Vous êtes sur la page 1sur 14

See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/263573883

Predicting Consumer Behavior and Media


Preferences: The Comparative Validity of
Personality Traits and...

Article in Psychology and Marketing November 2013


DOI: 10.1002/mar.20657

CITATIONS READS

4 303

3 authors, including:

Samuel D Gosling
University of Texas at Austin
151 PUBLICATIONS 15,821 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Great Ape Personality and Behaviour View project

All content following this page was uploaded by Samuel D Gosling on 11 March 2015.

The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original document
and are linked to publications on ResearchGate, letting you access and read them immediately.
Predicting Consumer Behavior and
Media Preferences: The Comparative
Validity of Personality Traits and
Demographic Variables
Carson J. Sandy and Samuel D. Gosling
The University of Texas

John Durant
Mindset Media

ABSTRACT

It is common practice for organizations selling a product to divide potential consumers into
segments to allow them to target those most likely to buy their products. Two broad approaches to
market segmentation can be delineated. The most common approach relies on segmenting by
demographic variables (e.g., age, gender). The second approach (known as psychographics)
identifies market divisions in terms of psychological variables such as values, attitudes, and
personality traits. There has been little research comparing the efficacy of the two approaches. Based
on analyses of over 45,000 participants, the present research empirically compares the effectiveness
of the two approaches among segmentation variables ranging from cell phones and lottery tickets to
newspapers and television shows. Overall, both approaches explained surprisingly small amounts of
variance in consumer behavior. Nonetheless, for the variance that was predictive, the relative
contribution of demographics and psychographics varied dramatically across consumer behaviors;
for some behaviors (e.g., electronic purchases), demographics had superior predictive potential but
for others (e.g., television shows) psychographics were more useful. Therefore, an approach that
integrates both methods is recommended.  C 2013 Wiley Periodicals, Inc.

It is common practice for organizations that have a Market Segmentation and Psychographics
product to sell to divide their potential consumers
into segments. Doing so allows the organizations to The early 1930s saw a revitalization of economic
target their marketing efforts, directing advertising theory in which it became apparent that components
to the consumers most likely to buy their product of the supply side (i.e., production) and the demand
and crafting their pitches to appeal to the intended side (i.e., consumption) of manufacturing were no
audience (Vyncke, 2002). Two broad approaches to longer homogenous (Smith, 1956). Marketers began to
market segmentation can be delineated. The first ap- recognize that people were growing in their desire for
proach, and by far the largest, relies on segmenting the variety. Customers needs and desires were growing
market in terms of sociodemographic variables such more complex. These factors, as well as a number of
as age, gender, race, and class. The second approach other practical issues, drove manufacturers and busi-
sometimes referred to as psychographicsidentifies nesses to segment their large heterogeneous market
market divisions in terms of psychological variables into smaller, more homogenous groups (Smith, 1956).
such as values, attitudes, goals, and personality. The term psychographics as it is used in the field of
There has been surprisingly little research directly market segmentation was coined by Demby in 1965
comparing the efficacy of the two approaches. The goal (Demby, 1994). To enrich the fields understanding
of the present research was to conduct just such an of consumer behavior and to enhance advertising
analysis, empirically comparing the relative effective- strategies, Dembys idea was add to demographic
ness of the two approaches in segmenting markets for segmentation by further segmenting people based on
a wide variety of products ranging from cell phones their tendency to think or act in a certain way (Demby,
to lottery tickets. To provide context for the current 1994). Subsequent marketers and researchers began
approach, the fields of psychographic and demographic to use psychological variables such as values, motiva-
segmentation are first briefly reviewed. tions, and personality traits to segment their potential

Psychology and Marketing, Vol. 30(11): 937949 (November 2013)


View this article online at wileyonlinelibrary.com/journal/mar

C 2013 Wiley Periodicals, Inc. DOI: 10.1002/mar.20657

937
customers (e.g., Engel, Kollat, & Blackwell, 1969; applications in advertising and marketing. For exam-
Wells, 1975). ple, Yankelovic and Meer (2006) use the example of the
Early psychographic segmentation was heavily psychographic-based Pepsi Generation campaign. They
rooted in personality profiling. One of the most com- argued that though this campaign reinforced youth cul-
mon inventories used for this purpose was the Edwards ture in Pepsis brand image, it did nothing to influence
Personal Preference Schedule (Edwards, 1959), a scale commercial activity. One of their main complaints with
on which people rated themselves on 15 normal needs psychographic segmentation was the lack of empirical
or motives. For example, Evans (1959) attempted to data in predicting actual consumer behavior.
predict brand preference of Ford and Chevrolet own- Other researchers, such as Wells (1975) have offered
ers based on psychological needs (e.g., autonomy, dom- slightly more positive critiques of psychographic seg-
inance, order). Unfortunately Evans and similar re- mentation. Wells points out that when psychological
searchers attempts at predicting consumer behavior constructs have been relevant to the consumer behav-
were largely unsuccessful (Wells, 1975). ior being studied that predictive correlations have been
Eventually, the use of personality profiling as a ba- in the 0.20s and 0.30s. He notes that although some
sis for psychographic segmentation was replaced by have been quick to dismiss numbers of this magnitude,
the lifestyle approach (Wells, 1975). First introduced they are often high enough to offer substantial predic-
by Lazer (1963), lifestyle is generally defined as the tive validity. He also points out that accounting for dif-
patterns in which people live and spend their time ferences in segments is often more important than ex-
and money (Vyncke, 2002). Often, lifestyle question- plaining the variance in individual consumer behavior.
naires were designed within the context of the prod- Despite these occasional positive critiques, there has
uct being sold. So, for example, heavy users of shot- been a broader lack of systematic research on the effi-
gun ammunition were identified by such questions as: cacy of demographic and psychographic segmentation
I like hunting and I would like to be a policeman (Novak & MacEvoy, 1990).
(Plummer, Erickson, & Groves, 1975; Wells, 1975). Ex- It is clear that the study of psychographics in con-
panding upon this method, Plummer (1974) and oth- sumer behavior is fraught with complex methodological
ers took a more person-centered approach to lifestyle issues as well as a number of misconceptions by those
segmentation (e.g., Vyncke, 2002). The most widely both inside and outside of the academic field of con-
accepted model of lifestyle segmentation is the AIO sumer behavior. Three factors in particular may con-
approachactivities, interests, and opinions. Activities tribute to the failure of psychographics as a prominent
include actions such as vacations, shopping, sports, and method of segmentation.
work. Interests are defined as the degree of excitement
that is affiliated with topics such as fashion, media,
food, family, and home. Lastly, opinions include any de- Criticisms of Studies that Have Used
scriptive beliefs that one holds (e.g., politics, business, Psychographics as a Predictor of Consumer
education, culture; Plummer, 1974). The AIO batter- Behavior
ies were often quite large, with as many as 300 items
(e.g., Wells & Tigert, 1971). The AIO approach has seen One factor driving the methodological issues of using
some popularity over the years largely because psycho- psychographics in market segmentation is the aban-
graphic lifestyle profiles (like the one derived in the donment of more traditional personality measures,
hunting study) and product-specific psychographic pro- which occurred after the 1960s (e.g., Wells, 1975). One
files can be useful in providing focused marketing ef- factor driving this abandonment was that marketers
forts for a specific product (e.g., Plummer, 1974). commonly adopted measures developed in clinical con-
A closely related method of segmentation is based texts; such instruments, designed to provide clinical
on values. In the field of psychographics, values can be diagnoses, are not well suited to large-scale assess-
defined as goals that extend across situations, are en- ments of personality. For example, The Edwards Per-
during, and serve as guiding principles in peoples lives sonal Preference Schedule (Edwards, 1959), which was
(Vyncke, 2002). The two most common measures used used in early psychographic research, was primarily de-
for values-based segmentation are the Values and Life veloped for personal counseling. Its effectiveness in con-
Styles (VALS; Mitchell, 1983) and the List of Values sumer research was largely unsuccessful (Wells, 1975).
(LOV; Kahle, 1983; Veroff, Kulka, & Douvan, 1981). Additionally, marketers have favored ad hoc (rather
Utilizing values as a segmentation method has two than established) measures that either relate to a
advantages over the AIO approach: values capture a specific product or that relate to a theory of a particular
broader array of behavior and they can be measured marketing group (e.g., Novak & MacEvoy, 1990; Wells,
with shorter instruments. 1975). Many of these ad hoc measures were developed
The lifestyle (or AIO) approach and the values ap- by nonpsychologists who typically are not trained in
proach are the main methods of psychographic segmen- psychometrics; as a result issues such as reliability
tation that remain in use today. These methods have and validity have often been neglected, resulting in
received mixed reviews from researchers. Some authors measures that are not well placed to provide predictive
have argued that psychographic methods in general do power in the consumer-behavior context (e.g., Vyncke,
not provide enough predictive validity to offer practical 2002). Additionally, ad hoc scales in general do not

