THE EFFECTS OF DISPOSITIONAL AND SITUATIONAL ATTRIBUTIONS ON
EVALUATIVE RESPONSES TO NONCELEBRITY TESTIMONIAL ADVERTISING
by
KYOO-HOON HAN
(Under the Direction of Spencer F. Tinkham)
ABSTRACT
Testimonial advertising has been widely used to convey brand recommendations using an
endorser or a group of endorsers on behalf of the advertiser. This type of advertising appeal
involves a complex context in which consumer responses to the message are likely influenced by
many perceptual and motivational factors. While a number of studies have examined
endorsement effects over the years, relatively little attention has been paid to how consumers
process such testimonial messages from a theory-driven point of view. This research attempts to
examine causal associations between consumer perceptions generated by the endorser and the
message’s persuasive effects, using an attributional approach.
The present study is concerned with how consumers make causal inferences regarding the
endorser’s motivation to support the brand – specifically, dispositional versus situational
attributions – and how such inferences affect advertising effectiveness. Based on correspondent
inference theory and additional theoretical perspectives such as the source credibility model and
the elaboration likelihood model, it is hypothesized that consumers will generate predictable
patterns of attributional responses to testimonial messages according to the level of source
credibility and their personal involvement (product involvement and ad involvement). Further, it
is posited that these responses, in turn, will influence the consumers’ evaluations of the
advertisement and the brand being endorsed. Mediating roles of endorser-generated attributions
in the causal links from source credibility and personal involvement to the selected indicators of
persuasion are also proposed for hypothesis testing, as well as the effects of correspondence bias,
a psychological trait derived from the correspondent inference theory.
Two-stage randomized field experiments were administered to attain reliability and
generalizability of findings, by using replicated measurements and a consumer panel sample for
the second experiment. After three pretests designed to develop and refine the experimental
instruments, Experiment I and Experiment II were conducted with 231 college students and 356
consumer panelists, respectively. Results support the differential effects of dispositional and
situational attributions on the persuasiveness of noncelebrity testimonial advertisements. That is,
while participants generated situational attributions to a greater extent than dispositional
attributions, advertising effectiveness and the impacts of source credibility and personal
involvement on persuasion were better predicted by dispositional attributions than situational
attributions. An additional investigation of correspondence bias also confirmed its positive
effect on consumer responses to testimonial messages, consistent with the effects of dispositional
attributions. Based upon the empirical findings from the two experiments, theoretical and
managerial implications, as well as directions for future research, are discussed.
INDEX WORDS: attribution, testimonial advertising, product endorsement, source credibility,
involvement, consumer attitude, advertising effects
THE EFFECTS OF DISPOSITIONAL AND SITUATIONAL ATTRIBUTIONS ON
EVALUATIVE RESPONSES TO NONCELEBRITY TESTIMONIAL ADVERTISING
by
KYOO-HOON HAN
B.A., Sogang University, South Korea, 1993
M.A., The University of Missouri – Columbia, 2001
A Dissertation Submitted to the Graduate Faculty of the University of Georgia in Partial
Fulfillment of the Requirements for the Degree
DOCTOR OF PHILOSOPHY
ATHENS, GEORGIA
2004
© 2004
Kyoo-Hoon Han
All Rights Reserved
ii
THE EFFECTS OF DISPOSITIONAL AND SITUATIONAL ATTRIBUTIONS ON
EVALUATIVE RESPONSES TO NONCELEBRITY TESTIMONIAL ADVERTISING
by
KYOO-HOON HAN
Major Professor:
Spencer F. Tinkham
Committee:
Dean Krugman
Karen King
James Hamilton
Ellen Day
Electronic Version Approved:
Maureen Grasso
Dean of the Graduate School
The University of Georgia
December 2004
iii
ACKNOWLEDGEMENTS
It was six years ago that I decided to quit my job at an advertising agency to study
advertising at the graduate level in the United States. My decision was welcomed by nobody in
my family at the time. Subsequently, however, I was lucky to have unwavering encouragement
and support from many people – most of all, my wife and my parents who soon became the
greatest mental supporters of my studies. I dedicate my dissertation to them – my wife and
longtime friend, Seung-Hee, and my parents, Seung-Hun Han and Song-Ja Kim. Without their
love and devotion, I could not have reached this point.
I would like to truly thank my dissertation advisor Dr. Spencer Tinkham. I cannot forget
the e-mail he sent me with his lengthy comments on my research from North Carolina the day
after his mother passed away. I deeply appreciate his genuine care and considerate guidance
through the entire process of developing my dissertation. It was my great fortune to have his
mentorship during my doctoral program.
I also would like to thank the other members of my dissertation committee: Dr. Dean
Krugman, Dr. Karen King, Dr. Ellen Day, and Dr. James Hamilton. Their warm support and
many insightful suggestions greatly helped me improve my dissertation. I always have been
proud of my committee – for their expertise, caring, and commitment.
I want to thank Dr. Leonard Reid, who sincerely supported me for all the years of my
doctoral program. Also, I thank Debbie Sickles and Donna LeBlond for their administrative help.
There are still more people to thank. Dr. Glen Cameron, Dr. Fritz Cropp, Dr. Won Ho
Chang, Dr. Sekang Kim, and Prof. Henry Hager – they helped open my eyes to the world of
iv
advertising research during the years of my Master’s program at the University of Missouri –
Columbia. Federico de Gregorio, Yongjun Sung, and Dr. Doyle Yoon – collaboration with them
served as invaluable experience for improving my research skills. Also, Jung-Suk Han – he
motivated me to study advertising in academia. Without his initial encouragement, I would not
have been able to taste this world.
Finally, I would like to give my special thanks to Seung-Jin, my adorable son. He has
always been a primary reason why I have to achieve my goals. And he has always provided me
with the energy to achieve those goals. I believe that he truly contributed to the completion of
my dissertation just by being. Thanks, my son – your daddy did it!
v
TABLE OF CONTENTS
Page
ACKNOWLEDGEMENTS …………………………………………………………………iv
LIST OF TABLES ………………………………………………….…..………………… viii
LIST OF FIGURES ………………………………………………………………… ……… x
CHAPTER I.
INTRODUCTION ………………………………………………………………….…….…. 1
CHAPTER II.
CONCEPTUAL BACKGROUND …………………………………………………………. 5
1. Product Endorsements and Testimonial Advertising …………………………………. 5
2. Celebrity versus Noncelebrity Endorsers ………………………………………...…… 7
3. Source Credibility: Expertise and Trustworthiness …………………………………… 9
4. The Role of Source Credibility: An Elaborative Processing Perspective …………… 12
5. Product Involvement and Ad Involvement .…………………………………..………13
CHAPTER III.
THEORETICAL FRAMEWORK ………………………………………………...………. 16
1. Endorsement Motives: Application of Correspondent Inference Theory …………… 16
2. Dispositional Attributions versus Situational Attributions …………………..……… 18
3. Correspondence Bias: A Review of the Concept …………………………….……… 20
4. Antecedents of Dispositional and Situational Attributions …………………. ………23
5. Consequences of Dispositional and Situational Attributions ………………..……… 26
6. The Mediating Role of Dispositional and Situational Attributions ………….……… 29
vi
CHAPTER IV.
RESEARCH HYPOTHESES ……………………………………………….……………. 32
CHAPTER V.
METHODOLOGY ………………………………………………………………...……… 41
1. Overview …………………………………………………………………….……… 41
2. Conceptual Definitions of Key Variables …………………………………………… 42
3. Pre-Experimental Procedure: Pretests ………………………………………. ………43
4. Experimental Design, Sampling, and Procedure …………………………….……… 46
5. Dependent Measures ………………………………………………………...……… 54
CHAPTER VI.
RESULTS ………….……………………………………………………………..………. 58
1. Experiment I ………………………………………………………………………… 58
2. Experiment II ………………………………………………………………..……… 65
CHAPTER VII.
DISCUSSION …………………………………………………………………….……… 77
1. Summary of Research ……………………………………………………………… 77
2. Major Findings and Interpretations ………………………………………………… 78
3. Theoretical and Managerial Implications …………………………………..……… 88
4. Limitations and Future Research ………………………………………….……….. 94
BIBLIOGRAPHY ……………………………………………………………….………100
APPENDICES …………………………………………………………………..……… 108
APPENDIX A. TABLES …………………………………………………………109
APPENDIX B. QUESTIONNAIRES ..…………………………………..……… 125
APPENDIX C. STIMULUS ADVERTISEMENTS ……………………...…….. 151
vii
LIST OF TABLES
Page
Table 1: Independent Samples T-Tests: Manipulation Checks (Experiment I)…………….. 110
Table 2: Factor Analysis: Dispositional vs. Situational Attributions (Experiment I) ……… 110
Table 3: Independent Samples T-Tests: Source Credibility Effects on Attributions
(Experiment I) ……………………………………………………………………. 110
Table 4: Independent Samples T-Tests: Source Credibility Effects on Ad/Brand Evaluations
(Experiment I) ……………………………………………………………………. 111
Table 5: MANOCA: Effects of Source Credibility and Product Involvement on Attributions
(Experiment I) ……………………………………………………………………. 111
Table 6: Multiple Regression Analyses: Attribution Effects on Ad/Brand Evaluations
(Experiment I) ……………………………………………………………………. 112
Table 7: Hierarchical Regression Analyses (Step 4): Mediating Effects of Attributions
(Experiment I) ………………………………………………………………..……112
Table 8: Demographic Profiles of Sample (Experiment II) …………………………………113
Table 9: Chi-Square Tests: Group Equivalency Check (Experiment II) ……………………114
Table 10: One-Way ANOVA: Manipulation Check (Experiment II) ……………………… 115
Table 11: Factor Analysis: Dispositional vs. Situational Attributions (Experiment II) …….115
Table 12: Paired Samples T-Test: Dispositional vs. Situational Attributions
(Experiment II) ………………………………………………………………… 115
Table 13: Independent Samples T-Tests: Gender Effects (Experiment II) …………………116
Table 14: One-Way ANOVA: Age Effects (Experiment II) ………………………………. 116
Table 15: One-Way ANOVA: Source Credibility Effects on Attributions
(Experiment II) ………………………………………………………………... 117
Table 16: One-Way ANOVA: Source Credibility Effects on Ad/Brand Evaluations
(Experiment II) ………………………………………………………………... 117
viii
Table 17: MANCOVA: Effects of Three Independent Variables on Attributions
(Experiment II) ……………………………………………………………….. 118
Table 18: One-Way ANOVA: Product Involvement Effects on Attributions
(Experiment II) ……………………………………………………………….. 119
Table 19: One-Way ANOVA: Product Involvement Effects on Ad/Brand Evaluations
(Experiment II) ……………………………………………………………….. 119
Table 20: One-Way ANOVA: Ad Involvement Effects on Attributions
(Experiment II) ……………………………………………………………….. 120
Table 21: One-Way ANOVA: Ad Involvement Effects on Ad/Brand Evaluations
(Experiment II) …………………………………………………….…………. 120
Table 22: Multiple Regression Analyses: Attribution Effects on Ad/Brand Evaluations
(Experiment II) ……………………………………………………………….. 121
Table 23: Hierarchical Regression Analyses (Step 4): Mediating Effects of Attributions
(Experiment II) ……………………………………………………………….. 121
Table 24: Chi-Square Tests: Effects of Three Independent Variables on Correspondence
Bias (Experiment II) …………………………………………………………. 123
Table 25: Three-Way ANOVA: Effects of Three Independent Variables on Correspondence
Bias (Experiment II) …………………………………………………………. 124
Table 26: Independent Samples T-Tests: Effects of Correspondence Bias on Ad/Brand
Evaluations (Experiment II) …………………………………………………. 124
ix
LIST OF FIGURES
Page
Figure 1: Kelley & Michela’s (1980) Mediation Model of Attributions …………………. 30
Figure 2: A Path Diagram of the Mediating Effect ………………………………………. 31
Figure 3: A Proposed Mediation Model of Endorser-Generated Attributions …………… 38
Figure 4: Factorial Design for Experiment II ……………………….…............................. 49
Figure 5: Path Diagrams for Attribution Effects: Experiment I ……………………….…. 62
Figure 6: Hierarchical Regression Analyses (the first three steps): Experiment I ……….. 64
Figure 7: Path Diagrams for Attribution Effects: Experiment II …………………………. 72
Figure 8: Hierarchical Regression Analyses (the first three steps): Experiment II ………. 74
Figure 9: A Revised Mediation Model of Endorser-Generated Attributions …………….. 85
x
CHAPTER I
INTRODUCTION
Using a person to speak on behalf of the advertiser has been a popular advertising
technique over the years, evoking ongoing interest among both advertising practitioners and
scholars with regard to its effectiveness (Goldsmith, Lafferty & Newell, 2000; Joseph, 1982;
Petroshius & Kenneth, 1989; Till & Busler, 2000). Product endorsements by a spokesperson
(hereafter an endorser) are thought to build credibility, which generally influences
persuasiveness of the message. A testimonial is a common approach of such product
endorsements. In this type of advertising appeal, an endorser typically talks about positive
feelings or outcomes attained from his/her own experience with or belief in the advertised
product. As a consequence, the endorser is likely to be a frequent object of causal inferences in
that advertising readers might suspect whether or not the endorser’s claim about product benefits
is genuinely motivated.
This research investigates the circumstances in which perceived causality of the
endorser’s motivation – specifically, dispositional versus situational attributions – is generated
while consumers are exposed to testimonial advertising, also it considers how these attributions
influence message persuasiveness. Based on a review of literature, this study has developed a set
of hypotheses, which posit that consumers will generate predictable patterns of attributional
responses to product endorsements in testimonial advertising that vary in terms of endorser
credibility, product involvement, and ad involvement. In addition, they posit that such
attributional responses in turn will influence ad and brand evaluations, playing a mediating role
1
in the relationships between attributional antecedents and consequences. Despite a number of
studies on consumer responses to various aspects of endorser presence (e.g., Erdogan, Baker &
Tagg, 2001; Kamins, 1989; Ohanian, 1991; Petroshius & Kenneth, 1989; Till & Busler, 2000;
Tripp, Jensen & Carlson, 1994), these specific issues have received little attention.
The hypothesized propositions in the present research are based upon several theoretical
perspectives. First, persuasion models such as the source credibility model, the elaboration
likelihood model, and the heuristic-systematic processing model provide base for the
assumptions that endorser credibility and personal involvement (with the product class and the
advertisement) would influence the attributional process of a testimonial advertisement as well as
the persuasive effects of the advertisement. Second, correspondent inference theory, a specific
attribution model, leads to the hypotheses regarding the differential impacts of two modes of
attributions – dispositional and situational – on the indicators of message effectiveness. Finally,
the correspondence bias perspective and empirical research testing the concept (e.g., Cronley et
al., 1999; Sorum, Grape & Silvera, 2003) support the possible occurrence of dispositional
attributions even when message receivers recognize one or more situational forces for the
endorsement. The use of these multiple theoretical perspectives is considered essential for this
research to predict the intricate processes by which product endorsements affect consumers’
attitudinal and motivational responses.
Two experiments were administered to test the proposed hypotheses with the samples of
college students (Experiment I) and consumer panelists (Experiment II). Due to the complicated
nature of the research design, each stage of the experiments was designed to complementarily
apply to the hypothesis tests or contribute to enhancing reliability through repeated
measurements. To select appropriate endorser types and product classes that would be used for
2
the experimental stimuli (i.e., testimonial advertisements created by the researcher), a series of
pretests were conducted before the experiments. This procedure was to ensure high levels of
internal and measurement validity and reliability, which are critical for the rigorous assessment
of hypothesized propositions.
The present study extends endorsement research in two important ways. First, although
attribution perspectives have been used in several endorsement studies (e.g., Choi, 2002; Cronley
et al., 1999; Kamins, 1989; Hunt, Kernan & Mizerski, 1983; Tripp, Jensen & Carlson; 1994),
they have usually focused on addressing direct relations between product endorsements and
ad/brand evaluations. On the other hand, this research attempts to extensively assess the role of
the attributional process that accounts for these associations. Testing the mediating effect of
causal inferences is important because a consumer’s evaluative responses to a product
endorsement are likely to be thoroughly explained by a more complex perceptual mechanism
than illustrated by the simple “attributions to attitude change” path.
Second, whereas prior research has typically examined consumer responses to celebrity
endorsements in advertising (e.g., Atkin & Block, 1983; Goldsmith, Lafferty & Newell, 2000;
Kahle & Homer, 1985; Ohanian, 1991; Tripp, Jensen & Carlson, 1994), this study is concerned
with testimonial messages conveyed by noncelebrity endorsers. Due to the different
characteristics between celebrity and noncelebrity endorsers – such as attractiveness, familiarity,
and similarity – perceived by message receivers, findings restricted to celebrity endorsement
studies are considered to have limited implications and generalizability for consumer processing
of endorsement messages and its consequences. Accordingly, this research is a meaningful
attempt to better understand how people generate attributional responses to product
endorsements and how these attributional responses affect persuasion in a broad scope of
3
endorsement situations. These efforts are also expected to produce managerial implications with
respect to appropriate selections of product endorsers and testimonial messages.
4
CHAPTER II
CONCEPTUAL BACKGROUND
This chapter reviews the literature regarding product endorsements, source credibility,
and personal involvement. Testimonial advertising is defined first, and then different aspects of
celebrity and noncelebrity endorsements are compared. This is followed by conceptual reviews
of source credibility and personal involvement, which are assumed to function as attributional
antecedents. Relevant theoretical perspectives and empirical findings are appraised to support
predictions about those antecedent conditions of attribution evocation.
1. Product Endorsements and Testimonial Advertising
Product endorsements have long and frequently been used by advertisers for the purpose
of influencing consumers’ attitudes and purchase intentions (Friedman, Termini & Washington,
1976; Homer & Kahle, 1990; Kamins, 1989). A common belief is that endorsements in
advertising can be a powerful tool because they generate immediate trust among consumers, who
look for clues to make them confident of their purchase decisions (Erdogan, Baker & Tagg,
2001; Shimp, 2002). Consumers tend to believe what others say about a product more than what
an advertiser says about its own product. As a result, testimonials by someone else, such as
celebrities, experts, and other regular consumers, are deemed to offer greater credibility than
self-proclamations (Clow & Baack, 2001).
The term testimonial is often confused with endorsement. The Federal Trade
Commission (FTC) treats testimonials and endorsements identically in the context of its
enforcement of the FTC Act. According to its provision regarding this part of the law, a
5
testimonial means “any advertising message which message consumers are likely to believe
reflects the opinions, beliefs, findings, or experience of a party other than the sponsoring
advertiser” (§255.0[b]). In contrast with the FTC’s view for its regulatory function, however, it
is more accurate to regard testimonials as one specific type of endorsement. As indicated in the
American Advertising Federation (AAF)’s platform on advertising ethics and principles,
testimonial advertising limits its definitive scope to involving verbal statements spoken by an
endorser (or a group of endorsers) who is (are) reflecting a real and honest opinion about or
experience with the product being advertised.
A general approach of testimonial advertising is to persuade consumers that the
advertised product is capable of delivering the claimed benefits and to provide reassurance that
others (e.g., endorsers) have made the same decision and have been satisfied with the results.
Like the effect of word-of-mouth recommendations, therefore, this approach can be especially
effective when a consumer is initially uncertain about the benefits of the product. Testimonial
strategies aim to lead target audience to have positive attitude toward the advertised product, and
it is generally achievable when the message claimed by the endorser is believed (Batra, Myers &
Aaker, 1996). Thus, it is likely that the persuasive effects of a testimonial advertisement are in
part the consequence of message believability as perceived by the target audience.
Testimonial advertising, however, may often confront legal and/or ethical challenges in
relation to truthfulness of messages conveyed by the endorser. Turning back to the FTC’s
definition of a testimonial, “reflection of opinions, beliefs, findings, and experience” indicates
how an endorser’s claim in testimonial advertising should be phrased. That is, the endorser in
testimonials must provide supportive claims regarding the advertised product based upon his/her
belief in and expertise or personal experience with the product. Further, these claims must be
6
verifiable otherwise the message can be regarded as deceptive (Wells, Burnett & Moriarty, 2002).
Despite this normative nature, however, the believability of testimonial messages is often
questionable. This is because false claims or misleading testimonials in advertising may be
common, subsisting away of regulatory scrutiny. As a relevant illustration, a recent report by the
FTC (2002) indicates that nearly 50% of the sampled testimonial advertisements for weight-loss
products have claimed that users can lose weight without dieting or exercising, and most of those
advertisements are very likely false or misleading.
2. Celebrity versus Noncelebrity Endorsers
Numerous studies have shown that a communicator’s characteristics significantly affect
message persuasiveness (e.g., Atkin & Block, 1983; Erdogan, Baker & Tagg, 2001; Hass, 1981;
Ohanian, 1990; Sternthal, Dholakia & Leavitt, 1978). Thus, selecting an appropriate endorser
for a brand is always an important decision for advertisers. In testimonial advertising, it is a
frequent practice to employ celebrities, such as well-known actors, entertainers, and athletes, due
to their familiarity and attention-gaining power. Industry professionals and general consumers
are also often chosen as product endorsers because of their perceived expertise with the product
category advertised or similarity to the target audience.
Perhaps the most common type of endorsement advertising involves a celebrity endorser
(Erdogan, Baker & Tagg, 2001; Ohanian, 1991; Tripp, Jensen & Carlson; 1994). A recent study
estimates that around one quarter of all television commercials screened in the U.S. include
celebrity endorsers (Shimp, 2002). The term celebrity refers to “an individual who enjoys public
recognition and who uses this recognition on behalf of a consumer good by appearing with it in
an advertisement” (McCracken, 1989, p.310). As implied from this definition, celebrities’ high
profiles, popularity, and personalities lead advertisers to expect positive consumer responses to
7
their product. For this reason, it is believed that celebrity endorsers help brands stand out from
the surrounding clutter, thus improving brand image and enhancing the persuasive effects of
marketing communications (Atkin & Block, 1983; Erdogan, Baker & Tagg, 2001).
