___ TAKING YOU STEP ga2 THE RESEARCH PROGESS / PO aa OUea NYY PSYCHOLOGY IWLINSWIa ad Xa ANNE Myers (Osc UGHeA ADO MRAFIFTH EDITION EXPERIMENTAL PSYCHOLOGY ‘Anne Myers Christine H. Hansen Oakland University WADSWORTH ——— igopore ® Spain # Unied Kingiom « United SatesWADSWORTH Publisher Vit Knight Manuscript ator: Robin Gad dito Assistant: le Dillemath and Permissions Editor Mary Kay Poiemen ‘Den Moneypenny Cover Design: Dense Davidion Marketing Manager Joome Teraar Cover lustaton: Een Photography Marketing Asstt fasting Fergison Frnt Bayer Neny Panciera ‘sistant EaeorJemer Wilkinson ‘Composit GEX Pablishing Services roject Eitr Kirk Bont Priming and Binding: RR Dowel, Production Service: Forbes Mil Press ‘Cranford COPYRIGHT © 2002 Wadsworth Group, Wadeworth isan imprint ofthe Wadsworth Group, SGiion of Thomson Leaming Ine Thomson Lesring™ isa trademark sed berln under oer. For mor infomation abot hs o any ther Wadsworth product, coma WADSWORTH 311 Forest Lodge Road Pace Grove, CA 93950 USA form wadaworth.com T7p0-4250563 (Thomson Learing Academie Resource Cente) ALL RIGHTS RESERVED. No part ofthis work covered by the copyright Keon may be rolced or edi any form or by any meene— grape eke or nechanial. nla Ter photocopying. econing. ang. Web dsttbuton, of iormaton sorage and reel ‘Suton tout the waten persion ofthe publisher For pemision to use mae frm dhs work conc! sy Wed wwethomsonnghiscom Fax, 18007302215, Phone! 1-900-730.2314 Printed in he United States of Americ woe Tes 4 Leary of Congres Cataloging Puleaton Daa yer me periment pycology Anne Myer, Chrisie H.Harsen~ Sth inte biti references and index, tsanosseseioe3 1 Paycholoy, Experiment Hansen, Cistine WT BIBI. MB5 2001 150.7 24-de2 2001035756 ‘Te my Mom, for prayer and laughter. AM. To RDH, my favorite colleague CHPREFACE Goals of the Text Experimental Peychology, Pith Edition, is an introduction to the basic princi ples of research in psychology. It explains the Key principles of esearch, par ticularly experimental research, clearly and within the context of concrete examples, It teaches students how to design and execute an experiment, ana- Iyze and interpret the results, and write a research repor. Although the main focus is on experimentation, many alternative approaches are discussed as ‘important complements to controlled laboratory experiments. This text was designed to be as comprehensive as possible—without over ‘whelming the beginning researcher. The principles of experimentation and the skepticism of the scientific approach are concepts that are new to students and ran counter to their commonsense notions about causal inference; for most psychology students, mastering experimental methods requires making ‘a quantum leap from their original ideas about psyehology as well as chal lenging them to leam an entirely new language. This text has been designed to address these issues and provide practical solutions to them. It was writ ten with flexibility in mind forthe instructor as well, Each chapter can stand fon its own, and instriciors can select text assignments to match individual course content Special Features of the Text This text introduces the experimental process in a structured way that allows students to gain a thorough grasp of the scientific method. First, it Is orga nized to carry students through the entre process of conducting an exper- ment. The major sections—Introduction, Method, Results, and Discussion— parallel the major sections of the research report in order to clarify the relationships among designing, conducting, and reporting the experiment. The Publication Manual of the American Psychological Assocation, Fifth Edition—_—_—_—_—_—__— PREFACE (2000), includes a numberof ehanges and addons along with new equ Gen and recommendations. Throughout the text, we have inchded new ‘aerial that inconporstes Use changes Second, many pacal ais ae provided. Research ethics ae discussed in deta ae ect tecniques for developing a teserch hypothesis In resenting ecasch methods, we have sessed the fnteal relationship among the expetmenal hypotbess, the research design, andthe satis analysis The process of seeing a design fas been broken down into basic sepst0 Provkle stctue fr te student. A dtd chapter on report wing incdes P'Srmple journal ale o lsat the mast current reporting conventions {ro ad scrdens th producing an APA-syle repr, the manoserpt version of thivarcl is eprodaced in Appendix C) The raucrale behing al procedures 1D Sagnined wohelp students apply them. Important tems ae introduced in bolder ope throughou the tex and are Inted atthe end ofeach chapter. Each chap aso includes a summary nd review and study questions. New totus elton are chapter objectives, erical thinking excess, and online Tespurss At the end of he book, andor number table (Appenix B), gos Sp, and index ate Inluded ‘fur exampes ate drawn from many diferent research ates emp sete inporare of ound veench methods troughs A ssc ner of pocholgy. The examples, both clase and euren, provide cle, Pare Mcmeatns ofthe conepts under dacsson, The eect choice of Samples ges strc the freedom to supplement the text with conten nented readings in areas of thelr chote rial mates included to help students fterpretreseach ndings ‘Te rls ston ofthe text provides students wih a conceptual overview Sith proces oss iene and spb sep isco for fag and crying ot some ofthe tests commonly used n simple experiments ae enns are revewed, snd sats! tables ae inchaded (Append B)s0 that al de equed information is avalable in ds singe source. The proces of iarpeing ar desciing staal results diseuced in de Organization of the Fifth Edition “Those who used the text in che fourth edition wil find that the overall plan and focus ofthe book have remained unchanged. Many interesting new exam ples have been nchided throughout. Some topics have been vpdated and Expanded in response to reviewer and user feedback few topics have been. faded. In Part 1 Introduction, Chapter 1 focuses on introducing the student {o the need for scientific explanations of behavior and now includes more real ‘world examples of nonscientfic inference as well as added coverage of fal Cation. Chapter 2 presents extensive coverage of research ethics and includes broadened coverage of informed consent, consent forms, and scientific miscon uct, An interesting new case study on OCD and increased discussion of mail fand Internet surveys have been added to the nonexperimental approaches dis- cussed in Chapter 3. Chapter 4 includes correlational and quasi-2xperimental PREFACE desig and has been expanded to include more information about neat and multiple regression techniques. Computation ofa simple comelation hasbeen tied to Appendix A. Chiper 5 teches sens the bass of forming fypotes now includes expanded coverage of ela-analyss andthe use of computer dates forthe Heat sarchee Th Far 2, Method, more information on reliably and validty has been included in Chapter 6, which teaches the bases of experentation. roc Gres for controling extaneousvarcls,inclting many prac ps, have teen moved forward and ate now fond in Chapter 7 Chapes 8 and9 fos on betwoen-subess designs: two group design, mle group desgns, and facials, Expanded discussion of elle sze estimates can now be found In Chapter 8. Chapter 9 has expanded coverage of interactions, Chaper 10 Gestries wisubecte and nee desis ant now inte more ‘noe mution about coumterbelaningtechaiques, including techniques fo con Sructng else Latin Squares and bok andomiaton of ese com Giions Chapter 11 focuses on small N designe, It now Includes more Giseusion ot the pros and cone of stall Nand large N designs and more Caviadons of ABA desig in far 3, Results: Coping with Data many nw examples have been added tothe already wellecened chapters on sats. Chapter 12 dmses ypoth tsi testng probably deen errors, measines of cena tendency. and va Shee, and ic now includes infomation above comping mean sore. Chap ter 13 concenates on the selection and interpretation of sais fo (90 {Soup ces, and we have augoented sats hypothe eing wh fle Eezsand confidence neve, Chapter 14 focuses ot anand two aay Ses of vatance, The chapter fas been updated to rele more sophisticated ANOVA techniques and Includes expanded coverage of eft ses tn ar 4, Disusion, Chapter 15 now inches even mote practical infor ration for sens sus tnerpretng Bndings end evant ls om Sais tests Ir also includes more discussion ofthe ned to examine ale. tative expansion for ndings Chaper 16 festres sn nesting ne 2 Bie ance Cefecs of a Brit Witvatona Intervension with College Student Brunke), by Bran Born and late Cavey whichis annotated deal o help students write their own APAstjle report, We think you wil find the fit edition even more comprehensive than beloe—but sll user iendly Acknowledgments Many people contributed to the development of this exbook. We are espe- cially gratefol to Robert D. Nye, who served as an untiring sounding board in the early days, Howard Cohen, James Halpern, David Schiflman, Mark Sherman, and Jodi Solomon deserve special mention for reading portions of the original manuscript, Phyllis Freeman, Joanne Green, Zanvel Lif, Barbara Novick, David Morse, Robert Prsbie, Richard Slaon, and Carol Vazquez also helped, Special thanks to Dave Carroll, Donna Lewandowski, Deanna Hall, Andrea Kozak, Mark Hoheisel, Gerth’Preuthun, Cynthia Shantz, Jo AnnPREFACE ‘Swaney, and Jane Youngs, who read and commented on previous editions from a students and teaching assistant’ point of view: Special thanks to all the students in Research Methods courses who read and commented on the fourth edition, particularly those stedents from across the country who took the time and energy to write to us. We hope we have Clinfed aey ambiguities ard given you interesting new examples and car toons. Very special thanks to all our research and teaching assistants for the snumerous literature searches you conducted for the fifth edition, especially ‘Michelle Pelker and Patrick Fairclath, who contributed so much to this man tseript. Thanks, also, to three methodological wizards—Bill Crano, Randy Hansen, and Latty Messé—as well as Don Campbell and his colleagues, who hhave taught so many generations of students. We are also deeply indebred to the many researchers whose work inspired much of this text and to the many futhors and publishers who permitted reproductions of portions of their Grorke. They ate cited throughout the text, Particular thanks to Robert A, Baron and Robert Rosenthal, whose work fils so many pages, and especially to Robert Zajonc, cher ami, or his imagination and good humor. We are grate- fal to the literary executor ofthe late it Ronald A. Fisher, FRS., to Dr. Frank Yates, FRS., and to the Longman Group UK, Lid. for permission to reprint portions of statistical tables Tables BI, B2, and B3 in our Appendix B). And { Sidney Haris, immeasurable thanks for the new cartoons! They brought new smiles to our faces. ‘We gratefully acknowledge the contributions of Barbara S. Chaparro, \Wiehita Sate University, Danna Dahlgren, Indiana University-Southeast; Lauren Freedman, Montclair State University, William Kelemen, University of Misour, ‘5 Louis, Richard Topolski, Augusta Sate University, Ronald R. Ulm, Salisbury Suate University, Luis A. Veg, Califomia State University, Bakersfield; Burrton Woodnil Butler Univesity. and Cita Zinser, East Tennessee State University ‘who reviewed the fourth edition. Their constructive suggestions improved the book greatly nally, we would also like to thank the people at Wadsworth for thei careful handling ofthe revision, patticularly Vicki Knight, publisker, Joanne Terhaar, marketing manager; Jennifer Wilkinson, assistant editor, Julie Dille uth and Dan Moneypenny, editorial assistants; Justine Ferguson, market- ing assistant, Kirk Bomont, project editor, Nancy Panziera, print buyer, and Veron Boes, design director. Finally, out very special thanks to Robin Gold at Forbes Mil Press for her skills and patience (and her unflagging good spirits) Anne Myers Christine Hansen BRIEF CONTENTS part one INTRODUCTION 1 1 Experimental Psychology and the Scientific Method 2 2 Research Ethics 27 3 Altemauves to Experimentation: Nonexperimental Designs 53 4 Atematves to Experentaton:Coneltional and Quasi-Experimental Designs 87 5 Formulating the Hypothesis 117 rant two METHOD 141 6 The Basics of Experimentation 142 7 Solving Problems: Controlling Extraneous Variables 186 8 Basie Between-Subjects Designs 212 9 Berween-Subjects Factorial Designs 244 10 Within Subjects Designs 267 LL Within-Subjects Designs: Small N 297 rant tueee RESULTS: COPING WITH DATA 319 12 Why We Need Stauistics 320 13 Analyzing Results: Two Group Examples 359) 14 Analyzing Results: Mulkiple-Groups and Factorial Experiments 387 rant rour DISCUSSION 427 15 Drawing Conclusions: The Search forthe Elusive Bottom Line 428 16 Writing the Research Report 453499 ‘4 Computational Form ® Statistical Tables 495 € A Journal Article in Manuscript Form Glossary References Index CONTENTS. INTRODUCTION 1 EXPERIMENTAL PSYCHOLOGY AND THE SCIENTIFIC METHOD 2 ‘The Need for Sciemtific Methodology 4 The Characteristics of Modern Science 9 The Tools of Psychological Science 15 Scientific Explanation in Psychological Science 19 ‘The Organization of the Text 22 Summary 23 Key Terms 24 Review and Study Questions 25 Critical Thinking Exercise 26 Online Resources 26 CHAPTER 2 RESEARCH ETHICS 27 Research Ethics 28 ‘The American Psychological Association Guidelines 31contents Protecting the Welfare of Animal Subjects 38 Fraud in Science 46 Plagiarism — 48 Summary 49 Key Terms 50 Review and Study Questions 51 Critical Thinking Exercise 52 Online Resources 52 cuAprER 3 ALTERNATIVES TO EXPERIMENTATION: NONEXPERIMENTAL DESIGNS 53 Describing Research Activities 55 Nonexperimental Approaches 56 Sampling 75 Summary 82 Key Terms 83 Review and Study Questions 85 Critical Thinking Exercise 86 Online Resources 86 ALTERNATIVES TO EXPERIMENTATION: CORRELATIONAL AND QUASI-EXPERIMENTAL DESIGNS 87 Correlational Designs 90 Quasi-Experimental Designs 102 Summary 113 Key Terms 114 Review and Study Questions 115 Critical Thinking Exercise 116 Online Resources. 116 CONTENTS xt FORMULATING THE HYPOTHESIS 117 The Characteristics of an Experimental Hypothesis 119 The Inductive Model 123 The Deductive Model 124 Combining Induction and Deduction 125 Building on Prior Research 126 Serendipity and the Windfall Hypothesis 127 Intuition — 128 When All Else Fails 131 Searching the Research Literature 132 Summary 136 Key Terms 137 Review and Study Questions 137 Critical Thinking Exercise 138 Online Resources. 139 METHOD 141 cuAPrER 6 THE BASICS OF EXPERIMENTATION 14? Independent and Dependent Variables 144 Operational Definitions 151 Evaluating Operational Definitions 161 Evaluating the Experiment: Internal Validity 168 Planning the Method Section 178 Summary 181 Key Terms 182 Review and Study Questions 184CONTENTS Critical Thinking Exercise 185 Online Resources 185 cwAPTER T SOLVING PROBLEMS: CONTROLLING EXTRANEOUS VARIABLES — 186 Physical Variables 187 Social Variables 191 Personality Variables 203, Context Variables 206 Summary 208 Key Terms 209 Review and Study Questions 210 Critical Thinking Exercise 211 Online Resources 211 BASIC BETWEEN-SUBJECTS DESIGNS 212 Selecting and Recruiting Subjects 214 (One Independent Variable: Two-Group Designs 217 ‘wo Independent Groups 218 ‘Two Matched Groups 229 Multiple Groups 233 Summary 239 Key Terms 241 Review and Study Questions 242 Critical Thinking Exercise 243 Online Resources 243 BETWEEN-SUBJECTS FACTORIAL DESIGNS 244 ‘More Than One Independent Variable 245 CONTENTS Laying Out a Factorial Design 248 A Research Example 253 Choosing a Between-Subjects Design 259 Summary 264 Key Terms 265 Review and Study Questions 265 Critical Thinking Exercise 266 Online Resources 266 WITHIN-SUBJECTS DESIGNS — 267 ‘A Within-Subjects Experiment: Perceptual Bias for Forward Motion 270 Within Subjects Factorial Designs 272 Mixed Designs 273 Advantages of Within-Subjects Designs 274 Disadvantages of Within-Subjects Designs 276 Controlling Within-Subjects Designs 278 How Can You Choose a Design? 291 Summary 292 Key Terms 293 Review and Study Questions 204 Critical Thinking Exercise 296 Online Resources 296 cuAPTER 11 WITHIN-SUBJECTS DESIGNS: SMALL N 297. ‘Small N Designs 298 ABA Designs 303 ‘Multiple-Baseline Design 309 Discrete Trials Designs 311 ‘When to Use Large Nand Small N Designs 312aot CONTENTS Summary 315 Key Terms 316 Study Questions 317 Critical Thinking Exercise 318 Online Resources 318 Review an RESULTS: COPING WITH DATA 319 WHY WE NEED STATISTICS 320 Weighing the Evidence 322 Statistical Inference: An Overview 323 Applying Statistical Inference: An Example 330. ‘The Odds of Finding Significance 339 ‘Test Statistics 344 Organizing and S warizing Data 345 Summary 354 Key Ta 335 Review and Study Questions 357 Critical Thinking Exercise 358 Online Resources 358 ANALYZING RESULTS: TWO GROUP EXAMPLES 359 Which Test Do I Use? 360 The Chi Square Test 363 The (Test 368 Summary 382 Key Terms 383 CONTENTS. awit Review and Study Questions 383 Critical Thinking Exercise 386 0: Resources 386 cCMAPTER 14 ANALYZING RESULTS: MULTIPLE-GROUPS AND FACTORIAL EXPERIMENTS 387 Analysis of Variance 388 Sources of Variability 389 A One-Way Between-Subjects Analysis of Variance 392 ‘One-Way Repeated Measures Analysis of Variance 405 Analyzing Data from a Between-Subjects Factorial Experiment 405 A Two-Way Analysis of Variance 408 Repeated Measures and Mixed Factor Designs 419 Summary 421 Key Terms 422 Review and Study Questions 423 Critical Thinking Exercise 426 Online Resources 426 DISCU a7 DRAWING CONCLUSIONS: THE SEARCH FOR THE ELUSIVE BOTTOM LINE 428 Evaluating the Experiment from the Inside: Internal Validity 430 Taking a Broader Perspective: The Problem of External Validity 435 Handling a Nonsignificant Outcome 45so CONTENTS Summary 449 Key Terms 450 Review and Study Questions 450 Critical Thinking Exercise 452 Online Resources 452 WRITING THE RESEARCH REPORT 453 ‘The Written Report: Purpose and Format 454 Major Sections 456 Looking at a Journal Article 464 Preparing Your Manuscript: Procedural Details 478 Making Revisions 481 Summary 484 Key Terms 484 Review and Study Questions 485 Critical Thinking Exercise 486 Onling Resources 487 APPENDICES ‘A Computational Formulas 489 B Statistical Tables 495 CA Journal Article in Manuscript Form 505 Glossary 531 References 545 Index PoavAscoR oo]: I Introduction (CuaPTER 1 +) Experimental Psychology the ‘Cuapter 2 + Research Ethics ciemtific Method ‘Chapter 3 + Alternatives to Experimentation: Nonexperimental Designs Graeren 4 + Alternatives to Experimentation: Correlational and Quast Experimental Designs CempreR 5 + Formulating the HypothesisCHAPTER 1 Experimental Psychology and the Scientific Method ‘The Need for Scent Methodology Nonsense Sores of Disa NonscetifcInrence The Characteristics of Modern Science he Scene Mey Gatberng Emp Dae Seskng Genes Pencils Good inking Seconection Pablczng Ress Fepleation “The Tools of Faychoogcl Science Otverraion Measurement Eperimeration Sclntc Explanation in PychologelSeence Identyng Antecedents Conitons Compe tesieent Cris “he Fjchlogy Experiment Esablahing Cus tnd ct Necessity verse Sufficene condions The Organization ofthe Text Setunaty "erm Review and Study Questions Creal Thinking Exerdse Online Resomees ee (CHAPTER 1 EXPERIMENTAL PSYCHOLOGY AND THE IENTIFIC METHOD CHAPTER OBJECTIVES * Understand why we rely on scientific methods rather than common sense to explain behavior + Learn the principles ofthe scientific method * Learn the basic tools of psychological research * Understand how “cause and effec is explained by experimentation T_T Bctoiogy iste scence of tehavor As psyhcogs, we ke a scien tc apprach to understanding behavior, our knowledge abot poboinl Processes is based on scientific evidence accumulated through earch we Ens, we rely on scientific methods when we conduct payelone eas Such as specifying the conditions under which we nae oat ose” chserving in 8 systematic or otderly way, and accepting ox mca tive explanations of behaviors on the basis of what we oben sei Tesarch about the psychological processes underlying behavine Keo Psychologeal science i, Word selence comes lrom the Latin word senta, which simply means Knowledge. As the word is used today howeveh it has et Soy tatlons-content and process The content of scence is what ar how ar the fects we eat tn our psychology of chemistry Cousee Bae wae is also process—tha i, an activity that includes the speonane ee dich we go abou gathering data, noting reltionships, and sop oo tations, Explaining the proces of paychologial science the peek ns ofthis text, In the chapters that follow, we will examine some of the basic tacies used in poyehological research. We will study methodology, the sienhe tet gues used to collect and evaluate psehologial date Ge ct ont Rony fathered in research studies), All areas of poychology use scene eee, Iethods. For example, researchers investigating preeption calles ease thal laboratory experiments designed o provide the mest pate werner Foychologits interested in understanding sttudes and saeal he {eg GR Metlvord senings, Clinklans may collet data by adminitetng a Fancy of tests or by observing personality functioning dung sessions wah Patients. Whether the data come from laboratory experiments “eal wend {ines Psychological testing, or therapy sessions, all psyeholegise tee cose tile criteria to evaluate their data 3PART ONE INTRODUCTION The Need for Scientific Methodology In ow diy ies, all of ws collet and se peychologal data to understand the behavior of her and to guide our own behavior, When you notice st Your roommate in bad mee, you dont ak for favor You do noe Yat and Tey tothe same pry because you know they dont like each other: You dressup when you a gomng fora ob interview because you know fst Impresion are impotent. You an probaly ink of many noe examples ef stuatons in which you used psychologeal data to predic he bchavior of ates ano gue jour or tetra The a ery. nonce tor dvard oles has ben called commonsense poychology (Heder, 1938) Iescems to work wellenough for us mst of the ime, We might be quite suc Ces finding the best time to ak our oommae fora favor or choosing the fight onthe tan ines At oter tines, tough, nonsienti data gubeing can leave us up in the at Suppose, fox example tha yur signin othe es jst announced tha she his been acted for summer program s the Sorbonne in Pais Shou you be worrea Some of the nonscenuie data you pthersbout absent pares rersuning absence makes the hea pow fonder’ some isnt Cout of opi. on of ming’). Most of your daa sere to suppor the former concsions6 you see her off tthe airport good spins, On the dre home, you remember al the stones you have heard about Pas nighlfe and sexy Parsan men. Wuhout knowing which ootome i realy tee probable this gece suaton, you me Hkh to spend an aos Ss A dong of cara ae of on Since poyhology migat nat help th demas ke hs one, bu maybe you become a beter poychologealsclemstif only because i clay demon Batcs the need for amore scenic approach to undestanding and predic nee commontense psychologists, we find that our aly to gather data in a systematic and impart yay consined by to very pean farors the sources of poecogal information and our inerenial sates. Com- mmonseoe bees abut betavor ae detved fom data we col fom out town observations ad experience and what we have learned fom others. The data we eolec in our everyday ive have been gered fom very stall Sarpl of beavis nd te conclsions we daw fom them ae subject fo a number of inherent tendencies, o ise, that ht thet sceutaey a we fulness, Frequent, the sures of our nuored eles about behavior can beunrlable and the explanations and prediedon that we dive from them ze lly to e imperect Do bids ofa feahe lock rogeter? Or do oppo seta? Gu guage led wih he Kins of config, commen. ge ant preiel beavor i ny singe tance. Let us Tok Diy a fe ofthe protemy encountered byte commonsense poychlopetbefoe we thm tothe senilic approaches tsed in poyehologi sence, (CHAPTER 1 EXPERIMENTAL PSYCHOLOGY AND THE SCIENTIFIC METHOD Nonscientific Sources of Data ‘Some of the data we gather as commonsense psychologists come from sources that seem credible and trustworthy—fiiends and relatives, people in author ity, people we admire, reports from the media, books we heve read, and so forah—but, i fact, these sources are not always very good ones for obtaining valid information about behavior (see Box 1-1), Nevertheless, psychological information, particularly when it is offered by people we like, respect, or admire, is typically accepted without question. These belies tend to become stable because we rarely, if ever, est them. Once we believe we know some thing, we tend to overlook instances that might disconfirm ou: beliefs and wwe seek, instead, confirmatory instances of behavior. If you believe that the full moon brings out psychotic behavior (the word funacy comes from the Latin word for *moon’), you will notice and remember instances when peo: ple acted abnormally while the moon was full, and you will ignore the many, any more instances in which no unusual behavior occurred Its also unlikely that anyone can completely avoid assimilating some myths, superstitions, and pop psychology explanations for behavior. Do you believe in the power of crystals? (ome people believe that wearing amethyst will increase your intuition and that clear quartz will build inner strength) Do you ever Tead your horescope? Do you believe that dreams foretell the fature? Do you ever speculate about the “le line” on your palm? Do you feel a bit anxious on Fei. clay the 13th? Interestingly, only the later superstition —fear of Friday the 13th has produced any supportive scientific documentation (ee Box 1-2 Research has shown that we are more likely to believe information if it comes from cerain kinds of individuals: People who ate popular, atracive, high in status, seemingly expert, ar who appest highly confident are more powerful sources of information than others are. But other people ate not out sole source of data about psychological processes, We gather a lot of infor: ‘ation about behavior from our own observations and interactions with oth: ers and the conclusions we draw from them, Children learn ve-y early ttt their smiles are rewarded and that touching a hot stove can have negative con. sequences. We leam to predict consequences and direct our behavior toward desired goals. Frequently, we use our beliefs and feelings about how things ‘operate to explain behavior—our own as well as that of others Researchers have discovered that we are not always privy to our own decision-making processes (Nisbett & Wilson, 1977). North, Hargreaves, and MeKendrick (1999) conducted an interesting experiment that demonstrates this ‘inability. tm their experiment, French or German music was played on alter nate days from a supermarket display featuring two French and German wines of similar price and sweetness. As predicted, on days that French music was being played, the French wine outsold the German wine; whereas German wine outsold French wine on days that German music played (both by margins of about 3:1). Clearly the music had an influence on purchases. W>en queried about the reasons for their choices, however, only | out of 44 people men: tioned the music. Even when asked specifically whether they fet the music might have influenced their choice, only 6 out of 44 said that it might have 5BuO) iia Nonscientific Sources: When Court Decisions Substitute for Scientific Data tn 1962, an American woman received the fist slicone bess implants. th 197f, an Ohio woman teceved a $170,000 sclement from Dow Coming, she manulacurer ofthe Impans, et csiing pein and al feig ered by ruptured implanis and subsequent operations, in 194, 2 Sun Fandsco rman was evarded $21,000 in compensatory dan ! tees and $1.5 lion in purtve datages when the jury was convinced if “expens who hypuhesed a nk betwen slicone nd somone alesse In 1901, anoint San Fancsco woman was avarded 873 mal fon shen aur deeded tat her comestne tase dase was eased by her siicone breast implans. tn 1992, a Houston woman won $25 | muon, jury decided tbat her connective wes and some d= casts, chtonle path, muscle pa, lt ain, Headaches, and zine Were caused by her brea implans, giicone bess implants were Femoved fom the wakes in 1982 By 1994, 19.092 india lavuts had been filed against Dow Commit and by 19954900000 worne nd feet fo ines | f panel cas acon suit Nota singe publed research dy had Genonstated a conection between leone best implants =n any pe of dese. In December 1994, the American College of Rbeuratlogy Sued a saremen hat implans did no ene oyemic dessin june 1995, rel of the Harvard Nurses Epidemiologe Siody, published In the Now Eglnd Journal of Metcne, found no Increase in diese in semen ils brat ila In 1997, the Ameren Assen of Nek logy renewed all esting research data ad concaded that Gere was | no link berween implants snd neusogal Gordes Despite suong scene dence fom medical expens, in Decem ter 1988, the Netada Supreme Cour upheld» compenssony damage yard ol $41 milion toa Nevada wom for her lple-serose ke |. Simpcoms in amar 1995, an atorney won $10 milion in damages aning that her inplant cased err tn June 1993, the Inset of Med pars ofthe eminent Nasional Academy of Sciences, concluded in a 400-page report slicone best Implants were not oposite fray major deses To dat, sores of lag, well contol neical suis have found no lalondp between Stine beatin and disease. Despue he ck of scenic event tron (CHAPTER | EXPERIMENTAL PSYCHOLOGY AND THE SCIENTIFIC METHOD Gontaed that silicone breast implants are harmful they remain off the market in the ‘Unuted States except for women having breas surgery for medical condi- tions. The current status of silicone breast implants Can be found at the Iutp/varweda gov Web ste aye torn FRONTINE/WGBH Educonal Foundation og tw Copy 1 198 Pas and WCRLVFRONTLINE Rape by perio We are very often unaware of factors that influence our attitudes and behavior. For example, when we uncnitically accept information from an attractive source, we art unlikely to be aware of what actually persuaded us. I someone were to ask us, we would probably believe chat the information must have been extremely persuasive. We would be unlikely to realize that wwe were really persuaded because the person who communicated the infor- ‘mation was highly attractive We frequently use data from our own experiences to come up with com ‘monsense assumptions about cause and effect—but, if we were to rely only ‘on commonsense psychology, we would frequently be wrong. The inferential strategies we use when we process data are sometimes too simple to be com- pletely accurate. Lets look at a few areas in which the commonsense psy hologist is ikely to make errors, Nonscientific Inference One of the fst kinds of data we collet about others comes in the form of tats we assign to them, Commonsense psychologsts are trait therits—at least when it comes to explaining the behavir of others. When we understand other peo- pleS behavior, there is strong bias to overlook situational data in favor of data ‘hat substantiate trait explanations (Ross & Nisbet, 1991). When we notice that Michelle sporting another new designer outfit, we conclude that she is vain about her appearance and spends money fvolously. We tend to miss or ignore ther important, causally elated information about the situation (for instance, that, Michelles mother ges ffee samples because she designs for the manulacture) Clearly, perceiving others in terms of their tats can be useful for pre- dicting their behavior, but it can also lead to overestimations ofthe likelihood that they wil actin trai-consistent ways ina wide variety of situations, In fact, the research literature suggests that people may overestimate this kind of cross-situational behavioral consistency by as much as a factor of ten. And, apparently, this bias is hard to overcome, even with taining. In one amusing, study (Kunda & Nisbett, 1986), it was discovered that 2 group of tained research paychologisis was not much better at estimating the actual predictive [power of traits than were people in general—even when they were reminded of the bias by the presence of a very well-known personality researcher. 