 

 

€33 UBRARY
V Michigan State
University

 

 

This is to certify that the
dissertation entitled

REVISITING THE EFFECT OF MODELS’ RACE
ON EXPLICIT AND IMPLICIT ATTITUDES TOWARD
ADVERTISEMENTS

presented by

EUNSUN LEE

has been accepted towards fulﬁllment
of the requirements for the

Ph.D degree in Advertising

 

 

gag/4M;

 

Major Professor’s Signature
8/!7/ 2:” 7

Date

 

MSU is an Afﬁrmative Action/Equal Opportunity Employer

 

 

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 KzlPrqlec5Pres/CIRC/DateDue.indd

 

REVISITING THE EFFECT OF MODELS’ RACE
ON EXPLICIT AND IMPLICIT ATTITUDES TOWARD ADVERTISEMENTS

By

Eunsun Lee

A DISSERTATION

Submitted to
Michigan State University
In partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Advertising

2009

ABSTRACT

REVISITING THE EFFECT OF MODELS’ RACE
WITH EXPLICIT AND IMPLICIT ATTITUDE MEASURES

By

Eunsun Lee

Historically. studies of racial attitudes are fraught with controversy. Those who
study racial attitudes recognize that Caucasian Americans‘ racial attitudes are complex
and variable. Likewise. this is true for advertising—related minority studies. Previous
studies suggest that African Americans have positive attitudes toward ads containing
strong ethnic cues. However, the literature is not so clear regarding how Caucasian
Americans form their attitudes toward African American models in ads.

Adding to the ambiguity. recent ﬁndings suggest that people may have two
different attitudes toward an object at the same time: One is explicit. i.e.. measured by
traditional attitudinal measurements (e.g.. self-report): the other is implicit. i.e., measured
by implicit measurement (e.g., implicit association test, word-fragment completion task
etc). This dual attitude theory implies that consumers may have two different attitudes
toward the model’s race simultaneously, and that previous advertising-related minority
studies only capturing explicit attitudes by employing self-report measures may have
missed critical information.

This study attempts to rectify such a deﬁciency by examining the ability of
participants to hold dual attitudes toward the same attitude object. Speciﬁcally. the
current study employs both explicit and implicit measures of attitude in response to

advertisements containing models of different ethnic backgrounds. Both Caucasian

 

American and African American participants viewed ads with models of both Caucasian
American and African American models.

This study demonstrates that consumers may hold dual attitudes. However.
unlike previous psychological studies which show Caucasian American’s attitude
preference for models of their own race, no such bias was found in the current study. In
addition, African American participants also demonstrated consistency in both their
implicit and explicit attitudes. Contrary to previous advertising-related minority studies.
African American participants did not demonstrate in-group bias and in fact were more
positive toward ads with Caucasian models. This result supports the reasoning that,
since members of a society develop shared category norms through implicit
communication of category norms in speech and writing, participants assume that white

models in ads is the norm.

Copyright by
Eunsun Lee
2009

This dissertation is dedicated to

my parents for their unconditional love and support.

 

ACKNOWLEDGMENTS

Without the guidance of my committee members, help from friends and support
from my family, I would never have been able to ﬁnish my doctoral program.

I thank my advisor. Dr. Steve Edwards. for his generous time and commitment.
Throughout my doctoral work he encouraged me to develop independent thinking and
research skills. He continually stimulated my analytical thinking and greatly assisted
my scientiﬁc writing. Ultimately. his invaluable advice and thoughtful guidance have
helped me complete this dissertation successfully. I appreciate the countless hours he
spent reading my drafts and the numerous meetings we had to make this research better
and more valuable.

Additionally, I am most grateful for having an exceptional doctoral committee
and thank Drs. Carrie La F erle. Theresa Mastin. and Sandi Smith for their continual
support and encouragement. Throughout my years in the doctoral program, they helped
me in choosing course materials. ﬁnishing my preliminary paper, and guiding my
dissertation.

I offer special thanks to my dearest friend, Hyun-Sun Park for assisting me with
miscellaneous tasks such as modifying ads and entry data. Also, I acknowledge my
wonderful colleagues, Jung-Sun Ahn, Jounghwa Choi, Mikyoung Kim, Elizabeth
Quilliam and Myung-Ja Yang for being ﬁne helpers and friends.

Finally, I thank my parents, brother, sister-in-law and cousin. Jung-Eun Lee.

They always supported me and encouraged me with their best wishes.

vi

TABLE OF CONTENTS

LIST OF TABLES .......................................................................................................... ix
LIST OF FIGURES ........................................................................................................ x
CHAPTER I .................................................................................................................... 1
INTRODUCTION .......................................................................................................... 1
Statement of Problem .............................................................................................. 4
Purpose of Study ...................................................................................................... 6
CHAPTER II .................................................................................................................. 10
ETHNICITY AND ADVERTISINGL ........................................................................... 10
Ethinicity ................................................................................................................ 10
Ehtnicity Theories .................................................................................................... 12
Ethnicitu—related Advertising Studies ...................................................................... 14
CHAPTER III ................................................................................................................. 21
ATTITUDE .................................................................................................................... 21
Attitude .................................................................................................................... 21
Attitude toward the Ad ............................................................................................ 22
Measures of Attitude toward the Ad ....................................................................... 24
lnforrnation Processing Perspective: Spontaneous vs. Deliberative Process ..........
24
Thories of Implicit and Explicit Attitudes ............................................................. 25
Explicit Attitude Measures .................................................................................... 28
Implict Attitude Measures ..................................................................................... 29
CHAPTER IV ................................................................................................................. 31
HYPOTHESIS .............................................................................................................. 31
Racial Attitude: Prejudice ........................................................................................ 31
Implicit Attitude and Prejudice ............................................................................. 33
Hypothese .............................................................................................................. 35
CHAPTER V .................................................................................................................. 40
METHODOLOGY ......................................................................................................... 40
Stimuli Development ............................................................................................. 4O
Pretest .................................................................................................................... 42
Experimental Design ............................................................................................. 43
Participants ............................................................................................................ 45
Procedures ............................................................................................................. 45
Dependent Variables ............................................................................................... 48
CHAPTER IV ................................................................................................................. 50
DATA ANALYSIS ........................................................................................................ 50
Descriptive Statistics ............................................................................................... 50

vii

Puriﬁcation of Response Latencies ....................................................................... 53

Manipulation Check for Measures .......................................................................... 56
Hypotheses Testing ................................................................................................. 57
Analysis of Control Group (Condition 5) ................................................................ 60
Puriﬁcation of Response Latencies for Condition 5 ...................................... 60
Hypotheses Testing for Condition 5 .............................................................. 61
CHAPTER IV ................................................................................................................. 64
DISCUSSION AND CONCLUSIONS .......................................................................... 64
Discussion ................................................................................................................ 64
Attitude toward Ads ....................................................................................... 64

Attitude Measures: Response Latency ........................................................... 66

Control Group: Condition 5 ........................................................................... 67
Implications ............................................................................................................. 67
Limitations ............................................................................................................... 70
Suggestions for Future Study .................................................................................. 71
Conclusion ............................................................................................................... 73
APPENDICES .............................................................................................................. 75
REFERENCES ............................................................................................................... 81

viii

LIST OF TABLES

Table 1. Experimental and Control Groups. ................................................................. 44
Table 2. Sequence of Experiment ............... ' .................................................................. 4 6
Table 3. Descriptive Statistics. Correlations. and Scale Reliabilities ......................... 50
Table 4. Descriptive Statistics and Correlations for Caucasian American ................... 51
Table 5. Descriptive Statistics and Correlations for African American ....................... 52

Table 6. Correlations and Descriptive Statistics for Implicit and Explicit Measures... 56
Table 7. Participant Race by Model’s Race on Explicit Attitudes ............................... 57
Table 8. Participant Race by Model’s Race on Response Latencies ............................ 58
Table 9. Participant Race by Model’s Race by Attitude Measure (Implicit/Explicit) . 59
Table 10. Participant Race by Model‘s Race on Explicit Attitudes in Condition 5. 62
Table 1 1. Participant Race by Model‘s Race on Response Latencies in Condition 5.. 62

Table 12. Participant Race by Model‘s Race by Attitude Measure
(Implicit/Explicit) in Condition 5 ......................................................................... 63

LIST OF FIGURES

Figure 1. FCB Grid — 15 Chosen Products ...................................................................... 41
Figure 2. Overall Response Latencies for Conditions ..................................................... 55
Figure 3. Overall Response Latencies for Conditions (ConditionS) ............................... 61

 

CHAPTER I

INTRODUCTION

With the growing diversity of the US. population, ethnic target marketing has
become an increasingly crucial component of marketing strategy (Pieres et a1. 2003,
Torres 2007). As a matter of fact, the US. Census estimates that by 2050 the Hispanic,
African American, and Asian American segments of the US. population will collectively
exceed 50%, compared to less than 30% in 2003 (US. Census Bureau 2003). As the
size and spending power of ethnic groups continues to grow, a greater number of
marketers are paying more attention to targeting issues (Nelson and Lukas 1990). While
some commercials like the recent Dove advertisement showing differently shaped women
of various ethnicities try to register with a broad group of people, research ﬁndings
present conﬂicting results regarding consumer response to the inclusion of ethnic cues in
advertising. In light of these trends, it is important to examine both the pros and cons of
targeting ethnic minorities with ethnic cues.

Many advertising-related minority studies have examined the inﬂuence of models’
ethnicities in advertisements on consumer attitudes (Aaker et a1. 2000; Briley et a1. 2007;
Deshpande and Stayman 1994; Dimofte et a1. 2003; Forehand and Deshpande 2001;
Whittler 1989; Whittler and DiMeo 1991; Wooten 1995). Several studies have found
that minority consumers respond positively to ethnic cues when they mirror consumers’
ethnic memberships (Whittler 1989). For example, Aaker, Brumbaugh, and Grier
(2000) argue that consumers react more favorably to advertisements in which the models

are of the same ethnicity as the consumers. Beyond race (Whittler 1989). researchers

 

have also examined ethnic identiﬁcation (Williams and Qualls I989), cultural knowledge
(Brumbaugh 2002). and ethnic salience (Deshpande and Stayman 1994) as targeting cues
that may increase a consumer‘s afﬁnity for a brand. These studies suggest that since
advertising cues can trigger a feeling of similarity between the consumer and the model,
effective ads need to include ethnic models, ethnic identiﬁcation and/or salient ethnic
cues, when targeting ethnic populations (Forehand and Deshpande 2001; Forehand and
Reed 2002; Dimofte et a1. 2003).

In addition, the inclusion of ethnic minority cues in ads does not seem to detract
from the effectiveness of ads targeted toward broader populations. Several studies
found that Caucasian Americans do not react negatively toward advertisements featuring
African American models (Bush et a1. 1972: Solomon at a1. 1976; Whittler 1989). For
example, Whittler (1989) found no difference between C aucasians" attitudes in response
to advertisements that contained either African American models or Caucasian models.
Bush, Gwinner and Solomon (1972) further found that Caucasian American consumers
did not purchase differently because of displays containing Caucasian Americans, African
Americans, or integrated models. In contrast, Whittler and DiMeo (1991) found that
Caucasian Americans have less favorable attitudes toward advertisements featuring
African American models than those featuring Caucasian American models. In
reconciling the different results between his two studies, Whittler (1991) argued that
today’s younger Caucasian Americans have more opportunity for interaction with
individuals of different races, and they may be more willing to accept them than old
people.

However some psychological studies would suggest that Caucasian Americans

ls)

 

may in fact prefer Caucasian models, but could be modifying their responses to be
perceived as socially acceptable, whether conscious or not of this process (Petty et a1.
1999; Brunel. Tietje and Greenwald 2004; White and Harkins 1994). Therefore, it is
important for advertisers to examine not only traditional explicit measures of attitudes
toward ads but also implicit measures when assessing the effectiveness of ethnic minority
models in ads. Research projects focusing on racial attitudes are often fraught with
controversy due to the sensitivity of the subject matter, the potential for respondent biases,
as well as contextual effects, and a host of other measurement issues. Therefore, the
current study examines the potential role of social desirability in inﬂuencing Caucasian
consumer responses to ads with models of a different race.

Recent theories in social psychology assume that people may have two different
attitudes toward an object at the same time: One is explicit. i.e., measured by traditional
attitudinal measurements (e.g.. self-report); the other is implicit. i.e., measured by
implicit measurement (e.g., implicit association test, word-fragment completion task etc)
(F riese et a1. 2006). According to Fazio and Olson (2003), attitudes can exert inﬂuence
through relatively spontaneous or more deliberative process/es. The former involve
judgments of, or behavior toward, an object being inﬂuenced by one’s construal of the
object in the immediate situation. The judgments are perceptions that themselves can be
affected by individuals’ attitudes having been automatically activated upon encountering
the attitude object. In contrast. deliberative processing involves a more effortful, cost-
beneﬁt analysis of the utility ofa particular behavior. Therefore. implicit attitudes may
exert inﬂuence through automatic activation, while explicit attitudes may be formed

through effortful activation. Brunel, Tietje and Greenwald (2004), applying these

 

theoretical concepts to consumer social cognition, found that participants did not exhibit
racial preferences at the explicit level but showed strong preference for ads containing
Caucasian American Models.

This implies that consumers may have two different attitudes toward the model’s
race simultaneously, and that previous advertising-related minority studies only explored
explicit attitudes by employing self-report measures. As more ads in ‘general’
magazines and TV programs increase their presentation ofethnic minority models in an
effort to reach a broad range of consumers. the ﬁndings of this study become even more
critical. Through a technique utilizing both explicit self-reports and implicit attitude

measures, the viability of dual-attitudes toward the same stimuli are investigated.

Statement of Problem

Unfortunately, methodological problems lead uncertain ﬁndings in the
advertising-related minority studies. Many such studies failed to consider issues of l)
attitudinal measures. 2) advertising components, and 3) product types.

First, the previous advertising-related minority studies adopted a technique
utilizing explicit self-reports. Several studies found that attitude measurement is highly
context-dependent, and minor changes in question wording, format or order and a variety
of factors can impact, profoundly, the obtained attitude reports (e. g., Fazio and Olson
2003; Greenwald and Banaji 1995; Hotgraves 2004). As a matter of fact, many
psychologists also have demonstrated that participants “self-edit” attitude-related answers
for reasons of social desirability and self presentation (Fazio et a1. 1995: Greenwald et a1.

1998).

Second, the previous advertising-related minority studies failed to examine, solely,
models' effects in ads. The studies used ﬁction- or nonﬁction-ads which manipulated
only models‘ races alone. i.e., except race. all advertising components in the ads were
similar. However, they failed to control other advertising component (i.e., color, theme.
and background etc.) inﬂuence on consumer perceptions. Thus. it is hard to justify the
previous studies having solely examined models’ race effect in ads.

Finally, the previous studies employed a limited number of product types in their
experiments, such as a watch (Torres 2007), a liquid detergent (W hittler 1991), and sport
equipment (Brunel et a1. 2004). According to extensive content analysis in Henderson
and Baldasty (2003). some products are more closely associated with Caucasian
Americans than with people of color, and vice versa. For example, whereas more
expensive cars were advertised solely by Caucasian American models, less expensive
cars were advertised by models of color. Caucasian American models dominate in ads
for many products most associated with home: non-fast food (requiring cooking) and
domestic and household product (toothpaste, soap, household cleaners etc). On the
other hand. people of color were closely tied to low-cost and low-nutrition products (i.e.,
fast food, soft drinks. candy. gum etc). Unlike Caucasian Americans who were shown
preparing foods at home, when people of color were shown using non-fast food products.
they appeared mostly with convenience foods. Therefore, the use of product, i.e.,
choosing neutral products or including diverse product types, is critical in advertising-
related minority studies.

Considering these issues, can we take the results of Caucasian American (or

African American) attitudes toward African American (or Caucasian American) models

’JI

from the previous advertising-related minority studies at face value?

Purpose of Study

The experiment discussed here uses rigorous experimental design methods to
examine the effects of model race on viewers’ attitudes in diverse product categories.
Unlike the previous advertising-related minority studies to examine attitude through the
limited technique utilizing only participant self-reports, participants’ spontaneous and
automatic responses were also measured using response latencies. Using multiple
measures captures greater variability and may allow for the exploration of dual attitudes
toward the ethnic cues shown in the ads.

Second, to exclusively examine model’s race effect in the ads, advertising
components (i.e., brand. product. color. background) other than model were removed
deliberately and completely through experimental manipulation. In this current study,
participants were exposed to four different advertisement types including: background
only, background and product only, background, product and African American model,
and background, product and Caucasian American Model. These four different group
designs permit examination of different advertising component effects. In data analysis,
participants’ response latencies for those exposed to background, product and model ads
were subtracted from participants’ response latencies that were exposed to background
and product only ads. Subsequently. the reﬁned response latencies were used for further
data analysis resulting in the examination of model race and not other extraneous
advertising cues.

Finally, to remove unexpected product effects on attitudes toward the ads, 15

different types of products were used, deliberately chosen from F oote, Cone and
Belding’s Grid (FCB Grid). The FCB grid dimensionalizes consumers’ attitudes toward
products in terms of two dimensions: involvement and affect versus cognition (Ratchford,
1987; Vaughn, 1986). This allows eliminating participants’ involvement and
emotional/cognitive effects on their attitudes toward the ads.

Executing such a rigorous experimental design assures that the current study
contribute to our understanding about attitude formation in general and the role of dual
attitudes in response to advertising. Importantly, capturing both explicit and implict
attitudes in response to the same stimuli, allows for the comparison of those attitudes and
for bias or self-censuring to be detected. Such an examination is critical to truly
understanding a topic where political correctness is demanded, such as race in America.
Speciﬁcally for the ﬁeld of advertising, this study clariﬁes the multiple conﬂicting
ﬁndings regarding the use of ethnic cues.