938 SANDY, GOSLING, AND DURANT


Psychology and Marketing DOI: 10.1002/mar
provide the same robustness of reliability and validity and has experienced little variation, but psychographic
as do established scales (Furr, 2011). measurement is more complex and has seen a great deal
A second and related methodological concern is the of development over the past few decades. In particu-
narrow scope of behavioral outcomes that are typically lar, the Big Five model has emerged as a widely used
examined in traditional segmentation studies. Study- personality framework, with established reliability and
ing such a narrow scope of behaviors makes compara- validity (e.g., Goldberg, 1992; John & Srivastava, 1999;
bility and generalizability practically impossible. For McCrae & Costa, 1999). The Big Five captures five
example, it would be hard to generalize results about broad dimensions of personality: Extraversion, Agree-
personality predictors in shotgun ammunition to other ableness, Conscientiousness, Neuroticism, and Open-
types of purchasing behavior such as selecting a brand ness. Researchers have conducted many studies using
of painkillers (OConnor & Sullivan, 1995). The narrow the Big Five as a framework, with topics ranging from
focus may not be problematic for market practitioners their impact on health outcomes to their role in pre-
because they will tend to focus on a problem and then dicting pet preference (Gosling, Sandy, & Potter, 2010;
move on to the next one. But more generalizable mea- Roberts et al., 2007). When behaviors are predicted us-
sures are needed for scientists trying to understand ing a common framework (e.g., Big Five), it becomes
broader patterns of consumer behavior. easier to compare findings across studies and to gener-
Finally, segmentation research has largely over- ate informed hypotheses in future research.
looked how psychographic and demographic variables Second, a wider scope of behaviors needs to be exam-
may differentially contribute to predicting consumer ined than has been previously been studied in psycho-
behavior. Wells (1975) cites a couple of studies from graphic research. The narrow scope of previous studies
the 1960s and 1970s that examined the contributions may have served the needs of the product makers and
of both demographic and psychographic variables (e.g., advertisers at the time but a broader array of vari-
Bass, Pessemier, & Tigert, 1969; Burger & Schott, 1972; ables is needed to provide a more general picture of the
King & Sproles, 1973). The psychographic variables relative predictive ability of demographic and person-
rarely accounted for a great deal of the variance but ality variables. The narrow scope of previous research
they often captured greater differences than the demo- has also limited the generalizability of results. In the
graphic variables. For example, Bass, Pessemier, and same way that the Big Five can provide a common mea-
Tigert (1969) examined factor dimensions of various surement framework, using a broad array of consumer
consumer behaviors (e.g., magazine readership, media) behaviors can provide a broad pattern of results on
and found that of the 31% total variance explained, which future researchers may build. This framework
psychographic variables (responses to AIO items such may allow researchers to develop a model from which
as Husband oriented, Risk avoidance, Information to generate predictions about consumer behavior more
seeker) accounted for 16.7% while demographic vari- generally.
ables accounted for 14.3%. One drawback to this study Lastly, it is important to evaluate psychographics
and others that Wells cites is that the sample size was and demographics concurrently when predicting con-
relatively small (N = 344) and restricted (housewives sumer behavior. Doing so will allow researchers to eval-
in Lafayette, Indiana). uate cases in which one method may be more advanta-
Aside from the studies examining psychographic and geous than the other (e.g., Novak & MacEvoy, 1990;
demographic studies in the 1960s and 1970s there is Roberts et al., 2007). Roberts et al. (2007) recently used
one other more recent exceptiona study conducted by this strategy in a meta-analytic study of the incremen-
Novak and MacEvoy (1990). The primary goal of their tal validity of personality variables over demographics
study was to compare the efficacy of the two values ap- in terms of a variety of important life outcomes (e.g.,
proaches to segmentation (i.e., VALS and LOV). But marital status, mortality, occupational attainment).
the researchers also evaluated the predictive ability of Thus, the present study sought to build on previous
psychographics over demographics. They found in 60 of research by addressing the issues highlighted above.
64 consumer outcomes, that demographics offered more To address inconsistent and unreliable measurement,
predictive ability than psychographics (as measured by the analyses were based on the Big Five framework.
the VALS and LOV). Unfortunately, investigating the This framework brings sound psychometric quality to
relative predictive validity of demographics and psy- the analyses of consumer behavior and it will allow
chographics has fallen out of practice. future researchers to build on the current hypotheses
and findings.
To address the lack of generalizability in psycho-
What is Needed? graphic research, a diverse array of consumer behaviors
was examined, ranging from purchasing lottery tickets
To understand the predictive power of personality and automobiles to endorsing political candidates and
relative to demographic factors, three things are watching television shows. To understand broader be-
needed. First, to address the measurement issues, well- havioral patterns, exploratory factor analysis was used
established scales need to be implemented in psycho- to derive meaningful latent variables. By predicting
graphic research. The measurement of demographic these patterns using standard demographics and psy-
variables (e.g., age, sex, income) is fairly uncomplicated chographics, researchers can begin to understand how