7PART ONE MVTRODUCTION 0) xpah a) The Power of Negative Thinking | Friday the 13th has « mythical history of being unlucky, and if Fay falls fon the 13¢h of the month, the superstition often comes to mind (or some= fone reminds us!) Do you feel a moment of anxiety when you realize its Friday the 13ih? Have you ever wondered whether itis really unlucky? Apparently i is, bu: probably not because of any dark and powerful unseen force that exerts its wll upon us. study conducted in West Sus- sex in Great Britain and reported in the Britsh Medical Journal (Scanlon, Iuben, Scanlon, & Singleton, 1993) found that Friday the 13th did appear to be an unlucky day for drivers, When researchers compared each Friday the 13th with each Friday the 6th since 1989, looking tthe num ber of emergency oom visite from automobile accidents, more accident vitims (as many as 52% more) were treated on Friday the 13th even. though fewer cars were driven that day. According to the researches, the: higher accident rate for Friday the 13th was probably caused by increased. trepidation about the date: Anxiety caused reduced attention to driving, | and mote accidents ocurred, Whether their reason isthe correc one oF not, it makes sense to be extra cautious if you ae driving that day because ‘other drivers might 22 mote anxious and accident ‘sual The process of stereotyping illustrates a related problem of nonscientific inference. Once we know that Carol isa librarian, we automatically assume that she is probably also serious because that characteristic is part of & ibrar Jan stereotype (Hamilton & Rese, 1980). And, similar to what can happen to individuals who believe the full moon myth, sterecypic expectations can lead us to seek confirmatory instances of behavior: "See! She is so serious, she always has her nose in a book” (Of course she does! Its her job!) ‘Additonal problems in prediction occur because people are not very good at using data to estimate the tue probabilities of events. The well-nown “gamblers fallacy” ts a good example ofthis problem. When we see tha a certain sloc machine has not pad off in along time, we tend to belive that i s overdue for a pay Jn reality st makes no diflerence whether the machine bas reeencly paid of or not (ness the machine is rigged). Each spin is entirely random and independent, so your odds ofa jeckpot this time are idericl to your odds on any other ty" (Genin ihe te ds xy you ow heme of wh nd he nb of He on ech whe Tie rhinestone our can of ong ee She {sorte omy ra fra pagel a 8 Ws 1 = 115 oF 8 depen (CHAPTER | EXPERIMENTAL PSYCHOLOGY AND THE SCIENTIFIC METHOD FIGURE 1-1 Do you believe the gambler fllacy? nbs alec, Post Finally, compounding our inferential shortcomings is a phenomenon known as the "overconfidence bias,” Our predictions, guesses, and explana: tions tend to feel much more correct than they actually are, and the more dlata we have available (accurate or no), the more confidence we have in our judgments about behavior (Dunning, Griffin, Milojkovic, & Ross, 1990). “These and many other inferential biases exist in human information process: ing. They are believed to be the brain$ way of coping with an immense vol ‘ume of information. They are shortcus, and most of the time they allow us to function well enough. However, if we want to be able to rely on our con- clusions and use them as general principles to predict behavior across many settings and conditions, we need to proceed more systematically and objec- livelyin other words, scientifically. The steps scientists take 10 gather and verify information, answer questions, explain relationships, and communicate this information to others ere know as the scientific method. We will now turn to several important characteristis of the scientific method, he Characteristics of Modern Science ‘The Scientific Mentality The psychologists goal of prediction rests on a simple, but important, assumption: Behavior must follow a natural order, therefore it can be pre: dicted. This elementary assumption lies at the heart of what Alfred North10 PARTONE INTRODUCTION CGARRELD, © 1993 Paws, ne. Dy Univeral Pee Sdae, Rested with emi Al gh ‘Whitehead called the “scientife mentality,” Whitehead (1861-1947) was a philosopher of sclence who traced the development of science in his now clas- sic book Science athe Moder. World (1925). He postulated that “Eth” in an organized universe is essential 0 science, If no inherent order existed, there ‘would be no point in looking for one and no need to develop methods for doing so. Research psychologsts share the belief that there are specifable Galthough not necessarily simple or obvious) reasons for the way people ‘behave and that these reasons can be discovered through research Gathering Empirical Data ‘Whitehead raced che beginings of modem scence to the works of Arto the fourt-centary ac. Greck philosopher. Like contemporary scientists, Aris totle assumed that order eit nthe Universe, and he set about desenbing that order im systematic way by sollecng empirical data—that is, data that ae observable oF experienced. Arse advocated systematic observation and ca {ul clasfcation of naturally oearring evens. From his observations, he argued that heavy objects fall fase than ight objecs because hei “natural” pace is down. Later observations by Galileo (1564-1682), however, led 10 the ines Capable contusion that sf we st up the proper testing candtion (a vacuum), light objets wl fal jut as fast as Reavy Ones. Clearly, gathering empirical data in'a systematic and orderly way is preferable to commonsense data collection, ‘bur it cannot guarantee thatthe comect conclusions wl be reached Seeking General Principles Modern scientists go beyond cataloging observations to proposing general principles—laws of theories—ihat will explain them, We could observe end- less pieces of data, adding to the content of science, but our observations would be of limited use withcut general principles to structure them, When these principles have the generality to apply to all situations, they are called laws. For example, atronome: Tycho Brahe (1546-1601) painstakingly gath- fered observations of the sta for nearly a lifetime. But Johannes Kepler Tr CHAPTER 1 EXPERIMENTAL HYCHOLOGY AND THE SCIENTIFIC METHOD (1571-1630) made these observations useful by explaining them through a system of equations now known as Kepler’ la Typically, we do not have enough information to state a general law. We ‘may then propose an interim explanation, commonly called a theory. Theo- rier pull together, or unify, diverse eete of scientific facts into an organizing scheme, such as 2 general principle or et of rules, that can be used to pre dict new examples of behavior. Theories can explain many, but not al, instances ofa situation or behavior—the more a theory can explain, the bet teritis. Sir Karl Popper (1902-1994), a modem philosopher of science, wrote that science progresses only through progressively better theories (Popper, 1963). Old theories are replaced by new theories with greater explanatory power. Laws are seldom determined outside the physical sciences, so the behavioral sciences like psychology largely progress by developing better and better theories, Theories also guide the course of future observations: “We must remet ber that what we observe is very much determined by what theory suggests should be observed: and we must remember also thatthe way in which obser vation will be reported and interpreted is a function of the theory that isin the observers mind” (Schlegel, 1972, p. 11). Theory-based expectancies can cause us to pay more attention to behavioral information that is predicted by the theory and to overlook nonpredicted behaviors. The next characteristic of the scientific method, good thinking, is essential to offset a predisposition to find only what we are seeking, Good Thinking ‘A central feature of the scientific method is good thinking. Our approach to the collection and interpretation of data should be systematic, objective, and rational. The scientist avoids letting private beliefs or expectations influence ‘observations or conclusions. Good thinking includes being open to new ideas and avoiding woodenheadedness. Woodenheaded thinking is “assessing a sit ation in terms of preconceived fixed notions while ignoring or rejecting any ‘couteary signs” Tuchman, 1984, p. 7), Good thinking also follows the rules ff logic. Conciusions wil follow from the data, whether they are in agreement ‘with our expectations or not Another important aspect of good thinking is the principle of parsimony, sometimes called Occams razor. William of Occam was a fourteenth-century Philosopher who cautioned us to stick to a basic premise: Entities should not bbe multiplied without necessity What Occam had in mind was simplicity, pre cision, and clarity of thought. We must avoid making unnecessary assump- tions to support an argument or hypothesis. When two explanations are equally defensible, the simplest explanation is preferted until it is ruled out by conflicting data Lewis (1978) applied the idea of parsimony to some developmental ndings fn an interesting way. He reported that infants in poor families spent mare time Jn their mother’ laps than middle-class infans did. Infants in poor families alsoPART ONE INTRODUCTION ee ae INVISIBLE RAYS CaaS iid Werraania.4 ECS UNIFIED AND U1 A TeoRy, Bur W'S NOT Te UNE THEORY WEYE ALL BEEN LOOKING FOR. copy © 1966 by Soy Has tended to vocalize less than infants in middle-class families did. We could spec- ulate on all sors of differences in attitudes, cultura factors or parental expecta- tions that might lead to differences in mothers’ behaviors, which in turn might alet infant development. But Lewiss more parsimonious explanation made a Compelling esee for » simpler environmental difference Eventhough the mother lp is the most frequent place fr the infant, the cil {sles likely to make sounds there than in any other situation, Mother tend to Noval more wih ther children in teit arms and their vocalization inhibi their infans from making sounds, Surpesingy, some of the least frequent situa tions-—such asin the playpen and the Door—account forthe highest percentage of infant vocalization ed the data by socal lass, the importance of situational dit fees became ees moe ce. Nofoeeame moe ever put he ink tn the floor, but midlecacs babies spent three percent oftheir time there. Why? {E far example the Mocs ofthe poor are unsale—cold, lacking mugs, and with the Tr (CHAPTER 1 EXPERIMENTAL PSYCHOLOGY AND TH IENTIFIC METHOD selded danger of anacking rodents—a poor mother would be unlikely to allow her {tal to ply onthe oor Therefore, inlnt vocalization i greater when the cil fs out ofthe motherS arms, then socal cs diferences in infant vocalization may rot be function of diferent aitudes or desires of mothers of diferent clases, but of stutional diferences as mundane as what shape the Noor is in. (p. 22) Ik is more parsimonious, or simpler, to explain the findings in terms of the physical conditions in poor homes than in terms ofthe attitudes or desies of different classes, Self-Correction Modern sess acept the uneraimy of thet o¥m conusions, The con tent of sence changes se we acquire new cent information and ld infor tuatinis evaluated in ight of new face Changes scent explanations dnd theories are an important pat of scenic progress. Experience favors & reighvo-eidence" approach The tote evidence that accumulates fo sp pore praca explanation or tory te more confidence we Rave the the theory core Old explanations on give way simply because the weight of supporung evidence pe the sales In vor of a diferent scentie expla tution, For example for more than twenty yea, the ink between media o- lence and aggresive behavior was explained by Soci esring Theory arn ing to iat others aggressive behave) Today, Cogtve Priming Theory isised wo expan thee lec (bering volencetiggts cope Tepe sentatons of aggressive behavior stored in memory in our own cognitive Schemas) beat the newer theory can expan peoples tendencies fo per form similar as well as identical behaviors after being exposed to them ony in ln, or in uci (erkowits © Rope 1989) Popper aged ht hones ne bes ese trough ater alscton — noc verton. What hs means i Ut scents challenge cxising explanations td thos by esting hypoiees tha alow lay fom the the text $hows 3 hypothesis ale then the ong thon shouldbe mode or bam done for be tat explains the new fdig.Fyptheses hat have not been proven fase ae not however, necesaly Tue Pethape or testing methods te Eo sullen senie to accomplish sca ex Tegan, in fact, angie that we can never really prove that a statement is rueve can only prove ati ae. Thnk abo te fllowing pres ‘All eows ae Hack Conon sence tls tha hie ete seme. st Gan it ever realy be proven? No. Been, oo rater how urtly sem is posible tha someone il someday come vipon prop of eros thet tre red And, only takes one case ofa contary Insane To prove the sale ment false (Runyan, Coleman, & Piterger, 2000). This is the prindple of mots tolls, ao called the “procedure of lticatlon” Statements ay be argh oh ice by min TH SCNCS rir eae NY 0021 oral -ROOTHE MAS ri BPART ONE. INTRODUCTON proven false by a singe, contrary observation; whereas, statements can never be proven to be true Eecause there is always the possibilty that a single, con: trary example might exist (or will exist in the future, or has existed in the past), but has simply not yet been observed. Publicizing Results Because of is dynamic nature, modem science has become a highly public activ 'iy, Scientists meet Requently through professional and special interest groups and atend professional conferences to exchange information about their current ‘work? The numberof scientific papers published each year in scientific journals, fs growing, and new journals are constanly being added in specialized disciplines. ‘This continuous exchange of information is vial tothe scene proces. It would do litle good for scerists to work in isolation. The opportunity to incorporate ‘the most reeent findings of others would be missed. There would be much wasted flor a researchers duplicated failures as wel a Succ Replication Replication is another important part of the scientific approach, We should be able to repeat our procedres and get the same results again if we have gat sd data objectively and if we have followed good thinking. Findings that can be obtained by only one researcher have only limited scientific value. For exam. ple, people sometimes repor. dreams that seem to predict the fuure. A woman sms ofa stranger and meets him the following day; a man dreams of a cat accident and then heas of the fatal crash of a frend. Have these people seen into the future through their dreams? We cannot provide a scientific answer to that question. Is impossible to recreate the original condition that led to these fevents, so we cannot replicate these experiences. Its also difficult t evaluate them objectively because the dreamer is the only observer of the dream Im contrast, a researcher predicts that childven will hit a doll ater chey hhave seen an adult hizing a stall child on television. The prediction is con- firmed, In this instance we can apply scientific criteria to the researchers find: ings. We can replicate the findings by setting up the same or similar condi- tions and observing whether the outcome is the same. Replication of research, findings by others can be important; we have a great deal more confidence that we have explained something ifthe predicted effects are repeatable by other researchers ‘Generally, replication i more common in the physical than in the behav- ioral sciences. For example, several years ago investigators in Texas reported that they had created muclat sion in the laboratory without heat (2 monumental > tei ages profesional npr spring the exhage oration sou ecology ae ie hnrean Fycsopel Hexen (AN) ad the Aen Rtg Socy (AS). APA he ‘hen nie ed 1892. Te ines of APA ar pn al ane win yeh, 262 ABA fone te reas of geycbologs engaged bh reach aNd ep) Pea) oh a ag sy A as xs CHAPTER 1 EXPERIMENTAL FSYCHOLOGY AND THE SCIENTIFIC METHOD selenific discovery), The report led to worldwide attempts to replicate their experiment in other laboratories. To date, other researchers have not substant- ated the claim for cold fusion. As in other sciences, published replications of psychological research are more common when the reported findings ether have Important implications or when reported result directly contradict current con- ventional wistom, The Tools of Psychological Science By now you are familar with the scientific approach to research: We gather information objectively and systematically, and we base our conclusions on the evidence we obtain. Let us riow begin t0 look at the dhree main tools of the Scientific approach: observation, measurement, nd experimentation. These are also the basic tools of the psychological scientist The research example Aeseribed in Box 1-3 ls a good Mlusration of observation, measurement, and experimentation, Observation ‘The frst tool, observation, isthe systematic noting and recording of events. COnly events that are observable can be studied scientifically At this point it sight seem as though we are restricting what we can study’ in psychology 10 a very narrow range of events. Many behaviors are observable (smoking, smil- ing, talking), but what about internal processes such as felings, thinking, oF problem solving? How can we explore those areas? The key is how we apply the scientific method, It is perfectly legitimate 10 study events that take place inside a person, such as thinking and feeling—if we can develop observable signs of these events The key to studying internal processes is defining them in terms of events that can be observed: the time it takes a person to solve a problem; a per- sont answers to a mood questionnaire; the amplitude of someone’ palmar skin conductance responses. (Finding @ suitable definition is one of the prob- lems we will discuss in Chapter 6.) in the experiment described in Box 1-3, Baron and his colleagues needed to observe people’ moods, Moods, however, cannot be observed directly in any systematic manner. It is not possible to Judge a person mood reliably just by looking; instead, researchers like Baron typically ask people to report on their own moods by using questionnaires or other instrument, Within the scientific framework, observations also must be made system atically—once the researcher has devised a system for observing, the same sys tem must be applied consistently to each observation. For example, the same ‘mood questionnaire would be given to each person in the study. Of equal importance, observations must be made objectively, another objective ‘observer mist be able to obtain the same record ofthese events. And, clearly swe must avoid distorting data by allowing our preconceived notions of the 516 sO bal pcp | ‘An Experimental Example: What's in the Air? Wiis dy, setonal winds, Ike te sco ls the spate Medes ‘ae or the Sonia Ana winds in Calor, have been blamed for every thing fom insomnia to homiide, and some research evidence seems fo |. Suppor this These winds increase a temperature, reduce humdi, and ale the atmosphere’ eleeon blance,splting atoms nto postvey and Segatvely charged panicles, calle ont, Dung these seasonal winds, there ie lightly higher concenuauon of postive sons sn he ai, and ¢ Smal amount of etarch evidence suggests that postive lone can pro- alice negative shod shits, Could negate loss have the oppose elec, Taking people fel beter? Robew A Baron and his colleagues (Baton, Rise Arms, 1985) tested the later Bypoties in an interesting lb ratory expetiment. (The atal experiment was more complex than hs Scd.we al ceo € again in ltr chaps) Tose up the condons fo testing, the researchers used a machine that generate egauve lots tts do) to change the atin Laboratory toom so tha the concentration of negative tons wes ele low (aol amb x0, moderate, oF high. Tn tach session mae under fraduate was led to belle he tas involved in an expesent Sout laming: His tak ie ing the session toed tan ing another undergraduate to reduce his hear rate with oleae The emer” wat scully tained conederate of the researcher, and He erfomance was completely Scriped in advance. Sesons vere setup so that hal of he ‘olunteer were intentionally nin FIGURE 1-2 Rober A. Baron Tr (CHAPTER L_ EXPERIMENTAL FSYCHOLOGY AND THE SCIENTIFIC METHOD eontued angered by nasty comments from the “learner” during the session; the other half were not Each volunteerS mood was measured at the end of the session. The results were not quite what Baron and his colleagues had predicted, Instead of inducing more postive moods in everyone, higher concentrations ‘ol negative ions seemed to increase the strength of whatever mood the vol ‘unieers repored good or bad. At the end of the session, nonangered vol ‘unters who had been exposed to higher levels of negative ins reponed that, they fl less angry, ess depressed, anc less fatigued than did the nonangered volunteers exposed to normal levels. In contrast, the angered volunteers tended up feling more angry if they had been exposed to higher levels of negative ions than if they had been exposed to normal levels. One explana- tim is that negative ions may be physiologically arousing, and arousal may increase the strength of whatever emotions people are feeling nature of events to alter our records, Baron and his colleagues recorded and reported ther subjects’ exact responses to the questionnaires, eventhough the responses did not quite match what the tesearchers had predicted Measurement “Measurement is assigning numerical values to objects or events or their chat acteristics according to conventional rules, When we do research, we assign ‘numbers to different sizes, quantities, or qualities of the events under obser vation. We are all familiar with such conventional physical dimensions as length, width, and height, Rather than relying on global impressions (‘It was really big"), we use standardized units, agreed-upon conventions that such measures as the minute, the meter, and the ounce Standards are not always as clear-cut for dimensions of human behavior. We use standardized intelligence tests and a varity of standardized personality measures, but very often our standards are determined by the context of a par ticular study. We often wish to deseribe the behaviors of individuals in a pre~ determined situation (How much di they tal with others in a stressful situa tion?), Other times, we want to measure individual’ reactions to the situation we have created (How depressed did they fel ater a stressful situation”). Or we may wish to quantify their evaluations of an objector another person (In a stressful situation, isa stranger judged more favorably on dimensions such as attractiveness and intelligence?). When Baron and his colleagues measured each volunteers mood, they designed their mood questionnaire using numbered scales to represent progressively higher levels of anger, depression, and so forth. ‘Asin Baron’ experiment, we ae typically interested in comparing the behav Jor of individuals exposed to diffrent sets of conditions. Our measurements musta8 PARTONE INTRODUCTION be consistent across each set of conditions, If measurement Is inconsistent, we cannot compare our measured observations direcly (its like comparing oranges and apples) The same unit of measurement needs o be used each time we mea- sure our observations; we would not ue ounces one ay and teaspoons the nex. And, to be consistent, we need to use the same instruments and procedures each time the event is observed. In Barons investigation, for example, the queston- naires, as wel a the way they were aiministred, were identical in each session, Recause we use statistics to evaluate research findings, we need mumbers, oF scores, to represent diferent levels or amounts ofthe behavior of interest, As you vill seein upcoming chapters, we often compare the average scores ofall sub jects exposed to one set of conditions with scores from other groups of subjects ‘exposed to different conditions. Experimentation Experimentation isa process undetaken to show that certain kinds of events are predictable under certain, specifiable