First, Muehling and MaCann (1993) suggest that “future research needs to
provide further insight regarding whether attitude toward the ad is an automatic,
involuntary reaction to an ad or whether it is a conscious more deliberate element (or
both)” (p.51). Since traditional measures of attitude toward ads typically rely on
individuals’ conscious self-reports of affect (or cognition) toward ads, the previous
advertising-related minority studies employing these measures were difﬁcult for
identifying the “automatic” nature of attitude formation. In fact, unlike marketers
expect, consumers may not spend much time to read print advertisements or TV
commercials. They tend to page through magazines and skip many ads. unless

intentionally looking for speciﬁc information. Therefore. it is true that consumers may

have automatic or spontaneous attitudes toward the ads that they brieﬂy scanned. From
this viewpoint. the implicit measure newly employed in this study is worthy of pursuing
relevant to the deﬁnitional issue of attitude toward the ads.

Second, the study helps to clarify the use of models’ of different races in ads.
Advertising strategies may need to consider model ethnicity, ethnicity of the targeted
consumers and situational or biographical reasons for awareness of it, and product appeal
involving ethnic cues. Therefore, placing ethnic minority models in media popular with
minorities clearly indicates that advertisers aim to reach and appeal to ethnic minority
consumers. Previous studies showed that African Americans were likelier to show
positive attitude toward ads containing African American models than toward ads
containing Caucasian Americans (Appiah 2001b; Forehand and Deshpande 2001;
Forehand and Reed 2002; Dimofte et a1. 2003). However, there has been controversy
over use of African American models in mainstream ads. Recent content analysis study
showed that Caucasian American models are the norm in mainstream magazine ads, and
that African American models had second-highest frequency (Knobloch-Westerwick and
Coates 2006). It is not clear how Caucasian Americans perceive African American
models in the ads. Whereas some studies suggested that Caucasian Americans do not
show negative attitude toward African American models in ads (Bush et a1. 1972;
Solomon at al. 1976; Whittler 1989). other studies suggested that Caucasian Americans
have less favorable attitude toward African American Models in ads (Whittler and DiMeo
1991). By employing rigorous experimental design, this current study examines the
effect of model ethnicity on African American and Caucasian American consumers'

attitudes.

Finally, the current study provides an example of how implicit measures may be
employed in advertising research to uncover hidden biases (Brunel 2004; Gawronski
2006). Historically. advertisers have been reluctant to use African American models in
ads, fearing that African American characters might offend Caucasian American
consumers and adversely affect sales of the advertised product and other products offered
by the sponsoring company (Appiah 2001b; Bush et a1. 1979; Qualls and Moore. 1990).
The use of African American models in mainstream media is not only about attitude
toward the ad, but also about racial attitudes toward African Americans. According to
Fazio, Jackson, Dunton and Williams (1995), some individuals do experience guilt and
self-recrimination after becoming aware that they would respond more negatively than
they believe they should, thus are reluctant to express their opinions when reporting their
attitudes. From this viewpoint. implicit attitude measure provides unique opportunity to

study Caucasian American attitudes toward African American models in ads.

CHAPTER II

ETHNICITY AND ADVERTISING

Ethnicity

According to the theory of group differences in ad processing, individuals process
ad information based on their social background (Bhat, Leigh and Wardlow 1998;
Brumbaugh and Grier 1999). A social group has different attitudes. personal
characteristics, product needs, media habits. lifestyles. experiences and motives from
other groups (Bhat, Leigh and Wardlow 1998). Hence, each social group differently
processes information based on its shared subculture (Scott 1994).

Social identity theory, the similarity-attraction paradigm, suggests that
demographically-similar individuals perceive themselves as more alike and are more
inclined to identify with one another than with members of dissimilar dyads (Byme,
1971). Among demographic factors, ethnic identity is a component of acculturation
related to how members of an ethnic group relate to the group as a subset of larger society
(Green 1999). Phinney (1996) deﬁnes ethnic identity as the subjective sense of ethnic
group membership. Ethnic identity constitutes “a basic part of the ethnic individual’s
personality, and is a powerful contributor to ethnic group formation, maintenance, and
social ties” (Bernai and Knight 1993, p. 1). To understand the role of ethnicity in the
attitudes and behavior of an individual, it is critical to take ethnic identity into account.

Ethnic identity is a set of self-ideas about one’s own ethnic group membership
that has several dimensions along which these self-ideas vary (Bernai and Knight 1993).

The degree of ethnicity identity is expressed by two terms: ethnic salience and ethnic

strength (Forehand et. a1. 2002). They are distinct constructs in terms of their duration.
Whereas ethnic salience is the momentary activation oftheir own ethnicity, ethnic
intensity is an enduring association between an individual’s sense of self and the person’s
ethnic identity (Forehand et. a1. 2002). In other words. ethnic salience is most often
elicited when people process ethnicity-related information and they categorize
themselves along ethnicity—related criteria.

Many researchers have viewed ethnic identity as an enduring and fundamental
aspect of the self that includes (a) a sense of membership in an ethnic group (i.e.. self-
identiﬁcation) and (b) the attitudes and feelings associated with that membership (i.e..
self-afﬁliation) (Bernai and Knight 1993; Cokley et. a1. 2007; Keefe 1992; Phinney 1990).
Self-identiﬁcation and self-afﬁliation varies in importance and strength among ethnic
group members (Keefe 1992).

Self identiﬁcation refers to “the ethnic labels or terms that people use in
identifying themselves and the meaning of these labels” (Bernai and Knight 1993, p. l).
Casas and Pytluk (1995) stated that because ethnic identiﬁcation is a product of a
socialization process that can be unique to each individual, it is not the same for all
members of any respective group. As another dimension of ethnic identiﬁcation, ethnic
affiliation is “the preferences, feelings. and values that people may embrace, reject. or
have neutral feelings and preferences about their ethnic families, companions, and
cultural values” (Bernai and Knight 1993, p. 1).

In consumer behavior, a great deal of research has discussed the inﬂuence of
ethnic identiﬁcation on consumer attitudes and judgments. Research has been

conducted on various advertising components such as media, ethnic cues. product types.

and racial composition (Green 1999; Grier et. a1. 2006; Ragoonan et. a1. 2005; Williams
and Grantham 1999). In the following sections. the major theories and studies related to

ethnicity are reviewed.

Ethnicity Theories

There are several theories relevant to the current study focused on ethnicity
which each explain how self-identity is formed and maintained by minority groups
including: homophily theory, distinctiveness theory, in-group bias theory, and
accommodation theory.

Most general is the the phenomenon of perceived similarity between two people
which is referred to as homophily (Rogers and Bhowmik 1970). Homophily is deﬁned
as “the degree to which pairs of individuals who interact are similar with respect to
attributes, such as beliefs. values. education. social status. etc” (p.526). Perceived
similarity leads to greater identiﬁcation. Then, the greater identiﬁcation causes greater
spokesperson credibility, which results in a positive attitude toward the ad (Deshpande
and Stayman 1991). According to homophily theory, members of ethnic minorities feel
connected to a message source that is not only a member of their own ethnic group but
also of a different minority group because their connection is based on the fact that they
are both minority members. not that they belong to the same minority (Torres 2007).

Race is one way peoplejudge similarity and group membership as it is distinctive
and often easily recognized. According to distinctiveness theory. an individual’s
distinctive traits in relation to other people in the environment are more salient to the

individual than are more common traits (Aaker et. a1. 2000; Deshpande and Stayman

1994; Forehand et. al., 2002). Thus, ethnic salience occurs when an individual is
prompted to categorize himself or herself along ethnicity-oriented criteria. Thus,
individuals who belong to a distinctive or numerically rare group tend to be highly aware
and mindful of characteristics shared by that group and are likelier to incorporate that
group identity into their self-concept than are individuals who do not belong to such a
group. As a result. the lower the population of minority group members in the overall
population. the likelier that ethnically targeted advertising models are effective as a
means to activate ethnic identity (Deshpande and Stayman 1994).

Relatedly, in-group bias theory suggests that positive bias exists toward members
of one’s own group, but also suggests that just because people prefer in-group members,
does not mean they automatically disfavor out-group members (Brewer 1979). The
theory implies that ethnic minority members are predisposed to look favorably on cultural
aspects of the in-group that deﬁne membership (e.g., language, customs, values, and
social views) over cultural aspects of the out-group (Green 1999). Thus, according to
the theory, people prefer members of their own ethnic group as information sources over
members of other ethnic groups as sources. As an example, Kerin (1979) found that in
evaluating print ads, Caucasian Americans associated better product quality with their
own ethnic group’s model, whereas African Americans associated better quality with
African American models in the ads. Thus, people tend to positively bias judgments
about those perceived to be in-group members and these judgments can further impact

judgments of the information presented by in-group members.

Once ethnic identity is activated. accommodation may take place between the
recipients ofa message and the source ifthe recipients deem similarity between
themselves and the source. Accommodation theory has its roots in the sociopsychological
theories addressing similarity on attraction (Byme 1871) and has been applied in diverse
consumer research contexts (Kiran 2005). The theory ﬁrst predicted that “the greater
the amount of effort in accommodation that a blingual speaker of one group was
perceived to put into this message, the more favorably he would be perceived by listeners
from another ethnic group, and also the more effort they in turn would put into
accommodating back to the speaker” (Koslow 1994. p. 576). It suggests that as A
becomes more similar to B. the likelihood that B will favorably evaluate A is increased
(Green 1999). Accommodation theory provides useful insight about how members of
ethnic minority group respond to a message. The theory predicts that members of ethnic
minority groups should react positively to the use of ethnic cues in ads by attributing their
use to the advertiser’s sensitivity to and respect for their own ethnicity (Koslow 1994).
Thus, the targets of a message reward that sensitivity by paying attention to the message
and ultimately are more likely to purchase the advertised product.

Given all ofthe many ways that ethnicity is conceptualized and explained. the
importance of ethnicity as an informational cue becomes apparent. Therefore ethnicity

as a cue in advertisement will now be reviewed.

Ethnicity-related Advertising Studies
Ethnicity has been shown to be an important variable in the targeting of

advertising to different segments (Anonymous. 2001). The use of race as a

segmentation variable assumes that those in each group share similar characteristics
distinguishing them from members of another group. This is important to advertising
because messages designed for a speciﬁc group have been shown to be more effective
than messages targeted toward a heterogeneous mass (Aaker et al. 2001). Furthermore.
ads with racial or ethnic indicators are seen as more relevant, elicit more positive
attitudes towards the advertisements, and are believed to offer more credible messages
(Green1999; La Ferle and Morimoto 2004; Morimoto and La Ferle 2005).

Models in ads are one way to include ethnic cues in advertising. Therefore,
choosing the right model is critical and this is especially true given research ﬁndings that
show the same message delivered by different sources produce varying outcomes among
message recipients (Whittler and Spira 2002). For this reason, the effect of race on
consumers’ evaluations of advertising messages has received considerable attention.
particularly as it relates to consumers’ preferences for ads featuring models of a particular
race rather than another. The main focus of such research has been how consumers
evaluate ads featuring models of their own race compared to ads featuring models of
another race.

As mentioned in the introduction, African American consumers have been found.
to respond more favorably to ads featuring African American models, than to ads
featuring Caucasian American models (e.g.. Choundhury and Schmid 1974; Kerin 1979;
Williams et a1. 1995). However. this same beneﬁt has not typically been found for
Caucasian Americans’ attitudes where these consumers have not rated ads featuring
African American models differently from those featuring Caucasian American models

(Whittler and DiMeo 1991; Williams et a1. 1995). It seems that members of a racial

majority (i.e., Caucasian Americans) are less mindful of a model’s race in their
advertising assessments than members from an ethnic minority group (Appiah 2001a).

Variations in response patterns between Caucasian and African Americans toward
other-race models may stem from differences in the importance placed on ethnic identity.
For many Caucasian Americans, ethnicity is not as important to their identity compared
to members of ethnic minority groups (Waters 1990). Phinney and Alipuria (1990)
showed ethnic identity to be a more salient part ofthe self for ethnic minority groups.
while Alba (1985) states that Caucasian Americans do not think ofthemselves as
belonging to an ethnic group. Similarly, Waters (1990) contends that Caucasian
Americans can choose what role ethnicity plays in their lives, while Deaux (1992)
believes that ethnicity is salient for ethnic minority individuals because group
membership is more evident.

These phenomena could be explained by group position theory and norm theory.
Group position theory (Blumer and Duster 1980) suggests that ethnic minority groups
perceive competitive threats from other minority groups. Thus, when an out-group
overreaches its own group’s accustomed place in society or threatens its status and power,
an individual feels fear, apprehension, resentment, and anger (Torres 2007). These
feelings translate into a general negative response to the perceived challenge by the out-
group to its own sense of group position (Blumer and Duster 1980). Also, group
position theory argues that antagonism toward one’s out-group is best understood as a
general attitude involving normative ideas about where one’s own group should stand
relative to others in the social order (Torres 2007). According to Aaker, Brumbaugh and

Grier (2001 ). viewers in the non-target market may perceive dissimilarities between

themselves and the intended target in the ads. Consequently. they may infer that their
tastes and preferences differ from those of the intended target. thus may fail to adopt a
favorable attitude toward the ads. Also. potential backlash may result. depending on the
group targeted, for example, gays, lesbians, and African Americans.

Additionally, norm theory suggests that exemplars which come to mind in
response to a category label, determine implicit expectations for that category. Thus, the
most frequent features among the generated exemplars are known as “category norms”
(Kahneman and Miller 1986; Pratto et. a1. 2007). The concept of norm is applied to
events ranging in complexity from single visual displays to social interactions.
According to Pratto. Korchmaros and Hegarty (2007), adult Americans automatically
encode the race of individuals. Hence, if an exemplar’s race contrasts with the
category's normative features, it will become salient and the focus of explanatory
attention (Hegarty and Pratto 2001; 2004). According to Knobloch-Westerwick and
Coates (2006), Caucasian American models are the norm in mainstream magazine ads,
and that African American models have second-highest frequency. Resultantly, since
African American models are less frequent in mainstream advertising than are Caucasian
American models, whiteness would be norm and blackness would be salient.

As a result, several advertising-related minority studies have shown that racial
similarity (Whittler 1989), ethnic intensity (Williams and Qualls 1989), and ethnic
salience (Deshpande and Stayman 1994) evoke positive effects among ethnic minority
audiences. The researchers recommend using strong ethnic cues in ads, to target
speciﬁc ethnic groups. For example, in assessing the inﬂuence of advertising execution

variables on consumer social identiﬁcation and advertising response, Dimofte, Forehand

and Deshpande (2003) found evidence of the power of advertising to elicit self-awareness
when using saliently targeted ads. Such results indicate that atypical targeting strategies
stimulate the processing of ethnic information and increase consumers’ awareness of their
group memberships, while also positively influencing consumers’ responses to ads.
Dimofte, Forehand and Deshpande (2003) suggest that unusual targeting techniques may
be worth pursuing.

At the same time, minimal adverse effects on the responses of Caucasian
Americans have been found, as the majority group appears to be indifferent to the race of
the model used in an advertisement. For example, in a study that measured the sales
response of Caucasian American consumers to a point-of—purchase display of African
American models. there were no differences in Caucasian American consumers’
purchases from displays containing Caucasian American. African American, or integrated
model conditions (Solomon, Bush, and Hair 1976). Also, Forehand and Reed’s (2002)
work conﬁrmed that participants in the majority group did not react negatively to
communications targeted on a minority group.

Alternatively, ﬁndings from the psychology literature have suggested that
responses by Caucasian Americans might be inﬂuenced by how their responses could
appear to others and themselves in the context of racial interactions. Research has
shown that some Caucasian Americans were more strongly motivated to pay attention to
a minority source so as not to appear unfair or prejudice (Petty et a1. 1999; Brunel. Tietje
and Greenwald 2004; White and Harkins 1994).

Investigating the effects of models” race on persuasion using the Elaboration

Likelihood Model, White and Harkins (1994) found in a series of studies that Caucasian

Americans had the same attitudes regardless of source’s race (e.g., African American,
Hispanic American and Caucasian American). However, when they examined their
motivations to scrutinize the information, the result showed that Caucasian Americans
were more motivated to scrutinize information presented by stigmatized source (e.g.,
African American or Hispanic American) than by nonstigmatized source (e.g., Caucasian
American).

Furthermore. Petty. Fleming and White (1999) found that people were likelier to
process stigmatized groups’ messages than those of non-stigmatized groups. They
suggest that low-prejudiced individuals were especially conscious about others in a
society which did not share their unprejudiced beliefs. Thus the individuals might
carefully scrutinize the actions of stigmatized others in an effort to prevent others from
discriminating. As detailed in these studies’ ﬁndings, Caucasian American consumers
rated ads featuring African American models the same as ads featuring Caucasian
American models (Whittler and DiMeo 1991; Williams et a1. 1995), which may be the
result of participants scrutinizing their prejudiced behaviors (motivational factor) or
socially desirable responding (personality factor).

Recently, employing an implicit measure, the Implicit Association Test (IAT),
Brunel, Tietje and Greenwald (2004) demonstrated that implicit measures of attitude
toward the ad revealed an inﬂuence of ethnic preference on attitude toward the ad that
explicit measures did not. This ﬁnding underscored the potential utility of the IAT in
gaining better understanding of the impact of ethnic stereotypes on attitude toward the ad.
They argued that use of the IAT assists advertising studies to understand implicit attitudes

toward race better. because political correctness and social desirability might tend to

inﬂuence explicit answers on the issue.