As well as celebrities, noncelebrity endorsers are often used for testimonial advertising.
Unlike celebrity endorsers, however, no clear definition of a “noncelebrity endorser” exists. As
the term indicates (exclusively opposite to “celebrity”), a noncelebrity is generally considered as
any person who is unknown or little known to the public. Common types of noncelebrity
endorsers used for testimonial advertising include typical consumers, experts, and CEOs (some
experts and CEOs can be celebrities depending on the level of public awareness) (Friedman,
Termini & Washington, 1976; Kamins, 1989; Stephens & Faranda, 1993).
Typical consumers are often used as product endorsers in testimonial advertising because
of their personal and/or demographic similarity to message recipients (Ohanian, 1990; Sorum,
Grape & Silvera, 2003). This type of endorser is expected to induce a heightened sense of
familiarity among target consumers, thus having referent power and being seen as credible (Hass,
1981; Sorum, Grape & Silvera, 2003). Several studies have shown that people tend to evaluate
similar others more favorably than dissimilar others and that others who bear a resemblance to
oneself are generally regarded as more competent (e.g., Howard & Ferris, 1996; Lancaster,
Royal & Whiteside, 1995).
Experts are employed as endorsers for their perceived credibility and authority regarding
recommendations, which are believed to come from their professional knowledge of the product
class advertised (Friedman, Termini & Washington, 1976). For example, doctors, lawyers, and
athletes can be considered as appropriate endorsers for products or services related to their
respective professions. However, product endorsements by some experts, such as physicians,
8
dentists, and pharmacists, tend to be regulated by Federal-level authorities, such as FTC and
FDA, or occupational codes of behavior (Wells, Burnett & Moriarty, 2002).
CEOs, persons at the top of sponsoring organizations, sometimes can be an effective
endorser choice because they are presumably perceived as knowledgeable, competent, and able
to make reliable claims about their company or brand (Stephens & Faranda, 1993). However, as
Stephens and Faranda’s (1993) study implies, consumers may not see CEOs as credible
endorsers due to their extreme bias in favor of their own brand.
Despite ongoing and widespread uses of noncelebrity endorsers in advertising, most
endorsement studies have focused on celebrity endorsers (e.g., Atkin & Block, 1983; Choi, 2002;
Erdogan, Baker & Tagg, 2001; Ohanian, 1990, 1991; Till & Busler, 2000; Tripp, Jensen &
Carlson, 1994), whereas only several studies have explored communication effects of
noncelebrity endorsers (e.g., Kamins, 1989; Sorum, Grape & Silvera, 2003; Stephens & Faranda,
1993). While some researchers have attempted to compare the differential effects between
celebrity and noncelebrity endorsers, their results have been mixed. For instance, Friedman,
Termini, and Washington (1976) found no significant differences among four types of endorsers
– celebrity, typical consumer, expert, and company president – on believability measures; while
a later study by Friedman and Friedman (1979) found that a celebrity endorser would lead to
higher believability and more favorable brand attitude than a noncelebrity endorser. In any case,
further exploration of noncelebrity endorsements is demanded, considering relatively scant
investigations of this issue and different source characteristics of celebrities and noncelebrities.
3. Source Credibility: Expertise and Trustworthiness
Whether an endorser in testimonial advertising is a celebrity or noncelebrity, message
persuasiveness is likely to be in part dependent upon the endorser’s characteristics (Hass, 1981;
9
Homer & Kahle, 1990; Sternthal, Dholakia & Leavitt, 1978). In his milestone article on the
effects of source characteristics, Hass (1981) defined three sets of characteristics that affect the
persuasive impact of a message when they are attributed to its source (i.e., endorser or
spokesperson): the source’s credibility, attractiveness, and power. Although these three
characteristics may be combined in most natural situations (Hass, 1981), they have been
considered and treated as conceptually distinct (e.g., Erdogan, Baker & Tagg, 2001; Hass, 1981;
Ohanian, 1990; Till & Busler, 2000). Among the three characteristics, source credibility has
received considerable attention in the advertising literature (e.g., Goldsmith, Lafferty & Newell,
2000; Gotlieb & Sarel, 1991; Homer & Kahle, 1990; Joseph, 1982; Ohanian, 1990; Sternthal,
Dholakia & Leavitt, 1978; Till & Busler, 2000).
Source credibility refers to “the extent to which the source is perceived as possessing
expertise relevant to the communication topic and can be trusted to give an objective opinion on
the subject” (Belch & Belch, 1994, pp.189-190). Accordingly, this dimension implies a
communicator’s positive characteristics that affect the receiver’s acceptance of a message
(Ohanian, 1990). Advertising and marketing researchers have expanded the dimensions of the
source credibility construct. For example, Applbaum and Anatol (1972) considered source
credibility as encompassing the dimensions of trustworthiness, expertness, dynamism, and
objectivity. DeSarbo and Harshman (1985), on the other hand, included attractiveness and
likability in the source credibility dimensions, while Wynn (1987) identified believability and
sociability as other dimensions of source credibility. Although there has been inconsistency
among researchers in the dimensions that source credibility is considered to comprise (see
Ohanian [1990] for a review), it is most common to view expertise and trustworthiness as the
10
two primary dimensions of source credibility (e.g., Goldsmith, Lafferty & Newell, 2000; Gotlieb
& Sarel, 1991; Homer & Kahle, 1990; McGinnies & Ward, 1980).
In their early study, Hovland, Janis, and Kelley (1953) analyzed the factors leading to
perceived credibility of the communicator, and identified expertness (expertise) and
trustworthiness as those that underscore the concept of source credibility. They defined
expertise as “the extent to which a communicator is perceived to be a source of valid assertions,”
and trustworthiness as “the degree of confidence in the communicator’s intent to communicate
the assertions he considers most valid.” These definitions are still widely used in the source
credibility literature (e.g., Choi, 2002; Erdogan, Baker & Tagg, 2001; Ohanian, 1990). Similarly,
Gotlieb & Sarel (1991) defined expertise as the receiver’s perception of the source as a
knowledgeable person, while trustworthiness is viewed as the receiver’s perception that the
source’s opinions are unbiased. From these definitions, it can be recognized that these two
dimensions (1) refer to what receivers perceive the source as having; (2) are concerned with the
source’s knowledge of the subject (expertise) and honesty or believability of the claim
(trustworthiness); and thus (3) may not always be internally consistent (e.g., a person who has
expertise in a subject may not be trustworthy, and vice versa).
Empirical evidence suggests that source expertise and trustworthiness influence attitude
change and product evaluations (see McGuire [1985] for a review). Crisci and Kassinove’s
(1973) early experimental study indicated that respondents’ compliance with the source’s
recommendations is positively associated with the perceived level of expertise. Studies have
also supported the positive combined effects of expertise and trustworthiness on attitude change.
For example, McGinnies and Ward (1980) found that a communicator who is perceived to be
both an expert and trustworthy would generate the greatest opinion change. Although a
11
significant relationship between expertise or trustworthiness and persuasion has not been fully
supported by several empirical studies (e.g., Johnson & Scileppi, 1969; McGarry & Hendrick,
1974; Mcginnies, 1973), the source credibility literature generally suggests that a highly credible
source tends to induce more positive attitude toward the position advocated by the source and to
encourage more behavioral change than a less credible source (e.g., Goldsmith, Lafferty &
Newell, 2000; Sternthal, Phillips & Dholakia, 1978; Ohanian, 1991).
4. The Role of Source Credibility: An Elaborative Processing Perspective
Source credibility has been incorporated into the two major theoretical perspectives
concerning information processing of persuasive messages: the elaboration likelihood model
(e.g., Gotlieb & Sarel, 1991; Homer & Kahle, 1990; Petty & Cacioppo, 1986; Petty, Cacioppo &
Goldman, 1981) and the heuristic-systematic processing model (e.g., Chaiken, 1980; Chaiken,
Liberman & Eagly, 1989; Chaiken & Maheswaran, 1994). Both theories propose two distinct
routes to attitude change: central route (or systematic processing) and peripheral route (or
heuristic processing). An underlying dimension that differentiates these routes is involvement,
which means personal relevance with a particular issue or product (Chaiken, 1980; Petty &
Cacioppo, 1981; 1986; Zaichkowsky, 1986). That is, according to these dual-process models,
under high involvement situations (i.e., when an object under consideration has a high degree of
personal relevance to the recipient), people tend to form or change their attitudes by actively
attending to and cognitively elaborating on issue-relevant arguments. On the other hand, under
low involvement situations (i.e., when the personal relevance with the object is trivial or low),
people tend to form or change their attitudes by invoking peripheral or heuristic cues.
Source credibility – as well as source attractiveness, music, and mood generated by the
persuasive message – has been considered as a peripheral or heuristic cue because it is a non-
12
content element perceived by individual receivers and not directly associated with the message
arguments (Chaiken & Maheswaran, 1994; Petty, Cacioppo & Goldman, 1981). Thus, the
construct of source credibility has been posited to influence persuasion via the peripheral route or
heuristic processing under low involvement situations. Empirical findings suggest that source
credibility has a direct attitudinal impact under low involvement conditions (e.g., Chaiken &
Maheswaran, 1994; Petty & Cacioppo, 1981; Petty, Cacioppo & Goldman, 1981) and even under
limited conditions of high involvement (e.g., Chaiken & Maheswaran, 1994).
Homer and Kahle’s (1990) experimental study, however, yielded an inconsistent result
regarding the role of source credibility in the persuasive environment, suggesting that source
credibility can act as a central cue, rather than a peripheral cue, under both high and low
involvement conditions. That is, according to their result, the receivers of persuasive messages
may treat source credibility as equivalent to or as a surrogate for central message arguments in
certain situations (Homer & Kahle, 1990). This assumption seems to be plausible, especially for
situations in which consumers are exposed to and process a testimonial advertisement because of
the direct, inseparable connection between the source (i.e., endorser) and the central argument
(i.e., testimonial message) in this particular type of advertising context. Consequently, the
construct of source credibility needs to be further investigated as an important determinant of
persuasive effects, while not simply assuming that this property of the message source will be
peripherally processed.
5. Product Involvement and Ad Involvement
According to the dual process models, personal involvement may influence the role of
source credibility in the persuasive communication context. Involvement, however, is not a
simple construct that can be precisely and universally conceptualized across various situations
13
and different domains of research. Indeed, conceptualization and measurement of involvement
has long been an important and controversial issue among communication and marketing
researchers (e.g., Andrews, Durvasula & Akhter, 1990; Day, Stafford & Camacho, 1995;
Mitchell, 1981; Zaichkowsky, 1985; 1986; 1994).
A general definition of involvement, which is applicable to most conceptualizations of
involvement, is “a person’s perceived relevance of the object (e.g., an issue, a product class, an
advertisement) based on inherent needs, values, and interests” (Zaichkowsky, 1985, p.342).
Specifically, involvement is considered as a motivational state – not a generalized trait or a
process – influenced by such perceived relevance, as defined by several researchers (e.g.,
Andrews, Durvasula & Akhter, 1990; Day, Stafford & Camacho, 1995; Mitchell, 1981). Unlike
the common dichotomous operationalization of the involvement construct (e.g., Chaiken, 1980;
Petty & Cacioppo, 1981), a person’s motivational state is, in fact, likely to lie on a continuum
though the continuum is hard to measure in operational terms (Andrews, 1988; Andrews,
Durvasula & Akhter, 1990; Petty and Cacioppo, 1986). Also, involvement as a perceived level
of personal relevance will vary by a given object (or stimulus) or a specific situation.
In her comprehensive review of the advertising and marketing literature using the
involvement construct, Zaichkowsky (1986) identified three streams of involvement research,
which were categorized in terms of the object of involvement: (1) involvement with the product
class; (2) involvement with the advertisement; and (3) involvement with purchase decisions.
Among these domains of involvement, the first two types of involvement – involvement with the
product class (hereafter product involvement) and involvement with the advertisement (hereafter
ad involvement) – have been most frequently incorporated and discussed in advertising research
(see Andrews, Durvasula & Akhter [1990] and Zaichkowsky [1986] for the review). Although
14
both dimensions share the same underlying theme focusing on personal relevance with the object
(Zaichkowsky, 1986), they should not be treated as identical constructs when conducting
consumer research under the relevant topic. It is because they may have different attributes or
tendencies with respect to motivation stability (i.e., enduring vs. situational involvement),
evaluative processing, antecedent conditions, and resulting attitudes and behaviors of
involvement (Andrews, Durvasula & Akhter, 1990; Batra & Ray, 1985; Day, Stafford &
Camacho, 1995; Lee, 2000; Mitchell, 1981; Zaichkowsky; 1985; 1986).
In many cases, product involvement precedes ad involvement, influencing consumers’
cognitive processing and attitude formation (Day, Stafford & Camacho, 1995). As such, product
involvement leads to message-processing motivation (Gill, Grossbart & Laczniak, 1988; Michael,
1994; Mitchell, 1981; Petty, Cacioppo & Schumann, 1983), which is fundamentally related to ad
involvement. Product involvement, however, will not always positively influence ad
involvement because of the confounding effects of other potential antecedents of involvement,
such as the receiver’s characteristics, physical characteristics of the stimulus, and the varying
situation (Zaichkowsky, 1986). In addition, while product involvement is relatively enduring, ad
involvement represents an internal and situational state evoked by an advertising message at a
particular point of time (Batra & Ray, 1985; Lee, 2000). Consequently, although both product
involvement and ad involvement can determine information processing and message
persuasiveness directly or indirectly, they need to be conceptually distinguished and thus
measured separately to yield more valid and precise results of involvement effects.
15
CHAPTER III
THEORETICAL FRAMEWORK
This chapter provides a fundamental frame that underlies predictions for hypothesis tests.
Based on a comprehensive review of psychology and advertising literatures, theoretical and
empirical propositions and concepts are introduced. Correspondent inference theory, proposed
and developed by social psychologists, is reviewed first, as well as an evaluation of the theory’s
applicability to the context of noncelebrity testimonial advertising. Following the review,
antecedents, consequences, and mediating effects of causal inferences are constructed for their
applied assessments. Correspondence bias, a sophisticated attribution-related concept, is
additionally reviewed to test its evocation in the endorsement context.
1. Endorsement Motives: Application of Correspondent Inference Theory
Endorsers in testimonial advertising are supposed to be credible so that advertisers can
convince their target consumers of positive experience with and benefits of the product. That is,
the information value of the endorser’s message may be valid only to the extent that consumers
believe the endorsement to be genuine (Sorum, Grape & Silvera, 2003). Consumers may suspect
whether the endorser truly recommends the product or acts primarily for money or other external
benefits. If consumers are suspicious of the endorser’s motives, such as perceiving the
endorser’s lack of true belief in the product, they will be less receptive to persuasion (Cronley et
al., 1999; Settle & Golden, 1974; Smith & Hunt, 1978; Sorum, Grape & Silvera, 2003). Due to
this responsive nature of the testimonial approach, attribution perspectives appear to be useful in
16
understanding consumer perceptions of the endorser’s reasons for communicating (e.g., I believe
the endorser is supporting the brand because…).
Attribution theory, developed largely by social psychologists, seeks to describe the
cognitive process involved when an individual assigns an observable event to its underlying
causes (Forkes, 1988; Heider, 1958; Kelley, 1973; Kelley & Michela, 1980; Settle & Golden,
1974). According to this theory, causal analyses are inherent in an individual’s need to
understand social events, such as why another person communicates as he/she does (Heider,
1958; Jones & Davis, 1965; Kelley, 1967). The basic premise of attribution theory is that an
observer’s response to the actor’s behavior depends largely on how the observer views the causes
of the behavior (Bemmels, 1991; Heider, 1958; Kelley, 1967). Accordingly, a consumer’s
attributions concerning why the communicator (e.g., endorser) takes a particular position in a
message has an important impact on how the consumer responds to the message (Gotlieb & Sarel,
1991). If a consumer is engaged in such an attributional process regarding true endorsement
motives, the rationale behind the product endorsement might be trusted or flawed, thus further
affecting persuasive effectiveness of the message (Sorum, Grape & Silvera, 2003).
The correspondent inference theory, a specific attribution model, is more suitable to
explain or predict a consumer’s attributional process in a communication context than the
general attribution theory (e.g., Cronley et al., 1999; Smith & Hunt, 1978; Sorum, Grape &
Silvera, 2003). This theory, proposed by Jones and Davis (1965), is concerned with particular
types of attributions rather than causal inferences in general (Smith & Hunt, 1978). Specifically,
the theory addresses the situations in which a person naively attributes an event either to actual
dispositions of the actor (e.g., real affection toward the product) or to situational constraints (e.g.,
money). A correspondent inference will be made when a person attributes an event to the
17
actor’s dispositions (internal causes). The term “dispositional attribution” has been used to
represent this kind of causal inference (Gilbert & Malone, 1995; Jones & Davis, 1965; Sorum,
Grape & Silvera, 2003). On the other hand, if a person attributes an event to environmental or
situational factors (external causes), his/her causal inference will be noncorrespondent because
there is no perceived relationship between the event and the actor’s actual dispositions (Jones &
Davis, 1965; Smith & Hunt, 1978). This condition points to a “situational attribution,” as
conceptually opposed to the dispositional attribution (e.g., Bierbrauer, 1979; Kerber & Singleton,
1984; Pilkonis, 1977). An in-depth exploration of both types of attributions is necessary because
they lie at the center of the correspondent inference theory.
2. Dispositional Attributions versus Situational Attributions
Although attribution theory suggests that people have the potential to generate multiple
and interactive attributional responses (Laczniak, DeCarlo & Ramaswami, 2001), it has been
widely accepted to view causal inferences as two-sided, namely dispositional and situational, as
the correspondent inference theory proposes. As indicated earlier, a dispositional attribution
refers to the observer’s assumption that the actor’s behavior reflects his/her internal disposition,
such as personal traits, attitude, and belief, whereas a situational attribution means the observer’s
assumption that the actor’s behavior has been caused by situational constraints (Gilbert &
Malone, 1995; Jones & Davis, 1965). By implication, a dispositional attribution indicates the
tendency for observers to underestimate the influence of situational forces on a behavior and to
overestimate dispositional causes, while a situational attribution specifies the opposite tendency
(Bierbrauer, 1979).
An important issue drawn here is whether dispositional and situational attributions are
mutually exclusive of each other. Kelley’s (1967) attribution model seems to imply that both
18
types of attributions are negatively correlated: i.e., “if one can point to dispositional causes for a
behavior, then presumably one is left with less to say about situational influences” (Pilkonis,
1977, p.269). Should this assumption be true, the attribution process then will conform to a
“zero-sum” model of dispositional versus situational causations of behavior (Bierbrauer, 1979).
However, such an inverse relationship between the two types of attributions has been questioned
by several researchers (e.g., Bierbrauer, 1979; Pilkonis, 1977; Storms, 1973). For example, in
his study regarding the impact of consensus information on causal inferences, Pilkonis (1977)
found no significant psychological connection between dispositional and situational attributions,
suggesting that a person may be able to assess dispositional influences without implying
anything about his/her opinion of situational pressures, and vice versa. Accordingly, it may be
appropriate to presume that dispositional and situational attributions will be independently made
in consumers’ mind.
Although both dispositional and situational attributions have received academic attention
across disciplines (e.g., Bierbrauer, 1979; Jones & Davis, 1965; Pilkonis, 1977; Sorum, Grape &
Silvera, 2003), most of the research have been more concerned with dispositional attributions
than situational attributions. Among them, two studies are noteworthy in relation to the interest
of this research. First, Cronley and her colleagues (1999) attempted to apply the perspective
from the correspondent inference theory to evaluating celebrity-endorsed advertising. Their
experimental study treated the subjects’ causal inferences by manipulating paid versus nonpaid
conditions on a selected celebrity endorser (Cindy Crawford). The results revealed that even in
the paid condition (i.e., when a celebrity’s endorsement is highly constrained by situational
factors), consumers might make dispositional inferences about the celebrity’s attitude and
preferences for the endorsed brand. In short, Cronley et al.’s study (1999) suggests the limited
19
impact of the situational inferences on the effectiveness of celebrity-endorsed advertisements,
enabling the conclusion that this type of advertising can be successful in general despite large
endorsement fees.
Sorum, Grape, and Silvera (2003), on the other hand, investigated the attitudinal effect of
peer (i.e., typical consumer) endorsed advertising based on the correspondent inference theory.
To the researcher’s knowledge, their study is the only attempt as yet to apply the attributional
perspective to the perceptual dimension of noncelebrity endorsements. Replicating the
experimental frame constructed by Cronley et al. (1999), they tested the associations between
causal inferences concerning a peer endorser’s motives and receivers’ product evaluations. The
results were not consistent with those obtained in Cronley et al.’s (1999) study: that is, unlike the
case of celebrity endorsements, the participants in their study did not make correspondent
inferences regarding the peer endorser’s preferences when they knew she was paid.
Empirical findings and theoretical implications from the two studies suggest the need to
further explore the actual consequences of correspondent inferences within the different context
of product endorsements. This effort requires first an understanding of the concept of
correspondence bias, a psychological process derived from the correspondent inference theory.
In both studies reviewed above, the effectiveness of celebrity or noncelebrity endorsements was
predicted in terms of the hypothesized impact of this perceptual variable.