PERSONALITY AND CONSUMER DECISIONS 939


Psychology and Marketing DOI: 10.1002/mar
these variables may predict consumer behavior more Table 1. Demographics.
generally. This breadth, like the Big Five framework, Survey 1 Survey 2 Survey 3
will provide the foundation for broader theorizing re-
garding the factors that drive consumer behavior. Total N 8878 31,883 5570
Most importantly, the analyses focused on the incre- Age 1893 18100 1890
mental validity of personality variables over and above (M = 50.23, (M = 49.12, (M = 50.36,
SD = 13.46) SD = 12.76) SD = 12.74)
demographic variables. The outcome variables mea-
Sex
sured in the present study can be grouped into three
Female 72.2% 77.9% 73.9%
different consumer markets (Vynke, 2002): media (e.g., Male 27.8% 22.1% 26.1%
television programs), indirect buying behavior (e.g., en- Ethnicity
vironmental attitudes), and consumer goods (e.g., cars, White 81.6% N/A 74.9%
lottery tickets). Black 7.6% N/A 7.3%
Asian or Pacific 1.4% N/A 1.0%
Islander
METHODS American Indian 1.0% N/A 1.0%
and Alaska
Native
The data were collected as part of three surveys and Other, including 3.2% N/A 3.4%
were administered by Mindset Media, a marketing com- multiple
pany focused on psychographic measurement and tar- ethnicities
geting. Unknown 5.2% N/A 12.4%
Income
$0$24,999 17.8% N/A 15.9%
Participants $25,000$49,999 32.5% N/A 29.4%
$50,000$99,999 32.7% N/A 30.3%
Participants were 45,843 volunteers who provided de- $100,000+ 9.1% N/A 9.7%
mographic and personality information over the Inter- Unknown 7.8% N/A 14.6%
net from 2008 to 2009. Participants age ranged from Education
18 to 100 (mean = 49.6; SD = 12.9) and 76.4% were Grammar school 0.1% N/A 0.04%
female. See Table 1 for the demographic breakdown Some high school 1.7% N/A 1.5%
High school grad 21.5% N/A 18.1%
across surveys.
Some college 33.7% N/A 31.2%
Associate degree 10.6% N/A 10.8%
Bachelors 17.9% N/A 17.3%
Measures degree
Post grad degree 9.2% N/A 8.6%
Personality. The Big Five personality traits were Unknown 5.2% N/A 12.4%
measured using items from the NEO PI-R (Costa & Note: N/A indicates that information was not available. Ethnic-
McCrae, 1992). Unfortunately, as a result of opera- ity, income, and education information was not available for Survey 2
tional considerations at Mindset Media, Neuroticism participants.
items were not collected. Thus, the analyses of per-
sonality presented here focus on Extraversion, Agree-
ableness, Conscientiousness, and Openness. Across the
three surveys, reliability was acceptable for all the Big
Five personality factors: Extraversion ( = 0.820.84), stance, there are a number of general items directed at
Agreeableness ( = 0.740.81), Conscientiousness ( = determining how restrictive parents are of their chil-
0.84.85), and Openness ( = 0.710.73). drens behavior (e.g., How much do you restrict your
childs use of video games?). This item is not a di-
Consumer Behavior. Three categories of variables rect measure of consumer behavior but it is indicative
related to consumer behavior were measured: media, of how much money that particular individual might
indirect buying behavior, and consumer goods. The me- spend on video games for his or her child. This category
dia category includes items such as types of television of outcome variables is not always treated as a form
shows watched and amount of hours spent per week of consumer behavior in the previous literature but it
consuming various forms of media (e.g., TV, newspa- is potentially influential and of interest to real-world
pers, magazines; see Table 2, for a break down of the practitioners.
items). Consumer goods include types of automobiles pur-
Indirect buying behavior includes items such as en- chased, electronic (e.g., flat screen TVs, cell phones; see
vironmental attitudes, political behavior, and parental Table 4 for breakdown of the items) purchases, and lot-
restrictions (see Table 3 for a breakdown of the items). tery ticket purchases.
This category of variables is not a pure measure of di- As with the personality variables, the selection of
rect consumer behavior. However, the variables all di- behavioral variables was determined by the operational
rectly or indirectly influence buying behavior. For in- needs of Mindset Media.

940 SANDY, GOSLING, AND DURANT


Psychology and Marketing DOI: 10.1002/mar
Table 2. Demographic and Personality Variable Effects on Media Consumption.
Personality Demos Personality Personality
Item Survey Demographic Variables Variables Only R2 Only R2 + Demos R2 R2 Change

Movies watched per 1 Age, gender, education, E, A, C, O 0.042 0.033 0.067 0.025
month (in theaters) income
Do you buy movie tickets 1 Age, gender, education, E, A, C, O 0.125 0.043 0.149 0.024
online? income
Hours spent watching TV 2 Age, gender E, A, C, O 0.014 0.013 0.029 0.015
(per week)
Network 2 Age, gender E, A, C, O 0.011 0.014 0.023 0.012
Cable 2 Age, gender E, A, C, O 0.004 0.005 0.008 0.004
Premium channels 2 Age, gender E, A, C, O 0.007 0.012 0.016 0.009
Crime shows 2 Age, gender E, A, C, O 0.002 0.003 0.005 0.003
Sports 2 Age, gender E, A, C, O 0.046 0.012 0.057 0.011
Reality competitions 2 Age, gender E, A, C, O 0.024 0.007 0.032 0.008
Reality living shows 2 Age, gender E, A, C, O 0.087 0.004 0.089 0.002
Relationship dramas 2 Age, gender E, A, C, O 0.023 0.004 0.027 0.004
Action dramas 2 Age, gender E, A, C, O 0.035 0.012 0.042 0.007
Sitcoms 2 Age, gender E, A, C, O 0.032 0.004 0.034 0.002
Adult animation 2 Age, gender E, A, C, O 0.118 0.055 0.151 0.033
Morning shows 2 Age, gender E, A, C, O 0.016 0.008 0.022 0.077
Soap operas 2 Age, gender E, A, C, O 0.027 0.004 0.030 0.003
Hourly news 2 Age, gender E, A, C, O 0.026 0.014 0.040 0.014
Political shows 2 Age, gender E, A, C, O 0.04 0.030 0.076 0.036
Evening news 2 Age, gender E, A, C, O 0.063 0.009 0.073 0.01
Local news 2 Age, gender E, A, C, O 0.035 0.014 0.043 0.008
Cable news 2 Age, gender E, A, C, O 0.024 0.017 0.043 0.019
Financial news 2 Age, gender E, A, C, O 0.026 0.022 0.051 0.025
Hours spent reading the 3 Age, gender, education, E, A, C, O 0.041 0 0.041 0
newspaper (per week) income
Hours spent reading 3 Age, gender, education, E, A, C, O 0.01 0.001 0.01 0
magazines (per week) income
Hours spent on the 3 Age, gender, education, E, A, C, O 0.015 0.001 0.015 0
Internet (per week) income
Hours spent listening to 3 Age, gender, education, E, A, C, O 0.006 0.002 0.01 0.004
the radio income
Note: Change in r was estimated using Nagelkerke method. r change indicates incremental validity, or change in variance explained by
personality variables when demographics are accounted for.

Items that are binary.

