kusserow (1989) discusses the concept of scien­
In Chapter 1, we discussed the distinctive con­
tific misconduct.
tribution of science to society and how social re­ 2. See Blum (1989) and D'Antonio (1989)for details
search is a source of knowledge about the social on this case. Also see Goldner (1998) on legal ver­
world. The perspectives and techniques of social sus scientific views of misconduct. Gibelman
research can be powerful tools for understand­ (2001) discusses several cases and the changing
ing the world. Nevertheless, with that power to defmition of misconduct.
discover comes responsibility-a responsibility 3. See Lifton (1986) on Nazi experiments, and
to yourself, a responsibility to your sponsors, a Williams and Wallace (1989) discuss Japanese ex­
responsibility to the community of scientific re­ periments. Harris (2002) argues that the Japanese
searchers, and a responsibility to the larger soci­ experiments were more horrific, but the United
States did not prosecute the Japanese scientists as
ety. These responsibilities can conflict with each
the Germans were because the U.S. military
other. Ultimately, you personally must decide to
wanted the results to develop its own biological
conduct research in an ethical manner, to up­ warfare program.
hold and defend the principles of the social sci­ 4. See Jones (1981) and Mitchell (1997) on "Bad
ence approach you adopt, and to demand ethical Blood."
conduct by others. The truthfulness of knowl­ 5. Diener and Crandall (1978:128) discuss examples.
edge produced by social research and its use or 6. A discussion of physical harm to research partici­
misuse depends on individual researchers like pants can be found in Kelman (1982), Reynolds
you, reflecting on their actions and on the seri­ (1979, 1982), and Warwick (1982).
ous role of social research in society. In the next 7. For a discussion, see Diener and Crandall
chapter, we examine basic design approaches (1978:21-22) and Kidder and Judd (1986:481­
and issues that appear in both qualitative and
8. See Monaghan (1993a, 1993b, 1993c).
quantitative research.
9. Broadhead and Rist (1976) discuss gatekeepers.
10. See "U'W Protects Dissertation Sources," Capital
Times (Madison, Wisconsin), December 19,1994,
Key Terms p.4.
11. See Hirschman (1970) on loyalty, exit, or voice.
anonymity 12. See Edward Fiske, "The Misleading Concept of
confidentiality 'Average' on Reading Test Changes, More Stu­
crossover design dents Fall Below It," New York Times (July 12,
informed consent 1989). Also see Koretz (1988) and Weiss and Gru­
institutional reviewboard (IRB) ber (1987).
13. See "State Sought, Got Author's Changes of Lot­
tery Report," Capital Times (Madison, Wiscon­
principle of voluntary consent
sin), July 28, 1989, p. 21.
public sociology 14. Andrew Revkin, "Bush Aide Edited Climate Re­
research fraud ports," New York Times (June 8, 2005). "White
scientific misconduct House Calls Editing Climate Files Part of Usual
special populations Review," New York Times (June 9,2005). Union of
whistle-blower Concerned Scientists, "Politics Trumps Science at

U.S. Fish and Wildlife Service" (February 9, ber 10, 2004). James Glanz, "Scientists Say Ad­
2005)." Specific Examples ofthe Abuse of Science ministration Distorts Facts," New York Times
www.ucsusa.org/global_environment/rsiJpage.ef (February 19, 2004). Dylan O. Krider, "The Politi­
m?pageID=1398, downloaded August 3, 2005. cization of Science in the Bush Administration,"
"Summary of National Oceanic & Atmospheric Skeptic Vol. 11, Number 2 (2004) at www. Skep­
Administration Fisheries Service Scientist Survey" tic.com. C. Orstein, "Politics Trumps Science in
by Union of Concerned Scientists (June 2005). E. Condom Fact Sheet," New York Times (December
Shogren, "Researchers Accuse Bush ofManipulat­ 27, 2002). "Scientist Says Officials Ignored Advice
ing Science," LosAngeles Times (July 9, 2004). Jef­ on Water Levels," Washington Post (October 29,
frey McCracker, "Government Bans Release of 2002).
Auto-Safety Data," Detroit Free Press (August 19, 15. See Adler and Adler (1993).
2004). Garddiner Harris, "Lawmaker Says FDA 16. See Neuman (2003, Chapter 16) for a discussion
Held Back Drug Data," New York Times (Septem­ of political issues in social research.

In the past three

about the main
Reviewing the Scholarly search, discovered
a study, and erami..
Literature and Planning cial research. You
specifics of how to
Recall from LUo~"IIl:II
a Study begins with a ~
topic down into a
then makes d ..
signing a stud> ~
~. . 1
question. .
Where do topidj
come from man~~
Introduction vision or film.
Literature Review with friends and ­
Where to Find Research Literature about in a book,
How to Conduct a Systematic Literature Review
topic often begins 1
Taking Notes
YO.ill curiosity,
nutments or ~~
Writing the Review is really wrong and ~
What a Good Review Looks Like cial research, a ~
Using the Internet for Social Research terns that opera~.
empirically meas
Qualitative and Quantitative Orientations toward Research
out topics about
Linearand Nonlinear Paths vour boy/girlfriend
Preplanned and Emergent Research Questions your friend's littkl
Qualitative Design Issues teacher), or one ~
family), or somethi
The Language of Cases and Contexts
even indirectly I e.g.j
Grounded Theory
pernatural POWer5. ~
The Context Is Critical interesting topics. b
The Case and Process remain to be inYestiJI
Interpretation How you proces
Quantitative Design Issues on whether you ado!
rive approach. Cd
The Language of Variables and Hypotheses
researcher, those ....
Causal Theory and Hypotheses proach and gather «I
Aspects of Explanation much more time to ~
From the Research Question to Hypotheses tion very precisely all
a study in advance. )
velop the judgments
might be better to •
quantitative or an iD


address a topic and research question. Three

things can help you learn what is the most effec­
In the past three chapters, you have learned tive type of study to pursue for a question:
about the main principles and types of social re­
search, discovered how researchers use theory in 1. Reading studies that others have conducted
a study, and examined the place of ethics in so­ on a topic
cial research. You are now ready to get into the 2. Grasping issues that operate in qualitative
specifics of how to go about designing a study. and quantitative approaches to research
Recall from Chapter 1 that a researcher usually 3. Understanding how to use various research
begins with a general topic, then narrows the techniques as well as their strengths and
topic down into a specific research question, and limitations
then makes decisions about the specifics of de­
This chapter introduces you to the first two
signing a study that will address the research
of these, whereas many of the remaining chap­
ters of the book discuss the third item in the list.
Where do topics for study come from? They
come from many sources: previous studies, tele­
vision or film, personal experiences, discussions
with friends and family, or something you read
about in a book, magazine, or newspaper. A Reading the "literature," or the collection of
topic often begins as something that arouses studies already published on a topic, serves sev­
your curiosity, about which you hold deep com­ eral very important functions. First, it helps you
mitments or strong feelings, or that you believe narrow down a broad topic by showing you how
is really wrong and want to change. To apply so­ others conducted their studies. The studies by
cial research, a topic must be about social pat­ others give you a model of how narrowly fo­
terns that operate in aggregates and be cused a research question should be, what kinds
empirically measurable or observable. This rules of study designs others have used, and how to
out topics about one unique situation (e.g., why measure variables or analyze data. Second, it in­
your boy/girlfriend dumped you yesterday, why forms you about the "state of knowledge" on a
your friend's little sister hates her school topic. From the studies by others, you can learn
teacher), or one individual case (e.g., your own the key ideas, terms, and issues that surround a
family), or something one can never observe, topic. You should consider replicating, testing,
even indirectly (e.g., unicorns, ghosts with su­ or extending what others already found. Third,
pernatural powers, etc.). This may rule out some the literature often stimulates your creativity and
interesting topics, but many tens of thousands curiosity. Last, even if you never get to conduct
remain to be investigated. or publish your own research study, a published
How you proceed differs slightly depending study offers you an example ofwhat the final re­
on whether you adopt an inductive or a deduc­ port on a study looks like, its major parts, its
tive approach. Compared to an inductive form, and its style of writing. Another reason is
researcher, those who choose a deductive ap­ more practical. Just as attentively reading a lot of
proach and gather quantitative data will devote top-quality writing can help you improve your
much more time to specifying the research ques­ own writing skills, reading many reports of
tion very precisely and planning many details of good-quality social research enables you to grasp
a study in advance. It will take you a while to de­ better the elements that go into conducting a re­
velop the judgment skillsfor deciding whether it search study.
might be better to conduct a more deductive­ It is best to be organized and not haphazard
quantitative or an inductive-qualitative study to as you locate and read the scholarly or academic

' CHA"~

literature on a topic and associated research ontele ·on~·

questions. Also, it is wise to plan to prepare a newss
written literature review. There are many spe­ plete reports
cialized types of reviews, but in general a literature .
1. To demonstrate a familiarity with a body of knowl­
literature review is a carefully crafted summary
of the recent studies conducted on a topic that edge and establish credibility. A review tells a summanes~
audience, and
includes key findings and methods researchers reader that the researcher knows the research in needed for a
used while making sure to document the an area and knows the major issues. A good re­ Textbooks and •
sources. For most purposes, you must first lo­ view increases a reader's confidence in the re­ densed snm~
searcher's professional competence, ability, and
cate the relevant studies; next, read thoroughly who are new to.
to discover the major findings, central issues, adequate for ~
and methods of the studies, and take conscien­ 2, To show the path of priorresearch and how a cur­ cause many essei
tious notes on what you read. While the reading rent project is linked to it. A review outlines the di­ absent. i
is still fresh in your mind and with the notes in rection of research on a question and shows the It is easv for~
development of knowledge. A good review
front of you, you need to organize what you have ature revi~ to I
places a research project in a context and
learned and write clearly about the studies in a types of periodid
demonstrates its relevance by making connec­
way that builds a context around a specific re­ to distinguish ul
tions to a body of knowledge.
search question that is of interest to you. persandma~
A literature review is based on the assump­ 3. To integrate and summarize what is known in an lic, (2) pop~
area. A review pulls together and synthesizes
tion that knowledge accumulates and that peo­ opinionm~
different results. A good review points out areas
and express ~
ple learn from and build on what others have
done. Scientific research is a collective effort of
many researchers who share their results with
where prior studies agree, where they disagree,
and where major questions remain. It collects
mic journals in.
findings of studiIj
what is known up to a point in time and indicates
one another and who pursue knowledge as a the direction for future research.
cation to the sd
community. Although some studies may be es­ viewed empirial
4. To learn from others and stimulate new ideas. A re­
pecially important and individual researchers complete form ~
view tells what others have found so that a re­
may become famous, a specific research project tion, although ali
searcher can benefit from the efforts of others.
is just a tiny part of the overall process of creat­ sionally talk abod
A good review identifies blind alleys and sug­
ing knowledge. Today's studies build on those of Mass markd
gests hypotheses for replication. It divulges pro­
yesterday. Researchers read studies to compare, cedures, techniques, and research designs worth
Time, Newsweek; j
replicate, or criticize them for weaknesses. copying so that a researcher can better focus can Spectator, aDli
Reviews vary in scope and depth. Different hypotheses and gain new insights. newsstands and d
kinds of reviews are stronger at fulfilling one or public with n~'SI
another offour goals (see Box 4.1). It may take a A researcher mil
researcher over a year to complete an extensive source on currenl
professional summary review of all the literature vide full reports •
on a broad question. The same researcher might books, dissertations, government documents, or needed to prepan
complete a highlyfocusedreviewin a veryspecial­ policy reports. They also present them as papers Popularized I
ized area in a fewweeks. When beginning a review, at the meetings of professional societies, but for professional pol
a researcher decides on a topic, how much depth the most part, you can find them only in a col­ Psychology Todav;
to go into, and the kind of reviewto conduct. legeor university library. This section briefly dis­ Their purpose is 1
cusses each type and gives you a simple road cated lay public a
map on how to accessthem. or a cornmentan
Where to Find Research Literature
original research
Researchers present reports of their research Periodicals. You can find the results of social social science ill
projects in several written forms: periodicals, research in newspapers, in popular magazines, other sources in a

on television or radio broadcasts, and in Internet It is harder to recognize serious opinion

news summaries, but these are not the full, com­ magazines (e.g., American Prospect, Commen­
plete reports of research required to prepare a tary, Dissent, and Public Interest). Larger book­
literature review. They are selected, condensed stores in major cities sell them. Leading scholars
summaries prepared by journalists for a general often write articles for opinion magazines about
audience, and they lack many essential details topics on which they may also conduct empirical
needed for a serious evaluation of the study. research (e.g., welfare reform, prison expansion,
Textbooks and encyclopedias also present con­ voter turnout). They differ in purpose, look, and
densed summaries as introductions to readers scope from scholarly journals of social science
who are new to a topic, but, again, these are in­ research findings. The publications are an arena
adequate for preparing a literature review be­ where intellectuals debate current issues, not
cause many essential details about the study are where researchers present findings of their stud­
absent. ies to the broader scientific community.
It is easy for someone preparing a first liter­
ature review to be confused about the many Scholarly Journals. The primary type of period­
types of periodicals. With skill, you will be able ical to use for a literature review is the scholarly
to distinguish among (1) mass market newspa­ journal filled with peer-reviewed reports of re­
pers and magazines written for the general pub­ search (e.g., American SociologicalReview, Social
lic, (2) popularized social science magazines, (3) Problems, American Journal of Sociology, Crimi­
opinion magazines in which intellectuals debate nology, and Social Science Quarterly). One rarely
and express their views, and (4) scholarlyacade­ finds them outside of college and university li­
mic journals in which researchers present the braries. Recall from Chapter 1 that researchers
findings of studies or provide other communi­ disseminate findings of new studies in scholarly
cation to the scientific community. Peer-re­ journals.
viewed empirical research findings appear in a Some scholarly journals are specialized. In­
complete form only in the last type of publica­ stead of reports of research studies, they have
tion, although articles in the other types occa­ only book reviewsthat provide commentary and
sionally talk about findings published elsewhere. evaluations on a book (e.g., Contemporary Soci­
Mass market publications (e.g., McCleans, ology), or they contain only literature review es­
Time, Newsweek, Economist, The Nation, Ameri­ says (e.g., Annual Review of Sociology, Annual
can Spectator, and Atlantic Monthly) are sold at Review ofPsychology, and Annual Review ofAn­
newsstands and designed to provide the general thropology) in which researchers give a "state of
public with news, opinion, and entertainment. the field" essayfor others. Publications that spe­
A researcher might occasionally use them as a cialize in literature reviews can be helpful if an
source on current events, but they do not pro­ article was recently published on a specific topic
vide full reports of research studies in the form of interest. Many other scholarly journals have a
needed to prepare a literature review. mix of articles that are literature reviews, books
Popularized social science magazines and reviews, reports on research studies, and theo­
professional publications (e.g., Society and retical essays.
Psychology Today) are sometimes peer reviewed. No simple solution or "seal of approval"
Their purpose is to provide the interested, edu­ distinguishes scholarly journals, the kind of pub­
cated lay public a simplified version of findings lications on which to build a serious literature
or a commentary, but not to be an outlet for review from other periodicals, or instantly dis­
original research findings. At best, popularized tinguishes the report on a research study from
social science magazines can supplement to other types of articles. One needs to develop
other sources in a literature review. judgment or ask experienced researchers or pro­

fessionallibrarians. Nonetheless, distinguishing and only then if a library pays for a special on­ Scholarly join
among types of publications is essential to build line subscription service. once a year or 35 i
on a body of research. One of the best ways to Once you locate a scholarly journal that re­ pear four to six
learn to distinguish among types of publications ports on social science research studies, you need Sociological Ouars
is to read many articles in scholarly journals. to make sure that a particular article presents the To assist in loca
The number of journals varies by field. Psy­ results of a study, since the journal may have scholars have des
chology has over 400 journals, whereas sociol­ other types of articles. It is easier to identify scholarly journals
ogy has about 250 scholarly journals, political quantitative studies because they usually have a issue is assigned a
science and communication have slightly fewer methods or data section and charts, statistical sue number. This
than sociology, anthropology-archaeology and formulas, and tables of numbers. Qualitative re­ locate an article. S
social work have about 100, urban studies and search articles are more difficult to identify, and details such as aut!
women studies have about 50, and there are many students confuse them with theoretical es­ is called an article
about a dozen journals in criminology. Each says, literature review articles, idea-discussion ographies. When
publishes from a few dozen to over 100 articles a essays, policy recommendations, book reviews, begins with volUIJI
year. and legal case analyses. To distinguish among increasing the D1
Many, but not all, scholarly journals may be these types requires a good grasp of the varieties most journals faa
viewed via the Internet. Usually, this is limited to of research as well as experience in reading many enough exception
selected years and to libraries that paid special articles. tention to citation
subscription fees. A few Internet services provide Your college library has a section for schol­ nals, each volume!
full, exact copies of scholarly journal articles arly journals and magazines, or, in some cases, issue with volumi
over the Internet. For example, JSTOR provides they may be mixed with books. Look at a map of means that the .iOll
exact copies, but only for a small number of library facilities or ask a librarian to find this sec­ 52 years. Most, bI
scholarly journals and only for past years. Other tion. The most recent issues, which look like thin publishing cycle it
Internet services, such as EBSCO HOST, offer a paperbacks or thick magazines, are often physi­ Most journals
full-text version of recent articles for a limited cally separate in a "current periodicals" section. by issue. The first
number of scholarly journals, but they are not in This is done to store them temporarily and make gins with page 1, OIl
the same format as a print version of an article. them available until the library receives all the is­ throughout the ell
This can make it impossible to find a specific sues of a volume. Most often, libraries bind all is­ first page of volun
page number or see an exact copy of a chart. It is sues of a volume together as a book before Most journals hal
best to visit the library and see what a full-print adding them to their permanent collections. and a table of cons
version of the scholarly article looks like. An Scholarly journals from many different title, the author's
added benefit is that it makes it easy for you to fields are placed together with popular maga­ page on which till
browse the Table of Contents of the journals. zines. All are periodicals, or serials in the jargon as few as 1 or 2 ..
Browsing can be very useful for generating new of librarians. Thus, you will find popular maga­ have 8 to 18 artidl
ideas for research topics, seeing an established zines (e.g., Time, Road and Track, Cosmopolitan, long. The article
topic in creative ways, or learning how to expand and Atlantic Monthly) next to journals for as­ summaries on til
an idea into new areas. Only a tiny handful of tronomy, chemistry, mathematics, literature, grouped together!
new Internet-only scholarly journals, called e­ and philosophy as well as sociology, psychology, Many librarie
journals, present peer-reviewed research studies social work, and education. Some fields have copies of older jOil
(e.g., Sociological Research Online, Current Re­ more scholarly journals than others. The "pure" they retain only.
search in Social Psychology, andlournal of World academic fields usually have more than the "ap­ hundreds of schol
Systems Research). Eventually, the Internet for­ plied" or practical fields such as marketing or so­ fields, with each ~
mat may replace print versions. But for now, 99 cial work. The journals are listed by title in a card Only the large res
percent of scholarly journals are available in catalog or a computerized catalog system. Li­ of them. You IruIII
print form and about one-third of these are also braries can provide you with a list of the period­ photocopy of ~
available in a full-text version over the Internet icals to which they subscribe. through an interl

Scholarly journals are published as rarely as which libraries lend books or materials to other
once a year or as frequently as weekly. Most ap­ libraries. Few libraries allow people to check out
pear four to six times a year. For example, recent issues of scholarly journals. You should
Sociological Quarterly appears four times a year. plan to use these in the library. Some, not all,
To assist in locating articles, librarians and scholarly journals are available via the Internet.
scholars have developed a system for tracking Once you find the periodicals section, wan­
scholarly journals and the articles in them. Each der down the aisles and skim what is on the
issue is assigned a date, volume number, and is­ shelves. You will see volumes containing many
sue number. This information makes it easier to research reports. Each title of a scholarly journal
locate an article. Such information-along with has a call number like that of a regular library
details such as author, title, and page number­ book. Libraries often arrange them alphabeti­
is called an article's citation and is used in bibli­ cally by title. Because journals change titles, it
ographies. When a journal is first published, it may create confusion if the journal is shelved
begins with volume 1, number 1, and continues under its original title.
increasing the numbers thereafter. Although
most journals follow a similar system, there are Citation Formats. An article's citation is the
enough exceptions that you have to pay close at­ key to locating it. Suppose you want to read the
tention to citation information. For most jour­ study by Weitzer and Tuch (2005) on percep­
nals, each volume is one year. Ifyou see a journal tions of police misconduct discussed in Chapter
issue with volume 52, for example, it probably 2. Its citation is as follows:
means that the journal has been in existence for
52 years. Most, but not all, journals begin their Weitzer, Ronald, and Steven Tuch. 2005.
publishing cycle in January. "Racially Biased Policing: Determinants of
Most journals number pages by volume, not Citizen Perceptions." Social Forces
by issue. The first issue of a volume usually be­ 83:1009-1030.
gins with page 1, and page numbering continues
throughout the entire volume. For example, the This tells you that you can find the article in
first page ofvolume 52, issue 4, may be page 547. an issue of Social Forces published in 2005. The
Most journals have an index for each volume citation does not provide the issue or month, but
and a table ofcontents for each issue that lists the it gives the volume number, 83, and the page
title, the author's or authors' names, and the numbers, 1009 to 1030.
page on which the article begins. Issues contain There are many ways to cite the literature.
as few as 1 or 2 articles or as many as 50. Most Formats for citing literature in the text itself
have 8 to 18 articles, which may be 5 to 50 pages vary, with the internal citation format of using
long. The articles often have abstracts, short an author's last name and date of publication in
summaries on the first page of the article or parentheses being very popular. The full citation
grouped together at the beginning of the issue. appears in a separate bibliography or reference
Many libraries do not retain physical, paper section. There are many styles for full citations of
copies of older journals. To save space and costs, journal articles, with books and other types of
they retain only microfilm versions. There are works each having a separate style. When citing
hundreds ofscholarly journals in most academic articles, it is best to check with an instructor,
fields, with each costing $50 to $2,500 per year. journal, or other outlet for the desired format.
Only the large research libraries subscribe to all Almost all include the names of authors, article
of them. You may have to borrow a journal or title, journal name, and volume and page num­
photocopy of an article from a distant library bers. Beyond these basic elements, there is great
through an interlibrary loan service, a system by variety. Some include the authors' first names,

others use initials only. Some include all authors, scriptions and complex theoretical or philo­
others give only the first one. Some include in­ sophical discussions usually appear as books. Fi­ FIGURE 4.1
formation on the issue or month of publication, nally, an author who wants to communicate to
The oldest jouma/4
others do not (see Figure 4.1). scholarly peers and to the educated public may ginity pledges by_
Citation formats can get complex. Two ma­ write a book that bridges the scholarly, academic 2001 issue (nu~
jor reference tools on the topic in social science style and a popular nonfiction style. journal's 106th ~
are Chicago Manual of Style, which has nearly 80 Locating original research articles in books logical Review (ASRJj
pages on bibliographies and reference formats, can be difficult because there is no single source 1
and American Psychological Association Publica­ listing them. Three types of books contain col­ ASRStyle
tion Manual, which devotes about 60 pages to lections of articles or research reports. The first is Bearman, Peter ~
the topic. In sociology, the American Sociological designed for teaching purposes. Such books, Americanj~
Review style, with 2 pages of style instructions, is called readers, may include original research re­
widely followed. ports. Usually, articles on a topic from scholarly APA Style
journals are gathered and edited to be easier for Bearman, P., and Bit
Books. Books communicate many types of in­ nonspecialists to read and understand. journalof~
formation, provoke thought, and entertain. The second type of collection is designed for
Other Styles 1
There are many types of books: picture books, scholars and may gather journal articles or may
textbooks, short story books, novels, popular fic­ contain original research or theoretical essayson Bearman, P., and H.~
tion or nonfiction, religious books, children's a specific topic. Some collections contain articles ofSociology 1
books, and others. Our concern here is with from journals that are difficult to locate. They Bearman, Peter and
those books containing reports of original re­ may include original research reports organized "Promising t:he~
search or collections of research articles. Li­ around a specialized topic. The table of contents Bearman, P. and BUll
braries shelve these books and assign call lists the titles and authors. Libraries shelve these can joumal of~
Bearman, Peter and
numbers to them, as they do with other types of collections with other books, and some library
"Promising 1:he.
books. You can find citation information on catalog systems include them.
them (e.g., title, author, publisher) in the li­ Citations or references to books are shorter
Bearman, P. and H. III
brary's catalog system. than article citations. They include the author's
It is not easy to distinguish a book that re­ name, book title, year and place of publication, Peter Bearmanand ~
ports on research from other books. You are and publisher's name. joumalof~
more likely to find such books in a college or
university library. Some publishers, such as uni­ Dissertations. All graduate students who re­
versity presses, specialize in publishing them. ceive the Ph.D. degree are required to complete
Nevertheless, there is no guaranteed method for a work of original research, which they write up example, Dis~
identifying one without reading it. as a dissertation thesis. The dissertation is bound dissertations wnh ~
Some types of social research are more likely and shelved in the library of the university that versities. This ~
to appear in book form than others. For exam­ granted the Ph.D. About half of all dissertations contains an al>str.iIl
ple, studies by anthropologists and historians are are eventually published as books or articles. Be­ can borrow IIlO5t ~
more likely to appear in book-length reports cause dissertations report on original research, loan from the ~
than are those of economists or psychologists. they can be valuable sources of information. university permits •
Yet, some anthropological and historical studies Some students who receive the master's degree
are articles, and some economic and psycholog­ conduct original research and write a master's Governmmt~
ical studies appear as books. In education, social thesis, but fewer master's theses involve serious ment of the c~
work, sociology, and political science, the results research, and they are much more difficult to lo­ other nations, state,
oflong, complex studies may appear both in two cate than unpublished dissertations. ments, the Lnited I
or three articles and in book form. Studies that Specialized indexes list dissertations com­ tional agencies SUI
involve detailed clinical or ethnographic de­ pleted by students at accredited universities. For sponsor studies _


FIG U RE 4. 1 Different Reference Citations for a Journal Article

The oldest journal of sociology in the United States, American Journal of Sociology, reports on a study of vir­
ginity pledges by Peter Bearman and Hannah BUckner. It appeared on pages 859 to 91 3 of the January
2001 issue (number 4) of the journal, which begins counting issues in March. It was in volume 106, or the
journal's 106th year. Here are ways to cite the article. Two very popular styles are those of American Socio­
logical Review (ASR) and American Psychological Association (APA).

Bearman, Peter and Hannah BUckner. 2001. "Promising the Future: Virginity Pledges and First Intercourse."
American Journal ofSaciology 106:859-912.

Bearman, P., and BUckner, H. (2001). Promising the future: Virginity pledges and first intercourse. American
Journal of Sociology 106, 859-912.

Other Styles
Bearman, P., and H. Buckner. "Promising the Future: Virginity Pledges and First Intercourse," American Journal
of Sociology 106 (2001 ), 859-912.
Bearman, Peter and Hannah Buckner, 2001.
"Promising the future: Virginity pledges and first Intercourse." Am.J. af Sociol. 106:859- 91 2.
Bearman, P. and Buckner, H. (2001). "Promising the Future: Virginity Pledges and First Intercourse." Ameri­
canJournal of Sociology 106 (January): 859-912.
Bearman, Peter and Hannah Buckner. 2001 .
"Promising the future: Virginity pledges and first Intercourse." American Journal of Sociology 106
Bearman,P.and H. BUckner. (2001 ). "Promising the future: Virginity pledges and first intercourse." American
Journal of Sociology 106,859-912.
Peter Bearman and Hannah BUckner, "Promising the Future: Virginity Pledges and First Intercourse," American
Journal of Sociology 106, no. 4 (2001): 859-912.

example, Dissertation Abstracts Internationallists search. Many college and university libraries
dissertations with their authors, titles, and uni­ have these documents in their holdings, usually
versities. This index is organized by topic and in a special "government documents" section.
contains an abstract of each dissertation. You These reports are rarely found in the catalog sys­
can borrow most dissertations via interlibrary tem. You must use specialized lists of publica­
loan from the degree-granting university if the tions and indexes, usually with the help of a
university permits this. librarian, to locate these reports. Most college
and university libraries hold only the most fre­
Government Documents. The federal govern­ quently requested documents and reports.
ment of the United States, the governments of
other nations, state- or provincial-level govern­ Policy Reports and Presented Papers. A re­
ments, the United Nations, and other interna­ searcher conducting a thorough review ofthe lit­
tional agencies such as the World Bank, all erature will examine these two sources, which
sponsor studies and publish reports of the re­ are difficult for all but the trained specialist to

obtain. Research institutes and policy centers Design a Search. After choosing a focused re­
(e.g., Brookings Institute, Institute for Research search question for the review, the next step is to able Yia con. .'"
search process,
on Poverty, Rand Corporation, etc.) publish pa­ plan a search strategy. The reviewer needs to de­ .-\hsb:a:n or ­
pers and reports. Some major research libraries cide on the type of review, its extensiveness, and . . . bamjl-al. . . .
purchase these and shelve them with books. The the types of materials to include. The key is to be .IIowa~.
only way to be sure ofwhat has been published is
to write directly to the institute or center and re­
careful, systematic, and organized. Set parame­
ters on your search: how much time you will de­
- .- 1"­
quest a list of reports. vote to it, how far back in time you will look, the -~~-.
Each year, the professional associations in
academic fields (e.g., sociology, political science,
minimum number of research reports you will
examine, how many libraries you will visit, and
* -iiwjew
Sa - f
; Cz::srI
psychology) hold annual meetings. Thousands so forth.
of researchers assemble to give, listen to, or dis­ Also, decide how to record the bibliographic
cuss oral reports of recent research. Most of citation for each reference you find and how to ,
these oral reports are available as written papers
to those attending the meeting. People who do
not attend the meetings but who are members of
take notes (e.g., in a notebook, on 3 X 5 cards, in
a computer file). Develop a schedule, because
several visits are usually necessary. You should --..-h ."5.

the association receive a program of the meeting, begin a file folder or computer file in which you II~--.I
listing each paper to be presented with its title, can place possible sources and ideas for new
..* ........

author, and author's place of employment. They

can write directly to the author and request a
copy of the paper. Many, but not all, of the pa­
sources. As the review proceeds, it should be­
come more focused. _I..........

pers are later published as articles. The papers Locate Research Reports. Locating research
may be listed in indexes or abstract services (to reports depends on the type of report or "outlet"
be discussed). of research being searched. As a general rule, use
multiple search strategies in order to counteract
the limitations of a single search method.
How to Conduct a Systematic
Literature Review
Articles in Scholarly Journals. As discussed ear­
Define and Refine a Topic. Just as a researcher lier, most social research is published in schol­
must plan and clearly define a topic and research arly journals. There are dozens ofjournals, many
question when beginning a research project, you going back decades, each containing many arti­
need to begin a literature review with a clearly cles. The task of searching for articles can be for­
defined, well-focused research question and a midable. Luckily, specialized publications make
plan. A good review topic should be as focused the task easier.
as a research question. For example, "divorce" You may have used an index for general
or "crime" is much too broad. A more appro­ publications, such as Reader's Guide to Periodical
priate review topic might be "the stability of Literature. Many academic fields have "ab­
families with stepchildren" or "economic in­ stracts" or "indexes" for the scholarly literature
equality and crime rates across nations." If you (e.g., Psychological Abstracts, Social Sciences In­ ,
conduct a context review for a research project, dex, Sociological Abstracts, and Gerontological
• »
. . c.- ..