3. Correspondence Bias: A Review of the Concept
Correspondence bias refers to “the tendency to draw inferences about a person’s unique
and enduring dispositions from behaviors that can be entirely explained by the situations in
which they occur” (Gilbert & Malone, 1995, p.21). Thus, it is believed that people with the
correspondence bias tend to favor dispositional attributions for other’s behavior (Boven, Kamada
20
& Cilovich, 1999; Jones & Davis, 1965). Here correspondence bias is often mixed with the
meaning of a dispositional attribution. To put it specifically, however, it is appropriate to view
correspondence bias as the psychological propensity by which the observer exaggeratedly makes
dispositional attributions – that may be correct or incorrect by occasions – from the actor’s
behavior (Gilbert & Malone, 1995; Sorum, Grape & Silvera, 2003).
The concept of correspondence bias has been extensively discussed by social
psychologists since Jones and Harris (1967) first empirically examined Jones and Davis’ (1965)
correspondent inference theory (see Gilbert & Malone [1995] for a comprehensive review of the
historical background of correspondence bias). In their classic experiment, Jones and Harris
(1967) predicted and tested how observers would make dispositional attributions about an actor
if the observers knew the constraints on the actor’s behavior. Test participants were exposed to
essays that reflected either a pro- or anti-position on Cuba’s leader, Fidel Castro, and then they
were told that the essay’s author had been free to choose his written position (group 1) or had
been instructed by the debate coach to defend an assigned position (group 2). Participants’
attribution of the essay writer’s true attitude was inconsistent with the prediction by Jones and
Davis’ (1965) normative theory, which posited that if observers knew certain situational
constraints on the actor’s behavior, they would be less inclined to make dispositional attributions
about the actor. Jones and Harris’ (1967) found that even if participants knew the writer’s
position was assigned, they regarded that the writer actually believed what he/she had written.
After the finding, this tendency was labeled as a bias. (As reviewed earlier, Cronley et al.’s
[1999] study of celebrity endorsements yielded a consistent result with this finding.)
As did Jones and Harris’ (1967) experiment participants, consumers may often make
dispositional attributions to a larger extent than is logically justified. Thus, correspondence bias
21
can occur even when situational constraints are obvious to the observer (Gilbert & Malone,
1995; Sorum, Grape & Silvera, 2003). This proposition is well applied to the context of product
endorsements. In such a situation, consumers play the role of observers and an endorser
becomes the actor of behavior (i.e., endorsement). Dispositional attributions here signify
consumers’ tendencies to regard the endorsement as originating from the endorser’s true belief in
the advertised product, while a monetary incentive (model fees) for the endorser represents a
typical situational constraint in this context (e.g., Cronley et al., 1999; Sorum, Grape & Silvera,
2003). As Cronley et al. (1999) stated, most consumers today are so savvy that they seem to
know that endorsers (especially celebrities) in advertising are well-paid for their endorsement;
therefore, they would not be so sensitive to the situational constraints of the endorser’s behavior.
In the light of correspondence bias, this can cause receivers to assume that the endorser actually
has had positive experience with and/or attitude toward the endorsed product and that the act of
endorsing is based on his/her genuine affection for the product (Cronley et al., 1999; Sorum,
Grape & Silvera, 2003).
In summary, correspondence bias is a sophisticated concept derived from the
correspondent inference theory, and its presence is a consequence of complex relations among
the actor (his/her behavior and perceived dispositions), the observer, and situational forces
(Gilbert & Malone, 1995). Empirical findings suggest that the evocation of correspondence bias
may differ by the actor’s characteristics (e.g., Cronley et al., 1999; Sorum, Grape & Silvera,
2003), the actor’s behavioral option (e.g., Jones & Harris, 1967; Ross, Amabile & Steinmetz,
1977), and the observer’s awareness or recognition of situational forces (Gilbert & Malone,
1995). Due to their conceptual and operational similarities, it is assumed that correspondence
bias will play a similar role to that of dispositional attributions in persuasive effects. Strictly
22
speaking, however, the test of correspondence bias is a more sensitive measure than that of
dispositional attributions because correspondence bias premises the condition in which the
observer recognizes the situational constraints. Hence, to rigorously assess the correspondent
inference theory, the occurrence and impact of correspondence bias must be explored as well
when the causes and consequences of dispositional attributions are investigated.
4. Antecedents of Dispositional and Situational Attributions
The following theoretical issues are centered on antecedents and consequences of
dispositional and situational attributions. What factors might cause the evocation of
dispositional and situational attributions? And what outcomes might result from the activation of
each type of attribution? These issues represent one of the most important domains in this
research.
A significant portion of attribution research deals with how people make causal
inferences (Folkes, 1988; Green et al., 1985). According to Kelley and Michela’s (1980), there
are three types of antecedents of attributions: information, beliefs, and motivation. First,
information about a behavior, such as how frequently it occurs and with what other behaviors it
covaries, forms the basis for attributional situations (Folkes, 1988; Kelley & Michela, 1980).
Kelley (1967) suggested that how people use information influences how they make causal
inferences, depending on consensus (with other’s responses), distinctiveness (comparing with
other behaviors), and consistency over time and modality (of an individual’s response) (Folkes,
1988; Sparkman & Locander, 1980). Furthermore, consensus information (i.e., information
about how most people behave) or information about the consequences of alternative behaviors
can be used to make dispositional or situational inferences about the intention behind a particular
behavior (Jones & Davis, 1965; Kelley & Michela, 1980; Pilkonis, 1977).
23
Second, people’s beliefs about a behavior’s causes and effects, including the suppositions
about its causes and the expectations about its effects, may lead to causal inferences (Kelley &
Michela, 1980). As a result, presumably the observer’s beliefs (or expectations) about the effects
associated with the actor or about the likelihood of a behavior’s occurrence in a particular
situation (e.g., Kelley & Michela, 1980) may affect the evocation of dispositional or situational
attributions. Prior beliefs often interplay with information in their influence on the attributional
process (Folkes, 1988). In such a case, beliefs tend to have more weight than information, as
revealed in several experiments (e.g., Sparkman & Locander, 1980; Yalch & Yoshida, 1983). It
may only rarely happen, therefore, that the processing of information proceeds without some
influence from the observer’s beliefs about causes and effects (Kelley & Michela, 1980).
Finally, people’s motivation, elicited by a behavior’s expected consequences, is thought
to influence the processing of information about the behavior and thus induces certain
attributions (Folkes, 1988; Kelley & Michela, 1980). Research demonstrates that motivational
factors, such as hedonic relevance (e.g., Jones & Davis, 1965), personal relevance (e.g., Higgins,
Kuiper & Olson, 1980), self-esteem (e.g., Folkes, 1988; Kelley & Michela, 1980), the need for
control (e.g., Heider, 1958; Kelley & Michela, 1980), and the desire to predict (e.g., Heider,
1958), can instigate the attributional process. In view of that, a person’s interests and needs
might be closely relevant to and entangled with the attribution process, thus affecting when the
person will become motivated to make attributions and when he/she will prefer to arrive at
certain attributions – like dispositional or situational attributions – rather than others (Kelley and
Michela, 1980).
Although most attribution research has defined a problem in terms of only one type of
these antecedents (Folkes, 1988), they are pertinent to frame the general model regarding the
24
causal relation between attributions and their antecedents. In addition, since Kelley and
Michela’s (1980) classification of attributional antecedents is based upon what Jones and Davis
(1965) illustrated in their correspondent inference theory, it is likely that those types of
antecedents – information, beliefs, motivation – can be well applied to the case of dispositional
and situational attributions.
To frame this study, it is necessary to estimate if source credibility and product/ad
involvement, the proposed determinants of endorsement effects, can be rationally predicted as
antecedent variables for the evocation of dispositional or situational attributions in the context of
testimonial advertising. From Kelley and Michela’s (1980) perspective, is each factor explicitly
relevant to one of attributional antecedents (i.e., information, beliefs, motivation)? Because
Kelley and Michela’s (1980) antecedent model is described in general classificatory terms
(Green et al., 1985), the current research needs to verify and empirically assess specific,
measurable variables within the range of potential attributional antecedents.
Source Credibility as an Attributional Antecedent
To the best of the researcher’s knowledge, no empirical studies have explicitly assumed
and tested a communicator’s credibility as an antecedent variable of attributions. Thus, a new
approach to reasoning with this factor is necessary.
The probable effect of source credibility on causal inferences can be partially explained
by the discounting principle, which represents a type of belief about how causes are related (e.g.,
Kelley, 1973). This principle entails the tendency that people discount the effect of an
attribution for a certain behavior when an alternative attribution could account for the behavior
(Folkes, 1988). Implications from Hunt, Kernan, and Mizerski’s (1983) study suggest that a
credible endorser can discount the receiver’s expectation that the endorser has bias in his/her
25
product description, by evoking the receiver to infer alternative, internal (dispositional) causes
rather than external (situational) causes for endorsing the product; and vice versa. Hence, an
endorser’s credibility, as a perceptual trait, is associated with the receiver’s belief about the
expected link between credibility and internal/external causes of the endorsing behavior, thus
enabling the researcher to presume this factor as one of the attributional antecedents.
Product Involvement and Ad Involvement as Attributional Antecedents
As conceptually reviewed earlier, involvement refers to personal relevance with the
object, which often becomes a motivational drive of cognitive activities (e.g., Petty & Cacioppo,
1986; Zaichkowsky, 1986). A highly involved person with a certain object, therefore, is likely to
have a stronger interest in and/or need to elaborately process information about the object than a
less involved person (Chaiken & Maheswaran, 1994; Petty, Cacioppo & Goldman, 1981).
Although people do not always make attributions in a rational and orderly manner (Green et al.,
1985), causal inferences represent the cognitive activity involved when an individual assigns an
observable event to its underlying causes (Bemmels, 1991; Smith & Hunt, 1978; Weiner, 1986).
By this nature of attributions, the likelihood an individual makes causal inferences may partly
depend upon his/her inherent involvement with the object in question, which instigates cognitive
information processing. Accordingly, personal involvement – with the product class or the
advertisement – is assumed to limit the conditions under which consumers generate causal
inferences about the endorser’s motives (Folkes, 1988; Petty & Cacioppo, 1986).
5. Consequences of Dispositional and Situational Attributions
As well as attributional antecedents, attitudinal and behavioral consequences of causal
inferences have been important concerns in attribution research (e.g., Folkes, 1988; Kelley and
Michela, 1980; Laczniak, DeCarlo & Ramaswami, 2001; Settle & Golden, 1974; Smith & Hunt,
26
1978; Sparkman & Locander, 1980). Nevertheless, theoretical statements about this issue are
very limited because attribution theory itself deals only with the processes by which attributions
are derived from informational input (Kelley, 1973). Based on the perspectives and findings
from attribution research, it can generally be assumed that causal inferences may affect the
perceiver’s thought, emotion, expectancy, motivation, attitude, or behavior (e.g., Harvey &
Weary, 1984; Kelley & Michela, 1980; Olson & Ross, 1985; Settle & Gorden, 1974; Weiner,
2000), even though some outcome variables (e.g., emotion, motivation) are likely to be confused
with antecedent conditions of the inferences (e.g., Weiner, 1986). More specifically, attributions
influence the perceiver’s feelings about past events, expectations about future events, attitudes
toward other persons or objects, behavioral reactions to an action, conceptions of him/herself, or
efforts to improve fortunes (Kelley & Michela, 1980).
In the field of advertising research, the consequences of causal inferences seem to have
been discussed and tested in a rather simple domain. Typical consequences (i.e., dependent
variables) of attributions examined in advertising research include consumers’ attitude toward to
the ad and attitude toward the brand being advertised (e.g., Cronley et al., 1999; Sorum, Grape &
Silvera, 2003; Tripp, Jensen & Carlson, 1994). Also, attitude toward the endorser (e.g., Cronley
et al., 1999), purchase intention (e.g., Tripp, Jensen & Carlson, 1994), advertiser truthfulness
(e.g., Smith & Hunter, 1978), and brand confidence and expectancy values (e.g., Settle & Golden
1974) have been tested as the perceptual or attitudinal outcomes drawn when causal inferences
are evoked in the advertising context. All these attributional consequences universally focus on
persuasive effects of the advertising content manipulated.
Despite their conceptual pervasiveness, dispositional and situational attributions – in
relation to the concept of correspondence bias – have not been much discussed in terms of their
27
ultimate consequences. Instead, although they dealt with the issue in a very psychological
domain, Gilbert and Malone’s (1995) points are useful in predicting the directional effects of
dispositional and situational attributions. They viewed correspondence bias as a sort of “wishful
thinking” that satisfies an observer who wishes to explain or forecast others’ behavior in a way
that ignores situational constraints. Thus, correspondence bias is likely to have positive – rather
than negative – consequences for the observer as long as the situation he/she has ignored is an
effect or a cause of the actor’s dispositions (Gilbert & Malone, 1995) and the behavior aims at
positive reactions. This proposition suggests that, if an observer attributes a behavior to the
actor’s true dispositions, the inference (dispositional attribution) would have a positive impact on
the observer’s perceptual and attitudinal responses to the behavior even when the observer (e.g.,
consumer) knows that the actor (e.g., endorser) was under very constraining situational
circumstances. In contrast, when the situational attribution is dominant, a certain degree of
negative consequences will result from the attribution.
This study assesses the scope of attributional consequences consistent with prior
advertising research. That is, for prospective consequences of dispositional and situational
attributions, the present study selects the common indicators of advertising effectiveness: attitude
toward the advertisement (hereafter ad attitude), attitude toward the brand (hereafter brand
attitude), and behavioral intention. Since causal inferences concerning the endorser’s motives
are likely to affect how the receiver perceives and attitudinally responds to the endorsement
message, persuasive effects of the message will be largely dependent upon the receiver’s attitude
toward the message content (e.g., the advertisement), the message source (e.g., the brand or the
advertiser), and behavioral motivation (e.g., purchase intention). Also, this setting of the
attributional consequences is desirable for other reasons such as (1) straight comparisons with
28
prior research – especially, with the studies of Cronley et al. (1999) and Sorum, Grape, and
Silvera (2003); (2) sequential and analytic parsimony of the effect flow; (3) reliable
measurement schemes of the variables; and (4) providing managerial implications. Although
these attitudinal and motivational factors do not represent all consequences that dispositional or
situational attributions might evoke (including emotion and behavior), they are considered as
significant outcome variables for communication effectiveness, as posited and investigated in
many advertising studies.
6. The Mediating Role of Dispositional and Situational Attributions
The next important concern of this research is whether or not dispositional and situational
attributions play a mediating role in the relationship between their antecedents and consequences.
In their general model of the attribution field, Kelley and Michela (1980) categorized two major
domains of attribution research: (1) a research paradigm primarily focusing on the antecedents –
attributions link (“attribution research” by their term) and (2) another research paradigm for the
attributions – consequences link (“attributional research” by their term). As exhibited in
Figure 1, Kelley and Michela’s (1980) model can be used to posit the mediating effect of causal
inferences. Their model suggests that perceived causes would mediate the relation between
antecedents of attributions (i.e., information, belief, motivation) and consequences of attributions
(i.e., behavior, affect, expectancy). However, because most attribution studies have typically
examined only one dimension – antecedents or consequences – of causal inferences (e.g., Choi,
2002; Cronley et al., 1999; Sorum, Grape & Silvera, 2003; Settle & Golden, 1974; Smith &
Hunter, 1978; Tripp, Jensen & Carlson, 1994), little is actually known about whether and how
attributions play a mediating role between what causes them and what they affect (Folkes, 1988;
Kelley & Michela, 1980).
29
Antecedents
Attributions
Consequences
Information
Beliefs
Motivation
Perceived
Causes
Behavior
Affect
Expectancy
Figure 1. Kelley & Michela’s (1980) Mediation Model of Attributions
Mediating effects have been commonly examined in consumer research. As an
illustration, advertising researchers have tested the mediating effects of many types of receiveroriented variables, such as cognition (e.g., Arias-Bolzmann, Chakraborty & Mowen, 2000),
emotion (e.g., Holbrook & Batra, 1987), attention (e.g., MacKenzie, 1986), and attitude toward
the ad (e.g., Shimp, 1981; Stayman & Aaker, 1988), to explore the processes that underlie
consumer responses to advertising. However, most research articles testing a mediating effect
lack conceptual explanations of the effect. According to Baron and Kenny (1986), who
distinguished the properties of mediator and moderator variables, a given variable may function
as a mediator to the extent that it accounts for the relationship between the predictor
(independent variable) and the criterion (dependent variable). Mediators, thus, explain how and
why a certain causal effect occurs – whereas moderators specify when a certain effect holds –,
typically helping researchers account for the causal relations under investigation (Baron &
Kenny, 1986).
To assess any proposed mediating effect, it is critical to understand how this type of
effect can be conceptually and statistically verified. A general test for the mediator effect is to
examine three types of relationships: (a) between independent and dependent variables; (b)
between independent and mediator variables; and (c) between mediator and dependent variables
(see Figure 2). To determine whether a certain variable is a mediator, all of these correlations
should be significant, and the relationship between independent and dependent variables should
30
be weakened after controlling for the relationship between the mediator and the dependent
variable. Theoretically, a significant reduction in the relational strength indicates that a given
mediator is indeed present (Baron & Kenny, 1986).
(b)
Independent
Variable
Mediator
(c)
Dependent
Variable
(a)
Figure 2. A Path Diagram of the Mediating Effect
Applying this conceptual function of the mediating effect to the current research,
dispositional or situational attributions would mediate the relationship between its antecedents
(e.g., source credibility) and consequences (e.g., ad attitude, brand attitude, behavioral intention)
if they could explain why the causal links between those variables exist. For instance, the
mediating effect of dispositional attributions will occur when the direct impact of source
credibility on ad attitude is reduced as the factor of dispositional attributions is controlled for the
relationship between the two variables. It is important to examine the mediating role that
dispositional and situational attributions play in consumers’ information processing. By doing so,
the researcher can explicitly address the mechanism by which endorser-generated inferences and
their antecedents affect persuasion of testimonial messages. Therefore, this study is an effort to
investigate a rarely explored area in both advertising and attribution research.
31
CHAPTER IV
RESEARCH HYPOTHESES
In this chapter, multiple research hypotheses are drawn based on the reviewed literature.
Since both conceptual and theoretical foundations were discussed at length in the preceding
chapters, this chapter is devoted to deriving the hypotheses on the basis of the established
framework, adding supportive statements for predictions.
Overview
The primary purpose of this research is to provide both theoretical and managerial
implications with regard to the role of causal inferences in the persuasive effects of noncelebrity
endorsements. Focal points for the investigation include: (1) antecedents – source credibility and
product/ad involvement – of dispositional/situational attributions; (2) consequences – ad attitude,
brand attitude, and behavioral intention – of dispositional/ situational attributions; (3) the
mediating effects of dispositional/situational attributions on the relationships between their
antecedents and consequences; and (4) evocation and persuasive effects of correspondence bias.
Noncelebrity testimonial advertising is considered as an appropriate context within which to
examine these issues because message believability, presumably affected by causal inferences
about the endorser’s motives, is more likely to determine persuasion in this type of advertising
than other contexts such as celebrity advertising.
Antecedents of Dispositional and Situational Attributions
The first set of research hypotheses is concerned with the antecedent conditions that
might cause the evocation of causal inferences, specifically dispositional and situational
32
attributions. Based on the reviewed literature, this study assumes the receiver’s perceived source
credibility, product involvement, and ad involvement as influential factors on endorser-arousing
attributions.
First, it is a general expectation that communicator credibility will positively affect
persuasiveness of a testimonial message (e.g., McGinnies & Ward, 1980; McGuire, 1985;
Ohanian, 1991). One of the underlying questions for this study is whether such source
credibility also influences the receiver’s perception of the endorser’s motivation to support the
brand. Source credibility, the concept that incorporates the dimensions of the communicator’s
expertise and trustworthiness, is important in that this factor is likely to be associated with a
receiver’s perception about the endorser’s genuine motives to communicate as he/she does.
Credibility is assumed to be a crucial property particularly for noncelebrity endorsers, who tend
to lack attractiveness and familiarity in contrast to celebrity endorsers. (See p.25 for an
alternative argument about source credibility effects.)
Under the circumstances of causal inferences being made about an endorser, a receiver
may believe that the endorsement is motivated by true confidence in the product being endorsed
(dispositional attribution) and/or by self-interest (situational attribution). It is assumed that in
this attributional process source credibility has a directional effect on the evocation of
dispositional or situational attributions. That is, if a noncelebrity endorser is perceived to lack
credibility, the message’s truthfulness will be doubted and thus the receiver’s attribution about
endorsement motivation will not correspond to the endorser’s dispositions. In this situation, the
receiver is likely to generate situational attributions, ascribing the endorserment to one or more
environmental constraints – commonly including a monetary incentive for the endorsement. In
contrast, a highly credible endorser will cause the perceiver to be receptive to endorsing
33
messages, resulting in the endorsement being attributed to the endorser’s actual belief in what
he/she says. In short, the level of source credibility will determine whether the receiver
attributes the endorsement motivation to the endorser’s true dispositions or to situational forces.
Thus, the following hypotheses are proposed:
H1: Source credibility will significantly influence the evocation of dispositional and
situational attributions in the context of noncelebrity testimonial advertising.
Specifically,
H1-a: Dispositional attributions are more likely to occur when source credibility
is high than when it is low.
H1-b: Situational attributions are more likely to occur when source credibility is
low than when it is high.
Two dimensions of involvement – product involvement and ad involvement – are also
presumed as potential determinants of the generation of dispositional or situational attributions.
The basis of this assumption is derived from the elaboration likelihood model (e.g., Petty &
Cacioppo, 1981) and the heuristic-systematic model (e.g., Chaiken, 1980), both of which predict
cognitive processing of message arguments when an individual is involved with a certain object
or issue. Since involvement is positively related to the individual’s motivation for message
processing (e.g., Andrews, 1988; Mitchell, 1979; Petty, Cacioppo & Schumann, 1983), the
likelihood of message-relevant thoughts, including attribution evocation, will increase as the
receiver’s involvement is high and thus the product-related cues are evaluated in a reasoned
manner.