RESULTS To assess the incremental variance explained over


demographic variables, stepwise regressions were per-
To assess the unique and incremental variance ac- formed. In the first block of each regression, the demo-
counted for by demographic and personality variables, graphic variables were entered. In the second block of
multiple regressions were computed for each outcome the regressions, demographic and personality variables
variable. In cases of binary outcomes (e.g., Are you were included. This method accounted for the variance
a registered voter?), logistic regressions were com- provided for by the demographic variables and allowed
puted and the variances were estimated via the Nagelk- us to assess the change in variance accounted for due
erke method (Nagelkerke, 1991). The goal of the cur- to the addition of personality variables.
rent paper was to assess the unique predictive valid- First, it should be noted that the full predictive
ity of personality variables as a whole over and above models (i.e., demographics and psychographics com-
demographic variables, not to examine the predictive bined) explained, on average, only 5% of the variance.
effect of particular personality (e.g., Extraversion) or Personality variables on average explain 2% of the
demographic (e.g., gender) variables. To this end, de- variance while demographics on average explain 3% of
mographic and personality variables were grouped into the variance. Thus overall, the R2 for both personality
blocks for the analyses. Therefore, beta coefficients for and demographic variables were quite low. Next, the
individual personality or demographic variables were incremental validity of personality over and above
not reported. Researchers interested in the specific ef- demographic predictors was evaluated, the overall
fects regarding individual variables may contact the small effects notwithstanding. After computing the
authors for more information. change in variance (i.e., the change from Step 1 with
demographics only to Step 2 with demographics and

PERSONALITY AND CONSUMER DECISIONS 941


Psychology and Marketing DOI: 10.1002/mar
Table 3. Demographic and Personality Variable Effects on Indirect Consumer Behavior.
Personality Demos Personality Personality
Item Survey Demographic Variables Variables Only R2 Only R2 + Demos R2 R2 Change

Are you a registered 1 Age, gender, education, income E, A, C, O 0.114 0.044 0.131 0.017
voter?
Are you a Democrat? 1 Age, gender, education, income E, A, C, O 0.015 0.012 0.026 0.011
Are you a Republican? 1 Age, gender, education, income E, A, C, O 0.023 0.035 0.056 0.033
Are you an 1 Age, gender, education, income E, A, C, O 0.011 0.013 0.023 0.012
Independent?
Did you vote for John 1 Age, gender, education, income E, A, C, O 0.042 0.044 0.078 0.036
McCain?
Did you vote for Barack 1 Age, gender, education, income E, A, C, O 0.034 0.056 0.08 0.046
Obama?
Global warming
Scientists will find a 1 Age, gender, education, income E, A, C, O 0.023 0.051 0.054 0.031
solution to global
warming without
people having to
make big changes
to their lifestyles.
Recycling
I go out of my way to 1 Age, gender, education, income E, A, C, O 0.013 0.091 0.113 0.100
buy recycled
products.
Sustainability
I go out of my way to 1 Age, gender, education, income E, A, C, O 0.012 0.037 0.042 0.030
buy sustainable
products.
Parental restrictions
If you are a parent, how restrictive are you of the following?
Sugar 1 Age, gender, education, income E, A, C, O 0.016 0.012 0.028 0.012
TV 1 Age, gender, education, income E, A, C, O 0.004 0.006 0.011 0.007
Video games 1 Age, gender, education, income E, A, C, O 0.027 0.004 0.029 0.002
Movies 1 Age, gender, education, income E, A, C, O 0.002 0.003 0.004 0.002
Music 1 Age, gender, education, income E, A, C, O 0 0.004 0.004 0.004
Internet 1 Age, gender, education, income E, A, C, O 0.043 0.002 0.044 0.001
Note: Change in r was estimated using Nagelkerke method. r change indicates incremental validity, or change in variance explained by
personality variables when demographics are accounted for.

Items that are binary.

psychographics combined), the ratio of change was ple, when evaluating how many hours of television per
calculated by dividing R2 for the second regression week an individual watches, R2 increased from 0.01 to
block by the R2 for the first regression block. The 0.03, an increase of 2.07. However, when examining a
average factor increase for all outcome variables more specific genre such as Sitcoms no change could
assessed was 1.9. Thus, when adding personality to be observed in R2 . One exception to this genre-specific
a model of overall consumer behavior, on average the limitation was Crime shows and Political shows. Crime
variance almost doubled. shows increased by 2.5 in R2 when adding personality
The primary aim of this paper was to evaluate the variables to the regression. Similarly, political shows
overall value of personality variables in offering predic- increased by 1.9 in R2 .
tive ability over and above demographics. It is, however, As shown in Table 3, the indirect buying behav-
worth examining the areas, if any, in which personality ior category also revealed a complex pattern of find-
is more beneficial than others. As shown in Tables 24, ings. Personality played an important role in predict-
certain areas of consumer behavior benefited more than ing items related to attitudes about recycled products
others did from the addition of personality variables to and sustainable products. For example, demographic
the model. variables alone predicted whether or not an individ-
As shown in Table 2, in the media category, predic- ual will purchase recycled products with an R2 = 0.013
tions of general television watching items (e.g., How but after adding personality to the model, R2 increased
many hours of TV do you watch?, Do you subscribe to to 0.113an increase of a factor of 8.69. In the cate-
premium channels?) benefited from adding personal- gory of political items, the two behaviors that benefited
ity to the model. For most of the more specific genre most from personality variables were whether or not
questions, personality did not account for a great deal the person was a registered Republican (increase in R2
of variance over and above demographics. For exam- of 2.43) and whether or not the person voted for Barack

942 SANDY, GOSLING, AND DURANT


Psychology and Marketing DOI: 10.1002/mar
Table 4. Demographic and Personality Variable Effects on Consumer Products.
Personality Demo Personality Personality
Item Survey Demographic Variables Variables Only R2 Only R2 + Demos R2 R2 Change

Lottery
Do you play? 1 Age, gender, education, income E, A, C, O 0.02 0.019 0.044 0.024
How often? 1 Age, gender, education, income E, A, C, O 0.02 0.007 0.024 0.004
Scratch-off games 1 Age, gender, education, income E, A, C, O 0.084 0.007 0.089 0.005
Daily or numbers 1 Age, gender, education, income E, A, C, O 0.014 0.008 0.027 0.013
games
Weekly or jackpot 1 Age, gender, education, income E, A, C, O 0.031 0.008 0.037 0.006
games
Electronics
Flat screen TVs 1 Age, gender, education, income E, A, C, O 0.06 0.019 0.067 0.007
Computers 1 Age, gender, education, income E, A, C, O 0.031 0.019 0.046 0.015
Cell phones 1 Age, gender, education, income E, A, C, O 0.088 0.031 0.111 0.023
Mac or PC? 1 Age, gender, education, income E, A, C, O 0.021 0.031 0.045 0.024
Cars
Number of cars 3 Age, gender, education, income E, A, C, O 0.181 0.001 0.182 0.001
Minivan 2 Age, gender E, A, C, O 0.01 0.014 0.026 0.016
Sedan 2 Age, gender E, A, C, O 0.012 0.006 0.02 0.008
Sports car 2 Age, gender E, A, C, O 0.004 0.01 0.012 0.008
Station wagon 2 Age, gender E, A, C, O 0.001 0.007 0.009 0.008
SUV 2 Age, gender E, A, C, O 0.009 0.005 0.015 0.006
Truck 2 Age, gender E, A, C, O 0.038 0.008 0.041 0.003
Note: Change in r was estimated using Nagelkerke method. r change indicates incremental validity, or change in variance explained by
personality variables when demographics are accounted for.

Items that are binary.