situations. Psychologists use experi- mentation to demonstrate the conditions under which a particular behavior can be expected to oocur with regularity. When we experiment, we systema ically manipulate aspects of a setting to verily our predictions about observ able behavior under specific conditins. Experimentation is not always possi- ble. To do an experiment, our precictions must be testable. Two minimum requirements must be mei: Fits, we must have procedures for manipulating the seting, Second, the predicted cutcome must be observable. Suppose we have predictions about the observable effecs on human travelers of making a 20-year journey through outer space. Our predictions are not testable because ‘we do not yer have the technology to make that long a journey. Clearly, some hypotheses that cannot be tested now may become testable in the future Baron’ prediction wat testable because he could manipulate the setting to cre- ate the conditions he wanted to investigate, and he could observe the outcome. Experimentation must also be objective, Ideally, we do not bias our results by seting up situations in which our predictions can always be confirmed. We o not stack the deck in our favor by giving subjects subtle cues to respond in the desized way. (We will have moze to say about this in Chapter 7.) Nor do we prevent them from responding in the nonpredicted direction, ‘AL times experimentation might be possible, but it cannot be careed out for ethical reasons. For example, we would not test the effects of smoking on fetal development in pregnant women by asking a group of nonsmoking women to stoke during their pregnancies, We would not peep through win- dows to study people’ sexual behavors. We would not change students’ exam grades to leara about how people respond to success and fallure, In a moral Society, there are many experiments that should never be conducted because it would be unethical to do so, (We will explain ethical guidelines in the next chapter) This is not to say that these things cannot be studied by psycholo- ists simply means we must study them in ethical ways. Sometimes this is done by using nonexperimental me-hods (Chapters 3 and 4) or by designing experiments that pose less risk to participants (Chapter 6) (CHAPTER 1 EXPERIMENTAL PSYCHOLOGY AND THE SCIENTIFIC METHOD Gcientific Explanation in Psychological Science Identifying Antecedent Conditions Ina scientific context, explanation means specifying the antecedent conditions of an event or behavior, Antecedent conditions, or antecedents, are the ci cumstances that come before the event cr behavior that we want to explain. In Baron experiment, for example, diferent concentrations of negative ions were the specified antecedent conditions, and mood was the. behavior explained by these conditions. If we can identify all the antecedents of a behavior, we can explain that behavior in the following way: When XYZ is the set of antecedent conditions, the ouscome is a particular behavior. This, explanation allows us to make predictions about future behaviors. Ifthe XYZ set of antecedents occurs again, we expect the same outcome Comparing Treatment Conditions In psychology, it would be vitally impossible to identify all che antecedents that affect the behavior of research participants (also called subjects) at a par ticular time, But although we cannot identify all the antecedent conditions, we can focus on particular antecedents that we believe have an effect on behavior In the paychology experiment, we create specie vets of antecedent conditions that we call treatments. We compare diferent treatment conditions so that we ‘an test our explanations of behaviors systematically and scientifically. Keep in rmind that the word treatment, as used in experimentation, does not necessar- ily mean that we raust atively do something to “teat” each subject (although itcan mean this in some experiments like the sample research article in Chap- ter 16). Rather, It means that we treat some subjects differently than we do others, We expose them to diflerent artecedent conditions. Sometimes, as Baron and his colleagues did, we test our explanations of behavior by creating tweatment conditions in which some people are exposed to one set of antecedent conditions and others are exposed 10a diferent set of antecedents. ‘Then, we compare the effects of these different antecedents on a particular behavior When we are able to specify the antecedents, or treatment condi- tions, that lead to a behavior, we have essentially explained that behavior. ‘The Psychology Experiment A poychology experiment is a controled procedure in which atleast ro diferent treatment conditions ace sppled fo subjects. The subjects’ behav tors ae then measured and compared to testa hypothesis abowt the elects of those treatments on behavior, Note that we ms have at eat two di and observe the way behavior changes tetmentcondiions change. Note fico that the procedures In the pajehalogy experiment are carefully con: trolled so we canbe sure we ate measuring what we ilend to measute For 1»» PARTONE INTRODUCTION this reason, characteristics of subjects receiving diferent treatments are also controlled by special techniques (Chapter 8). We want to ensure that peo. ple who receive one kind of treatment are equivlent (as similar as possible) to subjects receiving a different treatment. If subjects who received one treat rent had diferent characteristics than did subjects who received another treatment, we would have no way of knowing whether we were measuring behavioral differences produced by differences in the antecedent conditions we had created or whether we were just measuring behavioral differences that alteady existed When you were a child, did anyone ever ask you: "Which falls faster, 2 feather or a rock?” If 50, you probably said, *A rock” And of course, you would have been right ifthe test were made unéer uncontrolled conditions. Rocks do fal faster than feathers, unless we contzl the effects of ait currents and air resistance by measuring how fast chey flim a vacuura, AS Galileo ‘discovered, the aceleration caused by gravity is really the same forall objects. Successful experimentation relies heavily on che principle of control. For experimentation to produce valid conclusions, all explanations except the ‘one(s) being tested should be clearly ruled out. Other factors that could be producing the effect we want to explain are carefully controlled. We can achieve the greatest degree of control with experiments that are run in the Inboratory where the psychologist can insulate subjects from factors that could alfect behavior and lead to incorrect conclusions. Some critics have argued, however, dat lburaory situations can be artifical und unrealistic and that Jaboratory results might not be applicable to everyday life. Afterall, not many rocks fll to earth in a vacuurm. We sometimes sterfice a certain amount of realism and generalizability to gain precision, but contol is critical 10 exper Iimentation, The principle of conteol will be a part of discussions throughout the balance of the text. Establishing Cause and Effect The greatest value of the psychology experiment is that, within the expen ment, we can infer a cause and effect relationship between the antecedent conditions and the subjects’ behaviors. Ifthe XYZ set of antecedents always leads toa particular behavior, whereas other treatments do not, we can infer that XYZ causes the behaviot. For example, with all other factors constant, Baron and his colleagues demonstrated that high concentrations of negative fons altered moods. In Chapter 12, however, we will discover that our infer- ences about cause and effect relationships are stated in the form of probabil The type of cause and effect relationship we establish through experiments is called a temporal relationship. A time diference holds in the relationship: The treatment conditions come before the behavior. We look for diferences In behavior after subjects are exposed to the trectment—not before. In fact, if you got differences before the treatments, you would need to look for another cause! We expose subjects to different levels of negative ions, for example, then we see how they feel. Temporal relationships are built into our T (CHAPTER 1 EXPERIMENTAL PSYCHOLOGY AND THE SCIENTIFIC METHOD experiments. We give subjects various insructions then see how they behave ‘We show children various cartoons, then observe their play. (Other kinds of relationships can suggest cause and effect, but they are less convincing to the scientist because other potential causal explanations are never completely ruled out. Lets look at examples of two other types of rel tionships that people use: spatial and logical. A very large and rambunctious Labrador retriever and a very tiny, but aflectionate, Himalayan cat reside with fone of the authors, The cat Is mulch more fond of the dog than ke is of her. ‘When the dog runs out of patience, he will chase her around the house. One day, the author heard a crash from the living room and ran to see what hap- pened. A prized figurine was in pieces cn the floor next to the collee table, find the cat was sitting on the edge of the table looking down interestely at the pleces—the cat and the broken figurine had a spatial relationship. Natu- rally, the author scolded the eat, But was the cat the real culprit? (Iso, why was the dog sinking away with his tall between his legs?) Using spatial rela Lonships to infer cause and effect can be compelling —but not always correct. Sometimes we use logical relationships to establish cause and effet. On another ocesion, the same author discovered a small hole in the wall above the sofa, She identified the dog, rather than the ea, as the culprit, because of a log seal relationship. At about 4:30 every afternoon, the dog sprints a fll speed around the house, carrying as many of hs prized, wolfsized, plastic bones as hhe an fi ino his mouth, Running across the soa is part ofthis nual. Most of the ume, he runs the course with great agility, bt he has been known to run into walls, furniture, or people who happen to be in his way. Logically, then, although other causes might be possible, he seemed the most hkely perpetrator ‘As we search for cause and eflectieationships through our research, we generally look for temporal relationships, OF course, the simple fact that one event precedes another isnot sufficient, ane, to establish a causal relationship The Satish philosopher David Hume (1711-1776) argued that we can never establish causality in this way (According to Whitehead, the logical conclusion fof Hume’ philosophy t that there can be no science. For that reason, scientists fare not Hurne’ disciples!) HumeS objections were based on the argument that Just because one event precedes another, t does not necessarily mean that the fit causes the second. For example, on December 18, 1995, the Dow Jones Industrials dropped dramatically (along with other World markets). This event was preceded by an unusual astronomical event, the Sun passing very close 10 the postion of jupiter. Even so, few people would be willing to believe that stock prices are determined by the postion of remote celestial bodies. Obvi- ‘ously, many other factors could have had an impact on the economy. The advan- tage ofthe experiment in bringing us closer to establishing cause and effect re: tonships i that in the experiment only one factor is allowed to change Necessary versus Sufficient Conditions AAs we seek cause and effect relationships in science and psychology, we try to identify the conditions under which events will occur, We distinguish between necessary and sufcient conditions. Cutting down on fat intake might—K—X—mKX,s—nansxwrol— PARTONE INTRODUCTION be a sufficient condition to produce weight loss. But is it a necessary condition? Pio wwe aso could lose weight by increasing our activity level, so reducing ft pecs nota necessary condition. n concras, a snowmobile will mot fun with eee Therefore [uel is a necessity condition for renning a snowmobile. ‘The cause and effect relationshipe established through scientific research commonly involve identifying suficent conditions. For example, « number St psychologial studies have shown that being ina good mood increases or Shthngncss fo help others (sen, 1987), However, many other factors (char vrterusce of the person needing help, the number of other potential helpers, sees can ago determine whether we will help oF not (Latané & Dar- fey 1970), Seing in a good mood is not a necessary condition to increase helpfulness—but ic is sufficient ‘When we seek causes, we rarely eek conditions that are both necessary and sullicient Ta do so would inzolve a search for the first or primary aoe ce Cause with a capital C! Given the complexity of our universe, we SeSSTd make slow progress in our search for order if we Tefused to setle for Joything less than eauses that were both necessary and sufficient, Researchers Ae stedy helpfulness would probably stll be uying to trace the Cause of wine aie Rehavior-—right down to the molecular chain that produces the bio themieal changes associated with helping. How chat chain got there, the Chose’ Causes would lengthen their search even further. The scientific Sppreach to causality is moe practcal, relying on suficient causes as expla nations for events. ‘Phe Organization of the Text You may be required to design an conduct a esearch project in the courst Tot ahich you are reading this tex. Writing @ research report of your find igs may aso be a required part ofthe course. You will ind tha this ext is ‘itaed into four major pans: Introduction, Method, Results, and Discussion. ‘These parte parallel both the process of conducting an experiment and the donecponiding sections of a research repos. Part I, Introduction, gives 2” cratl orientation to the field of research methods, much as a [erature Chieu ges an overall picture of the state of research in particular content cere Research ethics are covered fst, Later chapters focus on the differences Ronvcen experimental and other research methods in psychology t0 help develop your understanding of te experiments. This section of the book eeus wee a chapter on formulating a testable hypothesis In short, Paxt 1 wil provide you with all the information you need to begin thinking about an iperiment in a parciculr area and wil provide you with report wnting ips along the way. ‘Bar Ik Method, includes all the basic procedures used in conducting sim ple experiments, selecting subjects, and collecting data in a scientific way, Part Fit eeults: Coping with Data, reviews the common statistical procedures {ised to analyze data, Examples of experiments and actual computations are Fr CHAPTER 1 EXPERIMENTAL -HOLOGY AND THE SCIENTIFIC METHOD tecluded 10 help you understand how these procedues ae used and what eymean Pur Decason lotsa he major ws nvlved in avg Corclasns fom da. We exinine protlens of genlsing fra lbore tory experiment tothe real ond, The chapter on report tng inches information on how each section of a research report is organized and writ- ten Append © shows what your fished report shuld ook ike, Before we ig ling semen rc, po ot he topic of Chapter 2. a leaideletle ie Summary our knoe stot paychlogalpoees is bed on sce ede eumulata frogh wear td a scents we en sec eta. Seip oes psig mech he, pes ey late ey exer ol al mle ert eoplaa goer ing observable, objective data and the search for general laws or theories to orga Ses a of tcl pce Me Ring an ange mat ing et ror across esi snd nc at ee neg then demonsete at sendy bnew rene lence apan nde plc Sgr er alge we say that we have explained that behavior, and we can predict the outcome ‘when the same set of antecedents occurs again. eaten aleast two different treatment conditions are applied to research participants, or subjects The subjects’ behaviors are then measured and compared so that we esta hypothesis about the ellects of those treatments on behav 1a cause and efectrelaionsip between the antecedent treatment Gitions and the subjects behaviors; we may say that the particular treatment Causes the behavior Psychologists generally look for the sufficient conditions ‘explain behavior rather than looking for the ultimate causes of behavior. ‘As experimental poychologisis, we begin the experimental process with a view of the research Iteratute to suggest a hypothesis about behavior. Next Wwe design a procedure to test that hypothesis in a systematic way. We use sta Istical procedures to analyze our observations and to decide whether the data hypothesis, We then reevaluate our procedures and write a research report of the findings. ‘Rey Terms Antecedent conditions All cicumstances that iehaviorto be explained: also called antecedents Cause and effect relationship The reaton between 2 partcuae behavior and a set antecedents or enst before the event or nt always precedes whereas other antecedents do not—s0 hat the st is inferred to cause the behavior Commonsense psychology. Everyday, nonsientic collection of psychological data sed to undersand the socal world and guide our behavior Data Facts and figures gathered om observations in research, (Data isthe plural form ofthe Lain word datum.) Experimentation The process undertaken 10 strate that events tht have already occured will occu again under a specified eof conditions @prineipal 00] ofthe scientific method. Falsfcation To challenge an existing explanation oF theory by esting a hypothesis hat follows logically from it and demonstrating that ths hypothesis false Good thinking Organized and rational thought, chaacenized by open-minded objectivity and parsimony, « principal tool of the scent method Laws, General senticprincple that explain our universe and predict e ‘Measurement The sjstematic estimation ofthe quanti siz, or qualy of an observ able event principal tool ofthe scent method Methordology The scent techniques used to callc and evaluste psychological dat Observation The systemati noting and recording of events; a pneipal tool of scientific method Parsimony An aspect of good thinking, tating chat the simplest explanation i ps fered until ruled out by conflicting evidence; also known as Occam’ razor Psychology experiment A contolled procedure in which a least vo di ment conditions a applied to subjects whose behaviors ae then measured and [Compared totes hypethests sbout the eflects ofthe treatments on behavior Replication The p repeating research procedures to very hat the outcome Wil be the same as before # principal tool of the scientific method Science The sysiematic gathering of data to provide desenpions of evens taking place under specific condition, enabling researchers to explain, predict, and con= trol evens Tr (CHUPTER | EXPERIMENTAL PSYCHOLOGY AND THE SCIENTIFIC METHOD 25 Scientific method Steps scientists take to gather and verify information, tions, explain relationships, and communicate findings Sbject The selene term for an individua. who participates in cesarch. Testable Capable of being texted; typically used in reference to a hypothesis. Two ments mus be met in order to have a testable hypothesi. procedures for manipulating the setting must exist, and the predicted outcome must be ‘Theory. A set of general p ‘Treatment A specie at of antecedent conditions crested by the expe resented fo subjects t tet ts elect on beh pes that attempts to expla behavior ot Beview and Study Questions 1, What s science? 2. Why do we need scent 3 1, Give an examp ‘of ow you used it within che past wee. What ate the characterise of modem se the scientific method? * 6. a. Define an experiment. b. Male up an experiment to test the saying, “Absence males the heart grow 7. Suppose for instance, that «researcher believed that Srtborn children have higher Qs than thei later born sisters and brothers because the mo ae younger when firstborns are conceived, This hypothesis would no testable. Give two reasons why and expan them ‘What ate antecedent conditions and how are in scienifc explanation? ‘course, you mis attend every 6 fram’ fe clas attendance a ne ‘What are treatment condtons? What ste purpose of wsng atleast two teatment conditions inn exp 0 1 For each of the following examples, sats which bsc pines ofthe ihe ave been las and expat why each privtle was lated 3 waned to co an cxpertnent om gu mene tose wheter the name Brands ve bewer mage She filed et tank wih FoelUp one we and wrth welcnowm brand the flowing were tthe end ofthat tine he Though thing over and sade Wel, Sant noice mach diferncebetieen the brands filed te cor wih FurlUp ona csday and nested gh ag ‘must be about the same, . nif26 PART ONE. INTRODUCTION J b. Mike has been telling all his frends that is read. One evening Mike invites some of is fends over for cofee and to give a demonstation of his daughters remarkable ski. Ali cen appears to Tead a small stonook that Mike keeps on the coffe table. One of the ends i notconvinced and asks the litle gi to read a page from a die ent but equally simple storybook. allie i, Mike explaine Ales havior by saying, “hes Just shy with tangs” c. An author sdvveates the use of large doses of vitamin C to protong be. In Sr nverview he says he bas tried this tearment only on himsel 4. A researcher report that flowers val move aay {them of Others ry unsucesflly to duplat ths finding Ti rps that skeptical ches are jst insensitive to the plans 13. Name and describe each ofthe four main sectors o experimental report CHAPTER 2 Critical Thinking Exercise Research Ethics Research Ethics The myth: Wi we are angry, it helps to vent our feelings by taking The American Psychological Association anger out on inanimate objects, like punching bags or pillows, Suis 7enoloe The scent ‘Venting our anger by physical aggression actually Deception and Full Disclosure makes us Beshman Baumeister & Stack, 1999) ‘Anonymity and Confidentiality Protecting the Welfare of Animal Subjects Pb: Des esos wy heh ay ave coe pt eee z our culture Fraud in Science Problem 2: Reality based talk shows, like the Jery Springer Show Plagiarism often erupt in violent behavior when angry guests confront ea Summary oer, Use the scientific findings to predict what fic beirg on Key Terms ‘how might have on these guests ata later time & Online Resources publisher of your textbook (Brools/Cole, part of Thompson Learning) rowdes online workshops and ober -eeources in the areas of esearch met fds and statistics. You can view the information on the Web site or download itas a Microsoft Word docur Review and Study Questions Critical Thinking Exercise hutpsipsychologywadsworth.comvorkshops/workshops hum! For a humotous introduction tothe scientific method, you might want co try frontier osrbe.edw/hsscience/bsimeth hum ow28 PART ONE INTRODUCTION CHAPTER OBJECTIVES ‘© Understand the roles of IRBs and the APA Guidelines in the ethical conduct of research using human participants Learn the meaning of animal welfare and how itis protected ‘Learn the meaning of animal rights and the views of animal activists Understand scientific fraud and how to avoid plagiarism Avsy seach poet ates desions sot he sbjts who wil pare igen te sd, Who vl hey be an bow wl e std then? in Caper 1 Wrest the general cong of eben, datebsed scence. Scene des Yih face, with uth seeing and wid ondesanding our univee.Saene Is {omonly though ofa ama fom asec pom of ev as Goer through sence se eter moral or inmate js bxppen to ees tet tis ou nape fo repo our ndings tly beter we ke Ths sence pers does notice values. A researches, howeve, we do bring ou own ves, eb, moral ard sense of gh and wrong othe work seco We have to deal wth the ehical and moral questions hat are eight to dscver way old more fice weapon? Is tht to postpone death by anifcal mean? is ght to study wage to crete more pesee olling through behavioral eognie Is it ght or prude to investigate techniques at could be used to brainwash people? Do we have height to perform any exe ten maghable jst for the sake of pe Ine Tn hs chaps we wil oes on the le question. We wll cos the ethics ofthe researches relationship wih naman nd nonhuman subjects and the researchers espouses fn rey experimen Research Ethics The researcher’ foremost concern in recruiting and using subjects is weating them ethically and responsibly. Whether we work with animals or humans, we must consider their safety and well. Responsible psychological research iS not an attempt to satisly idle curosity about other peoples innermost thoughts and experiences, Rather, resronsible esearch is aimed at advancing ‘oar understanding of feelings, thoughts, and behaviors in ways that wall ult ‘mately benefit humanity. But the well-being of the individual research partic ‘pant is no less important than the search for general knowledge: Research that is barmful to participants is undesirable even though it may add to the sore of knowledge. For instance, early experience is an important aspect of child developmient—but we would not raise children in isolation just to asses the effects of deprivation. There is no way we can justify such a study, no ‘matter how important the knowledge we might gain A researcher is also legally responsible for what happens to research par- ticipants ofa study. He of she is lable for any harm to tesearch participants, even if occurs unintentionally, This means that a researcher could be sued for damages if an experiment hurt someone, whether the injury was physical ‘or psychological, intentional or accidental. To protect the subjects of psycho: logical research, legal and ethical guidelines have been formulated, From a legal standpoint, human participants are provected by a federal law (Tile 45, Section 46.1061b)) This law requires each institution engaging in research to set up a review committee, called an institutional review board, or IRB, to evaluate pro- posed studies before they are conducted. Both laypeople and researchers serve fn these committees to guarantee that the views of the general community, as ‘well a5 those of scientists, are taken into consideration. The primary duty of 8 review board is to ensure thatthe safety of research participants is adequately protected (see Box 2-1). The IRBS fist task isto decide whether the proposed study puts the subjects at risk, According to the regulations, “subject at risk? means any individual who may be exposed to the possibly of injury, ncuding physical. psychological, o socal injury, aes consequence of Pst Ueiption as subjeet in any researc, development, or related actvy which tlspans fom the applicrtion of those established and accepted methods neces sary to meet his needs, of which increases the ordinary risks of dally fe, includ Ing the risks inherent in a chosen occupation or feld of scence. (US. Depart ‘ent of Heals, Education, and Welle, 1979, p. 11834) In essence, the regulations say that a subject at risk is one who is more likely to be armed in some way by participating in the research. The IRB raust determine whether any risks to the individual ae outweighed by potential benefits or the importance ofthe knowledge to be gained. This is called a risk/beneft analysis. ‘An understanding of research design is crtcal to such an analysis, and atleast some members ofa review board must be skilled in research methodology. As we will see in the following chapters, there are important diferences in the kinds of conclusions that can be drawn from diferent research design, and research that is improperly designed has few benefits, Psychologist Robert Rosenthal (1994) has given three important reasons why poorly designed research can be unethical 1. Students, teachers, and administrators’ time will be taken from potentially more beneficial educational experiences, 2. Poorly designed research can lead to unwarranted and inaccurate concla- sions that may be damaging tothe solely that directly or indinectly pays for the research, 3, Allocating time end money to poor-quality science will keep those finite resources from better quality science.