Besides. some advertising-related minority studies suggest that attitude formation
may be inﬂuenced by stereotype expectation related to the product being advertised.
Lee, Mastin and Edwards (2005) revealed that Caucasian Americans, in their attitudes
toward automobiles, a highly involved “thinking” product, were impacted by the model’s
race. However, it seemed that they did not have any impact in the lowly involved
“feeling” product, online dating service that is less important and relies highly on
affective information processing. Thus. response to different race models in the ads is
likely to depend on the product being advertised (Cohen 1992: Kanungo and Pang 1973).
For example, Cohen (1992) found that for stereo speakers. responses to Asian models
were signiﬁcantly more positive than were responses to Caucasian American models.
Also, Green (1999) found that for race-based products, model race is signiﬁcant.

Consequently, when advertising-related minority studies are conducted, caution is
advised. Since Caucasian Americans worry about their response to other-race models in
ads, the use of types of measure and product types should receive attention to examining

attitudes.

CHAPTER III

ATTITUDE

Attitude

Study of attitudes is vital in contemporary social psychology. Eagly and
Chaiken (1993) conceptually deﬁne attitude as “a psychological tendency that is
expressed by evaluating a particular entity with some degree of favor or disfavor” (p. 1).
This deﬁnition implies that attitude is an internal state that lasts for at least a short time
(Eagly and Chaiken 1993). It is an acquired behavioral disposition — a learned state that
creates an inclination to respond in particular ways. Thus, it endures for a relatively
long time, but some attitudes are relatively temporary and changeable. especially if they
are unimportant to those who have them. Second, attitude is an evaluative state
accounting for covariance between certain classes of stimuli and certain classes of
responses (Eagly and Chaiken 1993). It differs in valence and intensity, and it is
evaluated as a location on a bipolar continuum or dimension ranging from extremely
positive to extremely negative.

In consumer research, attitude is a chief concern among consumers’ mental states.
Fishbein and Ajzen (1975) deﬁne attitude as “a learned predisposition to respond in a
consistently favorable or unfavorable manner with respect to a given object (p. 6).”

This deﬁnition implies that people learn feelings of favorability or unfavorability via
information about the attitude object or via direct experience with the attitude object.
Also, attitude researchers believe that an attitude is a predisposition to respond overtly.

and that this predisposition leads to actual overt behavior (Lutz 1991). It gives rise to a

21

consistently favorable or unfavorable pattern of responses. Thus, marketers and
advertisers attempt to create or modify attitudes toward their brands through the use of
marketing communications tools such as advertising.

Zanna and Rarnpel (1988) deﬁne attitude as the categorization of an entity along
the evaluative dimension. They suggest that attitudes are evaluations based on beliefs,
feelings. and/or past behaviors. implying that one may have different attitudes about the
same object at the same time. Thus they distinguish between evaluation and affect.
which has implications for the measurement of attitudes (Tesser and Shaffer 1990).
Most measures of attitude involve self-report. As long as affect and evaluation were
seen as equivalent, self-report indexes seemed adequate. But with the distinction
between affect and evaluation has come greater sensitivity to the idea that all aspects of
attitudes can be represented variously, verbal representation being only one.
Researchers believe that emotions have physiological components that may or may not be

cognitively represented (Tesser and Shaffer 1990).

Attitude Toward the Ad
Attitude toward the ad has been useful for assessing consumers’ affective
reactions to ads as opposed to purely cognitive reactions (Mackenzie et al. 1986). Most
advertising research appearing in the literature today comes from Shrimp (1981) and
Mitchell and Olson (1981) who introduced the notion that consumers’ brand/choice
behavior is likely to be inﬂuenced by attitude toward the ad. Since Mitchell and Olson
(1981) demonstrated that attitude toward the ad is a mediating influence on brand attitude.

attitude toward the ad has been considered as a distinct construct from beliefs and brand

attitudes (Muehling and McCann 1993).

In the literature. attitude toward the ad is conceptualized broadly as liking of an
advertisement, and has been regarded by some as the best indicator of advertising
effectiveness (Haley and Baldinger 1991; Brown and Stayman 1992). MacKenzie et a1.
(1985) deﬁned attitude toward the ad in their study as a “predisposition to respond in a
favorable or unfavorable manner to a particular advertising stimulus during a particular
exposure occasion” (p.130). Also. Muehling and McCanne (1993) suggest that attitude
toward the ad represents “individuals’ evaluations of the overall advertising stimulus”
(p.26).

Two positions exist to explain what attitude toward the ad may represent. First,
some attitude researchers in psychology have adopted a unidimensional deﬁnition of
attitude as a summary evaluation. Although attitudes are assumed to encompass
affective, behavioral and cognitive responses, some researchers believe that these three
classes of responses are not necessarily separable from each other, and do not necessarily
represent three independent factors (Bohner and Wanke 2002). Some consumer
behavior researchers have claimed that attitudes referred to affects or general evaluative
reactions, and recent research trends in the area have linked the concept to an affective
construct rather than a cognitive one (Mowen 1995). A review of literature on Attitude
toward the ad suggests that researchers generally used four or ﬁve items to measure
Attitude toward the ad (Lutz and Belch 1983; Mackenzie et a1. 1986; Muehling 1987;
Olson and Sentis 1983). However, most attitudes do not serve merely a single purpose,
but may be multifunctional, and have multiple types of determinants constituting

separable dimensions. Some attitudes are more automatically determined while others

may be more deliberatively determined. Brunel, Tietje and Greewald (2004) found that
dissociation exists between implicit and explicit attitudes. They argue that whereas the

explicit measures of Attitude toward the ad reﬂected participants’ intended responses, the
implicit measures reflected less-controllable automatic associations.

.laniszewski (1998) demonstrates that some antecedents to affective responses are
formed independently of conscious processing. In Muehling and McCanne’s (1993)
extensive review of the literature, they proposed three categories of antecedents of
Attitude toward the ad: personal antecedents, ad-related antecedents and other
antecedents. Personal antecedents include ad credibility, ad perceptions, attitude toward
the advertiser, attitude toward advertising and mood. Ad-related antecedents are humor
appeal. the use of celebrities. ad content. model’s race. number of ad exposures and ad
placement. Finally. other antecedents are product-related elements (novelty and

involvement). affective priming. zipping and time/delay.

Measures of Attitude Toward the Ad

Information Processing Perspective: Spontaneous vs. Deliberative Process

Empirically. attitude measurement is highly dependent on context. The
underlying dynamics of attitudes are well known as complicated cognitive and
communicative processes. MODE is an acronym for motivation (e.g., concern about
evaluation) and opportunity (e.g.. sufficient time) as determinants of whether the attitude-
to-behavior process is primarily spontaneous or deliberative in nature. According to the
MODE model of attitude-behavior processes, attitudes inﬂuence judgments and

behaviors (Fazio and Olson 2003).

24

The model proposes that attitudes can exert inﬂuence through either a
spontaneous or deliberative process. Spontaneous processing involves judgments of an
object being inﬂuenced by one’s analysis of the object in the immediate situation. while
deliberative processing involves a more extended analysis of the utility of a particular
behavior. Given the extended reﬂection required for deliberative processing, some
motivating force is necessary to induce individuals to engage in the processing. Time
and resources to deliberate. which the model refers to as opportunity. also must exist.
Consequently, attitude researchers have searched for a means of assessing attitudes not
subject to participants’ self-presentational concerns about social desirability. To
eliminate these opportunities and motivations to control individuals’ true attitudes.
attitude researchers have attempted to assess individuals’ attitudes by using a variety of

unobtrusive measures of attitude (i.e., Implicit Association Test [IAT] or priming test).

Theories of Implicit and Explicit Attitudes

 

Attempting to resolve challenges to the traditional view ofa uniﬁed attitude,
recent attitude research proposes several theories from social psychology (Gawronski and
Bodenhausen 2006; Wilson et a1. 2000) and consumer research (Cohen and Reed 2006).

Wilson, Lindsey and Schooler’s (2000) model of dual attitudes, proposes that
people can have dual attitudes which are different evaluations of the same attitude object
— automatic (implicit) attitude and deliberative (explicit) attitude. The attitude that
people hold at any particular point depends on “whether they have the cognitive capacity
to retrieve the explicit attitude and whether the explicit attitude overrides the implicit

one” (Wilson et al. 2000. p.102). This dual-attitude hypothesis comes from two main

[J
'Jl

research streams on attitudes: stored evaluation and context-sensitive constructions.

Attitudes are stable evaluations that are automatically activated, storied in
memory. and persist over time. At the same time. people construct on-the-spot attitudes
on the basis of available (accessible) information at that point. In the model, Wilson,
Lindsey and Schooler (2000) suggest that even when people change their attitude Al to
A2, A1 is not replaced and remains in memory with A2. Thus, explicit attitudes and
implicit attitudes toward the same object can coexist in memory. However, their
coexisting attitudes are differently retrieved. Whereas implicit attitude is activated
automatically. explicit attitude requires more capacity and motivation. With the right
capacity and motivation. an explicit attitude can override implicit attitude. and people
report their explicit attitude. While explicit attitude is activated, implicit attitude still
inﬂuences implicit responses such as nonverbal behaviors. On the other hand, without
explicit attitudes, people report implicit attitudes. Finally, explicit attitudes change with
relative ease, whereas implicit attitudes change more slowly. For types of dual attitudes,
Wilson, Lindsey and Schooler (2000) propose repression, independent systems,
motivated overriding and automatic overriding.

Gawronski and Bodenhausen (2006) argue that “implicit and explicit attitudes
should be understood in terms of their underlying mental processes, which are associative
and propositional processes” (p.693). They propose the associative-propositional
evaluation (APE) model. In other words, they argue that the underlying mental process
for each attitude is different. Associative process builds the basis for implicit attitudes,
and its associative evaluations are characterized as automatic affective reactions resulting

from particular associations activated automatically when people encounter relevant

stimuli. Also, it is pattern activation, which means that the activation of particular
associations in memory is determined by the relative ﬁt between preexisting associations
in memory and external input stimuli. Propositional process builds on the basis for
explicit attitudes. and its propositional evaluations are characterized as evaluative
judgments based on syllogistic inferences from any relevant information for a given
judgment. Thus. whereas implicit attitude can be activated regardless whether people
consider the associations true or not, explicit attitude is activated only if it is true, based
on validation of evaluations and beliefs. Its propositional process depends on its
subjective validity.

Finally, Cohen and Reed (2006) propose a Multiple Pathway Anchoring and
Adjustment (MPAA) model. In the model, they suggest that “attitudes develop at
different points in time through a variety of mechanisms involving personal experience,
transmitted information, and inferential/analogical reasoning” (p.7). Cohen and Reed
(2006) view the dual attitudes model as a special case of non-overlapping cognitive
elements produced by different focal thought about the same object. They suggest how
different pathways lead to attitude formation. As the foregoing distinctions in attitude
formation process, they suggest that an attitude object can be assessed from the inside-out
(explicit attitude) or outside-in (implicit attitude). If the initial attitude (implicit) was
formed in outside-in (object-centered) mechanism. an evaluative judgment about the
attitude object can be made later in an inside-out (person-centered) manner. The factors
used to test adequacy of retrieved or constructed attitudes are whether people have
representational sufﬁciency and functional sufﬁciency.

In summary, it is important to note that implicit and explicit attitudes coexist

27

toward the same objects. Unlike previous studies, attitudes are not unidimensional
summary statements. Although different explanations exist for each attitude formation
and type, it is clear that people have two different attitudes (i.e., implicit and explicit

attitudes) simultaneously. with different processes (or pathways).

Explicit Attitude Measures

 

Explicit attitudes usually are equated with deliberative and self-reported
evaluations (Gawronski and Bodenhausen 2006). Since attitudes are not directly
observable, their existence can be inferred only from overt responses or indicators on
measures as evaluative tendencies. A measure is deﬁned as “an observed score gathered
through self-report, interview, observation, or some other means” (Edwards and Bagozzi
2000, p 156). A measure is a quantiﬁed record taken as an empirical analog to a
construct that is a conceptual term used to describe a phenomenon of theoretical interest.
Since All'port (1935) asserted that “the concept of attitude is probably the most distinctive
and indispensable concept in contemporary American social psychology” (p. 798), a
variety of quantitatively-sophisticated methods for attitude measurement had been
developed to investigate attitude (i.e., psychophysical scaling, Thurstone judgment and
magnitude estimation, psychometrics, and Guttman scaling).

Measuring attitude is an important component of advertising research.
Traditionally, the advertising ﬁeld considers attitudes to be good predictors and
evaluative variables. However. for attitude researchers. measuring attitudes is not a
simple task. With no universal oprationalization of Attitude toward the ad. the measures

typically used to assess individuals’ ad attitudes are often similar. Of many studies

incorporating a measure ofAttitude toward the ad, most have used semantic differential
items-pairs (Muehling and McCann 1993). Typical endpoints used in Attitude toward
the ad research include: “good-bad.” “like-dislike.” “favorable-unfavorable,”

“informative-uninfonnative.” “appealing-unappealing,” “enjoyable-unenjoyable.”

99 66'

“interesting-aninteresting, irritating-not irritating,

’9 66 93 66

pleasant-unpleasant, nice-

1 99 66
9

awfu entertaining-unentertaining.” and “offensive-inoffensive.”

Implicit Attitude Measures

 

One ofthe most important contributions in social cognition research within the
last decade was the development of implicit measures of attitudes (e.g., Fazio and Olson
1995; Greenwald and Band 1995). Implicit measures are based on reaction times in
response to compatibility tasks and intended to assess relatively automatic mental
associations difﬁcult to gauge with explicit measures (Hofmann et al. 2005). Research
on implicit attitudes relies on a large variety of measures, such as the Implicit Association
Test (Greenwald and Banaji 1995). affective priming (Fazio and Olson 1995), semantic
priming (Wittenbrink et a1. 1997). the go/no-go association task (Nosek and Banaji 2001),
the extrinsic affective Simon task (De Houwer. 2003), and the affect misattribution
paradigm (Payne et a1. 2005).

The best-known implicit measurement technique is the IAT, developed by
Greenwald, McGhee and Schwartz (1998). The test procedure seeks to measure implicit
attitudes by measuring their underlying automatic evaluation. According to Greenwald.
McGhee, and Schwartz (1998), a possible property of the IAT - one similar to a major

virtue of cognitive priming methods - is that it may resist masking by self-presentation

strategies. That is, the implicit association method may reveal attitudes and other
automatic associations even for subjects who prefer not to express those attitudes.

IAT procedure assesses the strength of an association between a target concept
and an attribute dimension. by measuring the latency with which a participant can employ
two response keys when each has been assigned a dual meaning (Fazio & Olson 2003).
Respondents" tasks are to categorize stimuli appearing on screen, as soon as possible, to
prevent monitoring or controlling their responses. For example, in the Greenwald,
McGhee and Schwartz (1998) study, racial attitudes were examined using the IAT.
Caucasian American participants ﬁrst were asked to categorize names as typical of
African American versus Caucasian Americans by responding to two keys, “African
American” or “Caucasian American.” Next. Caucasian American participants
categorized a variety of clearly-valenced words as pleasant or unpleasant (attribute
dimension) in combination with a name that appeared. Results indicated that Caucasian
American participants were much faster at responding when an African American name
was paired with an unpleasant word versus a pleasant one. That is, Caucasian American
participants found it much easier to associate the target concept “African American” with
the attribute unpleasant than with the attribute pleasant. Thus, it is possible that
seemingly indifferent responses or explicit attitude measures on ethnic models in
advertisements could be a controlled or adjusted response rather than based on true

feelings.

30

CHAPTER IV
HYPOTHESIS
Racial Attitude: Prejudice

Partially in response to the limitations of direct self-report attitude measures. the
self— report measures have been criticized. As a result, racial-attitude researchers long
have searched for new measures they could employ in accurate measurement of prejudice.
Crandall and Eshleman (2003) deﬁned prejudice as “a negative evaluation of a social
group or a negative evaluation of an individual that is signiﬁcantly based on the
individual’s group membership” (p. 414). Socially or politically, expressing prejudice is
marked by deep conﬂict between desire to express an emotion and, simultaneously, to
maintain values and self-concepts contrary to prejudice.

According to Crandall and Eshleman’s (2003) justiﬁcation-suppression model of
the expression and experience of prejudice. prejudice is genuine, “the ﬁrst-formed
affective component ofthe evaluation ofa group or one of its members” (p.418).

Unlike previous studies conceptualizing prejudices as byproducts of beliefs, values and
ideology, Crandall and Eshleman (2003) conceptualized “most of the personality,
attitudinal, and religious variables that correlate with prejudice not as causes but as
beliefs that serve as justiﬁers of prejudice” (p.416). This genuine prejudice is a negative
reaction usually not directly available, yet crucial and dominant. Crandall and
Eshleman (2003) suggest that all people have some degree of prejudice from their social-,
cultural-, cognitive- and developmental experiences. However. individuals may
suppress their prejudices in an effort to report socially-desirable responses because of

social norms, personal standards. beliefs and values. Therefore, in studying prejudice, the

31

important task is to examine not the existence ofprejudice, but the factors that enhance or
minimize its expressions.

Consequently, it can be assumed that Caucasian Americans have genuine
prejudices against African American models. However, they may suppress their
prejudices in an effort to report socially-desirable responses because of social norms,
personal standards, beliefs and values. According to Devine (1989), prejudice has both
an automatic and a controlled component. The factor distinguishing someone
prejudiced from someone unprejudiced is whether prejudice is controlled. The
unprejudiced are presumed to hold personal beliefs motivating them to restrain
automatically-activated cultural stereotyping. Accordingly. in examining their attitudes.
they tend to report socially-desirable responses to create a good impression. but their
responding may threaten the validity of explicit self-report measures (Hotgraves 2004).

Therefore, participants worry about their answers, if the question directly relates
to their sensitivities, especially about prejudice. A socially-desirable response is one
that is biased that reﬂects individuals’ efforts to present themselves in a positive manner
(Abrams, and Trusty 2004'). This biased self-presentation may reﬂect an unconscious
self-deception or a conscious attempt to appear positive to others (Paulhus, 1991). This
is known as social desirability, which refers to “a tendency to respond to self-report items
in a manner that makes the participant look good rather than to respond in an accurate
and truthful manner” (Holtgraves 2004, p. 161). According to Paulhus (1991), due to
political and social reforms in the United States, many Caucasian Americans are
concerned with maintaining their self-image as fair people committed to racial equality.