Obama (increase in R2 of 2.35). One area that did not and 2.60, respectively). However, for sedans, SUV, and
see much benefit from adding personality variables to truck owners, the factor of change when adding per-
the model was whether or not the individual is a regis- sonality variables induced much smaller factors of R2
tered voter (an increase in R2 of only 1.15). Among the change (1.60, 1.60, and 1.00, respectively).
parental restriction items, restriction of television and To get a clearer grasp of how demographics and per-
movies (R2 change of 2.75 and 2.00, respectively) ben- sonality variables predict general consumer-behavior
efited more than Internet, video games, or music (R2 patterns, an exploratory factor analyses was performed
change of 1.07, 1.02, and 1.00, respectively). on the pools of dependent variables. Separate factor
As shown in Table 4, the findings for the consumer- analyses were performed on each of the three sur-
goods category were also complex. Similar to the finding veys. Across all three surveys, 11 dimensions were
in the media category, general patterns of consumer derived. Three dimensions pertained to indirect buy-
behavior often benefited more from adding personal- ing behavior: Conservative Political Behavior, Recy-
ity to the model. For example, adding personality to cle/Sustainable Behavior, and Parental Restrictive-
the model predicting lottery-ticket behavior increased ness. Five dimensions pertained to media consumption:
the R2 from 0.02 to 0.04 (2.2 factor increase). However, News/Information shows, Entertainment shows, Type
when evaluating the more specific behavior of whether of Television, Traditional Entertainment, Modern En-
the person plays scratch-off games, the R2 was 0.08 for tertainment. And three dimensions pertained to con-
demographics alone, increasing to 0.09 when personal- sumer products: Electronic, Lottery, Car Ownership.
ity was added to the model (factor of change of 1.06). The component matrices for the 11 dimensions can be
Contrary to this specificity effect, the categories of elec- seen in Tables 5 through 7.
tronics and car ownership presented a slightly different Some dimensions were better predicted by demo-
picture. For electronics, personality added almost noth- graphic variables and others were better predicted by
ing to explained variance for items such as flat screen demographic variables. As shown in Table 8, personal-
TVs per household and cell phones per household (with ity variables were superior in predicting the following
factor of changes of 1.00 and 1.2, respectively). How- dimensions: Conservative Political Behavior, Recy-
ever, when evaluating whether the participant owned a cle/Sustainable Purchases, Type of Television, and Tra-
Macintosh or a PC, personality was a strong contributor ditional Entertainment. Demographic variables were
(R2 increased from 0.02 to 0.05, a 2.14 factor increase). superior in predicting the remaining dimensions: Elec-
The findings for car ownership were split. When pre- tronic Purchases, Lottery, Parental Restrictiveness,
dicting station wagon, sports car, and minivan owners, News/Information Shows, Entertainment Shows, Car
adding personality variables increased explained vari- Ownership, and Modern Entertainment. Results also
ance by double or more (factor of change of 9.00, 3.00, revealed that for about half of the dimensions, adding

PERSONALITY AND CONSUMER DECISIONS 943


Psychology and Marketing DOI: 10.1002/mar
Table 5. Rotated Component Matrix for Exploratory Factor Analysis of DVs in Survey 1.
Conservative Electronics/TV Lottery Green Products
I II III IV

John McCain 0.801 0.014 0.053 0.009


Republican 0.791 0.000 0.058 0.052
Barack Obama 0.679 0.070 0.002 0.072
Global warming 0.407 0.055 0.093 0.076
American dream 0.366 0.043 0.017 0.091
Democrat 0.605 0.009 0.145 0.026
Electronics: Computers 0.035 0.698 0.040 0.012
Electronics: Cell phones 0.003 0.649 0.120 0.121
Electronics: Flat screens 0.127 0.595 0.027 0.003
Movie tickets online 0.060 0.456 0.083 0.024
Movies per month 0.040 0.449 0.071 0.184
MAC versus PC 0.029 0.243 0.093 0.156
Lottery: Scratch-off games 0.025 0.024 0.708 0.032
Lottery: Daily or numbers games 0.017 0.068 0.615 0.046
Lottery: Weekly or jackpot games 0.044 0.123 0.525 0.073
Lottery: How often do you play? 0.010 0.070 0.309 0.033
Recycled products 0.091 0.063 0.044 0.783
Sustainability products 0.020 0.036 0.044 0.755

Table 6. Rotated Component Matrix for Exploratory Factor Analysis of DVs in Survey 2.
Parental News/Information Entertainment Type of Car
Restrictiveness Shows Shows Television Ownership
I II III IV V

Parental restriction: TV 0.980 0.052 0.05 0.031 0.032


Parental restriction: Movies 0.980 0.045 0.059 0.031 0.032
Parental restriction: Music 0.974 0.042 0.059 0.031 0.029
Parental restriction: Video games 0.972 0.035 0.054 0.031 0.032
Parental restriction: Sugar 0.970 0.034 0.051 0.026 0.028
Parental restriction: Internet 0.970 0.024 0.044 0.025 0.035
Minivan 0.233 0.011 0.029 0.070 0.195
TV: Politics 0.099 0.732 0.087 0.132 0.039
TV: Cable news 0.049 0.711 0.047 0.312 0.004
TV: Financial news 0.056 0.694 0.208 0.196 0.025
TV: Evening news 0.119 0.676 0.160 0.308 0.023
TV: Hourly news 0.066 0.642 0.284 0.142 0.015
TV: Local news 0.026 0.592 0.180 0.346 0.024
TV: Morning shows 0.016 0.544 0.309 0.232 0.016
TV: Sports 0.006 0.426 0.247 0.105 0.035
TV: Hours per week 0.118 0.197 0.055 0.016 0.033
TV: Sitcoms 0.027 0.030 0.685 0.100 0.005
TV: Relationship dramas 0.000 0.074 0.661 0.142 0.008
TV: Action dramas 0.018 0.049 0.623 0.022 0.012
TV: Reality competition 0.043 0.146 0.581 0.140 0.042
TV: Reality living 0.123 0.162 0.575 0.317 0.028
TV: Soaps 0.030 0.239 0.519 0.102 0.006
TV: Adult animation 0.070 0.148 0.513 0.262 0.069
TV: Crime dramas 0.043 0.093 0.494 0.073 0.011
TV: Food shows 0.031 0.293 0.448 0.264 0.039
TV: Network 0.002 0.186 0.034 0.903 0.049
TV: Cable 0.010 0.053 0.250 0.806 0.034
TV: Premium channels 0.019 0.029 0.007 0.233 0.064
Sedan 0.183 0.000 0.003 0.033 0.946
SUV 0.082 0.009 0.025 0.028 0.731
Truck 0.019 0.025 0.036 0.032 0.217
Station wagon 0.058 0.038 0.040 0.045 0.126
Sports car 0.024 0.006 0.010 0.021 0.104