— Box 2-1 | A Terrifying Research Experience | | consider ts example of rach conducted before the exablshent of Conse ns eal of sca ded lo ts te | | SPs athe Ge ofan event Cassel onaaaning At] | sets Ge Seodhon enn epee CS | | response is known as the conditioned response (CR) because its occur- Se nc savin Oo es alata chat | brag he bes MoS a seer | i Raa hacen SDN Sr oe to cae | | Seon eames | area) in ost subjects The emovonal eacton persisted (has, alld Pre navel) and actually increased wath repeated presentations of the 1 ee Nbuld you have wanted tbe in ths study? Do yoo think the | fetahes could have demonsted cas codtenngit anther way? | “Anothce very important task of an TRB isto safeguard the rights of individ als by making cerain that each subject at nsk gives informed consent to partic pate Informed consent means thatthe subject agrees to participate afer having een fully informed about the nature and purpose of the study. Several aspects of Informed consent are particulary relevant 10 psychological research. First ind “duals must give thet consent freely, without the use of force, duress, or coer Toon. Second, they must be fee to drop out ofthe experiment at any time. Third, (tovarchers must give subjects a full explanation ofthe procedures to be followed Ral ofir to answer any questions about them. Fourth, researchers must make tear the potential risks and berefs ol the expenment. if there is ary postbalty of pain or njury.esearchers rst expan his in vance, so dat subjects Iw ‘itt they are geting into before agreeing to participate. Fifth, researchers must provide asurances that ll daa wil remain private and confidential. Final CHAPTER? RESEARCH ETHICS according to Federal guidelines (4SCFR 46.115), subjects may not be asked 10 release the researchers (or study sponsors, institutions, or other agents) from li Dility orto waive their legal righs in the case of negligence ‘Consent should be obtained in writing, and subjects should receive a copy to keep. When the subject is a minor otis cognitively impaired, researchers should obtain consent from a patent or legal guardian. Even in these cases, hhowever, subjects should sll be given as much explanation as they can under- sand and be allowed to refuse to participate, even though the parent or guardian has given permission ‘Consent forms need to be writen in ly language atthe appropriate reading level for the study participants. Simply allowing your subjecs to read and sign the consent form, however, may not be enough to guarantee that they fully under stand what they ate signing, An interesting experiment by Mann (1994) demon ‘Stated that most subjects retain litle information inthe consent forms they have signed. Fewer than half of Manns subjects understood the procedures ot what ‘would happen to them in the case of injury: And even though it was explcly Stated in the consent form that subjects were not signing away any oftheir legal rights the simple act of signing a consent form made the majority of subjects, believe that they had given up thei rights to sue. Therefore, to ensure that informed consent is rally being obtained, researchers need to verbally reinforce ‘formation that is imporant for subjects Box 2-2 shows an example of the kind of writen informed consent form typically used in psychological experiments. The consent form provides subjects ‘with information relevant to their participation in the experiment: the nature of the experiment, an overview ofthe procedures that wil occur, how long it wall, take, and what they wil be required to-do. The specific bypothess of the exper iment, however, typically not dislosed cn the form. I subjects are made aware of the researchers expectations, their reactions during the experiment may be ‘unintentionally or intentionally altered by this information. (We will talk about this and other demand characteristics in Chapter 7.) The purpose of informed ‘consent i to give subjects enough information about the experiment so that they fan make an informed decision about whether or not they want to participate The American Psychological Association Guidelines ‘Although the law contains specific provisions for the way research isto be con- ducted, questions may sil ase in actual situations. For this reason, the Amer ‘can Psychological Association (APA) publishes its own set of ethical standards (1992). The code applies to psychologists and students who assume the role of psychologists by engaging in research or practice. Psychologists conducting {esearch ae tesponsbl for maintaining eal standards in all esearch conducted by them or by others they supervise. The APA standards include the same gen: eal requirements for ensuring subjects! welfare as set forth in civil lw "the an idles ath son rein of he nga APA. ules pubis 1971, The Ssh comme ncn ato, ba Hcy am eg a ss| <=> Sample Consent Form for Participation of Undergraduates in Psychological Research Informed Consent for Participation of Human Subjects in a Research Project Project tile Name of investigator: How to contact the investigator: I voluntarily agree co participate in this study. 1 understand that I can terminate miy partiipation at any point without penalty and that ter ‘mination wil in no way jeopardize my standing at (name of tniversiy) The investigation has been described to me by the experimenter who has answered all my questions. I understand that {will be asked to (deuals of specific procedure sh of tim and 50 forth) My participation is subject to the following eo 1. That adequate safeguards will be rovided to maintain the p confidentiality of my espe 2 That my name and my family’ name wall not be used t0 v idenufy ssid material; instead, code numbers will be used 3. That my individual scores will nox be reported: that data will be reporeed as aggregate or group scores. (paricipands signature) Gave 7 signature) @ Whenever any question arises about the ethics of an cedure, the researcher should seek advice from an IRB oF & employ ll possible saleguar require IRB approval of each study befor itis periment of pro rm colleagues and to clarly the procedures that apply in your int ‘ation oval for an experiment from an instructor a review beard begin) Keep in mind that even after procedutes have been approved, the individual researcher has the final responsibilty for carrying out the study in an ethical way. Obtaining cially important Folly informed consent is required fom all participants who are at risk because of the nature of the research. In some cases participants are consi ered to be at minimal risk. The law defines minimal risk as “risk that is no greater in probability and severity than that ordinarily encountered in daily ‘or during the performance of routine physical o ions or tests’ (Title 45, Section 46.102(p)). Basically, this means that research does not alter the participants’ eing harmed. For example ‘of what proportion of the population uses soap and water to wash heir hands in a public restroom would be considered minimal-rsk resea the participants are engaging in a public activity in a public place. Wh or not their behavior is recorded as pat of a study is unlikely to affect th in eny way, There is a small chance they could be injured di (for example, by falling on a slippery flor). increase the chances ofthis happening. bservations of public behavior, anonymous questionnaires, and certain Ikinds of archival esearch (discussed in Chapter 3) fall in the minimal-isk category I sent isnot always mandatory in minimal-isk research, AAs a safeguard, however, st is usually desirable (0 obtain informed consen whenever possible, and many institutions now require written documentation of informed consent as a matter of course in all research using human par- cipants. The APA standards for informed consent are reprinted in Box 2-3. The responsibilities of both subject and experimenter must be agreed upon in advance, and the experimenter must honor all commitments made (© subjects. Commitments include promises to pay subjects, t0 give them credit in courses, 1o maintain confidentiality, and to share the results of the study with them nformed consent is considered espe rever, the research does not Deception and Full Disclosure The relationship between researcher and panicipants should be as open and honest as possible. In many psychological studies, however, the true purpose ofthe study s disguised, Currently, some form of on is used in approx: mately 60% ofp fal studies (Christensen, 1988). Remember the neg ative ion study (Baron, Russell, & Arms, 1985) described in the first chapte Besides merely testing the ellects of dilleren levels of ionization on subjects roods, the researchers wanted to know if diferent levels of negative ions also ‘would affect aggressive behavior-—as the seasonal winds ate believed to do. Fo tes this hypothesis in the kind of controlled laboratory condi sary to manipulate ion levels, they needed to use deception to «1 ion where aggressiveness could occur and where matically Parts of their experimental procedures were based on the famous laboratory studies of obedience conducted in the 1960s and 1970s by Stan ley Milgram (Milgram, 1 7 Milgram experiments and the controversy surrounding Milgram’s use of deception, you can explore the online resources at the end of this chapter,oe nae Box 2-3 APA Standard 6.11: Informed Consent to | Research* a. Pojhologs use Imgunge ths, sesonably underandable to research paricpans in obtaining their appropriate lelormed consent Seep & provided in Sandal 6 12, Dupensing wh Informed Consent (lseused inthe tet as “nimalnske suds) Such informed consent approprtely documented > Using lngunge thats esonaby understandable opaccipans py hologss infor parcipans tthe natre ofthe research they tnlomn paripants that they ae fe co parhate orto decine © | ‘participate or to withdraw from the research; they explain the foresee- | | Bie consequences of decinng or withdrawing, they inform pare | Seater a ne trilngness to paripate auch asst, discomfort, adverse elects, oT Cee ease Gee ence ee ee ee eee a eae Saree cae hes ope epi gs ne 9 Eee ee ea beets Se eee Oe eee ae ee ees cae Re ite ee eee etre eee eee eee ara penta a Ae ere eee Bere eee eee ee oe eee See ee enna oie oc“ Pancps of Fylogia ad Code af anda” 1992, Amen Pas Toor eitCopmngh 61092 Aenea Pell Asocaton Rep meh erin To create a situation in which aggressiveness could occur, Baron, Russell, and Arms enlisted the help of a confederate (in this case, a student accomplice). ‘When one of the male subjects showed up for a session, the contederate was CHAPTER 2 RESEARCH ETHICS ready there, posing as another subject. The experimenter explained that the experiment involved learning to decrease heart rate using biofeedback. One subject was to be the “teacher,” the othe the “learner." They drew slips of paper to select rolee—but the drawing was rigged, and the subject was always the teacher. During the “wrining session," the learner made mistakes, Every ime hhe made a mistake, the teacher was supposed to punish him by pushing a but: ton on @ machine that (supposedly) sent an vnpleasant, but rot dangerous, burst of heat to a cuff on the learner wrist. The ten butions on the machine were labeled from very low to very intense, and the teacher was told to select any level he wanted, During the course of the session, the teacher had 20 ‘opportunities to punish (Le. display aggression toward) the learner. ‘Of course, no real heat was ever delivered in this experiment, although subjects believed that they were punishing the learner. The intensity levels, selected by each subject, however, wee recorded. (In addition, you will recall, this confederate angered half the subjects. Before the “learning” tals began, he provoked the real subject through a series of scripted nasty comments, These, of course, were part of the deception, too.) The entire nase was thor- oughly explained to each subject at the end of the session How do research subjects feel sbout being deceived in this way? Many studies have been conducted to determine the answer to this question. In a recent review of these studies, Christensen (1988) reports that most research subjects are apperently not bothered by deception, This review ofthe Hterature, which has attempted to document the impact of deception on research participants, hae consstmly revealed that research pt ‘espns do not perelve thatthe’ are harmed and do nat seem to mind being misled. Infact, evidence exists suggesting that deception experiments are more ‘joy and beneficial dha nondecepion experiments. (P680) Interestingly, Cistensen also reports that research professionals are much more bothered by deception than subjects are. The consensus among researchers, how: ‘ever, seems to be thatthe use of deception is justified by the knowledge that is fined Guls & Rosnow, 1988). Many important psychological problems cannot be studied without the use of deception, and it may be argued, as Christensen has, that failing to study important problems is even less justifiable. The standards covering the use of deception are shown in Box 2-4 ‘Sometimes a small omission, or outright deception, is necessary to make an appropriate test of the experimental hypothesis. How can this be recon: caled with the principles of informed consent? The answer isin APA standard 6.150), shown in Box 2-4. The deception must not influence a subjects deci sion to take patt in the research—any deception that is used must be such that subjects would not refuse to participate if they knew what wes really hap: pening. For instance, ic would not be ethical to reeruit subjects fora learn {ng experiment without telling them that we intend to punish thelr incorrect, responses by exposing them to the sound of scratching nails on a blackboard Because many subjects might decline to participate in such a study, our decep: ton would be unethical. In contast, as far as we know, Baron and colleagues as6 PARTONE msTRODUCTION Bo ae) | APA Standard 6.15: Deception in | Research* 4 Psychologists do not conduct a study involving decepsion unless they | have determined thatthe use of deceptive techniques is justified by the study prospective scientific, educational, ot applied value and that equally effective altemative procedures that do not use deception are not feasible , Peychologiss never deceive research participants about significant aspects that woul affect cher willingness to participate, such as phys teal nsls, discomfort, or unpleasant emotional experiences ‘© Any other deception that isan integral leature ofthe design and con: duct of an experiment must be explained co participants as early asi feasible, preferably tthe conclusion of thee participation, but no later than avthe conclusion ofthe research (APA Standard 6,15) fro Eo Pines Phlogis and Co of Cond 992, Ancram Bvt 47, | 1SGT- 16. Cope 0109 Anencn Pela Aston. Reprod wes perms subjects would probably have consented to participate in the experiment even Af they knew that they were not really punishing the “learners” errors. Fur thermore, the researchers adhered to the principle of full disclosure by com. pletely debriefing subjects at the end of the experiment—that is, explining, the true nature and purpose of the study. Even if subjecs are debriefed, will the explanation completely undo the effects of deception? Perhaps not. Bramel’ (1963) study of projection is an example of an experiment in which debriefing might have been insufficient. Bramel was interested in studying auributive projection, which is the process of projecting our own traits onto another person. The tals projected are waits the person is consciously aware of having, making this phenomenon different from the classical projection that Freud described. (In classical projection, the traits projected are thse the person is not consciously aware of possessing ) To ensure that subjects were aware of possessing a trait they might pro} Brame] employed a procedural deception. Male subjects were shown pho- tographs of males in various stages of nudity. Subjects were given false fed: back about their degree of sexual arousal to the pictues, which led them 10 believe they possessed homosexual tendencies. Bramel then tested the subjects CGHUAPTER 2 RESEARCH ETHICS for projection of sexual arousal onto others. He asked them to estimate how aroused other people would be when they saw the pictures. Bramel found that Subjects projected arousal onto people who were similar to themselves (other students) but not onto people who were unlike themselves (criminals) ‘Of course, Bramel debriefed all his scbjects at the end of the experiment; he told ther the feedback had been false and that there was no indlation of homosexual tendencies in their responses because this was not even being measured. But was the explanation sufficient? Te As possible that for subjects who had doubts about their sexual identity, the bogus feedback aroused con- siderable anxiety and discomfort. It is also possible that subjects may have doubted Bramelsfnal full disclosue. if he admitted deceiving them about the feedback, perhaps he was also deceiving them about theit real responses. At the very least, subjects may have felt somewhat foolish at having been duped the experimenter, Whether the effects of deception can ever be fully rrsed by debriefing remains a serious ethical question. Regardless of any later explanation, the subjects’ anxiety and discomfort during the experiment were real, Once done, these experiences cannot be undone, We cannot always reverse our experimental effects, so it makes sense to avoid using procedures that are potentialy harmful, painful, or upsetting to subjects whenever possible. In addition te ther dublous ethical standing, such procedures often add litle to our understanding of behavioral processes. For Instance, we already know that high anxety has debilitating eflects on many behaviors. What would it add to our understanding of the thinking process to find that subjects learn nursery rhymes or solve riddles less efficiently alter they are made extremely afraid by a noxous laboratory procedure? ‘Some kinds of important research wil have unavotdably negative conse- quences fr subjects (because negative consequences are precisely the eflect being investigated), In cases like this, special swps need to be taken to remove any harmful effecs, Experiments on the eflecs of violent pornography ate a good ‘example. When subjects, typically male undergraduates, are exposed to high lev- cls of violent pornography in psychology experiments, they become more accept- ing of the “rape myth’ than nonexposed men do (Donnersiein, Linz, & Penrod, 1987). They are more likely to report callous attitudes toward women, as reflected in such statements a8 “Women really want co be raped." They are also more likely o say that they would consider raping a wornan if they knew they ‘would not be caught, Unfortunately once esablished, ches attitudes do not seer to disappear quickly on their own. Researchers discovered that a simple debrief ing at the end of the experiment was not enough to erase the negtve ellects produced by the experimental treatment. Instead, extensive postexperimental {ebrefings that detail the unralty of the themes and images in violent pornog- raphy are necessary In one experiment designed to assess the impact of exten: sive debriefing, a follow-up study was conducted two to four months ater men hhad participated in a violent pornography study. The researchers found that extensive debrefings helped to remove the harmful belies created by the films. “Two to four months later, men who had rot yet been debriefed showed signif cantly more accepting attitudes about the rape myth than did men who had received extensive debrefings (Donnerstein & Berkowitz, 1981) Fa8 PART ONE INTRODUCTION Anonymity and Confidentiality Maintaining anonymity and confidentiality is another important consideration for researchers It is our responsibility to protect the prvacy of research par- ticipants. When possible, dita should be collected anonymously and ident Ged only by code numbers, Practically, we do not seed wo adentily subjects by rname—most psychological research uses aggregated or group data and reports results as average scotes for each treatment group (these procedures will be ‘covered in later chapters), Data collected are Kept confidential. They must be stored in a secure place, end they may not be used for any purpose not ‘explained to the subject. Trey do not become items of gossip to be shared with friends. When shared with colleagues, data must also be treated with discretion and subjects’ identities protected. Fictitious names of mumbers (uch as subject 17) are used. Identifying details are disguised if there is a chance that a subject will be recognizable Protecting the Welfare of Animal Subjects The principles governing informed consent, debriefing, anonymity, and confi entity are clearly important aspects of ethical research involving human par Upants, There ate other standards that protec animal welfare—the humane care and treatment of animus. The care and treatment of animals in research 's regulated by the Animal Welfare Act of 1966, 2s amended in 1970 (Public Law 91-579), The Animal Welfare Act deals with general sandards for animal care. The act was recently reamended by Congress (Federal Register, February 15, 1991) to include new regulations that address the psychological well-being of higher animals. It requires, for example, that all dogs receive exercise and {hat primates are housed aleng with others of their species, As isthe case with research using human subjects, institutions engaged in animal research must hhave a teview board, called an institutional animal care and use committee ACUC), to evaluate experiments before they are conducted. Federal rel tlons stipulate that before animal experimentation ean be approved, possible skernatives muct be carefully considered, The JACUC must determine that the researchers have appropriately researched the alternatives and provided vit- ten documentation that no other altematves ate available, Several professional organizations around the country encourage ongoing ‘monitoring of care of laboratory animals through sell-policing by members of the scientific community The American Association for Laboratory Animal Sci= ence (ALAS), for example. publishes detalled taining manuals that include information on adequate hosing, sanitation, and nutrition forall animals used in research, The American Association for Accreditation of Laboratory Animal Care (AAALAC) also promotes uniform standards, AAALAC uses the USS. Department of Health, Education, and Welfare (1978) Guide for the Use and Care of Laboratory Animals as its principal set of guidelines. That publication provides specific standards for cage sizes, food and bedding, waste disposal, drugs and anesthesia, and many other aspects of animal care. The guide also FIGURE 2-1 A close elaionship can develop between researc human subjects. Uevseuman addresses issues of safety and sanitation for the benefit of animal handlers ‘working in the labs. APAS Committee on Animal Research and Ethics (CARE) has been influential in establishing national guidelines for animal welfare, APA inchudes standards for animal care among its ethical principles (see Box 2-5), ‘As with human participants, our concern about animal welfare involves avoiding any unnecessary pain or risk to the subject. Research involving any procedure that might be painful fo the animals, such as surgery, drug use, or Shock, must be closely supervised by a researcher specially trained in the field Despite the existence of legal and ethical guidelines. some ents have argued that animals have been sbused in some psychological studies—a violation of ‘thicel principles, In his book Animal Liberation (1975), Peter Singer chronicled numerous cases of animal abuse’ Many of the examples dealt with studies involving eletric shock or food deprivation. (However, an even greater portion of the book dealt with the treatment of animals being raised for food or used. to test consumer products.) Lets look at one of SingerS examples This case comes from the work of Joseph V. Brady (10958). For several years Brady studied the emotional behavior of rhesus monkeys. The monkeys were kept in restraining chairs, these allowed them to move their heads and limbs bbut not their bodies (ee Figure 2-2). They were placed in the chairs so that they could be tained through various conditioning procedures involving elec trie shock, The experimental setup, according to Brady, seemed to be highlyBOX 2-5 APA Standard 6.20: Care and Use of Animals in Research* ee er eae ratee merece cc ee Deena ett a eter ee carne ant. DE ire os oases el earreae | lls une each mets and expereince cere sins eve nui rave or tatep krona ersten ote co eee ee Ee eee lease TE ety anette Aibea daihar eal ee neonates vested ae See aed eee eames eee e fepee Mlistnd arias darmpsalssseing nares po Pa ee eae eR een bee eet Ce a ene Teg ese yates oem mye ene eee eae ee Se eee ae AG ee hae reer ae ere ne rae ee eee ee) ee {Aten appropri hte anal eb eles done Se ee ee ere ee ‘fron Eh Pini of Popo and Cie of Cass,” 1992, Amen Pag 1297-61 opyig ©1983 neta Pochlagial Aan Rene wh emiscen, stressful for the animals, Many of them died during the preliminary study ‘Autopsies showed that many of the dead subjects had developed ulcers, which are unusual in laboratory animals. Restraint alone was not the explanation; FIGURE 2-2. Rhesus mous being hep resting clas. 5 ra ena some animals had been kept in the retaining chairs for 6 months, received no shock, and did not develop uleers. Therefore, in subsequent work Brady explored the eect ofthe conditioning procedures. Brady trained two monkey, designating one an executive and one acontol. Both monkeys were given brief shocks on the feet. However, the executive monkey could prevent shock by pressing 4 lever. Unless the executive acted appropriately it (and its pare, the control monkey) would be shocked on the feet once every 20 seconds ‘The monkeys Were exposed to altemating 6-haue periods of shock avoidance and rex (no shock). This procedure continved for 23 days, afer which the exec- lutive monkey collapsed and’ was sacrificed. Subsequent experiments were con: ducted with diferent pars of monkeys and with diferent time inervals, ‘There are several ethical objections to this line of research. Fist, the use ofthe restraining chatrs alone is probably extremely distressing tothe animals (Gear in mind that some animals spent 6 months in these chairs.) Second, thea PART ONE. I TRODUCTION use of electric shock concerns many cris of animal research. Brady’ ori nal article (1958) does mot contain the specific level of shock used in the experiment. We are told only that the shocks were brief. Most of us try to avoid electric shock of any magnitude, no matter how brief. Monkeys appar- tently share this preference becanse they quickly earn to work to avoid shock Some of the animals were so distressed physiologealy by Bradys experiment ‘that they actually developed ulcers and other gastrointestinal problems. Many dled or were sacrificed so that their tissues could be studied for signs of ulcers. Third, ethical objections also can be raised because the study’ was poorly designed, Brady’ Findings were of diminished benefit because ofthe way the executive and control monkeys were selected ir the experiment. (We will return to this issue in Chapter 8.) Thus, in this case, it seems that Singer’ criticisms were probably justified although, in faimess, it should be pointed fut that Brady work predates the ethical guidelines that currently apply to research psychologists But what about more recent allegations of animal abuse? Psychologists Cole and Miller (1984) conducted a study to evaluate some extreme all gations of abuse. They reviewed all 608 articles reporting animal research that appeared in the major psychology journals between 1979 and 1983. Table 2-1 summarizes « portion of theit findings. Overall, the psychology research literature for this five-year period simply does not support the claims of extreme critics like Singer. Although shock and food deprivation ‘were used in some studies, the manipulations were by no means as extreme as reported by the critics of animal research, Infact, not a single one of the specific extreme allegations was supported by the literature. Shock was used in 10% of the studies; inescapable shocks stronger than .001 ampere? were used in 3.9% of all the studies and in just 0.2% of the studies involving monkeys, dogs, ot cats. The longest period of food or water deprivation reported was 48 hours, Less than 19 of