it should be slightly broader than the specific re­ Abstracts). For education-related topics, the Ed­
search question being tested. Often, a researcher ucational Resources Information Center (ERIC)
will not finalize a specific research question for a system is especially valuable. There are over 100 . . . . . . ._ ...._1
study until he or she has reviewed the literature. such publications. You can usually find them in
The review helps bring greater focus to the re­ the reference section of a library. Many ab­ ~_-.I
...a .....1III1
search question. stracts or index services as well as ERICare avail- ..· . .11

able via computer access, which speeds the most computer-based searches and consider
search process. several synonyms. The computer's searching
Abstracts or indexes are published on a reg­ method can vary and most only look for a key­
ular basis (monthly, six times a year, etc.) and word in a title or abstract. If you choose too few
allow a reader to look up articles by author name words or very narrow terms, you will miss a lot
or subject. The journals covered by the abstract of relevant articles. If you choose too many
or index are listed in it, often in the front. An in­ words or very broad terms, you will get a huge
dex, such as the Social Sciences Index, lists only number of irrelevant articles. The best way to
the citation, whereas an abstract, such as learn the appropriate breadth and number of
Sociological Abstracts, lists the citation and has a keywords is by trial and error.
copy of the article's abstract. Abstracts do not In a study I conducted on how college stu­
give you all the findings and details of a research dents define sexualharassment(Neuman, 1992),
project. Researchers use abstracts to screen arti­ I used the following keywords: sexual harass­
cles for relevance, then locate the more relevant ment, sexual assault, harassment, gender equity,
articles. Abstracts may also include papers pre­ gender fairness, and sexdiscrimination. I later dis­
sented at professional meetings. covered a few important studies that lacked any
It may sound as if all you have to do is to go of these keywords in their titles. I also tried the
find the index in the reference section of the li­ keywords college student and rape, but got huge
brary or on the Internet and look up a topic. Un­ numbers of unrelated articles that I could not
fortunately, things are more complicated than even skim.
that. In order to cover the studies across many There are numerous computer-assisted
years, you may have to look through many issues search databases or systems. A person with a
of the abstracts or indexes. Also, the subjects or computer and an Internet hook-up can search
topics listed are broad. The specificresearch ques­ some article index collections, the catalogs ofli­
tion that interests you may fit into several subject braries, and other information sources around
areas. You should check each one. For example, the globe if they are available on the Internet.
for the topic of illegal drugs in high schools, you All computerized searching methods share a
might look up these subjects: drug addiction, similar logic, but each has its own method of op­
drug abuse, substance abuse, drug laws, illegal eration to learn. In my study, I looked for
drugs, high schools, and secondary schools. Many sources in the previous seven years and used five
of the articles under a subject area will not be rel­ computerized databases of scholarly literature:
evant for your literature review. Also, there is a 3­ Social Science Index, CARL (Colorado Area Re­
to 12-month time lag between the publication of search Library), Sociofile, Social Science Citation
an article and its appearance in the abstracts or Index, and PsychLit.
indexes. Unless you are at a major research li­ Often, the same articles will appear in mul­
brary, the most useful article may not be available tiple scholarly literature databases, but each
in your library. You can obtain it only by using an database may identify a few new articles not
interlibrary loan service, or it may be in a foreign found in the others. For example, I discovered
language that you do not read. several excellent sources not listed in any of the
The computerized literature search works computerized databases that had been published
on the same principle as an abstract or an index. in earlier years by studying the bibliographies of
Researchers organize computerized searches in the relevant articles.
several ways-by author, by article title, by sub­ The process in my study was fairly typical.
ject, or by keyword. A keyword is an important Based on my keyword search, I quickly skimmed
term for a topic that is likely to be found in a ti­ or scanned the titles or abstracts of over 200
tle. You will want to use six to eight keywords in sources. From these, I selected about 80 articles,

reports, and books to read. I found about 49 of though you may not use some and later erase
the 80 sources valuable, and they appear in the them. Do not forget anything in a complete bib­
bibliography of the published article. liographic citation, such as a page number or the
name of the second author; you will regret it
Scholarly Books. Finding scholarly books on a later. It is far easier to erase a source you do not
subject can be difficult. The subject topics of li­ use than to try to locate bibliographic informa­
brary catalog systems are usually incomplete and tion later for a source you discover that you need
too broad to be useful. Moreover, they list only or from which you forgot one detail.
books that are in a particular library system, al­ I recommend creating two kinds of Source
though you may be able to search other libraries Files, or divide a master file into two parts: Have
for interlibrary loan books. Libraries organize File and Potential File. The Have File is for
books by call numbers based on subject matter. sources that you have found and for which you
Again, the subject matter classifications may not have already taken content notes. The Potential
reflect the subjects of interest to you or all the File is for leads and possible new sources that
subjects discussed in a book. Once you learn the you have yet to track down or read. You can add
system for your library, you will find that most to the Potential File anytime you come across a
books on a topic will share the main parts of the new source or in the bibliography of something
call number. In addition, librarians can help you you read. Toward the end ofwriting a report, the
locate books from other libraries. For example, Potential File will disappear while the Have File
the Library of Congress National Union Catalog will become your bibliography.
lists all books in the U.S. Library of Congress. Li­ Your note cards or computer documents go
brarians have access to sources that list books at into the Content File. This file contains substan­
other libraries, or you can use the Internet. tive information of interest from a source, usu­
There is no sure-fire way to locate relevant ally its major findings, details of methodology,
books. Use multiple search methods, including a definitions of concepts, or interesting quotes. If
look at journals that have book reviews and the you directly quote from a source or want to take
bibliographies of articles. some specific information from a source, you
need to record the specific page number(s) on
which the quote appears. Link the files by
Taking Notes
putting key source information, such as author
As you gather the relevant research literature, it is and date, on each content file.
easy to feel overwhelmed by the quantity of in­
formation, so you need a system for taking notes. What to Record. You will find it much easier Photoc~
The old-fashioned approach is to write notes to take all notes on the same type and size of pa­ will save you ~
onto index cards. You then shift and sort the note per or card, rather than having some notes on sure that you 1'~
cards, place them in piles, and so forth as you sheets of papers, others on cards, and so on. Re­ you can make ~
look for connections among them or develop an searchers have to decide what to record about an several warnings •
outline for a report or paper. This method still article, book, or other source. It is better to err in tocopying can be c
works. Today, however, most people use word­ the direction of recording too much rather than search. Second, be
processing software and gather photocopies or too little. In general, record the hypotheses laws. U.S. cop~
printed versions of many articles. tested, how major concepts were measured, the for personal resea
As you discover sources, it is a good idea to main findings, the basic design of the research, record or photooq
create two kinds of files for your note cards or the group or sample used, and ideas for future all citation inform
computer documents: a Source File and a study (see Box 4.2). It is wise to examine the re­ tire articles can be ~
Content File. Record all the bibliographic infor­ port's bibliography and note sources that you eral different par1I
mation for each source in the Source File, even can add to your search. used. Finally, unk

take good notes, you may have to reread the en­

tire article later.
Organize Notes. After gathering a large num­
1. Read with a clear purpose or goal in mind. Are ber of references and notes, you need an orga­
you reading for basic knowledge or to apply it to nizing scheme. One approach is to group
a specific question? studies or specific findings by skimming notes
and creating a mental map of how they fit to­
2. Skim the article before reading it all. What can
you learn from the title, abstract, summary and
gether. Try several organizing schemes before
conclusions, and headings? What are the topic, settling on a final one. Organizing is a skill that
major findings, method, and main conclusion? improves with practice. For example, place
notes into piles representing common themes,
3. Consider your own orientation. What is your
or draw charts comparing what different re­
bias toward the topic, the method, the publica­
tion source, and so on, that may color your
ports state about the same question, noting
reading? agreements and disagreements.
In the process of organizing notes, you will
4. Marshal external knowledge. What do you al­
find that some references and notes do not fit
ready know about the topic and the methods
and should be discarded as irrelevant. Also, you
used? How credible is the publication source?
may discover gaps or areas and topics that are
5. Evaluate as you read the article. What errors are
relevant but that you did not examine. This ne­
present? Do findings follow the data? Is the ar­
cessitates return visits to the library.
ticle consistent with assumptions of the ap­
There are many organizing schemes. The
proach it takes?
best one depends on the purpose of the review.
6. Summarize information as an abstract with the Usually, it is best to organize reports around a
topic, the methods used, and the findings. As­
specific research question or around core com­
sess the factual accuracy of findings and cite
mon findings of a field and the main hypotheses
questions about the article.
Source: Adapted from Katzer, Cook, and Crouch (1991;
199-207). Writing the Review
A literature review requires planning and good,
clear writing, which requires a lot of rewriting.
This step is often merged with organizing notes.
Photocopying all relevant articles or reports All the rules of good writing (e.g., clear organi­
will save you time recording notes and will en­ zational structure, an introduction and conclu­
sure that you will have an entire report. Also, sion, transitions between sections, etc.) apply to
you can make notes on the photocopy. There are writing a literature review. Keep your purposes
several warnings about this practice. First, pho­ in mind when you write, and communicate
tocopying can be expensive for a large literature clearly and effectively.
search. Second, be aware of and obey copyright To prepare a good review, read articles and
laws. U.S. copyright laws permit photocopying other literature critically. Recall that skepticism
for personal research use. Third, remember to is a norm of science. It means that you should
record or photocopy the entire article, including not accept what is written simply on the basis of
all citation information. Fourth, organizing en­ the authority of its having been published. Ques­
tire articles can be cumbersome, especiallyif sev­ tion what you read, and evaluate it. The first
eral different parts of a single article are being hurdle to overcome is thinking something must
used. Finally, unless you highlight carefully or be perfect just because it has been published.

Critically reading research reports requires sources. The Internet continues to expand and
skills that take time and practice to develop. De­ change at an explosive rate. FIGURE 4.2
spite a peer-review procedure and high rejection The Internet has been a mixed blessing for
rates, errors and sloppy logic slip in. Read care­ social research, but it has not proved to be the
fully to see whether the introduction and title re­ panacea that some people first thought it might Bearman.~_
ally fit with the rest of the article. Sometimes, be. It provides new and important ways to find and Fnt
titles, abstracts, or the introduction are mislead­ information, but it remains one tool among oth­ 00.4).
ing. They may not fully explain the research pro­ ers. It can quickly make some specific pieces of
ject's method and results. An article should be information accessible. For example, from my
logically tight, and all the parts should fit to­ home computer, I was able to go to the U.S. Fed­
gether. Strong logical links should exist between eral Bureau of Prisons and in less than three
parts of the argument. Weak articles make leaps minutes locate a table showing me that in 1980,
in logic or omit transitional steps. Likewise, arti­ 139 people per 100,000 were incarcerated in the
cles do not always make their theory or ap­ United States, whereas in 2004 (the most recent
proach to research explicit. Be prepared to read data available), it was 486 per 100,000. The In­
the article more than once. (See Figure 4.2 on ternet is best thought of as a supplement rather Since 1993...
taking notes on an article.) than as a replacement for traditional library re­
search. There are "up" and "down" sides to us­

miIion tJ!IeI!5
fected the -
ing the Internet for social research:
What a Good Review Looks Like rlOIlJ*c Iti it
An author should communicate a review's pur­ The UpSide p6edge~"
pose to the reader by its organization. The wrong
1. The Internet is easy, fast, and cheap. It is
way to write a review is to list a series of research ~sj.
widely accessible and can be used from many lo­
reports with a summary of the findings of each.
cations. This near-free resource allows people to
This fails to communicate a sense of purpose. It
find source material from almost anywhere-lo­
reads as a set of notes strung together. Perhaps
cal public libraries, homes, labs or classrooms,
the reviewer got sloppy and skipped over the im­
or anywhere a computer is connected to the In­
portant organizing step in writing the review.
ternet system. Also, the Internet does not close;
The right way to write a review is to organize
it operates 24 hours a day, seven days a week.
common findings or arguments together. A well­
With minimal training, most people can quicldy
accepted approach is to address the most imp or- .
perform searches and get information on their
tant ideas first, to logically link statements or
computer screens that would have required
findings, and to note discrepancies or weaknesses
them to take a major trip to large research li­
in the research (see Box 4.3 for an example).
braries a few years ago. Searching a vast quantity
of information electronically has always been
easier and faster than a manual search, and the
USING THE INTERNET FOR Internet greatly expands the amount and variety
SOCIAL RESEARCH of source material. More and more information
(e.g., Statistical Abstract of the United States) is
The Internet (see Box 4.4) has revolutionized
available on the Internet. In addition, once the
how social researchers work. A mere decade ago,
information is located, a researcher can often
it was rarely used; today, most social researchers
store it electronically or print it at a local site.
use the Internet regularly to help them review
the literature, to communicate with other re­ 2. The Internet has "links" that provide ad­
searchers, and to search for other information ditional ways to find and connect to many other

FIG U R E 4.2 Example of Notes on an Article

I Bearman, Peter, and Hannah Buckner. 2001. "Promising the Future: Virginity Pledges
and First Intercourse." American Journal of Sociology 106:859-91 2. (January, issue
no. 4).


Bearman and Buckner 2001 Topics: Teen pregnancy & sexuality,

pledges/promises, virginity, first sexual
intercourse, S. Baptists, identity movement

Since 1993, the Southern Baptist Church sponsored a movement among teens
whereby the teens make a public pledge to remain virgins until marriage. Over 2.5
million teens have made the pledge. This study examines whether the pledge af­
fected the timing of sexual intercourse and whether pledging teens differ from
non pledging teens. Critics of the movement are uncomfortable with it because
pledge supporters often reject sex education, hold an overly romanticized view of
marriage, and adhere to traditional gender roles.

Adolescents will engage in behavior that adults enjoy but that is forbidden to
them based on the amount of social controls that constrain opportunities to en­
gage in forbidden behavior. Teens in nontraditional families with greater freedom
and less supervision are more likely to engage in forbidden behavior (sex). Teens
in traditional families and who are closer to their parents will delay sexual activ­
ity. Teens closely tied to "identity movements" outside the family will modify be­
, havior based on norms the movements teach.

Data are from a national health survey of U.S. teens in grades 7-12 who were in
public or private schools in 1994-1995. A total of 90,000 students in 141
schools completed questionnaires. A second questionnaire was completed by
20,000 of the 90,000 students. The questionnaire asked about a pledge, im­
portance of religion, and sexual activity.

The study found a substantial delay in the timing of first intercourse among
pledgers. Yet, the effect of pledging varies by the age of the teen. In addition,
pledging only works in some social contexts (i.e., where it is at least partially a so­
cial norm). Pledgers tend to be more religious, less developed physically, and from
more traditional social and family backgrounds.

Example of Bad Review undergraduates at a medium-sized university in

Sexual harassment has many consequences. Adams, groups of 15 to 25. They found disagreement and
Kottke, and Padgitt (1983) found that some women confusion among students. ..... _ s .......
students said they avoided taking a class or working
Example of Better Review
with certain professors because of the risk of harass­
ment. They also found that men and women students The victims of sexual harassment suffer a range of
consequences, from lowered self-esteem and loss of ~i5c11i1.-..il1
reacted differently. Their research was a survey of
1,000 men and women graduate and undergraduate self-confidence to withdrawal from social interaction,
changed career goals, and depression (Adams, Kot­
The .........

students. Benson and Thomson's study in Social
Problems (1982) lists many problems created by tke, and Padgitt, 1983; Benson and Thomson, ...ldMde.lBt
sexual harassment. In their excellent book, The Lech­ 1982; Dziech and Weiner, 1990). For example,
erous Professor, Dziech and Weiner (1990) give a Adams, Kottke, and Padgitt (1983) noted that 13
long list of difficulties that victims have suffered. percent of women students said they avoided taking
Researchers study the topic in different ways. a class or working with certain professors because of
Hunter and McClelland (1 991) conducted a study the risk of harassment.
of undergraduates at a small liberal arts college. They Research into campus sexual harassment has
had a sample of 300 students and students were taken several approaches. In addition to survey re­
given multiple vignettes that varied by the reaction of search, many have experimented with vignettes or
the victim and the situation. Jaschik and Fretz presented hypothetical scenarios (Hunter and Mc­
(1991) showed 90 women students at a mideastern Clelland, 1991 ;Jaschikand Fretz, 1991; Popovich et .
university a videotape with a classic example of sex­ al., 1987; Reilley, Carpenter, Dull, and Barlett, Uliwitdlt:d.. ......t. .
ual harassment by a teaching assistant. Before it was 1982; Rossi and Anderson, 1982; Valentine-French ~ c::uwit!dBll
labeled as sexual harassment, few women called it and Radtke, 1 989; Weber-Burdin and Rossi, 1982). c!a5e ~ c::u.-.[iiI"
that. When asked whether it was sexual harassment, Victim verbal responses and situational factors ap­ lha ..cMdts
cmc'~.... ..GII1rt1
98 percent agreed. Weber-Burdin and Rossi (1982) pear to affect whether observers label a behavior as D5l. _
replicated a previous study on sexual harassment, harassment. There is confusion over the application
of a sexual harassment label for inappropriate behav­
only they used students at the University of Massa­ ~"'.-.c
chusetts. They had 59 students rate 40 hypotheti­ ior. For example, [aschik and Fretz (1 991) found that
cal situations. Reilley, Carpenter, Dull, and Bartlett only 3 percent of the women students shown a
(1982) conducted a study of 250 female and 150 videotape with a classic example of sexual harass­
male undergraduates at the University of California ment by a teaching assistant initially labeled it as
at Santa Barbara. They also had a sample of 52 fac­ sexual harassment. Instead, they called it "sexist,"
ulty. Both samples completed a questionnaire in "rude," "unprofessional," or "demeaning." When
asked whether it was sexual harassment, 98 percent
3. . . . . . .
which respondents were presented vignettes of sex­
agreed. Roscoe and colleagues (1987) reported ......-1
ual-harassing situations that they were to rate.
similar labeling difficulties. dkrt. h pnJri*:IlI
Popovich and Colleagues (1986) created a nine­
item scale of sexual harassment. They studied 209
nw:ion ~
kKlg disuncn
sources of information. Many websites, home link indicator (usually a button or a highlighted month. the ill'.- ."I
pages, and other Internet resource pages have word or phrase). This connects people to more 00II:k ar DO COiL
"hot links" that can call up information from re­ information and provides "instant" access to tiom 00 .-0<.­
lated sites or sources simply by clicking on the cross-referenced material. Links make embed­ wt:I3l apw:eus c.

ficulty publishing or disseminating their materi­

als can now do so with ease.
4. The Internet is the provider of a very
The Internet is not a single thing inone place.Rather, wide range of information sources, some in for­
the Internet is a system or interconnected web of mats that are more dynamic and interesting. It
computers around the world. It is changing very can send and be a resource for more than
rapidly. Icannot describe everythingon the Internet; straight black and white text, as in traditional
many large books attempt to do that. Plus, even if I academic journals and sources. It transmits in­
tried, it would be out of date in six months. The In­ formation in the form of bright colors, graphics,
ternet is changing, ina powerful way, howmanypeo­ "action" images, audio (e.g., music, voices,
ple communicate and share information. sounds), photos, and video clips. Authors and
The Internet provides low-cost (often free), other creators of information can be creative in
worldwide, fast communication among people with their presentations.
computers or between people with computers and
information in the computers of organizations (e.g.,
universities, government agencies, businesses). The Down Side
There are special hardware and software require­ 1. There is no quality control over what gets
ments,but the Internet potentiallycan transmit elec­ on the Internet. Unlike standard academic pub­
tronic versions of text material, up to entire books,as lications, there is no peer-review process or any
well as photos, music, video, and other information. review. Anyone can put almost anything on a
To get onto the Internet, a person needs an ac­ website. It may be poor quality, undocumented,
count in a computer that is connected to the Inter­ highly biased, totally made up, or plain fraudu­
net. Most college mainframe computers are lent. There is a lot of real "trash" out there! Once
connected, many businessor governmentcomputers
a person finds material, the real work is to dis­
are connected, and individuals with modemscan pur­
tinguish the "trash" from valid information.
chase a connection froman Internet service provider
that provides access over telephone lines, special One needs to treat a webpage with the same cau­
DSL lines, or cabletelevision lines. In addition to a mi­ tion that one applies to a paper flyer someone
crocomputer, the person needs only a little knowl­ hands out on the street; it could contain the dri­
edge about usingcomputers. vel of a "nut" or be really valuable information.
A less serious problem is that the"glitz" ofbright
colors, music, or moving images found in sites
can distract unsophisticated users. The "glitz"
ding one source within a network of related may attract them more than serious content,
sources easy. and they may confuse glitz for high-caliber in­
formation. The Internet is better designed for a
3. The Internet speeds the flow of informa­
quick look and short attention spans rather than
tion around the globe and has a "democratizing"
the slow, deliberative, careful reading and study
effect. It provides rapid transmission of infor­
of content.
mation (e.g., text, news, data, and photos) across
long distances and international borders. In­ 2. Many excellent sources and some of the
stead of waiting a week for a report or having to most important resource materials (research
send off for a foreign publication and wait for a studies and data) for social research are not
month, the information is often available in sec­ available on the Internet (e.g., Sociofile, GSS
onds at no cost. There are virtually no restric­ datafiles, and recent journal articles). Much in­
tions on who can put material on the Internet or formation is available only through special sub­
what appears on it, so many people who had dif­ scription services that can be expensive.


Contrary to popular belief, the Internet has not inal materials and read them for ideas or to build
made all information free and accessible to
everyone. Often, what is free is limited, and
fuller information is available only to those who
on them. Also, it is easy to copy, modify, or dis­
tort, then reproduce copies of a source. For ex­
ample, a person could alter a text passage or a

pay. In fact, because some libraries redirected photo image then create a new webpage to dis­
funds to buy computers for the Internet and cut seminate the false information. This raises issues .... , d
the purchases for books and paper copies of doc­
uments, the Internet's overall impact may have
about copyright protection and the authenticity
of source material.
actually reduced what is available for some users. There are few rules for locating the best sites
on the Internet-ones that have useful and
3. Finding sources on the Internet can be
truthful information. Sources that originate at
very difficult and time consuming. It is not easy
universities, research institutes, or government
to locate specific source materials. Also, different
agencies usually are more trustworthy for re­
"search engines" can produce very different re­
search purposes than ones that are individual
sults. It is wise to use multiple search engines
home pages of unspecified origin or location, or
(e.g., Yahoo, Excite, and Google), since they
that a commercial organization or a political/so­
work differently. Most search engines simply
cial issue advocacy group sponsors. In addition
look for specific words in a short description of
to moving or disappearing, many webpages or
the webpage. This description may not reveal the
sources fail to provide complete information to
full content of the source, just as a title does not
make citation easy. Better sources provide fuller
fully tell what a book or article is about. In addi­
or more complete information about the author,
tion, search engines often come up with tens of
date, location, and so on.
thousands of sources, far too many for anyone to
As you prepare a review of the scholarly lit­
examine. The ones at the "top" may be there be­
erature and more narrowly focus a topic, you
cause they were recently added to the Internet or
should be thinking about how to design a study.
because their short description had several ver­
The specifics of design can vary somewhat de­
sions of the search word. The "best" or most rel­
pending on whether your study will primarily
evant source might be buried as the 150th item
found in a search. Also, one must often wade
through a lot of commercials and advertise­
ments to locate "real" information.
employ a quantitative-deductive-positivist ap­
proach or a qualitative-inductive-interpretive/
critical approach. The two approaches have a -
~ .........~
great deal in common and mutually comple­ iR IioIralID
4. Internet sources can be "unstable" and ment one another, but there several places where Reseu:dItu
difficult to document. After one conducts a "branches in the path" of designing a study di­ ~com ...."
search on the Internet and locates webpages with verge depending on the approach you adopt. \Jlber. but thr
information, it is important to note the specific iitrles are IJ:BJhUlilJl
"address" (usually it starts http://) where it re­ dion to uo:ll.ln1Ci1111
sides. This address refers to an electronic file sit­ ~.:.anbe
ting in a computer somewhere. If the computer
file is moved, it may not be at the same address


two months later. Unlike a journal article that

will be stored on a shelf or on microfiche in hun­ Qualitative and quantitative research differ in
dreds oflibraries for many decades to come and many ways, but they complement each other, as search, The path ill
available for anyone to read, webpages can well. All social researchers systematically collect \Ji things to do: ..
quickly vanish. This means it may not be possi­ and analyze empirical data and carefully exam­ researcher has bell
ble to check someone's web references easily, ine the patterns in them to understand and ex­ --here he or she ill
verify a quote in a document, or go back to orig­ plain social life. One of the differences between -urn and marked

the two styles comes from the nature ofthe data. other researchers have trod. Alternatively, it may
Softdata, in the form of impressions, words, sen­ be a new path into unknown territory where few
tences, photos, symbols, and so forth, dictate others have gone, and without signs marking the
different research strategies and data collection direction forward.
techniques than hard data, in the form of num­ In general, quantitative researchers follow a
bers. Another difference is that qualitative and more linear path than do qualitative researchers.
quantitative researchers often hold different as­ A linear research path follows a fixed sequence of
sumptions about social life and have different steps; it is like a staircase leading in one clear di­
objectives. These differences can make tools rection. It is a way of thinking and a wayoflook­
used by the other style inappropriate or irrele­ ing at issues-the direct, narrow, straight path
vant. People who judge qualitative research by that is most common in western European and
standards of quantitative research are often dis­ North American culture.
appointed, and vice versa. It is best to appreciate Qualitative research is more nonlinear and
the strengths each style offers. . cyclical. Rather than moving in a straight line, a
To appreciate the strengths of each style,it is nonlinearresearch path makes successive passes

I important to understand the distinct orienta­

tions of researchers. Qualitative researchers of­
ten rely on interpretive or critical social science,
follow a nonlinear research path, and speak a
through steps, sometimes moving backward and
sidewaysbefore moving on. It is more ofa spiral,
moving slowly upward but not directly. With
each cycleor repetition, a researcher collects new
language of "cases and contexts." They empha­ data and gains new insights.
size conducting detailed examinations of cases People who are used to the direct, linear ap­
that arise in the natural flow of social life. They proach may be impatient with a less direct cycli­
usually try to present authentic interpretations cal path. From a strict linear perspective, a
that are sensitive to specific social-historical cyclicalpath looks inefficient and sloppy. But the
contexts. diffuse cyclical approach is not merely disorga­
Almost all quantitative researchers rely on a nized, undefined chaos. It can be highly effective
positivist approach to social science.They follow for creating a feeling for the whole, for grasping
a linear research path, speak a language of "vari­ subtle shades of meaning, for pulling together
abies and hypotheses," and emphasize precisely divergent information, and for switching per­
measuring variables and testing hypotheses that spectives. It is not an excuse for doing poor­
are linked to general causal explanations. quality research, and it has its own discipline and
Researchers who use one style alone do not rigor. It borrows devices from the humanities
always communicate well with those using the (e.g., metaphor, analogy, theme, motif, and
other, but the languages and orientations of the irony) and is oriented toward constructing
styles are mutually intelligible. It takes time and meaning. A cyclical path is suited for tasks such
effort to understand both styles and to see how as translating languages, where delicate shades of
they can be complementary. meaning, subtle connotations, or contextual dis­
tinctions can be important.
Linear and Nonlinear Paths
Preplanned and Emergent Research
Researchers follow a path when conducting re­
search. The path is a metaphor for the sequence
of things to do: what is finished first or where a Your first step when beginning a research proj­
researcher has been, and what comes next or ect is to select a topic. There is no formula for
where he or she is going. The path may be well this task. Whether you are an experienced re­
worn and marked with signposts where many searcher or just beginning, the best guide is to

conduct research on something that interests Typical research questions for qualitative
you. researchers include: How did a certain condition
All research begins with a topic but a topic is or social situation originate? How is the condi­
only a starting point that researchers must nar­ tion/situation maintained over time? What are
row into a focused research question. Qualita­ the processes by which a condition/situation 1. Examine the liusl
tive and quantitative researchers tend to adopt changes, develops, or operates? A different type excellent 5OUra!
They are usuaty
different approaches to turn a topic to a focused of question tries to confirm existing beliefs or as­
ficity and s~
research question for a specific study. Qualita­ sumptions. A last type of question tries to dis­
tive researchers often begin with vague or un­ cover new ideas.
clear research questions. The topic emerges Research projects are designed around re­ a. Replicate a pi
slowly during the study. The researchers often search problems or questions. Before designing a or with slignt
combine focusing on a specific question with the project, quantitative researchers focus on a spe­ b. Explore uneII
process of deciding the details of study design cific research problem within a broad topic. For previous resI
that occurs while they are gathering data. By example, your personal experience might sug­ c. Follow sugge
contrast, quantitative researchers narrow a topic gest labor unions as a topic. "Labor unions" is a research at 11
into a focused question as a discrete planning topic, not a research question or a problem. In d. Extend an eIIii
step before they finalize study design. They use it any large library, you will find hundreds of new topic Of" ~
as a step in the process of developing a testable books and thousands of articles written by soci­ e. Challenge ml
hypothesis (to be discussed later) and to guide ologists, historians, economists, management lationship. .
the study design before they collect any data. officials, political scientists, and others on f. Specify the ill
The qualitative research style is flexible and unions. The books and articles focus on different linking relatiG
encourages slowly focusing the topic throughout aspects of the topic and adopt many perspectives
2. Talkoverideas"
a study. In contrast to quantitative research, only on it. Before proceeding to design a research
a small amount of topic narrowing occurs in an project, you must narrow and focus the topic. a. Ask people •
the topic for
early research planning stage, and most of the An example research question is, "How much
have thoughl
narrowing occurs after a researcher has begun to did U.S. labor unions contribute to racial in­
collect data. equality by creating barriers to skilled jobs for b. Seek out thea
The qualitative researcher begins data gath­ African Americans in the post-World War II from yours aI
ering with a general topic and notions of what period?" research quei
will be relevant. Focusing and refining contin­ When starting research on a topic, ask your­ 3. Apply to a s~
ues after he or she has gathered some of the data self: What is it about the topic that is of greatest a. Focus the tOIl
and started preliminary analysis. Qualitative re­ interest? For a topic about which you know little, riod or time II
searchers use early data collection to guide how first get background knowledge by reading b. Narrow the 111
they adjust and sharpen the research question(s) about it. Research questions refer to the rela­ ographic urjt
because they rarely know the most important is­ tionships among a small number of variables. c. Consider w+W
sues or questions until after they become fully Identify a limited number of variables and spec­ people/units.
immersed in the data. Developing a focused re­ ify the relationships among them. are differerxa
search question is a part of the data collection A research question has one or a small num­ 4. Define the aim Of'"i
process, during which the researcher actively re­ ber of causal relationships. Box 4.5 lists some
a. Will the re~
flects on and develops preliminary interpreta­ ways to focus a topic into a research question.
ploratory. ~
tions. The qualitative researcher is open to For example, the question, "What causes di­
unanticipated data and constantly reevaluates b. Will the ~
vorce?" is not a good research question. A better
the focus early in a study. He or she is prepared research question is, "Is age at marriage associ­
to change the direction of research and follow ated with divorce?" The second question sug­
new lines of evidence. gests two variables: age of marriage and divorce.