Both product involvement and ad involvement, however, will not have a direct impact on
what sort of attributions – dispositional or situational attributions – will occur because
involvement itself is unlikely to guide a specific direction for message-relevant thinking (e.g.,
favorable or unfavorable, believable or unbelievable) when advertising becomes a motivational
34
object. Product involvement and ad involvement will instead play a moderating role in
persuasive situations, as many involvement researchers have suggested (e.g., Maoz & Tybout,
2002; Muehling, Laczniak & Stoltman, 1991; Petty, Cacioppo & Schumann, 1983). That is,
based on the theoretical association between source credibility and involvement in a receiver’s
information processing (as reviewed in Chapter II), both types of involvement are assumed to
influence the direction and strength of the source credibility’s effect on dispositional or
situational attribution evocation. Thus, the second set of hypotheses are drawn as follows:
H2: Personal involvement will moderate the effect of source credibility on the
evocation of dispositional or situational attributions in the context of noncelebrity
testimonial advertising.
Specifically,
H2-a: Product involvement will moderate the effect of source credibility on the
evocation of dispositional or situational attributions
H2-b: Ad involvement will moderate the effect of source credibility on the evocation
of dispositional or situational attributions
Consequences of Dispositional and Situational Attributions
The next concern of this study is the persuasive outcomes of dispositional and situational
attributions that may occur when an individual is exposed to a testimonial advertisement. As
stated earlier, this study limits attributional consequences to the primary indicators of advertising
effectiveness, which include ad attitude, brand attitude, and behavioral intention. Although these
variables alone are unable to provide a full assessment of possible attribution effects, they are
deemed to be appropriate outcome variables because all three factors have been consistently used
as barometers of persuasion in the advertising context (e.g., Choi, 2002; Goldsmith, Lafferty &
Newell, 2000; Gill, Grossbart & Laczniak, 1988; Kamins, 1989). In addition, those variables,
which have been found to be interrelated (e.g., Lord, Lee & Sauer, 1995; MacKenzie, Lutz &
35
Belch, 1986), explicitly reflect the receiver’s attitudes and expectations that Kelley and Michela
(1980) proposed as the major dimensions of attributional consequences.
Dispositional and situational attributions are likely to have directional effects on ad/brand
evaluations. Dispositional attributors will positively evaluate the ad and the advertised brand,
based on their congruent association between the object (e.g., testimonial messages) and what the
object is supposed to be (e.g., testimonial messages must reflect the endorser’s real experience
with and/or belief in the advertised brand). On the other hand, situational attributors will
generate unfavorable evaluations about the ad and the brand, because their attribution of the
endorsing behavior to self-motivations will lessen the probability that they trust the testimonial
messages. This tendency seems to be particularly plausible in the advertising environment where
a noncelebrity endorser speaks for the product to give consumers the impression of truthful
recommendations – compared to celebrity endorsements by which the endorser’s well-known
image, authority, and attractiveness might dilute the influence of perceived endorser motivations.
Hence, the subsequent research hypotheses are drawn as follows:
H3: Dispositional and situational attributions will have significant and differential
effects on persuasion in the context of noncelebrity testimonial advertising.
Specifically,
H3-a: Dispositional attributions will positively influence attitude toward the ad,
whereas situational attributions will negatively influence attitude toward the ad.
H3-b: Dispositional attributions will positively influence attitude toward the brand,
whereas situational attributions will negatively influence attitude toward the
brand.
H3-c: Dispositional attributions will positively influence behavioral intention,
whereas situational attributions will negatively influence behavioral intention.
36
The Mediating Role of Dispositional and Situational Attributions
The fourth set of research hypotheses is concerned with whether and how dispositional
and situational attributions could influence the effects of source credibility on attitudinal
responses to noncelebrity testimonial advertisements. That is, the issue here is centered on the
mediating role of dispositional and situational attributions in the relationships between their
antecedent (source credibility) and consequences (ad attitude, brand attitude, and behavioral
intention). The mediating effect of causal inferences has already been suggested by attribution
researchers (e.g., Kelley & Michela, 1980). However, to the researcher’s knowledge, no
empirical research has tested it yet. Incorporating the proposed predictor and outcome variables,
the mediating role of endorser-arousing attributions are posited as below:
H4: Dispositional and situational attributions will mediate the effect of source
credibility on persuasion in the context of noncelebrity testimonial advertising
Specifically,
H4-a: Dispositional and situational attributions will mediate the effect of source
credibility on attitude toward the ad.
H4-b: Dispositional and situational attributions will mediate the effect of source
credibility on attitude toward the brand.
H4-c: Dispositional and situational attributions will mediate the effect of source
credibility on behavioral intention.
Figure 3 illustrates the mediation model of dispositional/situational attributions, which
integrates the first four sets of hypotheses proposed in the present study. The model presupposes
multiple dependence relations and mediating links: that is, (1) the main effect of source
credibility on attributions (Hypothesis 1); (2) the moderating effects of product involvement and
ad involvement on attributions (Hypothesis 2); (3) the main effects of attributions on ad attitude,
brand attitude, and behavioral intention (Hypothesis 3); and (4) the mediating effects of
37
Product Involvement
Source Credibility
Ad Attitude
Dispositional
vs. Situational
Attributions
Brand Attitude
Ad Involvement
Behavioral Intention
Attributional Consequences
Attributional Antecedents
Figure 3. A Proposed Mediation Model of Endorser-Generated Attributions
attributions on the causal relationships between source credibility and ad attitude, brand attitude,
and behavioral intention (Hypothesis 4). The intention of these model tests is to
comprehensively assess the probable antecedents and consequences of dispositional and
situational attributions in a single study, thereby properly evaluating the mediating role of
attributions in the persuasive circumstance.
A Further Look: Occurrence and Impact of Correspondence Bias
The current research will also examine the conditions under which correspondence
bias occurs and assess its influence on the perceiver’s attitudes. Correspondence bias, by
definition, is an indispensable concept in discussing the impact of dispositional attributions. The
underlying assumption is that correspondence bias is evoked when a perceiver recognizes one or
more situational constraints surrounding a certain behavior (e.g., a monetary incentive for the
endorsement), but still attributes the behavior to the actor’s dispositions (e.g., true belief in the
endorsed product). Therefore, correspondence bias always occurs in company with dispositional
attributions. Theoretically, however, dispositional attributors do not necessarily possess
correspondence bias because its evocation is dependent on the receiver’s perceiving situational
38
forces as determinants of the actor’s behavior (Gilbert & Malone, 1995). For this reason, several
studies have assessed correspondence bias by manipulating subjects’ awareness of situational
factors that is linked to the actor’s behavior (e.g., Cronley et al., 1999; Jones & Davis, 1965;
Sorum, Grape & Silvera, 2003).
On the same bases as drawn in the preceding hypotheses, three independent variables –
source credibility, product involvement, and ad involvement – are posited as the antecedent
conditions for the generation of correspondence bias, while ad attitude, brand attitude, and
behavioral intention are set as outcome variables. It is a basic question whether correspondence
bias will be evoked in the context of noncelebrity testimonial advertising, as it was in the
celebrity endorsement context in Cronley et al.’s study (1999). Also, a credible endorser is
assumed to increase the possibility a receiver makes dispositional attributions while disregarding
any situational force that might cause the endorsing behavior; and the receiver’s product/ad
involvement will moderate such a causal link from source credibility to the evocation of
correspondence bias. Thus, the following hypotheses are established:
H5: Correspondence bias will be evoked in the context of noncelebrity testimonial
advertising.
Specifically,
H5-a: Correspondence bias will more likely be generated when source credibility is
high than when it is low
H5-b: Product involvement will moderate the effect of source credibility on the
evocation of correspondence bias.
H5-c: Ad involvement will moderate the effect of source credibility on the evocation
of correspondence bias.
It is predicted that correspondence bias, in turn, has a positive effect on persuasion
because it can be generated only when the receiver makes dispositional attributions (even though
39
situational factors might explain the communicating behavior), which have been hypothesized
earlier to positively influence advertising effectiveness. Empirical finding from Cornley et al.’s
(1999) and Sorum, Grape, and Silvera’s (2003) endorsement studies support such a positive
causal link between correspondence bias and its consequences. Thus, the following hypotheses
are drawn with respect to the persuasive effects of correspondence bias:
H6: Correspondence bias will positively influence persuasion in the context of noncelebrity
testimonial advertising.
Specifically,
H6-a: Correspondence bias will have a positive effect on the receiver’s attitude
toward the ad.
H6-b: Correspondence bias will have a positive effect on the receiver’s attitude
toward the brand.
H6-c: Correspondence bias will have a positive effect on the receiver’s behavioral
intention.
All these hypotheses require rigorous investigations based on the integrated model of
dispositional and situational attributions in order to provide theoretical and managerial
implications with respect to: (1) the applicability of the attribution perspective to consumer
processing of noncelebrity testimonial advertising and (2) appropriate selections of product
endorsers and message strategies. The methodological process to test the established hypotheses
will be described in the next chapter.
40
CHAPTER V
METHODOLOGY
Detailed methods used for the two experiments are presented in this chapter. Due to the
methodological complexity involved in this research, a general overview of the experimental
design and procedure is first provided, followed by conceptual definitions of main variables,
orderly descriptions of pretests and main experiments, and finally operationalizations of major
concepts for measurements.
1. Overview
To test the proposed hypotheses, two experiments were conducted after the pretests. This
two-stage experimental process is advantageous to assure the reliability of the findings and to
resolve the complexity involved in the research design. In the first experiment (hereafter
Experiment I), participants were selected from a population of college students and were
randomly assigned to one of four experimental conditions with a 2 (high vs. low source
credibility) x 2 (high vs. low product involvement) factorial design. On the other hand, the
second experiment (hereafter Experiment II) was administered with participants from a
nationwide online consumer panel, using a quasi-experimental design. In this stage, two
variables – ad involvement and correspondence bias – were additionally tested for full
assessment of the hypotheses. For both experiments, each participant was presented with a
magazine-style testimonial advertisement featuring a noncelebrity (typical consumer) endorser.
After exposure to the stimulus advertisement, participants’ levels of endorser-generated
41
attributions, ad/brand attitude, behavioral intention, source credibility, and product/ad
involvement were measured on multi-item scales.
2. Conceptual Definitions of Key Variables
It was necessary for this study to conceptually redefine some important variables in order
to appropriately operationalize them and thereby achieve an acceptable level of construct validity
before performing the actual experiments. Based on conceptualizations in the relevant literature,
the key concepts of this research are redefined as follows:
•
Attribution: The cognition a receiver generates to infer the cause of an actor’s behavior. In
this study, it specifically refers to an individual’s cognitive activity of ascribing the
endorser’s motivation for product endorsement to one or more causes.
•
Dispositional attribution: A condition in which the endorser’s motive for supporting the
brand is attributed to the endorser’s true disposition(s) – specifically, the endorser’s honest
feeling or belief about the advertised product
•
Situational attribution: A condition in which the endorser’s motive for supporting the
brand is attributed to certain situational force(s) – specifically, the endorser’s self-interest,
such as making money or enhancing personal goals, which is not based on his/her actual
confidence in the advertised product
•
Correspondence bias: The tendency for an individual to draw a dispositional attribution
from the endorsement even though he/she has recognized the situational constraint(s) that
might actually cause the endorsement
•
Source credibility: The extent to which the endorser is perceived as possessing expertise in
the communication subject (e.g., the advertised product class) and being trustworthy in
42
his/her opinion on the subject. Thus, this concept incorporates two sub-dimensions:
expertise and trustworthiness.
•
Product involvement: The extent to which an individual is personally relevant to the
advertised product class based on his/her enduring needs, values, or interests
•
Ad Involvement: The extent to which an individual pays attention to the advertising content
based on his/her enduring or temporary needs, values, or interests
3. Pre-Experimental Procedure: Pretests
To construct a valid set of experimental instruments, a series of pretests were conducted.
Each pretest was designed to measure and determine its own object that would become a key
element in the stimulus advertisements: i.e., the first pretest for the selection of product classes
(to manipulate product involvement); the second pretest for the selection of endorser occupations
(to manipulate source credibility); and the third pretest for the selection of fictitious brand names
that would be well matched with the products being endorsed. This procedure was expected to
ensure successful manipulations of the treatment variables and thus to contribute to enhancing
the internal validity of the experiments. Details are illustrated below.
Product Class Selection
The first pretest was designed to determine appropriate product classes that would be
advertised in the stimulus materials: one for a high involvement product and the other for a low
involvement product. Twelve initial product classes – six manufactured good categories and six
service categories – were chosen for the questionnaire based on two criteria: (1) the likelihood
that the product class could reasonably target college students, the population comprising the
experimental subjects for Experiment I in which product involvement would be manipulated; and
(2) the likelihood that advertisers in the product class could reasonably use a testimonial
43
approach. Measurement scales were adopted from the theoretical propositions regarding product
involvement that Zaichkowsky (1985) had used to test the construct validity of her personal
involvement inventory (see the pretest questionnaire in Appendix B-1).
Based on data from the pretest survey with 72 undergraduate students, athletic shoes and
low-fat yogurt were selected as the high involvement and low involvement products, respectively.
These product classes were especially suitable for the experiments because (1) both of them
belong to the manufactured goods category, thus avoiding the potential confounding effect of
product tangibility (i.e., tangible goods versus intangible services) and (2) they could appeal with
similar functional and hedonic benefits, such as becoming mentally and physically healthy with a
good shape of body by using the product. These kinds of characteristics of the selected product
classes were later reflected in the testimonial copy of the stimulus advertisements (particularly
for Experiment I).
Endorser Occupation Selection
The second pretest was to determine the distinguished levels of source credibility. In its
preliminary stage before the experiments, typical consumers were selected for use because this
type of endorser was deemed to appear much more frequently in testimonial advertising than
other kinds of noncelebrity endorsers, such as experts and CEOs who are little known to the
public. Prior experimental studies on source credibility have typically manipulated the
endorser’s occupation, which could directly or indirectly influence the endorser’s perceived
expertise and trustworthiness (e.g., Gotlieb & Sarel, 1991; Homer & Kahle, 1990). Following
this approach, the second pretest gathered evaluative data regarding perceived credibility over a
variety of occupations.
44
Seventy-one undergraduate students participated in the second pretest. They were first
given an operational definition of source credibility, as comprising the endorser’s expertise with
the advertised product category and trustworthiness in providing opinions about the advertised
brand. They were then requested to choose the most credible and the least credible person in
occupation each among a list of 30 occupations for the endorsements of athletic shoes and lowfat yogurt (see Appendix B-2 for the questionnaire). Based on its results, aerobics instructor and
dietician were selected as highly credible endorser occupations for athletic shoes and low-fat
yogurt, respectively. On the other hand, sales representative was chosen as the least credible
endorser occupation for both product classes. To minimize the confounding effect of endorser
gender, it was determined to consistently use an endorser of the same gender (female) for every
stimulus advertisement.
Brand Name Selection
For this study’s experiments, it was vital to create and use fictitious brands in order to
eliminate the possible effects of brand familiarity and perceived brand quality that might have
existed and varied in the potential participants’ minds. The third pretest was designed to
determine the most plausible and favorable brand names for athletic shoes and low-fat yogurt.
Considering product images and benefits, the researcher created 12 brand names (six for each
product class), which did not exist in the relevant markets at that time of the test (see Appendix
B-3 for the list of the brand names). Sixty-eight undergraduate students took part in the third
pretest, and voted for the most and second most favorable brand names for each product category.
According to the results, Stellar and YoSlim were selected as the brand names for athletic shoes
and low-fat yogurt, respectively.
45
4. Experimental Design, Sampling, and Procedure
After the development of a methodological frame guided by the pretests, two experiments
– Experiment I and Experiment II – were performed to empirically test the hypotheses drawn.
Detailed descriptions regarding the implementation of each experiment follow.
Experiment I
Objectives
The primary objectives of Experiment I were to test (1) the main effect of source
credibility on the evocation of dispositional/situational attributions (Hypothesis 1); (2) the
moderating effect of product involvement on the causal relation between source credibility and
attributions (Hypothesis 2-a); (3) the main effects of dispositional/situational attributions on
ad/brand evaluations (Hypothesis 3); and finally (4) the mediating effect of attributions on the
causal links between source credibility and ad/brand evaluations (Hypothesis 4). A 2 x 2
between-subjects design was used for this experiment, with the high versus low source
credibility condition crossed with the high versus low product involvement condition. Specific
treatments for source credibility and product involvement were derived from the pretest results.
Stimuli
The experimental stimuli were black-and-white, full-page, magazine-style advertisements
that employed a typical testimonial approach with a single endorser. For the endorser in the
stimulus advertisements, a fictitious female character named Lisa Jones was created. By using a
fictional endorser, variations in and level of the receiver’s knowledge and existing attitude
toward the endorser were minimized.
Four testimonial advertisements, which vary by source credibility (high/low) and product
involvement (high/low), were created for this stage of the experiment. In order to minimize
46
potential effects of uncontrolled variables, it was necessary to make these four different versions
of stimulus advertisements have minimal variances in their message and creative components.
Thus, each advertisement was designed to contain an identical model (same image and name),
layout, and messages (e.g., testimonial copy, outline of body copy), with the exceptions of the
endorser’s occupation, product picture (athletic shoes and low-fat yogurt), brand name (Stellar
and YoSlim), and brand logo (see Appendix C-1 for the advertisements used in Experiment I).
Based on the pretest data, two conditions of source credibility – high versus low – were
manipulated by differently labeling the endorser’s occupation: (1) for the athletic shoes brand
Stellar (a high involvement product), “aerobics instructor” as an endorser with high credibility
and “sales representative” as an endorser with low credibility; (2) for the low-fat yogurt brand
YoSlim (a low involvement product), “dietician” as an endorser with high credibility and “sales
representative” as an endorser with low credibility. These manipulations for the two treatment
variables were assured by measuring and comparing the participants’ ratings on perceived source
credibility and product involvement in the actual experiment.
Participants
A total of 231 undergraduate students enrolled in an introductory advertising class at a
major southeastern university participated in the first experiment. Participants’ average age was
20.6 and gender ratio was female 76.5% to male 23.5%. Each subject was randomly assigned to
one of four experimental groups (57 or 58 subjects per group). During this random assignment
process, gender ratios of the participants among the four groups were kept almost equal. To
assure successful differentiation in experimental conditions among the groups, participants
whose perceptual responses to the manipulations were inconsistent with the intended treatments
were dropped after the data collection, with an effort to make the sample size of every group
47
equivalent. Thus, the final sample size (after filtering the data) became 200; that is, 50 for each
of the four groups.
Procedure
Every participant was given a booklet containing a consent form, instructions, a
testimonial advertisement (one of four versions), and a questionnaire. The experimental task
required each subject to be exposed to a stimulus advertisement (the instructed exposure time
was 20-30 seconds, or longer if the respondent wished) and then fill out the given questionnaire,
which included the measures of the selected independent and dependent variables.
The participants were informed that the advertised brand (Stellar for athletic shoes and
YoSlim for low-fat yogurt) was a fictitious one created only for the experiment. They were asked
to imagine that the brand exists in the U.S. market and many people might have used it even
though the respondent has never tried the brand. The participants were additionally instructed to
assume that (1) they found the given advertisement while they were reading a magazine; (2) the
black-and-white advertisement is actually shown in full color in the magazine they were reading;
and (3) the endorser (Lisa Jones) in the advertisement is a real person. After the instructions, the
participants evaluated their involvement with the given product category – for a manipulation
check – before they were exposed to a stimulus advertisement, which was followed by the
measures on perceived credibility of the endorser, inferred motives of the endorser (i.e.,
dispositional/situational attributions), ad attitude, brand attitude, and behavioral intention.
Experiment II
Objectives
The main objectives of Experiment II were: (1) to examine if the hypothesized dependent
relations would yield consistent results on replicated measurements with the general consumer
48
sample; (2) to investigate the causal effects of ad involvement and correspondence bias, the
variables untested in Experiment I; and (3) to verify source credibility effects by running a
control group in addition to the two treatment groups. For these purposes, Experiment II
employed a 2 (manipulated by source credibility: high and low) x 3 (measured by product
involvement: high, moderate, and low) x 3 (measured by ad involvement: high, moderate, and
low) between-subject factorial design (see Figure 4). In this stage of the experiment, all the
hypotheses tested in Experiment I were reassessed using the repeated measurements along with
additional tests for: (1) the moderating effect of ad involvement on the causal relation between
source credibility and attributions (Hypothesis 2-b); (2) the evocation of correspondence bias in
the context of noncelebrity testimonial advertising (Hypothesis 5); and (3) the main effects of
correspondence bias on ad/brand evaluations (Hypothesis 6).
Source Credibility
(2 Levels)
High
Low
X
Product Involvement
(3 Levels)
X
Ad Involvement
(3 Levels)
High
Moderate
Low
High
Moderate
Low
Control Group
Figure 4. Factorial Design for Experiment II
Using a consumer panel sample is a critical component of Experiment II. Because it is
often questionable whether college student subjects are an appropriate surrogate for adult
consumers in marketing research (e.g., Burnett & Patrick, 1986; Peterson, 2001), Experiment II
extended its sample to a heterogeneous group of consumers, who vary by age, gender, and
occupation, in order to evaluate external validity of the findings. By doing so, it was expected
49
that the researcher could determine if the empirical results would be generalizable across the
consumer population.
Tested Variables and Product Class Selection
For Experiment II, the researcher manipulated or measured three independent variables:
source credibility, product involvement, and ad involvement. Among them, only source
credibility was manipulated on two treatment conditions (high vs. low), while the variable was
also controlled in one experimental group. On the other hand, both product involvement and ad
involvement were measured on multi-item scales and then split into the three groups (high,
moderate, and low) based on their statistical values. Basic equivalencies over the multiple
experimental cells were successfully achieved.