the deprivation studies left animals without food or water for more than 24 hours, Most used 24 hours of depri vation, which corresponds to feeding once a day as is recommended for most house pets [Nor were any studies done ovt of idle curity, as entice had alleged. AS Colle and Miller (1984) noted: Experiments involving inescapable shock . .. were almed at understand sng mechanisms believed likely to be invalved in human depression, a con dition that often causes its victims such intense suflering that they ate dt. fe For young adults, suicide isthe third leading cause of death es occur occasionally, but they are by no means encouraged or condoned by the vast majority of psychological researchers. ron ir ge Cae Maes, 188, p70), T CHAPTER2 RESEARCH ETHICS 4 ‘TABLE 2-1 Accusations of animal abuses by psychologis and the percentage of aces in which such teatments andor resus were reported Percentage of Arles egstion 1m Which Reported “Annals are given intense repeated electric sheeks which 00% they cannot escape, unl they lee the ably even steam in pam ay longer” “They are deprived of foo and water to sulle and die 0% slowly fom hunger and shi” “They are pt in ttl ioation chambers unl hey are 00% driven insane, or even dle, fom despair and teror They ane subjected to crushing forces which smash thelr 00% bones and ruptae thet Internal organs Thee bs are mutated or ampuisted to predce 00% behavior changes “They ae the viims of extreme pain and sues, ince 00% pon them ovtof dle const, m nighemansh experiments ach (1977) ted sy aia ies en ama 0,“4 PART ONE. INTRODUCTION decide yes every time we open a can or a bag of pet food (Herzog, 1991) Although we might not want to give up our pels or become vegetarians, we tay still question whether animals suler needlessly. Singer made a strong case for this point of view. Other waiters share his opinion: Many research projects ate pointess, gatuitously repetive, fatally awed, ox concerned withthe obvious or the tiv If animal sufering is of any sg cance whatsoever, is infliction can only be justified fr leary-defined goals. But many experiments ae il-conceived or actully unconceved in that no hypothe sis has been formulated belore conducting the experiment. One ot avery few repetitions of an experiment may be needed to confirm the results, but some txperimens ae repeied time and again with no epstemologia!” jstifcation. (ler & Wiliams, 1983, p. 326) The animal rights movement has gained momentum in this country. AS fan indication of the numbers of people who consider themselves to be involved in the movement for animal rights, 24,000 people attended a rally jn Washington, D.C. in June 1990, in-a survey of 574 highly committed activists attending the rally, Plous (1991) found that approximately 85% of them would like to see a total ban on the use of animals for esearch; the comparison figure from a sample of 54 nonactvists (people who Just bap- pened to be walking by the protest) was only 17%. Sixty-one pereént of the Activists surveyed alzo said they favored laboratory brealcins, whereas only 14% in the nonactivist sample favored them. Clearly, most people see some Dbenefits in animal research, See Box 2-6 for @ report on one recent breakin There is litle doubt that humans have benefited from animal research Miller has pointed out that psychological experiments with animals have led to treatments of many psychological problems (such as depression, obsessive compulsive disorder, enuresis [bedwetting], and anorexia nervosa) Studying animals has also advanced our knowledge of human behavior in the areas of tion, conditioning, learning, memory, and stress (Domjan & Purdy, A policy statement by AALAS (October 26, 1984) presen the case for continued experimentation using animals this way ‘The se of experimental animals has been shown tobe essential tothe develop- rent af an undersanding af many biochemical, hehavorl, physiologeal, nd disease proceses. Many of the factors that affect both animal and human life can only be sted in intact animal systems by syxematically manipulating specie research varables. Given an incomplete knowledge of biological systems, iti Inconcevable that animal experimentation can be replaced, in the foresceable future, by mechanical models or other incomplete biological systems. The charges of animal rights activists, although exaggerated and overly dra matic at times, do heighten Our awareness ofthe issues. And discussion of the {ssues has affected the amount of animal research that conducted: Fewer an mals ae being used in research each year (Rowan & Andrutis, 1990), However, Johnson (1990) angues that “the animal rights movement has misled the law. ‘makers and citizens of the country into sarifcing human lives for animal lives That is the stark realy” (p. 214) * pcm eso he hrng ge Box 2-6 | Activists Target Minnesota Psychology Lab* | (On April 5, 1999, animal rights activists representing the Animal Lib ration Front vandalized the animal research laboratory at the Uni sity of Minnesota, destroying equipment and stealing research data They also took more than 100 research animals and released them in a | nearby field. The damage was estimated at more than $2 milion. Van- dalism of animal research laboratories has been a federal offense since | | | 1992, but this was not the frst ime Minnesota had been a terget of n= ‘mal activists, During a previous protest, activists chained themselves together inside the olfce of @ psychologist who conducted research on primates. Once, a Minnesota student hoping to force a dialogue with “university officals, ung for several hours from a rope connected to the roof of a building In addition to monetary damages, the vandalism destroyed many months of researchers’ work For example, the research of one doctoral student who had been investigating learning and memory was com: pletely destroyed: She had spent months training 27 pigeons who were Erolen. Fourteen of her pigeons were eventually recovered, but it was not kniown whether they could ever be used in the research again, | Releasing laboratory animals used to captivity into the wild often ‘means certain death. Some of the released animals were found by the side of the road, but a number of released rats had already died. AS a result of the animal deaths, the Minnesota Senate passed a law making. it a crime punishable by up to a year in jail plus civil damages for a | person or organization to claim responsibilty for the unauthorized release of animals “Aap fo “Seracive Lab Atak Sens Wako Cal by B.A APM Mer One psn 1992, 300. 16. Copyright ©1993 by the nerean ecopal fancason | duped eh pein | Certainly, as long as animal experimentation continues, researchers have an obligation to behave responsibly. The situations causing the grievances of the past cannot be undone, but the picture for the fucure of animal welfare is hope- ful, Before 1987, the E in CARE stood for “experimentation.” It now stands for ‘ethis." The challenge to contemporary researchers is to ensure that each bbuman or animal subject contributes something wortherhle to scientific knowl: edge. In the chapters that follow, we will eturn to ethical questions about hhuman and animal subjects in our examples of actual research projects,6 PART ONE. INTRODUCTION Fraud in Science So far, we have discussed the ethical conoems of researchers designing exper iments involving human ot animal subjects. Reporting on research sa nec: essary part of the scientific proces. [ts also an ethical matter. We report our procedures and findings honestly and accurately. When we think of fraud in EXience, we typically think about researchers publishing false data, Data fl freation is a reach ofthe ettical principles stated in APA standard 6 21, “Py Chologiste do not fabricate data or falsify results in their publications.” As well Ge individual ethics, several important safeguards are built into the public process of sclentific reporting that help to keep fraud in check. Tse, with very few exceptions, research articles submitted for publication ave reviewed by the editor cf the periodical and by several experts inthe field before they car be accepted for publication. This procedure, called peer review, ie am important part of the Teporing process. The reviewers task is to assess the worth of a submission; reviewers go over each submission with a fine-tooth tomb, ooking lor problems and suggesting improvements. I there is something ‘Stange about the findings, expert reviewers are likely to discover it. (Only about 5% to 20% ofall artiles submited for publication in psychological journals ‘ever take it Into print) Reviewers make recommendations about publication to the editor of the journal who takes the final decision. The reviewers and the editor, then, ae a fst hne of defense against data falsification Replication isthe second line of defense, Researchers often attempt 10 replicate the published findings of others, particularly if those findings are sur- pissing, novel, or important. If data have been falsified, itis unlikely that the Experiment will be successfully replicated! Third, the competitive natute of academic peychology works against fraud. (even though it may also be the primary cause!) Tenure and promotion within teademic departments of psychology are partly based on research producti ity You have probably heard the saying, "Publish or perish.” There is strong, pressure on researchers to publish, and the pressure is probably strongest on those whose work has resulted in a series of failures. Fabricating some impres tive data might seem very tempting when keeping one’ job depends on pub- lishing the results of a successful experiment. However, competition among, colleagues for limited research resources can be a strong deterrent to fraud. is federal funding sourees become increasingly scarce, the competition sti ns, as resources become rare limited, other researchers in the same area will bbe on the alert for fraud, Despite these safeguards, fraud does occu from time ‘One of the most specracular cases involved the famous English psychol- cogist Sir Cyril Burt, who was awarded APAS prestigious Thorndike Award for ‘lstinguished service to educational psychology shortly before he died in 1971 (Green, 1992). Burt had achieved fame for his published studies on the inher ttablty of 1 in identical twins and fraternal twins who were raised apart (Bur, 1966, 1972). His data showed that the IQ scores of identical twins {twins whose genes are exactly the same) were almost identical even though (CHAPTER? RESEARCH ETHICS Wass 10? Tears Pee Review?" hey had been exposed to very diferente experiences conversely the 1Q sores of aterm wins (vin who share only ction of tht ene) could die realy. This wa very portant esac in ts tine. Eventually, ter esearch rote that some ofthe statistical values he reported in diferent articles were ‘say Ue same even tough But ad eponed using diferent nombers of tn. Siastally speaking, this isa highly improbable, aot inposabl, event. I vreely beloved tha Da fed ores, although his dcendes apse per Ssvely dat he wat merely "sloppy about reporting his data (hich he had detally’ topped colecting in 1939) They make the point that if Burt had isended fod be would have done a beter jo of (ren, 1992, p. 330) ate een de led das fon sds oh een of sypeactviy in children ith simula drug In the 1980s, a poyhologt named Stephen Breungadmted labreating daa showing hat sna sch fs Rian and Dexedrine could cram reduce hyperctivty He received « ing fede! gran om the Nal Lite for Metal Hels (NIMH on the bass of fae data. One of is colleagues suspected tht Breuning had alsfed Is data and reported these suspicions to IMAL Alera tre yea scene mi conde vestigation conducted by NIME,Brening ated in a plea brats ovPART ONE. INTRODUCTION that he had falsified his data on two occasions; in return NIMH dropped per- jury charges (Byrne, 1988). Breuning’ case, among others, has resulted In natlonal discussions about the definition ofscientfic miscondiict used by the two major granting agencies [NIMH and the National Science Foundation (NSF). The current defitions are vague and subjective; for example, NSE describes misconduct as "fabrication, falsication, plagiarism, or other serious deviation from accepted practices in proposing, carrying out, or reporting results from activites funded by NSF" (NSF [45CFR Part 6891 (aX1)). Whatever definition is used, the consequences ‘of scientific misconduct can be severe indeed for a researcher, anging from sus- pension of fring by the university to prison term if convicted in court. Uni- versie can be penalized as wel; they can be forced to return monies and can be excluded from future funding, In Breuning’ case, his university was required to retum more than $135,000 in funds to NIMH, and Breuning no longer has research carer. In almost all cases, funding from federal grant like Breuning’ is actually given to universes, rather than to individual researchers, The uni versity controls and monitors all spending, As funding becomes even scarcer, more frequent investigations and lighter contols ate expected Plagiarism Let us tum to what is probably a more common kind of fsud, plagiarism, We ‘must be careful in the way we draw upon the work of others in our presenta tions, To plagisrize means to represent someone elses ideas, words, of writen ‘work as your own. Plagiarism is a serious breach of ethics andl can result in legal action. It's not only “borrowing” facts and figures from someone else, plagiarism includes using someone else ideas without giving proper credit. A wellnown author was sued by # former professor who claimed the ideas in the authors best selling book carne directly ftom the professr$ leeures, The professor won his Sule and now recewves share of the authors royale. Although some fons of plagiarism are intentional, others occur through sim: ple oversights. It would be a rather large oversight to forget to give authors credit for direct quotations used, but what about paraphrasing their work—using theit ideas but changing the words around? Paraphrasing without giving credit is rep- resenting someane eles idea as your own tis lzo plagiarism, Even if you believe that the information you wish to use is “common knowledge,” if you have read it im someone else’ book, give the author credit Ifyou have done a thorough review of the literature. your research report should contain many citations? This way the reader wall know that you have read the important research in your topic area, Throughout this book you wil find figures, passages, and photographs wsed with the writen permissions of the original authors, ass, and publishers. * The ato gen or Rep (196) en tenting met ed pouty don ea oes aps ‘he bes nerf nis 2 pod xml lhe Eee ha rte a hs gue Grd {>be we), bt we ce pn Ren bled ail wh eiung eee on rand ce (CHAPTER 2 RESEARCH ETHICS Unfortunately cis easy 10 plagiarize without being aware of it, We read many things, jot down notes, and later forget where we got the information. we use that information elsewhere, we may inadvertently forget or be unable to name the source, In preparing a research report, you should follow several, guidelines to avoid the possibility of plagiarism (these guidelines also apply to works that are unpublished or not yet published and to sources in the media of on the Internet) 1, Take complete notes, which include a complete citation of the source authors name, tile of aricle, journal rame, volume number, year of publ cation, and page numbers. For books, include the authors name, te tile of the book, and the publisher’ name and city 2, Within your repor, identify the source of any ideas, words, or information that are not your own. 3. Identify any direct quotes by quotation marks at the beginning end end of the quotes and indicate where you got the quote (include page numbers) 4, Be careful with paraphrasing (restating someone else words). There isa great temptation to lift whole phrases or catchy words from another source. Use your own words instead or use quotes. Be sure to give credit to your source 5. Include a complete list of references at the end of the report. References should include all the information listed in item 1 6, Ifin doubt about whether a citation is necessary, cite the source anyway. You will do no harm by being especially cautious. Summary A well-planned experiment includes carefil reatment ofthe subjects who par ticipate. Today federal law regulates some aspects of psychological research. An ‘ntitution must, for example, have an institutional review board (RB) to approve cach study using human participants. The IRBS main tasks are 10 ensure the salety of human subjects and to conduct a ris/benfi analysis, and many of the law’ provisions are reflected in the ethical guidelines of the American Psycho- logical Association. Most important is obtlting the informed consent ofall those who will be participants in an experiment, particulary f the study places them ct rk. This consent must be given freely. without force or coercion, The per son must algo understand that he or she is free to drop out of the experiment at any time. In adeition, subjects must be given as much information about the txperiment as possible so that they can make a reasonable decision about whether to participate or not, Many institutions require that informed consent be obtained even ifthe research is considered minimal risk, Sometimes a researcher may need to disguise the true purpose of the study to ensure that subjects will behave naturally and spontaneously. In experiments that require some deception, subjects must be debrieed. But, because simply {debriefing subjects does not guarantee that ve can undo any upset we have caused them, researchers iy t0 avoid exposing subjects to any unnecessary pain o risk.ART ONE uvTRODUCTION When posbl, data should be collted anonymously and identified nly by ode tumbes, Onc colle, dita ate hept conden they may bot be used or Ey purpse nt expaned tothe sect Wen they ae repoed, daa should Sted cote numbers tous ames o pte ese, eal principles apply in research with animals too, rsitaons engage in animal reach mst have a reve board, called a ntiatonal animal are and use comme QACUO, to evant experiments Defoe they are conducted searcher havea respons o promote animal welfare whenever they use hima subjecs, Animals must Tecuve adequate physal care to say healihy fa comforable dros, surgery. or any potemally panfl procedures are Grolved,areeicher who is specially trained Inthe ld must closely super vise te nial, Despealegations by eis of animal esearch, tere i i= is evdence to support accusations of widespread abuse im psyehologcal research, Some cuts advocate anima gh, he poston that al species ae tually valued and have equal ight, and the animal rights movement has Tecenty gained momentum. Most people, however, se a need for animal Cxperimeraton becuse of beneis for human wellze Fraud in scence ts yplaly thought ofa ialying or Gbricaung dau clea fraud i unethical The peer review process, replication, and sry by coleagues hep old fraud in check. Paprom, representing someone eles Work as Jour ovn, iva seus beach of eles andi also considered a ype Saud, Reseateher mast be crf to give proper credit to others who con tributed words or ideas to thelr work ‘Rey Terms Animal rights The concept that all sensste species who fee pain are of equal value nd have rights ‘Animal welfare’ The humane care and treatment of animals, [Atisk ‘The Hisehood ofa subjects being harmed in soe way because ofthe nature ofthe research Debriefing The principle of full disclosure at the end of an experiment that is, explaining to the subject the nature and purpose of the sud Fraud’ The unethical practice of falsying of fabricating data; plagiarism i also a form of faud Informed consent. A subjects voluntary agreement to participate ina research pro ct afer the nature and purpose ofthe study have been explained. Institutional animal care and use committee (IACUC) An institutional commit- tee that reviews proposed research to safeguard the welfare of animal subjects. stitutlonal review board (IRB) An institutional commie that reviews proposed research 10 safeguard the safety and rights of human participants. Minimal risk. The subjects ods of being harmed are not increased by the research Plagiarism The representation of someone ese ideas, words, or wren work as ones own; a serious breach of ethics that can resul in legal action Rish/benefit analysis A deermination, made by an institutional review board, tha ‘any risks to the individual ae outweighed by potential benefits or the Impor- tance ofthe knowledge to be gained Beview and Study Questions 1. What age the tasks ofan institutions review board (RB)? 2, What is informed consent? 3, What isthe principle of fll discosute? 4. Ac the end of the semeser, all stdems in a general psychology course are told they oll not receve ced forthe cause unless they take pat in the instru tors research project. Students who wehuse 1 participate are given “Incompletes" tnd do not get credit forthe couse Is this Unethical? Which principles of the [APA guidelines have been wolated? 5. An experimenter studying the effects of stress gave subjects a series of maze prob- Jems to solve. The subjects were led to believe that she problems were all gute cas. In fac, several had no solution. Some ofthe subjects were vis upset by ‘hts inability to sole the problems, At the end of the study the experimenter fave no explanation ofthe procedures Wht ethical principles epply i this ase? ‘What should the experimenter have done? 6. In study of sexual attiudes, a stdent experimenter finds chat Pat, a frends spouse, has responded yes to the question, Have you ever had an extramarital alla?” The student i sre thatthe fend is unawate of Pats behavior. The st Gent decides to show Pats answers tothe frend. What ethical principles have ‘ben violated? How could this stuatin have been avoided? Have you ever been a subject ina restarch project? Ifo, how were tial pin pe app a dy? . . 8, What are the responsiblies ofan tsituional animal care and se commitee aacuo? 9. What ethical principles apply wien we do research with animal 10, Explain the concepes of animal welfare and animal rights. 11, To study the effect of a new drug to reduce depression, researchers mst sac fice animal subjects and lssect thelr brains, Discuss the ethical pres and eons. ofthis line of research, 12, What efraud? Describe the external presutes that can produce fad, Describe the saleguards tht Keep icin check, What ave the posible penalies for scien bc miseondct? 13, Lee had put off doing a lab report ut the end ofthe term, He ws badly pressed for ume. His roommate sai "No problem | took that course last year. You can. tase my report. Just put yout name en i” Lee decides i all right 0 use the paper because he has the authors consent. Do you agree? Why or why not?PARTONE INTRODUCTION Critical Thinking Exercise The myth: Opposites attract The scientific findings: We are much more likely to be attracted to someone who shares our atutudes, values, and preferences; in other words, someone who curselves (Sprecher & Duck, 1994) The problem: Suppose a researcher conducted an experiment to test this, In one condition of the experiment, she used confederates who were instructed to agree with everything an opposite-sex subject said whale they were geting acquainted, At the end ofthe experiment, she measured how strongly the subject was atracted to the conlederat The researcher could not tll subjects that they would be lied to before the experiment, but she did explain the deception atthe end. What ethical arguments could the researcher use to justify this experiment? & Online Resources ind the APA guidelines, Eth Conduct (1992), online at wwwapa.org/ rincpls of Psychologists and Cade of ebicslcode html Reviews of the controversy surrounding Milgram’ obedience studies can be accessed atthe following Web sites bap /wwwrfde.calpolyeduibarts/eslenv/L01/Obey/Ethics html bp stanford edu/-krollagorg_site/soc_psychmilgram. Te APA guideline fr research with animals can be found atthe following sit: sworwapa.org/science/angeide html CHAPTER 3 Alternatives to Experimentation: Nonexperimental Designs Describing Research Activities Nonexperimental Approaches, Phenomenology Case Studies Field Studies Survey Research opting Probability Sampling Nonprobability Sampling Summary Key Terms Review and Study Questions Critical Thinking Exercise Online Resources *PART ONE. INTRODUCTION CHAPTER OBJECTIVES Learn about techniques for studying behavior that do not ‘manipulate antecedent conditions: phenomenology, case studies, field studies, and survey research . Learn the factors involved in designing questionnaires Learn the pros and cons of different sampling techniques Ta ine tradnional poe reste specie st athe tradtonal psychology experiment, we create specif sets of te the experimental approich, the teseacher must be able to Set up these Conaitons Yor any indvidesl who ends up as. subject in he experiment Many times this requrement cannot and sometimes should no) be met. For trample, we may wish to study characteristics of the parttpants (atch a5 gender or personality) to see how they influence behavior, Because we eat~ dita in other ways. Sometimes the conditions a researcher wants to sty has pray or aol) cant be cat thie an rent not feasible or desable They ae also wed whenever texting = typothests nan existing real ie steaton is necessary or important. None perimental methods ae wed to study behavots in natura sexing (dren Playing, chimps pening, of les 4 gang, o explore unig or ae occur fences (a case of multiple personality presidental election: or to sample persona information (attitudes, opinions, preferences). Because each of these fonexperimental approaches can provide wsefl data, ether fom singe ind Silo rom age groups of people tis important to understand Row they ate applied in peyhologal researeh ‘Ave dictated in Cape 1, the primary purpose of an experiment is 0 establish asa raonsip between spate se of antecedent condlons Geant) and the subsequent observed behavior. The degree t which & ‘ecnch design allows us ome caus statement calle the itera val ity ofthe design. Research high i internal vay fe can conclude that tHe sntecedents caused the observed dilerenct in btavor among tbe vous groups of subjects inthe experimen. Researches ile tose labortory exper nents in prt becase they are potently high in itera vality Ce sy oteally” because experiments an fl sor of ths gal na vary of ways, thi wil be dlacused in Inter chapter) Nevertheless, experiments ae fe Surly eticzed in the media, by students, by ype, and by seasoned profesionals—for being arial and nels, and to sme exert this ex Sm ay be used. re experients olen seem to lak external validity that, prnerliblty or appleby to stuatons outside the research setting When dberatons canbe generalized to athe stings and other people hey CHAPTER’ ALTERNATIVES TO Exes MENTATION: NONEXPERIMENTAL DESIGNS are high in enteral validly: Nonesperimental designs are oten prefered ‘cause they may ave getr eral val. ead 10 theta world may be more appuent. There i, however, some trade. What we gan imecernal vl we gt ein nena vay A sty that fechingy relic might brig is no clower tothe "ru than one chat seems painflly contrived Describing Research Activities All approaches to research can be described along two major dimensions: (1) the jee of manipulation of antecedent conitions, and (2) the degre of pos tion of units (Willems, 1969) The degree of manipulation of antecedents thee ‘ell varies fom lw to igh, om leting things happens they wall to set ting up caeflly contolled conditons. For example. a stady' could involve simply vacking behavior along with subjects normal diets (ow matipulaion of amuecedents). Or atthe other exreme, we could place subjects on fixed des ‘wher sll meals are provided (high manipulation of antecedents), Selecting «high Gegeeof manipulation has ts pros and cons: We could then vary meal in pe- determined ways and evaluate subsequent changes in behavior, but we would be imposing on subjects artificial conitons that might have lle meaning i real hfe. Experiments ate typically high in degree of manipulation; nonexpen: tents ae stall lw “The degree of imposition of units isan equally important