Another technique for focusing a research

question is to specify the universe to which the
answer to the question can be generalized. All re­
search questions, hypotheses, and studies apply
1. Examine the literature. Published articles are an to some group or category of people, organiza­
excellent source of ideas for research questions. tions, or other units. The universe is the set of all
They are usually at an appropriate level of speci­ units that the research covers, or to which it can
ficity and suggest research questions that focus be generalized. For example, your research ques­
on the following: tion is about the effectsof a new attendance pol­
a. Replicate a previous research project exactly icy on learning by high school students. The
or with slight variations. universe, in this case, is all high school students.
b. Explore unexpected findings discovered in When refining a topic into a research ques­
previous research. tion and designing a research project, you also
e. Followsuggestions an author gives for future need to consider practical limitations. Designing
research at the end of an article. a perfect research project is an interesting acad­
d. Extend an existing explanation or theory to a emic exercise, but if you expect to carry out a re­
new topic or setting. search project, practical limitations will have an
e. Challenge findings or attempt to refute a re­ impact on its design.
lationship. Major limitations include time, costs, access
to resources, approval by authorities, ethical
f. Specify the intervening process and consider
linking relations.
concerns, and expertise. If you have 10 hours a
week for five weeks to conduct a research proj­
2. Talk over ideas with others.
ect, but the answer to a research question will
a. Ask people who are knowledgeable about take fiveyears, reformulate the research question
the topic for questions about it that they more narrowly. Estimating the amount of time
have thought of. required to answer a research question is diffi­
b. Seek out those who hold opinions that differ cult. The research question specified, the re­
from yours on the topic and discuss possible search technique used, and the type of data
research questions with them. collected all play significant roles. Experienced
3. Apply to a specific context. researchers are the best source of good estimates.
a. Focus the topic onto a specific historical pe­ Cost is another limitation. As with time,
riod or time period. there are inventive ways to answer a question
b. Narrow the topic to a specific society or ge­ within limitations, but it may be impossible to
ographic unit. answer some questions because of the expense
e. Consider which subgroups or categories of involved. For example, a research question
people/units are involved and whether there about the attitudes of all sports fans toward their
are differences among them. team mascot can be answered only with a great
4. Define the aim or desired outcome of the study. investment of time and money. Narrowing the
research question to how students at two differ­
a. Will the research question be for an ex­
ent colleges feel about their mascots might make
ploratory, explanatory, or descriptive study?
it more manageable.
b. Will the study involve applied or basic re­ Accessto resources is a common limitation.
search? Resources can include the expertise of others,
special equipment, or information. For example,
a research question about burglary rates and

family income in many different nations is al­ 4.1). In addition, researchers tend to adopt a dif­ ,
most impossible to answer because information ferent language and approach to study design,
on burglary and income is not collected or avail­ which we will consider next.
able for most countries. Some questions require
the approval of authorities (e.g., to see medical
records) or involve violating basic ethical princi­ QUALITATIVE DESIGN ISSUES
ples (e.g., causing serious physical harm to a per­

The Language of Cases and Contexts
son to see the person's reaction). The expertise
or background of the researcher is also a limita­
tion. Answering some research questions in­
Qualitative researchers use a language of cases
and contexts, examine social processes and cases
volves the use of data collection techniques,
statistical methods, knowledge of a foreign lan­
in their social context, and look at interpreta­
tions or the creation of meaning in specific set­
guage, or skillsthat the researcher may not have. tings. They try look at social life from multiple ~QII-"
Unless the researcher can acquire the necessary points of view and explain how people construct a aJ
training or can pay for another person's services, identities. Only rarely do they use variables or
the research question may not be practical.
In summary, styles of qualitative and quan­
titative researchers have much in common, but
the researchers often differ on design issues,
test hypotheses, or try to convert social life into
Qualitative researchers see most areas and
activities of social life as being intrinsically qual­
such as taking a linear or nonlinear research path itative. To them, qualitative data are not impre­
and developing a research question (see Table cise or deficient; they are highly meaningful.

TABLE 4.1 Quantitative Reasearch versus Qualitative Research
.. c
A a 5
Test hypothesis that the researcher begins with. Capture and discover meaning once the researcher

becomes immersed in the data.

Concepts are in the form of distinct variables. Concepts are in the form of themes, motifs,

generalizations, and taxonomies.

Measures are systematically created before data Measures are created in an ad hoc manner and are

collection and are standardized. often specific to the individual setting or


Data are in the form of numbers from precise Data are in the form of words and images from

measurement. documents, observations, and transcripts.

Theory is largely causal and is deductive. Theory can be causal or noncausal and is often
a . ._ .

Procedures are standard, and replication is Research procedures are particular, and replication

assumed. is very rare.

Analysis proceeds by using statistics, tables, or Analysis proceeds by extracting themes or

charts and discussing how what they show relates generalizations from evidence and organizing data

to hypotheses. to present a coherent, consistent picture.


Instead of trying to convert social life into vari­ car's license number before proceeding? After
ables or numbers, qualitative researchers borrow radioing the car's location, does the officer ask
ideas from the people they study and place them the motorist to get out of the car sometimes, but
within the context of a natural setting. They ex­ in others casually walk up to the car and talk to
amine motifs, themes, distinctions, and ideas in­ the seated driver? When data collection and the­
stead of variables, and they adopt the inductive orizing are interspersed, theoretical questions
approach of grounded theory. arise that suggest future observations, so new
Some people believe that qualitative data are data are tailored to answer theoretical questions
"soft," intangible, and immaterial. Suchdata are that came from thinking about previous data.
so fuzzy and elusive that researchers cannot re­
ally capture them. This is not necessarily the
The Context Is Critical
case. Qualitative data are empirical. They in­
volve documenting real events, recording what Qualitative researchers emphasize the social
people say (with words, gestures, and tone), ob­ context for understanding the social world. They
serving specific behaviors, studying written doc­ hold that the meaning of a social action or state­
uments, or examining visual images. These are ment depends, in an important way, on the con­
all concrete aspects of the world. For example, text in which it appears. When a researcher
some qualitative researchers take and closely removes an event, social action, answer to a
scrutinize photos or videotapes of people or so­ question, or conversation from the social con­
cial events. This evidence is just as "hard" and text in which it appears, or ignores the context,
physical as that used by quantitative researchers social meaning and significance are distorted.
to measure attitudes, social pressure, intelli­ Attention to social context means that a
gence, and the like. qualitative researcher notes what came before or
what surrounds the focus of study. It also im­
plies that the same events or behaviors can have
Grounded Theory
different meanings in different cultures or his­
A qualitative researcher develops theory during torical eras. For example, instead of ignoring the
the data collection process. This more inductive context and counting votes across time or cul­
method means that theory is built from data or tures, a qualitative researcher asks: What does
grounded in the data. Moreover, conceptualiza­ voting mean in the context? He or she may treat
tion and operationalization occur simultane­ the same behavior (e.g.,voting for a presidential
ously with data collection and preliminary data candidate) differently depending on the social
analysis. It makes qualitative research flexible context in which it occurs. Qualitative re­
and lets data and theory interact. Qualitative re­ searchers place parts of social life into a larger
searchers remain open to the unexpected, are whole. Otherwise, the meaning of the part may
willing to change the direction or focus of a re­ be lost. For example, it is hard to understand
search project, and may abandon their original what a baseball glove is without knowing some­
research question in the middle of a project. thing about the game of baseball. The whole of
A qualitative researcher builds theory by the game-innings, bats, curve balls, hits-gives
making comparisons. For example, when a re­ meaning to each part, and each part without the
searcher observes an event (e.g., a police officer whole has little meaning.
confronting a speeding motorist), he or she im­
mediately ponders questions and looks for sim­
The Case and Process
ilarities and differences. When watching a police
officer stop a speeder, a qualitative researcher In quantitative research, cases are usually the
asks: Does the police officer always radio in the same as a unit of analysis, or the unit on which

variables are measured (discussed later). Quan­ tative researcher gives meaning by rearranging,
titative researchers typically measure variables of examining, and discussing textual or visual data
their hypotheses across many cases. For exam­ in a way that conveys an authentic voice, or that
ple, if a researcher conducts a survey of 450 indi­ remains true to the original understandings of
viduals, each individual is a case or unit on the people and situations that he or she studied.
which he or she measures variables. Qualitative Instead of relying on charts, statistics, and
researchers tend to use a "case-oriented ap­
proach [that] places cases, not variables, center
stage" (Ragin, 1992:5). They examine a wide va­
riety of aspects of one or a few cases.Their analy­
displays of numbers, qualitative researchers put
a greater emphasis on interpreting the data.
Their data are often "richer" or more complex
and full of meaning. The qualitative researcher
..- .

ses emphasize contingencies in "messy" natural interprets to "translate" or make the originally
settings (i.e., the co-occurrence of many specific gathered data understandable to other people.
factors and events in one place and time). Expla­ The process of qualitative interpretation moves
nations or interpretations are complex and may through three stages or levels.
be in the form of an unfolding plot or a narrative A researcher begins with the point of view
story about particular people or specific events. of the people he or she is studying, and the re­
Rich detail and astute insight into the cases re­ searcher wants to grasp fully how they see the
place the sophisticated statistical analysis of pre­ world, how they define situations, or what
cise measures across a huge number of units or
cases found in quantitative research.
The passage of time is integral to qualitative
research. Qualitative researchers look at the se­
quence of events and pay attention to what hap­
pens first, second, third, and so on. Because
things mean to them. A first-order interpretation
contains the inner motives, personal reasons,
and point of view of the people who are being
studied in the original context. As the researcher
discovers and documents this first-order inter­
pretation, he or she remains one step removed


qualitative researchers examine the same case or from it. The researcher offers a second-order in­
set of cases over time, they can see an issue terpretation, which is an acknowledgment that
evolve, a conflict emerge, or a social relationship however much a researcher tries to get very
develop. The researcher can detect process and close and "under the skin" of those he or she is
causal relations. studying, a researcher is still "on the outside
In historical research, the passage of time looking in." In the second-order interpretation,
may involve years or decades. In field research, the researcher tries to elicit an underlying co­
the passage of time is shorter. Nevertheless, in herence or sense of overall meaning in the data.
both, a researcher notes what is occurring at dif­ To reach an understanding of what he or she
ferent points in time and recognizes that when sees or hears, a researcher often places the data
something occurs is often important. into a context of the larger flow of events and
behaviors. A qualitative researcher will often
move to the third step and link the understand­
ing that he or she achieved to larger concepts,
Interpretation means to assign significance or a generalizations, or theories. The researcher can ~lhe.-;JI_rlll
coherent meaning to something. Quantitative share this broader interpretation with other
and qualitative researchers both interpret data, people who are unfamiliar with the original
but they do so in different ways. A quantitative data, the people and events studied, or the social
researcher gives meaning by rearranging, exam­ situations observed by the researcher. This level
ining, and discussing the numbers by using of meaning translates the researcher's own un­
charts and statistics to explain how patterns in derstanding in a way that facilitates communica­
the data relate to the research question. A quali­ tion with people who are more distant from the

original source, and it represents a third-order able. It describes the intensity or strength of at­
interpretation. tachment to attitudes, beliefs, and behaviors as­
sociated with the concept of masculine within a
culture. "Married" is not a variable; it is an at­
tribute of the variable "marital status." Related
QUANTITATIVE DESIGN ISSUES ideas such as "number of years married" or
The language of Variables and "depth of commitment to a marriage" are vari­
ables. Likewise, "robbery" is not a variable; it is
an attribute of the variable "type of crime."
Variation and Variables. The variable is a "Number of robberies," "robbery rate,"
central idea in quantitative research. Simply de­ "amount taken during a robbery," and "type of
fined, a variable is a concept that varies. Quanti­ robbery" are all variables because they vary or
tative research uses a language of variables and take on a range of values.
relationships among variables. Quantitative researchers redefine concepts
In Chapter 2, you learned about two types of interest into the language of variables. As the
of concepts: those that refer to a fixed phenom­ examples of variables and attributes illustrate,
enon (e.g., the ideal type of bureaucracy) and slight changes in definition change a nonvariable
those that vary in quantity, intensity, or amount into a variable concept. As you saw in Chapter 2,
(e.g., amount of education). The second type of concepts are the building blocks of theory; they
concept and measures of the concepts are vari­ organize thinking about the social world. Clear
ables. Variables take on two or more values. concepts with careful definitions are essential in
Once you begin to look for them, you will see theory.
variables everywhere. For example, gender is a
variable; it can take on two values: male or fe­ Types ofVariables. Researchers who focus on
male. Marital status is a variable; it can take on causal relations usually begin with an effect, then
the values of never married single, married, di­ search for its causes. Variables are classified into
vorced, or widowed. Type of crime committed is three basic types, depending on their location in
a variable; it can take on values of robbery, bur­ a causal relationship. The cause variable, or the
glary, theft, murder, and so forth. Family income one that identifies forces or conditions that act
is a variable; it can take on values from zero to on something else, is the independent variable.
billions of dollars. A person's attitude toward The variable that is the effect or is the result or
abortion is a variable; it can range from strongly outcome of another variable is the dependent
favoring legal abortion to strongly believing in variable. The independent variable is "indepen­
antiabortion. dent of' prior causes that act on it, whereas the
The values or the categories of a variable are dependent variable "depends on" the cause.
its attributes. It is easy to confuse variables with It is not always easy to determine whether a
attributes. Variables and attributes are related, variable is independent or dependent. Two
but they have distinct purposes. The confusion questions help you identify the independent
arises because the attribute of one variable can variable. First, does it come before other vari­
itself become a separate variable with a slight ables in time? Independent variables come be­
change in definition. The distinction is between fore any other type. Second, if the variables
concepts themselves that vary and conditions occur at the same time, does the author suggest
within concepts that vary. For example, "male" that one variable has an impact on another vari­
is not a variable; it describes a category of gender able? Independent variables affect or have an im­
and is an attribute of the variable "gender." Yet, pact on other variables. Research topics are often
a related idea, "degree of masculinity," is a vari­ phrased in terms of the dependent variables be­


cause dependent variables are the phenomenon punishment for criminal acts. A multicause ex­
to be explained. For example, suppose a re­ planation usually specifies the independent vari­
searcher examines the reasons for an increase in able that has the greatest causal effect.
the crime rate in Dallas, Texas; the dependent A complex theoretical explanation contains
variable is the crime rate. a string of multiple intervening variables that are
A basic causal relationship requires only an linked together. For example, family disruption
independent and a dependent variable. A third causes lower self-esteem among children, which
type of variable, the intervening variable, appears causes depression, which causes poor grades in
in more complex causal relations. It comes be­ school, which causes reduced prospects for a
tween the independent and dependent variables
and shows the link or mechanism between them.
Advances in knowledge depend not only on doc­
umenting cause-and-effect relationships but
good job, which causes a lower adult income.
The chain of variables is: family disruption (in­
dependent), childhood self-esteem (interven­
ing), depression (intervening), grades in school

4. ­

. . .MiI....
also on specifying the mechanisms that account (intervening), job prospects (intervening), adult
for the causal relation. In a sense, the intervening income (dependent).
variable acts as a dependent variable with respect Two theories on the same topic may have
to the independent variable and acts as an inde­ different independent variables or predict differ­
pendent variable toward the dependent variable. ent independent variables to be important. In
For example, French sociologist Emile addition, theories may agree about the indepen­
Durkheim developed a theory of suicide that dent and dependent variables but differ on the
specified a causal relationship between marital intervening variable or causal mechanism. For
status and suicide rates. Durkheim found evi­ example, two theories say that family disruption
dence that married people are less likely to com­ causes lower adult income, but for different rea­
mit suicide than single people. He believed that sons. One theory holds that disruption encour­
married people have greater social integration ages children to join deviant peer groups that are
(i.e., feelings of belonging to a group or family). not socialized to norms of work and thrift.
He thought that a major cause of one type of sui­ Another emphasizes the impact of the disrup­
cide was that people lacked a sense of belonging tion on childhood depression and poor acade­ **.IQS ......
to a group. Thus, his theory can be restated as a mic performance, which directly affect job
three-variable relationship: marital status (inde­ performance.
pendent variable) causes the degree of social in­ A single research project usually tests only a
tegration (intervening variable), which affects small part of a causal chain. For example, a re­
suicide (dependent variable). Specifying the search project examining six variables may take
chain of causality makes the linkages in a theory the six from a large, complex theory with two
clearer and helps a researcher test complex ex­ dozen variables. Explicit links to a larger theory
planations.' strengthen and clarify a research project. This
Simple theories have one dependent and applies especiallyfor explanatory, basic research,
one independent variable, whereas complex the­ which is the model for most quantitative re­
ories can contain dozens of variables with multi­ search.
ple independent, intervening, and dependent
variables. For example, a theory of criminal be­
Causal Theory and Hypotheses
havior (dependent variable) identifies four inde­
pendent variables: an individual's economic The Hypothesis and Causality. A hypothesis is
hardship, opportunities to commit crime easily, a proposition to be tested or a tentative state­
membership in a deviant subgroup of society ment of a relationship between two variables.
that does not disapprove of crime, and lack of Hypotheses are guesses about how the social

picture of the research process by focusing on a

single research project that tests one hypothesis.
Knowledge develops over time as researchers
throughout the scientific community test many
1. It has at least two variables. hypotheses. It grows from shifting and winnow­
2. It expresses a causal or cause-effect relationship ing through many hypotheses. Each hypothesis
between the variables. represents an explanation of a dependent vari­
3. It can be expressed as a prediction or an ex­ able. If the evidence fails to support some hy­
pected future outcome. potheses, they are gradually eliminated from
4. It is logically linked to a research question and a consideration. Those that receive support re­
theory. main in contention. Theorists and researchers
5. It is falsifiable; that is, it is capable of being
constantly create new hypotheses to challenge
tested against empirical evidence and shown to those that have received support. Figure 4.3 rep­
be true or false. resents an example of the process of shifting
through hypotheses over time.
Scientists are a skeptical group. Support for
a hypothesis in one research project is not suffi­
cient for them to accept it. The principle of repli­
world works; they are stated in a value-neutral cation says that a hypothesis needs several tests
form. with consistent and repeated support to gain
A causal hypothesis has five characteristics broad acceptance. Another way to strengthen
(see Box 4.6). The first two characteristics define confidence in a hypothesis is to test related
the minimum elements of a hypothesis. The causal linkages in the theory from which it
third restates the hypothesis. For example, the comes.
hypothesis that attending religious services re­
duces the probability of divorce can be restated Types a/Hypotheses. Hypotheses are links in a
as a prediction: Couples who attend religious theoretical causal chain and can take several
services frequently have a lower divorce rate forms. Researchers use them to test the direction
than do couples who rarely attend religious ser­ and strength of a relationship between variables.
vices. The prediction can be tested against em­ When a hypothesis defeats its competitors, or of­
pirical evidence. The fourth characteristic states fers alternative explanations for a causal relation,
that the hypothesis should be logically tied to a it indirectly lends support to the researcher's ex­
research question and to a theory. Researchers planation. A curious aspect of hypothesis testing
test hypotheses to answer the research question is that researchers treat evidence that supports a
or to find empirical support for a theory. The hypothesis differently from evidence that op­
last characteristic requires that a researcher use poses it. They give negative evidence more im­
empirical data to test the hypothesis. Statements portance. The idea that negative evidence is
that are necessarily true as a result of logic, or critical when evaluating a hypothesis comes
questions that are impossible to answer through from the logic of disconftrming hypotheses? It is
empirical observation (e.g., What is the "good associated with Karl Popper's idea of falsification
life"? Is there a God?) cannot be scientific hy­ and with the use of null hypotheses (see later in
potheses. this section).
A hypothesis is never proved, but it can be
Testing and Refining Hypothesis. Knowledge disproved. A researcher with supporting evi­
rarely advances on the basis of one test of a sin­ dence can say only that the hypothesis remains a
gle hypothesis. In fact, it is easy to get a distorted possibility or that it is still in the running. Nega­

FIG U R E 4. 3

How the Process of Hypotheses Testing Operates over Time


for the hypothes

on him and his u
destroy the hvpo
There are five possible hypotheses.
straightforward 1
Many quantitam
Remain under
imenters, frame I
Two of the original five hypotheses
pothesis based OIl
are rejected.

hypotheses. Thes
A new one is developed.

evidence that \\iI

the null hypotbes
1986 pothesis as a wa~
null hypothesis d
Two hypotheses are rejected.
Two new ones are developed. relationship. FOJ
students who live
higher grades thai
and commute to
1996 that there is no I
Remain under and grades. Resell
Three hypotheses are rejected.

,. ---· A new one is developed. with a correspoa

experimental b~
0 8
:1 9 :1
pothesis says thal
alternative hypod
pus residence bas
Reject Remain under
One hypothesis is rejected. For most pel
~ Consideration
.----.. .----.. Two new ones are developed.
proach is a backa

iJ 9 0 8

L~~J L~~J
Null hypothesis t
tion that research
ship, so hypothesi
In 2006, 3 hypotheses are in contention, but from 1966 to 2006, 11 hypotheses were considered, and over make fmding a rei
time, 8 of them were rejected in one or more tests. researcher who 1
proach only dimIl
evidence supports
cept the null h~
tive evidence is more significant because the hy­ For example, a man stands on a street cor­ the tested relatioa
pothesis becomes "tarnished" or "soiled" if the ner with an umbrella and claims that his um­ plies that the altes
evidence fails to support it. This is because a hy­ brella protects him from falling elephants. His the other hand, iI
pothesis makes predictions. Negative and dis­ hypothesis that the umbrella provides protec­ dence to reject till
confirming evidence shows that the predictions tion has supporting evidence. He has not had a ternative hypo~
are wrong. Positive or confirming evidence for a single elephant fallon him in all the time he has researcher cannes
hypothesis is less critical because alternative hy­ had his umbrella open. Yet, such supportive ev­ by testing the nul
potheses may make the same prediction. A re­ idence is weak; it also is consistent with an alter­ the alternative h~
searcher who finds confirming evidence for a native hypothesis-that elephants do not fall null hypothesis til
prediction may not elevate one explanation over from the sky. Both predict that the man will be evidence, the argu
its alternatives. safe from falling elephants. Negative evidence esis can grow strOl

for the hypothesis-the one elephant that falls Many people find the null hypothesis to be
on him and his umbrella, crushing both-would confusing. Another way to think of it is that the
destroy the hypothesis for good. scientific community is extremely cautious. It
Researchers test hypotheses in two ways: a prefers to consider a causal relationship to be
straightforward way and a null hypothesis way. false until mountains of evidence show it to be
Many quantitative researchers, especially exper­ true. This is similar to the Anglo-American legal
imenters, frame hypotheses in terms ofa null hy­ idea of innocent until proved guilty. A re­
pothesis based on the logic of the disconfirming searcher assumes, or acts as if, the null hypothe­
hypotheses. They test hypotheses by looking for sis is correct until reasonable doubt suggests
evidence that will allow them to accept or reject otherwise. Researchers who use null hypotheses
the null hypothesis. Most people talk about a hy­ generally use it with specific statistical tests (e.g.,
pothesis as a way to predict a relationship. The r-test or F-test). Thus, a researcher may say there
null hypothesis does the opposite. It predicts no is reasonable doubt in a null hypothesis if a sta­
relationship. For example, Sarah believes that tistical test suggests that the odds of it being false
students who live on campus in dormitories get are 99 in 100. This is what a researcher means
higher grades than students who live off campus when he or she says that statistical tests allow
and commute to college. Her null hypothesis is him or her to "reject the null hypothesis at the
that there is no relationship between residence .01 level of significance."
and grades. Researchers use the null hypothesis
with a corresponding alternative hypothesis or
Aspects of Explanation
experimental hypothesis. The alternative hy­
pothesis says that a relationship exists. Sarah's Clarity about Units and Levels ofAnalysis. It
alternative hypothesis is that students' on-cam­ is easy to become confused at first about the
pus residence has a positive effect on grades. ideas of units and levels of analysis. Neverthe­
For most people, the null hypothesis ap­ less, they are important for dearly thinking
proach is a backward way of hypothesis testing. through and planning a research project. All
Null hypothesis thinking rests on the assump­ studies have both units and levels ofanalysis, but
tion that researchers try to discover a relation­ few researchers explicitly identify them as such.
ship, so hypothesis testing should be designed to The levels and units of analysis are restricted by
make finding a relationship more demanding. A the topic and the research question.
researcher who uses the null hypothesis ap­ A level of analysis is the level of social reality
proach only directly tests the null hypothesis. If to which theoretical explanations refer. The level
evidence supports or leads the researcher to ac­ of social reality varies on a continuum from mi­
cept the null hypothesis, he or she concludes that cro level (e.g., small groups or individual
the tested relationship does not exist. This im­ processes) to macro level (e.g., civilizations or
plies that the alternative hypothesis is false. On structural aspects ofsociety). The level includes a
the other hand, if the researcher can find evi­ mix of the number of people, the amount of
dence to reject the null hypothesis, then the al­ space, the scope of the activity, and the length of
ternative hypotheses remain a possibility. The time. For example, an extreme micro-level
researcher cannot prove the alternative; rather, analysis can involve a few seconds of interaction
by testing the null hypotheses, he or she keeps between two people in the same small room. An
the alternative hypotheses in contention. When extreme macro-level analysis can involve billions
null hypothesis testing is added to confirming of people on several continents across centuries.
evidence, the argument for an alterative hypoth­ Most social research uses a level of analysis that
esis can grow stronger over time. lies between these extremes.

The level of analysis delimits the kinds of hand, a study that compares the amounts differ­
assumptions, concepts, and theories that a re­ ent colleges spend on their football programs
searcher uses. For example, I want to study the would use the organization (the college) as the
topic of dating among college students. I use a unit of analysis because the spending by colleges
micro-level analysis and develop an explana­ is being compared and each college's spending is
tion that uses concepts such as interpersonal recorded.
contact, mutual friendships, and common in­ Researchers use units of analysis other than
terests. I think that students are likely to date individuals, groups, organizations, social cate­
someone with whom they have had personal gories, institutions, and societies. For example, a
contact in a class, share friends in common,
and share common interests. The topic and fo­
cus fit with a micro-level explanation because
they are targeted at the level of face-to- face in­
researcher wants to determine whether the
speeches of two candidates for president of the
United States contain specific themes. The re­
searcher uses content analysis and measures the
themes in each speech of the candidates. Here,
~ ...•--.
teraction among individuals. Another example
topic is how inequality affects the forms ofvio­ the speech is the unit of analysis. Geographic
lent behavior in a society. Here, I have chosen a units of analysis are also used. A researcher in­
more macro-level explanation because of the terested in determining whether cities that have f T
topic and the level of social reality at which it a high number of teenagers also have a high rate E
operates. I am interested in the degree of in­ of vandalism would use the city as the unit of
equality (e.g., the distribution of wealth, prop­ analysis. This is because the researcher measures
erty, income, and other resources) throughout the percentage of teenagers in each city and the
a society and in patterns of societal violence amount of vandalism for each city.
(e.g., aggression against other societies, sexual The units of analysis determine how a re­
assault, feuds between families). The topic and searcher measures variables or themes. They also
research question suggest macro-level concepts correspond loosely to the level of analysis in an
and theories. explanation. Thus, social-psychologicalor micro
The unit of analysis refers to the type of unit levelsof analysisfit with the individual as a unit of
a researcher uses when measuring. Common analysis, whereas macro levels of analysis fit with
units in sociology are the individual, the group the social category or institution as a unit. Theo­
(e.g., family, friendship group), the organization ries and explanations at the micro level generally
(e.g., corporation, university), the social cate­ refer to features of individuals or interactions
gory (e.g., social class, gender, race), the social among individuals. Those at the macro levelrefer
institution (e.g., religion, education, the family), to social forces operating across a society or rela­
and the society (e.g., a nation, a tribe). Although tions among major parts of a society as a whole.
the individual is the most commonly used unit Researchers use levels and units of analysis
of analysis, it is by no means the only one. Dif­ to design research projects, and being aware of
ferent theories emphasize one or another unit of them helps researchers avoid logical errors. For
analysis, and different research techniques are example, a study that examines whether colleges

associated with specific units of analysis. For ex­ in the North spend more on their football pro­
ample, the individual is usually the unit of analy­ grams than do colleges in the South implies that
sis in survey and experimental research. a researcher gathers information on spending by
As an example, the individual is the unit of college and the location of each college. The unit
analysis in a survey in which 150 students are of analysis-the organization or, specifically, the
asked to rate their favorite football player. The college-flows from the research problem and
individual is the unit because each individual tells the researcher to collect data from each
student's response is recorded. On the other college.