Product involvement, one of the manipulated variables for Experiment I, was not
manipulated in this phase because it was expected that the individual variance on this factor
would be considerable when participants are drawn from the consumer panelists whose age and
lifestyle are likely to vary to a greater extent than those of the college student sample. Athletic
shoes were selected as an appropriate single product for this stage of the experiment due to their
generally high relevance to consumers across age and gender (Simmons Market Research, 2003).
Thus, the participants were exposed to an athletic shoes advertisement endorsed by an aerobics
instructor (for the high source credibility condition), by a sales representative (for the low source
credibility condition), or with no given occupational information about the endorser (for the
controlled source credibility condition).
Stimuli
The stimulus advertisements used in Experiment I were refined in order to be
appropriately utilized for Experiment II. Since this stage of the experiment tested only one
50
product class, the testimonial copy and the body copy about the advertised product (Stellar
athletic shoes) were revised to tailor specific benefits of the endorsed brand. It was expected that
such a modification would lead participants to perceive a better fit between the product and its
commercial messages than did the stimulus advertisements for Experiment I. For the
manipulation or control of source credibility, the occupational label of the endorser (i.e., aerobics
instructor, sales representative, and no occupational label) was differentiated over the
experimental groups. However, like the condition in Experiment I, the endorser’s gender, image,
and name, overall layout, and message frame were held consistent for the three different
advertisements in order to minimize the possible confounding effect of any uncontrolled
variables (see Appendix C-2 for the advertisements used in Experiment II).
The original black-and-white advertisements were converted to full-color ones to provide
participants with realistically reproduced experimental stimuli. Instead, because participants
would be exposed to the assigned advertisement in the online circumstance, the researcher
requested them to assume that the magazine-style advertisement they would see on the computer
screen would be actually shown in a magazine of their interest.
Participants
A random stratified sample of 1500 consumer panelists were drawn from the Zoomerang
database, a division of Market Tool Inc., which possesses over 2 million registered adult
members of the nationwide consumer panel in the United States. The initial sample comprised
750 males and 750 females; and 1125 panelists aged under 45 (75%) and 375 panelists aged over
45 (25%). These ratios are compatible with those reported in the Simmons consumer data (2003)
concerning the purchase of athletic shoes (or sneakers/sports shoes) in the last 12 months
(male:female = 45:55; Under 45:Over 45=79:21). Those initial mail recipients also varied in
51
their race, occupation, educational level, household income, and state in which they currently
reside. With an effort to make such demographic characteristics equivalent over the three groups
(two treatment groups and one control group), each potential participant was randomly assigned
to one of the three experimental conditions (i.e., initial 500 panelists per group).
A week after the questionnaire (including a stimulus advertisement) had been provided to
the selected consumer panelists, a total of 356 responses were collected (response rate: 23.7%).
However, responses contrary to the treatment of source credibility or diluted by any technical
problems reported (e.g., the ad was not loaded) were excluded from the final data. After this
filtering process, 335 participants were determined as the final sample for statistical inference
(116 for the high source credibility group, 112 for the low source credibility group, and 107 for
the control group). As shown in Table 8, the respondents comprise: (1) by gender, 168 males
(50.1%) and 167 females (49.9%); (2) by age group, 26 aged 18-24 (7.8%), 109 aged 25-34
(32.5%), 124 aged 35-44 (37.0%), 31 aged 45-54 (9.3%), 31 aged 55-64 (9.3%), and 14 aged
over 65 (4.2%); (3) by race, 290 Caucasians (86.6%), 9 African-Americans (2.7%), 14 Asians
(4.2%), and 12 Latinos (3.6%); and (4) by occupation, 101 professionals (30.1%), 49
management people (14.6%), 39 homemakers (11.6%), 33 self-employed people, 11 college
students (3.3%), and others. Table 8 shows these demographic profiles of the experimental
subjects. As reported in Table 9, the equivalence over the three experimental groups by gender
and age, two major demographic factors, was ensured through two Chi-square tests (gender:
χ2=.116, df=2, p=.944; age: χ2=3.647, df=10, p=.962).
Procedure
Because Experiment II employed a field experiment method using an online consumer
panel, the data-gathering procedure was different from that used previously in the offline setting.
52
According to the treatment/control of source credibility, three electronic files, in which
experimental materials (i.e., instructions, a stimulus ad, and a questionnaire) are saved, were
generated; then, an invitation e-mail containing a link to one of these files was sent out to the
randomly selected consumer panelists. (All of them were opt-in for e-mails from the Zoomerangauthorized research).
Those who wanted to participate in the current experiment were guided to access the
assigned experimental content by clicking a link in the invitation e-mail at their convenience.
Once participants had accessed the experimental content, they were instructed to: (1) look at a
given advertisement for 20-30 seconds or longer if they wished; (2) assume that they find the
advertisement while reading a magazine; and (3) carefully read and follow the directions in each
section when filling out the post-exposure questionnaire. Referring back to the advertisement
whenever answering was allowed because it was not restricted in Experiment I and there were
many memory-based questions spanning the whole questionnaire.
The experimental content given to each participant was basically similar to that used for
Experiment I. However, because more questions were created due to the addition of two
variables – ad involvement and correspondence bias – for testing in this stage of the experiment,
the overall length of the questionnaire was longer than that of the previous experiment. Also, to
control for potential order effects, the logically ordered question sections in Experiment I (i.e.,
measures for independent variables came first, followed by attribution measures, and then
measures for dependent variables in the last) were altered for a new order, which can be briefed
as “attitude measures → attribution measures → source credibility and involvement measures,”
in this phase of the experiment.
53
5. Dependent Measures
Both Experiment I and Experiment II employed basically identical scales on the same
variables with some minor differences because most of the measurement scales were adapted
from prior research by which the scales had been verified to be internally reliable and valid. The
measurement of source credibility, comprising the dimensions of endorser expertise and
trustworthiness, followed Ohanian’s (1990) source-credibility scale. A seven-point, semantic
differential scale was used for five items – “expert/not an expert,” “experienced-inexperienced,”
“knowledgeable-unknowledgeable,” “qualified/unqualified,” and “skilled/unskilled” – on
expertise (avg. [Cronbach’s] alpha=.93 for the two experiments); and another five items –
“trustworthy/untrustworthy,” “honest/dishonest,” “reliable/unreliable,” “sincere/insincere,” and
“dependable/undependable” – on trustworthiness (avg. alpha=.94 for the two experiments). This
measure was for the manipulation check of source credibility and the mediation tests of
attributions.
Product involvement was also measured for the purpose of a manipulation check (for
Experiment I) or individual evaluations (for both Experiment I and Experiment II). For
dependable measurements, Zaichkowsky’s (1994) revised personal involvement inventory was
employed because of its well-proven content validity and internal scale reliability. This study
used her 10-item, seven-point, semantic differential scale, which was anchored by
“important/unimportant,” “relevant/irrelevant,” “appealing/unappealing,” “needed/not needed,”
etc., with no modification.
To measure ad involvement (for Experiment II only), the participants were asked to rate:
their levels of attentiveness to multiple advertising elements (e.g., the endorser, the copy, and the
product picture) and the overall advertisement; the perceived importance of and relevance to the
54
advertising message; and the task relevance on a seven-point Likert scale (from very little to very
much or from strongly disagree to strongly agree). Laczniak and Muehling’s (1993) original
scale for the measurement of ad message involvement was utilized in this phase. The recall test
of ad messages was additionally used to ascertain an individual participant’s objective (not selfevaluated) attentiveness to the given advertisement. The multiple-choice recall questions were
concerned with the respondent’s attention to brand benefits described in the body copy and brand
slogan.
Dispositional and situational attributions were measured in a forced condition: that is,
participants were asked to rate how much they agree or disagree with each of the listed causes of
the endorsement. Although people make causal inferences naively by natural activation of
attributional thinking (Heider, 1958), this method of measuring attributions was necessary to
evoke the participants’ dormant inferential activity and thereby to ensure a sufficient number of
attributional responses for statistical analyses. Dispositional versus situational attributions had to
be measured on separate rating scales because, as Pilkonis (1977) found, these two kinds of
attributions can independently occur although they are conceptually inverse. Thus, the
participants were requested to evaluate their level of agreement or disagreement on all the
statements regarding the endorser (Lisa Jones)’s motives to endorse the brand (Stellar or
YoSlim): for the measurement of dispositional attributions, “to convey her real belief in the
brand,” “to express her feeling about the brand based on her actual experience,” and “to speak
about brand benefits based on her knowledge about the product”; and for the measurement of
situational attributions, “to earn money,” “to become better known,” and “to receive nonmonetary compensation (e.g., free products).”
55
Correspondence bias, a variable tested in only Experiment II, was initially measured on
the scale used by Cronley et al. (1999) and Sorum, Grape, and Silvera (2003). Participants were
asked to rate on three descriptive items using a seven-point Likert scale (from strongly disagree
to strongly agree), which include “the endorser (Lisa Jones) frequently uses the brand (Stellar),”
“the endorser likes the brand,” and “the endorser regards the brand as good.” The responses on
these items were combined with those about dispositional attributions (three items described
earlier), thus generating a composite score of extended dispositional perceptions for individual
participants. A total of six items for this measure were found to be internally consistent
(alpha=.94). The calculated outcome per participant was then compared to the mean score of
four-item ratings on recognition of situational forces (alpha=.82), which include “Lisa Jones
received money for her endorsement” and “Lisa Jones took the opportunity to get publicity by
her endorsement.” Considering its conceptual definition, it was determined that correspondence
bias was evoked if a participant’s level of situational force recognition is high or moderate and
his/her level of extended dispositional attribution is high. (The measurement of correspondence
bias will be addressed in detail in the next chapter.)
Finally, ad attitude and brand attitude, the dependent variables posited as attributional
consequences, were measured using the scales adapted and slightly modified from the attitude
studies of MacKenzie and Lutz (1989) and Madden, Allen, and Twible (1988). Both ad attitude
and brand attitude were evaluated with 10-item, seven-point, semantic differential items each,
which were anchored by “good/bad (in general),” “pleasant/ unpleasant,”
“favorable/unfavorable,” “interesting/boring,” “useful/useless,” “convincing/ unconvincing,” and
so on. These items were internally reliable for both attitudes (avg. alpha=.93 for ad attitude; avg.
alpha=.95 for brand attitude). On the other hand, behavioral intention was measured on six
56
descriptive items with a seven-point Likert scale. It was considered that behavioral intention is a
more inclusive measure than purchase intention, a dependent variable tested in many advertising
studies, and thus appropriate for this study, which uses a fictitious brand that is not on the market.
The descriptive items for this measure include “the next time I purchase athletic shoes (or lowfat yogurt), I will buy Stellar (or YoSlim),” “I will consider using Stellar (or YoSlim),” “if a
special sale is offered, I will buy Stellar (or YoSlim),” “I will search for more information about
Stellar (or YoSlim).” A reliability test confirmed that these items were also internally consistent
(avg. alpha=.86). The questionnaires for both experiments are demonstrated in Appendix B-4
and B-5.
57
CHAPTER VI
RESULTS
Summarized results of the two experiments are reported in this chapter. Results of basic
analyses, such as manipulation and group equivalency checks, are reported first, and then each
hypothesis is assessed using an appropriate statistical technique. The reporting is arranged by
the order of the hypotheses, being divided into two sections – Experiment I and Experiment II.
1. Experiment I
Manipulation Checks
Multiple item measures, adapted from Ohanian’s (1990) 10-item source credibility scale
and Zaichkowsky’s (1994) 10-item revised personal involvement inventory, were used to
determine whether the manipulations of source credibility and product involvement were
successful. Reliability tests confirmed a sufficient level of internal consistency for both
measures (alpha=.94 for source credibility; alpha=.95 for product involvement), suggesting no
need to remove any items from the original scales for the manipulation checks. As shown in
Table 1, two independent sample t-tests revealed statistically significant differences between
treatment groups for both factors of source credibility (t= 4.999, p<.001) and product
involvement (t=9.039, p<.001). Thus, group differences on the dependent measures could be
reasonably ascribed to the treatments.
Validation of Dispositional/Situational Attribution Measures
To evaluate the structure of the scales developed to measure dispositional and situational
attributions, the six items of attributional statements were factor analyzed. As indicated in Table
58
2, a principal component analysis with a Varimax rotation supported a two-component solution –
dispositional and situational attribution scales (consisting of three items for each scale) –
accounting for 62.4% of the total variance in the attributional responses. Thus, the original
scales were used without removal of any items. The mutually exclusive nature of dispositional
and situational attributions, suggested by several researchers (e.g., Pilkonis, 1977; Storms, 1973),
was partly verified by an extremely low correlation coefficient between the two summated factor
scores found from a Pearson correlation analysis (r=-.017, p=.809). Both dispositional and
situational attributions were treated as metric variables, each of which had been derived from
composite mean scores of the three items (maximum score = 7).
Effects of Source Credibility
In the present study, source credibility has been posited to have two causal paths: to
dispositional/situational attributions (Hypothesis 1) and to ad/brand evaluations (a partial
requirement for Hypothesis 4). Multiple statistical methods were used to assess each set of the
hypothesized relations.
a. Impact on Dispositional/Situational Attributions
An independent samples t-test was conducted to find whether there was a statistically
significant difference between high and low source credibility treatments in the subjects’
attributional strength. As Table 3 shows, the results revealed a significant treatment effect of
source credibility for only dispositional attributions, although correspondent results to the
hypothesized directions of effects were found for both variables (dispositional: t=2.373, p<.05;
situational: t=-1.592, p=.113). A MANOVA test, which set product involvement (measured) as
another independent variable (moderator), also yielded consistent results (dispositional:
F(1,195)=5.131, p<.05; situational: F(1,195)=2.337, p=.128). From those results, it was found that
59
participants exposed to an endorser with high credibility are more likely to make dispositional
attributions than participants exposed to an endorser with low credibility; but there is no
significant difference in making situational attributions between the two different conditions of
source credibility. Thus, H1-a was supported whereas H1-b was not.
b. Impact on Ad/Brand Evaluations
Although the causal routes from source credibility to ad/brand evaluations were not
specifically hypothesized in this study, it was necessary to test them in order to assess the
mediating role of attributions (Hypothesis 4). As displayed in Table 4, results from an
independent samples t-test indicate that a high source credibility treatment induced more
favorable ad attitude, brand attitude, and behavioral intention than a low source credibility
treatment in the experiment; but the differences between the two treatments were statistically
significant only for ad attitude (t=3.397; p<.01). Therefore, it was found that credibility of a
noncelebrity endorser would have a partial impact on attitudinal factors. (These results were
used for the mediation tests in conjunction with the path analysis results, which will be explained
later.)
Effects of Product Involvement
Product involvement in this study has been hypothesized as a moderating variable in the
relationship between source credibility and dispositional/situational attributions. The
MANCOVA results, which were used earlier for the test of source credibility effects, were
utilized again here to test the posited interaction effects of source credibility and product
involvement. Table 5 shows that there were no significant interaction effects between the two
variables on both dispositional and situational attributions (dispositional: F(1,196)=.106, p=.745;
situational: F(1,196)=1.833, p=.177). Thus, H2-a was not supported, indicating that product
60
involvement does not moderate the causal link between source credibility and attributions. Main
effects of product involvement on both dispositional and situational attributions were also found
to be not statistically significant.
Effects of Dispositional versus Situational Attributions
A series of multiple regression analyses were conducted to assess the predictability of
dispositional and situational attributions on attitudinal and motivational factors. In Hypothesis 3,
it has been assumed that dispositional and situational attributions would have opposite effects on
dependent variables in their direction: i.e., positive impact of dispositional attributions and
negative impact of situational attributions. As presented in Table 6, both dispositional and
situational attributions significantly influenced ad attitude and their directional effects were
consistent with the hypothesized relations (dispositional: β=.345, p<.001; situational: β=-.150,
p<.05). The model with these two variables yielded a F-ratio of 16.503 (p<.001) and explained
14.4% of variance in ad attitude. Thus, H3-a was supported. On the other hand, brand attitude
was predicted by only dispositional attributions (dispositional: β=.222, p<.01; situational: β=.088, p=.203), while behavioral intention was not predicted by either dispositional or situational
attributions (dispositional: β=.126, p=.076; situational: β=.065, p=.358). From these results,
H3-b was partly supported and H3-c was not supported.
To additionally assess multiple direct and indirect causal routes to the dependent
variables, path analyses with two models were performed. The first model (Model 1) posited the
“ad attitude → brand attitude” path whereas the second model (Model 2) assumed the “brand
attitude → ad attitude” path. This comparative analysis was necessary due to a relatively high
correlation between ad attitude and brand attitude (r=.642, p<.001), by which alone the
researcher could not ascribe causal priority to one or the other. As displayed in Figure 5,
61
Model 1 appeared to better explain and predict the causal paths among the hypothesized
variables than Model 2. This “ad attitude → brand attitude” path was also thought to be more
reasonable than the “brand attitude → ad attitude” path because, under the experimental
condition, participants had taken no opportunity to form any sort of brand attitude before they
were exposed to the stimulus advertisement. Also, this path model is consistent with prior
D.A.: Dispositional Attributions
S.A.: Situational Attributions
Model 1
.345*
Dispositional
Attributions
Ad Attitude
.000
.642*(D.A.) .643*(S.A.)
-.009
-.150*
Brand Attitude
.008
.031 (D.A.)
.049 (S.A.)
.525*(D.A.) .523*(S.A.)
Situational
Attributions
Behavioral Intention
.119
Model 2
.224*
Dispositional
Attributions
Brand Attitude
.215*
.594*(D.A.) .633*(S.A.)
-.004
-.092
Ad Attitude
-.097
.031 (D.A.)
Situational
Attributions
.049 (S.A.)
Behavioral Intention
.119
* A path coefficient that is significant at the .05 level
Figure 5. Path Diagrams for Attribution Effects: Experiment I
62
.525*(D.A.)
.523*(S.A.)
findings by MacKenzie, Lutz, and Belch (1986). Thus, Model 1 was chosen for data
interpretation.
The results confirmed that the path of “dispositional and situational attributions → ad
attitude → brand attitude → behavioral intention” is the most significant. In particular, it was
found that ad attitude clearly mediates the effects of dispositional and situational attributions on
brand attitude and behavioral intention. In other words, dispositional and situational attributions
could have indirect effects on brand attitude and behavioral intention via ad attitude, and thus
full rejections of H3-b and H3-c are suspended by this extended analysis. Nonetheless, in
conjunction with the earlier multiple regression results, these findings suggest that ad attitude is
the most predictable consequence of both dispositional and situational attributions.
Mediating Role Tests
The hypothesized mediating effects of attributions on the links between source credibility
and ad/brand evaluations were assessed using the hierarchical regression procedures suggested
by Baron and Kenny (1986). For each mediation hypothesis to be supported, all of the following
four conditions should be met: (1) source credibility has a direct effect on dispositional or
situational attribution; (2) source credibility has a direct effect on ad attitude, brand attitude, or
behavioral intention; (3) dispositional or situational attribution has a direct effect on ad attitude,
brand attitude, or behavioral intention; and (4) the direct effect of source credibility on ad
attitude, brand attitude, or behavioral intention is significantly reduced when dispositional or
situational attribution is included as a covariate
in the regression model.
Although the previous hypothesis tests revealed no statistical significance for several
relations in the first two conditions, all the causal relationships were reassessed to ensure
63
accurate and reliable measurements by converting the categorical nature of the source credibility
factor to a metric trait (utilizing the measured, not manipulated, values of the variable). As
exhibited in Figure 6, a series of regression analyses revealed three significant causal routes,
which satisfy the first three conditions for the mediation test. They are (1) source credibility –
dispositional attribution – ad attitude; (2) source credibility – dispositional attribution – brand
attitude; and (3) source credibility – situational attribution – ad attitude.
β3
β1
Source Credibility
β2
Dispositional
Attribution
β6
β7
Ad Attitude
β9
Brand Attitude
β4
Situational
Attribution
β10
β11
β8
Behavioral Intention
β5
* Standardized Coefficients:
β1=.368 (p<.001); β2=-.147 (p<.05); β3=.439 (p<.001); β4=.328 (p<.001); β5=.121 (p=ns);
β6=.348 (p<.001); β7=.224 (p<.01); β8=.125 (p=ns); β9=-.156 (p<.05); β10=-.092 (p=ns);
β11=.063 (p=ns)
* A bold arrow indicates a significant main effect at the .05 level.
Figure 6. Hierarchical Regression Analyses (first three steps): Experiment I
The final-step tests – corresponding to the fourth condition for the mediation test – were
conducted for only the three routes that had passed the first three conditions for mediation.
Subsequent regression analyses found that, for all the three links, the strength (beta coefficient)
of source credibility effect was reduced when dispositional or situational attribution was included
in each regression equation. Specifically, as indicated in Table 5, the effect of source credibility
dropped by 82.0% in the first route (from β=.439 to β=.360); by 97.0% in the second route (from
β=.439 to β=.426); and by 86.6% in the third route (from β=.328 to β=.284). However, as the
64
source credibility effect still remains significant for all three routes after controlling for the
attribution factor, it was found that the endorser-generating attribution only partially mediates
the links between source credibility and ad/brand attitude. Noticeably, nevertheless, the
mediation strength of dispositional attribution appeared to be much greater than that of
situational attribution for both attitudinal factors. Based on these results, H4-a was supported,
whereas H4-b was partly supported and H4-c was rejected.
2. Experiment II
Group Equivalency Checks
For Experiment II, it was important to attain an acceptable level of equivalency in terms
of demographic profiles of subjects and distributions of scores on two measured independent
variables – product involvement and ad involvement – over three experimental groups (two
treatment groups and one control group). As demonstrated in Table 9 and as stated earlier, no
significant differences among the groups were found in the proportions of both gender and age
(gender: χ2=.116, p=.944; age: χ2=3.647, p=.962 ). Following these tests, whether or not the
levels of both types of involvement significantly differed among the three groups was examined.