dimension, This term refers tothe extent to which the researcher consis, of limits, the responses a subject may contribute tothe data, In sty of teenagers, for example, we might simply watch a group of teenagers and record whatever they say oF do (ow impostion of unis). With sch a plan, we would be imposing relatively litle constraint on the teens responses. But suppose that instead we were interested only in 8 single behavior, such asthe amount of time they lnten to rock music, We might then limit our subjects responses to the answer to the question, “How much time do you spend lstering to rock music each day?” Our study would then score high onthe dimension of inmposed ums. As you wil se i later chapters, mest experiments limit sub jee Inputs to a sarow range of responses, placing such experiments high on the acle of imposed units: Nonexpeimertalcesigo. as you wil discover tn this chapter and the next, can vay from low to high imposition of units ‘Antecedent conditions and imposed units can change independently and we can represent the various reseatch approaches vitally as shown Figure 3-1. As the figure lustrates, a range of posible research approaches exists, Laboratory experiments, which tend to al in the high-high range represent only one approach. As we discuss nonexperimental designs, We onl refer to this gure again s36 PARTONE. INTRODUCTION 2g 4 tom f Meum Madam 5 law Median High Degeeof manipulation fantcadent condions FIGURE 3-1 A space for desenbing research activites, Foe ‘Png Rail fr even Ranh by EP Wiens In EE iene at, Ran (Es), 98, Newt w ore n Pha Reach Coy 1989 Hol, Rena nd Win Reprod by perms. Nonexperimental Approaches Let us now discuss four common nonexperimental approaches used by psy- chologists: (1) phenomenology, (2) case studies, (3) field studies, and ( survey research. Together, these approaches form an important source of data on human as well as animal behavior. They permit us to gather infor- mation and gain understanding when experimentation is not possible or desirable Phenomenology So far we have discussed the scientific method in terms of observing and recording events that are assumed ro be external to the observer. An impor- tant supplement to the scientific method is the phenomenological approach, Phenomenology isthe description of one’ own immediate experience. Rather than looking at behaviors and events that are external to us, we begin with our own experience as a source of dat [Asa research approach, phenomenology often falls near the low-low end fon our graph of research activities (Figure 3-1). Antecedents are not manip: Slated, and the data may consist of any immediate experience; no constraints fare imposed. Much early work in psychology was based on the phenomeno- logical approach, Boring (1950) cites Purkinje as a good example of the phe nomenologically based researcher, Purkinje was interested in the physiology of vision and noticed that colors seemed to change as twilight deepened; reds appeared black, and blues retained their hue. This observation (now called the Purkinje phenomenon) eventually led to our understanding of the spec tral sensitivity of the rods and cones ofthe eye FIGURE 3.2 Willam James (1881-1910) nial ayo eine Wiliam James Figure 3-2) lo used the phenomenclogil approach. In his Principles of Psychology (1930, ongnal 1690, James dealt with base py chological ses, including hai, emotions, conciousness, and the seam of thought. James approached many tess fom the petpective of hs own experience. One of his most appealing pastages dels wis hie own dificlty Invgeting up in the morning. He pomted ov that out resistance to geting up ibis our movement, While we concenixe onthe pleasure of wart ‘heats and the ded of cold oor, we ae parle. Sad Jes," may gen Cris fa my egress nt pp win ay Stugale o dectsion a al. We suddenly find that we have got op (p. 528), Thetlore if we dont ves, we ought fo beable to rise wiout efor. The phenomenologial approach precludes experimental manipustion Compation of behaviors unde diferent restmentcondons isnot feed When sing thi pprosch we simply sind wo our own experience. As Dor ing 1950) explained Since phenomenology deals with immediate experience, its conclusions are instantaneous. They emerge at once and need not wait upon the results of cal calations derived from measurements Nor does a phenomenclogs use states, Since a frequency does not occur at an Instant and cannot be Immediately observed. (p. 602), Thus, the phenomenological approach is applied to a small sample of subjects — a sample of one. We cannot be sure that the process we are observing in our selves is not altered in some way by our atention to it, Because the observer is also the person whose process is observed, we may not be able to achieve the degree of accuracy and objectivity through phenomenology that we might achievePART ONE. INTRODUCTION through other methods. Nor are our private experiences observable in public it wal be dificult for others to replicate our experiences and apply scien criteria to our findings. Without replication, it cannot be known if others would have the samme experiences. If Purkinje had been color blind, his experience at sundown would have been very different from that of most people Phenomenology cannot he used to understand the causes of behavior. Like other nonexperimental designs, phenomenology describes, but cannot explain, behavior. Purkinje could not be certain that his experience of altered color was caused by an external change (a change in the amount of light that would affect all observers ina similar manner. Had he not been a scientist, he might hhave explained his experience in terms of a demon that had taken possesion cof is sense organs. In the absence of further evidence, one explanation would have been as good as the other ‘Phenomenology may lead us into areas of discovery that might otherwise {go unnoticed. It can be a useful source of information that may lead us to formulate hypotheses (we will return (0 this issue in Chapter 5), but experi- ‘mentation is stil equired to determine which antecedent conditions produce the behavior or experience, Elements of phenomenology are frequently com: bined with other research methods, however. In fact, that is protably one of the most commor. uses ofthe approach today. Experimentation, for example often relies on phenomenological, or self-repor, data to study the effects of ‘arious experimental manipulations. Recall that in the experiment by Baron and his colleagues (1985) cited in Chapter 1, subjects reported on their own toods. Our next method, the case study, males extensive use of se-reports. Case Studies Like phenomenology, the case study method involves no manpulation of antecedent conditions, The ease study is a descriptive record of an indiv uals experiences, or behaviors, or both, kept by an outside obse-ver. Such a record may be produced by systemaucilly recording experiences and behav- iors as they have occurred overtime. Clinical psychology, in particular, has relied heavily on case studies. There are generally few restrictions on the type of data to be included in 1 case study, Thus, case studies would be expected to fll inthe low-low por tion of our graphic scheme, The exact procedures used to producea case study. will depend on the purpose of the study: Sometimes, as in the clinical case, swe may work from a record made after the fact; the cient or ether know cedgeable source provides information concerning events inthe client life and the clients reactions and behaviors Kazdin (1992) has argued that case studies serve five major purposes: 1, They ate a source of inferences, hypotheses, and theories. 2. They are a source for developing therapy techniques, 3. They allow the study of rare phenomena (CHAPTER ALTERNATIVES TO EXPERIMENTATION: NONEXPERIMENTAL DESIGNS 4, They provide exceptions, or counterinstances, to accepted ideas, theories, oF practices 5. They have persuasive and motivational value Let us look at each point in turn. First, we may use a case study to make inferences about the impact of life events, the origin of disorders, or developmental processes (Gee Box 3-1). For cexample, this approach provided the first systematic data on the development of children motor, cognitive, and linguistic abilities. By making extensive records of {the behaviors of individual cilden,ealy researchers like Velen (1943) and Piaget, (1954) arrived at descriptions of hormal developmental sequences, Psychody. natn development has been inferred from case studies. Freud case of Lite Hans (Freud, 1933) is an example of how an individual cate may suggest a develop- ‘mental proces. Hans was afraid of horses. Freud analysis of Hans’ conversations with his father and the dzeams he reported suggested thatthe fear of horses was symbol for Hanst far of his father and arety about castration, Such case stud: tes led to Freud formulation of the theory of the Oedipus complex The case study provides information ebout the impact of significant events Jn a person’ life. We may evaluate whether changes occurred in the indivi uals adjustment following such entcal events as loss of a job or birth of a child. Knowledge about these events may lead to a better understanding of the psychodynamics of experience. For example, the fact that an early loss, like the death of a parent, is associated with depression in later life is ind cated by many cases (Jacobson, 1971), Second, as we understand the impact of such events more fully, we may be able 10 devise more appropriate treatment techniques, as well as preven: tive measures, The “talking cure” in psychotherapy, for instance, began a5 a result of treatment of hysterical symptoms in Anna O., one of Freud’ early cases (Breuer & Freud, 1957). The case ofa fearful boy named Peter (Jones, 1924) formed the basis of todays behavioral therapy techniques, Third, the case study ia perect forum or investigating unique cases or rae problems, Some variants of paraphilia (cross-dressing and sexual masochism, for Instance) are very rarely diagnosed in cincal stings, here, case studies can be useful soures of information, Fourth, case studies can provide evidence that cass doubt on theories or generally accepted practices. For example, the psychoanalytic notion that it Could be harmful to treat avert symptoms of dysfunction without treating their base causes was negated as counterinstances accumulated. Simply treating out ward symptoms with behavioral therapy techniques is often very elective, Filth, case studies are sometimes @ dramatic way to illustrate abstract con cepts. Seeing can be believing; hence, advertisers frequently use case studies (highly selective ones) to sell products. Because they are so compelling, how. lever, case studies can excite interest and bring about new research, Funhermore, the case study is used ir. clinical work to evaluate an individ uals overall level of psychological functioning. We compare our case against« TART ONE INTRODUCTION _— a BOX 3-1 Case Study: An Example of the Role of | ‘Traumatic Experiences in the Development of Obsessive-Compulsive Disorder Miss M, a 24-year-old single woman, was referred with a history of a | serious sexual assault whist on holiday abrosd. Immediately after the | traumatic event, she felt “quite diny” and spent a long time washing | herself and everything she had with ber at the time. Alter her return | home, she continued to feel dirty and said that she could not stop ot | resist the urge to wash repeatedly, She washed both her person and her clothes and other things inher flat she would spend hours doing this, She also suffered many symptoms of PTSD [Post Traumatic Stress Dis- | order), including flashback experiences, numbing, nightmares, poor | sleep, and hypervigilance. She had in fact had full-blown PTSD for some time alter the attack, for which she had received some professional coun- seling. By the time she was seen by us, he main complaint was OCD [Obsessive-Compulsive Disorder], end she had a clear diagnosis of the | disorder. She had obsessional thoughts about being dirty and unclean | Clam diny* “lam filtiy” “everyting is unclean, etc), which were linked to the washing compulsions, Miss M agreed that her washing was excessive and irrational, yet, despite her attempts to resis the compul- sive urges, she continued to engage in these rituals. It was this problem, that she specifically sought help fo:* ‘Rega om “The fle Tuma Eger he Gee of Cbste- Compe Dior Gey Si and MM, Bh eh and Thy 998,37, 981981. Cor cal standard of normal” behavior. Based on this comparison, we may suspect some form of psychopathology. We may then compare our ease against other cases to assess the degree of similarity or difference, This is the process underlying psychological diagnosis. The development of standard diag, noste citera for categories of mental cisorders appearing in the Diagnostic and Statistical Manual of Mental Disorders, 4th ed. (1994; abbreviated as DSM-IV} rellects groupings of many different patients’ case histories. Clinicians have noted similarities among patients that permit their problems to be classified into {goups. The records of persons previously diagnosed with “antisocial personal- ity disorder,” for example, have certain important similarities (see Box 3-2) (CHAPTER 3 ALTERNATIVES TO EXPERIMENTATION: NONEXPERIMENTAL DESIGNS ‘The deviant case analysis (Robinson, 1976) is an extension of the eval uative case study discussed earlier. Here cases of deviant and normal indi- viduals are compared for significant diferences. These differences may have Important implications for the etiology, or origin, of the psychopathology in question. Mednick and his colleagues have used this procedure to study the etiology of schizophrenia (Mednick, 1969; Mednick, Schulsinger, & Venables, 1981), They found, for example, that the autonomic systems of normal and schizophrenic children function differently Someday it may be possible to use this difference to predict which children will become schizophrenic, Clearly, the case study is a useful source of information. 1 is especially useful when we cannot experiment because of practical or ethical reasons (Obviously, we would not subject an individual 10 a stressful life experience, such as loss of a parent, simply to observe the outcome.) However, this approach has several limitations. Fitst, working with only one or perhaps a few subjects, we cannot be sure the people we are evaluating are representa tive of the general population; we would obtain a very distored picture of language development if we stodied one exceptional child. Second, if we are notable to observe an individual directly ll the time, we cannot be sure that, \we are aware of all the relevant aspects ofthat persont life, Third, subjects ot others providing data for case studies might neglect to mention important ‘information, either because they believe itis irelevant ot because they find it embarrassing ‘An even more severe problem is that case studies frequently rely on retrospective data, Retrospective data are data collected in the present that are based on recollections of past events. Information collected long after the fact is ap to be inaccurate for several reasons, People often cannot accurately remember all that happened at a particular point in time, We also know that Jhuman memories become altered or “reconstructed” over time by the cogni- tive system, Retrospective data can also be easily biased by the situation in which the data are collected. The mood of the data provider, for example, will alfect recollections, we tend to recall more postive events when we are in @ good mood and more negative events when we are in a bad mood.* 1m addition, aspects of the situation can trigger particular kinds of recol- lections. Something that seems as innocuous as a reproduction of da VinciS ‘Mona Lisa on a therapists wall might bring to mind past experiences with enigmatic, dark-haired. women that otherwise would not have been recalled aval. Even unintended hints from a researcher that cerain kinds of data are ‘more interesting than others can bias the kind of information that is brought to mind. For these reasons, reliance on retrospective data is a shortcoming, Records made at the time of an event ate always much prefered. The use of retrospective data is not limited to case studies but frequently occurs in this research method, Finally, because we have not ceated the antecedent condi tions in case studies, we cannot make cause and effect statements about thee Blox) 3. a) Antisocial Personality Disorder 1 is estimated that about 39% of adult American men and 1% of adult ‘American women could be diagnosed with antisocial personality disorder These individuals were formerly called “psychopaths” or “sociopaths” Diagnosis requires that the individual meet several cetera. The diagnos: tle cnteria were determined by the clinical judgments of experts in psj- chiatry and psychology and evolved from many, many case studies. The terms psychopath of sociopath bring convicted serial killers like Ted Bondy of Jelftey Dahmer to mind. However, psychologists Davison and "Neale (1986) described an important fact about people with an antisocial personality disorder—they can be found anywhere! "Business executives, ppoliclans, and physicians, plumbers, salespeople, carpenters, and bar tenuders—they are tobe found in all walks of life. Prostitute, pimps, con- fidence men, murderers, and drug dealers are by no means the only sociopaths? (p. 233) Some individuals with aspects of antisocial personality disorder are quite a bit more successful at managing not to break the law than oth: crs are_ What all these people have in common are personal histories that reflect similar kinds of behaviors and life events, Antisocial per. sonality always begins in childhood or adolescence with occasions of truaney, aggression, vandalism, stealing, lying, physical crue, oF other antisocial behaviors. Before the age of 18, euch individuals would be ‘iagnosed as having a “conduct disorder” Their behavior rarely goes unnoticed, they often find themselves in trouble with the authorities ‘Afier age 18, a diagnosis of antisocial personality disorder is made if individuals full specific behavioral criteria associated with the dis order. Individuals with the disorder carry the pattern of childhood anti- Social behavior into adulthood, where they are seen to engage in Fepested instances of behaviors that are assaultive, destructive, sre sponsible, of illegal. In addition, these individuals do not suffer from ((rue) remorse for their actions, From collections of case studies, several predisposing childhood factors have also been identified: inconsistent parental discipline, substance abuse, atention deficits, and yperactw ty, for example, Sociopathy is also more likely if one or both parents had the disorder DSM-IV is the standard for diagnosing mental disorders, but not all experts agiee with the diagnostic enteria. For example, many psychologists omtnued) believe chat sociopaths are also characterized by a great deal of charm that can make them exceptionally charismatic. Underneath itall, though, : thers Cleckley, 1976), Pare of the prob: Jem with compiling information fom case studies on individuals diag nosed with anuisocial personality disorder is that generally the individ ual availble for study are the extreme cases—typically incarcerated criminals or psychiatric patients, One can argue, as have Davison and Neale, that the database for sociopathy probably lacks valuable infor ‘mation about the most successful sociopaths ve, For example, we cannot say that an early loss causes later depression; we can merely say that there seems to be a relationship i just as plausible that some related factor, such as moving or changing schools because ofa parents death, might explain later depression. Field Studies Field studies are nonexperimental approaches used in the fleld or in ral fe settings. Researchers doing field studies ofter fe various types of date gathering to capitalize on the richness anc range of behavior fo ouside the laboratory. Antecedent conditions are not manipulated in fl studies, but the degree of constraint on responses varies considerably ‘om study to study. Depending on the measurenents used, field studies can fall anywhere along the continuum of low-low to low-high in our aphic scheme (see Figure next at some different f of field studies Naturalistic Observation Naturalistic observation is the technique of observing behaviors as they ‘occur spontaneously tn natural setings. Its a descriptive method: Like phe rnomenology and the case study method, naturalisie observation invalves no manipulation of antecedent conditions, Subjects’ wesponses are free to vary Because few constraints are imposed by the researcher (Sackett, 1978), nat ralistic observation would be considered low-low in Figure 3-1, The natural iste observation approach has been used most extensively in animal behav ior research (ethology), but itis often applied to human behavior as During naturalistic observation, observers attempt to remain unobtrusive (for example, behind « duck blind) so that the behaviors observed are not altered by the o presence. Ev pt is made to keep the seting as na ural as possible, so that the naturally occurring events will not be altered inPARTONE INTRODUCTION A he cating teatnet ie Je | cobras Teeameh te ere won oD, No HAVE Exiougt MATERIAL FoR MY EXPOSE OF PHKHOANALYSS, Copy ©1996 by Sy Hare any way: “The primary feature of such research is that human perceptual and judgmental abilities ae necessary to extract quantitative data [rom the low of responses" (Sackett, 1978, p. 2). This element creates special challenges for the field study In a typeal laboratory experiment, the researcher has only 2 small set of behaviors to record. In contrast, researchers who conduct field studies must contend with a vast array of responses, often including unantic- ‘pated, unconventional responses, Deciding who and when to observe and what to record and analyze draws heavily on both the researcher’ judgment and observational skills, The use of naturalistic observation is often essential and even preferable to experimentation. Psychological phenomena are rarely simple; a researcher can fail to take all the important antecedent conditions of @ behavior into account. Years of experiments on helpfulness are « casein point. As discussed in Chapter 1, so many factors influence our willingness to help others that it r (CHAPTER'S ALTERNATIVES TO EXPERIMENTATION: NONEXPERIMENTAL DESIGNS would be impossible to manipulate all of them in a single e could look a their combined ellects, The Geld researcher accepts the hallenge of dealing with data as they occur in real-life setings. Impos great deal of constraint on subjects’ responses is not always desirable, either Allowi sto behave as they usually would provides a much rich picture of the complex, multiple relationships among behaviors, and results com these studies may have more external validity, or generalizability, than periment so th In an example of naturalistic observation, Wheeler (1988) spent & yea observing Chinese residents of Hong Kong and noted many interesting di ferences between these Chinese and typical Americans. For example, he ved that during construction of the new headquarters of the Hong Kong and Shanghai Bank (one of the most ditional feng shut (wind and water) resive bualdings in the world), tra ert had to be consulted all along the 3¢ building was not constructe guidance of these expens in safely people would not have used the bank! Frequently, a naturalistic observation is designed to answer specif ns. Yepez (1994), for example, wanted to investigate whether teachers of English asa Seconcl Language coutses gave equal amounts of atention to their nd female students. With permission of the teachers (but without explaining her hypothesis beforehand), Yepes observed and recorded the num classroom interactions between four differen upil teacher was observed six times, Years of past rese der inequities are often found in the classtoom. Male students tend to be given more attention by teachers. Surprisingly, in Yepez’ observation, she found chat only one of the four teachers seemed to give more attention to the Yepez recorded behaviors in dl using a special coding system, called INTERSECT, developed for kinds of teacher-student interactions (Sadker & Sadker, 1982). I allowed her to collect data in a more objective and systematic manner. Many such coding systems have ‘ped for observational research in a wide variety of situations. These sys allow observations to be quantified, permit statistical analyses, and allow researchers to compare results across different studies that used the same co« * en she used INTERSECT, Yepee was engaging in the cechnique recorded using specific rules or guidelines, so that observations are ‘objective. For example, INTERSEC lines for Scoring four kinds of resp fare likely to use when interacting with students: pos ive reinforcement, acceptance, remediation, and criticism, Each type has | fully defined tna scoring manual, and many examples for coding behav are provided (Sadker & Sadker, 1982). Observers need to learn and prac tice the coding system thoroughly before they c 3 recording observations in whi66 PART ONE INTRODUCTION fa research setting, and the results should be reproducible by different trained ‘observers, In Yepezs study, for instance, she and an assistant coded practice responses until they consistently generated identical coding results (Occasionally, naturalistic observation is carried out in the laboratory. At times it is necessary to compare laboratory findings with behavior in natural seltings to confirm the uselulness of the laboratary setting for a particular research topic. Because some behaviors may be distoned by bringing them {nto the laboratory setting, such behaviors are best observed where they occur naturally. Naturalistic observation provides a wealth of descriptive informa- tion, but this research method does not lend itself to testing eausal antecedents of behavior, We would not know, for example, why the single teacher in Yepez’ study treated men somewhat diferenty than women. A further limitation of naturalistic observation is that we are dealing with specific samples of time that may or may not cortain the behaviors we want to observe, Ifwe bring our study into the laboratory, we may be abe to cte ate conditions to elicit the behavior of interest within a very circumscribed time and space. But when we do this, we must be aware of the possibilty that behaviors inthe laboratory setting might not be the same as they are in the real world. We might find that behaviors become very different when the subjects know they are being watched. This effect, the tendency of subjects to alter their behavior or responses when they are aware of the presence of an observer, is known a reactivity. We may even find that subjects try 10 guess the purpose of the study so that they can either confirm the researchers expectations or sabotage the results. (We will discuss these issues further in later chapters) Clearly, itis important for observers to remain as unobtrusive as possible in Geld research, Researchers can make themselves less obtrusive by hiding behind a duck blind, by observing from behind a one-way mirror, or by blending in with the Social surroundings so that they are not noticed. Many times, behavioral indicators can be observed without the subjects knowledge; such indicators ae called unobtrusive measures, The amount of wear and tear on your textbook, for example, could indicate how much time you spend studying, The number of candy bar wrappers in your garbage could indicate your penchant for sweets. Your physical distance from another pezson could Indicate how much attraction you feel toward that person. Frequently, researchers make inferences about behavior [rom observations of aspects of the environment. For example a researcher coulé study the trafic pattern in a supermarket by assessing the frequency of replacement of floor ules in each aisle (or even the wear and tear on les in different pars of the store). The subjects—shoppers—would never know their behaviors were being measured. Unobirusive measures are olten preferred over conspicuous measurement