Hence, Caucasian Americans are strongly motivated to appear non-racist, and may not be

Lu
IO

completely honest with explicit measures regarding racial issues (Holtgraves 2004).
These conﬂicts between desirable values and undesirable feelings may lead Caucasian
Americans to be highly sensitive in evaluating issues related to minorities (e.g..

Holtgraves 2004; Paulhus 1984; Paulhus and Reid 1991).

Implicit Attitude and Prejudice

It is common that conﬂict exists between individuals’ true attitudes and their self
presentations. Individuals are unaware of their true sentiments (Greenwald and Banaji
1995) or are reluctant to reveal negative sentiments toward African Americans (Fazio et.
a1. 1995). White and Harkins (1994) found that Caucasian American participants were
strongly motivated to process messages presented by an African American source. They
interpreted this effect as the result of political and social reforms in the United States.
which made it unfashionable to display prejudice openly. In such a context, people are
strongly motivated to mask any negative racial attitudes, although they may be unaware
of their prejudices. These conﬂicts between desirable values and undesirable feelings
may lead Caucasian Americans to be highly sensitive in evaluating objects related to
African Americans (e.g., Holtgraves 2004; Paulhus 1984; Paulhus and Reid 1991).

Attitude research has found the effect of social and political reforms on people’s
self-report attitude measures. According to Tourangeau and Rasinski (I988), answering
an attitude question entails ﬁve tasks: 1) question comprehension, 2) recall. 3) judgment.
4) formatting and 5) editing the response. Among the ﬁve, participants may want to edit
their private judgments before they report it to the researcher, especially regarding

sensitive issues such as race. In the reporting stage. at least two processes play an

33

important role. The format of most attitude items requires a participant to select a
response from among a pre-established set of answer categories. Thus, a participant
must map his or her judgment onto one of the response options. Additionally, answers
may undergo an editing process in which the answer is checked for consistency with prior
answers or for social desirability. The reporting process may be a compromise between
the respondent’s judgment and what dictates consistency or social pressure.
Consequently, self-report measures have been attacked heavily because oftheir
susceptibility to artifacts such as demand characteristics. evaluation apprehension and
impression management (Greenwald et a1. 2002; Holtgraves 2004).

Implicit attitudes are manifest as actions or judgments under the control of
automatically activated evaluation, without the actor’s awareness of that causation
(Greenwald and Banaji 1995). Measuring implicit attitudes is to provide an estimate of
construct of interest without having to ask the participant directly for a verbal report. Its
major appeal is that the indirect estimate is likely free of social-desirability concern
(Fazio and Olson 2003).

In this current study. implicit attitude measure is to measure participants’ latencies
of responses to target ads like most implicit attitude measures in Psychology. The
difference from previous psychology measures is non-use of any priming factor in this
study. As Wilson, Lindsey and Schooler (2000) propose, people can have dual attitudes,
which are different evaluations of the same attitude object —- automatic (implicit) attitude
and deliberative (explicit) attitude. Simultaneously. people construct on-the-spot
attitudes on the basis of available (accessible) information at that point. In a model of

dual attitudes, Wilson, Lindsey and Schooler (2000) suggest. explicit attitudes and

34

implicit attitudes toward the same object can coexist in memory. However. their
coexisting attitudes are differently retrieved. Whereas implicit attitude is automatically
activated. explicit attitude requires more capacity and motivation.

Also, Fazio (1986) suggests that whether attitude activation occurs, thus, whether
selective perception occurs, depends on attitude accessibility. More accessible attitudes
are likelier to be activated in a speciﬁc situation. In this current study, the attitude
object is the advertisement, about which people tend to form automatic and spontaneous
attitudes. Thus, measuring participants’ latencies of response can reveal more about
how people may respond to ads encountered by them in their everyday lives. However,
the main goal in measuring implicit attitudes is to compare them with explicit attitudes
for Caucasian and African Americans to see if differences exist and if so, discuss the

many implications for advertisers.

Hypotheses
In light of the previously presented literature, a study was undertaken to examine
explicit and implicit attitudes toward ads when models of different races are presented.
Speciﬁcally, Caucasian and African American respondents were presented with ads using
a black or a white model. Explicit and implicit attitude measures were assessed as well as
response times. The following hypotheses were tested based on the goals of the study and

the literature reviewed.

H I: Caucasian Americans will report no differences in attitudes toward ads with
African American or Caucasian American models on the explicit self-report
measures.

As members of a racial majority, Caucasian Americans are less mindful of a
model’s race in the ad than members from an ethnic minority group (Appiah 2001a).
Additionally. the previous advertising studies have shown no difference in Caucasian
American’s preference on explicit measures between African American and Caucasian
American models in ads (Bush et a1. 1972', Solomon et a1. 1976; Whittler 1989; Whittler
and DiMeo 1991; Williams et a1. 1995). Consequently, it is hypothesized that Caucasian

Americans do not report any difference related to model’s ethnicity.

H2: African Americans will report more positive attitudes toward ads with
African American models over ads with Caucasian American models on the
explicit self-report measures.

According to Accommodation theory. members of ethnicity minority groups
would react positively to the use of ethnic cues in ads by attributing their use to the
advertiser’s sensitivity to their own ethnicity (Koslow 1994). Also, previous advertising
studies have shown African Americans to prefer ethnic cues in ads such as same race
models over Caucasian models on explicit attitude measures (Aaker et a1. 2000; Briley et
al. 2007; Deshpande and Stayman 1994; Dimofte et a1. 2003; Forehand and Deshpande
2001; Whittler 1989; Whittler and DiMeo I991; Wooten 1995). Therefore, it is
hypothesized that African Americans report more positive attitude toward the ad
containing African American models than Caucasian American models.

A socially-desirable response is one that is biased and reﬂects individuals’ efforts
to present themselves in a positive manner (Abrams and Trusty 2004). The biased self-
presentation may reﬂect an unconscious self-deception or a conscious attempt to appear

positive to others (Paulhus 1984). But, due to political and social reforms in the United

36

States, most Caucasian Americans are concerned with maintaining their self-image as fair
people committed to racial equality. Hence. Caucasian Americans are strongly
motivated to appear non-racist. and may tend to not be completely honest with explicit
measures regarding racial issues (Holtgraves 2004). These conﬂicts between desirable
values and undesirable feelings may lead Caucasian Americans to be highly sensitive in
evaluating issues related to minorities (e.g.. Holtgraves 2004: Paulhus 1984; Paulhus and
Reid 1991). Assuming that all people have some degree of prejudice which is
considered as an automatic and controlled component (Devine 1989), the following three

hypotheses are proposed.

H3: Caucasian Americans’ reaction times will be longer when viewing ads with
African American models than when viewing ads with Caucasian American
models.

Caucasian Americans are strongly motivated not to express their negative racial
attitudes, although they may be unaware of their prejudices. These conﬂicts between
desirable values and undesirable feelings may lead Caucasian Americans to be highly
sensitive in evaluating objects related to African Americans (e. g., Holtgraves 2004;
Paulhus 1984; Paulus and Reid 1991). Consequently, research in psychology suggests
that Caucasians may monitor their answers or spend time to assess the perception of their
answers in relation to race based issues and this monitoring would not similarly occur
when Caucasian models were presented in ads (Petty et a1. 1999; Brunel, Tietje and
Greenwald 2004:. White and Harkins 1994). It is hypothesized that when Caucasian

Americans report their attitude toward the models. they tend to spend more time on the

ads containing African American models than Caucasian American models.

37

H4: African Americans’ reaction times will be similar when viewing ads with

either African American models or Caucasian American models.

African Americans respondents will not experience impression management
issues with respect to monitoring people’s perceptions about their racial beliefs
(Holtgraves 2004). When they view the ads with African American models. the ads
elicit self-awareness due to the model’s distinctive trait. race (Acker et. a1. 2000;
Despande and Stayman 1994: Forehand et. al. 2002). Therefore. this ethnic salience
helps African American respondents report their attitudes fast. At the same time. it is
Caucasian American models are the norm in mainstream magazine ads. there is no need
for spending more time to report their attitude toward the ad (Kahneman and Miller 1996;
Pratto et. a1. 2007). Consequently, it is hypothesized that when viewing ad with either
Caucasian or African American model. African American respondents do not show any

difference in reaction time.

11;: There will be a signiﬁcant difference between Caucasian American’s

implicit and explicit attitudes toward the ad.

According to the literature on dual and implicit attitude measures, explicit
attitudes in relation to race base issues require more deliberative thought processing by
Caucasian respondents either to help reduce unwarranted racial prejudices or as an
attempt to provide socially desirable responses in comparison to implicit attitude
measures (Brunel, Tietje and Greenwald 2004; Fazio and Olson 2003; Greenwald,

McGhee and Schwartz 1998: Wilson. Lindsey and Schooler 2000). As a result. it is

hypothesized that Caucasian American respondents would report less positive attitudes in

implicit than explicit measures due to time pressure.

H6: There will be no signiﬁcant difference between African American’s implicit

and explicit attitudes toward the ad.

When it comes to racial issues, members of a racial minority group are typically
less mindful of monitoring responses in order to produce socially desirable opinions than
are their Caucasian counterparts (Abrams and Trusty 2004; Holtgraves 2004; Paulhus
1984). It is hypothesized that African Americans do not show any different attitude in

either implicit or explicit measures. They would maintain the similar attitude.

39

CHAPTER V

METHODOLOGY

This study is designed to examine two attitudes toward the ads: Caucasian
American participants’ attitudes toward African American models in the ads and African
American participants’ attitudes toward Caucasian American models in the ads. To
measure their dual attitudes. two measures will be used: implicit and explicit attitude
measures (i.e., response latency and semantic differential scale). To examine the sole
effect of model’s race on participant’ attitudes. all other possible antecedents of attitude
toward the ad from the ads (i.e.. backgrounds, products. and brands) deliberately were

excluded in this current study.

Stimuli Development

Products and services presented in this current study were chosen according to the
following three criteria: F irst, ethnic products (i.e., hair or skin care) were carefully
excluded to control relevance. Second, different types of products and services were
chosen randomly across different dimensions in the FCB GridI in order to disperse
feelings of involvement and affect-cognition effects on their attitudes toward the ad.
Finally, products or services were chosen that were relevant to participants. Based on
these criteria, 15 products were selected for the college student sample in this present

study (Figure 12).

 

' The FCB Grid was created by Richard Vaughn. With this model. messages are categorized by "thinking"
and "feeling", "low" and "high": Low Think (practicality. pragmatism) ; High Think; Low Feel (sensuality,
pleasure); High Feel (product as extension of self) (Ratchford 1987).

The products are placed on the basis of Ratchford’s (1987) major study grid (p. 31). Besides, other

40

Figure l FCB Grid - 15 Chosen Products

Think 5 Feel
A . : Insurance 5
’ rmy 3 .
High 0; Family car
Involvement . 5 9
D l' ; Cosmetic*
e rvery : O Facial soap
servrce* :
IO....I.CI........‘C..".I.-I-.I.E..II...III..II.. ..... .UIICIICI
. Jeans . *
Body Toothpaste 0 Hotel
I..otron* ; Alcohol .
w E .
L0 0 Shampoo : beveragqrast food
0 i
Disposal
razor

To prevent participant from noticing this study’s true intention, 40 ﬁller ads and
20 target ads were used in the experiment. Four criteria were used to create target ads in
this study. First, the target advertisements needed to be selected from major magazines
that both African American and Caucasian American consumers read. Second, target
advertisements needed to include either African American or Caucasian American models.
Third, target advertisements must contain one of the selected product types based on the
criteria described above. Finally, to control for brand effects, a diverse set of brands
were used in the ﬁller and target ads. Since about 51 brands were used in this study (18
in target ads; 33 in ﬁller ads), speciﬁc brand effects were not expected. Based on these

criteria, 20 potential advertisements were selected from advertisements in popular

 

products that marked by asterisk are placed on the basis on their dimensional features that Ratchford (1987)
describes.

41

American magazines such as People, GQ and Glamour.

Pretest

To select the ﬁnal set of test stimuli, the 20 potential advertisements were
pretested. The pretest involved 12 undergraduate students presented with 20 ads and
asked to respond to a survey. Previous research suggests that a model’s characteristics
(familiarity, attractiveness, age. and appropriateness) can impact viewers’ attitudes
toward the ads. Since the 20 ads were to be manipulated with African American and
Caucasian American models. it was required that each of the potential model’s
characteristics should be similar. Thus. the models’ characteristics were compared
between African American and Caucasian American models. Familiarity was measured
on a 7-point scale ranging from “unfamiliar” to “familiar.” Attractiveness was scored
on ﬁve 7-point scales (attractive-unattractive, classy-not classy, beautiful-ugly, elegant-
plain, and sexy-not sexy) (Ohanian 1990). Scores were averaged and used for future
analysis. Also. models’ ages were estimated, and their appropriateness to the advertised
products measured on a 7-point scale. Finally, participants’ previous exposure to ads
was measured on a 7-point scale. and they were asked to estimate models’ race. to
examine whether viewers could identify the races correctly (See Appendix A).

T-test suggested that participants did not perceive attractiveness differently
between African American (m=4.42) and Caucasian American (m=4.41) models in each
of the ads (t(320) =1.628, ns). The result showed that their familiarity with African
American (m=2.04) and Caucasian American (m=2.20) models were not signiﬁcantly

different (1(320) =.868. us). Also. the result showed that they were not familiar with

 

models in the ads. Participants did not show any signiﬁcant difference between African
American (m=3.42) and Caucasian American (m=3.18) model’s cuteness (t(320)= .622,
ns). There was no signiﬁcant difference in perceiving model’s age between African
American (m=25.29) and Caucasian American (m=25.84) models (1(320) =.476. ns).
The result showed that participants felt the advertised products were appropriate to
African American (m=5.59) and Caucasian American (m=5.34), and its difference was
not signiﬁcant (I(320) =2.68, ns). The result showed they were rarely exposed to ads
with African American (m=1.77) and Caucasian American (m=l .78) models (1(320)
=.068, ns). Finally, there was no incorrect answer to indicate models’ race in the ads.
Overall, African American and Caucasian American models in the 20 potential ads were
similarly perceived by participants regarding familiarity. attractiveness. cuteness, age,
product appropriateness. and previous exposure.

Accordingly, 80 target advertisements were created with identical layouts. but
differently manipulated depending on experimental conditions (product only; product and
background only; product, background, and African American model; product,

background and Caucasian American model).

Experimental Design
The study uses a 2 (respondent’s race: African American and Caucasian
American) x 2 (model’s race: African American and Caucasian American model) x 3
(advertising components: product only vs. product and background vs. product,
background and African American model vs. product, background and Caucasian

American model) x 2 (attitude measures: implicit vs. explicit measure) factorial design

was used to examine attitude toward the ad. Models’ race and advertising components
are between subject variables, and attitude measures are within subject variables. Also,
to examine advertising component effects, the ﬁfth control group was added.
Participants in the control group randomly were exposed to different ads from the four
experimental groups’ ads (Table 1). To control the order effect of ad exposure. ten ads
were grouped and randomly chosen for implicit and explicit conditions. There was no
signiﬁcant difference between different ad sets (F=.296. p>0.10).

Table 1 Experimental and Control Groups

 

Group Name Subjects Advertising Components
Group 1 20(W)/15(B) Product
Group 2 15(W)/18(B) Product + Background
Group 3 15(W)/17(B) Product + Background + African American model
Group 4 15(W)/15(B) Product + Background + Caucasian American model
Group 5 16(W)/15(B) Randomly chosen 5 ads from 4 conditions

 

 

 

 

 

 

 

 

 

A control group study uses a control group to compare to experimental groups in
testing a causal hypothesis. In this current study. only the effect of models’ race was
examined. Since participants in each experimental condition were designed to expose
speciﬁc type of ads, it was required to examine the effect of speciﬁc ad exposure. For
example, a Caucasian American subject in a product, background and African American
condition may feel or seem unnatural because he or she continuously evaluates ads
containing African American models. In the real world, although there were ﬁller ads
among target ads, it may not seem natural, considering that advertisers typically use more
Caucasian American models than African American models. Thus, adding the control
group involved evaluating all different types of ads to remove the effect of speciﬁc ad

exposure. The speciﬁc ad exposures not actually major causal factors of participants’

44

 

attitudes toward ads. logic would dictate that that this should manifest itself more

signiﬁcantly in the experimental than in the control group.

Participants
Eighty-one Caucasian American and eighty African American undergraduate
students were solicited from classes campus wide, across a major mid-western university.
The participants were randomly assigned to view ads with either white or black models.
The presentation order for all treatment ads was randomized. Attitudes were measured
within-subjects using both explicit and implicit methods. Speciﬁc demographic

infomiation will be discussed in detail.

Procedures

Participants in the experiment were assigned randomly to one of the experimental
or control groups before entering the computer laboratory. Upon entering the lab.
participants were seated at a computer and informed that they would be asked to evaluate
advertisements. The sequence ofthe ﬁve-part experimental session is presented in Table
2.

Participants then were introduced to a series of advertisements via one of the ﬁve
experimental conditions. In the product-only condition, participants were asked to 20
target advertisements containing only products. In the product and background
condition, participants were exposed to 20 target advertisements containing products and
backgrounds. The advertisements in this condition contain all advertising components

except a human model. In the product, background and African American model

condition, participants were asked to evaluate 20 target advertisements containing all
advertising components with African American model/s. Likewise. in the product.
background and Caucasian American model condition. participants were exposed to 20
target advertisements containing all advertising components with Caucasian American
model/s. Finally, in the control group, participants were asked to evaluate
advertisements randomly chosen from four experimental conditions. Order of ad
presentation was randomized across conditions and sessions.