Researchers choose among different units to the unit on which you collect data (see Box
or levels of analysis for similar topics or research 4.7).
questions. For example, a researcher could con­
duct a project on the topic of patriarchy and vi­ Example. Tomsville and Ioansville each have
olence with society as the unit of analysis for the about 45,000 people living in them. Tomsville
research question, "Are patriarchal societies has a high percentage of upper-income people.
more violent?" He or she would collect data on Over half of the households in the town have
societies and classify each society by its degree of family incomes of over $200,000. The town also
patriarchy and its level of violence. On the other has more motorcycles registered in it than any
hand, if the research question was "Is the degree other town of its size. The town of Ioansville has
of patriarchy within a family associated with vi­ many poor people. Half its households live be-
olence against a spouse?" the unit of analysis
could be the group or the family, and a more mi­
cro level of analysis would be appropriate. The
researcher could collect data on families by mea­
suring the degree of patriarchy within different
families and the level of violence between Researchers have criticized the famous study Suicide
spouses in these families. The same topic can be ([1897] 1951) by Emile Durkheim for the ecologi­
addressed with different levels and units of calfallacy of treating group data as though they were
analysis because patriarchy can be a variable that individual-level data. In the study, Durkheim com­
describes an entire society, or it can describe so­ pared the suicide rates of Protestant and Catholic
cial relations within one family. Likewise, vio­ districts in nineteenth-century western Europe and
lence can be defined as general behavior across a explainedobserved differences as due to differences
society, or as the interpersonal actions of one between people's beliefsand practices in the two re­
spouse toward the other. ligions. He said that Protestants had a higher suicide
rate than Catholics because they were more individ­
Ecological Fallacy. The ecological fallacy arises ualistic and had lower social integration. Durkheim
from a mismatch of units of analysis. It refers to and early researchers only had data by district. Since
a poor fit between the units for which a re­ people tended to reside with others of the same re­
searcher has empirical evidence and the units ligion, Durkheim used group-level data (i.e., region)
for which he or she wants to make statements. It for individuals.
is due to imprecise reasoning and generalizing Later researchers (van Poppel and Day, 1996)
beyond what the evidence warrants. Ecological reexamined nineteenth-century suicide rates only
fallacy occurs when a researcher gathers data at with individual-level data that they discovered for
some areas. They compared the death records and
a higher or an aggregated unit of analysis but
looked at the official reason of death and religion,
wants to make a statement about a lower or
but their results differed from Durkheim's. Appar­
disaggregated unit. It is a fallacy because what
ently, local officials at that time recorded deaths dif­
happens in one unit of analysis does not always
ferently for people of different religions. They
hold for a different unit of analysis. Thus, if a re­ recorded "unspecified" as a reason for death far more
searcher gathers data for large aggregates (e.g., often for Catholics because of a strong moral prohi­
organizations, entire countries, etc.) and then bition against suicide among Catholics. Ourkheim's
draws conclusions about the behavior of indi­ larger theory may be correct, yet the evidence he
viduals from those data, he or she is commit­ had to test it was weakbecause he used data aggre­
ting the ecological fallacy. You can avoid this gated at the group level while trying to explain the
error by ensuring that the unit of analysis you actions of individuals.
use in an explanation is the same as or very close

low the poverty line. It also has fewer motorcy­ the ecological fallacy or reductionism. They
cles registered in it than any other town its size. make a mistake about the data appropriate for a
But it is a fallacy to say, on the basis of this infor­ research question, or they may seriously over­
mation alone, that rich people are more likely to generalize from the data. t
own motorcycles or that the evidence shows a You can make assumptions about units of Suppose )QI JIIid1
relationship between family income and motor­ analysis other than the ones you study empiri­
cycle ownership. The reason is that we do not
know which families in Tomsville or ]oansville
cally. Thus, research on individuals rests on as­
sumptions that individuals act within a set of Americ:GIf1'Kr~
CHi Rit;I«s &.
jOOty• .....
own motorcycles. We only know about the two social institutions. Research on social institu­
variables-average income and number of mo­ tions is based on assumptions about individual
torcycles-for the towns as a whole. The unit of behavior. We know that many micro-level units ~liIII:~
analysis for observing variables is the town as a
whole. Perhaps all of the low- and middle-in­
form macro-level units. The danger is that it is
easy to slide into using the causes or behavior of ~":::~~.
come families in Tomsville belong to a motorcy­ micro units, such as individuals, to explain the
cle club, and not a single upper-income family
belongs. Or perhaps one rich family and five
actions of macro units, such as social institu­
tions. What happens among units at one level
poor ones in ]oansville each own motorcycles. does not necessarily hold for different units of
In order to make a statement about the relation­
ship between family ownership of motorcycles
analysis. Sociology is a discipline that rests on
the fundamental belief that a distinct level of so­
foremost eM..,
and family income, we have to collect informa­ cial reality exists beyond the individual. Expla­ This says::
tion on families, not on towns as a whole. nations of this level require data and theory that
go beyond the individual alone. The causes,
U.S. race ret_..
Reductionism. Another problem involving forces, structures, or processes that exist among If you know
mismatched units of analysis and imprecise rea­ macro units cannot be reduced to individual seeaproblela.
soning about evidence is reductionism, also behavior. its successes ~
called the fallacy ofnonequivalence (see Box 4.8). one individual
This error occurs when a researcher explains Example. Why did World War I occur? You and guide a
macro-level events but has evidence only about may have heard that it was because a Serbian The idea of iI
specific individuals. It occurs when a researcher shot an archduke in the AustroHungarian Em­ force is reduced 1D
observes a lower or disaggregated unit of analysis pire in 1914. This is reductionism. Yes, the as­ cial phel1OlneflOrl'-i
but makes statements about the operations of sassination was a factor, but the macro-political are the actions of
higher or aggregated units. It is a mirror image of event between nations-war-cannot be re­ (marches. cOtn
the mismatch error in the ecological fallacy. A duced to a specific act of one individual. If it sit-ins, rioting.
advancing a shaRd
researcher who has data on how individuals be­ could, we could also say that the war occurred
have but makes statements about the dynamics because the assassin's alarm clock worked and
of macro-level units is committing the error of woke him up that morning. If it had not worked, l
reductionism. It occurs because it is often easier there would have been no assassination, so the but individual ~
to get data on concrete individuals. Also, the op­ alarm clock caused the war! The event, World Thus, it is ~ ~
eration ofmacro-level units is more abstract and War I, was much more complex and was due to out at about that ~
nebulous. As with the ecological fallacy, you can many social, political, and economic forces that had not occurred. j
avoid this error by ensuring that the unit of came together at a point in history. The actions I
analysis in your explanation is very close to the of specific individuals had a role, but only a mi­ Spuriousness. ~
one for which you have evidence. nor one compared to these macro forces. Indi­ variables spurious.
Researchers who fail to think precisely viduals affect events, which eventually, in Researchers get eIj
about the units of analysis and those who do not combination with larger-scale social forces and found a spurious ~
couple data with the theory are likely to commit organizations, affect others and move nations, show that what a(I

Suppose you pick up a book and read the following: The movement's ideology, popular mobilization, pol­
itics, organization, and strategy are absent. Related
American race relations changed dramatically during the macro-level historical events and trends that may
Civil Rights Era of the 796 as. Attitudes among the ma­ have influenced the movement (e.g., Vietnam War
jority, white population shifted to greater tolerance as protest, mood shift with the killing of John F.
laws and court rulings changed across the nation. Op­ Kennedy, African American separatist politics, African
portunities that had beenlegally and officially closed to American migration to urban North) are also ignored.
all but the white population-in the areas of housing, This error is not unique to historical explanations.
jobs, schooling, voting rights, and so on-were opened Many people think only in terms of individual actions
to people of allraces. From the Brown vs. Board of Ed­ and have an individualist bias, sometimes called
ucation decision in 7955, to the Civil Rights Act of methodological individualism. This is especially true in
7964, to the Waron Poverty from 7966 to 7968, a the extremely individualistic U.S. culture. The error is
new, dramatic outlook swept the country. This was the that it disregards units of analysis or forces beyond
result of the vision, dedication, and actions of America's the individual. The errorof reductionism shifts expla­
foremost civil rights leader, Dr. Martin Luther King Jr. nation to a much lower unit of analysis. One could
continue to reduce from an individual's behavior to
This says: dependent variable = major change in biological processes in a person, to micro-level neu­
U.S. race relations over a 10- to 13-year period; rochemical activities, to the subatomic level.
independentvariable = King's vision and actions. Most people live in "social worlds" focused on lo­
If you know much about the civil rights era, you cal, immediate settings and their interactions with a
see a problem. The entire civil rights movement and small set of others, so their everyday sense of reality
its successes are attributed to a single individual. Yes, encourages seeing social trends or events as individ­
one individual does make a difference and helps build ual actions or psychological processes. Often, they
and guide a movement, but the movement is missing. become blind to more abstract, macro-level enti­
The idea of a social-political movement as a causal ties-social forces, processes, organizations, institu­
force is reduced to its major leader. The distinct so­ tions, movements, or structures. The idea that all
cial phenomenon-a movement-is obscured. Lost social actions cannot be reduced to individuals alone
are the actions of hundreds of thousands of people is the core of sociology. In his classic work Suicide,
(marches, court cases, speeches, prayer meetings, Emile Durkheim fought methodological individualism
sit-ins, rioting, petitions, beatings, etc.) involved in and demonstrated that larger, unrecognized social
advancing a shared goal and the responses to them. forces explain even highly individual, private actions.

but individual actions alone are not the cause. and a more complex relation exists. Any associ­
Thus, it is likely that a war would have broken ation between two variables might be spurious,
out at about that time even if the assassination so researchers are cautious when they discover
had not occurred. that two variables are associated; upon further
investigation, it may not be the basis for a real
Spuriousness. To call a relationship between causal relationship. It may be an illusion, just
variables spurious means that it is false, a mirage. like the mirage that resembles a pool of water on
Researchers get excited if they think they have a road during a hot day.
found a spurious relationship because they can Spuriousness occurs when two variables ap­
show that what appears on the surface is false pear to be associated but are not causally related

you tell whether a relationship is spurious, and

how do you find out what the mysterious third
factor is? You will need to use statistical tech­
niques (discussed later in this book) to test
For many years, researchers observed a strong posi­ whether an association is spurious. To use them,
tive association between the use of a night-light and you need a theory or at least a guess about possi­
children who were nearsighted. Many thought that ble third factors. Actually, spuriousness is based
the night-light was somehow causing the children to on commonsense logic that you already use. For
develop vision problems (illustrated as a below). example, you already know that there is an asso­
Other researchers could think of no reason for a ciation between the use of air conditioners and
causal link between night-light use and developing
ice cream cone consumption. If you measured
nearsightedness. A 1999 study provided the answer.
the number of air conditioners in use and the
It found that nearsighted parents are more likely to
number of ice cream cones sold for each day, you
use night-lights; they also genetically pass on their vi­
would find a strong correlation, with more cones
sion deficiency to their children. The study found no
link between night-light use and nearsightedness
sold on the days when more air conditioners are
once parental vision was added to the explanation in use. But you know that eating ice cream cones
does not cause people to turn on air condition­
(see b below). Thus the initial causal link was mis­
leading or spurious (from New York Times, May 22,
2001 ).
ers. Instead, both variables are caused by a third
factor: hot days. You could verify the same thing
through statistics by measuring the daily temper­

a. Initial relationship

Night-Light Usage I--.. . . ~ '­ Nearsighted --J

ature as well as ice cream consumption and air
conditioner use. In social research, opposing the­
ories help people figure out which third factors


. . . . . .14


POSITIVE ASSOCIATION are relevant for many topics (e.g., the causes of
crime or the reasons for war or child abuse). 6r...~,..
b. Addition of the missing true causal factor
Example 1. Some people say that taking illegal
Parents Nearsighted drugs causes suicide, school dropouts, and vio­
lent acts. Advocates of "drugs are the problem"
Real Cause
position point to the positive correlations be­
tween taking drugs and being suicidal, dropping
Night-Light Usage ~-------~I Nearsighted ---J out of school, and engaging in violence. They ar­
gue that ending drug use will greatly reduce sui­
cide, dropouts, and violence. Others argue that
many people turn to drugs because of their emo­
tional problems or high levelsof disorder of their
because an unseen third factor is the real cause communities (e.g., high unemployment, unsta­
(see Box 4.9). The unseen third or other variable ble families, high crime, few community services,
is the cause of both the independent and the de­ lack of civility). The people with emotional prob­
pendent variable in the apparent but illusionary lems or who live in disordered communities are
relationship and accounts for the observed asso­ also more likely to commit suicide, drop out, and
ciation. In terms of conditions for causality, the engage in violence. This means that reducing 52.
unseen factor is a more powerful alternative emotional problems and community disorder
explanation. will cause illegal drug use, dropping out, suicide,
You now understand that you should be and violence all to decline greatly. Reducing drug
wary of correlations or associations, but how can taking alone will have only a limited effect be-

cause it ignores the root causes. The "drugs are treated differently because of it, such as having
the problem" argument is spurious because the different job opportunities and housing choices.
initial relationship between taking illegal drugs Discriminated-against people who are in some
and the problems is misleading. The emotional racial categories find limits in their housing
problems and community disorder are the true choices. This means they get separated or
and often unseen causal variables. grouped together in undesirable areas. Poor
housing gets combined with unequal schooling,
Example 2. In the United States and Canada, such that the lowest-quality schools are located
we observe an empirical association between in areas with the least desirable housing. Since
students classifiedas being in a non-White racial the relationship between school quality and test
category and scoring lower academic tests scores is very strong, students from families liv­
(compared to students classifed as in a White ing in less desirable housing areas with low-qual­
category). The relationship between racial classi­ ity schools get lower test scores.
fication and test scores is illusionary, because a We can now turn from the errors in causal
powerful and little-recognized variable is the explanation to avoid and more to other issues
true cause of both the racial classification and involving hypotheses. Table 4.2 provides a re­
the test scores (see Figure 4.4). In this case, the view of the major errors.
true cause operates directly on the independent
variable (racial classification) but indirectly
From the Research Question to
through an intervening process on the depen­
dent variable (test scores). A belief system that is
based on classifyingpeople as belonging to racial It is difficult to move from a broad topic to hy­
groups and assigning great significance to super­ potheses, but the leap from a well-formulated
ficial physical appearance, such as skin color, is research question to hypotheses is a short one.
the basis of what people call "race." Such a belief Hints about hypotheses are embedded within a
system also is the basis for prejudice and dis­ good research question. In addition, hypotheses
criminatory behavior. In such a situation, people are tentative answers to research questions (see
are seen as belonging to different races and Box4.10).

_ _ _ _mlllWltll'Wv::@@@i!ll%lillllllllll';,,"''''''' :< .. ·---W~~$riI

FIG U R E 4.4 Example of a Spurious Relationship between Belonging to a Non­

White "Race" and Getting low Academic Test Scores

Spurious Association
Students treated as belonging to Lower scores on
a racial social category ("White" academic tests
or "Non-White") based on for non-Whites
Discrimination against
superficial physical appearance than for Whites
non-Whites in jobs
and housing


Societywide racist beliefs and Real

Segregated housing
treatment of social categories
as if they had an
inherent-biological basis
Non-Whites attend

lower-quality schools

TAB L E 4.2

Summary of Errors in Explanation

.. c.-....

Ecological Fallacy The empirical observations are at too New York has a high crime rate. Joan
high a level for the causal relationship lives in New York. Therefore, she
that is stated. probably stole my watch.
Reductionism The empirical observations are at too Because Steven lost his job and did not
Iowa level for the causal relationship buy a new car, the country entered a
that is stated. long economic recession.
Spuriousness An unseen third variable is the actual Hair length is associated with TV
cause of both the independent and programs. People with short hair prefer
dependentvariab~. watching football; people with long hair
prefer romance stories. (Unseen: Gender)

Bad Research Questions Good Research Questions

Exploratory Questions
Not Empirically Testable, Nonscientific Questions
• Has the actual incidence of child abuse changed in
• Should abortion be legal? Wisconsin in the past 10 years?
• Is it right to have capital punishment?
Descriptive Questions
General Topics, Not Research Questions • Is child abuse, violent or sexual, more common in
• Treatment of alcohol and drug abuse families that have experienced a divorce than in
• Sexuality and aging intact, never-divorced families?

Set of Variables, Not Questions • Are the children raised in poverty households

more likely to have medical, learning, and social­

• Capital punishment and racial discrimination
emotional adjustment difficulties than non poverty
• Urban decay and gangs children?
Too Vague, Ambiguous Explanatory Questions
• Do police affect delinquency? • Does the emotional instability created by experi­ Study Citatic. ~
• What can be done to prevent child abuse? encing a divorce increase the chances that di­ ASA for1IIiIt slJIeI

vorced parents will physically abuse their

Need to BeStifl More Specific
• Has the incidence of child abuse risen?
• Is a lack of sufficent funds for preventive treat­
• How does poverty affect children?
ment a major cause of more serious medical prob­
• What problems do children who grow up in lems among children raised in families in poverty?
poverty experience that others do not?


Consider an example research question: "Is the chances that the marriage will end in divorce,
age at marriage associated with divorce?" The unless it is a marriage between members of a
question contains two variables: "age at mar­ tight-knit traditional religious community in
riage" and "divorce." To develop a hypothesis, a which early marriage is the norm."
researcher asks, "Which is the independent vari­ Formulating a research question and a hy­
able?" The independent variable is "age at mar­ pothesis do not have to proceed in fixed stages.A
riage" because marriage must logically precede researcher can formulate a tentative research
divorce. The researcher also asks, "What is the question, then develop possible hypotheses; the
direction of the relationship?" The hypothesis hypotheses then help the researcher state the re­
could be: "The lower the age at time of marriage, search question more precisely. The process is
the greater the chances that the marriage will interactive and involves creativity.
end in divorce." This hypothesis answers the re­ You may be wondering: Where does theory
search question and makes a prediction. Notice fit into the process of moving from a topic to a
that the research question can be reformulated hypothesis I can test? Recall from Chapter 2 that
and better focused now: "Are couples who theory takes many forms. Researchers use gen­
marry younger more likely to divorce?" eral theoretical issues as a source of topics. The­
Severalhypotheses can be developed for one ories provide concepts that researchers turn into
research question. Another hypothesis from the variables as well as the reasoning or mechanism
same research question is: "The smaller the dif­ that helps researchers connect variables into a
ference between the ages of the marriage part­ research question. A hypothesis can both answer
ners at the time of marriage, the less likely that a research question and be an untested proposi­
the marriage will end in divorce." In this case, tion from a theory. Researchers can express a hy­
the variable "age at marriage" is specified pothesis at an abstract, conceptual level or
differently. restate it in a more concrete, measurable form.
Hypotheses can specify that a relationship Examples of specific studies may help to il­
holds under some conditions but not others. For lustrate the parts of the research process. For ex­
example, a hypothesis states: "The lower the age amples of three quantitative studies, see Table
of the partners at time of marriage, the greater 4.3; for two qualitative studies, see Table 4.4.

TAB L E 4.3 Examples of Quantitative Studies

Study Citation (using Goar, Carla and Jane Musick, Mark, John Lauzen, Martha M. and
ASA format style) Sell. 2005. "Using Task Wilson, and William David M. Dozier. 2005.
Definition to Modify Bynum. 2000. "Race "Maintaining the Double
Racial Inequality Within and Formal Standard: Portrayals of
Task Groups" Volunteering: The Age and Gender in
Sociological Quarterly Differential Effects of Popular Films." Sex Roles
46:525-543. Class and Religion" 52: 437-446.
Social Forces 78:

-Methodological Experiment Survey Content Analysis


TABLE 4.3 (Continued) TA.L£ 4_:1

Topic Mixed race group Rates of volunteering by Age and Gender Sc

working on a task. A White and Black adults Stereotypes in U.s.
test of "expectation Mass Media
states theory"

Research Question If a group is presented Do different kinds of Do contemporary films

with a task that is resources available to show a double standard, 1'­
complex and requires Blacks and Whites in which males acquire
many diverse skills, does explain why Blacks are greater status and
this result in greater less likely to volunteer? leadership as they age,
equality in participation while females are not
across racial groups permittted to gain
because people believe status and leadership
different racial groups with increased age?
possess different skills?

Main Hypothesis Groups exposed to For Whites and Blacks, As with past popular
Tested instructions that social class and religion U.s. films and in other
suggest complex and affect whether a person popular mass media, a
TA8I.E 4_4
diverse skills are volunteers in different double standard still
required to complete a
task will show less racial
inequality in their
ways. exists.
...... ~

operations to complete
a task than groups
without such

Main Independent Whether groups were Social class, religious The age and gender of
Variable(s) told they were to a attendance, race. major film characters.
complete a complex
task that requires
diverse skills or not. ~

Main Dependent The amount of Whether a person said Whether a character has
Variable(s) time/involvement by he or she volunteered a leadership role, high
people of different for any of five occupational status, and
races to resolve a group organizations (religious, goals. .
task. education, political or
labor, senior citizen, or


Unit of Analysis Mixed race task group Individual adult The movie


TAB L E 4. 3 (Continued)

Specific Units in the 90 undergraduate Random sample of 100 top-grossing

Study females in 3-person 2,867 us. adults domestic u.S. films in
groups comprised of interviewed twice 2002.
one Black and two (panel) in 1986 and
White students. 1989.
Universe All task groups that All adult Whites and All films.
have a diverse set of Blacks in the United
members. States.


TAB L E 4.4 Examples of Qualitative Studies

Study Citation (using ASA Lu, Shun and Gary Fine. 1995. Molotch, Harvey, William
format style) "The Presentation of Ethnic Freudenburg, and Krista Paulsen.
Authenticity: Chinese Food as a 2000. "History Repeats Itself,
Social Accomplishment" but How? City Character, Urban
Sociological Quarterly Tradition, and the
36:535-53. Accomplishment of Place."
American Sociological Review
Methodological Technique Field Research Historical-Comparative Research

Topic The ways ethnic cultures are The ways cities develop a distinct
displayed within the boundaries urban "character,"
of being acceptable in the United
States and how they deploy
cultural resources.
Research Question How do Chinese restaurants Why did the California cities of
present food to balance, givinga Santa Barbara and Ventura,
feeling of cultural authenticity which appear very similar on the
and yet satisfying non-Chinese surface, develop very different
U.S. customers? characters?


TAB L E 4.4 (Continued)

trirwr nr,.


Crounded Theory Ethnic restaurants Americanize

their food to fit local tastes but
The authors used two
concepts-"Iash up" (interaction



also construct an impression of of many factors) and structure OI"-.r. . . .

authenticity. /t is a negotiated (past events create constraints JifaOas~
process of meeting the on subsequent ones)-to
customer's expectations/taste elaborate on character and <iI:ioa. i -J
conventions and the desire for an tradition. Economic, po/itical, IO~~

exotic and authentic eating
cultural, and social factors
combine to create distinct 1DIr:III:'
Q 5
"~ -~
cultural-economic places. Similar
forces can have opposite results
depending on context.

Social Process Restaurants make modifications Conditions in the two cities

to fit available ingredients, their contributed to two different
market niche, and the cultural economic development
and food tastes of local responses to the oil industry and
customers. highway development. The city
of Ventura formed an industria/­
employment base around the oil
industry and encouraged new
highways. The city of Santa
Barbara limited both the oil
industry and highway growth. It
instead focused on creating a
strong tourism industry.

Social Context or Field Site Chinese restaurants, especially The middle part of California's
four in Athens, Georgia. Pacific coast over the past 1 00



. . . . . . . . . .1
• .t*
In this chapter, you encountered the ground­
focus the research. The style that a researcher
uses will depend on the topic he or she selects,
the researcher's purpose and intended use of


. . . . .IIIIIIiI


work to begin a study. You saw how differences study results, the orientation toward social sci­ ' ­ 1

in the qualitative and quantitative styles or ap­ ence that he or she adopts, and the individual re­ .-.i 1tI
proaches to social research direct a researcher to searcher's own assumptions and beliefs.
prepare for a study differently. All social re­ Quantitative researchers take a linear path
searchers narrow their topic into a more specific, and emphasize objectivity. They tend to use ex­
focused research question. The styles of research plicit, standardized procedures and a causal ex­
suggest a different form and sequence of deci­ planation. Their language of variables and
sions, and different answers to when and how to hypotheses is found across many other areas of

science. The process is often deductive with a se­ strengths and limitations of each. The ultimate
quence ofdiscrete steps that precede data collec­ goal is to develop a better understanding and ex­
tion: Narrow the topic to a more focused planation of events in the social world. This
question, transform nebulous theoretical con­ comes from an appreciation of the value that
cepts into more exact variables, and develop one each style has to offer.
or more hypotheses to test. In actual practice, re­
searchers move back and forth, but the general
process flows in a single, linear direction. In ad­
dition, quantitative researchers take special care Key Terms
to avoid logical errors in hypothesis develop­
ment and causal explanation. abstract
Qualitative researchers follow a nonlinear alternative hypothesis
path and emphasize becoming intimate with the attributes
details of a natural setting or a particular cul­ citation
tural-historical context. They use fewer stan­ dependent variable
dardized procedures or explicit steps, and often ecologicalfallacy
devise on-the-spot techniques for one situation first-order interpretation
or study. Their language of cases and contexts hypothesis
directs them to conduct detailed investigations independent variable
of particular cases or processes in their search for intervening variable
authenticity. They rarely separate planning and level of analysis
design decisions into a distinct pre-data collec­ linear research path
tion stage, but continue to develop the study de­ literature review
sign throughout early data collection. The nonlinear research path
inductive qualitative style encourages a slow, null hypothesis
flexible evolution toward a specific focus based reductionism
on a researcher's ongoing learning from the second-order interpretation
data. Grounded theory emerges from the re­ spuriousness
searcher's continuous reflections on the data. third-order interpretation
Too often, the qualitative and quantitative unit of analysis
distinction is overdrawn and presented as a rigid universe
dichotomy. Adherents of one style of social re­ variable
search frequently judge the other style on the ba­
sis of the assumptions and standards of their
own style. The quantitative researcher demands Endnotes
to know the variables used and the hypothesis
tested. The qualitative researcher balks at turn­ 1. For a discussion of the "logic of the disconfirm­
ing humanity into cold numbers. The challenge ing hypothesis," see Singleton and associates
for the well-versed, prudent social researcher is (1988:456---460).
to understand and appreciate each style or ap­ 2. See Bailey (1987:43) for a discussion.
proach on its own terms, and to recognize the



to measure u.IiiI
Qualitative and to measure r.IIII:iII~
measure "'~.
Quantitative Measurement Reportsto~
social reseu~
explanation. ~

Why Measure?
Quantitative and Qualitative Measurement
Parts of the Measurement Process .::emed*-­
Quantitative Conceptualization and Operationalization ~raI.d~
disIinct *p ill
Qualitative Conceptualization and Operationalization
pDor1D cilia
Reliability and Validity
Reliability and Validity in Quantitative Research
cioIl km_"''''
drdr- tiw iIpI~"
Reliability and Validity in Qualitative Research then aaIIr
Relationship between Reliability and Validity .-daua-"
cxptCWII ill
Other Uses of the Terms Reliable and Valid
A Guide to Quantitative Measurement
_ _ 'WIrIY. . iI-."
Levels of Measurement
Specialized Measures: Scales and Indexes
~ ... alii"
Index Construction
The Purpose


Missing Data

.... .. .......
...._- -

Rates and Standardization

iE . .- nn.

The Purpose
** i ' "-..III. . . .

Logic of Scaling
Commonly Used Scales
- - . ... i11
~- t -



creative intelligence. Others suggest more types,

such as social-interpersonal, emotional, body­
You may have heard of the Stanford Binet IQ test kinesthetic, musical, or spatial. If there are many
to measure intelligence,the Index of Dissimilarity forms of intelligence but people narrowly limit
to measure racial segregation, the Poverty Line to measurement to one type, it seriously restricts
measure whether one is poor, or Uniform Crime how schools identify and nurture learning; how
Reports to measure the amount of crime. When larger society evaluates, promotes, and recog­
social researchers test a hypothesis, evaluate an nizes the contributions of people; and how a so­
explanation, provide empirical support for a the­ ciety values diverse human abilities.
ory, or systematically study an applied issue or Likewise, different policymakers and re­
some area of the social world, they measure con­ searchers conceptualize and operationalizepoverty
cepts and variables. How social researchers mea­ differently. How people measure poverty will
sure the numerous aspects of the social world­ determine whether people get assistance from
such as intelligence, segregation, poverty, crime, numerous social programs (e.g., subsidized
self-esteem, political power, alienation, or racial housing, food aid, health care, child care, etc.).
prejudice-is the focus of this chapter. For example, some say that people are poor only
Quantitative researchers are far more con­ if they cannot afford the food required to pre­
cerned about measurement issues than are qual­ vent malnutrition. Others say that people are
itative researchers. They treat measurement as a poor if they have an annual income that is less
distinct step in the research process that occurs than one-half of the average (median) income.
prior to data collection, and have developed spe­ Still others say that people are poor if they earn
cial terminology and techniques for it. Using a below a "living wage" based on a judgment
deductive approach, they begin with a concept about the income needed to meet minimal com­
then create empirical measures that precisely munity standards of health, safety, and decency
and accurately capture it in a form that can be in hygiene, housing, clothing, diet, transporta­
expressed in numbers. tion, and so forth. Decisions about how to con­
Qualitative researchers approach measure­ ceptualize and measure a variable-poverty­
ment very differently. They develop ways to cap­ can greatly influence the daily living conditions
ture and express variable and nonvariable of millions of people.
concepts using various alternatives to numbers.
They often take an inductive approach, so they
measure features of social life as part of a process
that integrates creating new concepts or theories
with measurement. We use many measures in our daily lives. For ex­
How people conceptualize and operational­ ample, this morning I woke up and hopped onto
ize variables can significantly affect social issues a bathroom scale to see how well my diet is
beyond concerns of research methodology. For working. I glanced at a thermometer to find out
example, psychologists debate the meaning and whether to wear a coat. Next, I got into my car
measurement of intelligence. Most intelligence and checked the gas gauge to be sure I could
tests that people use in schools, on job applica­ make it to campus. As I drove, I watched the
tions, and in making statements about racial or speedometer so I would not get a speeding
other inherited superiority measure only ana­ ticket. By8:00 A.M., I had measured weight, tem­
lytic reasoning (i.e., one's capacity to think ab­ perature, gasoline volume, and speed-all mea­
stractly and to infer logically). Yet, many argue sures about the physical world. Such precise,
that there are other types of intelligence in addi­ well-developed measures, which we use in daily
tion to analytic. Some say there is practical and life, are fundamental in the natural sciences.

We also measure the nonphysical world in observed (e.g., attitudes, ideology, divorce rates,
everyday life, but usually in less exact terms. deviance, sex roles, etc.). Like the natural scien­
We are measuring when we say that a restau­ tist who invents indirect measures of the "invis­
rant is excellent, that Pablo is really smart, that ible" objects and forces of the physical world, the
Karen has a negative attitude toward life, that social researcher devises measures for difficult­
Johnson is really prejudiced, or that the movie to-observe aspects of the social world. QvaliMiw
last night had a lot of violence in it. However, bctioRcbta
such everyday judgments as "really preju­ . . ID05l. of
diced" or "a lot of violence" are imprecise, lion.. Tbe cpaI....
vague measures.
Measurement also extends our senses. The

astronomer or biologist uses the telescope or the Both qualitative and quantitative researchers use .tID and 0II:II"­
microscope to extend natural vision. In contrast careful, systematic methods to gather high-qual­ ~ftftllO
to our senses, scientific measurement is more ity data. Yet, differences in the styles of research CIIl the pI'1laS
sensitive, varies less with the specific observer, and the types of data mean they approach the
.and yields more exact information. You recog­ measurement process differently. The two ap­
nize that a thermometer givesmore specific, pre­ proaches to measurement have three distinctions.
cise information about temperature than touch One difference between the two styles in­
can. Likewise, a good bathroom scale gives you volves timing. Quantitative researchers think
more specific, constant, and precise information about variables and convert them into specific .~aR51_"
about the weight of a 5-year-old girl than you get actions during a planning stage that occurs be­ aaJOCqJl.-'
by lifting her and calling her "heavy" or "light." fore and separate from gathering or analyzing a IleChniqIE. •
Social measures provide precise information data. Measurement for qualitative researchers ~~OI''''~
about social reality.
In addition, measurement helps us observe
what is otherwise invisible. Measurement ex­
occurs during the data collection process.
A second difference involves the data itself.
Quantitative researchers develop techniques that
Waw .....
tends human senses. It lets us observe things that can produce quantitative data (i.e., data in the . . . . ",,- iif'
were once unseen and unknown but were pre­ form of numbers). Thus, the researcher moves Q 5. ..
dicted by theory.
Before you can measure, you need a clear
idea about what you are interested in. For exam­
from abstract ideas to specific data collection
techniques to precise numerical information
produced by the techniques. The numerical in­

~~ .....

ple, you cannot see or feel magnetism with your

natural senses. Magnetism comes from a theory
formation is an empirical representation of the
abstract ideas. Data for qualitative researchers
about the physical world. You observe its effects sometimes is in the form of numbers; more of­
indirectly; for instance, metal flecks move near a ten, it includes written or spoken words, actions,
magnet. The magnet allows you to "see" or mea­ sounds, symbols, physical objects, or visual im­
sure the magnetic fields. Natural scientists have ages (e.g., maps, photographs, videos, etc.). The
invented thousands of measures to "see" very qualitative researcher does not convert all obser­
tiny things (molecules or insect organs) or very vation into a single medium such as numbers.
large things (huge geological land masses or Instead, he or she develops many flexible, ongo­
planets) that are not observable through ordi­ ing processes to measure that leaves the data in
nary senses. In addition, researchers are con­ various shapes, sizes, and forms.
stantly creating new measures. All researchers combine ideas and data to
Some of the things a social researcher is in­ analyze the social world. In both research styles,
terested in measuring are easy to see (e.g., age, data are empirical representations of concepts,
sex, skin color, etc.), but most cannot be directly and measurement links data to concepts. A third

difference is how the two styles make such link­ definition. A conceptual definition is a definition
ages. Quantitative researchers contemplate and in abstract, theoretical terms. It refers to other
reflect on concepts before they gather any data. ideas or constructs. There is no magical way to
They construct measurement techniques that turn a construct into a precise conceptual defin­
bridge concepts and data. ition. It involves thinking carefully, observing
Qualitative researchers also reflect on ideas directly, consulting with others, reading what
before data collection, but they develop many, if others have said, and trying possible definitions.
not most, of their concepts during data collec­ How might I develop a conceptual defini­
tion. The qualitative researcher reexamines and tion of the construct prejudice? When beginning
evaluates the data and concepts simultaneously to develop a conceptual definition, researchers
and interactively. Researchers start gathering often rely on multiple sources-personal experi­
data and creating ways to measure based what ence and deep thinking, discussions with other
they encounter. As they gather data, they reflect people, and the existing scholarly literature. I
on the process and develop new ideas. might reflect on what I know about prejudice,
ask others what they think about it, and go the li­
brary and look up its many definitions. As I
gather definitions, the core idea should get
clearer, but I have many definitions and need to
sort them out. Most definitions state that preju­
When a researcher measures, he or she links dice is an attitude about another group and in­
a concept, idea, or construct 1 to a measure (i.e., volves a prejudgment, or judging prior to getting
a technique, a process, a procedure, etc.) by specific information.
which he or she can observe the idea empirically. As I think about the construct, I notice that
Quantitative researchers primarily follow a de­ all the definitions refer to prejudice as an atti­
ductive route. They begin with the abstract idea, tude, and usually it is an attitude about the

follow with a measurement procedure, and end
with empirical data that represent the ideas.
Qualitative researchers primarily follow an in­
ductive route. They begin with empirical data,
members of another group. There are many
forms of prejudice, but most are negative views
about persons of a different racial-ethnic group.
Prejudice could be about other kinds of groups
follow with abstract ideas, relate ideas and data, (e.g.,people of a religion, ofa physical stature, or
and end with a mixture of ideas and data. Actu­ from a certain region), but it is always about a
ally,the process is more interactive in both styles collectivity to which one does not belong. Many
of research. As a quantitative researcher devel­ constructs have multiple dimensions or types, so
ops measures, the constructs become refined I should consider whether there can be different
and clearer, and as the researcher applies the types of prejudice-racial prejudice, religious
measures to gather data, he or she often adjusts prejudice, age prejudice, gender prejudice, na­
the measurement technique. As a qualitative re­ tion prejudice, and so forth.
searcher gathers data, he or she uses some preex­ I also need to consider the units of analysis
isting ideas to assist in data collection, and will that best fit my definition of the construct. Prej­
then mix old with new ideas that are developed udice is an attitude. Individuals hold and express
from the data. attitudes, but so might groups (e.g., families,
Both qualitative and quantitative researchers clubs, churches, companies, media outlets). I
use two processes: conceptualization and opera­ need to decide, Do I want my definition of prej­
tionalization in measurement. Conceptualizati­ udice to include only the attitudes of individuals
on is the process of taking a construct and or should it include attitudes held by groups, or­
refining it by giving it a conceptual or theoretical ganizations, and institutions as well? Can I say,