944 SANDY, GOSLING, AND DURANT


Psychology and Marketing DOI: 10.1002/mar
Table 7. Rotated Component Matrix for Exploratory not measured here that could account for some of the
Factor Analysis of DVs in Survey 3. variance in consumer behavior. Future work is needed
Traditional Modern to expand the predictive model.
Entertainment Entertainment Despite overall low R2 values the present research
I II suggests that personality variables can offer substan-
tial predictive validity over and above demographic
Newspaper hours 0.829 0.022
variables when evaluating consumer behavior. The ex-
Magazines hours 0.827 0.001
plained variance in more than half of the items in
Radio hours 0.328 0.186
Internet hours 0.176 0.784 this study increased by at least 60% as a result of
TV hours 0.191 0.736 adding personality variables to the model. Similar to
the findings of Roberts et al. (2007), the present re-
search revealed that the explanatory power of per-
sonality variables equaled and often times outper-
personality variables to a model of demographic
formed demographic variables. However, the increment
variables alone increased explained variance by 100%
in predictive ability varied widely across categories of
or more.
consumer behavior.
One factor that seemed to influence whether or not
DISCUSSION personality provided more predictive ability was the
level of specificity of the behavior. Previous research
Strikingly, the overall effect sizes and R2 values re- has shown that personality is better at predicting gen-
ported above were very small. Consistent with past eral or aggregated behaviors than specific behaviors
studies (e.g., Novak & MacEvoy, 1990), individual dif- (e.g., Barrick & Mount, 2005). As previously mentioned,
ference measureswhether they are demographics, personality is found to be consistent over time and
psychographics, lifestyles, etc.do not seem to pro- across situationsbut is known to fluctuate across nar-
vide a great deal of predictive power in the context row situations (e.g., across the course of a day; e.g.,
of buying behavior. There are a few possible explana- Fleeson, 2001). The media and consumer-goods cate-
tions for the generally low effect sizes. The first is that, gories seemed to be particularly affected by specificity.
like personality-related behaviors (e.g., Fleeson, 2001), For example, items such as number of hours spent
buying behaviors (e.g., electronics purchases) may vary watching TV per week and whether or not one plays the
a lot within person in the short term but show more lottery, benefited substantially from adding personality
stability over longer time periods. So buying behaviors to the model.
may need to be aggregated over longer periods of time in In the media category, general behavior (e.g., types
order to obtain reliable estimates of purchasing habits. of networks watched, amount of hours spent watching
A second possible explanation is that personality and TV) benefited most from personality. Factor analyses
demographic variables may predict greater variance in of the dependent variables showed that general behav-
the brand decisions (e.g., Toyota vs. Ford) than in the ior (e.g., TV Network Type) was predicted better by
class of product (e.g., minivan vs. SUV). A whole field of personality than by demographics. However, entertain-
research is devoted to understanding the personality of ment showseven when aggregated were still better
brands (e.g., Aaker, 1997) but the connection between predicted by demographic variables.
the brand personalities of the products and the person- In the indirect buying behavior category there
ality and demographics of the consumer (e.g., Huang, were several behavioral outcomes that benefited
Mitchell, & Rosenaum-Elliott, 2012) has yet to be exam- significantly from adding personality to the model
ined extensively. Lastly, there could many other situa- specifically, political behavior and attitudes about
tional variables and individual differences (e.g., values) recycled and sustainable products. There has been a

Table 8. Demographic and Personality Variable Effects on Latent Dependent Variables.


Personality Demos Personality Personality
Category Demographic Variables Variables Only R2 Only R2 + Demos R2

Conservative Age, gender, education, income E, A, C, O 0.025 0.063 0.077


Electronic Age, gender, education, income E, A, C, O 0.103 0.061 0.145
Lottery Age, gender, education, income E, A, C, O 0.011 0.005 0.018
Recycle/sustainable Age, gender, education, income E, A, C, O 0.01 0.091 0.098
Parental restrictiveness Age, gender, education, income E, A, C, O 0.016 0.004 0.022
News/information shows Age, gender E, A, C, O 0.031 0.028 0.06
Entertainment shows Age, gender E, A, C, O 0.06 0.009 0.064
Type of television Age, gender E, A, C, O 0.011 0.013 0.021
Car ownership Age, gender E, A, C, O 0.009 0.005 0.015
Traditional entertainment Age, gender E, A, C, O 0.002 0.002 0.005
Modern entertainment Age, gender E, A, C, O 0.005 0.002 0.007

PERSONALITY AND CONSUMER DECISIONS 945


Psychology and Marketing DOI: 10.1002/mar
substantial amount of research evaluating relation- predicting these types of car ownership but not as
ships between the Big Five and political orientation strongly as personality.
(e.g., Carney, Jost, Gosling, & Potter, 2008) as well Conversely, truck, SUV, and sedan ownership was
as environmental attitudes (e.g., Swami, Chamorro- predicted more by demographics than by personality.
Premuzic, Snelgar, & Furnham, 2010). Carney et al. Truck and sedan ownership was significantly related
(2008) found that liberals are consistently higher on to being a male and SUV ownership was related to
Openness than are conservatives. So it is not sur- being female. SUV owners tended to be younger and
prising that adding the Big Five, including Openness, truck owners tended to be older. In terms of person-
would drive an increase in explained variance when ality, truck owners were less agreeable and less open.
predicting outcomes related to political behavior. Like the truck owners, SUV owners were less agree-
In addition to political behavior, the predictions able and open. They were also more extraverted and
of endorsements of recycled and sustainable products conscientious. Sedan drivers were less extraverted and
benefited greatly from the addition of personality to agreeable and more conscientious and open. These per-
the model. Previous research (Swami et al., 2010) has sonality relationships were significant but the amount
shown that environmental concern is related to Agree- of explained variance over demographic variables was
ableness, Contentiousness, and low Neuroticism. The small in comparison to other types of car ownership.
present study replicated the effects of Agreeableness Additionally, demographics were more predictive of the
and Conscientiousness for recycled products and of general car-ownership factor. No published research ex-
Conscientiousness for sustainable products. It was not amining types of car ownership and Big Five traits was
possible to replicate the Neuroticism finding because, found but the results here suggest personality could be
as noted above, data on this trait were not collected. a fruitful factor to examine in future studies.
Interestingly, Swami et al. also found a positive rela- Overall, demographic variables were more predic-
tionship between Openness and both recycled products tive of lottery-ticket purchases than were the per-
and sustainable products. The present results do not sonality variables. Consistent with previous research
precisely mirror those past findings but it is clear that (Haisley, Mostafa, & Loewenstein, 2008), the findings
personality plays a large role in endorsing these types showed that lottery players tended to be young, male,
of products. and have a lower income. One exception was the item
One area within indirect buying behavior that Do you play the lottery? Past research has indicated
yielded mixed results was parental restrictiveness. The that lottery play is associated with being higher on
overall factor of parental restrictiveness was more ef- Extraversion and lower on Agreeableness and Open-
fectively predicted by demographics. However, when ness (Balabanis, 2002). The present study replicated
examining the more specific components of parental re- the negative Openness effect but the demographic
strictiveness, it was found that TV watching, movies, effects outweighed this contribution.
and sugar benefited significantly from adding person- Similar to lottery-ticket purchases, predictions of
ality to the model whereas restrictiveness with regard electronics purchases benefited most from demographic
to video games, Internet, and music did not. A meta- variables. Owning more computers, flat screen TVs,
analysis (Prinzie, Stams, Dekovic, Reijntjes, & Belsky, and cell phones per household was strongly related
2009) revealed that the more restrictive parents are to higher income. Cell phone and computer ownership
in general, the less open and agreeable they are and was also correlated with being younger. An exception
the more neurotic they are. The present results regard- to these items was the Macintosh versus PC question.
ing Openness and Agreeableness are consistent with Personality (specifically higher Openness among Mac-
this previous research and are also consistent across intosh users) almost doubled the explained variance
domains of restriction. in the model. Previous researchers have evaluated dif-
Like the media and indirect consumer behavior ferences between Macintosh and PC users in terms of
categories, the consumer-goods category showed vari- Internet use and Technophilia (Yang & Lester, 2003)
ability in how it benefited from personality variables. but there has been no examination of the personality
Automobile ownership in particular revealed a mixed differences between users.
pattern of results. Minivan, station wagon, and sports Overall, the patterns of findings reveal a complex
car ownership benefit the most from adding personality picture of how demographics and psychographics relate
to the model while truck, SUV, and sedan ownership to consumer behavior. Generally, where the present
benefits the least. Taking a closer look, minivan owners findings can be compared to previous research, the
tended to be more extraverted and agreeable and less findings replicate. However, a subset of findings ei-
conscientious and open. They also tend to be female. ther failed to replicate or had never been previously
Station wagon owners showed the exact opposite examined. Nonetheless, it is clear that certain out-
effects of personality (low Extraversion and Agreeable- come variables benefited from personality more than
ness, high Conscientiousness and Openness). They others (i.e., the general factors of political behavior, re-
also tended to be female. Sports car ownership was cycling and sustainable product attitudes, and broad
positively related to Extraversion and Openness but television watching behavior). To echo a point made
negatively to Agreeableness. There were no significant by Wells (1975), it is important to consider the nature
effects for age or sex. Demographics played a role in of ones outcome variables when deciding what type of