Table 2 Sequence of Experiment

 

 

 

 

 

 

 

 

 

 

Explicit Attitude Measures
Session 1 j 20 ads (10 Filler ads + 10 Target ads)

Implicit Attitude Measures
Session 2 Exercise 20 Filler ads
Session 3 Baseline Alphabet “”A
Session 4 Target 20 ads (10 Filler ads and [0 Target ads)

Paper Pencil Survey

Session 5 1 Ethnic. socioeconomic and demographic information

 

 

In the ﬁrst session of the study, using the computer, participants ﬁrst viewed 20
different ads and responded to traditional explicit attitude measures after each exposure.
Ten ﬁller and ten targeted ads were presented. Participants were instructed to press a
number key to express their evaluations of ads, ranging from 1 to 5 on the computer
keyboard. Also, they were instructed to have no time pressure to report their
evaluations of ads. Thus. it allowed participants to spend enough time prior, to report.
Explicit attitude measure was done using four 5-point differential scales (bad—good,
unfavorable-favorable, dislike-like, and poor quality-good quality). These items have
been used in past research on attitude toward the ad (MacKenzie, Lutz and Belch 1986).
Participants’ responses. along with latency times, were recorded to be compared with

implicit measures.

46

The implicit measure comprised four sessions. During measuring implicit
attitudes, participants were instructed to press a key “2” labeled favorable or “’/’ labeled
unfavorable as quickly as possible to indicate their evaluation of the ads. The second
session was the exercise session. purpose of which was to cause participants to maximize
the speed and accuracy oftheir responses. Twenty ﬁller ads were presented. In the
session. participants’ responses and latencies of response were recorded. but they were
not used in further data analysis. The third session, whose purpose was to obtain
baseline data, involved the presentation of a single alphabet character (A). In this
session, the participants were informed that a key “2” or “/” has no meaning, and they
merely pressed one of the designated keys as quickly as possible. Thus, their latencies
of responses were only recorded, and the average latency for this session served as
participants’ baseline latency for their responses to target ads. The fourth session was
designed to measure implicit attitudes toward the target ads. Ten ﬁller and 10 targeted
ads were presented. Participants’ responses were recorded. along with the latencies of
responses to the nearest millisecond.

In the ﬁfth session, participants completed a survey with attitude toward ads in
general, ethnic identiﬁcation, motivation to control prejudiced reaction, socioeconomic
information and demographic. Participants were asked to indicate their opinions about
advertising in general. Attitude toward the ad in general measure was constructed by
four items: “Advertising in general is useful.” “Advertising in general is truthful,”
“Advertising in general is believable,” and “Advertising in general is good.” These
items were adopted from previous studies (Karande 2005; Muehling 1987), and were

scored on 7-point scales ranging from “strongly disagree” to “strongly agree” (a=.77).

47

 

Ethnic identiﬁcation was measured by ﬁve items ((1:91): “I think a lot about how my life
will be affected by my ethnic group membership,” “I have a lot of pride in my ethnic
group and its accomplishments.” “I have a. strong sense of belonging to my own ethnic
group,” “I feel a strong attachment towards my own ethnic group,” and “I feel good about
my cultural or ethnic background” (Phinney 1992). These items were scored on 7-point
scales ranging from “strongly disagree” to "strongly agree.” Dunton and Fazio’s (1997)
Motivation to Control Prejudiced Reaction was used (See Appendix B) (a=.72).
Participants’ eco-social status (i.e.. family income, parents’ education levels, andjob title)
was measured. Finally, the participants were asked about their race, strength of ethnic

identiﬁcation (Karande 2005), age, academic year, and gender.

Dependent Variables

1n the ﬁrst session of the study. explicit attitudes were measured using four 5-
point differential scales (bad-good. unfavorable-favorable. dislike-like. and poor quality-
good quality). These items have been used in past research on attitude toward the ad
studies (MacKenzie, Lutz and Belch 1986). Respondents were told to answer the
questions by pressing a number key to express their evaluations of ads and the scale was
found to be reliable (a=.97). For this part of the experiment, there were no time
constraints for completing the task placed on respondents.

Implicit attitudes were measured by asking participants to make dichotomous
judgments in response to treatment ads as either favorable or unfavorable as fast as
possible while recording their response times in milliseconds. After cleaning the

responses for extreme cases. the average response latency was computed. After

48

cleaning the responses in the baseline session for extreme cases, a baseline response time
was also computed for each participant. Each participant’s baseline was then subtracted
from his or her average response latency for the treatment ads to reduce individual
variation in natural response times.

Finally, participants completed the last portion of the questionnaire assessing

demographic information.

49

CHAPTER IV

DATA ANALYSIS

Descriptive Statistics

Means. standard deviations. and correlations for all variables included in the

analysis. as well as reliability statistics for the multiple item indexes are shown in Table 3.

Generally. respondents have positive attitudes toward the ad. Additionally, their family

income is between $75,000 and $99,000 on average, and they consider themselves as the

lower middle class.

Finally, they tend to have strong ethnic identiﬁcation and ethnic

issues are highly related to their family income and status.

Table 3 Descriptive Statistics, Correlations, and Scale Reliabilities

 

 

Variables M SD 1 2 3 4 5 6 7 8
1.General
Attitude toward 4.80 1.00 (.97)
the ad
5" Fam'l—V 8.07 2.09 0.40 -
income
3. Family status 2.87 1.21 -,l7* ,45** -
4. Gender 1.67 0.47 .12 .06 -.02 -
5.Age 21.2 1.72 .10 -.01 .08 -.17‘ -
6' Ethnic 5 31 1 43 - 09 27** 27** 16’ 03 (91)
identiﬁcation ’ ’ ' " ' ‘ ' ’
7 Motivation to . .
' 7 *7: _ H: *4: _ _ *7:
avoid prejudice 4.5- 0.78 .22 .14 .23 .24 .15 .21 (.72)
8' cand'dm 6.05 1.45 -.02 -.22** .23** .215W .07 .63“ -.08 -

ethnicity

 

Note. Total respondents were examined. 11 = 161. *. Correlation is signiﬁcant at the .05 level (two—tailed).

**. Correlation is signiﬁcant at the 0.01 level (two-tailed). Cronbach’s alphas for multiple-item indexes are
shown on the diagonal.

. Higher scores indicate more positive attitude toward the ad.

1

2. Higher scores indicate higher family income.

3. l = the upper class; 2 = the upper middle class; 3 a: the lower middle class.
4. 1 = male; 2 = female.

\lO‘

. Higher scores indicate stronger identiﬁcation.
. Higher scores indicate stronger motivation

8. “How strongly do you identify yourselfas a member ofthe ethnic group you selected above? Please rate
your feeling.” 1 = strongly disagree; 7 = strongly agree.

50

Caucasian American and African American respondents’ mans, standard
deviations, and correlations for all variables are shown in Table 4 and Table 5 separately.
Caucasian American respondents showed a lower average score on Phinney’s (1992)
ethnic identity index than did African Americans. For Caucasian Americans, ethnicity
tends to not be an important part of their identity (Waters, 1990). This is evident in the
fact that many Caucasian Americans do not think of themselves as belonging to an ethnic
group (Alba. 1985). Additionally. Caucasian American respondents showed higher
scores in family income and status than did African Americans.

Table 4 Descriptive Statistics and Correlations for Caucasian American

 

 

Variables M SD 1 2 3 4 5 6 7 8
1. General
Attitude toward 5.1 1 .82 -
the ad
.2' ”my 8.59 1.57 .07 -
meome
3. Family status 2.36 .88 -.06 -_33** -
4. Gender 1.65 .47 .09 .01 .03 -
5.Age 21.2 1.33 .13 -.04 .03 -.13 -
.6' E’l‘mc . 4.44 1.26 .06 -.14 -.6 .17 -03 -
Identrﬁcatron
7. Motivation to 7 . .

. . 7 . . _ -. -. 5 4.. -. -. -

avoid prejudice 4 85 77 0 06 0 .36 10 06
8. Candidate 7 , *4 4*
ethnicity 5.47 1.7- .18 -.1._ .05 .30 .04 .58 .13 -

 

Note. Total respondents were examined. n = 81. *. Correlation is signiﬁcant at the .05 level (two-tailed).

**. Correlation is signiﬁcant at the 0.01 level (two-tailed). Cronbach’s alphas for multiple-item indexes are

shown on the diagonal.

1. Higher scores indicate more positive attitude toward the ad.

2. Higher scores indicate higher family income.

3. 1 = the upper class; 2 = the upper middle class; 3 = the lower middle class.

4. 1 = male; 2 = female.

6. Higher scores indicate stronger identiﬁcation.

7. Higher scores indicate stronger motivation

8. “How strongly do you identify yourself as a member of the ethnic group you selected above? Please rate
your feeling.” 1 = strongly disagree; 7 = strongly agree.

On the other hand. African Americans have stronger ethnic identiﬁcation and

candidate ethnicity than Caucasian Americans. Phinney and Alipuria (1990) showed

51

ethnic identity to be a more salient part of the self for ethnic minority groups than for
most Caucasian Americans. Deaux (1992) stated that a primary reason why ethnicity is
salient for ethnic minority groups is that their group membership is evident. A family
income and candidate ethnicity is negatively correlated. It indicates that as lower
income respondent’s family income, he or she has more strong ethnic identiﬁcation and
candidate ethnicity. According to Sherry (2006), low-income African Americans have
the strongest relation with ethnic identity. They are highly correlated with a higher
engagement in ethnic behaviors and more exploration into African American history and
tradition. This strong ethnic identiﬁcation caused by the experience of racism over
one’s lifetime. Thus, it was correlated negatively with one’s feelings of closeness
toward people from other ethnic groups.

Table 5 Descriptive Statistics and Correlations for African American

 

 

Variables M SD 1 2 3 4 5 6 7 8
1. General
Attitude toward 4.43 1.06 -
the ad
.2: Fm” 7.53 2.76 -.09 -
Income
3. Family status 3.39 1.28 -.02 -,43** -
4. Gender 1.7 0.46 .21 .11 -.10 -
5. Age 2‘2)" 2.04 -.33 .00 -. 12 .20 -
6' 5mm 6 20 0 97 24* - 20 07 18 - 91 -

identiﬁcation

7. Motivation to
avoid prejudice
8. Candidate
ethnicity

4.22 0.67 .17 .13 -.10 .19 22* .19 -
6.62 0.80 .06 -.25* .12 .21 -.15 .69** -.04 -

Note. Total respondents were examined, 11 = 80. *. Correlation is signiﬁcant at the .05 level (two-tailed).

**. Correlation is signiﬁcant at the 0.01 level (two—tailed). Cronbach’s alphas for multiple-item indexes are
shown on the diagonal.

1. Higher scores indicate more positive attitude toward the ad.

2. Higher scores indicate higher family income.

3. 1 = the upper class; 2 = the upper middle class; 3 = the lower middle class.

4. 1 = male; 2 = female.

6. Higher scores indicate stronger identification.

7. Higher scores indicate stronger motivation

8. “How strongly do you identify yourself as a member of the ethnic group you selected above? Please rate

52

your feeling.” 1 = strongly disagree; 7 = strongly agree.

In summary, the motivation to avoid prejudiced reactions was correlated with
gender. One-way ANOVA showed that female Caucasian American respondents (m =
5.03) have signiﬁcantly more motivation to avoid prejudice than male respondents (m =
4.45; F (1, 79) = 1.08. p = .001). However. African American respondent’s motivation

to avoid prejudice is not significantly different.

Purification of Response Latencies

Response latencies are notoriously impure, and in fact, are often characterized by
a positive skew and the prevalence of outliers (Ratcliff 1993). A variety of procedures
for dealing with distribution problems have been explored, but most common is the
transformation ofthe data to approximate a normal distribution and the use of cutoff
scores to limit the effects of outliers (Brunel et a1. 2004; Fazio et a1. 1995; Greenwald et
a1. 1998). Outlying scores typically indicate responses initiated prior to perceiving the
stimulus or momentary inattention (Greenwald et al. 1998). Therefore, short and long
outliners were identiﬁed and eliminated because they lack theoretical interest, distort
means, and inﬂate variance.

A variety of procedures for dealing with distribution problems have been explored.
Among notable conclusions of recent work is that the use of cutoff values to eliminate
outliers is difﬁcult to justify in most circumstances. Ratcliff(1993) recommended using
cutoffs that are the same across conditions and that eliminate a small percentage of
response. This is the most frequently used way of analyzing response time data, which
other studies used (Brunel et a1. 2004; F azio et a1. 1995; Greenwald et a1. 1998). Since
Van Zandt (2002) provides a compelling demonstration that the presence of even a single

53

outlier severely can bias the mean and the variance for an experimental condition. the
best solution to this dilemma is to retain all data, but to use cutoff analysis.

Following the procedure of Greenwald, McGhee and Schwartz (1998)
recommended for the application of the Implicit Association Test used in the current
study, response latencies faster than 300 ms and slower than 3.000 ms were eliminated to
reduce error. This resulted in the elimination of 5.15% ofthe scores for the
experimental treatment conditions (session 4). To correct the positive skew that
naturally characterizes response latency data. an inverse transformation of data was
conducted before analysis. The means reported in this study have been retransformed,
however, back into the millisecond metric for conceptual clarity.

For baseline measures taken in session 3, the ﬁrst two response latencies were
dropped because of the learning curve associated with the procedure (Greenwald.
McGhee and Schwartz 1998). Further. response latencies of faster than 150 ms and
slower than 1.500 ms were eliminated as outliers. A different criteria was used for the
baseline cut-off points. due to the fact that participantsjust responded to the presence of
the letter “A” on screen and did not need to make an evaluation prior to their response.
The cut-off criteria are in line with recommendations used by Wittenbrink, Judd, and Park
(1997) for a similar Lexical-decision Task. As a result of this puriﬁcation, 10.87% of
response latencies were eliminated in developing the baseline measure. The higher
error rates can be attributed to learning the procedure and participants’ anticipation about
the random presentation of the letter on the screen. Participants tried to guess when it
would appear. The positive skew ofthe baseline data was transformed in the same

manner as described above and retransformed back into the millisecond metric for

54

conceptual clarity in this paper.

F inally, for explicit measures in the ﬁrst session. response latencies from the value
faster than 300 ms were only recoded as missing values. Since participants were
instructed to spend as much time as they could. there was no cut-off point for slower
response. One hundred and ﬁve individual latencies were recoded (2.52%). Log and
inverse transformation data still showed evidence of considerable positive skew. The
means reported in this study have been retransformed, however, back into the millisecond
metric. Four response latencies for each ad were averaged and used for later analyses.

After data puriﬁcation. a total of 104 (26 per each condition) response latencies
were analyzed. Figure 2 displays mean latencies for the four conditions.

Figure 2 Overall Response Latencies for Conditions
1550
1500
1450
1400
1350
1300

1250
1200

I Response Latency

 

Conditionl Condition2 Conditi0n3 Condition4

As expected. respondents in conditions 3 (product, background and African
American model) and 4 (product, background and Caucasian American model) were
likelier to spend more time than were those in conditions 1 (product only) and 2 (product
and background). However, there were no signiﬁcant differences among conditions, F
(3, 103) = 1.591, p =.196. Then, to examine sole effect of model in the ads. response

latencies in condition 2 (product and background) were subtracted from response

latencies in conditions 3 (product, background and African American model) and 4
(product, background and Caucasian American model). Resultantly, a total of 52

response latencies were used for further analyses.

Manipulation Check for Measures
The degree of association between implicit and explicit measures was examined
in two ways: attitude and latency response (Table 6). First, implicit and explicit attitude
toward the ad was examined. When respondents report implicit attitude (m = 5.19).
they are more positive than they do explicit attitude (m=3.38). Also, these two attitudes
are negatively correlated (r = -.45, p<.001). Second. latency response of implicit and
explicit attitudes are highly related (r = .63. p<.001). It shows that respondents tend to

spend a reasonable amount ofthe time on processing the information in the ad.

Table 6 Correlations and Descriptive Statistics for Implicit and Explicit Measures

 

 

Variables M SD 1 2 3 4
1. Implicit Attitude
toward the Ad 5'19 2'3] "
2. Explicit Attitude 4 H
toward the Ade 3'3 8 0'51 "45 '
3. Lantency Response 46 13 549 65 07 09 _
for Implicit Attitude ' ' ' -
4. Lantency Response H
for Explicit Attitude 2059.8 54653 ~05 ‘21 -63

 

Note. Total respondents were examined. n = 52. **. Correlation is significant at the 0.01 level (two-tailed).

Previous research found that Caucasian Americans are unaware of their true
sentiments or reluctant to reveal negative sentiments toward African Americans (F azio et.
a1. 1995; Greenwald and Banaji 1995). In this study, implicit measures are employed to
measure Caucasian American’s true attitude. It is intended that respondents could not
edit their private judgments before they report it to the researcher, especially regarding

56

sensitive issues such as model race in ads. Implicit measure triggers respondents not to
check their answers for social desirability, and it yields them to report more positive
attitude than explicit measure. Additionally, the measures employed in this study are

successfully manipulated.

Hypotheses Testing
The ﬁrst two hypotheses were related to self-reported (explicit) attitudes toward
ads with models of different races. It was hypothesized that Caucasian Americans would
report no differences. while African Americans would show a preference for African
American models. A two-way ANOVA was conducted to test these hypotheses and a
summary of results are presented in Table 7.