techniques (obtrusive measures) because unobsrusive measures overcome the potential for react. (Do you know someone who would skip certain aisles at the supermarket, ike the junk food or candy ase, if they knew a researcher ‘was watching?) [CHAPTER 3 ALTERNATIVES TO EXPERIMENTATION: NONEXPERIMENTAL DESIGNS Partcpant-Observer Studies One special Kind of eld observation isthe partcipantabserver study. Here the veseacher actully Becomes pat ofthe group being studied. Sometimes ths the only metho tht ean be wed to stugya greup--parculty i te group would at reasonably be expected to coopente volun with feeach investigation. Tplaly group member are no tld that they ae pa ofa study Clea once tel presences known, researchers tun the ik at Sule beat may change ely mbes pce In soe pots, cislowure might ead to being ousted (or perhaps worse) Partpan server studies usualy do ot ince systematic obseration or measurement techniques, Instead, the data gathered tend to be usliatve he researchers Impressions are merely described, as they were in Wheels sta, arpa cbservers are typi ying to gather a much ifomation and deal as they can. Where would th Kind of study fl in Figure 31? As you might imagine, some very uselul information can be gathered using this method. Recent parucipant-observer studies, have. centered around such topes as einic identity in an urban high school hoteles- ness, selactualzation in spttial communi, and urban crack adds Glaiy each of these sues would be dificult to study in more typical Way. But the partclpantobserver needs to remain aware ofthe song pos Sbity thatthe mere presence ofan observer can alr subjects behaviors in unk way Ard somes oer in he ses id ch cult to remain objective and unbised: Ofen, particulary i rendships form, itis hard to remain an objective sclents. ° a i fas probably secured to you that observational sues ocasion spell ethical eoncems, and patipant observer sis are prticualy problematic Therefore, age del thought neds to go into cburatonalearh belre itis conducted. Ae you invading the privy of others? I ebical not to tll people they are beng studied? Is eto pretend tobe areal group mes- Eel Is your observation going to male an iporant contibutos to poche legal Knowedge? Keepin mind the bial guidlines fom Chapter ll you were subject 8 your ow sti ow would you fel about 1? Wen comes 1 cbservatonal research, opinions from an inttationd review board can be especialy vale for weighing tes sues objectively Conducting a Field Study The wide range of techniques that may be used in field studies i illustrated by the work of Bechtol and Williams (1977). These researchers were inter ested in studying California liter. They noticed that untegulated coastine areas atract large numbers of beachgoers, even though the supervised state beaches are considerably cleaner. The researchers set out to determine who were the users of the unregulated beaches, whether there was a pattern to the litering that occurs on such 2 beach, and how users of the beach felt about sunbathing in the midst of debris. Bechtol and Willams employed several of the techniques common in feld stuies or« PARTONE INTRODUCTION FIGURE 3-3. In this suing, thoughts sbout litering are probably far eva. Bechtol and Wiliams spent two years observing activities on an unregu- lated beach in Southern California. They used naturalistic observation to determine who used tine beach: They simply watched and recorded what sorts fof people appeared. They saw that young people were the principal users. By using an unobtrusive measure to assess the pattem of litering—collecting and courting all the cans Tet on the beach—these researchers assessed behavior ‘without their subjects' knowledge. From the number of cans in the sand, they inferred that people liter the beach. There was no need to see anyone liter AAs we might expect, Bechtol and Williams found that littering was greatest luring the summer, when beach use was greatest In addition to observation, Bechtol and Wilkams approached people on the beach and asked there how they felt about the condition of the beach, At this point the researchers deviated from naturalistic observation. Instead of contin- ling to remain inconspicuous, the researchers interviewed people to get their views. Their findings sere intiguing: Ftst, users of the beach reporced being disturbed about its ltered condition. Second, all the people interviewed reported that they always took their own wash with them when they left— although, in rwo years of observation, the researchers never saw a single per- son do so. This study isan excellent example of why interviews and queston- naires (discussed latet) should be supplemented with objective oaservations, including unobtrusive measures, whenever possible ‘As with the other approaches we have discussed so far inthis chapter, the field study does not involve direct manipulation of conditions: Behaviors are (CHAPTER ALTERNATIVES TO EXPERIMENTATION: NONEXPERIMENTAL DESIGNS ‘observed and recorded a they occur in the natural seting, and subjects may be interviewed in the "wild," where the contaminating effects of a laboratory setting are absent, [tis a useful way of gathering many types of data, partic- ularly when the researcher is studying behavior like litering, which we might ‘not see in the laboratory Note that a eld study fs not in be confused with a field experiment. A field experiment isa true experiment (the antecedent con: ditions are manipulated) that is conducted outside the laboratory. A clever ‘example of afield experiment is described in Box 3-3. Survey Research ‘We are all familiar with surveys because we are exposed to them all the time: telephone surveys, election polls, television ratings, product surveys Survey research is a useful way of obtaining informaticn about peoples opin- ions, attitudes, preferences, and experiences simply by asking them, Surveys can allow us to gather data about experiences, feelings, thoughts, and motives that are hard to observe directly. Survey data'can be useful for making infer ences about behavior, although they do not allow us to test hypotheses about causal relationships directly. They are used in conjunction with many kinds of research designs in the field and in the laboratory (we will see many appl cations of surveys throughout the text), but they can be particularly impor- tant in field research Surveys can allow us to gather large amounts of data efficiently. Surveys, of course, ae low in the manipulation of antecedents (in Figure 3-1), but they are considered high in the imposition of units. In surveys, we limit subjects’ inputs to a very nanow range of responses, We only allow them to answer the ques tions we pose to them, and, very often, we allow only a small numberof response alternatives (eg. answering yes or no, dcling one number on a 4-point scale) Written questionnaires and interviews are the two most common Survey tech: niques in psychology research. Questionnaires can be handed out or sen through the mal; sometimes, surveys are conducted by compute: on laboratory PCs or via the Internet. Interviews can be face-to-face or over the telephone. The gen- cazablity of survey and interview reslis is determined in large part by the procedures we use to select our subjects, so we will go into detail about alter native stmpling procedures later in the chapter. Interviews and Questionnaires Interviews and questionnaires are important techniques for field research. The kinds of interviews and questionnaires used in field studies commonly include fan assortment of open-ended and closed questions, Closed questions take the form of “Do you smoke?" “Should there be a ban on nuclear energy?" and “On scale from 1 10 10, how much do you like Madonna?” Closed questions must bbe answered by one of a limited number of alternatives. Open-ended questions sollet information about opinions and leelings by asking the question in such 1 way thatthe person must respond with more than a yes, no, of 1-10 rating, [Examples of open questions are “Why do you prefer powdered detergents over Tiquids?” and "What made you decide to come for treatment now?0 rARTONE INTRODUCTION BOX 3-3 A Field Experiment in Chicago ‘Cunningham (1989) provides an example of afield experiment conducted Jn seven suburban Chicago-area bars. Cunningham trained several college ssudents to approach opposite-sex bar patrons at random, delivering one of several diferent kinds of converstion-sarters lines"), The positivity of teach patron response to the ine was surreptitiously measured, Cunning- hham discovered that women were much more sensitive to the kind of line fan opposite-sex person delivered than were men. Women sesponded more posilively to lines chat were either sel- disclosing or ordinary lke a simple 7H" than they did 10 a fippanly delivered line like, “You remind me of someone [used to date” or "Bet I can outdenk you.” Men, however, did ‘ot appear to care which kind of conversational gambit «’woman Used: they responded equally positively to all thee. It is sometimes possible 10 achieve high degrees of Both excemal and tnteral validity as Cunmngham did, by conducting actual experiments inthe field. We will return to field experiments in later chapters about experimental approaches. FIGURE 5-4 Could some uf dese luppy-lovking people acasly be ub jee in a eld experiment? Spencer imesh Boson (CHAPTER 3 ALTERNATIVES TO EXPERIMENTATION: NONEXPERIMENTAL DESIG By asking a combination of questions, the researcher can gather a great deal of useful information. Open-ended questions can often be used ¢o cla ify or expand upon answers to closed questions. Lets look at an example Ia questionnaire of interview designed to study attitudes of 11-yearolds toward cartoon violence, you might begin by asking a closed question: “On the average, how much time do you spend watching Satueday morning cat toons?” You could allow the children to respond to one of the following foptions: “Less than an hour’ “Between one and two hours" “Between (wo and four hours’ "More than four hours" This could be followed up with fone of more open questions: “Why do you lke to watch cartoons?” "What do you think about characters who hit each other?” "Wat kinds of things might cause you to hit someone?" However, unless you are simply going 10 describe people’ responses verbatim, you will need to quantify the answers Answers to closed questions are much simpler to quantify than open ques tions are. For example, you. could simply report the number (or percent) of children who gave each of the four possible responses to the closed question about cartoon viewing time. (Statistical analysis has further measurement and scaling requirements see Chapter 13.) To quantify answers to open questions, however, a system must be designed to evaluate and categorize the content of ‘each answer. This process, called content analysis, is similar to coding behav lors using systematic observational techniques. Ina content analysis, responses ae assigned to categories according to objective rules or guidelines, Suppose that the children’ responses to the open question, “What kinds of things might cause you to hit someone?” appeared to fall into six categories (@) Someone looked at me funny, (2) Someone said something to me that 1 didnt like, (3) Someone wouldnt give me what 1 wanted, (4) Someone took something avay from me, (5) Someone hit me first, and (6) Other responses. You could evaluste each response and code it into the proper category, then you could report the frequency of each kind of response. You can probably {hink of other categorization schemes that could be constructed for the chil- dens responses: low, moderate of strong provocation; physical or nonphys- ‘al provecatn; the umber of provocations mentioned By each ei and Constructing questions properly can take a great deal of skill. Here are some tps for getting started: Keep items simple, and keep people involved. Many problems come about because subjects da not understand the meaning of a cerin question, Maybe you used words they did not understand. Per haps your sentences were too complex and included several ideas at once Ambiguous or incomprehensible questions cannot generate useful data. Get subjects involved right away by asking interesting questions. If you are col lecting demographic information (peoples vital statistics), consider placing those questions later. Make sure your questions are not value laden. Do not word your ques- tions in ways that would make a positive (or negative) response seem embar rassing or undesirable, Consider how differen the following two questions sound, even though both were designed to measure attitudes toward abortion:n PART ONE INTRODUCTION Version 1: Do you believe doctors should be allowed to kill unborn babies during the firs tnmester of pregnancy? Version 2: Do you believe doctors should be allowed to terminate « preg- nancy during the first trimester? Clearly, a person unopposed to abortion would find it much more ficult co answer yes to the frst question simply because of the way itis worded. Be sure to keep the ethical guidelines in mind when you wate survey questions If you are using a written questionnaire, be sure the instructions are simple and clear. Ifyou are sending a questionnaire in the mail, be sure to include a polite and professional cover letter (and include a stamped, self-addressed envelope), Make sure your questionnaire and return proce: dures protect subjects’ anonymity. Unless you have contacted people in advance about your survey, and they have agreed to fil it out, do not expect that everyone will fill out and return your questionnaire. you have the resources, including a smal gfe can increase the return tate (Fowler, 1993) (One incentive technique that can be successful and cost effective isto hold 4 drawing for a prize (maybe a bookstore or video sore gift certificate or movie tickets). Always keep track of the number of people who do not return the questionnaires; you will need to report it. Consider a second mail ing to people who did not return the fist survey; this can add an additional 50% 10 the number of surveys returned from your frst malling (Suskie, 1992), Always keep mail surveys as short as possible and include a con vineing cover letter. Tf the nonretam rate is high, interpreting your results can be dificult inded. Suppose you designed a questionnaire to gather data about drug use and only 40% of the subjects retumed it Suppose, alo, that 99% of the returned ques tionnaires were from people who reported they had never smoked marijuana (one or two said they had stoked, but never inhaled). Would it be reasonable to conclude that 99% of people had never smoked pot? Probably not. On sen- ve issues, some subjects just wont answer, pariculaey if their answer would indicate they had engaged in socially undesirable, deviant, or illegal activites, Individuals will differ in chek willingness to ge socially undesiable espouses (Crowne & Marlowe, 1964). (We will tak more about this issue in Chapter 6) Any time that nonreturn rates are high, be aware that the people who returned the questionnaire may be different in some way ffom those who did not. You wil need to be extremely cautious here in drawing conclusions. Ifyou ate handing out your questionnaite tn person, consider the pos sibiliy of reactivity. I possible, let subjects fill out the questionnaire in pri- vate. Weigh the pros and cons of group sessions. Even though its lot eas- fer to collect data from many people at once, subjects may not take yout survey a seriously ina group setting, Also consider whether sensitive ques- tions would cause mote embarrassment in group sessions. And unless the group consists of complete strangers, you are likely to find that your sub- jects spend as much time talking to each other as they do filing out your questionnaire Tr CHAPTER ALTERNATIVES TO EXPERIMENTATION: NONEXPERIMENTAL DESIGNS Response Styles Sometimes subjects’ own personal characteristics can inadvertently alter the ‘way they answer questions, and this creates a special problem whenever we gather data through interviews, questionnaires, or other written tess to which the individuals can respond selectively. Response styles are tendencies to respond to questions or test stems in specific ways, regardless of the content (Cronbach, 1950; Rorer, 1965). For example, people difer in response styles such as willingness to answer, postion preferences, and yea-saying and nay- saying, These response styles need to be considered and, if possible, controlled for when you design questions First, be aware that people diller in their willingness to answer questions they are unsure about. Willingness to answer comes ino play whenever ques tions requir specific knowledge about faci or issues. When unsure, some peo- ple will leave questions blank, others will take a guess. An unwillingness 10 answer is often a problem in questionnaire or survey research. Subjects might commit answers to Key questions, making both scoring and interpretation dif cult. Some researchers attempt to control for this factor by explicitly telling subjects to guess if they are not sure of the answer to a question, Other researchers say nothing—and hope for the best. There is no hard-and-iat rule; teach researcher must think this through and decide how best to handle i. your questions are multiple choice, response styles can influence the selec ‘ion of answers, When in doubt about the right answer on a multiple-choice ‘exam, pethaps you alvays answer b. This isan example of a position prefer- fence.? Because of position preferences, sophistcated test builders vary the farrangement of correct answers throughout a test. Most questionnaires and inter views are not tests with right or wrong answers, but you need to keep position preference in mind anyway as you design your questions. In « multiple-choice survey gathering data about attitudes toward abortion, fr instance, donot always put “pro-choice® responses option b. A third kind of response style shows up in answers to statements like those in Table 3-1, Would you say that each is true or false for you? The items are similar to ones you might find on the Minnesota ulliphasie Personality Inventory-2 (MMPI-2), a test made up of a long series of items ofthis type. The way you answer the items on the test can tell a psychologist various things about you—for instance, whether you are anxious or depressed, At firs glance, the way people answer such questions seems straightforward; a per- Son who feels happy ought to answer “True” tothe ist item. We would expect subjects to respond to the manifest content ofthe questions, the plain mean- ing of the words that actually appear on the page. When we give a question naire or other paper-and-pencil test, we are usually interested in the manifest > Een ts lamin tun ms town show potion pense or empl od ed shee eqs fr i we coe sole wh «pce urn wl es ‘ranage Rae onl rs ley varying the postin le cots) atthe a. ‘mast someting Fel sometmes np 2” PARTONE INTRODUCTION TABLE 3-1 Possible Items on a Personals Sale 1. eel sppy most of the time 2 Lenjoy beng with other people 3 Like paying acentin to deals 4+ When I can, 1 avoid noisy places 5. Sometimes I el frightened for no appaert reason content ofthe items. "Have you ever visited another county?” means just that; the manifest content ofthe item is simply foreign travel. Most people would answer based on thei actual travel histories, However, researchers have noticed that some interesting things can happen when subjects fill out questionnaires—especally ‘when the questionnaires ask about feelings or aides. Some subjects seem 10 respond to questions in a consistent way: They wall answer "Yes" or "Tue" to most items, or they wll sy "No" or False” to mos items. Some subjects are yea-sayers others are nay-sayers, Yearsayers are apt to agee with a question regardless of is manifest content, Nay-sayers tend to disagree no mater what they are asked Clearly, a yea-saying or nay-saying response style can pose a question about validity: Are we relly measuring what we set out to measure? How can Wwe avoid this problem? One way is by designing questions that force the sub- ject to think mote about the answer (Warwick & Lininger, 1975). For instance, think about the diference between these two items! Do you agree or disagree that she cast af ving has gone up tn the Iss year? In your opinion have prices gone up, gone down, or sayed about the same the past year, of dont you know? When we phrase questions to have simple yesino or agree/dsagree answers, we make it easy for subjects to respond based on response style. By building some specific content into the options, we encourage subjects to give more thought to each choice. If we must use yesino questions, we can still take some precautions. Table 3-2 shows two versions of the Unfounded Optimism Inventory. The ‘optimistic” choice ls underlined for each item. All the stems in version A ace written so that the optimistic response is a yes response. Subjects who are yea-sayers would score high on unfounded optimism:—even if they happen to be somewhat pessimistic. Now look at version B, To get a high unfounded optimism score, a subject would have to give both yes and no answers, Using version B would give more valid answers, subjects who tumed up high on unfounded optimism are probably not pessimistic yea-sayers, you decide o ask questions in face-to-face interviews, there are two more Important things to keep in mind: establishing rapport and maintaining inter viewer consistency. The interviewer’ appearance and bearing can aflect the ways subjects will spond. The more sensitive the questions, the more impor tant the interviewers demeanor becomes. To get honest answers to sensitive (CHAPTER 3 ALTERNATIVES TO EXPERIMENTATION: NONEXPERIMENTAL DESIGNS TABLE 32 The Unfounded Optimism Inventory 1 U know cat everyiing 1. Tow that everthing wl be allright | ues NO willbe all ht ve 2. ean ple he fastest line 2.1 aways sand inthe atthe bank Yes NO lowest in atthe Bank YES 3. often smile at nothing. Xes_-NO 3. I rarely smile, even when provoked ves 4 11a money, expect 4. If lose mone, expect ito be returned, Yes NO ito be retned us questions, the interviewer will need to spend time winning the subjects con- fidence and establishing trust, Remember that even subtle changes in the inter viewers behavior or tone of voice can influence subjects responses; your ques- tions must be asked the same way each time. To achieve the necessary consistency requires a lot of practice. You will also need to decide whether your interview will be structured, unstructured, ora litle of both. In a struc~ tured interview, the sare questions are asked in precisely the same way each time—deviations from this structure are not permitted Structured interviews provide more usable, quantifiable data. Unstructured interviews are more free Flowing: the interviewer is free to explore interesting issues as they come up, Dut the information may not be usable for a conten: analysis oF statistics Sampling Regardless of how you conduct your survey, one of the most critical isues is sampling, Selecting subjects is an important part of eny research regardless fof ts design, and it isa particularly critical issue in survey research, Fits, the researcher must decide who or what the subjects will be. (See Box 3-4 for an interesting approach.) Ideally, when we conduct research, we would include all the members of the population we wish to study. The population consists ofall people, animals, or objects that have at least one characteristic in com: ‘mon—for example, ali undergraduates form one population; all nursing home residents form another; all jelly beans form sill another. Clearly, itis almost never possible to study the entite population; instead we rely on samples ‘A sample of subjects is a group that is a subset ofthe population of inter- est. Data collected from samples can be used to draw inferences about @ popu lation without examining all its members. In this way, pollsters like Gallup are able to make predictions about the outcome of imporant elections. Diflerent samples may produce very diferent data. How accurately we can generalize our findings from a given sample o a population depends on its representativeness, cor how closely the sample mirrors the larger population—more precisely, how closely the sample responses we observe and measute reflect those we Would ‘Version A: No Contol for Response Siyle Version B: Centoling for Response Style Nobi Archival Studies Many opportunities for feld research can be found in survey data that hhave been collected for other purposes and stored in data archives. A wealth of statistical data is collected by government and private agencies hospitals, businesses, schools, and so on. Information about such things as crime and death rates, education levels, salaries, housing patterns, and isease rates are accessible to researchers ‘There are also archives housing ata from scientific research some using large probability sar ples. In addition t0 demographic information, some of these surveys Include information colected about people’ attitudes, This information can be used to analyze societal trends orto gather information about pop: ulation subgroups. A universiy bbraran is 2 good place to begin if you ‘use archival data, There may be a charge (0 ‘on popular culture a lw cost by using other ki is of existing materals rnewspapers, and magazines, for istance) * al data, Bowman (1992) studied some nan interesting use of ach controversial issues about Ainc discouragement over barriers to € problems with family rles—fom a positive respective. His study looked at factors related 10 fac adaptive coping mechanisms in a sample of 372 African- ‘American men who were responsible husbands and fathers. To gather data, he accessed a national data bank archived at the University of Michigan. Data were originally collected in 1979-1980 by face-to-face interviews with 2107 Afican Americans residing indifferent pars ofthe United States. Bowman found that kinship bonds and religious belils were stronger in men who were happy in their family roles, gested that family closeness and religion might help to met role difficulties, Because this was a correlational study (discussed in ‘Chapter 4), causal inferences cannot be justified However, valuable infor ‘mation for further study on these important issues was gained (nithout the great expense of fring the Kind of subjects he wanted and travel ing all across the county to interview them) by accessing and reanalyz- sng information from enisting data archives n a creative new wa} Gn out ot stot ached ech sed, M, Suh ER, and ade H. 98D) (CHAPTER ALTERNATIVES TO EXPERIMENTATION: NONEXPERIMENTAL DESIGNS ‘obtain if we could sample the entice population. We would be very wrong about the percentage of votes going to a candidate if we based our predictions on a preclction sample that included only Democrats! There are two general sam- pling approaches: probability sampling and nanprobablity sampling. Probability Sampling Probability refers to the study ofthe likelihood of events. What are your chances of rolling a 7 with a pair of dice? What are your chances of winning the lot tery today? What are your chances of being selected for a survey sample? Prob- ablity is @ quantitative discipline, to study probabilities, we must be able to count events and possible outcomes. From a theoretical standpoint, some form of probability sampling isthe preerred means of selecting subjects for research Probability sampling involves selecting subjects in such a way that the odds oftheir being in the study ate known of can be calculated. We begin by defin ing the population we want to study. For example, our target population might bbe all women born in 1075 and now living in Seattle, Washington. (It would take time, but we could count them and establish each womans odds of being in the sample) A second condition for probability sampling is that the researcher must use an unbiased method for selecting subjects, such as fipping 4 coin, drawing a number out of @ hat, or using a table of tandom numbers (see Box 3-5). This process is called random selection, meaning that any mem ber of the population has an equal opportunity to be selected, and the outcome of the