946 SANDY, GOSLING, AND DURANT


Psychology and Marketing DOI: 10.1002/mar
segmentation method to use. The present study pro- the results varied within and across categories. This
vides an initial framework of personality findings upon problem was especially acute with respect to studying
which future researchers can expand. What is clear is multiple consumer behaviors at once. Also, as noted in
that personality is capable of providing as much, and in the introduction, few studies have examined the rel-
some cases more, explanatory power than demographic ative of importance of demographic and psychographic
variables when predicting consumer behavior. variables in the domain of consumer behavior. It is thus
difficult, in some cases impossible, to make comparative
judgments about when and why personality should out-
LIMITATIONS AND FUTURE DIRECTIONS perform demographic variables and vice versa. Thus,
future researchers are encouraged to expand the list
There are a few limitations associated with the current of consumer behaviors and predictive variables they
study. First, as mentioned above, the absolute amounts study.
of variance explained by both personality and demo- There are a number of practical implications as-
graphic variables were very small. The magnitude of sociated with the current findings. If marketers and
the effects explains a very small amount of the vari- advertisers understand the psychological motivations
ance in the data and therefore limits the ability of for why individuals buy certain products, they can
the model to predict future outcomes. As previously then leverage that information to better promote their
mentioned, personality offers greater predictive abil- brands. Future research in this area should focus on un-
ity when examined in the context of repeated and ag- derstanding specific ways in which information about
gregated behavior. Therefore, it may be profitable for psychographic variables can be used to promote prod-
future research to collect repeated consumer behaviors uct sales. For example, Belisle and Bodur (2010) found
(e.g., How many days a week did you buy lottery tick- that the personality of online avatars reflected the ac-
ets over the past year?). It is also important to note tual personalities of the consumers who created them.
that, although personality effects were small, the de- They proposed that companies could utilize this infor-
mographic effects were virtually identical in magnitude mation in tandem with segmentation research to craft
across the prediction matrix. A major avenue of future company avatars that resonate with particular psycho-
inquiry is needed to determine what accounts for the graphic segments. For instance, extraverted consumers
lions share of the predictive variance. Is it situational may prefer to encounter an avatar who smiles and talks
factors, like effective advertising, marketing strategy, a lot.
and like? Or is it other demographic (e.g., geographic As previously mentioned, another potentially use-
location) and personality (e.g., values, self-esteem, at- ful area of future research is in brand personality. Re-
tachment styles) variables not evaluated here? searchers have established that product brands have
Second, the selection of variables was limited by the discernable personality dimensions (e.g., Aaker, 1997).
constraints of third-party data collection. In terms of A certain brand could be associated with any number
personality variables, the goal of Mindset Media was to of specific consumer products (e.g., cosmetics, hair care,
create a set of mindsets that could successfully pre- nail polish). Researchers have investigated the effects
dict consumer behavior. Personality researchers would of the presence or absence of certain brand personali-
be interested in evaluating pre-established personal- ties. For example, if a brand is seen as socially respon-
ity measures. Unfortunately the developers at Mind- sible, are consumers more willing to reward that brand
set Media did not deem the Neuroticism items of the by purchasing its products (Madrigal & Boush, 2008)?
NEO PI-R necessary to their data-collection goals. Thus Research has also been conducted to understand how
the current model was limited to the remaining factors new products will fit into an existing brand personality
of the Big Five. It is unknown how much the lack of (Lau & Phau, 2007). For instance, will a new car model
Neuroticism influenced the present findings. There has fit with the brand personality of the rest of the cars
been some work demonstrating effects for Neuroticism that a particular company manufactures? Less stud-
predicting consumer behavior and motives. For exam- ied, however, is the link between brand personality and
ple, Lynn (1994) demonstrated that the more neurotic human personality. How do certain human personality
an individual is, the more likely he or she is to tip at traits influence the decision to buy brands with partic-
a restaurant. In another study that examined shop- ular personalities? Understanding this link could help
ping motivations, Neuroticism predicted a utilitarian researchers and practitioners understand the patterns
shopping motivation (Guido, 2006). Based on these re- of human motivations associated with particular brand
sults, it seems plausible that the inclusion of Neuroti- personality profiles.
cism items would have increased the predictive power of In conclusion, personality variables do offer statisti-
the personality items. Obviously, future studies should cally significantoften times yielding R2 values larger
include all five of the Big Five factors. than those associated with demographic variables
Third, the majority of the literature directly relating predicative validity with regard to consumer behavior.
the specific consumer outcomes (e.g., lottery, electron- Future work is needed in this area to fully understand
ics, automobile, TV genres, parental restrictions) to the the role personality plays in accounting for variance
Big Five personality traits is very small. So previous in purchasing behavior. As it stands, the overall
research offers very little in the way of clues to why predictive validity of personality (and demographic)