Table 7 Participant Race by Model’s Race on Explicit Attitudes

 

 

 

 

 

 

 

Source SS df MS F P
Between Subjects .540 3 .180
Respondent’s Race .066 1 .066 .247 .621
Models Race .339 1 .339 1.273 .265
Respondent x Model .135 1 .135 .507 .480
Error 12.786 48 .266
Total 605.639 52

 

No signiﬁcant main effects or interaction effects were found from the analysis. Main
effect results revealed that respondent’s and model’s race were not signiﬁcantly different.
F (1, 48) = .247, p = .621, partial n 2 = .005 and F (1, 48) = .339, p = .265, partial n2
= .026 respectively. Also. no interaction effect was found. F (1. 48) = .507, p = .480,
partial r1 2 = .010.

Caucasian Americans and African Americans reported similar attitudes toward

both Caucasian and African American models, p>.05. This ﬁnding is in line with

57

expectations for hypothesis one. however fails to support hypothesis two. In contrast to
previous literature, African Americans did not report a preference for ads with African
American models.

Hypotheses three and four were related to the response latencies associated with
the formation of (implicit) attitudes toward ads with models of different races. It was
hypothesized that Caucasian Americans would take more time in evaluating ads with
African American models than they would in evaluating ads with Caucasian American
models. In contrast. it was argued that African Americans’ reaction times would be
similar when viewing ads with either African American models or Caucasian American
models. A two-way ANOVA was conducted to test these hypotheses and a summary of
results are presented in Table 8

Table 8 Participant Race by Model’s Race on Response Latencies

 

 

 

 

 

 

 

Source SS (If MS F P
Between Subjects 274302.272 3 91434.091
Respondent’s Race 2276.643 1 2276.643 .007 .933
Model’s Race 62728.097 1 62728.097 .199 .658
Respondent x Model 209297.533 1 209297.533 .664 .419
Error 15133667633 48 315284.742
Total 15518601.121 52

 

No signiﬁcant main effects or interaction effects were found from the analysis.
Main effect results revealed that respondent’s and model’s race were not signiﬁcantly
different, F (l, 48) = .007, p = .933, partial n2 =.000 and F (1, 48) = .199, p = .658, partial
n2 = .004 respectively. Also, no interaction effect was found, F (l, 48) = .664, p = .419,
partial n2 = .014. Both Caucasian Americans and African Americans’ response latencies
were similar across conditions. p>.05. This ﬁnding is in line with expectations for

hypothesis four. however fails to support hypothesis three. Interestingly. both Caucasians

58

and African Americans were faster to report attitudes toward a model of their respective
races over models of a different race. however not signiﬁcantly. Caucasians responded
faster to a Caucasian model (m = 45.43 ms) than toward an African American model (m =
68.22 ms). Similarly. African Americans responded faster to an African American
model (m = 10.80 ms) than to a Caucasian model (m = 150.92 ms).

Finally, hypotheses ﬁve and six examined the relationship between the explicit
and implicit responses provided by respondents. It was hypothesized that there would
be a signiﬁcant difference between Caucasian American’s explicit and implicit attitudes
toward the ad, but that we would see no such difference for African Americans. To test
these hypotheses. it was ﬁrst necessary to transform the data so it could be compared
across response formats. Both response latencies and explicit attitude measures were
converted into z-scores before running a one within (response format). two between
subjects (respondent/model race) ANOVA. A summary of results are presented in Table
9.

Table 9 Participant Race by Model’s Race by Attitude Measure (Implicit/Explicit)

 

 

 

 

 

 

 

 

 

 

 

Source SS df MS F P
Between Subjects 9.752 7 1.393
Respondent’s Race .649 l .649 .676 .413
Model’s Race 2.554 1 2.554 2.658 .106
Measure .000 l .000 .000 l .000
Respondent x Model .1 15 1 .115 .1 19 .731
Resfondem x .649 1 .649 .676 .413
Measure
<.05
Model x Measure 5.326 1 5.326 5.543 0
ReSpondem X MOde' .460 1 .460 .479 .491
x Measure
Error 92.248 96 .961
Total 1 02.000 104

 

59

No signiﬁcant main effects or interaction effects were found from the analysis.
Main effect results revealed that respondent’s and model’s race and measure were not
signiﬁcantly different, F (1, 96) = .676, p = .413, partial n2 = .007, F (1 , 96) = 2.658, p
= .106. partial r12 = .027. and F (1. 96) = .000. p = 1.000. partial n2 = .000 respectively.
Also, no signiﬁcant interaction effect was found for respondent’s and model’s race, and
respondent’s race and measure. F (l. 96) = .115. p = .731. partial n2 = .001 and F (I, 96)
= .649, p = .413. partial r12 = .007.

Both Caucasian and African Americans’ response latency z-scores were
distributed similarly to that of the explicit attitude measure z-scores across model
conditions, p>.05. This ﬁnding is in line with expectations for hypothesis six, however
fails to support hypothesis ﬁve. The ﬁndings indicate that social desirability did not
seem to effect the formation of attitudes towards models of different races in

advertisements.

Analysis of Control Group (Condition 5)
The following analyses examined whether the manipulation of ad differentially
affected respondents in different groups. These analyses reveal that the effect of

manipulated ad stimuli did not affect on respondents’ attitude toward the ads.

Puriﬁcation of Response Latencies for Condition 5

The same procedure applied for data reduction on response latencies in condition
5. Error rates for explicit. baseline and implicit are 1.29%. 16.12% and 12.25%
respectively. Thus. respectively 16, 40 and 38 individual response latencies were

recorded as missing value.

60

5

After data deduction. a total 31 respondents’ 23.) response latencies were analyzed.
Figure 3 displays mean latencies for the four conditions. As expected. respondents in
conditions 3 (product, background and African American model) and 4 (product,
background and Caucasian American model) were likelier to spend more time than those
in conditions 1 (product only) and 2 (product and background). However, there were no
signiﬁcant differences among conditions, F (3. 224) = .650. p =.583. Then, to examine
sole effect of model in the ads, response latencies in condition 2 (product and
background) were subtracted from response latencies in conditions 3 (product.
background and African American model) and 4 (product. background and Caucasian
American model). Resultantly, total 62 response latencies were used for further
analyses.

Figure 3 Overall Response Latencies for Conditions

 

6000
5000

4000
3000 I Response Latency
2000
1000
0 , . .

Conditionl Condition2 Condition3 Condition4

 

 

 

 

 

 

Hypotheses Testing for Condition 5

A two-way ANOVA was conducted for explicit attitudes; a summary of results are
presented in Table 10. Main effect results revealed that respondent’s race was not
signiﬁcantly different, F (1, 58) : .228, p = .635. partial n2 = .004. However, model’s

race showed signiﬁcant effect on explicit attitude measure. P (l. 58) = .6.021, p <05,

61

partial n2 = .094. Respondents have more positive attitude toward African American
models (m = 3.62) than toward Caucasian American models (m = 3.14) in the ads. No
signiﬁcant interaction effect was found. F (l. 58) = 1.198. p = .278. partial n2 = .020.

Table 10 Participant Race by Model’s Race on Explicit Attitudes in Condition 5

 

 

 

 

 

 

 

Source SS 4f MS F P ES
Between Subjects 4.433 3 1.478
Respondent’s Race .139 1 .139 .228 .635 .004
Model’s Race 3.666 1 3.666 6.021 <.050 .094
Respondent x Model .729 1 .729 1.198 .278 .020
Within Subjects 35.314 58 .609
Total 747.429 62

 

A two-way ANOVA was conducted for implicit attitude’s response latencies; a
summary of results are presented in Table 1 1. Main effect results revealed that
respondent’s and model’s race were not signiﬁcantly different. F (1. 58) = 3.162. p = .081.
partial n2 = .053 and F (l, 58) = .829. p = .367. partial n2 = .015 respectively. Also. no
signiﬁcant interaction effect was found, F (1, 58) = 1.930, p = .170, partial n2 = .033.

Table 11 Participant Race by Model’s Race on Response Latencies in Condition 5

 

 

 

 

 

 

 

Source SS df MS F P ES
Between Subjects 1442593842 3 480864.614
Respondent’s Race 779322.315 l 779322.315 3.162 .081 .053
Model’s Race 204270.204 1 204270.204 .829 .367 .015
Respondent x Model 475806.720 1 475806.720 1.930 .170 .033
Within Subjects 13803801323 56 246496.452
Total 16518308232 60

 

A three-way ANOVA was conducted for response latencies; a summary of results
are presented in Table 12. Response latencies were converted into z-score. Main
effect results revealed that respondent’s and model’s race and measure were not

signiﬁcantly different, F (1, 114) = 1.479. p = .226, partial n2 = .013, F (1, 114) = 6.222, p

= .014, partial n2 = .052, and F (1, 114) = .000, p= .985 partial 112 = .000 respectively.
Also, no signiﬁcant interaction effect was found for respondent’s and model’s race,
respondent’s race and measure, and model’s race and measure, F (l , 114) = .065, p = .799,
partial n2 = .001, F (1. 114) = .006. p = .940, partial n2 = .000, and F (1, 114) = .001. p

= .977, partial n2 = .000. Finally. there was no three-way interaction effect, F (l. 114) =
2.948. p = .089. partial 112 ——— .025.

Table 12 Participant Race by Model’s Race by Attitude Measure (Implicit/Explicit)
in Condition 5

 

 

 

Source SS df MS F P ES
Between Subjects 10.193 1.456
Respondent’s Race 1.424 1.424 1.479 .226 .013

 

 

 

—ih—Il—it—-l\l

 

 

 

 

 

Model’s Race 5.993 5.993 6.222 .014 .052

Measure .000 .000 .000 .985 .000

Respondent x Model .063 .063 .065 .799 .001

1’65”“de X .006 1 .006 .006 .940 .000
Measure

Model x Measure .001 1 .001 .001 .977 .000

Respondent x MOde] 2.840 1 2.840 2.948 .089 .025
x Measure

Within Subjects 109.807 1 14 .963
Total 1.424 1 1.424 1.479 .226 .013

 

63

CHAPTER IV

DISCUSSION AND CONCLUSIONS

This chapter summarizes ﬁndings from this study, discusses implications for
advertising practitioners and researchers. notes limitations and offers suggestions for

future study.

Discussion
This study tested both Caucasian and African American respondents’ attitudes
toward ads with models’ of different ethnic backgrounds. Additionally, the study also
examined the role of social-desirability in the process of reporting attitudes by examining
automatic processing in the form of response latencies. I wanted to investigate if
prejudice could be detected by examining explicit attitudes (subject to self-presentation)

in relation to implicit attitudes (not subject to self-presentation).

Attitude toward Ads

 

Although there was no signiﬁcant difference. there were some interesting
tendencies observed in the explicit attitude measure, using four 5-point differential scales
(bad-good, unfavorable-favorable, dislike-like, and poor quality-good quality). Unlike
previous studies, this study revealed that Caucasian American respondents explicitly
reported more positive attitudes toward a Caucasian American model (m = 3.37) than
toward an African American model (m = 3.31). Also it showed that Caucasian

American respondents implicitly expressed more positive attitudes toward a Caucasian

64

American model (m = 3.54) than toward an African American model (m = 3.28). These
ﬁndings conﬂict with previous studies. Those studies have found that African American
consumers responded more favorably to ads featuring African American models than to
ads featuring Caucasian American models (e.g.. Choundhury and Schmid 1974; Kerin
1979; Williams et a1. 1995). On the other hand. several studies have reported that
Caucasian American consumers did not rate ads featuring African American models any
differently from those featuring Caucasian American models (Whittler and DiMeo 1991;
Williams et a1. 1995).

Besides, there was no signiﬁcant difference in measuring implicit attitudes toward
the ads, but the ﬁnding showed interesting patterns. The result revealed that Caucasian
American respondents implicitly reported more positive attitudes toward a Caucasian
American model (m = 6.15) than toward an African American model (m = 6.00). Also.
the result showed that African American respondents implicitly expressed more positive
attitude toward an African American model (m = 4.39) than toward a Caucasian
American model (m = 4.23). Respondents were instructed that “To avoid mistakenly
pressing the wrong key, please keep your index ﬁngers on the “z” and “/” keys to enable
rapid response. DO THIS AS FAST AS YOU CAN while making as few mistakes as
possible.” Thus. under this instruction. it is possible to assume that respondents do not
have enough time to create deliberative attitudes toward the ads. As a result. they
reported their automatic and spontaneous attitudes.

Finally, results reveal that respondents are likelier to report positive attitudes to
Caucasian American models than to African American models. This likelihood is

signiﬁcantly more likely to be observed in explicit attitude measure than in implicit

65

measure. That is, respondents report more positive attitude when they explicitly express
attitudes toward Caucasian American models in ads.

This ﬁnding differs from what previous studies found that that African American
consumers responded more favorably to ads featuring African American models than they
did to ads featuring Caucasian American models (e.g.. Choundhury and Schmid 1974:
Kerin 1979; Williams et a1. 1995 ). Rather than. this study reveals the similar result from
the 1954 doll test used by pioneering psychologist Kenneth Clark to help make the case
for desegregation in the landmark Brown v. Board of Education Supreme Court decision
that outlawed segregated public schools. In that study, black children tended to prefer
white dolls over black dolls, leading analysts to conclude that they were internalizing
their second-class status -- believing themselves to be inferior (Craemer2005). In 2007.
the test was duplicated in Harlem. N.Y.. last year. more than a half-century after Brown.

and the results were unchanged (Iidney 2006).

Attitude Measures: Response Latency

Regarding response latency of implicit attitude, there was no signiﬁcant
difference. However, the result revealed that Caucasian American respondents were
faster to report attitudes toward a Caucasian American model (m = -45.43 ms) than
toward an African American model (m = 68.22 ms). Also. the result showed that
African American respondents were faster to report their attitudes toward an African
American model (m = 10.80 ms) than toward a Caucasian American model (m = 150.92
ms). C ontrarily. although no signiﬁcant difference was found in response latency of

explicit attitude, it differed from explicit attitude. The result revealed that Caucasian

66

American respondents were faster to report attitudes toward an African American model
(m = 2123.72 ms) than toward a Caucasian American model (m = 2737.50 ms). Also.
the result showed that African American respondents were faster to report their attitudes
toward an African American model (m = 1624.76 ms) than toward a Caucasian American
model (m = 1753.61 ms). There was no difference of African American respondents’
implicit and explicit response latencies. African American respondents were equally
faster to respond to African American models in the ads. However, Caucasian American
respondents showed different response latencies. If they were under time pressure, they
were faster in responding to Caucasian American models, but if they were not, they were

faster in responding to African American models in the ads.

Control Group: Condition 5

 

Employing within-subject design, Control group’s result was similar to between-
subject design. It is not signiﬁcant, but both ethnic groups preferred ads featuring their
own models. and were faster to respond to ads featuring their own models. On the other
hand. African American participants were observed to respond slower than Caucasian
Americans. Consequently. there was no artiﬁcial effect on exposing 'similar ads among

participants in conditions 1, 2, 3 and 4.

Implications
This study provides unique opportunity to examine consumer attitudes toward ads,
especially toward models. In this current study, all advertising components except

model ethnicity were deliberately removed, and two different types of attitude measures

67

were used. Although there was no signiﬁcant ﬁnding on respondent’s and model’s race
in explicit and implicit attitude. results indicate that whereas Caucasian Americans prefer
Caucasian American models in ads. African Americans have positive attitudes toward
African American models in ads. l-lowever. regardless oftheir race. respondents
reporting their deliberative (explicit) attitudes were likelier to report positive attitudes
toward Caucasian American models than toward African American models.

This result indicates that when they report their attitudes explicitly, they
signiﬁcantly have more positive attitudes toward Caucasian American models than
toward African American models. regardless race. It suggests that there may be some
difference between explicit and implicit attitude. As Muehling and McCann (1993)
suggested. consumers have dual attitudes toward ads: an automatic. involuntary reaction
and a conscious more deliberate evaluation. This current study showed the possibility
of the “automatic” nature of attitude formation that previous self-report measures of
attitude toward the ad have failed to identify. Considering that consumers tend to page
through a magazine and skip most of its ads, unless they are intentionally looking for
speciﬁc information, it is true that consumers may have automatic or spontaneous
attitudes toward ads they brieﬂy scan. From this viewpoint, the implicit measure newly
employed in this study provides unique opportunity for advertising researchers relevant
to the deﬁnitional issue ofattitude toward ads.

Also, regardless race and measurement. participants have signiﬁcantly more
positive attitudes toward Caucasian American models than toward African American
models. This result supports reasoning that since members of a society develop shared

category norms through the implicit communication of category norms in mass media,

68

 

participants assume that whiteness in ads is the norm (Kahneman and Miller 1986; Pratto
et. a1. 2007). Since Caucasian American models were the norm in mainstream magazine
ads, respondents’ overall attitudes toward the model are positive. This result is
consistent with Pratto, Korchmaros and Hegarty’s (2007) ﬁnding that typical race
features usually “go without saying” because category norms are assumed to be common
knowledge. Traditionally. general media advertising in the USA featured advertising
using ‘all-white’ models. with advertisements directed towards mass audience on the
assumption that African American as well as Caucasian American consumers would be
reached simultaneously. Due to the relatively higher status enjoyed by Caucasian
Americans, it was assumed that advertisements featuring Caucasian American models
would be equally-favorably evaluated by African American audiences (Kinra 1997).
However. research that supports the null should always be interpreted with
extreme caution. Differences between the groups may exist. but may not have been
captured in the current study. Therefore. while the current study shows no inherent
respondent bias in the reporting of attitudes toward ads. additional variables related to
how information is processed could be important to examine in future research. For
instance, the underlying processing of ethnic cues could be effected by level of ethnic
identiﬁcation for a given respondent. Those who deem their ethnic identity as central to
their self-concept could rely and seek out ethnic representation more so than others with
less ethnic identiﬁcation. Similarly. a measure of respondents’ motivation to control
prejudiced reactions could provide an indication of perceived pressure to conform to

societal norms.