sampling procedure cannot be predicted ahead of time by any known law (Kerlinger, 1973). Random selection is also a common assumption of the statistical tests used most often to analyze data. Now let us look at three types of probability samples—simple random sample, stratified random sample, and cluster sample Simple Random Sampling The most basic form of probability sampling is the simple random sample, whereby a portion of the whole population is selected in an unbiased way. Hull (1954) described the basic procedure in these colorful terms: 1 you have a barel of beans, some red snd some whit, there i only one way to find out exactly how many ofeach color you have: Count em. However, you fan find out approximately how many are red in much easier fashion by pling fut a handfol ef beans and counting just those, figuring that the proportion wil be the sume all trough the bare (p13) ‘Through random sampling we can find out about what the population is like without studying everyone, To obtain a simple random sample, all mem- bers of the population being studied must have an equal chance of being selected. If there are 1 mallion people in the population, the chance of any particular person being selected should equal one in a million. Even then, Using random selection procedures does not guarantee that our sample will bbe truly representative of the whole population. Suppose someone put all ”78 PARTONE INTRODUCTION ] CHAPTER ALTERNATIVES TO EXPERIMENTATION: NONEXPERIMENTAL DESIGNS 19° -¢ ruc) E How do we use the random number table? We begin by asaigning code pox 3-5 ‘numbers to all members of our subject pool. Al the nusing home, ve might ber the subjects in the order they appear on an alphabetical ist are 32 people avaiable, we assign them numbers 1 through 32. ‘we heed only 20 subjects, we go through the random murnber table in an ‘orderly manner (such as by reading vercally dowm each successive column Using the Random Number Table for Random Selection fof numbers) to find the firs 20 numbers berween 1 and 32 that appear in Suppose we want 0 collect data on beliefs about the reationship between the ble. Look at Appendix B, Table Bl. Beginning «: the top of the fis regular exercise and “qualiy of life’ among older persons. The population column and reading down, we find the numbers 03, 16, and 12. All are ve sted in studying is people over age 75, Clearly, we cannot between 1 and 32. So subjects 3, 16, and 12 will be m our sample. (The study the entire population; what We want is a random, representative sam- nd 95 also appear in the fis column, ple, In practice, w> often find that this goal cannot be achieved. We may, them.) We continue going thr jor example, only have acess to the residents of one nursing home. Thus eally unt we have a tual of 20 subjects. Ths group o the “population” avilable for study as already a select goup, although we ‘le ofthe 32 people available for the experiment. The sarting point foreach | would ulumately lke to male statements about the general population of study should be decided in a blind manner. For in older people, out sample must be taken from this smaller group, your eyes and put your finger on a part of the table. You would begin Suppose there are 32 5, but we only want to Selecion there interview 20 subjects. Assuring that all the residents iterviewed, how do we decide which 20 will take par ask everyone over 75 to report to the interview scom and allow the 20 arrivals tc be in the study, But some residents may arrive later the white beans into Hulls barel fist. I he took bis sample than do others for a variety of reasons. Those who are new to the home he might conclude incorrectly that the barrel contained only red beans, ight not know their way around the grounds and might take moze Random sampling yields only an estimate of what is likely to be tue. time to find the correct. health could be a factor. Indi nat feel well might not want to walk to the inter Stratified Random Sampling here could be significant personality differe When the population is known to contain distinct subgroups, researc up early and those who sleep late, Thus, a sam- often prefer another variation of probability sampling known ss stratified ple of subjects based on arrival time might not be representa random sampling. A stratified random sample is obtained by randomly uals wha de group as a whole; it would be biased to include a cisp-oportionate ss from people in each subgroup in the same proportions as they ber of healthy eatly risers who Know thelr way around the bullaing exist in the population. Here is an example. A particular factory is made We can get a farly good random sample of the nursing home res: up of 10% managers and 90% production workers. If you wanted to mea dents if we vite all the prospective subjects’ names on small pieces of sre employee morale in this plan, it would be desirable to use stratified paper, put them into a hat, mix well, and draw them out one by one ut random sampling. Managers and production workers might not have the we have as many 1s we need, The hat method is usualy adequate, but it sme feelings about the company. If you wanted a stratified sample of 100 's not foolproof 4 small variation in the ize of the papers may bias the ployers, you could randomly select 10 managers (10%) and 9 ion. The pafers might not be mixed enough, $9 you draw out only production-line workers (90%). Then your sample would reflect the s beginning with the lets M through Z, or only the lst numbers, makeup of the entie stall. A simple random sample could result in a sam. you wrote. A beer ‘might be to use the random mumber table (Gee Appendix 8), a uble of numbers generated by a computer so that very number to be used has an equal chance of being selected for each uded 0 tion-line workers. By using stratified random jement, you ensure that their views ate repre position inthe tabe. Unlike the hat method, the computer generated t het conducting a survey of urban attitudes might use stratified stoully unbiased. random sampling to mirror ethnic differences. A market researcher might use Gentine cto represent various age and income groups more accurately: There are 1¥0 4 advantages to stratified sampling: Fist, the subsets are sampled separately $080 PARTONE INTRODUCTION that important minorities or subgroups are represented in the total. Second, diffrent procedures can be used with different subgroups to maximize the usefulness of the data. This strategy is desirable when the various subgroups are likely to respond aillerenty? Cluster Sampling ‘When the population of interest i very large, itis often 00 costly oF imprac tical to randomly select subjects one by one. In such cases researchers may use another fo-m of probability sampling called cluster sampling. Instead cof sampling individuals from the whole population or from subgroups, researchers sample entire clusters, or naturally occurring groups, that exist within the population, As with all forms of probabiliy sampling, partici- pants are randomly selected, but whole groups of people are selected rather than individua's. Suppose you wanted to survey atitudes about the educa tion system in Connecticut. It would be very expensive and time consum- ing to randomly sample all 3.4 million people. Instead you could randomly. select clusters that already exist, such as zip code areas, school districts, cities, or counties. If you Selected six counties at random, then everyone in those six coumies would be surveyed The main advantage of cluster sampling is that the researcher can. sample data elliciently from relatively few locations. A potential disad! vantage of this approach is that subjects within clusters may resemble one another. The people in one county, for instance, might be quite similar in economic status, education, ethnicity, and even age. For that reason, it is desirable to sample many clusters and to obtain 2s large a sample as possible ‘Some form of probability sampling Is generally preferred for research pur poses because it increases the external validity of the study. Probability sam ples are more representative of the population, so research conclusions have Breater generalizability. There are, however, other ways of selecting samples, ‘hich fall under the heading of nonprobablity samples Nonprobability Sampling It is sometimes impossible to use the kinds of procedures we have just described, despite their advantages. Even though random selection of ub- Jects is accepted as a comerstone of good research, many surveys and other kinds of studies are based on nonprobability samples. As the name implies, in nonprobability sampling the subjects are not chosen at random. Let us look at we common examples of nonprobability samples—quota and. te se raig (CHAPTER ALTERNATIVES TO EXPERIMENTATION: NONEXPERIMENTAL DESIGNS Quota Sampling Tn quota sampling, researchers select samples through predetermined quots that are intended to rellect the makeup ofthe population. Samples can reflect the pro- portions of imporant subgroups, but the individuals are not selected at random. For example, a newspaper might want 10 document campus atitudes toward nuclear arms, A reporter is sen to & university with instractions to interview 40 ‘students, half male, half female, because the student body 1s roughly balf men, hall women, There ae no constainis on how the reporter selects people to inter view as long asthe quota is filed. Because the selection isnot random, the result {ng sample might of might not be a good representation ofthe college corammu- nity The teporer, for example, could arrive early in the moming and approach students gathered around the vending machines in the student union building ‘There are many sources of potential bias in this approzch—students who have late classes are missing: those who have money for the vending machines might, bbe overepresented; the reporter might want to approach only those who appear most friendly and cooperative. Such samples have human interest value and are sometimes used in public opinion polls, However, quota sampling lacks the igor required in scientific research and is low in exernal validity Convenience Sampling Convenience sampling is obeained by using any groups who happen to be nilable—for example, a church choir, a psychology class, a bowling league, fora supermarket checkout line. This i considered a weak form of sampling because the researcher exercises no control aver the representtiveness of the sample. Despite this drawback, convenience sampling (also called accidental sampling) is probably done more often than any other kind, It is convenient, i is certainly less expensive than sampling the whole population t random, and itis usualy faster However, researchers who rely on convenience samples must bbe cautious in the conclusions they draw from cheir data. Convenience samples agrealy limit any studys external validity: Whatever qualities distinguish choir members, classmates, or bowlers from the rest ofthe population can also lead to atypical research findings. It might not be valid to generalize observations beyond the group studied. We have our best chance of obtaining a sample that is representative of the whole population through random selection ring Samples THE apts snipes chosn influences what ean be conclude from the rel, The resnch report mst expan the ype of sample used and bow subjects were reroted so thatthe rau can be interpreted propery After alin many reumstanes college stidents might espond ferent than do fusing home residents, and results obsined from ene convenience sample inght hot be geeraized to everyone. So we need io tll the reader ea Tow the sample was obined. This includes denying the spect popula: tion sampled (or example cllege siden), at well as ging an exact description of where and how subjets wee btained. For example you could Sey, the subjects were 60 undergraduates, 30 men and 30 women, a the University of Oregon who responded to an ad inthe college newspaper”PART ONE. INTRODUCTION that important minorities or subgroups are represented in the total. Second, diferent procedures can be used with diflerent subgroups 1o maximize the usefulness of the data, This strategy is desirable when the varlous subgroups sxe likely to respond diferently? Cluster Sampling ‘When the population of interest is very lage, itis often too costly of imprac: tical to randomly select subjects one by one. In such cases researchers may use another form of probability sampling called cluster sampling. Instead of sampling individvals from the whole population or from subgroups, researchers sample entire clusters, or naturally occurring groups, that exist within the population. As with all forms of probability sampling, partici pans are randomly selected, but whole groups of people ae selected rather than individuals, Suppose you wanted to survey attitudes about the educa tion system in Connecticut. It would be very expensive and time consum: ing to randomly sample all 3.4 million people. Instead you could random select clusters that already exist, such as zip code areas, school districts, cities, oF counties. If you selected six counties at random, then everyone in those six counties weuld be surveyed, The main advan-age of cluster sampling is that the researcher can sample data efficiently from relatively few locations. A potential dise: vantage of this approach is that subjects within clusters may resemble one another. The people in one county, for instance, might be quite similar in economic status, education, ethnicity, and even age. For that reason, it is desirable to sample many clusters and to obiain as large a sample as possible Some form of probability sampling is generally preferred for research pur. poses because it increases the external validity of the study. Probability sam: ples are more representative of the population, so research conclusions have sreater generalizability, There are, however, other ways of selecting samples, ‘hich fall under the heading of nonprobability samples. Nonprobability Sampling Ik is sometimes impossible to use the kinds of procedures we have just described, despite their advantages. Even though tandom selection of sub- jects is accepted as a cornerstone of good research, many surveys and other kkinds of studies are based on nonprobability samples. AS the name implies, {in nonprobability sampling the subjects are not chosen at random. Let us look at two common examples of nonprobability samples—quota and (CHAPTER 3 ALTERNATIVES 10 EXPERIMENTATION: NONEXPERIMENTAL DESIGNS Quota Sampling In quota sampling, researchers selec samples through predetermined quowas that sae intended to reflect the makeup ofthe population. Samples can reflect the pro: portions of imporant subgroups, but the individuals are not selected at rancor, For example, a newspaper might want to document campus attitudes toward nuclear arms’ A reporter is sent to a university with instructions to interview 40 students, half male, half female, because the student body is roughly half men, half women, There are no constiints on how the reporter selects people to inter view as long a the quotas filled. Because the selection isnot random, the result ing sample might of might not be a good representation of the college commu nity The reporter, for example, could arrive early in the morning and approach students gathered around the vending machines in the student union building There are many sources of potential ias in this approach—students who have late clases are missing, those who have money forthe vending machines might be overrepresented; the reporter might want to appreach only those who appear ‘most fendly and cooperative. Such samples have human interest value and are sometimes used in public opinion polls. However, quota sampling lack the rigor required in scientific research and is low in exernal validity Convenience Sampling Convenience sampling is obtained by using any groups who happen to be available—for example, a church choir, a psychology clas, a bowling league, fr a supermarket checkout line. This is considered a weak form of sampling because the researcher exercises no control aver the represenativeness of the sample, Despite this drawback, convenience sampling (also called accidental sampling) is probably done more often than any other kind. It convenient, it is cenanly less expensive than sampling the whole population at random, and itis usually faster However, researchers who tely on convenience samples must, be cautious in the conclusions they draw from their data. Convenience samples greatly limit any stadys external validity. Whatever qualities distinguish choir rembers, classmates, or bowlers from the rest ofthe popilation can ala lea to atypical research findings. It might not be valid to generalize observations beyond the group studied, We have our best chance of obtaining a sample that is Tepresentative of the whole population through random selection, Reporting Samples {The say's sample is chosen influences what can be concluded from the rests The research repon must explain the ype of sample used and how ‘ubjets were recruited co thatthe reals can be interpreted propery. After alin many cumstances colege students might respond diferenty than do nsing home resident, and results obsined rom one convenience sample might not be generalized to everyone. So we need to tel the reader exaly Row the sample was obtained, This lncludes ening the specie popula tion sampled (or example college student) as well as giving am exact description of where and how subjere were cnained” For example, you could sy, “The sobjecte were 60 undergraduates, 30 ten and 30 women, atthe Uhlvesty of Oregon who responded to an ad inthe college newspaper”a PART ONE INTRODUCTON Include gender, age, end other important demographic characteristics (edu- cation level, racial and ethnic heritage, socioeconomic status, etc.) if they could be important in interpreting the results, ‘Any details chat might have influenced the type of subject participating in the study must be incled, If subjects were paid for participating or students participation fulfilled 2 course requirement, readers should be told that, coo. ‘Any limitations on who could patticipate should also be noted. In the nurs ing home example from Box 3-5, we might say ‘The participants were 20 (6 male and 14 femal) randomly selected residents of the Valley Nursing Home in Elmown, Ohio. To participate, subjects needed to be at leat 75 yeas old and in good health. The average age wae 77.3 years the ages ranged fom 75.2 years to 834 yeas Note that this statement immediately tells readers that, although we are di cussing atitudes about the values of exercise in healthy older persons, our par- ticipants were drawn at random from a very small pool. It also tells readers that we saudied both men and women who fell within a particular age range Occasionally, some panicipants who are selected are not included in the report: They dropped out, their data were discarded because they could not fo: low the instructions, and so on, These facts should also be reported. This may seem like a lot of information to report, but details about your sample are & eritical component of all research, and they are of major importance in survey research. They give the reader the information necessary to evaluate the gener- alizailty ofthe results orto compate the results of different studies. The details are also necessary for replication. Inthe next chapter, we wil learn about other nonexperimental designs—correlational and quasi-experimental designs—that, provide additional techniques for describing relationships among behaviors and for investigating individual diferences in attitudes, behavior, and personaly. Summary Research activities may be described slong two dimensions: degree of manip lation of antecedent conditions and degree of constraint on subjecs' responses, Laboratory experiments tend to lal at the high end of both dimensions, epre: senting only 2 small portion of the possible research options, Nonexperzmental Aesigns tend to be low on the frst dimension but vary from low to high on the second. This chapter covers four major nonexperimental approaches to data col- lection: phenomenology, case studies, field studies, and survey research, These approaches may be higher in external validity than aze laboratory experiments but they are probably lower in internal validity Phenomenology is the description of one’ own immediate experience Rather than looking out at behaviors in the world, the phenomenological approach requires us ‘0 begin with our own experience as a source of data, Phenomenological data are limited in three major respects: Because we da not compare subjects under different conditions, we cannot make cause and effect (CHAPTER ALTERNATIVES TO EXPERIMENTATION: NONEXPERIMENTAL DESIGNS statement about our experince. We have no way of knowing whether tend ing to our experience alters i t may not be completely accurate or objective Our experiences might or might not be generalizable fo others. “The ese study is used to study indvadale 1a descipve record made by an outside observer ofan Indvidals expences, behavior or both. Cae stud ies area sei source of ews, they can led to therapy techniques, they alow the Study of rae phenomena, they demonstate counternstances they can lead to more sesttch, The ecard may be mae sysemately ove a prod of neo afer the fact using rebpective data, a is len the casein cic practice. This approach tables sto study a variety of He events we would not say experimentally Fld studies are stocies done i reat ie settings. These studies allow us to explore behavior that we probably would not seein the laboratory. But these studies cannot be wed to make inrences about ease and elect eae tionships Field studies include a variety f techniques forcllecting data, Nat tral ebservation, the techni of observing events as they oer in thee attra settings, i a common component of field research, Duting natural tie observation, the observer remains anobrasive so thatthe Behaviors being hserved are not lieved by the presence ofan intruder. This approach reduces subject reactvly and allows study of behaviors tht would be dstoned or absent in the lboratory When the researcher wants to answer specific que tos aig natal ebseratons, ea devon hn be teed. In another feld technique, the papan-obsener study, the researcher hay inert wih subjects ae well ax observe them to obtain information: this tmethod i infequently used Surge (questionnaires and interviews) ae frequently used in fla stud. tes and can be part of many research desgns. When designing questions, the researcher needs fo consider response sie such a wiles (0 arse post ton reference, and yeasayng or nay sang Samping procedures ste por tant forall types of research snd are particular important for survey research, Oui abt to generalize research findings depends onthe epresetatveess of out sample of subjects, Some form of probably sampling (ample random sa Piling sttaifed random sampling, or ster sampling) is prefered, but nomprab- tdi samples, such as covenenc samples, are more frequently used ‘Rey Terms (Case study The descriptive record ofan individual experences, behaviors, oF both kept by an outside observer ‘Cluster sampling A form of probability simpling in which a researcher samples ‘ete clusters, or naturally occurring groups, tat exist within the population. ‘Content analysis” A system for quantifying responses to openended questions by categorizing them according to objective rules or guidelines. ‘Convenience sampling A convenience sarple is abtained by using any groups who Thappen to be corivenient, considered « weak form of stmpling because the researcher exercises no conttol over the repreentaiveness of the sample Called acldetal samping). ssPARTONE INTRODUCTION Deviant case analysis A frm of case study in whit deviant individuals ae com pared wih those who are not toate the significant varations between the External validity How well the ndings ofan ex iment generalize or apply to peo le and setings tha Field stady “A nonexpenimental research method used in the field or in a rea-lie ting. iypclly employing a vaney of techniques including natalie obs vation and unabirusive measures of survey tools, sch as questionnaires and Internal validity: The cerany thatthe changes i behavior observed across treatment ‘conditions inthe experiment were actually caused bythe independent variable Manifest content The plain meaning ofthe words or questions that actully appea ‘on the page Naturalistic observation A descriptive, nonexperimental method of observing behaviors as they occur spontaneous in natural settings. Nay-sayers People who ae apt to disagree witha question regardless ofits mane Nonprobabity sampling Semping procedures which subse are nt chosen rom; two common examples ae ques and convenience sample Parcipancobserverstdy A special Kind of fldobservaon i wh ‘tualy becomes pr Of the goup being sted Phenomenology A nonexpermenal uth of gathering Population All peo Position preference When in doubt sbout answers to muliplechoie que Probability sampling Selesing srtples in sack sway thatthe cas of any subject Quota sampling, Sekcting spl trough pre sniff he pope population sbgroup, bu he pare Random number table A ible of by atending to and are intended to te proportions of import Random selection An unbiased method for selecting subjects in such a way that each member ofthe popalation hasan equal opportunity be selected, and the futcome cannot be predicted ahead of time by any kav la Reactivity. The tendency of subjecs to ater their behavior or responses when they ‘av aware ofthe presence ofa observer Representativeness The extent to which the sample responses sue reflec those we would obtain if we could sample t entire population Response style Tendency for subject to respond to questions of test ens ita spe way, tegardlss ofthe content Retrospective data Data collected in the present based on recollections of past evens, ap to be inaccurate because of flty memory, bias, mood, and stato Sample of subjects selected subot of the populston of interes Sampling Deciding who or what the Simple random sampling The mo= portion of the whole poplati 3 selecting them, probably sampling whereby a Stratified random sampling A form of probability sample obtained by randomly sampling fem people i each Important popslation subgroup inthe same pro portion as they eatin the population, r (CHAPTER 3 ALTERNATIVES TO EXPERIMENTATION: NONEXPERIMENTAL DESIGNS Survey research A useful way of obtsning dts sbout people’ opinions atitudes, Systematic observation A syse Unobirusive measute A pr Willingness to answer The di Yemsayers People Who ate ap and experiences that are sbained using questonnai (9 observe directly; data may be recorded using specie rules o iowledge used to obtain more abe ces among people in ther syle of responding to questions t ‘people wil leave these questions blank, gree with 2 question regardless ofits man fest content Review and Study Questions nonexperimental approaches and give an example of how eternal vaiiy? Why are nonexperimental studies ote igher in exernal valcty than true experine ‘What is mean by invrnal validity? Why ate nonexperimental What is resospective dita? Why {6 the use of retospective data considered a shortcoming i scenic research? What ate unoberusive messures? Devise an unobtrusive measure co establish each ofthe following a Which profesor atthe univesty tthe most popu? 1, What ate the most popular library books? ©. Do people prefer tost on the let ot the right side ofthe theater when they a TE peope find addresed lees with stamps on them, wil they mal ther? Describe the difference between opencended and closed questions. How ate ded questions quantified? ee response styles and give an example ofeach, Explain how you would miroling for them when designing questions. Explain the ference between probeiliy sampling and nonprobsbiliy sampling and describe the diferent forms of each For each ofthe research topics listed hee, indicate the ype of nonexperimental approach that would be moet useful and explain why” (You may find more than fone approach potentially sell for some topics) a Pushing abead inline ®. Daydresming © Locating the most popular painsing in an art galery . Sibling nvalry 8