The school or newspaper was prejudiced? I also conceptualizati

must distinguish my construct from closely re­ ization, followe
lated ones. For example, I must ask, How is prej­ definition or D
udice similar to or different from ideas such as Quantitative res
discrimination, stereotype, or racism? 1. Remember the conceptual definition. The underly­ to rigorously lin!
Conceptualization is the process of carefully ing principle for any measure is to match it to procedures that
thinking through the meaning of a construct. At the specific conceptual definition of the con­ tive information
this stage, I believe that prejudice means an in­ struct that will be used in the study.
Figure 5.1
flexible negative attitude that an individual holds 2. Keep an open mind. Do not get locked into a sin­ process for two 1
and is directed toward a race or ethnic group gle measureor type of measure. Becreativeand in a theory and ..
that is an out-group. It can, but does not always, constantly lookfor better measures. els to consider. Q
lead to behavior, such as treating people un­ 3. Borrow from others. Do not be afraid to borrow pirical. At the mj
equally (i.e., discrimination), and it generally re­ from other researchers, as longas credit isgiven. is interested in 1
lies on a person's stereotypes of out-group Good ideas for measurescan be found in other two constructs.s
members. Thus, my initial thought, "Prejudice is studies or modified from other measures. level of operatiol
a negative feeling," has become a precisely de­ 4. Anticipate difficulties. Logical and practical prob­ interested in td
fined construct. Even with all my conceptualiza­ lems often arise when trying to measure vari­ determine the dlI
tion, I need to be even more specific. For ables of interest. Sometimes a problem can be dicators. This is I
example, if prejudice is a negative attitude about anticipated and avoided with careful fore­ statistics, questilll
a race or an ethnic group of which one is not a thought and planning. The third level.
member, I need to consider the meaning of race 5. Donot forget yourunitsof analysis. Yourmeasure the operational
or ethnic group. I should not assume everyone should fit with the units of analysis of the study questionnaires I I
sees racial-ethnic categories the same. Likewise, and permit you to generalize to the universe of struct (e.g., rac:ioII
it is possible to have a positive prejudgment, and interest. ture what happd
if so is that a kind of prejudice? The main point and relate it to ..
is that conceptualization requires that I become
very clear and state what I mean very explicitly
for others to see. etc.), and to the research techniques you know or
Operationalization links a conceptual defin­ can learn. You can develop a new measure from
ition to a specific set of measurement techniques scratch, or it can be a measure that is already be­
or procedures, the construct's operational defin­ ing used by other researchers (see Box 5.1). FIGURE 5.1
ition (i.e., a definition in terms of the specific op­ Operationalization links the language of
erations of actions a researcher carries out). An theory with the language of empirical measures.
operational definition could be a survey ques­ Theory is full of abstract concepts, assumptions,
tionnaire, a method of observing events in a field relationships, definitions, and causality. Empiri­
setting, a way to measure symbolic content in cal measures describe how people concretely
the mass media, or any process carried out by measure specific variables. They refer to specific
the researcher that reflects, documents, or repre­ operations or things people use to indicate the Concepb.lafi;aj
sents the abstract construct as it is expressed in presence of a construct that exists in observable I
the conceptual definition. reality.
There are usually multiple ways to measure a
construct. Some are better or worse and more or Quantitative Conceptualization and
less practical than others. The key is to fit your
measure to your specificconceptual definition, to
the practical constraints within which you must The measurement process for quantitative re­
operate (e.g., time, money, available subjects, . search flows in a straightforward sequence: first

conceptualization, followed by operational­ The measurement process links together

ization, followed by applying the operational the three levels, moving deductively from the
definition or measuring to collect the data. abstract to the concrete. A researcher first con­

Quantitative researchers developed several ways ceptualizes a variable, giving it a clear concep­
to rigorously link abstract ideas to measurement tual definition. Next, he or she operationalizes it
procedures that will produce precise quantita­ by developing an operational definition or set
tive information about empirical reality. of indicators for it. Last, he or she applies the
,- Figure 5.1 illustrates the measurement indicators in the empirical world. The links
process for two variables that are linked together from abstract constructs to empirical reality al­
in a theory and a hypothesis. There are three lev­ low the researcher to test empirical hypotheses.
els to consider: conceptual, operational, and em­ Those tests are logically linked back to a con­
pirical. At the most abstract level, the researcher ceptual hypothesis and causal relations in the
is interested in the causal relationship between world of theory.
two constructs, or a conceptual hypothesis. At the A hypothesis has at least two variables, and
level of operational definitions, the researcher is the processes of conceptualization and opera­
interested in testing an empirical hypothesis to tionalization are necessary for each variable. In
determine the degree of association between in­ the preceding example, prejudice is not a
dicators. This is the level at which correlations, hypothesis. It is one variable. It could be a de­
statistics, questionnaires, and the like are used. pendent variable caused by something else, or it
The third levelis the concrete empirical world. If co.uld be an independent variable causing
the operational indicators of variables (e.g., something else. It depends on my theoretical
questionnaires) are logically linked to a con­ explanation.
struct (e.g., racial discrimination), they will cap­ We can return to the quantitative study by
ture what happens in the empirical social world Weitzer and Tuch on perceptions of police bias
and relate it to the conceptual level. and misconduct discussed in Chapter 2 for an

FIGURE 5.1 Conceptualization and Operationalization

Abstract Construct to Concrete Measure

Independent Variable Dependent Variable

othetical Causal


Conceptual Definition

Conceptual Definition
} Level of

} Operational

Tested Empirical
Hypothesis } Empirical

example of how researchers conceptualize and sense" or organize the data and one's prelimi­ Just as qlWJli
operationalize variables. It is an explanatory nary ideas. ates from a rizid 4
study with two main variables in a causal hy­ As the researcher gathers and analyzes qual­ followed by ~
pothesis. The researchers began with the itative data, he or she develops new concepts, tual interaction. 1
conceptual hypothesis: Members of a nondomi­ formulates definitions for the concepts, and from beyond thd
nant racial group are more likely than a domi­ considers relationships among the concepts. ting. Qualitaliw:l
nant racial group to believe that policing is Eventually, he or she links concepts to one an­ howthe~
racially biased, and their experience with polic­ other to create theoretical relationships that may the researcher's.
ing and exposure to media reports on police or may not be causal. Qualitative researchers and CODcep6
racial bias increase the perception of racial bias. form the concepts as they examine their qualita­ emerged ciuriIIc
They conceptualized the independent variable, tive data (i.e., field notes, photos and maps.his­ qualitatiw
dominant racial group, as White and the non­ torical documents, etc.). Often, this involves a tually intel1l:kpc-t
dominant group as non-White subdivided into researcher asking theoretical questions about the '\Tecan~
Black and Hispanic. The researchers conceptual­ data (e.g., Is this a case of class conflict? What is tionalizatioo ill
ized the dependent variable, racially biased polic­ the sequence of events and could it be different?
of law bv Ed~"
ing, as unequal treatment by the police ofWhites Why did this happen here and not somewhere cussed in ......_----..0
and non-Whites and racial prejudice by police else?). de\-doped~
officers. The researchers operationalized the in­ A qualitative researcher conceptualizes by began ""ith ~ ­
dependent variable by self-identification to a
survey question about race. They operationalized
developing clear, explicit definitions of con­
structs. The definitions are somewhat abstract
r.nions CIIIa".
bte I~.
the dependent variable by using four sets ofsur­ and linked to other ideas, but usually they are tiutfums_­
vey questions: (l) questions about whether po­ also closely tied to specific data. They can be ex­ *¥",~mr -­

lice treat Blacks better, the same, or worse than pressed in the words and concrete actions of the .ooIies-.l
Whites, and the same question with Hispanics people being studied. In qualitative research,
substituted for Blacks; (2) questions about conceptualization is largely determined by the en -

whether police treat Black neighbhorhoods bet­ data. ~a.p. . .~..

ter' the same, or worse than Whites ones, with oiL. d"Id:t-'IS••14
the same question asked for Hispanic neighbor­ Operationalization. The operationalization

hoods; (3) a question about whether there is process for qualitative research significantly
racial-ethnic prejudice among police officers in differs from that in quantitative research and -~
the city; and (4) a question about whether police often precedes conceptualization. A researcher
are more likely to stop some drivers because they forms conceptual definitions out of rudimen­
are Black or Hispanic. tary "working ideas" that he or she used while
making observations or gathering data. Instead
Qualitative Conceptualization and of turning refined conceptual definitions into a
set of measurement operations, a qualitative re­
searcher operationalizes by describing how
6 S
Conceptualization. The conceptualization specific observations and thoughts about the
process in qualitative research also differs from data contributed to working ideas that are the
that in quantitative research. Instead of refining basis of conceptual definitions and theoretical
abstract ideas into theoretical definitions early in concepts.
the research process, qualitative researchers re­ Operationalization in qualitative research is
fine rudimentary "working ideas" during the an after-the-fact description more than a before­
data collection and analysis process. Conceptu­ the-fact preplanned technique. Almost in a re­
alization is a process of forming coherent theo­ verse of the quantitative process, data gathering
retical definitions as one struggles to "make occurs with or prior to full operationalization.

Just as quantitative operationalization devi­ sistance to reformulation to acceptance, and

ates from a rigid deductive process, the process with acceptance came new corporate policy. The
followed by qualitative researchers is one of mu­ researchers also drew on past studies to argue
tual interaction. The researcher draws on ideas that the "managerialization of law" illustrates
from beyond the data of a specific research set­ one role of top corporate managers-they inno­
ting. Qualitative operationalization describes vate and alter internal operations by creating
how the researcher collects data, but it includes new terms, justifications, and maneuvers that
the researcher's use of preexisting techniques help firms adjust to potential "disruptions" and
and concepts that were blended with those that requirements originating in the corporation's
emerged during the data collection process. In external political-legal environment.
qualitative research, ideas and evidence are mu­
tually interdependent.
We can see an example ofqualitative opera­
tionalization in the study on managerialization
of law by Edelman and associates (2001) dis­ Reliability and validity are central issues in all
cussed in Chapter 2. It is a descriptive study that measurement. Both concern how concrete mea­
developed one main construct. The researchers sures are connected to constructs. Reliability
began with an interest in how major U.S. corpo­ and validity are salient because constructs in so­
rations came to accept legal mandates from the cial theory are often ambiguous, diffuse, and
late 1970s to early 1990s. The mandates stated not directly observable. Perfect reliability and
that firms must institute policies to equalize and validity are virtually impossible to achieve.
improve the hiring and promotion of racial mi­ Rather, they are ideals for which researchers
norities and women, something the firms ini­ strive.
tially opposed. The researcher's empirical data All social researchers want their measures to
consisted of articles in magazines written for and be reliable and valid. Both ideas are important in
by corporate managers, or "managerial rhetoric" establishing the truthfulness, credibility, or be­
(i.e., debates and discussion within the commu­ lievability offindings. Both terms also have mul­
nity ofleading professional managers on impor­ tiple meanings. Here, they refer to related,
tant issues). After gathering numerous articles, desirable aspects of measurement.
the researchers operationalized the data by devel­ Reliability means dependability or consis­
oping working ideas and concepts from an in­ tency. It suggests that the same thing is repeated
ductive examination of the data. The researchers or recurs under the identical or very similar con­
discovered that as managers discussed and de­ ditions. The opposite of reliability is a measure­
liberated, they had created a set of new nonlegal ment that yields erratic, unstable, or inconsistent
terms, ideas, and justifications. The operational­ results.
ization moved inductively from looking at arti­ Validity suggests truthfulness and refers to
cles to creating working ideas based on what the match between a construct, or the way a re­
researchers found in the rhetoric. The researchers searcher conceptualizes the idea in a conceptual
conceptualized their working ideas into the ab­ definition, and a measure. It refers to how well
stract construct "managerialization of law." The an idea about reality "fits" with actual reality.
researchers saw that that corporate managers The absence of validity occurs if there is poor fit
had altered and reformulated the original legal between the constructs a researcher uses to de­
terms and mandates, and created new ones that scribe, theorize, or analyze the social world and
were more consistent with the values and views what actually occurs in the social world. In sim­
of major corporations. The researchers docu­ ple terms, validity addresses the question of how
mented a historical process that moved from re­ well the social reality being measured through

research matches with the constructs researchers Otherwise, it is impossible to determine which
use to understand it. concept is being "indicated." For example, the FIGURE 5.Z
Qualitative and quantitative researchers indicator of a pure chemical compound is more
want reliable and valid measurement, but be­ reliable than one in which the chemical is mixed A
yond an agreement on the basic ideas at a gen­ with other material or dirt. In the latter case, it is
eral level, each style sees the specifics of
reliability and validity in the research process
difficult to separate the "noise" of other material
from the pure chemical.
V. . . . . .
Increase the Level of Measurement. Levels of
Reliability and Validity in
measurement are discussed later. Indicators at
higher or more precise levels of measurement
Quantitative Research
are more likely to be reliable than less precise
Reliability. As just stated, reliability means de­ measures because the latter pick up less detailed
pendability. It means that the numerical results information. If more specific information is
produced by an indicator do not vary because of measured, then it is less likely that anything
characteristics of the measurement process or other than the construct will be captured. The
measurement instrument itself. For example, I
get on my bathroom scale and read my weight. I
get off and get on again and again. I have a reli­
able scale if it gives me the same weight each
general principle is: Try to measure at the most
precise level possible. However, it is more diffi­
cult to measure at higher levelsof measurement.
For example, if I have a choice of measuring
ota d:dia~
time-assuming, ofcourse, that I am not eating, prejudice as either high or low, or in 10 cate­ mapau..c..1
drinking, changing clothing, and so forth. An gories from extremely low to extremely high, it read~~
unreliable scale will register different weights would be better to measure it in 10 refined ~Iq.1Od5­
each time, even though my "true" weight does categories. ~ wwl'_lI:a"
not change. Another example is my car
speedometer. IfI am driving at a constant slow Use Multiple Indicators of a Variable. A third .:omn~
speed on a level surface, but the speedometer way to increase reliability is to use multiple indi­ iti.:atiom.. ... I
needle jumps from one end to the other, my cators, because two (or more) indicators of the or-edmiL!IWIP
speedometer is not a reliable indicator of how same construct are better than one. Figure 5.2 il­ ~dcridrs_
fast I am traveling. lustrates the use of multiple indicators in hy­ ~OIl_<III'I"'.
pothesis testing. Three indicators of the one ~(uItipIr ­
How to Improve Reliability. It is rare to have independent variable construct are combined
perfect reliability. There are four ways to in­ into an overall measure,A, and two indicators of ataOWl, '
crease the reliability of measures: (1) clearlycon­ a dependent variable are combined into a single the COII5I:IWI
ceptualize constructs, (2) use a precise level of measure, B. own uin..:
measurement, (3) use multiple indicators, and For example, I create three indicators of the ,;pJi"Stioo . . ~
(4) use pilot-tests. variable, racial-ethnic prejudice. My first indica­ but~_...
tor is an attitude question on a survey. I ask re­ ~ .S!'5I .
Clearly ConceptualizeAll Constructs. Reliability search participants their beliefs and feelings lDieaSWak..t.
increases when a single construct or subdimen­ about many different racial and ethnic groups.
sian of a construct is measured. This means For a second indicator, I observe research par­
developing unambiguous, clear theoretical defi­ ticipants from various races and ethnic groups
nitions. Constructs should be specified to elimi­ interacting together over the course of three
nate "noise" (i.e., distracting or interfering days. I look for those who regularly either (1)
information) from other constructs. Each mea­ avoid eye contact, appear to be tense, and sound
sure should indicate one and only one concept. cool and distant; or (2) make eye contact, appear


FIG U RES. 2 Measurement Using Multiple Indicators


Variable Measure
Variable Measure

Specific Indicators Specific Indicators

relaxed, and sound warm and friendly as they in­ a hypothesis-testing situation. This takes more
teract with people of their same or with people time and effort.
of a different racial-ethnic group. Last, I create The principle of using pilot-tests extends to
an experiment. I ask research participants to replicating the measures other researchers have
read the grade transcripts, resumes, and inter­ used. For example, I search the literature and find
view reports on 30 applicants for five jobs­ measures of prejudice from past research. I may
youth volunteer coordinator, office manager, want to build on and use a previous measure if it
janitor, clothing store clerk, and advertising ac­ is a good one, citing the source, of course. In ad­
count executive. The applicants have many qual­ dition, I may want to add new indicators and
ifications, but I secretly manipulate their racial compare them to the previous measure.
or ethnic group to see whether a research partic­
ipant decides on the best applicant for the jobs Validity. Validity is an overused term. Some­
based on an applicant's race and ethnicity. times, it is used to mean "true" or "correct."
Multiple indicators let a researcher take There are several general types of validity. Here,
measurements from a wider range of the content we are concerned with measurement validity.
of a conceptual definition. Different aspects of There are also several types of measurement va­
the construct can be measured, each with its lidity. Nonmeasurement types of validity are dis­
own indicator. Also, one indicator (e.g., one cussed later.
question on a questionnaire) may be imperfect, When a researcher says that an indicator is
but several measures are less likely to have the valid, it is valid for a particular purpose and def­
same (systematic) error. Multiple indicator inition. The same indicator can be valid for one
measures tend to be more stable than measures purpose (i.e., a research question with units of
with one item. analysis and universe) but less valid or invalid
for others. For example, the measure of preju­
Use Pretests, Pilot Studies, and Replication. dice discussed here might be valid for measuring
Reliability can be improved by using a pretest or prejudice among teachers but invalid for mea­
pilot version of a measure first. Develop one or suring the prejudice of police officers.
more draft or preliminary versions of a measure At its core, measurement validity refers to
and try them before applying the final version in how well the conceptual and operational defini­

tions mesh with each other. The better the fit, t-Content Validity. Content validity is a special similar construe
the greater the measurement validity. Validity is type of face validity. It addresses the question, Is similar results,
more difficult to achieve than reliability. We the full content of a definition represented in a
cannot have absolute confidence about validity, measure? A conceptual definition holds ideas; it
but some measures are more valid than others. is a "space" containing ideas and concepts. Mea­
The reason we can never achieve absolute valid­ sures should represent all ideas or areas in the
ity is that constructs are abstract ideas, whereas conceptual space. Content validity involves
indicators refer to concrete observation. This is three steps. First, specify fully the entire content
the gap between our mental pictures about the in a construct's definition. Next, sample from all
world and the specific things we do at particular areas of the definition. Finally, develop an indi­
times and places. Validity is part of a dynamic cator that taps all of the parts of the definition.
process that grows by accumulating evidence An example of content validity is my defin­
over time. Without it, all measurement becomes ition offeminism as a person's commitment to a
meaningless. set of beliefs creating full equality between men
and women in areas of the arts, intellectual pur­
Three Types ofMeasurement Validity suits, family, work, politics, and authority rela­
Face Validity. The easiest to achieve and the tions. I create a measure of feminism in which I
most basic kind of validity is face validity. It is a ask two survey questions: (1) Should men and
judgment by the scientific community that the women get equal pay for equal work and (2)
indicator really measures the construct. It ad­ Should men and women share household tasks?
dresses the question, On the face of it, do people My measure has low content validity because the
believe that the definition and method of mea­ two questions ask only about pay and household
surement fit? It is a consensus method. For ex­ tasks. They ignore the other areas (intellectual
ample, few people would accept a measure of pursuits, politics, authority relations, and other
college student math ability using a question aspects of work and family). For a content-valid
that asked students: 2 + 2 = ? This is not a valid measure, I must either expand the measure or
measure of college-level math ability on the face narrow the definition.
of it. Recall that in the scientific community, as­ Criterion Validity. Criterion validity uses some
pects of research are scrutinized by others. See standard or criterion to indicate a construct ac­
Table 5.1 for a summary of types of measure­ curately. The validity of an indicator is verified
ment validity. by comparing it with another measure of the
same construct that is widely accepted. There are


TAB L E 5. 1 Summary of Measurement

two subtypes of this kind of validity.
Concurrent Validity. To have concurrent valid­
Validity Types ity, an indicator must be associated with a preex­
isting indicator that is judged to be valid (i.e., it
has face validity). For example, you create a new
test to measure intelligence. For it to be concur­
Face-in the judgment of others rently valid, it should be highly associated with
Content-captures the entire meaning existing IQ tests (assuming the same definition
of intelligence is used). This means that most
Criterion-agrees with an external source
people who score high on the old measure
• Concurrent-agrees with a preexisting measure should also score high on the new one, and vice
• Predictive-agrees with future behavior versa. The two measures may not be perfectly
associated, but if they measure the same or a

similar construct, it is logical for them to yield qualitative researchers apply the principles dif­
similar results. ferently in practice.

Predictive Validity. Criterion validity whereby Reliability. Reliability means dependability or

an indicator predicts future events that are log­ consistency. Qualitative researchers use a variety
ically related to a construct is called predictive of techniques (e.g., interviews, participation,
validity. It cannot be used for all measures. The photographs, document studies, etc.) to record
measure and the action predicted must be dis­ their observations consistently. Qualitative re­
tinct from but indicate the same construct. Pre­ searchers want to be consistent (i.e., not vacillat­
dictive measurement validity should not be ing and erratic) in how, over time, they make
confused with prediction in hypothesis testing, observations, similar to the idea of stability reli­
where one variable predicts a different variable ability. One difficulty is that they often study
in the future. For example, the Scholastic As­ processes that are not stable over time. More­
sessment Test (SAT) that many U.S. high over, they emphasize the value of a changing or
school students take measures scholastic apti­ developing interaction between the researcher
tude-the ability ofa student to perform in col­ and what he or she studies.
lege. If the SAT has high predictive validity, Qualitative researchers believe that the sub­
then students who get high SAT scores will sub­ ject matter and a researcher's relationship to it
sequently do well in college. If students with should be a growing, evolving process. The
high scores perform the same as students with metaphor for the relationship between a re­
average or low scores, then the SAT has low searcher and the data is one of an evolving rela­
predictive validity. tionship or living organism (e.g., a plant) that
Another way to test predictive validity is to naturally matures. Most qualitative researchers
select a group of people who have specific char­ resist the quantitative approach to reliability,
acteristics and predict how they will score (very which they see as a cold, fixed mechanical in­
high or very low) vis-a-vis the construct. For ex­ strument that one repeatedly injects into or ap­
ample, I have a measure of political conser­ plies to some static, lifelessmaterial.
vatism. I predict that members of conservative Qualitative researchers consider a range of
groups (e.g., John Birch Society, Conservative data sources and employ multiple measurement
Caucus, Daughters of the American Revolution, methods. They accept that different researchers
Moral Majority) will score high on it, whereas or that researchers using alternative measures
members ofliberal groups (e.g., Democratic So­ will get distinctive results. This is because quali­
cialists, People for the American Way, Ameri­ tative researchers see data collection as an inter­
cans for Democratic Action) will score low. I active process in which particular researchers
"validate" the measure with the groups-that is, operate in an evolving setting and the setting's
I pilot-test it by using it on members of the context dictates using a unique mix of measures
groups. It can then be used as a measure of po­ that cannot be repeated. The diverse measures
litical conservatism for the general public. and interactions with different researchers are
beneficial because they can illuminate different
Reliability and Validity in Qualitative
facets or dimensions of a subject matter. Many
qualitative researchers question the quantitative
researcher's quest for standard, fixed measures.
Most qualitative researchers accept the princi­ They fear that such measures ignore the benefits
ples of reliability and validity, but use the terms of having a variety of researchers with many ap­
infrequently because of their close association proaches and may neglect key aspects of diver­
with quantitative measurement. In addition, sity that exist in the social world.

Validity. Validity means truthful. It refers to Relationship between Reliability 0'M!!1iia . .

the bridge between a construct and the data. and Validity ~
Qualitative researchers are more interested in nee '"ft5L
authenticity than validity. Authenticity means Reliabilityis necessary for validity and is easier to ~~dI
giving a fair, honest, and balanced account of achieve than validity. Although reliability is nec­ tioo.. &a_,,­
social life from the viewpoint of someone who essary in order to have a valid measure of a con­
lives it everyday. Qualitative researchers are cept, it does not guarantee that a measure will be
less concerned with trying to match an ab­ valid. It is not a sufficient condition for validity.
stract concept to empirical data and more A measure can produce the same result over and
concerned with giving a candid portrayal of over (i.e., it has reliability), but what it measures
social life that is true to the experiences of may not match the definition of the construct
people being studied. Most qualitative re­ (i.e., validity).
searchers concentrate on ways to capture an A measure can be reliable but invalid. For
inside view and provide a detailed account of example, I get on a scale and get weighed. The ~.l.. HitI"'1I
how those being studied feel about and under­ weight registered by the scale is the same each JIIft~
stand events. time I get on and off. But then I go to another ot~ . . . . .
Qualitative researchers have developed sev­
eral methods that serve as substitutes for the
quantitative approach to validity. These empha­
size conveying the insider's view to others. His­
torical researchers use internal and external
scale-an "official" one that measures true
weight-and it says that my weight is twice as
great. The first scale yielded reliable (i.e., de­
pendable and consistent) results, but it did not
give a valid measure of my weight.
... ...



~ 1"dilI ,...
criticisms to determine whether the evidence A diagram might help you see the relation­
they have is real or they believe it to be. Qualita­ ship between reliability and validity. Figure 5.3 1IIM:s 1IIr
tive researchers adhere to the core principle of illustrates the relationship between the concepts lks.-r......
validity, to be truthful (i.e., avoid false or dis­ by using the analogy of a target. The bull's-eye
torted accounts). They try to create a tight fit represents a fit between a measure and the defi­
between their understanding, ideas, and state­ nition of the construct.
ments about the social world and what is actu­ Validity and reliability are usually comple­
ally occurring in it. mentary concepts, but in some situations they


FIG U R E 5.3 Illustration of Relationship between Reliability and Validity

A BUll's-Eye = A Perfect Measure

, 7




Low Reliability High Reliability High Reliability

. .. - I ·
and Low Validity but Low Validity and High Validity

Source: Adapted from Babbie (2004:145). .. -"iii"


conflict with each other. Sometimes, as validity External Validity. Externalvalidity is used pri­
increases, reliabilityis more difficult to attain, and marily in experimental research. It is the ability
vice versa. This occurs when the construct has a to generalize findings from a specific setting and
highly abstract and not easily observable defini­ small group to a broad range of settings and peo­

I tion. Reliability is easiest to achieve when the

measure is precise and observable. Thus, there is a
strain between the true essence of the highly ab­
stract construct and measuring it in a concrete
ple. It addresses the question, If something hap­
pens in a laboratory or among a particular group
of subjects (e.g., college students), can the find­
ings be generalized to the "real" (nonlaboratory)
manner. For example, "alienation" is a very ab­ world or to the general public (nonstudents)?
stract, highly subjective construct, often defined High external validity means that the results can
as a deep inner sense of loss of one's humanity be generalized to many situations and many
that diffusesacross many aspects of one's life (e.g., groups of people. Low external validity means
t social relations, sense of self, orientation toward
nature). Highly precise questions in a question­
that the results apply only to a very specificsetting.