The result of a one-way ANOVA test indicates that there are no statistically significant
differences among the groups in the mean scores of those variables (product involvement:
F(2,334)=1.557, p=.212; ad involvement: F(2,334)=2.368, p=.095). Thus, the impact of both product
and ad involvement could be reasonably assessed, as they were found to be not different by the
manipulated variable (source credibility).
Manipulation Check
As in Experiment I, Ohanian’s (1990) 10-item source credibility scale was employed to
assess the successful manipulation of source credibility, the only manipulated variable in
65
Experiment II. This scale was repeatedly found to be internally reliable (alpha=.96). Since,
unlike Experiment I, this stage of the experiment utilized a control group in addition to the
treatment groups by source credibility, it was required to check whether or not this control group
had been successfully established (i.e., retaining a moderate level of source credibility).
As exhibited in Table 10, a one-way ANOVA test followed by LSD post-hoc tests
confirmed statistically significant differences among the three groups – high versus low source
credibility groups plus a control group – in the mean values of perceived (not manipulated)
source credibility composite scores (F(2,334)=23.566, p<.001). A comparative order of the mean
values was also consistent with the intended treatment and control of the source credibility
factor: i.e., a mean of the high source credibility group was the highest (Mhigh=4.5991) and that
of the low source credibility group was the lowest (Mlow=3.7241), while that of the control group
was in the middle (Mcontrol=4.1168). Hence, it was verified that both manipulation and control of
source credibility were successful.
Validation of Dispositional/Situational Attribution Measures
A factor analysis was conducted to assess if the initial scales for dispositional and
situational attributions were validly structured. Consistent with the results in Experiment I, a
principal component analysis with a Varimax rotation confirmed that two components – one for
dispositional attributions and another for situational attributions – provide the best solution for
data reduction, explaining 71.3% of the total variance for attributional responses in combination
(see Table 11). Thus, the original items for attribution measures (three items for each type of
attributions) were statistically validated for the subsequent tests. As in Experiment I, individual
composite scores for each of the dispositional and situational attributions were calculated and
then applied to the attribution tests.
66
Basic Comparisons of Dispositional and Situational Attributions
A paired samples t-test was conducted to find which type of attributions were greater in
an individual’s processing of testimonial messages. As Table 12 indicates, situational
attributions were found to be more greatly evoked than dispositional attributions
(Mdispositional=3.966; Msituational=4.879; p<.001). An interrelation between the two types of
attributions was tested using a Pearson correlation analysis. The result revealed no significant
correlation between dispositional and situational attributions (r=.006, p=.911). Although this
finding is inconsistent with the conceptual opposition of the two types of attributions, such a
mutual exclusiveness justifies separate measurements of each on the independently developed
scales.
Effects of Gender and Age
Since participants in Experiment II were equivalent in gender ratio (male : female =
50.1% : 49.9%; see Table 8) and varied by age groups (see Table 8), the gender and age effects
on attributional and attitudinal factors could be validly assessed. As shown in Table 13, the
results from a series of independent samples t-tests indicate that gender effects are significant for
perceived (measured) source credibility (t=2.022, p<.05), ad involvement (t=3.432, p<.01),
dispositional attributions (t=4.387, p<.001), brand attitude (t=2.439, p<.05), and behavioral
intention (t=4.585, p<.001). Specifically, females appeared to have higher levels of all these
dependent traits than males. It is assumed that these tendencies might be partly derived from the
use of a female endorser for stimulus advertisements. Consequently, the need to control for
gender effects (i.e., treating gender as a covariate) in testing the hypothesized relationships is
suggested by these results.
67
Age effects were not found to be significant for all the tested variables other than product
involvement. A significant impact of age on different levels of product involvement was
consistent with the original expectation, justifying no manipulation of this variable for
Experiment II, which recruited consumer panel subjects whose ages vary. As Table 14 indicates,
a one-way ANOVA test confirmed statistically significant differences in product involvement
among age groups (F(5,334)=3.463, p<.01). Briefly, participants in the younger age groups of 1845 were more involved in athletic shoes, the only selected product for Experiment II, than those
in the older age groups of 45+ (M45-=4.53, M45+=3.92; t=4.072, p<.001). Thus, age effects were
controlled in all tests involving product involvement.
Effects of Source Credibility
a. Impact on Dispositional and Situational Attributions
A one-way ANOVA was used to examine the effect of source credibility on attributions.
As reported in Table 15, the results suggest that source credibility significantly affected both
dispositional and situational attributions (dispositional: F(2,334)=10.176, p<.001; situational:
F(2,334)=3.484, p<.05). The subsequent LSD post-hoc tests specify treatment effects of source
credibility. It was found that there is a significant difference between high and low source
credibility groups in the level of dispositional attributions while both groups are also different
from the control group (Mhigh=4.33, Mlow=3.55, Mcontrol=4.01). However, no significant
difference was found between high and low source credibility groups in the level of situational
attributions (Mhigh=4.84, Mlow=4.74, p<.001). It is inferred that, nevertheless, a significantly
higher level of situational attributions for the control group than those for the two treatment
groups induced such a statistically significant difference among the three groups. From these indepth analyses, H1-a is supported whereas H1-b is rejected.
68
b. Impact of Ad/Brand Evaluations
A series of independent samples t-tests revealed significant differences between the two
treatment groups in terms of the effects of source credibility on ad attitude (t=2.882, p<.01),
brand attitude (t=2.819, p<.01), and behavioral intention (t=2.272, p<.05). As predicted in the
proposed hypotheses, a high source credibility treatment generated more positive ad attitude,
brand attitude, and behavioral intention than a low source credibility treatment. One-way
ANOVA tests (see Table 16) and LSD post-hoc analyses were subsequently conducted to verify
the control group effects on these causal links. Although the significance tests yielded somewhat
complex results, it was found that the control group’s composite score means for all three
dependent variables are higher than those of the low credibility group and lower than those of
high source credibility group. Therefore, the variances in those attitudinal and motivational
factors were reasonably attributable to the treatment effects. These results were used in the later
mediation tests.
Effects of Product Involvement and Ad Involvement
a. Classifying the Levels of Product/Ad Involvement
In order to test the proposed interaction effects of product involvement and ad
involvement with source credibility, metric values on both types of involvement were converted
into ordinal scales according to one-third percentile points by median, thus creating three levels
of involvement: high, moderate, and low. Testing a moderate level of involvement is
advantageous in dealing with the operational oversimplification when involvement is treated as a
dichotomy, typically high and low (e.g., Andrews, Durvasula & Akhter, 1990; Tavassoli, Shultz
& Fitzsimons, 1995). As stated earlier, the levels of product and ad involvement are statistically
independent from source credibility effects. Also, three experimental groups by source
69
credibility treatment and control had similar proportions for each level of product and ad
involvement (see Table 9); and this approximate equivalency was verified by two Chi-square
tests (product involvement: χ2=3.697, p=.449; ad involvement: χ2=5.739, p=.220).
b. Impact on Dispositional and Situational Attributions
It was hypothesized that product involvement and ad involvement will moderate the
effect of source credibility on endorser-generated attributions. A MANCOVA was employed for
the tests to comprehensively and simultaneously examine multiple causal routes to attributional
strength. In the analysis, gender and age were included as covariates because, as reported earlier,
these demographic variables were found to affect the levels of involvement and dispositional
attributions. As shown in Table 17, the analysis found no significant interaction effects between
source credibility and product involvement (dispositional: F=.822, p=.512; situational: F=.393,
p=.814) and between source credibility and ad involvement (dispositional: F=.784, p=.536;
situational: F=.619, p=.650) for both types of attributions. Instead, from the MANCOVA results,
both involvement factors were found to have significant main effects on the two types of
attributions, and these findings were assured by additional tests using one-way ANOVA (see
Table 19 and Table 20 for details). Thus, H2-a and H2-b are rejected, as the hypothesized
moderating effects of both involvement factors were revealed to be statistically insignificant.
c. Impact on Ad/Brand Evaluation (beyond hypothesis tests)
The results from one-way ANOVAs suggest that both product involvement and ad
involvement significantly influence all three dependent variables, i.e., ad attitude, brand attitude,
and behavioral intention (see Table 19 and Table 21). These findings were unexpected because
personal involvement has not been considered to influence attitudinal directions (e.g., positive or
negative). Based on these results about significant main effects of both involvement factors on
70
attributional and attitudinal factors, additional mediation tests (i.e., product/ad involvement →
attributions → ad/brand evaluations) were conducted to obtain deeper implications. (The
relevant results regarding these mediation tests will be reported later in this chapter.)
Effects of Dispositional versus Situational Attributions
As in Experiment I, a series of multiple regression analyses were performed to examine
the differential effects of dispositional and situational attributions on ad/brand evaluations. The
results indicate that all the dependent variables were significantly and positively influenced by
both dispositional and situational attributions. Specifically, as presented in Table 22,
attributional effects were significant for ad attitude (dispositional: β=.452, p<.001; situational:
β=.130, p<.01), brand attitude (dispositional: β=.497, p<.001; situational: β=.183, p<.001), and
behavioral intention (dispositional: β=.536, p<.001; situational: β=.100, p<.05), accounting for
22.2%, 28.2 %, and 29.8% of total variance respectively. Although such positive impacts of
dispositional attributions are consistent with the theory-based expectation, it is surprising that all
directions of situational attribution effects are also positive, which is contrary to the hypothesized
assumptions. Despite these findings, it is important to report that the strength of dispositional
attribution effects appeared to be much greater than that of situational attribution effects for all
the three dependent relations, considering comparative values of the beta coefficient. Based on
these results, all three parts of H3 are partly supported, as directional effects of situational
attributions are inconsistent with those implied in the proposed hypotheses.
Additionally, the same two paths as tested in Experiment I were reassessed in this phase
of the experiment. As demonstrated in Figure 7, the path of “ad attitude → brand attitude →
behavioral intention” (Model 1) appeared to better explain the way attributions influence
consumers’ motivations for brand choice than the path of “brand attitude → ad attitude →
71
behavioral intention” (Model 2). The substantial mediating effects of ad attitude in the multiple
routes from dispositional and situational attributions to brand attitude and behavioral intention
were reconfirmed by this analysis. All these results are consistent with those found in
Experiment I, thus enabling the researcher to make a confident deduction about the effect flows
among attributions and their consequences.
D.A.: Dispositional Attributions
S.A.: Situational Attributions
Model 1
.453*
Dispositional
Attributions
Ad Attitude
.227*
.599*(D.A.) .689*(S.A.)
.326*
.132*
Brand Attitude
.095*
.118 (D.A.)
.184*(S.A.)
.315*(D.A.) .431*(S.A.)
Situational
Attributions
Behavioral Intention
-.001
Model 2
.498*
Dispositional
Attributions
Brand Attitude
.138*
.633*(D.A.) .701*(S.A.)
.326*
.187*
Ad Attitude
.002
.118 (D.A.)
Situational
Attributions
.184*(S.A.)
Behavioral Intention
-.001
* A path coefficient that is significant at the .05 level
Figure 7. Path Diagrams for Attribution Effects: Experiment II
72
.315*(D.A.)
.431*(S.A.)
Mediating Role Tests
The mediating effects of attributions were examined using an identical procedure to that
employed for Experiment I. Also, like the mediation tests for Experiment I, the measured
(metric) values of source credibility were utilized to make the data better fit for regression
analyses. In addition to source credibility, the only independent variable tested for the mediation
tests in Experiment I, product involvement and ad involvement were newly assessed as
antecedent factors in the mediation links, as their main effects on attributions and ad/brand
evaluations appeared to be significant according to the preceding tests. As displayed in Figure 8,
a series of regression analyses revealed that all the tested causal routes are significant at the .05
level, thus satisfying the first three conditions for mediation. These results are in contrast with
those of Experiment I, which confirmed only three significant paths involving attributions.
As the final-step tests, hierarchical regression analyses were conducted. The results
indicate differential effects of dispositional and situational attributions in mediation strength and
significance. As exhibited in Table 23, dispositional attributions significantly mediated all the
effects of source credibility, product involvement, and ad involvement on three dependent
variables. Specifically, when dispositional attributions were controlled in each regression
equation model, the average percentage of the reduced strength of effect (beta coefficient) for
each independent variable was: (from 100% to) 62.0% for source credibility; 35.4% for product
involvement, and 64.3% for ad involvement. It is noteworthy that the mediating effect of
dispositional attributions was especially great for product involvement (e.g., by controlling
dispositional attribution effects, the impacts of product involvement on ad attitude and brand
attitude became no longer significant statistically). In contrast, the mediating effects of
situational attributions appear to be minimal or not significant. Situational attributions did not
73
mediate the effects of source credibility on all dependent variables. Also, its mediation strengths
for both product involvement and ad involvement are relatively very weak (average percentages
of the effect strength dropped by the mediation: 96.2% for product involvement; 97.7% for ad
involvement). From these results, H4-a, H4-b, and H4-c are each partly supported, as
dispositional attributions played significant mediating roles for all the tested causal links while
situational attributions did not.
β3
Source Credibility
β1
Product Involvement
Ad Involvement
β2
Dispositional
Attribution
Ad Attitude
β6
β7
β9
Brand Attitude
β4
Situational
Attribution
β10
β11
β8
Behavioral Intention
β5
* Source credibility, product involvement, and ad involvement were simultaneously regressed on
attributional and attitudinal factors.
* Standardized Coefficients for Source Credibility:
β1=.588 (p<.001); β2=-.150 (p<.05); β3=.369 (p<.001); β4=.423 (p<.001); β5=.247 (p<.001);
β6=.452 (p<.001); β7=.497 (p<.001); β8=.536 (p<.001); β9=.130 (p<.01); β10=.183 (p<.001);
β11=.100 (p<.05)
* Standardized Coefficients for Product Involvement:
β1=.588 (p<.001); β2=-.150 (p<.05); β3=.369 (p<.001); β4=.423 (p<.001); β5=.247 (p<.001);
β6=.452 (p<.001); β7=.497 (p<.001); β8=.536 (p<.001); β9=.130 (p<.01); β10=.183 (p<.001);
β11=.100 (p<.05)
* Standardized Coefficients for Ad Involvement:
β1=.183 (p<.001); β2=.233 (p<.001); β3=.209 (p<.001); β4=.228 (p<.001); β5=.463 (p<.001);
β6=.452 (p<.001); β7=.497 (p<.001); β8=.536 (p<.001); β9=.130 (p<.01); β10=.183 (p<.001);
β11=.100 (p<.05)
Figure 8. Hierarchical Regression Analyses (first three steps): Experiment II
74
Occurrence and Effects of Correspondence Bias
a. Measurement of Correspondence Bias
As briefly described in the previous chapter, the measurement of correspondence bias
was carried out by comparing both levels of the recognition of situational forces and the
extended dispositional attributions. Those measured values on each of the two factors were
classified into three groups – high, moderate, and low – by cutting the distribution on one-third
percentile points by median. Based on its conceptual definition, correspondence bias was
determined to exist when each subject’s level of situational force recognition was high or
moderate and his/her level of the extended dispositional attributions was high. Through this
procedure, a total of 70 participants (20.9%) were identified as having correspondence bias under
the experimental condition.
b. Correspondence Bias as a Criterion Variable
As in the preceding attribution tests, three independent variables – source credibility,
product involvement, and ad involvement – were posited as potential determinants of
correspondence bias. A series of Chi-square tests were performed to verify if there was a
statistically significant difference among the groups (high, moderate, and low) for each variable
in the number of correspondence bias holders. As shown in Table 24, significant levels of Chisquare were found for product involvement (χ2=21.270, p<.001) and ad involvement (χ2=33.721,
p<.001); and a Chi-square value for source credibility almost reached statistical significance at
the .05 level (χ2=5.770, p=.056). On the other hand, as displayed in Table 25, a three-way
ANOVA revealed no interaction effects between source credibility and product/ad involvement;
i.e., both product involvement and ad involvement do not play a moderating role in the causal
link between source credibility and correspondence bias as hypothesized (F=1.499, p=.202;
75
F=1.748, p=.139). Thus, H5-a, H5-b, and H5-c are not fully supported although their main
effects appeared to be significant.
These findings suggest that more participants in the high and controlled (moderate in
their values) source credibility groups generated correspondence bias than those in the low
source credibility group when they were exposed to a testimonial advertisement. On the other
hand, the high involvement groups for both product involvement and ad involvement possessed
considerably more holders of correspondence bias than the moderate and low involvement
groups. Additionally, further Chi-square analyses found that gender and age effects on
correspondence bias occurrence are not statistically significant (gender: χ2=1.882, p=.170; age:
χ2=3.154, p=.676); however, females had correspondence bias to a greater extent than males in
general, as did 18-45 age groups than 45+ age groups.
c. Correspondence Bias as a Predictor Variable
A series of independent samples t-tests were conducted to examine the effects of
correspondence bias on ad/brand evaluations. As indicated in Table 25, the results confirmed
significant differences between holders and non-holders of correspondence bias for all three
dependent variables. In more detail, those with correspondence bias showed more positive ad
attitude (t=4.192, p<.001), brand attitude (t=4.258, p<.001), and behavioral intention (t=6.117,
p<.001) than those without correspondence bias. These results are consistent with the
hypothesized assumptions; thus, H6-a, H6-b, and H6-c are supported. The findings from the
correspondence bias tests indirectly support the significant positive impacts of three independent
factors – source credibility, product involvement, and ad involvement – on advertising
effectiveness.
76
CHAPTER VII
DISCUSSION
In this chapter, the outline of the research is stated, followed by summary and
interpretations of major findings from the two experiments. Subsequent sections discuss
theoretical and managerial implications drawn from the present study, as are study limitations
and directions for future research.
1. Summary of Research
Testimonial advertising, one of the most popular endorsement strategies, entails a
complex context in which consumers’ attitudinal responses to the message are likely to be
affected by many perceptual and motivational factors. While a number of studies have examined
the effectiveness of endorsement approaches over the years, little attention has been paid to
consumer processing of testimonial messages on theoretical perspectives. This research is an
attempt to systematically investigate how consumers make causal inferences regarding the
endorser’s motivation to endorse the brand – specifically, dispositional versus situational
attributions – and how such inferences affect the persuasiveness of a testimonial advertisement.
Based on correspondent inference theory and several persuasion models, it was posited that
consumers will generate predictable patterns of attributional responses to testimonial messages,
which in turn will influence ad and brand evaluations. Mediating roles of endorser-generated
attributions in the causal links from source credibility, product involvement, and ad involvement
to advertising effectiveness factors were also assessed using statistical analyses.
77
Two-stage experiments were administered to attain reliability and generalizability of
findings, being assured by repeated measurements and use of a consumer panel sample. After
three pretests designed to develop experimental instruments, Experiment I was conducted with
231 college students. With the manipulations of source credibility and product involvement,
antecedents and consequences of dispositional versus situational attributions were examined, as
were the mediating roles of the two types of attributions. Although this first stage of the
experiment produced many meaningful results, findings were not highly ensured due to the
limited generalizability of a college student sample over the consumer population, the imbalance
in participants’ gender ratio, and the probable influence of ad involvement on attribution
evocation. Experiment II refined the measurement scales and extended the exploratory scope of
this research, employing 356 (335 after filtering) online consumer panelists as subjects and
additionally testing the effects of ad involvement and correspondence bias. While some findings
are consistent with those from Experiment I and others are not, cautious interpretations of
research results are required to adequately assess theories and proposed hypotheses. The
following section provides a summary of main findings from both experiments in company with
interpretations.
2. Major Findings and Interpretations
For both Experiment I and Experiment II, two-sided endorser-generated attributions –
dispositional and situational attributions – were separately assessed in terms of their antecedents,
consequences, and mediating effects. Since results from statistical analyses only partly support
the original hypotheses about multiple causal links, it is appropriate to say that consumer
attributions may not work for communication effects in exactly the way that was predicted. The
78
two experiments, nevertheless, found many noteworthy points. In the following, these key
findings are summarized and interpreted by research subject.
Antecedents of Dispositional and Situational Attributions
Both experiments yielded consistent results with respect to the positive impact of
source credibility on an individual’s level of dispositional attributions. However, a significant
causal link between source credibility and situational attributions was found only in Experiment
II while the direction of the effect was also inconsistent over the two experiments (i.e., negative
impact in Experiment I and positive impact in Experiment II). From these results, it became
clear that the endorser’s credibility has differential effects on dispositional and situational
attributions. That is, a highly credible endorser is more likely to evoke dispositional attributions
than a less credible endorser; however, a less credible endorser does not necessarily evoke more
situational attributions than a highly credible endorser. These findings suggest that source
credibility determines the likelihood of a perceptual congruence between the testimonial message
and its motivational force (i.e., the probability of evoking dispositional attributions), while
situational constraints surrounding the product endorsement, such as the endorser’s desire to take
model fees and/or self-image publicity, can be commonly attributed as the endorser’s behavioral
motives regardless of the level of credibility he/she has.
Interpreting such source credibility effects on attributions requires additional inferences
derived from other results in this research. One of the most significant findings from the two
experiments is that dispositional and situational attributions are unlikely to occur
interdependently, which was suggested by an extremely low correlation between them (r=-.017
in Experiment I; r=.006 in Experiment II). In other words, a higher level of dispositional
attributions is not necessarily associated with a lower level of situational attributions; and vice
79
versa. For instance, an individual may attribute the endorser’s motivation to both brand
confidence and making money. Another important finding is that consumers tend to have a
higher level of situational attributions than dispositional attributions when they are exposed to
testimonial advertising. Because most adult consumers are sophisticated receivers of advertising,
it is likely that they normally and potentially perceive that people will endorse a particular brand
for self-interested motives rather than because of their true belief in the brand. Findings from
both experiments suggest that this perceptual propensity would seldom change even though
source credibility is high and even though dispositional motivations for the endorsement are
highly attributed. Thus, it is concluded that the level of situational attributions consumers make
when they are exposed to a noncelebrity testimonial advertisement is not significantly affected
by the endorser’s credibility.