PERSONALITY AND CONSUMER DECISIONS 947


Psychology and Marketing DOI: 10.1002/mar
variables is low. Thus, these results offer compelling Huang, H. H., Mitchell, V., & Rosenaum-Elliott, R. (2012). Are
evidence for continued research on uncovering the consumer and brand personalities the same? Psychology &
connection between consumer behavior and individual Marketing, 29, 334349.
differences. John, O. P., & Srivastava, S. (1999). The Big Five Trait taxon-
omy: History, measurement, and theoretical perspectives.
In L. A. Pervin & O. P. John (Eds.), Handbook of personal-
ity: Theory and research (2nd ed.) (pp. 102138). New York,
REFERENCES NY, US: Guilford Press.
Kahle, L. R. (Ed.). (1983). Social values and social change:
Aaker, J. L. (1997) Dimensions of brand personality. Journal Adaptation to life in America. New York: Praeger.
of Marketing Research, 34, 347356. King, C. W., & Sproles, G. B. (1973). Predictive efficacy of
Balabanis, G. (2002). The relationship between lottery ticket psycho-personality characteristics in fashion change agent
and scratch-card buying behaviour, personality and other identification. Proceedings of the Annual Convention of the
compulsive behaviours. Journal of Consumer Behaviour, 2, American Psychological Association, pp. 847848.
722. Koponen, A. (1960). Personality characteristics of purchasers.
Barrick, M. R., & Mount, M. K. (2005). Yes, personality mat- Journal of Advertising Research, 1, 1612.
ters: Moving on to more important matters. Human Perfor- Lau, K., & Phau, I. (2007). Extending symbolic brands using
mance, 18, 359372. their personality: Examining antecedents and implications
Bass, F. M., Pessemier, E. A., & Tigert, D. J. (1969). A tax- towards brand image fit and brand dilution. Psychology &
onomy of magazine readership applied to problems in mar- Marketing, 24, 421444.
keting strategy and media selection. Journal of Business, Lazer, W. (1963). Lifestyle concepts and marketing. In S.
42, 337363. Greysser (Ed.), Toward scientific marketing (pp. 24352).
Belisle, J., & Bodur, H. (2010). Avatars as information: Percep- Chicago, IL: American Marketing Association.
tion of consumers based on their avatars in virtual worlds. Lynn, M. (1994). Neuroticism and the prevalence of tipping:
Psychology & Marketing, 27, 741765. A cross-country study. Personality and Individual Differ-
Burger, P.C., & Schott, B. (1972). Can private brand buyers ences, 17, 137138.
be identified? Journal of Marketing Research, 9, 219222. Madrigal, R., & Boush, D. M. (2008). Social responsibility as
Carney, D. R., Jost, J. T., Gosling, S. D., & Potter, J. (2008). a unique dimension of brand personality and consumers
The secret lives of liberals and conservatives: Personality willingness to reward. Psychology & Marketing, 25, 538
profiles, interaction styles, and the things they leave be- 564.
hind. Political Psychology, 29, 807840. McCrae, R. R., & Costa, P. R. (1999). A Five-Factor theory of
Costa, P. T., Jr., & McCrae, R. R. (1992). NEO PI-R pro- personality. In L. A. Pervin, O. P. John (Eds.), Handbook of
fessional manual. Odessa, FL: Psychological Assessment personality: Theory and research (2nd ed.) (pp. 139153).
Resources, Inc. New York, NY, US: Guilford Press.
Demby, E. H. (1994). Psychographics revisited: The birth of a Mitchell, A. (1983). The nine American life styles. New York:
technique. Marketing Research, 6, 2630. Warner.
Edwards, A. L. (1959). Edwards Personal Preference Schedule Nagelkerke, N. J. D. (1991). A note on a general definition of
[manual] (rev. ed.) New York: Psychological Corp. the coefficient of determination. Biometrika, 78, 691692.
Engel, J. F., Kollat, D. J., & Blackwell, R. D. (1968). Consumer Novak, T. P., & MacEvoy, B. (1990). On comparing alternative
Behavior. Oxford England: Holt, Rinehart & Winston. segmentation schemes: The list of values (LOV) and values
Evans, F. B. (1959). Psychological and objective factors in the and life styles (VALS). Journal of Consumer Behavior, 17,
prediction of brand choice: Ford vs. Chevrolet. Journal of 105110.
Business, 32, 340369. OConnor, P. J., & Sullivan, G. L. (1995). Market segmenta-
Fleeson, W. (2001). Toward a structure- and process- tion: A comparison of benefits/attributes desired and brand
integrated view of personality: Traits as density distribu- preference. Psychology & Marketing, 12, 613635.
tions of states. Journal of Personality and Social Psychol- Plummer, J. T. (1974). The concept and application of life style
ogy, 80, 10111027. segmentation. Journal of Marketing, 38, 3337.
Furr, R. M. (2011). Scale construction and psychometrics for Plummer, J. T., Erickson, D. L., & Groves, D. L. (1975).
social and personality psychology. London, UK: Sage Pub- Lifestyle portrait of the hunter. Journal of Instructional
lications. Psychology, 2, 3340.
Goldberg, L. R. (1992). The development of markers for the Prinzie, P., Stams, G. M., Dekovic, M., Reijntjes, A. A., &
Big-Five factor structure. Psychological Assessment, 4, Belsky, J. (2009). The relations between parents Big Five
2642. personality factors and parenting: A meta-analytic review.
Gosling, S. D., Sandy, C. J., & Potter, J. (2010). Personalities Journal of Personality and Social Psychology, 97, 351362.
of self-identified dog people and cat people. Anthrozoos, Roberts, B. W., Kuncel, N. R., Shiner, R., Caspi, A., & Gold-
23, 213222. berg, L. R. (2007). The power of personality: The compar-
Guido, G. (2006). Shopping motives, Big Five factors, and the ative validity of personality traits, socioeconomic status,
hedonic/utilitarian shopping value: An integration and fac- and cognitive ability for predicting important life outcomes.
torial study. Innovative Marketing, 2, 5767. Perspectives On Psychological Science, 2, 313345.
Haisley, E., Mostafa, R., & Loewenstein, G. (2008). Subjec- Smith, W. R. (1956). Product differentiation and market seg-
tive relative income and lottery ticket purchases. Journal mentation as alternative marketing strategies. Journal of
of Behavioral Decision Making, 21, 283295. Marketing, 21, 38.
Holak, S. L., Matveev, A. V., & Havlena, W. J. (2007). Nostal- Swami, V., Chamorro-Premuzic, T., Snelgar, R., & Furnham,
gia in post-socialist Russia: Exploring applications to ad- A. (2010). Egoistic, altruistic, and biospheric environmen-
vertising strategy. Journal of Business Research, 60, 649 tal concerns: A path analytic investigation of their determi-
655. nants. Scandinavian Journal of Psychology, 51, 139145.

948 SANDY, GOSLING, AND DURANT


Psychology and Marketing DOI: 10.1002/mar
Veroff, J., Kulka, R. A., & Douvan, E. (1981). Mental health in Yang, B., & Lester, D. (2003). PC versus Macintosh users.
America: Patterns of help-seeking from 1957 to 1976. New Psychological Reports, 92, 616.
York: Basic Books. Yankelovic, D., & Meer, D. (2006). Rediscovering market seg-
Vyncke, P. (2002). Lifestyle segmentation: From attitudes, in- mentation. Harvard Business Review, 84, 122132.
terests and opinions, to values, aesthetic styles, life visions
and media preferences. European Journal of Communica-
tion, 17, 445463.
Wells, W. D. (1975). Psychographics: A critical review. Journal Correspondence regarding this article should be sent to:
of Marketing Research, 12, 196213. Carson J. Sandy, Ph.D. Director of Behavioral Research
Wells, W. D., & Tigert, D. (1971). Activities, interests and Tip Tap Inc., 150 5th St. no. 3, Cambridge, MA 02141
opinions. Journal of Advertising Research, 11, 2735. (cjosandy@gmail.com).

PERSONALITY AND CONSUMER DECISIONS 949


Psychology and Marketing DOI: 10.1002/mar

View publication stats

Vous aimerez peut-être aussi