naire give reliable measures, but there is a danger Statistical Validity. Statistical validity means
oflosing the subjective essence of the concept. that the correct statistical procedure is chosen
and its assumptions are fully met. Different sta­
Other Uses of the Terms Reliable tistical tests or procedures are appropriate for
and Valid different conditions, which are discussed in text­
books that describe the statistical procedures.
Many words have multiple definitions, includ­
All statistics are based on assumptions about
ing reliability and validity. This creates confusion
the mathematical properties of the numbers be­
unless we distinguish among alternative uses of
ing used. A statistic will be invalid and its results
the same word.
nonsense if the major assumptions are violated.
Reliability. We use reliability in everyday lan­ For example, to compute an average (actually the
guage. A reliable person is one who is depend­ mean, which is discussed in a later chapter), one
able, stable, and responsible; a reliable car is cannot use information at the nominal level of
dependable and trustworthy. This means the measurement (to be discussed). For example,
person responds in similar, predictable ways in suppose I measure the race of a class of students.
different times and conditions; the same can be I give each race a number: White > 1, African
said for the car. In addition to measurement re­ American> 2, Asian > 3, others =: 4. It makes no
liability, researchers sometimes say a study or its sense to say that the "mean" race of a class of stu­
results are reliable. By this, they mean that the dents is 1.9 (almost African American?). This is a
method of conducting a study or the results misuse of the statistical procedure, and the re­
from it can be reproduced or replicated by other sults are invalid even if the computation is cor­
researchers. rect. The degree to which statistical assumptions
can be violated or bent (the technical term is
Internal Validity. Internal validity means robustness) is a topic in which professional statis­
there are no errors internal to the design of the ticians take great interest.
research project. It is used primarily in experi­
mental research to talk about possible errors or
alternative explanations of results that arise de­
spite attempts to institute controls. High inter­
nal validity means there are few such errors. Low
internal validity means that such errors are Thus far, you have learned about the principles
likely. of measurement, including the principles of reli­

ability and validity. Quantitative researchers have Four Levels ofMeasurement with less preosa
developed ideas and specialized measures to help Precision and Levels. The idea oflevels of mea­ tion and then l1li
them in the process of creating operational defi­ surement expands on the difference between
nitions that will be reliable and valid measures continuous and discrete variables and organizes Distinguishing olI
and yield numerical data for their variable con­ types of variables for their use in statistics. The levels from 10\<111:
structs. This section ofthe chapter is a brief guide four levels of measurement categorize the degree sion are noJ:lliDl
to these ideas and a few of the measures. of precision of measurement. Each level gives ,
Deciding on the appropriate level of mea­ (see Table 5.~ l.ll
surement for a construct often creates confu­ that there is a dill
Levels of Measurement
sion. The appropriate level of measurement for a religion: Prot~
Levels of measurement is an abstract but impor­ variable depends on two things: (1) how a con­ racial heritage: .4
tant and widely used idea. Basically, it says that struct is conceptualized and (2) the type of indi­ panic, other). 01
some ways a researcher measures a construct are cator or measurement that a researcher uses. ference, plus tiwl
at a higher or more refined level, and others are The way a researcher conceptualizes a ranked (e.g., letre
crude or less precisely specified. The level of construct can limit how precisely it can be mea­ measures: Stro~
measurement depends on the way in which a sured. For example, some of the variables listed Strongly DisagmI:;
construct is conceptualized-that is, assump­
tions about whether it has particular characteris­
earlier as continuous can be reconceptualized as
discrete. Temperature can be a continuous vari­
the first two
tics. The level of measurement affects the kinds able (e.g., degrees, fractions of degrees) or it can Celsius teIl1J>eT3t'
of indicators chosen and is tied to basic assump­ be crudely measured with discrete categories no, 125). AlQnn.
tions in a construct's definition. The way in (e.g., hot or cold). Likewise, age can be continu­ val measures; thII
which a researcher conceptualizes a variable lim­ ous (how old a person is in years, months, days, score. Ratio JDeaSI
its the levels of measurement that he or she can hours, and minutes) or treated as discrete cate­ levels do, plus~
use and has implications for how measurement gories (infancy, childhood, adolescence, young possible to state nj
and statistical analysis can proceed. adulthood, middle age, old age). Yet, most dis­ or ratios (e.g., IIIIlIl
crete variables cannot be conceptualized as con­ years of formal scl
Continuous and Discrete Variables. Vari­ tinuous variables. For example, sex, religion, and years).
ables can be thought of as being either contin­ marital status cannot be conceptualized as con­ In most pra.:Q
uous or discrete. Continuous variables have an tinuous; however, related constructs can be con­ between inten~l.
infinite number ofvalues or attributes that flow ceptualized as continuous (e.g., femininity, ference. The ­
along a continuum. The values can be divided degree of religiousness, commitment to a mari­ measures can be
into many smaller increments; in mathematical tal relationship, etc.). in temperature ­
theory, there is an infinite number of incre­ The level of measurement limits the statisti­ ally a doublingoft1
ments. Examples of continuous variables in­ cal measures that can be used. A wide range of I
clude temperature, age, income, crime rate, and powerful statistical procedures are available for
amount of schooling. Discrete variables have a the higher levels of measurement, but the types
relatively fixed set of separate values or variable of statistics that can be used with the lowest lev­ TABLE 5.2
attributes. Instead of a smooth continuum of els are very limited.
values, discrete variables contain distinct cate­ There is a practical reason to conceptualize
gories. Examples of discrete variables include and measure variables at higher levels of mea­
gender (male or female), religion (Protestant, surement. You can collapse higher levels of mea­
Catholic, Jew, Muslim, atheist), and marital surement to lower levels, but the reverse is not
status (never married single, married, di­ true. In other words, it is possible to measure a Ordinal
vorced or separated, widowed). Whether a construct very precisely, gather very specific in­ Interval
variable is continuous or discrete affects its formation, and then ignore some of the preci­
level of measurement. sion. But it is not possible to measure a construct

with less precision or with less specific informa­ numbers double, because zero degrees is not the
tion and then make it more precise later. absence of all heat.
Discrete variables are nominal and ordinal,
Distinguishing among the FourLevels. The four whereas continuous variables can be measured
levels from lowest to greatest or highest preci­ at the interval or ratio level. A ratio-level mea­
sion are nominal, ordinal, interval, and ratio. sure can be turned into an interval, ordinal, or
Each level gives a different type of information nominal level. The interval level can always be
(see Table 5.2). Nominal measures indicate only turned into an ordinal or nominal level, but the
that there is a difference among categories (e.g., process does not work in the opposite way!
religion: Protestant, Catholic, Jew, Muslim; In general, use at least five ordinal categories
racial heritage: African, Asian, Caucasian, His­ and obtain many observations. This is because
panic, other). Ordinal measures indicate a dif­ the distortion created by collapsing a continu­
ference, plus the categories can be ordered or ous construct into a smaller number of ordered
ranked (e.g., letter grades: A, B, C, D, F; opinion categories is minimized as the number of cate­
measures: Strongly Agree, Agree, Disagree, gories and the number of observations increase.
Strongly Disagree). Interval measures everything The ratio level of measurement is rarely
the first two do, plus it can specifythe amount of used in the social sciences. For most purposes, it
distance between categories (e.g., Fahrenheit or is indistinguishable from interval measurement.
Celsius temperature: 5°,45°, 90°; IQ scores: 95, The only difference is that ratio measurement
110,125). Arbitrary zeroes may be used in inter­ has a "true zero." This can be confusing because
val measures; they are just there to help keep some measures, like temperature, have zeroes
score. Ratio measures do everything all the other that are not true zeroes. The temperature can be
levels do, plus there is a true zero, which makes it zero, or below zero, but zero is an arbitrary
possible to state relations in terms of proportion number when it is assigned to temperature. This
or ratios (e.g., money income: $10, $100, $500; can be illustrated by comparing zero degrees
years of formal schooling: 1 year, 10 years, 13 Celsius with zero degrees Fahrenheit-they are
years). different temperatures. In addition, doubling
In most practical situations, the distinction the degrees in one system does not double the
between interval and ratio levelsmakes little dif­ degrees in the other. Likewise, it does not make
ference. The arbitrary zeroes of some interval sense to say that it is "twice as warm," as is pos­
measures can be confusing. For example, a rise sible with ratio measurement, if the temperature
in temperature from 30 to 60 degrees is not re­ rises from 2 to 4 degrees, from 15 to 30 degrees,
ally a doubling of the temperature, although the or from 40 to 80 degrees. Another common ex-

TAB LE 5.2 Characteristics of the Four Levels of Measurement

Nominal Yes
Ordinal Yes Yes
Interval Yes Yes Yes
Ratio Yes Yes Yes Yes

ample of arhitrary-not true-zeroes occurs Keep two things in mind. First, virtually
when measuring attitudes where numbers are every social phenomenon can be measured.
assigned to statements (c.g., -1 = disagree, 0 = Some constructs can be measured directly and
no opinion, +1 = agree). True zeroes exist for produce precise numerical values (e.g., family
variables such as income, age, or years of educa­ income). Other constructs require the use of
tion. Examples of the four levelsof measurement surrogates or proxies that indirectly measure a _,ill..........
are shown in Table 5.3. variable and may not be as precise (e.g., predis­
position to commit a crime). Second, a lot can
be learned from measures used by other re­
Specialized Measures: Scales
searchers. You are fortunate to have the work of
and Indexes thousands of researchers to draw on. It is not al­
• 5
Researchers have created thousands of different ways necessary to start from scratch. You can usc
'Ii. -
scales and indexes to measure social variables.
For example, scales and indexes have been de­
veloped to measure the degree of formalization
in bureaucratic organizations, the prestige of oc­
a past scale or index, or you can modify it for
your own purposes.

Indexes and Scales. You might find the terms

•- .......

cupations, the adjustment of people to a mar­ index and scale confusing because they are often
riage, the intensity of group interaction, the level used interchangeably. One researcher's scale is
of social activity in a community, the degree to another's index. Both produce ordinal- or inter­
which a state's sexual assault laws reflect feminist val-level measures of a variable. To add to the
values, and the level of socioeconomic develop­ confusion, scale and index techniques can be
ment of a nation. I cannot discuss the thousands combined in one measure. Scales and indexes
of scales and indexes. Instead, I will focus on give a researcher more information about vari­
principles of scale and index construction and ables and make it possible to assess the quality of
explore some major types. measurement. Scales and indexes increase relia­


TAB L E 5.3 Example of Levels of Measurement

Religion (nominal) Different religious denominations (Jewish, Catholic, Lutheran, Baptist) are not
ranked, just different (unless one belief is conceptualized as closer to heaven).
Attendance (ordinal) "How often do you attend religious services? (0) Never, (1) less than once a
year, (3) several times a year, (4) about once a month, (5) two or three times
a week, or (8) several times a week?" This might have been measured at a ratio
level if the exact number of times a person attended was asked instead.
IQ Score (interval) Most intelligence tests are organized with 100 as average, middle, or normal.
Scores higher or lower indicate distance from the average. Someone with a
score of 11 5 has somewhat above average measured intelligence for people
who took the test, while 90 is slightly below. Scores of below 65 or above 140
are rare.
Age (ratio) Age is measured by years of age. There is a true zero (birth). Note that a 40­
year-old has lived twice as long as a 20-year-old.

the Jewish category. Likewise,a variable measur­

ing type of city, with the attributes river port city,
state capital, and interstate highway exit, lacks
mutually exclusive attributes. One city could be
For most purposes, you can treat scales and indexes all three (a river port state capital with an inter­
as interchangeable. Social researchers do not use a state exit), anyone of the three, or none of the
consistent nomenclature to distinguish between three.
them. Exhaustive attributes means that all cases fit
A scale is a measure in which a researcher captures into one of the attributes of a variable. When
the intensity, direction, level, or potency of a variable measuring religion, a measure with the attrib­
construct. It arranges responses or observations on utes Catholic, Protestant, and Jewish is not ex­
a continuum. A scale can use a single indicator or
clusive. The individual who is a Buddhist, a
multiple indicators. Most are at the ordinal level of
Moslem, or an agnostic does not fit anywhere.
The attributes should be developed so that every
An index is a measure in which a researcher adds
possible situation is covered. For example,
or combines several distinct indicators of a construct
into a single score. This composite score is often a
Catholic, Protestant, Jewish, or other is an ex­
simple sum of the multiple indicators. It is used for clusive and mutually exclusive set of attributes.
content and convergent validity. Indexes are often
measured at the interval or ratio level. Unidimensionality. In addition to being mu­
Researchers sometimes combine the features of tually exclusive and exhaustive, scales and in­
scales and indexes in a single measure. This is com­ dexes should also be unidimensional, or one
mon when a researcher has several indicators that dimensional. Unidimensionality means that all
are scales (i.e., that measure intensity or direction). the items in a scale or index fit together, or mea­
He or she then adds these indicators together to sure a single construct. Unidimensionality was
yield a single score, thereby creating an index. suggested in discussions of content and concur­

I bility and validity, and they aid in data reduc­

tion; that is, they condense and simplify the in­
rent validity. Unidimensionality says: If you
combine several specific pieces of information
into a single score or measure, have all the pieces
work together and measure the same thing. Re­
searchers use a statistical measure called Cron­
formation that is collected (see Box 5.2). bach's alpha to assess unidimenionality. Alpha
ranges from a maximum of 1.0 for a perfect
Mutually Exclusive and Exhaustive Attributes. score to zero. To be considered a good measure,
Before discussing scales and indexes, it is impor­ the alpha should be .70 or higher.
tant to review features of good measurement. There is an apparent contradiction between
The attributes of all measures, including nomi­ using a scale or index to combine parts or sub­
nal-level measures, should be mutually exclusive parts of a construct into one measure and the
and exhaustive. criteria of unidimensionality. It is only an appar­
Mutually exclusive attributes means that an ent contradiction, however, because constructs
individual or case fits into one and only one at­ are theoretically defined at different levels of ab­
tribute of a variable. For example, a variable straction. General, higher-level or more abstract
measuring type of religion-with the attributes constructs can be defined as containing several
Christian, non-Christian, and Jewish-is not subparts. Each subdimension is a part of the
mutually exclusive. Judaism is both a non­ construct's overall content.
Christian religion and a Jewish religion, so a Jew­ For example, I define the construct "femi­
ish person fits into both the non-Christian and nist ideology" as a general ideology about gen­

der. Feminist ideology is a highly abstract and inflation, is created by totaling the cost of buying
general construct. It includes specific beliefs and a list of goods and services (e.g., food, rent, and
attitudes toward social, economic, political, fam­ utilities) and comparing the total to the cost of
ily, and sexual relations. The ideology's five be­ buying the same list in the previous year. The
lief areas are parts of the single general construct. consumer price index has been used by the U.S.
The parts are mutually reinforcing and together Bureau of Labor Statistics since 1919; wage in­
form a system of beliefs about the dignity, creases, union contracts, and social security pay­
strength, and power of women. ments are based on it. An index is a combination
If feminist ideology is unidimensional, then of items into a single numerical score. Various

--_ .
there is a unified belief system that varies from components or subparts of a construct are each

very antifeminist to very profeminist. We can test measured, then combined into one measure. ~ j
the validity of the measure that includes multiple There are many types of indexes. For exam­
indicators that tap the construct's subparts. If ple, if you take an exam with 25 questions, the
one belief area (e.g., sexual relations) is consis­ total number of questions correct is a kind of in­ a..:,. • -• ..
tently distinct from the other areas in empirical
tests, then we question its unidimensionality.
dex. It is a composite measure in which each
question measures a small piece of knowledge,

It is easy to become confused: A specific and all the questions scored correct or incorrect
measure can be an indicator of a unidimensional are totaled to produce a single measure.
construct in one situation and indicate a part of Indexes measure the most desirable place to -
a different construct in another situation. This is live (based on unemployment, commuting time,

possible because constructs can be used at dif­ crime rate, recreation opportunities, weather,
ferent levels of abstraction. and so on), the degree of crime (based on com­ -=. ..
For example, a person's attitude toward bining the occurrence of different specific
gender equality with regard to pay is more spe­ crimes), the mental health of a person (based on
cificand less abstract than feminist ideology (i.e., the person's adjustment in various areas oflife),
beliefs about gender relations throughout soci­ and the like.
ety). An attitude toward equal pay can be both a One way to demonstrate that indexes are
unidimensional construct in its own right and a not very complicated is to use one. Answer yes or
subpart of the more general and abstract unidi­ no to the seven questions that follow on the
mensional construct, ideology toward gender characteristics of an occupation. Base your an­
relations. swers on your thoughts regarding the following
four occupations: long-distance truck driver,
medical doctor, accountant, telephone operator.
Score each answer 1 for yes and 0 for no.
1. Does it pay a good salary?
The Purpose
2. Is the job secure from layoffs or unemploy­
You hear about indexes all the time. For example, ment?
U.S. newspapers report the Federal Bureau of In­ 3. Is the work interesting and challenging?
vestigation (FBI) crime index and the consumer 4. Are its working conditions (e.g., hours,
price index (CPI). The FBI index is the sum of safety, time on the road) good?
police reports on seven so-called index crimes 5. Are there opportunities for career advance­
(criminal homicide, aggravated assault, forcible ment and promotion?
rape, robbery, burglary, larceny of $50 or more, 6. Is it prestigious or looked up to by others?
and auto theft). It began with the Uniform Crime 7. Does it permit self-direction and the free­
Report in 1930. The CPI, which is a measure of dom to make decisions?

Total the seven answers for each of the four are threatened whenever data for some cases are
occupations. Which had the highest and which missing. There are four ways to attempt to re­
had the lowest score? The seven questions are solve the problem, but none fully solve it.
my operational definition of the construct good For example, I construct an index of the de­
occupation. Each question represents a subpart gree of societal development in 1975 for 50 na­
of my theoretical definition. A different theoret­ tions. The index contains four items: life
ical definition would result in different ques­ expectancy, percentage of homes with indoor
tions, perhaps more than seven. plumbing, percentage ofpopulation that is liter­
Creating indexes is so easy that it is impor­ ate, and number of telephones per 100 people. I
tant to be careful that every item in the index has locate a source of United Nations statistics for
face validity. Items without face validity should my information. The values for Belgium are 68 +
be excluded. Each part of the construct should 87 + 97 + 28; for Turkey, the scores are 55 + 36
be measured with at least one indicator. Of + 49 + 3; for Finland, however, I discover that
course, it is better to measure the parts of a con­ literacy data are unavailable. I check other
struct with multiple indicators. sources of information, but none has the data
because they were not collected.
Rates and Standardization
An important issue in index construction is
whether to weight items. Unless it is otherwise You have heard of crime rates, rates of popula­
stated, assume that an index is unweighted. Like­ tion growth, and the unemployment rate. Some
wise, unless you have a good theoretical reason indexes and single-indicator measures are ex­
for assigning different weights, use equal pressed as rates. Rates involve standardizing the
weights. An unweighted index gives each item value of an item to make comparisons possible.
equal weight. It involves adding up the items The items in an index frequently need to be stan­
without modification, as if each were multiplied dardized before they can be combined.
by 1 (or -1 for items that are negative). Standardization involves selecting a base
In a weighted index, a researcher values or and dividing a raw measure by the base. For ex­
weights some items more than others. The size ample, City A had 10 murders and City B had 30
of weights can come from theoretical assump­ murders in the same year. In order to compare
tions, the theoretical definition, or a statistical murders in the two cities, the raw number
technique such as factor analysis. Weighting of murders needs to be standardized by the city
changes the theoretical definition of the con­ population. If the cities are the same size, City B
struct. is more dangerous. But City B may be safer if it is
Weighting can produce different index much larger. For example, if City A has 100,000
scores, but in most cases, weighted and un­ people and City B has 600,000, then the murder
weighted indexes yield similar results. Re­ rate per 100,000 is 10 for City A and 5 for City B.
searchers are concerned with the relationship Standardization makes it possible to com­
between variables, and weighted and un­ pare different units on a common base. The
weighted indexes usually give similar results for process of standardization, also called norming,
the relationships between variables. removes the effect of relevant but different char­
acteristics in order to make the important differ­
ences visible. For example, there are two classes
Missing Data
of students. An art class has 12 smokers and a bi­
Missing data can be a serious problem when ology class has 22 smokers. A researcher can
constructing an index. Validity and reliability compare the rate or incidence of smokers by

standardizing the number of smokers by the size When combining several items into an index, it --­
of the classes. The art class has 32 students and is best to standardize items on a common base . Be'

the biology class has 143 students. One method (see Box 5.3). 5.3
of standardization that you already know is the
use of percentages, whereby measures are stan­ Sports fans in a.r
dardized to a common base of 100. In terms of SCALES "winning· at 1tIr 201
percentages, it is easy to see that the art class has most gold -.:i& I
more than twice the rate of smokers (37.5 per­ The Purpose sto~ft·_

cent) than the biology class (15.4 percent). IWOl"Icfs ridlE5l . . .

Scaling,like index construction, creates an ordi­
A critical question in standardization is de­ JOfI does WEI . .
nal, interval, or ratio measure of a variable ex­
ciding what base to use. In the examples given, Nltions.. To 5Iee . . . . .
pressed as a numerical score. Scalesare common dan:ize on ~ tasI! fill
how did I know to use city size or classsize as the in situations where a researcher wants to mea­
base? The choice is not always obvious; it de­ dardization JIid* '
sure how an individual feels or thinks about
pends on the theoretical definition of a construct. ~tfr IelIIls
something. Some call this the hardness or po­
Different bases can produce different rates. tency of feelings.
For example, the unemployment rate can be de­ Scales are used for two related purposes.
fined as the number of people in the work force First, scales help in the conceptualization and
who are out of work. The overall unemployment operationalization processes. Scalesshow the fit
rate is: between a set of indicators and a single con­

Unemployment rate ==
Number of
unemployed people
Total number of
people working
struct. For example, a researcher believes that
there is a single ideological dimension that un­
derlies people's judgments about specific poli­
cies (e.g., housing, education, foreign affairs,


etc.). Scaling can help determine whether a sin­

gle construct- for instance, "conservative/lib­
We can divide the total population into sub­ ru.r
groups to get rates for subgroups in the popula­
eral ideology"-underlies the positions people
take on specific policies.
7 .....,
tion such as White males, African American Second, scaling produces quantitative mea­ &
females, African American males between the 9
sures and can be used with other variables to test
ages of 18 and 28, or people with collegedegrees. TO
hypotheses. This second purpose of scaling is
Rates for these subgroups may be more relevant our primary focus because it involves scales as a
to the theoretical definition or research problem. technique for measuring a variable.
For example, a researcher believes that unem­
ployment is an experience that affects an entire
~ "llJ"'_...._rl~.
Logic of Scaling 'ti:15sarl5
household or family and that the base should be
~ "'lIIit""...~•••
households, not individuals. The rate will look As stated before, scaling is based on the idea of
like this: measuring the intensity, hardness, or potency of
a variable. Graphic rating scales are an elemen­
Number of households tary form of scaling. People indicate a rating by
with at least one checking a point on a line that runs from one ex­
New unemployed person
Unemployment treme to another. This type of scale is easy to
rate Total number construct and use. It conveys the idea of a con­
of households tinuum, and assigning numbers helps people
think about quantities. A built-in assumption of
Different conceptualizations suggest differ­ scales is that people with the same subjective
ent bases and different ways to standardize. feeling mark the graphic scale at the same place.


Sports fans in the United States were jubilant about ulations and wealth. The results show that the Ba­
"winning" at the 2000 Olympics by carrying off the hamas, with less than 300,000 citizens (smallerthan
most gold medals. However, because they failed to a medium-sized U.S. city), proportionately won the
standardize, the "win" is an illusion. Of course, the most gold. Adjusted for its population size or wealth,
world's richest nation with the third largest popula­ the United States is not even near the top; it appears
tion does well in one-on-one competition among all to be the leader only because of its great size and
nations. To see what reallyhappened, one must stan­ wealth. Sports fans in the United States can perpet­
dardize on a base of the population or wealth. Stan­ uate the illusion of being at the top only if they ig­
dardization yields a more accurate picture by nore the comparative advantage of the United
adjusting the results as ifthe nations had equal pop­ States.


1 USA 39 Bahamas 1 33.3 20.0

2 Russia 32 Slovenia 2 10 10.0
3 China 28 Cuba 11 9.9 50.0
4 Australia 16 Norway 4 9.1 2.6
5 Germany 14 Australia 16 8.6 4.1
6 France 13 Hungry 8 7.9 16.7
7 Italy 13 Netherlands 12 7.6 3.0
8 Netherlands 12 Estonia 1 7.1 20.0
9 Cuba 11 Bulgaria 5 6.0 41.7
10 Britain 11 Lithuania 2 5.4 18.2
EU15** 80 EU15 80 2.1 0.9
USA 39 1.4 0.4

Note: 'Population is gold medals per 10 million people and GDP is gold medals per $1 0 billion;

'*EU 15 is the 1 5 nations of the European Union treated as a single unit.

Source: Adapted from The Economist, October 7, 2000, p. 52.

Figure 5.4 is an example of a "feeling ther­ measure attitudes toward candidates, social
mometer" scale that is used to find out how peo­ groups, and issues.
ple feel about various groups in society (e.g., the
National Organization of Women, the Ku Klux
Commonly Used Scales
Klan, labor unions, physicians, etc.). This type
of measure has been used by political scientists "'tLikert Scale. You have probably used Likert
in the National Election Study since 1964 to scales; they are widely used and very common in


"Feeling Thermometer"
Graphic Rating Scale
forth. Keep the number of choices to eight or
nine at most. More distinctions than that are
probably not meaningful, and people will be­
come confused. The choices should be evenly
100 Very Warm
balanced (e.g., "strongly agree," "agree" with The Rosenberg !i
90 "strongly disagree," "disagree"). Aliinall,lam~
Researchers have debated about whether to 1. Almosta~
80 offer a neutral category (e.g., "don't know," "un­ 2. Often true 1
decided," "no opinion") in addition to the di­ 3. Sometimes til
rectional categories (e.g., "disagree," "agree"). 4. Seldom true j
60 A neutral category implies an odd number of 5. Never true

50 Neither Warm nor Cold categories.

A researcher can combine several Likert A Student EvaIIMl
40 scale questions into a composite index if they all Overall, I rate the~
measure a single construct. Consider the Social Excellent •
Dominance Index that van Laar and colleagues <
20 (2005) used in their study of racial-ethnic atti­ A Market ReseiIIIIIl,

tudes among college roommates (see Box 5.5). ~
As part of a larger survey, they asked four ques­ Brand ~
0 Very Cold tions about group inequality. The answer to each " J
question was a seven-point Likert scale with
choices from Strongly Disagree to Strongly
Agree. They created the index by adding the an­
swers for each student to create scores that WOrkGroup~
ranged from 4 to 28. Notice that they worded
My supervisor: i
survey research. They were developed in the question number four in a reverse direction from
1930s by Rensis Likert to provide an ordinal-level the other questions. The reason for switching di­
measure of a person's attitude. Likert scales usu­ rections in this way is to avoid the problem ofthe
Lets members
ally ask people to indicate whether they agree or response set. The response set, also called response
disagree with a statement. Other modifications style and response bias, is the tendency of some Is friendly and
are possible; people might be asked whether they people to answer a large number of items in the Treats all unit
approve or disapprove, or whether they believe same way (usually agreeing) out of laziness or a
something is "almost always true." Box 5.4 pre­ psychological predisposition. For example, if
sents several examples of Likert scales. items are worded so that saying "strongly agree"
Likert scales need a minimum of two cate­ always indicates self-esteem, we would not know
gories, such as "agree" and "disagree." Using whether a person who always strongly agreed had ated with . J
only two choices creates a crude measure and high self-esteem or simply had a tendency to indexuses~
forces distinctions into only two categories. It is agree with questions. The person might be an­ reliability. lhe~'
usually better to use four to eight categories. A swering "strongly agree" out of habit or a ten­ measure several
researcher can combine or collapse categories af­ dency to agree. Researchers word statements in 100 .IIIlpI'O¥l5
ter the data are collected, but data collected with alternative directions, so that anyone who agrees 5C~Tes give ~ ~
crude categories cannot be made more precise all the time appears to answer inconsistently or to ot a person's ~
You can increase the number of categories
at the end of a scale by adding "strongly agree,"
"somewhat agree," "very strongly agree," and so
have a contradictory opinion.
Researchers often combine many Likert­
scaled attitude indicators into an index. The
scale and indexes have properties that are associ-
500'S opinion ~

. ~

The Rosenberg Self-Esteem Scale

All in all, Iam inclined to feel that I am a failure:
1. Almost alwaystrue
2. Often true
3. Sometimes true
4. Seldom true
5. Never true

A Student Evaluation of Instruction Scale

Overall, I rate the quality of instruction in this course as:
Excellent Good Average Fair Poor

A Market Research Mouthwash Rating Scale

Dislike Dislike Dislike like like like
Brand Completely Somewhat a little a little Somewhat Completely


Work Group Supervisor Scale

My supervisor:
Never Seldom Sometimes Often Always

Lets members knowwhat is expected of them 2 3 4 5

Is friendly and approachable 2 3 4 5
Treats all unit members as equals 2 3 4 5

ated with improving reliability and validity. An Instead of scoring Likert items, as in the pre­
index uses multiple indicators, which improves vious example, the scores -2, -1, + 1, +2 could
reliability. The use of multiple indicators that be used. This scoring has an advantage in that a
measure several aspects of a construct or opin­ zero implies neutrality or complete ambiguity,
ion improves content validity. Finally, the index whereas a high negative number means an atti­
scores give a more precise quantitative measure tude that opposes the opinion represented by a
of a person's opinion. For example, each per­ high positive number.
son's opinion can be measured with a number The numbers assigned to the response cate­
from 10 to 40, instead of in four categories: gories are arbitrary. Remember that the use of a
"strongly agree," "agree," "disagree," and zero does not give the scale or index a ratio level
"strongly disagree." of measurement. Likert scale measures are at the

Example 1 that had yes or no answers to create two composite

In a study of college roommates and racial-ethnic indexes. The index for vicarious experiences was the
groups, van Laar and colleagues (2005) measured sum of items 2, 4, and 6, with "yes" scored as 1 and
Social Dominance (i.e., a feeling that groups are fun­ "no" scored as zero. An index of personal experience
damentally unequal) with the following four-item in­ was the sum of answers to items 1, 3, 5, and 7, with
dex that used a Likert scale, from 1 (Strongly "yes" scored as 1 and "no" scored as zero.
Disagree) to 7 (Strongly Agree).
1. Have you ever been stopped by police on the
street without a good reason?
1. It is probably a good thing that certain groups
are at the top and other groups are at the bot­ 2. Has anyone else in your household been
tom. stopped by police on the street without a good
2. Inferior groups should stay in their place.
3. Have the police ever used insulting language to­
3. We should do all we can to equalize the condi­
ward you?
tions of different groups.
4. Have the police ever used insulting language to­
4. We should increase social equality:
ward anyone else in your household?
'NOTE: This item was reverse scored. 5. Have the police ever used excessive force
against you?
The scores for the Likertresponses (1 to 7) for items 6. Have the police ever used excessive force
1 to 4 were added to yield an index that ranged from against anyone else in your household?
4 to 28 for each respondent. They report a Cron­ 7. Have you ever seen a police officer engage in
bach's alpha for this index as .74. any corrupt activities (such as taking bribes or
involvement in drug trade)?
Example 2
In a study of perceptions of police misconduct, Weitzer and Tuch (2004) report a Cronbach's
Weitzer and Tuch (2004) measured a respondent's alpha for the personal experiences index as .78 and
experiences with police by asking seven questions for vicarious experience index as .86.

ordinal level of measurement because responses cator measurement is possible. The scale has two
indicate a ranking only. Instead of 1 to 4 or -2 limitations: Different combinations of several
to +2, the numbers 100, 70, 50, and 5 would scale items can result in the same overall score or
have worked. Also, do not be fooled into think­ result, and the response set is a potential danger.
ing that the distances between the ordinal cate­
gories are intervals just because numbers are Bogardus Social Distance Scale. The Bogardus
assigned. Although the number system has nice social distance scale measures the social distance
mathematical properties, the numbers are used separating ethnic or other groups from each
for convenience only. The fundamental mea­ other. It is used with one group to determine
surement is only ordinal. how much distance it feels toward a target or
The simplicity and ease of use of the Likert "out-group."
scale is its real strength. When several items are The scale has a simple logic. People respond
combined, more comprehensive multiple indi­ to a series of ordered statements; those that are

most threatening or most socially distant are at cially distant items will refuse the socially closer

one end, and those that might be least threaten­ items (see Box 5.6).

ing or socially intimate are at the other end. The Researchers use the scale in several ways. For

logic of the scale assumes that a person who re­ example, people are given a series of statements:

fuses contact or is uncomfortable with the so- People from Group X are entering your country,

In 1993, Kleg and Yamamoto (1 998) replicated the gories, but they were worded slightly differently (see
original 1925 study by Emory Bogardus that first below). Third, both studies had seven categories
used the social distance scale. The original study had (called anchor points) printed left to right at the top.