Product involvement and ad involvement were also tested as potential predictors of
attributions. In the hypotheses, they were posited to play a moderating role in source credibility
effects on dispositional and situational attributions. Experiment I, in which product involvement
was manipulated and ad involvement was not tested, found no significant moderating effect of
product involvement. Inconsistent results with the original prediction were also observed in
Experiment II. Findings from the second experiment, which tested both product involvement
and ad involvement in the quasi-experimental setting, revealed that the two involvement factors
do not function as moderators as was hypothesized. However, this stage of the experiment
instead found significant main effects of both types of involvement on dispositional and
situational attributions. The findings from Experiment II imply that the higher a consumer’s
product involvement, the higher his/her levels of both dispositional and situational attributions;
and the same holds true for ad involvement.
80
As expected, the effects of product involvement and ad involvement were consistent in
their directions (i.e., positive for all causal routes). These findings can be interpreted in relation
to the predictable association between an individual’s involvement level and intensity of
information processing. The involvement literature suggests that a highly involved person
(whatever the object is a product class or an advertisement) is more likely to engage in messagerelevant cognitive activities than a less involved person (e.g., Michael, 1994; Petty, Cacioppo &
Schumann, 1983). This proposition and the current findings support that involvement may work
as a drive to evoke attributional thinking when consumers see a testimonial advertisement.
Another important issue here is how the level of personal involvement could determine
the strength of each dispositional and situational attributions. An interpretative cue is (again) a
lack of interrelation between the two types of attributions. That is, when people are highly
involved in a product class or an advertisement, they make a strong causal inference about the
endorser’s motivation no matter whether the attributional direction is dispositional or situational;
and the level of one type of attributions hardly affects the level of another.
However, it should be cautioned not to infer that an individual’s strengths for the two kinds of
attributions are equivalent to each other. Those attributional strengths are relative values that
were determined within a single type of attributions; and again, situational attributions were
more highly evoked than dispositional attributions in both experiments.
Consequences of Dispositional and Situational Attributions
Multiple regression analyses found that both dispositional and situational attributions
significantly influence the indicators of advertising effectiveness. The effects of dispositional
attributions were generally consistent over the two experiments. That is, the level of
dispositional attributions positively affected ad attitude, brand attitude, and behavioral intention
81
(although its positive effect on behavioral intention was not statistically significant in
Experiment I). On the other hand, situational attribution effects on the three dependent variables
were confusing and inconsistent over the experiments. In the first experiment, situational
attributions significantly influenced only ad attitude in a negative direction. In contrast, the
second experiment found that, like dispositional attributions, the level of situational attributions
positively influenced all three dependent variables. While the corresponding hypotheses about
the situational attribution effects were rejected by those results, it is questionable how situational
attributions could have positive (also statistically significant) impacts on the multiple indicators
of persuasion.
It sounds very reasonable that testimonial advertising will be more persuasive when the
level of dispositional attributions is high than when it is low, as inferred and confirmed by this
research. However, consumers are unlikely to have positive attitude and behavioral intention
when they attribute the endorsement motivation to situational factors. To understand this
unexpected result, it should be noted that such positive effects of situational attributions were
significant when this attribution factor operated in conjunction with dispositional attributions in
each regression model. Additional statistical tests using one-way ANOVA, which examined the
independent effects of situational attributions, did not support the same impacts on the three
dependent variables in a positive direction.
These findings suggest that situational attributions can predict persuasive effects only
when they are conjoint with predictive capacity of dispositional attributions (i.e., when both
attribution factors are regressed in the equation models for prediction). Although the regression
equation models explain a relatively small proportion (22.2% to 29.8%) of the variance of each
dependent variable, attitudinal and motivational outcomes predicted by the two independent
82
factors are considered to be stable due to absence of multicollinearity in the models (i.e., an
extremely low correlation between dispositional and situational attributions). The results
evidently indicate that situational attributions do not negatively influence persuasion; yet, its
positive impact, which was found in Experiment II and exactly opposite to the hypothesized
direction, remains a hard-to-interpret finding and thus needs further investigations and
discussions of the issue.
Noticeably, dispositional attributions appeared to have a greater impact on ad/brand
evaluations than situational attributions. For both experiments, the compared values of the beta
coefficient signify substantial differences in the extent to which each type of attributions
influences the dependent variables. That is, all the values of dispositional attributions were
consistently much higher than those of situational attributions for ad attitude, brand attitude, and
behavioral intention. It is thus concluded that dispositional attributions are the better predictor of
testimonial advertising effects than situational attributions. On the other hand, it was difficult to
determine which dependent factor is the most predictable consequence of dispositional and
situational attributions, as each stage of experiment yielded inconsistent results: i.e., by
comparing F-values of the three regression models, it was found that ad attitude was most
predictable in Experiment I whereas behavioral intention was most predictable in Experiment II.
These confounding results might be partly derived from an experimental limitation in that
accurate measurements of brand attitude and behavioral intention were, perhaps, somewhat
restricted due to very limited cues about the brand that had been created only for the experiments.
Mediating Roles of Dispositional and Situational Attributions
Assessing the mediating roles of endorser-generated attributions was an effort not only to
empirically test a theoretical mediation model (proposed by Kelley and Michela [1980]) but to
83
explore attributional effects in a broader scope, incorporating the proposed antecedents of
attributions therein. The two experiments confirmed that dispositional and situational
attributions differ in their mediating roles for multiple links between the tested attributional
antecedents and consequences. In Experiment I, it was revealed that an individual’s level of
dispositional attributions mediates the effects of source credibility on ad attitude and brand
attitude, while the mediation effects of situational attributions were minimal or insignificant.
Similar patterns were found in Experiment II. Specifically, source credibility effects on ad
attitude, brand attitude, and behavioral intention were all significantly mediated by the level of
dispositional attributions; in contrast, none of those causal links were significantly mediated by
the level of situational attributions.
As both product involvement and ad involvement were found to have significant main
effects on all designated dependent factors in Experiment II, more causal paths than originally
hypothesized were subject to mediation tests. Findings from the additional tests suggested that
the level of dispositional attributions also significantly mediates the effects of both types of
involvement on all dependent variables, whereas the mediation effects of situational attributions
for the same causal routes were negligible. It is an unexpected but interesting finding that
consumer involvement (with a product class or an advertisement) can have a positive impact on
attitude – other than motivation (e.g., behavioral intention) – via the effects of dispositional
attributions. In short, those consistent results purport that dispositional attributions apparently
play mediating roles in the paths from attributional antecedents to advertising effectiveness
factors; however, situational attributions do not significantly mediate any of those paths.
That a certain variable plays a mediating role signifies a greater effect of the variable on
dependent relationships than postulating merely its main impact on criterion factors – because a
84
mediator not only becomes a direct predictor of a designated dependent variable but also
determines the effect of other factor(s) on the dependent variable. According to the results from
this research, the level of dispositional attributions is likely to determine the extent to which
source credibility, product involvement, and ad involvement influence attitudinal and
motivational factors as indicators of advertising effectiveness. Consequently, dispositional
attributions, rather than situational attributions, are considered as a strong and direct determinant
of the persuasiveness of noncelebrity testimonial advertising. Synthesizing all the relevant
results yielded over the two experiments, this research suggests the following mediation model in
Figure 9, which revises the originally proposed model and has its explanatory power validated by
robust empirical tests:
Source Credibility
Product Involvement
Ad Involvement
Ad Attitude
Dispositional
Attributions
Situational
Attributions
Brand Attitude
Behavioral Intention
Attributional Consequences
Attributional Antecedents
Figure 9. A Revised Mediation Model of Endorser-Generated Attributions
Occurrence and Effects of Correspondence Bias
Originally, the investigation of correspondence bias was not a central part of this
research; however, in connection with new findings from the experiments, exploring this
psychological trait in the context of noncelebrity testimonial advertising became as important as
the integral issues regarding dispositional/situational attributions. Presence of correspondence
bias when consumers are exposed to testimonial messages was well hinted before its actual
85
testing, as the preceding results indicated that dispositional and situational attributions, which are
conceptually opposite to each other, occur in a mutually exclusive fashion (i.e., a presence of one
at a high level does not necessarily predict a presence of the other at a low level, and vice versa).
Correspondence bias, a variable tested in only Experiment II, was clearly evoked by
approximately one-fifth (20.9%) of the experiment participants. They strongly recognized
situational forces (e.g., an opportunity for earning money or self-publicity) that might motivate
an endorser’s decision to endorse the brand; nonetheless, they highly attributed the endorsing
behavior to the endorser’s genuine belief in, experience with, and/or knowledge of the brand
being endorsed. It was also found that such correspondence bias is likely to be potent when the
level of each predictor variable – source credibility, product involvement, and ad involvement –
is high. That is, generally, the more the endorser is credible and the more the consumer is
involved with the product class or the advertisement, the more likely correspondence bias occurs
in the consumer’s mind. However, this study’s finding about the presence of correspondence
bias in the context of noncelebrity-endorsed advertising is contrary to the results of Sorum,
Grape, and Silvery (2003)’s study, which indicated no evidence of correspondence bias in the
peer (typical consumer) endorsement situation. It is assumed that these conflicting findings are
partly due to different approaches to test correspondence bias (Sorum, Grape, and Silvery
[2003]’s study manipulated a situational condition – paid vs. unpaid for the endorsement – and
conducted t-tests) and possible effects of culture and age (their experimental subjects were
Norwegian students).
Not surprisingly, the effects of correspondence bias on advertising effectiveness factors
are very consistent with those of dispositional attributions. Findings suggest that those with
correspondence bias tend to have more favorable ad/brand attitude and behavioral intention than
86
those without correspondence bias. Such a positive role of correspondence bias for persuasive
communication effects is also consistent with the results from Cronley et al.’s (1999) study,
which explored the role correspondence bias plays in the context of celebrity-endorsed
advertising. Although correspondence bias was not evoked among the majority of participants
(in Experiment II), its presence and positive impact on consumer evaluations partly explains why
an advertising approach using a noncelebrity endorser could be effective, as Cronley and her
colleagues inferred their findings to the effectiveness of celebrity endorsements.
Indeed, these current findings about correspondence bias reinforce the comparative
importance of dispositional attributions for persuasion. As implied earlier from the empirical
results of this research, consumers can make both dispositional and situational attributions when
they see an advertisement in which a product endorser appears; however, communication effects
are mostly determined by dispositional attributions rather than situational attributions. It seems
that consumers normally perceive that endorsers in advertising endorse a certain brand primarily
for themselves, not for the brand; thus, the assumption that their endorsement might be
constrained by one or more situational causes has a minimal impact on consumer evaluations of
the advertisement (not advertising in general) and the brand. Instead, such evaluations may be
substantially influenced by the degree to which the endorsement is believed as a true reflection
of brand preference and confidence; and correspondence bias explicates a psychological
condition under which making dispositional attributions is not seriously hindered by consumers’
common expectations of self-interested motives for the endorsement. Accordingly, it is
reasonable to assume that most of those who make strong dispositional attributions possess a
certain degree of correspondence bias (though this research determined the presence or absence
of correspondence bias by a clear-cut standard for a statistical purpose).
87
3. Theoretical and Managerial Implications
Theoretical Implications
The present study proposed and tested hypothetical propositions based upon multiple
theoretical perspectives. Among them, Jones and Davis’ (1965) correspondent inference theory
provided this research with its primary conceptual and theoretical foundation. As reviewed
earlier, this theory suggests two main modes of causal inferences, that is, dispositional
(correspondent) versus situational (noncorrespondent) attributions according to whether or not
an actor (e.g., endorser)’s behavior is perceived to correspond to his/her internal beliefs. Such a
perspective from correspondent inference theory well fit with a primary concern of this research:
i.e., how do consumer inferences about the causes of an endorser’s behavior influence
advertising effectiveness? First of all, this research confirmed the usefulness of separately
assessing dispositional and situational attributions, since these two types of attributions were
found to be independent from each other and to have differential effects on message persuasion.
As Pilkonis (1977) implied, recognizing the mutual independence of dispositional and situational
attributions is important not only to accurately evaluate attribution effects but also to discover an
effective approach for attitude or behavior change.
Although Jones and Davis (1965) did not argue about particular effects of dispositional
and situational attributions, later attribution studies (e.g., Cronley et al., 1999; Sorum, Grape &
Silvera, 2003) suggested a positive impact of dispositional attributions and a negative impact of
situational attributions on consumer attitude. In this research, consistent results were yielded for
dispositional attributions; but results for situational attributions were inconsistent with the
prediction as their effects on attitudinal factors were found to be insignificant or not negative.
Rather, these current findings suggest a new theoretical proposition regarding a positive
88
association between attribution strength (in combination of dispositional and situational
attributions) and attitude. That is, the more strongly an individual attributes the endorser’s
behavior, the more positive attitude toward the message and the message sponsor (i.e., brand or
advertiser) the individual is likely to have. This proposition, however, will not be simply
addressed by a direct relation between attribution strength and attitude; instead, it seems
appropriate to assume that one or more moderating factors, which were untested in this study,
might influence their causal link. Nevertheless, all the positive and significant effects of
dispositional attributions on ad/brand attitude and behavioral intention clearly support a
theoretically and empirically drawn implication with regard to the role of correspondent
inferences.
It was one of the most imperative attempts in this research to test attribution’s mediating
role in multiple causal links between the designated attributional antecedents and consequences.
Kelley and Michela (1980) originally proposed a conceptual model about causal inferences’
mediating process; however, such a mediation effect of attributions has gained little attention in
advertising research. Although the present study incorporated only a limited number of
attributional antecedents and consequences, the multi-step mediation tests with the experimental
data partially support a mediating role of endorser-generated attributions in the testimonial
advertising context – based on the findings that dispositional attributions mediate all the effects
of source credibility, product involvement, and ad involvement on ad/brand attitude and
behavioral intention while situational attributions mediate none of those effects. Such significant
mediation effects of dispositional attributions demonstrate the importance of correspondent
inferences as a latent, but strong, determinant of testimonial advertising effectiveness.
89
Source credibility is an integral concept that a number of advertising scholars have
investigated. Yet, the impact of source credibility on attributional thought has been virtually
unexplored. The present study verified the significance of such an effect: as predicted, source
credibility was found to positively influence the evocation of dispositional attributions, which in
turn induce favorable ad/brand evaluations. This finding is consistent with the traditional
standpoint of the source credibility model, which purports that a highly credible source will have
more positive communication effects than a less credible source (Hass, 1981; Ohanian, 1990;
Sternthal, Phillips & Dholakia, 1978). It is another expected but meaningful discover that source
credibility affects the way the source (endorser)’s behavior is attributed in consumer processing
of testimonial advertisements.
The probable impact of personal involvement on attitude formation or change is well
illustrated in the duel-process models, such as the elaboration likelihood model and heuristicsystematic processing model (e.g., Chaiken, 1980; Petty & Cacioppo, 1985). This research’s
finding that both product involvement and ad involvement have main, not moderating as posited,
effects on attributional and attitudinal factors (other than motivational factors such as purchase
intention) was unexpected and somewhat surprising. Although there were arguments that
involvement can influence the evocation of attributional thinking (e.g., Lord & Smith, 1983) or
attitude strength (e.g., Pomerantz, Chaiken & Tordesillas, 1995), such involvement effects on the
direction (correspondent or noncorrespondent; positive or negative) of either attributions or
attitude were not explicitly suggested in the previous literature. This finding may be partly
explained by a possible inverse (contrary to the prediction) causal effect in the ad involvement –
attributions link: that is, correspondent inferences about the endorser’s motivation might lead
some participants to be more actively engaged in processing ad messages. In any case, such
90
great impacts of both involvement factors provides a rationale to continue to further explore
personal involvement as a strong determinant of persuasion.
Finally, results from the correspondence bias tests confirmed the importance of this
psychological concept in evaluating how consumers process testimonial advertising. As
predicted, occurring conditions and consequences of correspondence bias were exactly consistent
with those of dispositional attributions: i.e., being positively influenced by source credibility and
product/ad involvement; influencing positively ad/brand attitude and behavioral intention. These
findings seem quite reasonable because correspondence bias refers to the psychological
phenomenon of the observer (e.g., consumer) excessively making dispositional attributions from
the actor (e.g., endorser)’s behavior (Gilbert & Malone, 1995). However, correspondence bias is
capable to better explain the way testimonial messages reach persuasion than does mere
dispositional attributions in that most consumers recognize situational drives for an endorsing
behavior, but their “inferential errors” do not bring about unfavorable consequences for
themselves (Gilbert & Malone, 1995) and rather lead to desirable consequences for the advertiser.
In conjunction with Cronley et al.’s (1999) findings within the context of celebrity-endorsed
advertising, this research suggests that correspondence bias is present while consumers process
endorsing messages spoken by either celebrities or noncelebrities and it significantly influences
persuasive effects of the messages. Considering the usefulness of this factor in understanding
the endorsement effects, further investigations of correspondence bias are recommended for
advertising researchers.
Managerial Implications
Results from the present study also provide many implications for advertising practice. A
substantial advantage of this research is in the use of adult consumer panelists, who varied by
91
age, gender, and other demographic and geographic factors, as experimental subjects (in
Experiment II). Thus, findings derived from the data can be well projected over the consumer
population, providing advertisers whose target market may differ with dependable suggestions
for effective endorsement strategies.
Above all, a strong impact of dispositional attributions on advertising effectiveness puts
forward the need to refine advertising techniques to enhance consumer belief that the endorser is
promoting the brand based on his/her true opinion and experience. Findings suggest that
consumers generally perceive situational factors, such as money, self-publicity, or image
enhancement, surrounding the endorsement; however, such perceptions significantly affect
neither making dispositional attributions nor ad/brand evaluations. Thus, advertising
practitioners need to consider how they can stimulate their target consumers to have dispositional
attributions from their advertisement. It is deemed that such a tactical objective can be achieved
by careful selections of two critical elements of testimonial advertising: an endorser and a
testimonial message.
Findings about source credibility effects indicate that an endorser in testimonial
advertising should be credible to maximize persuasive effects. However, it may be somewhat
complex to determine what makes an endorser “credible” because sub-dimensions of source
credibility – expertise and trustworthiness – are not always internally consistent. For example, a
peer (typical consumer) endorser who is perceived as trustworthy may lack expertise with the
product class being endorsed, while the opposite case is possible for a CEO or expert endorser.
If so, it is questionable whether or not such an endorser would be a bad choice for advertising
effectiveness. An ideal selection of a noncelebrity endorser with high expertise and high
trustworthiness will certainly contribute to achieving desired communication effects. Unless this
92
is easy to fulfill, however, advertising practitioners would need to estimate which property of the
noncelebrity endorser will be more important in promoting their brand: e.g., should be the
endorser perceived as knowledgeable? Or is it more crucial to be perceived as honest? To do so,
they also need to take into account the prospective endorser’s visual image, occupational
background, and perceived fit with the product class so that their advertisement can highly evoke
dispositional attributions, which likely lead to persuasion on the part of target consumers.
How a testimonial message should be is also hinted by the considerable impact of
dispositional attributions on persuasion. From the extended inference based on the findings, it is
recommended that a testimonial message must be believable to be effective. This is a very basic
requirement for any good testimonial message; however, this seems to be often overlooked in
many testimonial advertisements. It seems that a frequent mistake is puffery or exaggeration of
product benefits. Since testimonial messages mostly come from product users based on their
actual experience with the brand, such user messages should be distinguished from what the
advertiser can or wants to tell in their own voice. That is, testimonial copy needs to be
comprised of a consumer’s or user’s words – not an advertiser’s words (unless the endorser is a
CEO) – to make its readers believe that the endorser’s message reflects his/her genuine belief in
the endorsed brand.
For a testimonial message to be read, however, the testimonial advertisement as a whole
must first attract consumer attention. Results about ad involvement effects suggest that the more
consumers are involved with a testimonial advertisement, the more they are likely to be
persuaded – via a mediating effect of dispositional attributions. Thus, making target consumers
pay attention to and elaborately process the testimonial message becomes an important
prerequisite of advertising effectiveness, although it is always challenging for advertising
93
practitioners to make their advertisements stand out in the clutter. To achieve desired
communication effects, therefore, advertising creators and planners need to search for the ways
to better present their testimonial advertisements – by improving originality and creativity of the
advertisements and/or by establishing an effective media strategy. This effort will be especially
essential for testimonial advertising that uses noncelebrities whose attention-gaining ability is
generally much lower than that of celebrities.
Lastly, it is necessary to consider whether implications from this research are for
testimonial advertising in general or noncelebrity testimonial advertising in particular. That is,
many findings and indications from the present study bring up the issue of generalizability over
the contexts of both celebrity and noncelebrity endorsements. This is because they differ in the
ways advertising messages are processed and achieve persuasion despite their similar format in
message presentation. For example, consumers may commonly form a positive attitude toward
the brand endorsed by a celebrity they like without processing testimonial messages spoken by
the celebrity; but this is unlikely to be the ordinary case for noncelebrity testimonial advertising.
Although this study does not provide comparative data regarding celebrity versus noncelebrity
endorsement effects, such predictable differences strongly suggest the need for advertising
practitioners to develop different criteria in selecting and presenting their product endorsers for
each approach of celebrity and noncelebrity endorsements. And this endeavor should be assisted
by further investigations of many important issues with respect to the two different streams of
product endorsement strategies.