69

Limitations

Results should be interpreted in light of limitations of methods used. Like most
experimental research on advertising effects. conditions surrounding exposure to the
stimuli were not completely natural. Viewing ads on a computer in one’s home or
university computer lab, or outside of a driver’s license ofﬁce, does not necessarily
simulate natural exposure conditions. Nonetheless, stimulus ads were created as
realistically as possible by modifying actual ads in ways consistent with the theoretical
constructs and to simulate normal exposure conditions as closely as possible. given
inevitable constraints.

The results indicate that model’s race may not matter in the formation of attitude
toward an ad for this particular population. It is not out-of-line to expect prejudice to be
lessened through exposure and education. In fact, the need to hide undesirable attitudes
may have been mitigated by the relatively homogeneous pool of college students, their
shared experiences. a shared environment, and similar backgrounds. Also, ince this
current study was conducted with a college student sample. the results cannot be
generalized to consumers as a whole. According to Brown and Stayman (1992). student
samples can be expected to be more homogenous and tend to yield higher correlations
than can non-student samples. Thus, using student subjects appears to have an upward-
biasing effect on the strength of some relationships. This effect is a limiting condition
on the generalizability of results generated from student samples. Also, Caucasian
American respondents’ economic backgrounds were richer than African American
respondents who participated in this current study. Since ethnicity and

majority/minority representation nearly always are confounded with status. respondents’

70

status affects how they respond to advertising (Brumbaugh and Grier 2006).

Additionally. in this present study. implicit measure assess attitude toward the ads
as identical to those assessed by explicit measures. In other words. implicit measure
employed in this current study was not about uncovering different associations (Brunel et
al.. 2004; Greenwald et a1. 1998). Rather. both explicit and implicit measures assess the
same construct, attitude toward the ad. As a result, it is hard to judge this study’s
ﬁnding against previous psychology studies considering implicit measures different from
explicit measures, because they ﬁrst primed speciﬁc constructs in a memory. When
people are aware of their attitudes. beliefs, or values but are unwilling to divulge them, no
priming or association is necessary (Brunel et al.. 2004; Fazio 1995: Greenwald et a1.
1998).

F inally, fatigue is always an issue with repeated measures experiments. While
every effort was taken to distribute the effects of fatigue through random presentation of
ads, these effects may have limited involvement in the tasks. Future research should
strike a balance between the number of exposures and participant involvement. While
incentives were offered for participation, they were not tied to performance outcomes.
Perhaps future research using the methods employed in the current study may wish to

design a reward structure that encourages active participation.

Suggestions for Future Study
Future research might explore other viewer/advertising matches like status,
media habit, lifestyle, etc. Further examination of the role of individual differences

might be necessary. While this current study tested the possibility of dual attitude

71

toward models in ads, the results indicate that other individual factors affect attitudes.

Researchers also should explore ethnic product and media placement as the
antecedents of attitude toward ads. Media could be a possible factor affecting
determinants of attitude toward ads. Depending on media types. implicit or explicit
attitudes may exist. For instance. the Internet seems likelier to be associated with
explicit attitude than television seems, due to inherent characteristics in the medium or its
audience (e.g., interactivity, and level of attention).

In addition, the underlying process in attitude formation is still unclear as is the
effectiveness of minority/majority models on other measures of advertising effectiveness
(e. g., perceived relevance, purchase intent, etc.). Future research must examine the role
of prejudice in the reporting of attitude using a more diverse population. Further,
research should examine the topic in an altemative advertising context. While the
current study used products relevant to college students. we did not measure involvement
in products nor in the experiment itself. Students may have been uninterested in the
topic or products and therefore did not care enough to evaluate the ads as they would in a
more externally valid setting.

Finally, future study should explore conceptual issues regarding attitude toward
ads. According to Greenwald and Banaji (1995), implicit attitude is deﬁned as
“introspectively unidentiﬁed (or inaccurately identiﬁed) traces of past experience that
mediate favorable or unfavorable feelings. thoughts, or action toward social objects” (p.8).
Also, Fazio. .lackson, Dunton and Williams (1995) consider implicit attitude measure as a
“bona ﬁde pipeline” that allows measuring of“true” attitude. Unlike psychology

studies, this study’s attitude object is advertising known as “cluttering and incidental”

72

exposure in a competitive environment. Thus. it is assumed that attitude toward the ad
is formed in preattentive processing (.laniszewski 1988; Leong et al., 1999).

Preattentive processing refers to “the simultaneous, preconscious monitoring of all
sensory channels for events that will require a shift in attention” (.laniszewski 1988, p.
201). This study showed the possibility of dual attitudes toward the ads. Thus. future
study should investigate how attitude formation occurs in different information

processing stages (i.e.. preattentive. attentive. and comprehensive stages etc).

Conclusion

In conclusion, although ethnicity is convenient and frequently appropriate
contextual reference in comparing groups, it is unclear how this group differs from the
comparison group on bases other than ethnicity (Brumbaugh and Grier 2006). However,
this current study’s rigorous experimental design allows examining the sole effect of
model ethnicity on consumers’ dual attitudes across different types of products.

This study demonstrates that consumers may have dual attitudes. Unlike
psychology studies to show Caucasian American’s strong preference to their own race
(Brunel et al.. 2004; Fazio 1995; Greenwald et a1. 1998), this study did not ﬁnd their
adverse attitudes toward African American models in the ads. It is not only for the use
of different implicit measure but also for the use of different attitude object. Also,
attempting to resolve conﬂicted ﬁndings among advertising-related minority studies. this
current study supports that, regardless of their ethnicities, consumers prefer majority race
models. In particular, congruent with Brunel, Tietje and Greenwald (2004), African

Americans did not show an extreme in-group bias.

73

Overall. the research attempted to examine a new measure in the ethnic minority
advertising literature. The study provided an introduction to the concept of underlying
racial biases that may inﬂuence attitudes. While there were no differences detected
between explicit and implicit attitude measures in this study, future research should still
examine this area in more detail with varied age groups, ethnicities, genders and product
categories. However, the fact that respondents from two different racial groups showed
no preferences for model by race may be an indication that the next generation is
becoming more integrated; a potential trend that will have profound effects on the
advertising industry as well as racial relations in the nation. In the end, the idea of dual-
attitudes is very provocative for future research as are the opportunities to develop better
implicit measures for assessing underlying attitudes, both within ethnic advertising

research and beyond.

74

APPENDIX A

PRETEST QUESTIONNAIRE

Brand Name:

 

The Mode1(s) in the Advertisement is (are)...

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Unfamiliar I 2 3 4 5 6 7 Familiar
Attractive 1 2 3 4 5 6 7 Unaftracfive
Not Classy I 2 3 4 5 6 7 Classy
Beautiful 1 2 3 4 5 6 7 Ugly
Plain 1 2 3 4 5 6 7 Elegant
Not Sexy 1 2 3 4 5 6 7 Sexy
Cute 1 2 3 4 5 6 7 Not cute
How old does the model (or models) appear in the ad?
How appropriate is the model for the product being advertised?
j Inappropriate j 1 j 2 j 3 I 4 j 5 I 6 I 7 f Appropriate j
Have you seen this advertisement before?
:II)’ at 1 2 3 4 5 6 7 Frequently

 

 

 

 

 

 

 

 

 

 

 

What Is the model's race?
1. American Indian

2. Asian or Pacific Islander

3. Black or African-American
4. White (Caucasian)

5. Latino or Hispanic

6. Others (Specify )

75

APPENDIX B

PAPER SURVEY

YOUR COMPUTER NUMBER IS

Now we have just a few more questions about you.
Please follow the directions with each question.

On the rating scales below, please circle the number that best describes your
Opinion about advertising in general.

 

 

 

 

 

 

 

Strongly Disagree Strongly Agree
Advertising in general is useful. 1 2 3 4 5 6 7
Advertising in general is truthful. l 2 3 4 5 6 7
Advertising in general is l 2 3 4 5 6 7
believable.
Advertising in general is good. i 2 3 4 5 6 7

 

 

On the rating scales below, please circle the number that best describes your
opinion about your ethnic group.

 

Strongly Disagree Strongly Agree

 

I think a lot about how my life
will be affected by my ethnic 1 2 3 4 5 6 7
group membership

 

I have a lot of pride in my ethnic

 

 

group and its accomplishments. I 2 3 4 5 <5 7
I have a strong sense of

belonging to my own ethnic l 2 3 4 5 6 7
group.

I feel a strong attachment

towards my own ethnic group. I 2 3 4 5 6 7

 

I feel good about my cultural or
ethnic background. I 2 3

 

 

 

 

76

On the rating scales below, please circle the number that best describes your

opinion about yourself.

 

Strongly Disagree

Strongly Agree

 

I feel guilty when I have a
negative thought or feeling
about an African American.

 

I feel guilty when I have a
negative thought or feeling
about an Asian American.

 

I feel guilty when I have a
negative thought or feeling
about a European American
(Caucasian).

 

I feel guilty when I have a
negative thought or feeling
about a Hispanic American.

 

I feel guilty when I have a
negative thought or feeling
about a Native American.

 

In today’s society. if is important
to not be perceived as
prejudiced in any manner.

 

I always express my thoughts
and feelings, regardless of how
controversial they might be.

 

I get angry With myself when I
have a thought or feeling that
might be considered
prejudiced.

 

Going though life worrying
about whether you might
offend someone is just more
trouble than it's worth.

 

 

 

77

 

On the rating scales below, please circle the number that best describes your
opinion about yourself.

 

Strongly Disagree Strongly Agree

 

It’s important to me that other
people do not think I'm I 2 3 4 5 6 7
prejudiced.

 

I feel it's important to behave
according to society’s I 2 3 4 5 4 7
standards.

 

I’m careful not to offend my

friends, but I don't worry about
offending people I don't know I 2 3 4 5 6 7
or don’t like.

 

I think it is important to speak
one's mind rather than worrying 1 2 3 4 5 6 7
about offending someone.

 

It's never acceptable to express
one’s prejudices. I 2 3 4 5 6 7

 

It bothers me a great deal when
I think I've offended someone,

so I’m always careful to I 2 3 4 5 6 7
consider other people's feelings.

 

If I have a prejudiced thought or
feeling, I keep it to myself. I 2 3 4 5 6 7

 

I would never tell jokes that
might offend others. I 2 3 4 5 6 7

 

I'm not afraid to tell others what
I think even when I know they I 2 3 4 5 6 7
disagree with me.

 

If someone who made me

uncomfortable sits next to me
on a bus, I would not hesitate to I 2 3 4 5 6 7
move to another seat.

 

 

 

 

78

Now we have some questions about your family.

Which of these categories best describes your family‘s total combined
household income for the past 12 months? This should Include income (before
taxes) from all sources, wages, rent from properties, social security, disability
and/or veteran's benefits, unemployment benefits, workman's compensation.
help from relatives (including child payments and alimony), and so on.

 

I. Less than $5,000 7. $50,000 through $74,999
2. $5,000 through $I I,999 8. $75,000 through $99,999
3. $12,000 through $I5,999 9. $100,000 through 199,999
4. $l 6,000 through $24,999 I0. $200,000 and greater
5. $25,000 through $34,999 I I. Don't know
6. $35,000 through $49,999 I2. No response
Do you consider your family as 2
I. The upper class 5. The lower class
2. The upper middle class 6. Don’t know
3. The (lower) middle class 7. No response
4. The working class

Now we have some questions about your father and mother.

What is the highest degree your father earned? Father Mother

1. High school diploma or 6. Professional (MD, JD, DDS,
equivalency (GED) etc.)

2. Associate degree (junior 7. None of the above (less than
college) high school)

3. Bachelor's degree 8. Don't know

4. Master's degree 9. No response

5. Doctorate ID. No applicable

With regard to your father's current or most recent job activity, what kind of work
does (did) your parents do? (Job Title)

Father
Mother
(For example: registered nurse, personnel manager, supervisor of order
department, gasoline engine assembler, grinder operator.)

 

 

79

About yourself
What is your race or ethnic background?:

American Indian

Asian or Pacific Islander

Black or African-American

White (Caucasian), or

Latino or Hispanic

Others (Specify )

99.6.0059.“

How strongly do you identify yourself as a member of the ethnic group you
selected above? Please rate your feeling?

 

Strongly Disagree Strongly Agree

 

 

I 2 3 4 5 6 7

 

 

In what year were you born?

 

Which year are you in?

I. Freshmen
2. Sophomore
3. Junior
4. Senior

Are you:
I. Male
2. Female

This completes our study. Thank you for your time!

80

BIBLIOGRAPHY

Aaker, Jeniffer, Ann M. Brumbaugh, and Sonya Grier (2000), “Non-Target Markets and
Viewer Distinctiveness: The Impact of Target Marketing on Advertising
Attitudes.“ Journal ofConsumer Psychology. 9 (3), 127-140.

Aherson, Christopher L.. Carl Shoemaker. and Christina Tomolillo (2004), “Implicit Bias
and Contact: The Role of Interethnic F riendships,” The Journal of Social
Psychology, 1 I4 (3), 335-347.

Alba, Richard D. (1985), Ethnicity and race in the U. S.A..' Toward the twenty-ﬁrst
century, London: Routledge.

Allport, G. W. (1935), Attitudes, in A Handbook of Social Psychology, C. Murchison eds.
Worcester, MA: Clark University Press, 798-844.

Anonymous (200] ). “A Market Kept in the Closet,” American Demographics. 23, 36-42.

Appiah, Osei (2001 a). “Ethnic Identiﬁcation on Adolescents’ Evaluations of
Advertisements,” Journal of Advertising Research, 41 (5), 7-22.

Appiah, Osei (2001b). “Black, White, Hispanic, and Asian American Adolescents’
Responses to Culturally Embedded Ads.” The Howard Journal of
Communications. 12. 29-48.

Atkin, Charles and Martin Block (1983), “Effectiveness of Celebrity Endorsers.” Journal
of Advertising Research, 23 (1), 57-61.

Bemal, Martha E. and George P. Knight (Eds) (1993), Ethnic Identity: Formation and
Transmission among Hispanics and Other Minorities, Albany: Sate University of
New York Press.

Bhat, Subodh, Thomas W. Leigh and Daniel L. Wardlow (1998), “The Effect of
Consumer Prejudices on Ad Processing: Heterosexual Consumers’ Responses to
Homosexual Imagery in Ads,” Journal of'Adver/ising. 27 (4), 9-25.

Blumer, Herbert. and Duster Troy ( I 980). “Theories of Race and Social Action,” in
Sociological Theories: Race and Colonialism. UNESCO eds., Paris. France:

UNESCO, 211-238.

Brewer, Marilynn B. (1979), “In-Group Bias in the Minimal Intergroup Situation: A
Cognitive-Motivational Analysis,” Psychological Bulletin, 86 (2). 307-324.

Briley, Donnel A., L. J. Shrum, and Robert S. Wyer, Jr. (2007), “Subjective Impressions

of Minority Group Representation in the Media: A Comparison of Majority and
Minority Viewers’ Judgments and Underlying Processes,” Journal of Consumer

81

Psychology, 17 (l ), 36-48.

Brown, Steven P. and Douglas M. Stayman (I992), “Antecedents and Consequences of
Attitude toward the Ad: A Meta-analysis,” Journal of Consumer Research. 19 (I),
34-51.

Brumbaugh. Anne M. (2002). “Source and Nonsource Cues in Advertising and Their
Effects on the Activation of Cultural and Subcultural Knowledge on the Route to
Persuasion,” Journal of'C'onsumer Research, 29 (September). 258—269.

Brumbaugh. Anne M. and Sonya A. Grier (2006), “Insights from a “Failed” Experiment:
Directions for Pluralistie. Multiethnic Advertising Research,” Journal of
Advertising. 35 (3). 35-46.

Brunel, Frederic, Brian C. Tietje, and Anthony G. Greenwald (2004), “Is the Implicit
Association Test a Valid and Valuable Measure of Implicit Consumer Social
Cognition?” Journal of Consumer Psychology, 14 (4), 384-404.

Bush, Ronald F., Robert F. Gwinner. and Paul J. Solomon (1972), “White Backlash to
Negro Ads: Fact or Fantasy?” Journalism Quarterly, 49 (Summer). 253-262.

Byme. Donn (1971). The attraction paradigm. New York: Academic Press.

Casas. J. M. and S. D. Pytluk (1995). “Hispanic identity development: Implications for
research and practice.” in Handbook of'ttlulticultural Counseling, J. G. Ponterotto.
J. M. Casas. L. A. Suzuki. C. M. Alexander eds.. Thousand Oaks. CA: Sage. 155-
180.

C houdhury, Pravat K. and Lawrence S. Schmid (1974), “African American Models in
Advertising to African Americans,” Journal of Advertising Research, 14, 19-22.

Cobb, Michael D. (2001), Unobtrusively Measuring Racial Attitudes: The Consequences
of Social Desirability Effects. Doctoral dissertation, University of Illinois at
Urbana-Champaign, 2001 .

Cohen, Judy (1992). “Caucasian American Consumer Response to Asian Models in
Advertising,” The Journal of Consumer Marketing, 9 (2), 17-27.

Cohen, Joel B. and Americus Reed II (2006). "A Multiple Pathway Anchoring and
Adjustment (MPAA) Model of Attitude Generation and Recruitment," Journal of
Consumer Research. 33(1). 1-15.

Cokley, Kevin, Meera Komarraju. Rachel Pickett and Frances Shen (2007), “Ethnic
Differences in Endorsement of the Protestant Work Ethic: The Role of Ethnic
Identity and Perceptions of Social Class,” The Journal of Social Psychology, 147
(I), 75-81.

82

C randaII. Christina S. and Amy Eshleman (2003), “A Justiﬁcation-Suppression Model of
the Expression and Experience of Prejudice.” Psychology Bulletin, 129 (3), 414-
446.

Craemer, Thomas (2005), “Racial Affect among African Americans and Non-African
Americans. An Implicit "Dolls Test" " in the Proceedings of the 2005 Conference
of the Midwestern Political Science Association, 1-28.