1 1 0 subjects from the Pacific Coast. Participants in­ In the Bogardus original it said: "According to my first
cluded 107 White Americans of non-Jewish Euro­ feeling reactions I would willingly admit members of
pean ancestry, 1 Jewish White, 1 Chinese, and 1 each race (as a class, and not the best I have known,
Japanese (about 70 percent were female). In their nor the worst members) to one or more of the clas­
1993 replication, Kleg and Yamamoto selected 1 35 sifications under which I have placed a cross (x)." In
middle school teachers from an affluent school dis­ the 1993 replication it said: "Social distance means
trict in a Colorado metropolitan area. There were the degree that individuals desire to associate with
1 1 9 non-Jewish Whites, 7 Jewish Whites, 6 African others. This scale relates to a special form of social
Americans, 1 American Indian, 1 Asian, and 1 un­ distance known as person to group distance. You are
known (65 percent were female). There were three given a list of groups. Across from each group there
minor deviations from the 1925 study. First, the are boxes identified by the labels at the top. Place an
original Bogardus respondents were given a list of "x" in the boxes that indicate the degree of associa­
39 groups. Those in the replication had a list of 36 tion you would desire to have with each group. Give
groups. The two lists shared 24 groups in common. your first reaction." The main finding was that al­
Three target groups were renamed: Negroes in though the average social distance declined a great
1925 versus African Americans in 1993; Syrians ver­ deal over over 68 years, the ranking of the 25
sus Arabs; and German-Jews and Russian-Jews vs. groups changed very little (see below).
Jews. Second, both studies contained seven cate-


1. To close kinship by marriage To marry into group

2. To my club as personal chums To have as best friend
3. To my street as neighbors To have as next-door neighbors
4. To employment in my occupation in my country To work in the same office
5. To citizenship in my country To have as speaking acquaintances only
6. As visitors only to my country To have as visitors to my country
7. Would exclude from my country To keep out of my country

social distance ~
dent or a depera
A researchc
see how distane
Results versus another.
ethnic groups. iI
tor-patient disIl
associates (200t
English 1.27 1 1.17 2 ported differem
Scottish 1.69 2 1.22 6 with different
Irish 1.93 3 1.14 1 would be willin
French 2.04 4 1.20 4 with arthritis, UI
Dutch 2.12 5 1.25 9 tion. Fewer than
Swedish 2.44 6 1.21 5 being a friend tel
tion. The social
Danish 2.48 7 1.23 7
way to determis
Norwegian 2.67 8 1.25 8
toward a social !
German 2.89 9 1.27 10
tations. First, an
Spanish 3.28 10 1.29 11 egories to Ci!- speci
Italian 3.98 11 1.19 3 Second, it ~not.
Hindu 4.35 12 1.95 23 how a responde!
Polish 4.57 13 1.30 12 groups unless th
Russian 4.57 14 1.33 13 lar social distaoo
Native American 4.65 15 1.44 16 same time. Of Cl
Jewish 4.83* 16 1.42 15 pletes the scale a
Greek 4.89 17 1.38 14 havior in specific
Arab 5.00* 18 2.21 24
Mexican 5.02 19 1.56 18 f Semantic DiffO!
provides an indi
Black American 5.10* 20 1.55 17
feels about a COl
Chinese 5.28 21 1.68 20
The technique III
Japanese 5.30 22 1.62 19 ward something I
Korean 5.55 23 1.72 21 cause people COIII
Turk 5.80 24 1.77 22 adjectives in spoI
Grand Mean 3.82 1.43 cause most adieci
light/dark, ha;d 51
"Slight change in name of group. po site adjectives
scale. The Semai
connotations as5lII
evaluated and pm
are in your town, work at your place of employ­ feel uncomfortable with the relationship. People The Semantic
ment, live in your neighborhood, become your may be asked to respond to all statements, or many purposes. I
personal friends, and marry your brother or sis­ they may keep reading statements until they are how consumers •
ter. People are asked whether they feel comfort­ not comfortable with a relationship. There is no advisers use it it
able "ith the statement or if the contact is set number of statements required; the number about a candidate
acceptable. It is also possible to ask whether they usually ranges from five to nine. The measure of

social distance can be used as either an indepen­ to determine how a client perceives himself or
dent or a dependent variable. herself (see Box 5.7).
A researcher can use the Bogardus scale to To use the Semantic Differential, a re­
see how distant people feel from one out-group searcher presents subjects with a list of paired
versus another. In addition to studying racial­ opposite adjectives with a continuum of 7 to 11
ethnic groups, it has been used to examine doc­ points between them. The subjects mark the
tor-patient distance. For example, Gordon and spot on the continuum between the adjectives
associates (2004) found that college students re­ that expresses their feelings. The adjectives can
ported different social distance toward people be very diverse and should be well mixed (e.g.,
with different disabilities. Over 95 percent positive items should not be located mostly on
would be willing to be a friend with someone either the right or the left side). Studies of a wide
with arthritis, cancer, diabetes, or a heart condi­ variety of adjectives in English found that they
tion. Fewer than 70 percent would ever consider fall into three major classes of meaning: evalua­
being a friend to someone with mental retarda­ tion (good-bad), potency (strong-weak), and ac­
tion. The social distance scale is a convenient tivity (active-passive). Of the three classes of
way to determine how close a respondent feels meaning, evaluation is usually the most signifi­
toward a social group. It has two potentiallimi­ cant. The analysis of results is difficult, and a re­
tations. First, a researcher needs to tailor the cat­ searcher needs to use statistical procedures to
egories to a specific out-group and social setting. analyze a subject's feelings toward the concept.
Second, it is not easy for a researcher to compare Results from a Semantic Differential tell a
how a respondent feels toward several different researcher how one person perceives different
groups unless the respondent completes a simi­ concepts or how different people view the same
lar social distance scale for all out-groups at the concept. For example, political analysts might
same time. Of course, how a respondent com­ discover that young voters perceive their candi­
pletes the scale and the respondent's actual be­ date as traditional, weak, and slow, and as
havior in specific social situations may differ. halfway between good and bad. Elderly voters
perceive the candidate as leaning toward strong,
f Semantic Differential. Semantic Differential fast, and good, and as halfway between tradi­
provides an indirect measure of how a person tional and modern.
feels about a concept, object, or other person.
The technique measures subjective feelings to­ f'<;uttman Scaling. Guttman scaling, or cumu­
ward something by using adjectives. This is be­ lative scaling, differs from the previous scales or
cause people communicate evaluations through indexes in that researchers use it to evaluate data
adjectives in spoken and written language. Be­ after they are collected. This means that re­
cause most adjectives have polar opposites (e.g., searchers must design a study with the Guttman
light/dark, hard/soft, slow/fast), it uses polar op­ scaling technique in mind.
posite adjectives to create a rating measure or Guttman scaling begins with measuring a
scale. The Semantic Differential captures the set of indicators or items. These can be ques­
connotations associated with whatever is being tionnaire items, votes, or observed characteris­
evaluated and provides an indirect measure of it. tics. Guttman scaling measures many different
The Semantic Differential has been used for phenomena (e.g., patterns of crime or drug use,
many purposes. In marketing research, it tells characteristics of societies or organizations, vot­
how consumers feel about a product; political ing or political participation, psychological dis­
advisers use it to discover what voters think orders). The indicators are usually measured in a
about a candidate or issue; and therapists use it simple yes/no or present/absent fashion. From 3
to 20 indicators can be used. The researcher se­

. -­
As part of her undergraduate thesis, Daina Hawkes restaurant, clothing store, or grocery store;
studied attitudes toward women with tattoos using boyfriend or not; average grades or failing grades.
the semantic differential (Hawkes, Senn, and Thorn, They used a semantic differential with 22 adjective
2004). The researchers had 268 students at a pairs. They also had participants complete two
medium-sized Canadian university complete a se­ scales: Feminist and Women's Movement scale and
mantic differential form in response to several sce­ Neosexism scale. The semantic differential terms
narios about a 22-year-old woman college student were selected to indicate three factors: evaluative,
with a tattoo. They had five scenarios in which they activity, and potency (strong/weak). Based on sta­
varied the size of the tattoo (small versus large) and tistical analysis three adjectives were dropped. The
whether or not it was visible, and one with no details 19 items used are listed below. Among other find­
about the tattoo. The authors also varied features of ings, the authors found that there were more nega­
the senario: weight problem or not; part-time job at tive feelings toward a woman with a visible tattoo.

Good Bad'
Beautiful Ugly
Clean Dirty
Kind Cruel'
Rich Poor'
Honest Dishonest'
Pleasant Unpleasant'
Successful Unsuccessful
Reputable Disreputable
Safe Dangerous
Gentle Violent'
Feminine Masculine
Weak Powerful'
Passive Active'
Cautious Rash'
Soft Hard
Weak Strong
Mild Intense
Delicate Rugged'

"These items were presented in reverse order.


lects items on the belief that there is a logical re­ Once a set of items is measured, the re­
lationship among them. He or she then places searcher considers all possible combinations of
the results into a Guttman scale and determines responses for the items. For example, three items
whether the items form a pattern that corre­ are measured: whether a child knows her age,
sponds to the relationship. (See Box 5.8 for an her telephone number, and three local elected
example of a study using Guttman scaling.) political officials. The little girl may know her

Crozat (1998) examined public responses to various strations), but not all who accepted modest forms
forms of political protest. He looked at survey data accepted the more intense forms. In addition to
on the public's acceptance of forms of protest in showing the usefulness of the Guttman scale, Crozat
Great Britain, Germany, Italy, Netherlands, and the also found that people in different nations saw
United States in 1974 and 1 990. He found that the protest similarily and the degree of Guttman scala­
pattern of the public's acceptance formed a Guttman bility increased over time. Thus, the pattern of ac­
scale. Those who accepted more intense forms of ceptance of protest activities was Guttman "scalable"
protest (e.g., strikes and sit-ins) almost always ac­ in both time periods, but it more closely followed the
cepted more modest forms (e.g., petitions or demon- Guttman pattern in 1990 than 1974.


Guttman Patterns

y y y y y
Other Patterns (examples only)

N y N Y N

age but no other answer, or all three, or only her tern. Alternative statistics to measure scalability As you are poroI
age and telephone number. In fact, for three have also been suggested. search involves doiI
items there are eight possible combinations of a study. Serious III
answers or patterns of responses, from not one phase can do i
knowing any through knowing all three. There is sults, even if the G
a mathematical way to compute the number of project were coodg
combinations (e.g., 2 3 ) , but you can write down In this chapter, you learned about the principles
all the combinations of yes or no for three ques­ and processes of measurement in quantitative
tions and see the eight possibilities. and qualitative research. All researchers concep­ Key Terms
The logical relationship among items in tualize-or refine and clarify their ideas into
Guttman scaling is hierarchical. Most people or conceptual definitions. All researchers opera­ Bogardus Social Dill
cases have or agree to lower-order items. The tionalize-or develop a set of techniques or conceptual cIe6wjrW
smaller number of cases that have the higher-or­ processes that will link their conceptual defini­ conceptual h~
der items also have the lower-order ones, but not tions to empirical reality. Qualitative and quan­ conceptualintie­
vice versa. In other words, the higher-order titative researchers differ in how they approach concurrent ~ .
items build on the lower ones. The lower-order these processes, however. The quantitative re­ content validity <

items are necessary for the appearance of the searcher takes a more deductive path, whereas continuous ,ai' " ~
higher-order items. the qualitative researcher takes a more inductive criterion v~
An application of Guttman scaling, known path. The goal remains the same: to establish un­ discrete variables I
as scalogram analysis, lets a researcher test ambiguous links between a reseacher's abstract empirical h~
whether a hierarchical relationship exists among ideas and empirical data. exhaustive ~
the items. For example, it is easier for a child to You also learned about the principles of re­ external v~ ~
know her age than her telephone number, and to liability and validity. Reliability refers to the de­ face validitv
know her telephone number than the names of pendability or consistency of a measure; validity Guttman~
political leaders. The items are called scalable, or refers to its truthfulness, or how well a construct index
capable of forming a Guttman scale, if a hierar­ and data for it fit together. Quantitative and internal~
chical pattern exists. qualitative styles of research significantly diverge interval-level
The patterns of responses can be divided in how they understand these principles. None­ IevelsOfJDe2!il1Rl1lllllll
into two groups: scaled and errors (or nonscal­ theless, both quantitative and qualitative re­ Likert scale
able). The scaled patterns for the child's knowl­ searchers try to measure in a consistent way, and
edge example would be as follows: not knowing both seek a tight fit between the abstract ideas
any item, knowing only age, knowing only age they use to understand social world and what
plus phone number, knowing all three. Other occurs in the actual, empirical social world. In
combinations of answers (e.g., knowing the po­ addition, you saw how quantitative researchers
liticalleaders but not her age) are possible but apply the principles of measurement when they
are nonscalable. If a hierarchical relationship ex­ create indexes and scales, and you read about
ists among the items, then most answers fit into some major scales they use.
the scalable patterns. Beyond the core ideas of reliability and va­
The strength or degree to which items can lidity, good measurement requires that you cre­
be scaled is measured with statistics that mea­ ate clear definitions for concepts, use multiple
sure whether the responses can be reproduced indicators, and, as appropriate, weigh and stan­
based on a hierarchical pattern. Most range from dardize the data. These principles hold across all
zero to 100 percent. A score of zero indicates a fields of study (e.g., family, criminology, in­
random pattern, or no hierarchical pattern. A equality, race relations, etc.) and across the
score of 100 percent indicates that all responses many research techniques (e.g., experiments,
to the answer fit the hierarchical or scaled pat­ surveys, etc.).

As you are probably beginning to realize, re­ multiple indicators

search involves doing a good job in each phase of mutually exclusive attributes
a study. Serious mistakes or sloppiness in any nominal-level measurement
one phase can do irreparable damage to the re­ operational definition
sults, even if the other phases of the research operationalization
project were conducted in a flawless manner. ordinal-level measurement
predictive validity
ratio-level measurement
Key Terms reliability
Bogardus Social Distance Scale Semantic Differential
conceptual definition standardization
conceptual hypothesis unidimensionality
conceptualization validity
concurrent validity
content validity
continuous variables Endnote
criterion validity
discrete variables 1. The terms concept, construct, and idea are used
empirical hypothesis more or less interchangeably, but there are differ­
exhaustive attributes ences in meaning between them. An idea is any
external validity mental image, belief plan, or impression. It refers
to any vague impression, opinion, or thought. A
face validity
concept is a thought, a general notion, or a gener­
Guttman scaling alized idea about a class of objects. A construct is a
index thought that is systematically put together, an or­
internal validity derly arrangement of ideas, facts, and impres­
interval-level measurement sions. The term construct is used here because its
levels of measurement emphasis is on taking vague concepts and turning
Likert scale them into systematically organized ideas.
measurement validity
Q 5 c

Qualitative and ...-Jt~...

• i5 S
- 111
Quantitative Sampling
"'of__ ~

Ia:tiaa ...

Win" SIIIIIIf. .L4
Nonprobability Sampling
Haphazard, Accidental, or Convenience Sampling
Quota Sampling
Purposive or Judgmental Sampling
Snowball Sampling
Deviant Case Sampling
Sequential Sampling
Probability Sampling
Populations, Elements, and Sampling Frames
Why Random?
Types of Probability Samples
Hidden Populations
How Large Should a Sample Be?
Drawing Inferences



Qualitative and quantitative researchers ap­ Qualitative researchers rarely draw a representa­
proach sampling differently. Most discussions tive sample from a huge number of cases to in­
of sampling come from researchers who use tensely study the sampled cases-the goal in
the quantitative style. Their primary goal is to quantitative research. Instead, they use non­
get a representative sample, or a small collec­ probability or nonrandom samples. This means
tion of units or cases from a much larger col­ they rarely determine the sample size in advance
lection or population, such that the researcher and have limited knowledge about the larger
can study the smaller group and produce accu­ group or population from which the sample is
rate generalizations about the larger group. taken. Unlike the quantitative researcher who
They tend to use sampling based on theories of uses a preplanned approach based on mathe­
probability from mathematics (called proba­ matical theory, the qualitative researcher selects
bility sampling). cases gradually, with the specific content of a
Researchers have two motivations for using
probability or random sampling. The first moti­
vation is saving time and cost. If properly con­
ducted, results from a sample may yield results
at 111,000 the cost and time. For example, in­
stead of gathering data from 20 million people,
case determining whether it is chosen. Table 6.1

TAB L E 6.1 Types of Nonprobability
a researcher may draw a sample of 2,000; the
data from those 2,000 are equal for most pur­
poses to the data from all 20 million. The sec­
Haphazard Get any cases in any manner
ond purpose of probability sampling is that is convenient.
accuracy. The results of a well-designed, care­
fully executed probability sample will produce Quota Get a preset number of cases in
each of several predetermined
results that are equally if not more accurate than
categories that will reflect the
trying to reach every single person in the whole
diversity of the population,
population. A census is usually an attempt to
using haphazard methods.
count everyone. In 2000, the U.S. Census Bu­
reau tried to count everyone in the nation, but it Purposive Get all possible cases that fit
would have been more accurate if it used very particular criteria, using various
specialized statistical sampling. methods.
Qualitative researchers focus less on a sam­ Snowball Get cases using referrals from
ple's representativeness or on detailed techniques one or a few cases, and then
for drawing a probability sample. Instead, they referrals from those cases, and
focus on how the sample or small collection of so forth.
cases, units, or activities illuminates key features Deviant Case Get cases that substantially
of sociallife.The purpose ofsampling is to collect differ from the dominant
cases, events, or actions that clarify and deepen pattern (a special type of
understanding. Qualitative researchers' concern purposive sample).
is to find cases that will enhance what the re­ Sequential Get cases until there is no
searchers learn about the processes of social life additional information or new
in a specific context. For this reason, qualitative characteristics (often used with
researchers tend to collect a second type of sam­ other sampling methods).
pling: nonprobability sampling.

shows a variety of nonprobability sampling in each category. Thus, the number of people in
techniques. various categories of the sample is fixed. For ex­ FIGURE 6.1
ample, a researcher decides to select 5 males and
5 females under age 30, 10 males and 10 females
Haphazard, Accidental, or
aged 30 to 60, and 5 males and 5 females over age
Convenience Sampling
60 for a 40-person sample. It is difficult to repre­
Haphazard sampling can produce ineffective, sent all population characteristics accurately (see
highly unrepresentative samples and is not rec­ Figure 6.1).
ommended. When a researcher haphazardly se­ Quota sampling is an improvement because
lects cases that are convenient, he or she can the researcher can ensure that some differences
easily get a sample that seriously misrepresents are in the sample. In haphazard sampling, all
the population. Such samples are cheap and those interviewed might be of the same age, sex,
quick; however, the systematic errors that easily or race. But once the quota sampler fixesthe cat­
occur make them worse than no sample at all. egories and number of cases in each category, he
The person-on-the-street interview conducted or she uses haphazard sampling. For example, Of 32 a
by television programs is an example of a hap­ the researcher interviews the first five males un­
hazard sample. Television interviewers go out on der age 30 he or she encounters, even if all five
the street with camera and microphone to talk to just walked out of the campaign headquarters of
a few people who are convenient to interview. a political candidate. Not only is misrepresenta­
The people walking past a television studio in tion possible because haphazard sampling is
the middle of the day do not represent everyone used within the categories, but nothing prevents
(e.g., homemakers, people in rural areas, etc.). the researcher from selecting people who "act
Likewise, television interviewers often select friendly" or who want to be interviewed.
people who look "normal" to them and avoid A case from the history of sampling illus­
people who are unattractive, poor, very old, or trates the limitations of quota sampling. George
inarticulate. Gallup's American Institute of Public Opinion,
Another example of a haphazard sample is using quota sampling, successfully predicted the
that of a newspaper that asks readers to clip a outcomes of the 1936, 1940, and 1944 U.S. pres­
questionnaire from the paper and mail it in. Not idential elections. But in 1948, Gallup predicted Purposive 5aJlII
everyone reads the newspaper, has an interest in the wrong candidate. The incorrect prediction situations. First, a
the topic, or will take the time to cut out the had several causes (e.g., many voters were unde­ unique cases that i1III
questionnaire and mail it. Some people will, and cided, interviewing stopped early), but a major example, a researda
the number who do so may seem large (e.g., reason was that the quota categories did not ac­ sisto study ma~
5,000), but the sample cannot be used to gener­ curately represent all geographical areas and all or she selects a speI
alize accurately to the population. Such haphaz­ people who actually cast a vote. zine to study becais
ard samples may have entertainment value, but Second, a Tesl
they can give a distorted view and seriously mis­ sampling to select
Purposive or Judgmental Sampling
represent the population. reach, specialized pj
Purposive samplingis used in situations in which lations later in t:h.iIl
an expert uses judgment in selecting cases with a researcher wants til
Quota Sampling
specific purpose in mind. It is inappropriate if it possible to list all I
Quotasampling is an improvement over haphaz­ is used to pick the "average housewife" or the domly from the Ii
ard sampling. In quota sampling, a researcher "typical school." With purposive sampling, the subjective informa
first identifies relevant categories of people (e.g., researcher never knows whether the cases se­ prostitutes solicit.
male and female; or under age 30, ages 30 to 60, lected represent the population. It is often used prostitutes associaa
over age 60, etc.), then decides how many to get in exploratory research or in field research. lice who work on ..

FIGURE 6.1 Quota Sampling

o 0 0

• ~9o • w°9°~ ol t
iiti~ ~i' il
• • • • Q•
I i 01
Of 32 adults and children in the street scene, select 10 for the sample:

g II• 0
4 Adult Males

4 Adult Females

1 Male Child 1 Female Child

Purposive sampling is appropriate in three etc.) to identify a "sample" of prostitutes for in­
situations. First, a researcher uses it to select clusion in the research project. The researcher
unique cases that are especially informative. For uses many different methods to identify the
example, a researcher wants to use content analy­ cases, because his or her goal is to locate as many
sis to study magazines to find cultural themes. He cases as possible.
or she selects a specific popular women's maga­ Another situation for purposive sampling
zine to study because it is trend setting. occurs when a researcher wants to identify par­
Second, a researcher may use purposive ticular types of cases for in-depth investigation.
sampling to select members of a difficult-to­ The purpose is less to generalize to a larger pop­
reach, specializedpopulation (see Hidden Popu­ ulation than it is to gain a deeper understanding
lations later in this chapter). For example, the of types. For example, Gamson (1992) used pur­
researcher wants to study prostitutes. It is im­ posive sampling in a focus group study of what
possible to list all prostitutes and sample ran­ working-class people think about politics.
domly from the list. Instead, he or she uses (Chapter 11 discusses focus groups.) Gamson
subjective information (e.g., locations where wanted a total of 188 working-class people to
prostitutes solicit, social groups with whom participate in one of37 focus groups. He sought
prostitutes associate, etc.) and experts (e.g., po­ respondents who had not completed collegebut
lice who work on vice units, other prostitutes, who were diverse in terms of age, ethnicity, reli­

gion, interest in politics, and type of occupation. case, and the lines represent friendship or other poverty rate and
He recruited people from 35 neighborhoods in linkages (see Figure 6.2). percent I s.eaioIl
the Boston area by going to festivals, picnics, Researchers also use snowball sampling in nonpublic SCITi.:r
fairs, and flea markets and posting notices on combination with purposive sampling as in the books, the 1nti:Dllll.
many public bulletin boards. In addition to ex­ case of Kissane (2003) in a descriptive field re­ ingdown~~
plaining the study, he paid the respondents well search study oflow-income women in Philadel­ tified 50 DOD~iI
so as to attract people who would not tradition­ phia. The U.S. policy to provide aid and services observed dw 3.
ally participate in a study. to low-income people changed in 1996 to in­
crease assistance (e.g., food pantries, domestic
violence shelters, drug rehabilitation services,
Snowball Sampling
clothing distribution centers) deliveredby non­
Snowball sampling (also called network, chain re­ public as opposed to government/public agen­
ferral, or reputationai sampling) is a method for cies. As frequently occurs, the policy change was
identifying and sampling (or selecting) the cases made without a study of its consequences in
in a network. It is based on an analogy to a snow­ advance. Noone knew whether the affected low­
ball, which begins small but becomes larger as it income people would use the assistance pro­
is rolled on wet snow and picks up additional vided by nonpublic agencies as much as that
snow. Snowball sampling is a multistage tech­ provided by public agencies. One year after the
nique. It begins with one or a few people or cases new policy, Kissane studied whether low-in­
and spreads out on the basis oflinks to the initial come women were equally likely to use nonpub­
cases. lie aid. She focused on the Kensington area of
One use ofsnowball sampling is to sample a Philadelphia. It had a high (over 30 percent)
network. Social researchers are often interested
in an interconnected network of people or orga­
nizations. The network could be scientists
around the world investigating the same prob­
lem, the elites of a medium-sized city, the mem­

FIG U R E 6.2 Sociogram of Friendship

bers of an organized crime family, persons who Relations
sit on the boards of directors of major banks and
corporations, or people on a college campus
who have had sexual relations with each other.
The crucial feature is that each person or unit is
connected with another through a direct or indi­
rect linkage. This does not mean that each
person directly knows, interacts with, or is influ­
enced by every other person in the network.
Rather, it means that, taken as a whole, with
direct and indirect links, they are within an in­
terconnected web oflinkages.
Researchers represent such a network by
drawing a sociogram-a diagram of circles con­
nected with lines. For example, Sally and Tim do
not know each other directly, but each has a
good friend, Susan, so they have an indirect con­ Dennis
o Larry

nection. All three are part ofthe same friendship Edith

network. The circles represent each person or

poverty rate and was a predominately White (85 previous research suggested that a majority of
percent) section of the city. First, she identified dropouts come from families that have low in­
nonpublic service providers by using telephone come, are single parent or unstable, have been
books, the Internet, referral literature, and walk­ geographically mobile, and are racial minorities.
ing down every street of the area until she iden­ The family environment is one in which parents
tified 50 nonpublic social service providers. She and/or siblings have low education or are them­
observed that a previous study found low-in­ selves dropouts. In addition, dropouts are often
come women in the area distrusted outsiders engaged in illegal behavior and have a criminal
and intellectuals. Her snowball sample began record prior to dropping out. A researcher using
asking service providers for the names of a few deviant case sampling would seek majority­
low-income women in the area. She then asked group dropouts who have no record of illegalac­
those women to refer her to others in a similar tivities and who are from stable two-parent,
situation, and asked those respondents to refer upper-middle-income families who are geo­
her to still others. She identified 20 low-income graphically stable and well educated.
women aged 21 to 50, most who had received
public assistance. She conducted in-depth,
Sequential Sampling
open-ended interviews about their awareness
and experience with nonpublic agencies. She Sequential sampling is similar to purposive sam­
learned that the women were less likely to get pling with one difference. In purposive sam­
nonpublic than public assistance. Compared to pling, the researcher tries to find as many
public agencies, the women were less aware of relevant cases as possible, until time, financial
nonpublic agencies. Nonpublic agencies created resources, or his or her energy is exhausted. The
more social stigma, generated greater adminis­ goal is to get every possible case. In sequential
trative hassles, were in worse locations, and in­ sampling, a researcher continues to gather cases
volved more scheduling difficulties because of until the amount of new information or diver­
limited hours. sity of cases is filled. In economic terms, infor­
mation is gathered until the marginal utility, or
incremental benefit for additional cases, levels
Deviant Case Sampling
off or drops significantly. It requires that a re­
A researcher uses deviant case sampling (also searcher continuously evaluate all the collected
called extreme case sampling) when he or she cases. For example, a researcher locates and
seeks casesthat differ from the dominant pattern plans in-depth interviews with 60 widows over
or that differ from the predominant characteris­ 70 years old who have been living without a
tics of other cases.Similar to purposive sampling, spouse for 10 or more years. Depending on the
a researcher uses a variety of techniques to locate researcher's purposes, getting an additional 20
cases with specific characteristics. Deviant case widows whose life experiences, social back­
sampling differs from purposive sampling in that grounds, and worldviews differ little from the
the goal is to locate a collection of unusual, dif­ first 60 may be unnecessary.
ferent, or peculiar cases that are not representa­
tive of the whole. The deviant cases are selected
because they are unusual, and a researcher hopes
to learn more about the social life by considering
cases that fall outside the general pattern or in­ A specialized vocabulary or jargon has devel­
cluding what is beyond the main flow of events. oped around terms used in probability sam­
For example, a researcher is interested in pling. Before examining probability sampling, it
studying high school dropouts. Let us say that is important to review its language.

Populations, Elements, and commercials, etc.) and geographical and time many types of sampI
Sampling Frames boundaries. tories, tax records, •
A researcher begins with an idea ofthe pop­ so on. Listing the
A researcher draws a sample from a larger pool
ulation (e.g., all people in a city) but defines it sounds simple. It is4
of cases, or elements. A sampling element is the
more precisely. The term target population refers may be no good 1ist4
unit of analysis or case in a population. It can be to the specific pool of cases that he or she wants
a person, a group, an organization, a writt~n A good samplil
to study. The ratio ofthe size ofthe sample to the sampling. A mismi
document or symbolic message, or even a social
size of the target population is the sampling ratio. frame and the rona
action (e.g., an arrest, a divorce, or a kiss) that
For example, the population has 50,000 people, can be a major 50IIl
is being measured. The large pool is the and a researcher draws a sample of 150 from it. match between ~I
population, which has an important role in sam­
Thus, the sampling ratio is 150/50,000 = 0.003, definitions of a ,dIiiI
pling. Sometimes, the term universe is used in­ or 0.3 percent. If the population is 500 and the ment, so a miSIWIII
terchangeably with population. To define the researcher samples 100, then the sampling ratio
population, a researcher specifies the unit being frame and the pops
is 100/500 = 0.20, or 20 percent. pling. Researchers b
sampled, the geographical location, and the tem­
A population is an abstract concept. How For example, you .~
poral boundaries of populations. Consider the
can population be an abstract concept, when in a region of the C.
examples of populations in Box 6.1. All the
there are a given number of people at a certain get a list of everyors
examples include the elements to be sampled
time? Except for specific small populations, one some people do 00
(e.g., people, businesses, hospital admissions,
can never truly freeze a population to measure it. the lists of those ""~
For example, in a city at any given moment, regularly, quickly glI
some people are dying, some are boarding or income tax records.. 1
getting off airplanes, and some are in cars dri­ some people cheat I
ving across city boundaries. The researcher must no income and do 1
decide exactly who to count. Should he or she died or have not ~
count a city resident who happens to be on vaca­ others have entered,
1. All persons aged 1 6 or older living in Singapore
tion when the time is fixed? What about the time taxes were due
on December 2, 1999, who were not incarcer­ tourist staying at a hotel in the city when the ries, but they are 00
ated in prison, asylums, and similar institutions time is fixed? Should he or she count adults, chil­ are not listed in a tell
dren, people in jails, those in hospitals? A popu­ ple have unlisted nl
2. All business establishments employing more
lation, even the population of all people over the cently moved. With
than 1 00 persons in Ontario Province, Canada,
that operated in the month of July 2005
age of 18 in the city limits of Milwaukee, Wis­ of all students enroU
consin, at 12:01 A.M. on March 1,2006, is an ab­ frames are almost all
3. All admissions to public or private hospitals in
stract concept. It exists in the mind but is
the state of New Jersey between August 1,
impossible to pinpoint concretely.
1988, and July 31,1993
Because a population is an abstract concept,
4. All television commercials aired between 7:00 except for small specialized populations (e.g., all FIGURE 6.3 AI
A.M. and 11 :00 P.M. Eastern Standard Time on
the students in a classroom), a researcher needs
three major U.S. networks between November 1
to estimate the population. As an abstract con­
and November 25, 2006
cept, the population needs an operational defin­
5. All currently practicing physicians in Australia ition. This process is similar to developing
who received medical degrees between January operational definitions for constructs that are
1, 1960, and the present
6. All African American male heroin addicts in the A researcher operationalizes a population
Vancouver, British Columbia, or Seattle, Wash­ by developing a specific list that closely approxi­
ington, metropolitan areas during 2003 mates all the elements in the population. This list
is a sampling frame. He or she can choose from

many types of sampling frames: telephone direc­ frame can include some of those outside the tar­
tories, tax records, driver's license records, and get population (e.g., a telephone directory that
so on. Listing the elements in a population lists people who have moved away) or might
sounds simple. It is often difficult because there omit some of those inside it (e.g., those without
may be no good list of elements in a population. telephones).
A good sampling frame is crucial to good Any characteristic of a population (e.g., the
sampling. A mismatch between the sampling percentage of city residents who smoke ciga­
frame and the conceptually defined population rettes, the average height of all women over the
can be a major source of error. Just as a mis­ age of 21, the percent of people who believe in
match between the theoretical and operational UFOs) is a population parameter. It is the true
definitions ofa variable creates invalid measure­ characteristic of the population. Parameters are
ment, so a mismatch between the sampling determined when all elements in a population
frame and the population causes invalid sam­ are measured. The parameter is never known
pling. Researchers try to minimize mismatches. with absolute accuracy for large populations
For example, you would like to sample all people (e.g., an entire nation), so researchers must esti­
in a region of the United States, so you decide to mate it on the basis of samples. They use infor­
get a list of everyone with a driver's license. But mation from the sample, called a statistic, to
some people do not have driver's licenses, and estimate population parameters (see Figure 6.3).
the lists of those with licenses, even if updated A famous case in the history of sampling il­
regularly, quickly go out of date. Next, you try lustrates the limitations of the technique. The
income tax records. But not everyone pays taxes; Literary Digest, a major U.S. magazine, sent
some people cheat and do not pay, others have postcards to people before the 1920, 1924, 1928,
no income and do not have to file, some have and 1932 U.S. presidential elections. The maga­
died or have not begun to pay taxes, and still zine took the names for the sample from auto­
others have entered or left the area since the last mobile registrations and telephone directories­
time taxes were due. You try telephone directo­ the sampling frame. People returned the post­
ries, but they are not much better; some people cards indicating whom they would vote for. The
are not listed in a telephone directory, some peo­ magazine correctly predicted all four election
ple have unlisted numbers, and others have re­ outcomes. The magazine's success with predic­
cently moved. With a few exceptions (e.g., a list tions was well known, and in 1936, it increased
of all students enrolled at a university), sampling the sample to 10 million. The magazine pre­
frames are almost alwaysinaccurate. A sampling dicted a huge victory for Alf Landon over

FIG U R E 6.3 A Model of the Logic of Sampling
What You Population
Would Like to
Talk About

Sampling Process

Franklin D. Roosevelt. But the Literary Digest sample. For example, if conducting a telephone research methods bQ
was wrong; Franklin D. Roosevelt won by a survey, the researcher needs to try to reach the era ted by a pure ~
landslide. specific sampled person, by calling back four or number has an equ.JIj
The prediction was wrong for several rea­ five times, to get an accurate random sample. any position. Comps
sons, but the most important were mistakes in Random samples are most likely to yield a duce lists of randorm
sampling. Although the magazine sampled a sample that truly represents the population. In You may ask, OIl
large number of people, its sampling frame did addition, random sampling lets a researcher sta­ the sampling frame"
not accurately represent the target population tistically calculate the relationship between the sampling frame or.
(i.e., all voters). It excluded people without tele­ sample and the population-that is, the size of common answer is d
phones or automobiles, a sizable percentage of the sampling error. A nonstatistical definition of stricted random sa.­
the population in 1936, during the worst of the the sampling error is the deviation between sam­ with replacemen~
Great Depression of the 1930s. The frame ex­ ple results and a population parameter due to after sampling it
simple random s.mJI
cluded as much as 65 percent of the population random processes.
and a segment of the voting population (lower Random sampling is based on a great deal of the researcher ignoRIl
income) that tended to favor Roosevelt. The sophisticated mathematics. This chapter focuses into the sample.
magazine had been accurate in earlier elections on the fundamentals of how sampling works, the The logic of ~
because people with higher and lower incomes difference between good and bad samples, how illustrated with an ~
did not differ in how they voted. Also, during to draw a sample, and basic principles of sam­ pling marbles from a
earlier elections, before the Depression, more pling in social research. This does not mean that 5,000 marbles, so.lIlill1
lower-income people could afford to have tele­ random sampling is unimportant. It is essential 5,000 marbles are rot
phones and automobiles. to first master the fundamentals. If you plan to meter I want to ~
You can learn two important lessons from pursue a career using quantitative research, you marbles in it. I ~
the Literary Digest mistake. First, the sampling should get more statistical background than close my eyes, shab
frame is cruciaL Second, the size of a sample is space permits here. and repeat the pr~
less important than whether or not it accurately a random sample ofl
represents the population. A representative sam­ ber of red marbles i:oI
Types of Probability Samples
ple of2,500 can give more accurate predications percentage of red ~
about the U.S. population than a nonrepresenta­ Simple Random. The simple random sample is population. This is ~
tive sample of 1 million or 10 million. both the easiest random sample to understand 5,000 marbles. ~f:, ill
and the one on which other types are modeled. red marbles.
In simple random sampling, a researcher devel­ Does this me:.m I
Why Random?
ops an accurate sampling frame, selects elements meter is 48 percent nl
The area of applied mathematics called proba­ from the sampling frame according to a mathe­ cause of random ~
bility theory relies on random processes. The matically random procedure, then locates the might be off. I can ~
word random has a special meaning in mathe­ exact element that was selected for inclusion in the 100 marbles ~
matics. It refers to a process that generates a the sample. bles, and drawing .a ~
mathematically random result; that is, the selec­ After numbering all elements in a sampling 100 marbles. On the.l.
tion process operates in a truly random method frame, a researcher uses a list of random num­ 49 white marbles ~
(i.e., no pattern), and a researcher can calculate bers to decide which elements to select. He or problem. \\b:ich is
the probability of outcomes. In a true random she needs as many random numbers as there are random sampling
process, each element has an equal probability elements to be sampled; for example, for a sam­ from the same
of being selected. ple of 100, 100 random numbers are needed. sultsr I repeat the
Probability samples that rely on random The researcher can get random numbers from a I have drawn 130
processes require more work than nonrandom random-number table, a table of numbers chosen bles each (see Box
ones. A researcher must identify specific sam­ in a mathematically random way. Random­ might empty the w
pling elements (e.g., person) to include in the number tables are available in most statistics and Kant to see what is

research methods books. The numbers are gen­ 130 different samples reveal a clear pattern. The
erated by a pure random process so that any most common mix of red and white marbles is
number has an equal probability of appearing in 50/50. Samples that are close to that split are
any position. Computer programs can also pro­ more frequent than those with more uneven
duce lists of random numbers. splits. The population parameter appears to be
You may ask, Once I select an element from 50 percent white and 50 percent red marbles.
the sampling frame, do I then return it to the Mathematical proofs and empirical tests
sampling frame or do I keep it separate? The demonstrate that the pattern found in Box 6.2
common answer is that it is not returned. Unre­ always appears. The set of many random sam­
stricted random sampling is random sampling ples is my samplingdistribution. It is a distribu­
with replacement-that is, replacing an element tion of different samples that shows the
after sampling it so it can be selected again. In frequency of different sample outcomes from
simple random sampling without replacement, many separate random samples. The pattern will
the researcher ignores elements already selected appear if the sample size is 1,000 instead of 100;
into the sample. if there are 10 colors of marbles instead of 2; if
The logic of simple random sampling can be the population has 100 marbles or 10 million
illustrated with an elementary example-sam­ marbles instead of 5,000; and if the population is
pling marbles from a jar. I have a large jar full of people, automobiles, or colleges instead of mar­
5,000 marbles, some red and some white. The bles. In fact, the pattern will become clearer as
5,000 marbles are my population, and the para­ more and more independent random samples
meter I want to estimate is the percentage of red are drawn from the population.
marbles in it. I randomly select 100 marbles (I The pattern in the sampling distribution
close my eyes, shake the jar, pick one marble, suggests that over many separate samples, the
and repeat the procedure 99 times). I now have true population parameter (i.e., the 50/50 split in
a random sample of marbles. I count the num­ the preceding example) is more common than
ber of red marbles in my sample to estimate the any other result. Some samples deviate from the
percentage of red versus white marbles in the population parameter, but they are less com­
population. This is a lot easier than counting all mon. When many different random samples are
5,000 marbles. My sample has 52 white and 48 plotted as in the graph in Box 6.2, then the sam­
red marbles. pling distribution looks like a normal or bell­
Does this mean that the population para­ shaped curve. Such a curve is theoretically
meter is 48 percent red marbles? Maybe not. Be­ important and is used throughout statistics.
cause of random chance, my specific sample The central limit theorem from mathematics
might be off. I can check my results by dumping tells us that as the number of different random
the 100 marbles back in the jar, mixing the mar­ samples in a sampling distribution increases to­
bles, and drawing a second random sample of ward infinity, the pattern of samples and the
100 marbles. On the second try, my sample has population parameter become more predictable.
49 white marbles and 51 red ones. Now I have a With a huge number of random samples, the
problem. Which is correct? How good is this sampling distribution forms a normal curve, and
random sampling business if different samples the midpoint of the curve approaches the popu­
from the same population can yield different re­ lation parameter as the number of samples
sults? I repeat the procedure over and over until increases.
I have drawn 130 different samples of 100 mar­ Perhaps you want only one sample because
bles each (see Box 6.2 for results). Most people you do not have the time or energy to draw
might empty the jar and count all 5,000, but I many different samples. You are not alone. A
want to see what is going on. The results of my researcher rarely draws many samples. He or she

Red White Number of Samples

42 58 1
43 57 1
45 55 2
46 54 4
47 53 8
48 52 12 Number of red and white marbles that were
49 51 21 randomly drawn from a jar of 5,000 marbles
50 50 31 with 100 drawn each time, repeated 1 30
51 49 20 times for 1 30 independent random samples.
52 48 13
53 47 9
54 46 5
55 45 2
57 43 1
Total 130

Number of Samples

variation in it­
42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 predict specific
Number of RedMarbles in a Sample -ith a great ~

usually draws only one random sample, but the Systematic Sampling. Systematic sampling is
central limit theorem lets him or her generalize simple random sampling with a shortcut for
from one sample to the population. The theorem random selection. Again, the first step is to num­
is about many samples, but lets the researcher ber each element in the sampling frame. Instead
calculate the probability of a particular sample of using a list of random numbers, a researcher
being off from the population parameter. calculates a sampling interval, and the interval
Random sampling does not guarantee that becomes his or her quasi-random selection
every random sample perfectly represents the method. The sampling interval (i.e., 1 in k,
population. Instead, it means that most random where k is some number) tells the researcher
samples will be close to the population most of how to select elements from a sampling frame
the time, and that one can calculate the proba­ by skipping elements in the frame before select­
bility of a particular sample being inaccurate. A ing one for the sample.
researcher estimates the chance that a particular For instance, I want to sample 300 names
sample is off or unrepresentative (i.e., the size of from 900. After a random starting point, I select
the sampling error) by using information from every third name of the 900 to get a sample of300.
the sample to estimate the sampling distribu­ My sampling interval is 3. Sampling intervals are
tion. He or she combines this information with easy to compute. I need the sample size and the
knowledge of the central limit theorem to con­ population size (or sampling frame size as a best
struct confidence intervals. estimate). You can think of the sampling interval
The confidence interval is a relatively simple as the inverse of the sampling ratio. The sampling
but powerful idea. When television or newspa­ ratio for 300 names out of900 is 300/900 = .333 =
per polls are reported, you may hear about 33.3 percent. The sampling interval is 900/300 = 3.
something called the margin of error being plus In most cases, a simple random sample and
or minus 2 percentage points. This is a version of a systematic sample yield virtually equivalent re­
confidence intervals. A confidence interval is a sults. One important situation in which system­
range around a specific point used to estimate a atic sampling cannot be substituted for simple
population parameter. A range is used because random sampling occurs when the elements in a
the statistics of random processes do not let a re­ sample are organized in some kind of cycle or
searcher predict an exact point, but they let the pattern. For example, a researcher's sampling
researcher say with a high level of confidence frame is organized by married couples with the
(e.g., 95 percent) that the true population para­ male first and the female second (see Table 6.2).
meter lies within a certain range. Such a pattern gives the researcher an unrepre­
The calculations for sampling errors or con­ sentative sample if systematic sampling is used.
fidence intervals are beyond the level of this dis­ His or her systematic sample can be nonrepre­
cussion, but they are based on the idea of the sentative and include only wives because of how
sampling distribution that lets a researcher cal­ the cases are organized. When his or her sample
culate the sampling error and confidence inter­ frame is organized as couples, even-numbered
val. For example, I cannot say, "There are sampling intervals result in samples with all hus­
precisely 2,500 red marbles in the jar based on a bands or all wives.
random sample." However, I can say, "I am 95 Table 6.3 illustrates simple random sam­
percent certain that the population parameter pling and systematic sampling. Notice that dif­
lies between 2,450 and 2,550." I can combine ferent names were drawn in each sample. For
characteristics of the sample (e.g., its size, the example, H. Adams appears in both samples, but
variation in it) with the central limit theorem to C. Droullard is only in the simple random sam­
predict specific ranges around the parameter ple. This is because it is rare for any two random
with a great deal of confidence. samples to be identical.

TAB L E 6. 2 Problems with Systematic
Sampling of Cyclical Data
The sampling frame contains 20 males and
20 females (gender is in parenthesis after each
name). The simple random sample yielded 3
TABLE 6.3 C~

males and 7 females, and the systematic sample

yielded 5 males and 5 females. Does this mean
that systematic sampling is more accurate? No.
Abrams, J. (M)
Husband To check this, draw a new sample using different
2 3
Wife random numbers; try taking the first two digits 02 Adams, H. (F)
3 Husband 03 Anderson. H. (I
and beginning at the end (e.g., 11 from 11921,
04 Arminond. L (Il
4 Wife then 43 from 43232). Also draw a new system­
05 Boorstein. A (1
5 Husband atic sample with a different random start. The
06 Breitsprechec I
63 Wife last time the random start was 18. Try a random 07 Brown, D. (F)
start of 11. What did you find? How many of 08 Cattelino. J. (f)
7 Husband
each sex? 09 Cidoni,S. (M)
8 Wife
10 Davis, L (F)
9 Husband Stratified Sampling. In stratified sampling, a 11 Droullard. C (Ij
10 3 Wife researcher first divides the population into sub­ 12 Durette, R. (F)
11 Husband populations (strata) on the basis of supplemen­ 13 Elsnau, K. (F)
12 Wife tary information. After dividing the population 14 Falconer. T. (MIl
into strata, the researcher draws a random sam­ 15 Fuerstenberg.. ~
Random start ~ 2; Sampling interval = 4. ple from each subpopulation. He or she can 16 Fulton, P. (F)
·Selected into sample.
sample randomly within strata using simple ran­ 17 Gnewuch. S. (F)
18 Green, C (M)


TAB L E 6.3 How to Draw Simple Random and Systematic Samples

Goodwanda. T_I
Harris, B. (M)

Excerpt from a RandI

1. Number each case in the sampling frame in the example), ignore the second occurrence.
sequence. The list of 40 names is in Continue until the number of cases in the
alphabetical order, numbered from 1 to 40. sample (lOin our example) is reached. 1501..0 18.5
90122 381
2. Decide on a sample size. We will draw two 25 4. For a systematic sample, begin with a random
67256 13.
percent (1O-name) samples. start. The easiest way to do this is to point
13761 231
3. For a simple random sample, locate a random­ blindlyat the random number table, then take
81994 666
the closest number that appears on the
number table (see excerpt). Before using 79180 r~
random-number table, count the largest samplingframe. In the example, 18 was chosen.
Start with the random number, then count the
07984 4;~
number of digits needed for the sample (e.g.,
with 40 names, two digits are needed; for 100 sampling interval, or 4 in our example,to come
to the first number. Mark it, and then count the
• Numbers that appeared ~
to 999, three digits; for 1,000 to 9,999, four
digits). Begin anywhere on the random number sampling interval for the next number. Continue
table (we will begin in the upper left) and take a to the end of the list. Continue counting the
set of digits (we will take the last two). Mark the sampling interval as ifthe beginning of the list dom or systematic sI
number on the samplingframe that corresponds was attached to the end of the list (like a piing, the researchers
to the chosen random number to indicate that circle). Keep counting until ending close to the each stratum, raiN
the case is in the sample. Ifthe number is too start, or on the start ifthe sampling interval processes control it. 1
large (over 40), ignore it. Ifthe number appears divides evenly into the total of the sampling tiveness or fixes t:bI
more than once (10 and 21 occurred twice in frame.
strata within a sampll

TAB LE 6. 3 Continued

01 Abrams, j. (M) 21 Hjelmhaug, N. (M)

02 Adams, H. (F) Yes Yes (6) 22 Huang, j. (F) Yes Yes (1 )
03 Anderson, H. (M) 23 Ivono, V. (F)
04 Arminond, l. (M) 24 jaquees, j. (M)
05 Boorstein, A. (M) 25 johnson, A. (F)
06 Breitsprecher, P. (M) Yes Yes (7) 26 Kennedy, M. (F) Yes (2)
07 Brown, D. (F) 27 Koschoreck, l. (F)
08 Cattelino, j. (F) 28 Koykkar, j. (M)
Cidoni, S. (M)
Davis, L. (F) Yes
. Yes (8)
Kozlowski, C. (F)
Laurent,j. (M)
Yes (3)
11 Droullard, C. (M) Yes 31 Lee, R. (F)
12 Durette, R. (F) 32 Ling, C. (M)
13 Elsnau, K. (F) Yes 33 McKinnon, K. (F)
14 Falconer, T. (M) Yes (9) 34 Min, H. (F) Yes Yes (4)
15 Fuerstenberg, j. (M) 35 Moini, A. (F)
16 Fulton, P. (F) 36 Navarre, H. (M)
17 Gnewuch, S. (F) 37 O'Sullivan, C. (M)
18 Green, C. (M) START, 38 Oh,j. (M) Yes (5)
Yes (10) 39 Olson, j. (M)
19 Goodwanda, T. (F) Yes 40 Ortiz y Garcia, L. (F)
20 Harris, B. (M)

Excerpt from a Random-Number Table (for Simple Random Sample)

1501.0 18590 00102 4221.0 94174 22099

90122 3822l 21529. ooon 0473A 60457
67256 13887 941ll 11077 01061 27779
13761 23390 12947 21280 445ilii 36457
81994 666ll 16597 44457 0762l 51949
79180 25992 46178 23992 62108 43232
07984 47169 88094 82752 15318 11921

• Numbers that appeared twice in random numbers selected.

dam or systematic sampling. In stratified sam­ supplemental information about strata is not al­

pling, the researcher controls the relative size of ways available.

each stratum, rather than letting random In general, stratified sampling produces

processes control it. This guarantees representa­ samples that are more representative of the pop­

tiveness or fixes the proportion of different ulation than simple random sampling if the stra­

strata within a sample. Of course, the necessary tum information is accurate. A simple example


illustrates why this is so. Imagine a population later chapter) oversampled African Americans.
that is 51 percent female and 49 percent male; A random sample of the U.S. population yielded
the population parameter is a sex ratio of 51 to 191 Blacks. Davis and Smith conducted a sepa­
49. With stratified sampling, a researcher draws rate sample of African Americans to increase the
random samples among females and among total number of Blacksto 544. The 191 Blackre­
males so that the sample contains a 51 to 49 per­ spondents are about 13 percent of the random
cent sex ratio. If the researcher had used simple sample, roughly equal to the percentage of
random sampling, it would be possible for a ran­ Blacks in the U.S. population. The 544 Blacksare
dom sample to be off from the true sex ratio in 30 percent of the disproportionate sample. The
the population. Thus, he or she makes fewer er­ researcher who wants to use the entire sample
rors representing the population and has a must adjust it to reduce the number of sampled
smaller sampling error with stratified sampling. African Americans before generalizing to the SuIf~
Researchers use stratified sampling when a U.S. population. Disproportionate sampling
stratum of interest is a small percentage of a helps the researcher who wants to focus on is­
population and random processes could miss sues most relevant to a subpopulation. In this ~~
the stratum by chance. For example, a researcher case, he or she can more accurately generalize to
draws a sample of 200 from 20,000 college stu­ African Americans using the 544 respondents
~ tftilllioii'"
dents. He or she gets information from the col­ than using a sample of only 191. The larger sam­
lege registrar indicating that 2 percent of the ple is more likely to reflect the full diversity of ~
20,000 students, or 400, are divorced women the African American subpopulation. Clets
with children under the age of 5. This group is ~fII;l/I
important to include in the sample. There would Cluster Sampling. Cluster sampling addresses
be 4 such students (2 percent of200) in a repre­ two problems: Researchers lack a good sampling ~stI
sentative sample, but the researcher could miss frame for a dispersed population and the cost to
them by chance in one simple random sample. reach a sampled element is very high. For exam­
With stratified sampling, he or she obtains a list ple, there is no single list of all automobile me­ Iow~-xl.
of the 400 such students from the registrar and chanics in North America. Even if a researcher ~T.iE!H '}L.
randomly selects 4 from it. This guarantees that got an accurate sampling frame, it would cost T~~..-

the sample represents the population with re­ too much to reach the sampled mechanics who
gard to the important strata (see Box 6.3). are geographically spread out. Instead of using a
In special situations, a researcher may want single sampling frame, researchers use a sam­
the proportion of a stratum in a sample to differ pling design that involves multiple stages and
from its true proportion in the population. For clusters.
example, the population contains 0.5 percent
Aleuts, but the researcher wants to examine
Aleuts in particular. He or she oversamples so
that Aleuts make up 10 percent of the sample.
With this type of disproportionate stratified
A cluster is a unit that contains final sam­
pling elements but can be treated temporarily as
a sampling element itself.A researcher first sam­
ples clusters, each of which contains elements,
then draws a second sample from within the
.._.-r" ­ ......
¥' i
,­ ..

sample, the researcher cannot generalize directly clusters selected in the first stage of sampling. In ia. . . or R:ldI. .1II
from the sample to the population without spe­ other words, the researcher randomly samples _-\R:kada
cial adjustments. clusters, then randomly samples elements from -dRaa-s."II&-.•
In some situations, a researcher wants the
proportion of a stratum or subgroup to differ
within the selected clusters. This has a big prac­
tical advantage. He or she can create a good sam­ ,,_,5
1 . . . . . . . ' -. . . .
. -
from its true proportion in the population. For pling frame of clusters, even if it is impossible to . . .... ~d.-II:I;.l"
example, Davis and Smith (1992) reported that create one for sampling elements. Once the re­ . .ftc Mi'-­
the 1987 General Social Survey (explained in a searcher gets a sample of clusters, creating a --'for'c i 5


Administrators 15 2.88 1 3 -2
Staff physicians 25 4.81 2 5 -3
Intern physicians 25 4.81 6 5 +1
Registered nurses 100 19.23 22 19 +3
Nurse assistants 100 19.23 21 19 +2
Medical technicians 75 14.42 9 14 +5
Orderlies 50 9.62 8 10 -2
Clerks 75 14.42 5 14 +1
Maintenance staff 30 5.77 3 6 -3
Cleaning staff 25 4.81 3 5 -2
-­ --­ -­ -­
Total 520 100.00 100 100

Randomly select 3 of 1 5 administrators. 5 of 25 staff physicians, and so on.

Note: Traditionally, N symbolizes the number in the population and n represents the number in the sample.

The simple random sample overrepresents nurses, nursing assistants, and medical technicians, but underrepresents

administrators, staff physicians, maintenance staff, and cleaning staff. The stratified sample gives an accurate representation

of each type of position.

sampling frame for elements within each cluster individuals from Mapleville. First, he or she ran­
becomes more manageable. A second advantage domly samples city blocks, then households
for geographically dispersed populations is that within blocks, then individuals within house­
elements within each cluster are physically closer holds (see Box 6.4). Although there is no accurate
to one another. This may produce a savings in list of all residents of Mapleville, there is an accu­
locating or reaching each element. rate list of blocks in the city. After selecting a ran­
A researcher draws several samples in stages dom sample of blocks, the researcher counts all
in cluster sampling. In a three-stage sample, stage households on the selected blocks to create a
1 is random sampling of big clusters; stage 2 is sample frame for each block. He or she then uses
random sampling of small clusters within each the list of households to draw a random sample
selected big cluster; and the last stage is sampling at the stage of sampling households. Finally, the
of elements from within the sampled small clus­ researcher chooses a specific individual within
ters. For example, a researcher wants a sample of each sampled household.

Cluster sampIi
than simple r~
curate. Each stage ..
sampling errors. Th
Goal: Draw a random sample of 240 people in Mapleville. sample has more 51
Step 1: Mapleville has 55 districts. Randomly select 6 districts. stage random samP
1 2 3* 4 5 6 7 8 9 10 11 12 13 14 15* 16 17 18 1920 21 22 23 2425 26 A researcher 1
27* 282930 31* 323334353637383940* 4142434445464748 must decide the II
4950 51 525354* 55 number of element!
* = Randomly selected. ample, in a two-stall
Step 2: Divide the selected districts into blocks. Each district contains 20 blocks. Randomly select 4 blocks ple from Mapleei
from the district. randomly select 1]
Example of District 3 (selected in step 1): ments from each. a
and select 120 ekmi
1 234* 5 678910* 11 12 13* 14 15 16 17* 18 1920
The general aJ1S'lo-~
* = Randomly selected.
clusters is better. ~
Step 3: Divide blocks into households. Randomlyselect households. clusters (e.g., peopI
Example of Block 4 of District 3 (selected in step 2): tend to be similar •
Block 4 contains a mix of single-family homes, duplexes, and four-unit apartment buildings. It is bounded by the same block ~
Oak Street, RiverRoad,South Avenue, and GreenviewDrive. There are 45 householdson the block. Randomly on different blocks)
select 1 a households from the 45. many similar e1emll
would be less repR:lll
1 #1 Oak Street 16 31 *
tion. For example.:
2 #3 Oak Street #154 River Road 32'
two blocks with rei
3* #5 Oak Street 18 # 1 56 River Road 33
4 19* #158 River Road 34 # 1 56 Greenview Drive
draw 120 people fnl
5 20* 35* representative than]
6 21 #13 South Avenue 36 city blocks and .1 inII
7 #7 Oak Street 22 37 When a researd
8 23 #11 South Avenue 38 ographical area andl
9* #150 River Road 24 #9 South Avenue 39 #158 Greenview Drive cluster sampling •
10* 25 #7 South Avenue 40 costs. As usual, tbeIll
11 26 # 5 South Avenue 41 racy and cost.
12 27 #3 South Avenue 42 For example. A
13 #152 River Road 28 #1 South Avenue 43 #160 Greenview Drive each plan to visit •
14 29* 44 sample of 1,500 ~
15 30 #152 Greenview Drive 45 ulation of all colle!!lli
Alan obtains an aal
* = Randomly selected.
students and uses si
Step 4: Select a respondent within each household. travels to 1,000 ditl
Summary of cluster sampling: one or two stu~
random sample of
1 person randomly selected per household
1 a households randomly selected per block
3,000 colleges, .
4 blocks randomly selected per district 500 students from ~
6 districts randomly selected in the city dom sample of ~
1 X 10 X 4 6 = 240 people in sample and selects 5 studell
erage $250 per loIj

Cluster sampling is usually less expensive $250,000, Ricardo's is $750, and Barbara's is
than simple random sampling, but it is less ac­ $75,000. Alan's sample is highly accurate, but
curate. Each stage in cluster sampling introduces Barbara's is only slightly less accurate for one­
sampling errors. This means a multistage cluster third the cost. Ricardo's sample is the cheapest,
sample has more sampling errors than a one­ but it is not representative at all.
stage random sample.
A researcher who uses cluster sampling Probability Proportionate to Size (PPS). There
must decide the number of clusters and the are two methods of cluster sampling. The
number of elements within each cluster. For ex­ method just described is proportionate or un­
ample, in a two-stage cluster sample of240 peo­ weighted cluster sampling. It is proportionate
ple from Mapleville, the researcher could because the size of each cluster (or number of
randomly select 120 clusters and select 2 ele­ elements at each stage) is the same. The more
ments from each, or randomly select 2 clusters common situation is for the cluster groups to
and select 120 elements in each. Which is best? be of different sizes. When this is the case, the
The general answer is that a design with more researcher must adjust the probability or sam­
clusters isbetter. This is because elements within pling ratio at various stages in sampling (see
clusters (e.g., people living on the same block) Box6.5).
tend to be similar to each other (e.g., people on The foregoing cluster sampling example
the same block tend to be more alike than those with Alan, Barbara, and Ricardo illustrates the
on different blocks). If few clusters are chosen, problem with unweighted cluster sampling. Bar­
many similar elements could be selected, which bara drew a simple random sample of 300 col­
would be less representative of the total popula­ leges from a list of all 3,000 colleges, but she
tion. For example, the researcher could select made a mistake-unless every college has an
two blocks with relatively wealthy people and identical number of students. Her method gave
draw 120 people from each. This would be less each college an equal chance of being selected­
representative than a sample with 120 different a 300/3,000 or 10 percent chance. But colleges
city blocks and 2 individuals chosen from each. have different numbers of students, so each stu­
When a researcher samples from a large ge­ dent does not have an equal chance to end up in
ographical area and must travel to each element, her sample.
cluster sampling significantly reduces travel Barbara listed every college and sampled
costs. As usual, there is a tradeoff between accu­ from the list. A large university with 40,000 stu­
racy and cost. dents and a small college with 400 students had
For example, Alan, Ricardo, and Barbara an equal chance of being selected. But if she
each plan to visit and personally interview a chose the large university, the chance of a given
sample of 1,500 students who represent the pop­ student at that college being selected was 5 in
ulation of all college students in North America. 40,000 (5/40,000 = 0.0125 percent), whereas a
Alan obtains an accurate sampling frame of all student at the small collegehad a 5 in 400 (5/400
students and uses simple random sampling. He = 1.25 percent) chance of being selected. The
travels to 1,000 different locations to interview small-college student was 100 times more likely
one or two students at each. Ricardo draws a to be in her sample. The total probability of be­
random sample of three collegesfrom a list of all ing selected for a student from the large univer­
3,000 colleges, then visits the three and selects sity was 0.125 percent (10 X 0.0125), while it
500 students from each. Barbara draws a ran­ was 12.5 percent (10 X 1.25) for the small­
dom sample of 300 colleges. She visits the 300 college student. Barbara violated a principle of
and selects 5 students at each. If travel costs av­ random sampling-that each element has an
erage $250 per location, Alan's travel bill is equal chance to be selected into the sample.

Sampling has many terms for the different parts of county or SMSA. This gave them a sample of 84 Vaquera and Kao (2
samples or types of samples. A complex sample illus­ counties or SMSAs. tion among adolesa
trates how researchers use them. Look at the 1980 For the second stage, the researchers identified were either from the
sample for the best-known national U.S.survey in so­ city blocks, census tracts, or the rural equivalent in Their data were from
ciology, the General Social Survey. each county or SMSA. Each sampling element (e.g., adolescent health I
The population is defined as all resident adults city block) had a minimum of 50 housing units. In or­ through 12 in 80
(1 8 years or older) in the U.S. for the universe of all der to get an accurate count of the number of hous­ schools. There were
Americans. The target population consists of all Eng­ ing units for some counties, a researcher counted schools. After the s
lish-speaking adults who livein households, excluding addresses in the field. The researchers selected 6 or mately 200 studenl
those living in institutional settings such as college more blocks within each county or SMSA using PPS from within those sci
dormitories, nursing homes, or military quarters. The to yield 562 blocks. the school, and stud
researchers estimated that 97.3 percent of all resi­ In the third stage, the researchers used the the school. Because t
dent adults lived in households and that 97 percent household as a sampling element. They randomly se­ size, ranging from 11
of the household population spoke sufficient English lected households from the addresses in the block. thors adjusted usinf;
to be interviewed. After selecting an address, an interviewer contacted size (PPS). They fau
The researchers used a complex multistage prob­ the household and chose an eligible respondent from dents had a relatiom
ability sample that is both a cluster sample and a it. The interviewer looked at a selection table for pos­ site sex in the Pf"et
stratified sample. First, they created a national sible respondents and interviewed a type of respon­ Blacks were more iii
sampling frame of all U.S.counties, independent cities, dent (e.g., second oldest) based on the table. In ships (90 percent) (]
and Standard Metropolitan Statistical Areas (SMSAs), total, 1,934 people were contacted for interviews ics (70 percent). Th
a Census Bureau designation for larger cities and sur­ and 75.9 percent of interviews were completed. This mixed-race couples d
rounding areas. Each sampling element at this first gave a final sample size of 1,468. We can calculate affection, but the inti
level had about 4,000 households. They divided the sampling ratio by dividing 1,468 by the total to do so in public thl
these elements into strata. The strata were the four number of adults living in households, which was
major geographic regions as defined by the Census about 150 million, which is 0.01 percent. To check
Bureau, divided into metropolitan and nonmetropol­ the representativeness of their sample, the re­
itan areas. They then sampled from each strata using searchers also compared characteristics of the sam­ Three kinds ot
probability proportionate to size (PPS) random selec­ ple to census results (see Davis and Smith, 1992: sampling frame iu
tion, based on the number of housing units in each 31-44). without telephooa
moved, and prop
Those without phQ
ucated, and transa
If Barbara uses probability proportionate to with 40,000 students will be 100 times more phone interview stI
size (PPS) and samples correctly, then each final likely to be selected than one with 400 students. general public ",itb
sampling element or student will have an equal (See Box 6.6 for another example.) cent in advanced •
probability of being selected. She does this by percentage of the PI
adjusting the chances ofselecting a college in the Random-Digit Dialing. Random-digit dialing creased, the perces
first stage of sampling. She must give large col­ (RDD) is a special sampling technique used in has also grown. Se"Il
leges with more students a greater chance of be­ research projects in which the general public is listed numbers: rc­
ing selected and small colleges a smaller chance. interviewed by telephone. It differs from the tra­ lection agencies; tt.
She adjusts the probability of selecting a college ditional method of sampling for telephone in­ want privacy and ,
on the basis of the proportion of all students in terviews because a published telephone directory salespeople, and pII
the population who attend it. Thus, a college is not the sampling frame. eas, the percentag

high as 50 percent. In addition, people change

their residences, so directories that are published
annually or less often have numbers for people
who have left and do not list those who have re­
Vaquera and Kao (2005) studied displays of affec­ cently moved into an area. Plus, directories do
tion among adolescent couples in which the couple not list cell phone numbers. A researcher using
were either from the same or differentracial groups. RDD randomly selects tel