Edney, Hazel T. (2006, September 14), “New ‘DoII Test’ Produces Ugly Results " F ('N
Publishing, Retrieved April 2, 2007, from
http://wwwftnalcalI.com/artman/publish/article_29 I 9.shtml

De Houwer. J. (2003). “A Structural Analysis oflndirect Measures ofAttitudes.” in
The Psychology ovaaluation: Affective Process in Cognition and Emotion, J.
Musch and KC. Klauer eds.. Mahwah. NJ: Erlbaum. 219-244.

Demetrius, Patterson (2006), “Buying Power of Blacks Taken for Granted,” Tri-State
Defender, 55 (1), 1A. (newspaper)

Deshpande, Rohit, Wayne D. Hoyer, and Naveen Donthu (1986), “The Intensity of
Ethnic Afﬁliation: A Study of the Sociology of Hispanic Consumption,” Journal
of Consumer Research. 13 (2). 214-220.

Deshpande, Rohit and Douglas Stayman (1994), “A Tale of Two Cities: Distinctiveness
Theory and Advertising Effectiveness,” Journal of Marketing Research, 31, 57-64.

Devine, Patricia G. ( l 989), “Stereotypes and prejudice: Their automatic and controlled
components." Journal ofl’ersonality and Social Psychology, 56. 5-18.

Dimofte, Claudiu V., Mark R. Forehand. and Rohit Deshpande (2003). “Ad Schema
Incongruity as Elicitor of Ethnic Self-Awareness and Differential Advertising
Response,” Journal ofAdvertising. 32 (4), 7-17.

Dovidio, John F., Kerry Kawakami, and Samuel L. Gattner (2002), “Implicit and Explicit
Prejudice and Interracial Interaction,” Journal of Personality and Social
Psychology, 82 (1), 62-68.

Eagly, Alice H. and Shelly Chaiken (1993), “The Nature of Attitudes.” in The
Psychology of Attitudes, A. H. Eagly and S. Chaiken eds.. Fort Worth, TX:
Harcourt Brace Jovanovich, pp. 1-21.

Edwards. Jeffrey R. and Richard P. Bagozzi (2000). “On the Nature and Direction of
Relationships between Constructs and Measures,” Psychological Methods. 5 (2).
155-174.

83

 

Fazio. Russell H. (1990). “Multiple Processes by Which Attitudes Guide Behavior: The
MODE model as an Integrated Framework." Advances in Experimental Social
Psychology: 23. 75-109.

Fazio. Russell H.. Joni R. Jackson. Bridget Dunton, and Carol J. Williams (1995),
“Variability in Automatic Activation as an Unobtrusive Measure of Racial
Attitudes: A Bona Fide Pipeline?” Journal of Personality and Social Psychology,
69 (6), 1013-1027.

F azio Russell H. and Michael. A. Olson (2003), “Implicit Measures in Social Cognition
Research: Their Meaning and Use,” Annual Review Psychology. 54, 297-327.

Fishbein, M. and I. Ajzen (1975), Beliefs, Attitude, Intention, and Behavior: An
Introduction to Theory and Research, Reading. MA: Addison-Wesley.

Forehand, Mark R. and Rohit Deshpande (2001). “What We See Makes Us Who We are:
Priming Ethnic Self-Awareness and Advertising Response,” Journal of
Marketing Research. 38 (3‘). 336-349.

Forehand, Mark. R. and Americus II Reed (2002). “Identity Salience and the Inﬂuence of
Differential Activation of the Social Self-Schema on Advertising Response,”
Journal of Applied Psychology, 87 (6), 1086-1099.

F riese, Malte, Michaela Wanke and Henning Plessner (2006), “Implicit Consumer
Preferences and Their Inﬂuence on Product Choice,” Psychology & Marketing.
23 (9), 727-740.

Deaux, Kay (1992). Personalizing identity and socializing self. In G. Breakwell (Ed.),
Social Psychology of Identity and the Self Concept (pp. 9-33). San Diego, CA:
Academic Press.

Gawronski. B., and Bodenhausen. G. V. (2006), “Associative and propositional processes
in evaluation: An integrative review of implicit and explicit attitude change,”
Psychological Bulletin, I 32. 692-731.

Green, Corliss L. (1999). “Ethnic Evaluations of Advertising: Interaction Effects of
Strength of Ethnic Identiﬁcation, Media Placement, and Degree of Racial
Composition,” Journal of A dvertising, 28 (I), 49-64.

Greenwald, Anthony G. and Mahzarin R. Banaji (1995), “Implicit social cognition:
Attitudes, Self-esteem, and Stereotypes,” Psychological Review, 102 (1), 4-27.

Greenwald, Anthony G., Debbie E. McGhee, and Jordan L. K. Schwartz (1998),
“Measuring Individual Differences in Implicit Cognition: The Implicit
Association Test,” Journal of Personality and Social Psychology, I 74, 1464-
1480.

84

Greenwald Anthony G... Mahzarin R. Banaji, Laurie A. Rudman, Shelly D. F amham,
Brian A. Nosek, and Deborah S. Mellott (2002), “A Uniﬁed Theory of Implicit
Attitudes, Stereotypes. Self-Esteem and Self-Concept,” Psychology Review, 109
(I ), 3—25.

Grier. Sonya A. and Ann M. Brumbaugh (I999). “Noticing Cultural Differences: Ad
Meanings Created by Target and Nontarget Markets.” Journal ofAdvertising. I8
(I). 79-83.

Grier, Sonya A. and Rohit Deshpande (2001), “Social Dimensions of Consumer
Distinctiveness: The Inﬂuence of Group Social Status and Identity on
Advertising Response,” Journal of Marketing Research, 38 (2), 216-224.

Grier, Sonya, Ann M. Brumbaugh and Corliss G Thornton (2006), “Crossover Dreams:
Consumer Responses to Ethnic-Oriented Products, “ Journal of Marketing. 70
(April), 35—51.

Hegarty, Peter and Felicia Pratto (2001), “The effects of social category norms on
explanations for intergroup differences,” Journal ofPersonality and Social
Psychology. 80. 723-735.

Hegarty. Peter and Felicia Pratto (2004). “The differences that norms make: Empiricism.
social constructionism and the interpretation of group differences” Sex Roles.
50, 445-453.

Henderson, Jennifer J. and Gerald J. Baldasty (2003), “Race, Advertising, and Prime-
Time Television,” The Howard Journal ofCommunication, 14, 97-112.

Himmelfarb, Samuel (1993), “The Measurement of Attitudes,” in The Psychology of
Attitudes, A. H. Eagly and S. Chaiken eds., Fort Worth, TX: Harcourt Brace
Jovanovich, pp. 23-57.

Hirschman, Elizabeth C. (1981), “American Jewish Ethnicity: Its Relationship to Some
Selected Aspects of Consumer Behavior.” Journal of Marketing, 45, 102-110.

Holtgraves, Thomas (2004). “Social Desirability and Self-Reports: Testing Models of
Socially Desirable Responding.” Personality and Social Psychology Bulletin. 30
(2). 161-172.

Janiszewski, Chris (1998), “Preconscious Processing Effects: The Independent of
Attitude Formation and Conscious Thought,” Journal of Consumer Research, 15

(3), 199-209.

Kahneman, Daniel and Dale Miller (1986), “Norm Theory: Comparing Reality to Its
Alternatives,” Psychological Review, 93 (2), 136-153.

85

Kanungo, Rabindra N. and Sam Pang (1973), “Effects of Human Models on Perceived
Product Quality,” Journal of Applied Psychology, 57, 172-178.

Karande, Kiran (2005), “Minority Response to Ethnically Similar Models in
Advertisements: An Application of Accommodation Theory.” Journal of
Business Research, 58 (I I), 1573-1580.

Keefe. Susan E. (1992). “Ethnic identity: The domain of perceptions of and attachment to
ethnic groups and cultures.” Human Organization. 51 (I). 35-43.

Kerin. Roger (1979). “African American Model Appearance and Production Evaluation,”
Journal ofCommunication. 4. 551-563.

Kinra, Neelam (1997). “The Communicative Effectiveness of Ethnically Oriented
Advertising,” International Journal of Advertising, 16 (3), 221-230.

Knobloch-Westerwick, Silvia and Brendon Coates (2006), “Minority Models in
Advertisements in Magazines Popular with Minorities,” Journalism & Mass
Communication Quarterly, 83 (3), 596-614.

Koslow, Scott, Prem N. Shamdassni and Ellen E. Touchstone (1994), “Exploring
Language Effects in Ethnic Advertising: A Sociolinguistic Perspective,” Journal
of Consumer Research, 20 (4), 575-585.

La Ferle, Carrie and Mariko Morimoto (2004). “Caucasian Models in Luxury Product
Advertisements: Attitudes of Japanese Female Consumers.” in the Proceedings of
the 2004 Conference of the American Academy of A dvertising. 108-1 15.

Lee, Eunsun, Teresa Mastin and Steve Edwards (2006), “Revisiting the Effect of Models'
Race on Attitudes in Advertisements,” presented at the American Academy of
Advertising Conference, April 2006, Reno, NV.

Lutz, Richard J. (1991), “The Role of Attitude Theory in Marketing,” in Perspectives in
Consumer Behavior (4th ed.), H.H. Kassarjian and TS. Robertson eds.,
Englewood Cliffs, NJ: Prentice-Hall, 317-39

MacKenzie, Scott B., Richard J. Lutz, and George E. Belch (1986), “The Role of Attitude
toward the Ad as 3 Mediator of Advertising Effectiveness: A Test of Competing
Explanations,” Journal of Marketing Research. 23 (2), 130-143.

McQuarrie, Edward F and Barbara J Phillips (2005), “Indirect Persuasion in Advertising:
How Consumers Process Metaphors Presented in Pictures and Words,” Journal

of Advertising, 34 (2), 7-22.

Morimoto, Mariko and Carrie La Ferle (2005), “Advertising Evaluations of Asian

86

Americans.” in the Proceedings ofthe 2005 Conference ofthe American
Academy of Advertising. 105-11 1.

Muehling, Darrel D. (1987), “An Investigation of Factors Underlying Attitude toward
Advertising in General,” Journal of Advertising, 16 (1), 32-40.

Muehling, Darrel D. and Michelle McCann (1993), “Attitude toward the Ad: A Review,”
Journal of Current Issues and Research in Advertising, 15 (2), 25-58.

Nelson. John R., Jr. and Jeanne E. Lukas (1990), “Target: Minorities.” Marketing &
Decisions, 25 (10), 70-71.

Nosek, Brian A. and Mahzarin R. Banaji (2005). “The go/no-go Association Task.” Social
Cognition. 19. 625—666.

Ohanian, Roobina (1990). “Construction and Validation of a Scale to Measure Celebrity
Endorsers‘ Perceived Expertise. Trustworthiness and Attractiveness. " Journal of
Advertising, 19 (3), 39-52.

PaUIhUS, Delroy L. (1984), “Two-Component Models of Socially Desirable Responding.”
Journal of Personality and Social Psychology, 43, 83 8-852.

Paulhus, Delroy L. and Douglas B. Reid (1991), “Enhancement and Denial in Socially
Desirable Responding,” Journal of Personality and Social Psychology, 60 (2),
207-317.

Payne, Keith 8., Clara Michelle Cheng. Olesya Govorun, and Brandon D. Stewart (2005),
“An Inkblot for Attitudes: Affect Misattribution as Implicit Measurement,”
Journal of Personality and Social Psychology, 89 (3), 277-293.

Petty. Richard E., Monique A. Fleming. and Paul H. White (1999). “Stigmatized Sources
and Persuasion: Prejudice as a Determinant of Argument Scrutiny.” Journal of
Personality and Social Psychology. 76 (I). 19-34.

Phinney, Jean S. (1992), “The Multi-group Ethnic Identity Measure: A New Scale for Use
with Adolescents and Youth Adults from Diverse Groups,” Journal ofAdoIescent
Research, 7, [56—176.

Phinney, Jean S. and Linda L. A1ipuria(1990), “Ethnic identity in college students from
four ethnic groups,” Journal of Adolescence, 13, 171-184.

Pires, Guilherrne, John Stanton, and Bruce Cheek (2003), “Identifying and reaching an
ethnic market: methodological issues,” Qualitative Market Research: An
International Journal, 6 (4), 224-235.

Pratto, Felicia, Josephine D Korchmaros, and Peter Hegarty (2007), “When Race and

87

Gender Go without Saying,” Social Cognition, 25 (2), 221-249.

Putrevu. Sanjay and Kenneth R. Lord (1994). “Comparative and Noncomparative
Advertising: Attitudinal Effects under Cognitive and Affective Involvement
Conditions.” Journal ofAdvertising. 23 (2), 77-90.

Ragoonan, Ashanti Y.. Nanda R. Shrestha and Wilbur 1. Smith (2005). “Advertising in
Black Magazines: Do Advertisements Reﬂect African American Consumer
Behaviro?” Services Marketing Quarterly. 27(2). 65-88.

Ratchford, Brian T. (1987), “New Insights about the FC B Grid,” Journal of Advertising
Research, 27 (4), 24-38.

Ratcliff, Roger (1993), “Methods for Dealing With Reaction Time Outliners,”
Psychological Bulletin, 114 (3). 510-532.

Rogers. Everett and D. P. Bhowmik (1970). “Homophily-Heterophily: Relational
Concepts for Communication Research.” Public Opinion Quarterly, 24, 523-538.

Scott. Linda M. (1994). “Images In Advertising: The Need For A Theory of Visual
Rhetoric.” Journal o/‘Consumer Research. 21 (2). 252-273.

Sherry, Allisa, Keith Wood. Emily B. Jackson and Nadine Kaslow (2006), “Racist Events
and Ethnic Identity in Low Income, African Americans,” Journal of Applied
Social Psychology, 36(6), 1365-1380.

Shimp, Terence A. (1981), “Attitude toward the Ad as a Mediator of Consumer Brand
Choice,” Journal of Advertising, 10 (2), 9-15.

Solomon, Paul J ., Ronald F. Bush and Joseph T. Hair (1976), “Caucasian American and
African American Consumer Sales Responses to African American Models,”

Journal of Marketing Research, 13, 431-434.

Tan, Chin T. and James McCullough (1985), “Relating Ethnic Attitudes and Consumption
Values in an Asian Context.” Advances in Consumer Research, 12, 126-128.

Thurstone, Louis L. (1931). “The Measurement of Attitudes.” Journal of A bnormal and
Social Psychology. 26. 249-269.

Tesser. Abraham and David Shaffer (I990). “Attitudes and Attitude Change,” Annual
Review of Psychology, 41, 479-523.

Torres, Ivonne M. (2007), “A Tale of Two Theories: Sympathy or Competition?” Journal
of Business Research, 60 (3), 197-205.

Tourangeau, Roger and Kenneth A. Rasinski (1988), “Cognitive Processes Underlying
Context Effects in Attitude Measurement,” Psychological Bulletin, 103 (3). 299-

88

314.

US. Census Bureau “Race and Ethnicity: US. Population Estimates and Population
Change, 2003 US. Census Bureau Race and Ethnicity: US. Population
Estimates and Population Change (2010—2050),” Washington DC: US.
Government Printing Ofﬁce (accessed June 30 2006), [available at
http://factﬁnder.census.gov/].

Vaughn. Richard (1986). “How Advertising Works: A Planning Model Revisited.”
Journal ofAdvertising Research. 26 (I). 57-66.

Waters. Mary C. (1990), Ethnic options: Choosing identities in America. Berkeley:

University of Califomia Press.

White, Paul H. and Stephen G. Harkins (1994), “Race of Source Effects in the
Elaboration Likelihood Model,” Journal of Personality and Social Psychology, 67
(5), 790-807.

Whittler, Tommy E. (1989), “Viewer’s Processing of Source and Message Cues in
Advertising Stimuli,” Psychology and Marketing, 6, 287-309.

Whittler, Tommy E. and DiMeo, J. (1991), “Viewers’ Reactions to Racial Cues in
Advertising Stimuli.” Journal of Advertising Research, 31 (6), 37-46.

Whittler, Tommy E. and Spira, Joan S. (2002). “Models Race: A Peripheral Cue in
Advertising Messages?” Journal of'Consumer Psychology. 12 (4). 291 -301.

Williams, Jerome D. and William .I. Qualls (1989), “Middle Class African American
Consumers and Intensity of Ethnic Identiﬁcation,” Psychology and Marketing. 6.
263-286.

Williams, Jerome D., William J. Qualls, and Sonya Grier (1995), “Racially Exclusive
Real Estate Advertising: Public Exclusive Real Estate Advertising: Public Policy

Implication for Fair Housing Practices,” Journal of Public Policy and Marketing,
14, 224-244.

Williams, Jerome D. and Kimberly D. Grantham (1999), “Racial and Ethnic Identity in
the Marketplace: An Examination of Nonverbal and Peripheral Cues,” Advances in

Consumer Research. 26. 451-454.

Wilson. Timothy D., Samuel Lindsey, and Tonya Y. Schooler (2000). “A Model of Dual
Attitudes,” Psychological Review. 107(1). 101-126

Wittenbrink. Bernd. Chick M. Judd. and Bernadette M. Park (1997), “Evidence for Racial
Prejudice at the implicit Level and Its Relationships with Questionnaire

Measures.” Journal of Personality and Social Psychology, 72, 262-274.

Wooten, David B. (1995), “One-of-a-Kind in a Full House: Some Consequences of

89

 

Ethnic and Gender Distinctiveness.” Journal of Consumer Psychology, 4 (3), 205-
224.

Zanna, M. P. and Rampel, J. K. (1988), Attitudes: A New Look at an Old Concept. In The

Social Psychology of Knowledge, D. Bar-Tal and A. W. Kruglanski eds., New
York: Cambridge University Press, 315-334.

90