4. Limitations and Future Research
Despite many significant findings and implications, several limitations are involved in
this research. First, only one type of noncelebrity endorser – a typical consumer – was used for
94
the experiments although, assumedly, this kind of endorser has most frequently appeared in
noncelebrity testimonial advertising. Thus, other major types of noncelebrity endorsers, such as
experts and company people (e.g., CEOs), and their predictable effects were not extensively
examined in this study. As Hass (1981) indicated, different source characteristics can yield
different communication effects. A typical consumer endorser has the unique characteristic of
source similarity, which is not the case for experts or CEOs, but tends to lack perceived expertise
with the endorsing product class in general; probably, such traits might significantly influence
participants’ responses to the given testimonial advertisement. Therefore, testing only one sort
of noncelebrity endorser limits this study’s capacity to fully represent all possible contexts of
noncelebrity-endorsed advertising.
Second, consumer attributions about the endorsement were measured on a forced
condition. That is, participants were asked to evaluate statements regarding particular motives
for the endorsement; thus, their attributional thinking was not voluntarily activated and, for this
reason, the situation in which no attributions are evoked in processing advertising messages was
not considered and tested in this research. As justified earlier, such a forced measure of
consumer attributions was used for the purposes of attaining an acceptable level of statistical
power (by obtaining enough responses on the measure) and encouraging participants to bring out
their latent attributional thoughts. Nonetheless, this might damage the construct validity of the
attribution measurements in that such an operationalization of the concept could impede this
study from achieving the maximum accuracy in relevant assessments.
Third, the two experiments (Experiment I and Experiment II) did not use exactly identical
settings and procedures, in terms of characteristics of human samples (i.e., college students
versus consumer panelists), manipulation of product involvement, use of the control group,
95
advertising copy, and experimental environment (e.g., offline versus online). Thus, several
inconsistencies between findings from the two experiments were partly attributable to one or
more of those initial discrepancies, obstructing the researcher to make confident interpretations
on some conflicting results over the experiments.
Fourth, like most experimental studies of advertising, a satisfactory level of ecological
validity is questionable in this research. Most of all, stimulus magazine-style advertisements
used for the experiments were not exposed in the same context as such an advertisement could be
actually placed, i.e., in a real magazine. The evaluation of brand attitude after showing only a
single advertisement also limits its accuracy of measurement because, in reality, consumers often
have a number of cues and experience that help them perceive a brand’s image and quality. In
addition, the well-justified use of a fictitious brand could discount the sense of a realistic
advertisement, thus possibly affecting participants’ responses on the attitudinal and motivational
measures.
Finally, this research cannot highly assure the generalizability of many findings over
different product classes and different advertising media. Athletic shoes, a product class selected
for both experiments, have a unique characteristic in that they incorporate both functional and
hedonic aspects and thus can appeal to target consumers in either way. Many other product
classes, however, do not have this kind of feature, which may affect consumer responses to
advertising of the product. Likewise, magazines, an advertising medium in which stimulus
advertisements for both experiments were assumed to be shown, differ from other types of
advertising media in their capability for information and image delivery, as well as consumer
habits with the medium. Therefore, the same results may not be guaranteed when testimonial
advertising appears in different media outlets, such as television, radio, and the Internet. The use
96
of only a female endorser for the stimulus testimonial advertisements is another factor to be
considered in that if a male endorser were used, participants’ responses might be a little different
from the current findings. Although a single gender was used to minimize a possible
confounding effect, such a potential impact of the endorser’s gender (e.g., influence on ad
attitude and purchase intention) is a considerable but untested area in this research.
The limited scope of the present study suggests several directions for future research.
First of all, there are still many issues that require further discussions and investigations
regarding comparative aspects of celebrity and noncelebrity endorsements. For instance, how do
they differ in the way the advertisement is processed? How do they differ in communication
effectiveness? And how do they differ in the effects of source characteristics on ad/brand
evaluations? Although noncelebrities are often employed to endorse a brand, most endorsement
studies have dealt with celebrity-endorsed advertising. Future studies of many unexplored issues
relevant to noncelebrity endorsements and comparative assessments of the two endorsement
approaches will help both advertising scholars and practitioners better understand similar or
different effects of these two major streams of endorsement strategies. Such studies will also
contribute to more effective practice of testimonial advertising (e.g., endorser and message
selections).
Consumer attributions in processing advertising messages are another domain for further
explorations. Despite previous efforts by some advertising and social psychology researchers,
attributional procedures and consequences have garnered relatively little attention in consumer
research (Weiner, 2000). Some people would dispute that currently advertising is not read or
viewed with cognitive elaborations such as making causal inferences about any observed object.
Indeed, advertising can be effective by making the brand salient in consumers’ minds through
97
repeated exposures (Moran, 1990); thus, cognitive processing of advertising messages is not
always requisite for persuasion. However, a still considerable amount of advertisements are
likely to be created with the hope that consumers actively read or view and think about the
central message therein. As empirical findings from this research suggests, consumers’
attributional thoughts can significantly influence the persuasive effects of advertisements using a
cognitive appeal (e.g., many testimonial advertisements); thus, the ongoing importance of this
perceptual activity cannot be ignored in future advertising research. In particular,
correspondence bias needs additional attention with respect to the situations under which it is
evoked and its impact on advertising effectiveness. Perhaps this psychological trait will be
indispensable in partly explaining how advertising currently works.
Finally, it is recommended for future research to reassess the hypotheses proposed and
tested in the present study – with more refined measurement scales and treatments. In order to
achieve an acceptable level of measurement validity, this research relied on existing scales,
which were verified to be internally reliable and valid by the originators of the scales, for most
variables tested. However, dispositional/situational attributions and correspondence bias were
not measured in the same way (using existing scales or after the scale developmental process)
because robust sets of measurement scales have not been developed for those variables.
Therefore, future research is suggested to establish more reliable and valid scales to measure
attribution-related variables and then to apply those scales to evaluating the test-retest reliability
of many findings from the current study. The manipulation of source credibility also needs to be
refined. Although its successful manipulation was statistically assured in both experiments, the
endorser’s occupation, the only treated factor for the manipulation, is unlikely to be the only cue
with which consumers can perceive credibility of the source. Accordingly, future studies on
98
these relevant issues are encouraged to develop and utilize a better way to manipulate this
variable; if this effort fails, it would be recommendable to separately assess the two subdimensions of source credibility, i.e., expertise and trustworthiness.
This research is an effort to expand the knowledge regarding a complex and not fully
explored perceptual mechanism in consumer processing of testimonial advertising, a creative
approach frequently used by advertisers. For this purpose, the present study attempted to provide
both theoretical and empirical frames that account for such a mechanism and found many notable
results through rigorous assessments of the proposed hypotheses. However, its contributions to
theory and practice will be ultimately determined by future advances in knowledge that can be
inspired by this study.
99
BIBLIOGRAPHY
Andrews, J. Craig (1988), “Motivation, Ability, and Opportunity to Process Information:
Conceptual and Experimental Manipulation Issues,” in Advances in consumer Research,
Vol.15, Michael J. Houston, ed., Provo, UT: Association for Consumer Research, 219225.
_______________, Srinivas Durvasula, and Syed H. Akhter (1990), “A Framework for
Conceptualizing and Measuring the Involvement Construct in Advertising Research,”
Journal of Advertising, 19(4), 27-40.
Applbaum, Ronald F. and Karl W. Anatol (1972), “The Factor Sturcture of Source Credibility as
a Function of the Speaking Situation,” Speech Monographs, 39(August), 216-222.
Atkin, Charles and Martin Block (1983), “Effectiveness of Celebrity Endorsers,” Journal of
Advertising Research, 23(Februry/March), 57-61.
Baron, Reuben M. and David A. Kenny (1986), “The Moderator-Mediator Variable Distinction
in Social Psychological Research: Conceptual, Strategic and Statistical Considerations,”
Journal of Personality and Social Psychology, 51(6), 1173-82.
Batra, Rajeev, John G. Myers, and David A. Aaker (1996), Advertising Management, 5 th ed.,
Upper Saddle River, NJ: Prentice Hall, Inc..
___________ and Michael L. Ray (1985), “How Advertising Works at Contact,” in
Psychological Processes and Advertising Effects: Theory, Research, and Application,
Linda Alwitt and Andrew Mitchell, eds., Hillsdale, NJ: Lawrence Erlbaum Associates,
13-43.
Belch, George E. and Michael A. Belch (1994), Introduction to Advertising and Promotion: An
Integrated Marketing Communications Perspective, Boston, MA: McGraw-Hill.
Bemmels, Brian (1991), “Attribution Theory and Discipline Arbitration,” Industrial & Labor
Relations Review, 44(3), 548-562.
Bierbrauer, Gunter (1979), “Why did He Do It?: Attribution of Obedience and the Phenomenon
of Dispositional Bias,” European Journal of Social Psychology, 9, 67-84.
Burnett, John J. and Patrick M. Dunne (1986), “An Appraisal of the Use of Student Subjects in
Marketing Research,” Journal of Business Research, 14(August), 329-343.
100
Chaiken, Shelly (1980), “Heuristic versus Systematic Information Processing and the Use of
Source versus Message Cues in Persuasion,” Journal of Personality and Social
Psychology, 39, 752-766.
_____________, Akiva Liberman, and Alice H. Eagly (1989), “Heuristic and Systematic
Information Processing within and beyond the Persuasion Context,” in Unintended
Thought, James. S. Uleman and John A. Bargh eds., New York: Guilford, 212-252.
_____________ and Durairaj Maheswaran (1994), “Heuristic Processing Can Bias Systematic
Processing: Effects of Source Credibility, Argument Ambiguity, and Task Importance on
Attitude Judgment,” Journal of Personality and Social Psychology, 66(3), 460-473.
Choi, Sejung M. (2002), “Attributional Approach to Understanding Celebrity/Product
Congruence Effects: Role of Perceived Expertise,” Doctoral Dissertation, Michigan State
University.
Clow, Kenneth E. and Donald Baack (2002), Integrated Advertising, Promotion & Marketing
Communications, Upper Saddle River, NJ: Prentice Hall, Inc..
Crisci, Richard and Howard Kassinove (1973), “Effects of Perceived Expertise, Strength of
Advice, and Environmental Setting on Parental Compliance,” Journal of Social
Psychology, 89(2), 245-250.
Cronley, Maria L., Frank R. Kardes, Perilou Goddard, and David C. Houghton (1999),
“Endorsing Products for the Money: The Role of the Correspondence Bias in Celebrity
Advertising,” Advances in Consumer Research, 26, 627-631.
Day, Ellen, Marla R. Stafford, and Alejandro Camacho (1995), “Opportunities for Involvement
Research: A Scale Development Approach,” Journal of Advertising, 24(3), 69-75.
Erdogan, B. Zafer, Michael J. Baker, and Stephen Tagg (2001), “Selecting Celebrity Endorsers:
The Practitioner’s Perspective,” Journal of Advertising Research, 41(May/June), 39-48.
Folkes, Valerie S. (1988), “Recent Attribution Research in C`onsumer Behavior: A Review and
New Directions,” Journal of Consumer Research, 14(March), 548-565.
Friedman, Hershey and Linda Friedman (1979), “Endorser Effectiveness by Product Type,”
Journal of Advertising Research, 19(October), 63-71.
_______________, Salvatore Termini, and Robert Washington (1976), “The Effectiveness of
Advertisements Utilizing Four Types of Endorsers,” Journal of Advertising, 5(3), 22-24.
Gilbert, Daniel T. and Patrick S. Malone (1995), “The Correspondence Bias,” Psychological
Bulletin, 117(1), 21-38.
101
Gill, James D., Sanford Grossbart, and Russell N. Laczniak (1988), “Influence of Involvement,
Commitment and Familiarity on Brand Beliefs and Attitudes of Viewers Exposed to
Alternative Ad Claim Strategies,” Journal of Advertising, 17(1), 33-43.
Goldsmith, Ronald E., Barbara A. Lafferty, and Stephen J. Newell (2000), “The Impact of
Corporate Credibility and Celebrity Credibility on consumer Reaction to Advertisements
and Brands,” Journal of Advertising, 29(3), 43-54.
Gotlieb, Jerry B. and Dan Sarel (1991), “Comparative Advertising Effectiveness: The Role of
Involvement and Source Credibility,” Journal of Advertising, 20(1), 38-45.
Green, Susan K., Mary A. Lightfoot, Carole Bandy, and Dale R. Buchanan (1985), “A General
Model of the Attribution Process,” Basic and Applied Social Psychology, 6(2), 159-179.
Harvey, John H. and Gifford Weary (1984), “Current Issues in Attribution Theory and
Research,” Annual Review of Psychology, 35, 427-259.
Hass, R. Glen (1981), “Effects of Source Characteristics on Cognitive Responses and
Persuasion,” in Cognitive Responses to Persuasion, Richard E. Petty, Thomas M. Ostrom,
and Timothy C. Brock eds., NJ: Erlbaum, 141-172.
Heider, Fritz (1958), The Psychology of Interpersonal Relations, New York: John Wiley & Sons.
Higgins, Edward T., Nicholas A. Kuiper, and Jim Olson (1980), “Social Cognition: A Need to
Get Personal,” in E. Higgins, C. Herman, and M. Zanna eds., Social Cognition: The
Ontario Symposium: Vol. 1, Hillsdale, NJ: Lawrence Erlbaum Associates, Inc., 395-420.
Homer, Pamela M. and Lynn R. Kahle (1990), “Source Expertise, Time of Source Identification,
and Involvement in Persuasion: An Elaborative Processing Perspective,” Journal of
Advertising, 19(1), 30-39.
Hovland, Carl I., Irving K. Janis, and Harold H., Kelley (1953), Communication and Persuasion,
New Haven, CT: Yale University Press.
Howard, Jack L. and Gerald R. Ferris (1996), “The Employment Interview Context: Social and
Situational Influences on Interviewer Decisions,” Journal of Applied Social Psychology,
26, 112-136.
Hunt, James M., Jerome B. Kernon, and Richard W. Mizerski (1983), “Causal Inference in
Consumer Response to Inequitable Exchange: A Case of Deceptive Advertising,” in
Advances in Consumer Research, vol.10, Richard P. Bagozzi and Alice M. Tybout, eds.,
Ann Arbor, MI: Association for Consumer Research, 136-141.
Jones, Edward E. and Keith E. Davis (1965), “From Acts to Dispositions,” Advances in
Experimental Psychology, Berkowitz, L. ed., New York: Academic Press, 219-266.
102
Jones, Edward E. and Victor A. Harris (1967), “The Attribution of Attitudes,” Journal of
Experimental Social Psychology, 3, 1-24.
Johnson, Homer and John Scileppi (1969), “Effects of Ego-Involvement Conditions on Attitude
Change to High and Low Credibility Communicators,” Journal of Personality and Social
Psychology, 13(September), 31-36.
Joseph, W. Benoy (1982), “The Credibility of Physically Attractive Communicators: A Review,”
Journal of Advertising, 11(3), 15-24.
Kahle, Lynn R. and Pamela M. Homer (1985), “Physical Attractiveness of Celebrity Endorser: A
Social Adaptation Perspective,” Journal of consumer Research, 11(March), 954-961.
Kamins, Michael A. (1989), “Celebrity and Noncelebrity Advertising in a Two-Sided Context,”
Journal of Advertising Research, 29(June/July), 34-42.
Kelley, Harold H. (1967), “Attribution Theory in Social Psychology,” in Nebraska Symposium
on Motivation, David Levine, ed., Lincoln, NE: University of Nebraska Press, 192-238.
______________ (1973), “The Process of Causal Attribution,” American Psychologist, 28(2),
107-128.
______________ and John L. Michela (1980), “Attribution Theory and Research,” Annual
Review of Psychology, 31, 457-501.
Kerber, Kenneth W. and Royce Singleton Jr. (1984), “Trait and Situational Attributions in a
Naturalistic Setting: Familiarity, Liking, and Attribution Validity,” Journal of Personality,
52(3), 205-219.
Lacznick, Russell N. and Darrel D. Muehling (1993), “The Relationship Between Experimental
Manipulations and Tests of Theory in and Advertising Message Involvement Context,”
Journal of Advertising, 22(3), 59-74.
________________, Thomas E. DeCarlo, and Sridhar N. Ramaswami (2001), “Consumers’
Responses to Negative Word-of-Mouth Communication: An Attribution Theory
Perspective,” Journal of Consumer Psychology, 11(July), 57-73.
Lancaster, Lisa, K. Elaine Royal, and Harold D. Whiteside (1995), “Attitude Similarity and
Evaluation of a Woman’s Athletic Team,” Journal of Social Behavior and Personality,
10, 885-890.
Lee, Yih Hwai (2000), “Manipulating Ad Message Involvement through Information
Expectancy: Effects on Attitude Evaluation and Confidence”, Journal of Advertising,
29(2), 29-43.
103
Lord, Kenneth R., Myung-Soo Lee, and Paul L. Sauer (1995), “The Combined Influence
Hypothesis: Central and Peripheral Antecedents of Attitude toward the Ad,” Journal of
Advertising, 26(1), 73-85.
Lord, Robert G. and Jonathan E. Smith (1981), “Theoretical, Information Processing, and
Situational Factors Affecting Attribution Theory Models of Organizational Behavior,”
Academy of Management Review, 8(1), 50-60.
MacKenzie, Scott B. and Richard J. Lutz (1989), “An Empirical Examination of the Structural
Antecedents of Attitude toward the Ad in an Advertising Pretesting Context,” Journal of
Marketing, 53(April), 48-65.
________________, Richard J. Lutz, and George E. Belch (1986), “The Role of Attitude toward
the Ad as a Mediator of Advertising Effectiveness: A Test of Competing Explanations,”
Journal of Marketing Research, 23(May), 130-143.
Madden, Thomas J., Chris T. Allen, Jacqueline L. Twible (1988), “Attitude toward the Ad: An
Assessment of Diverse Measurement Indices under Different Processing ‘Sets’”, Journal
of Marketing Research, 25(3), 242-252.
McCracken, Grant (1989), “Who Is the Celebrity Endorser?: Cultural Foundations of the
Endorsement Process,” Journal of Consumer Research, 16(December), 310-321.
McGarry, James and Clyde Hendrick (1974), “Communication Credibility and Persuasion,”
Memory and Cognition, 2(January), 82-86.
McGinnies, Elliott (1973), “Initial Attitude, Source Credibility, and Involvement as Factors in
Persuasion,” Journal of Experimental Social Psychology, 9(July), 285-296.
_______________ and Charles D. Ward (1980), “Better Liked Than Right: Trustworthiness and
Expertise as Factors in Credibility,” Personality and Social Psychology Bulletin, 6(3),
467-472.
McGuire, William J. (1985), “Attitudes and Attitude Change,” in Handbook of Social
Psychology, vol2, Gardner Lindzey and Elliot Aronson, eds., New York: Random House,
233-346.
Michael, Pfau (1994), “Impact of Product Involvement, Message Format, and Receiver Sex on
the Efficacy of Comparative Advertising Messages,” Communication Quarterly, 42(3),
244-258.
Mitchell, Andrew A. (1981), “Dimensions of Advertising Involvement,” in Advances in
Consumer Research, vol.8, Kent B. Monroe, ed., Ann Arbor, MI: Association for
Consumer Research, 25-30.
104
Moran, William T. (1990), “Brand Presence and the Perceptual Frame,” Journal of Advertising
Research, 30(October/November), 9-16.
Muehling, Darrel D., Ressell N. Laczniak, and Jeffrey J. Stoltman (1991), “The Moderating
Effects of Ad message Involvement: A Reassessment,” Journal of Advertising, 20(2),
29-38.
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.
_______________ (1991), “The Impact of Celebrity Spokespersons’ Perceived Image on
Consumers’ Intention to Purchase,” Journal of Advertising Research, 31
(February/March), 46-54.
Olson, James M. & Michael Ross (1985), “Attribution Research: Past Contributions, Current
Trends, and Future Prospects,” in Attribution in contemporary psychology. John H.
Harvey and Gifford Weary, eds., New York: Academic Press.
Peterson, Robert A. (2001), “On the Use of College Students in Social Science Research:
Insights from a Second-Order Meta-Analysis,” Journal of Consumer Research,
28(December), 450-461.
Petty, Richard E. and John T. Cacioppo (1981), “Issue Involvement as a Moderator of the Effects
on Attitude of Advertising Content and Context,” in Advances in Consumer Research,
vol.8, Kent B. Monroe, ed., Ann Arbor, MI: Association for Consumer Research, 20-24.
______________ and John T. Cacioppo (1986), Communication and Persuasion: Central and
Peripheral Routes to Attitude Change, New York: Springer-Verlag.
______________, John T. Cacioppo, and Richel Goldman (1981), “Personal Involvement as a
Determinant of Argument-Based Persuasion,” Journal of Personality and Social
Psychology, 41(5), 847-855.
______________, John T. Cacioppo, and Schumann, D. (1983), “Central and Peripheral Routes
to Advertising Effectiveness: The Moderating Role of Involvement,” Journal of
Consumer Research, 10(September), 135-146.
Petroshius, Susan M. and Kenneth E. Crocker (1989), “An Empirical Analysis of Spokesperson
Characteristics on Advertisement and Product Evaluation,” Journal of the Academy of
Marketing Science,” 17(3), 217-225.
Pilkonis, Paul A. (1977), “The Influence of Consensus Information and Prior Expectations on
Dispositional and Situational Attributions,” Journal of Social Psychology, 101(2), 267279.
105
Pomerantz, Eva M., Shelly Chaiken, and Rosalind S. Tordesillas (1995), “Attitude Strength and
Resistance Processes,” Journal of Personality and Social Psychology, 69(3), 408-419.
Ross, Lee D., Teresa M. Amabile, and Julia L. Steinmetz (1977), “Social Roles, Social Control,
and Biases in Social Perception Processes. Journal of Personality and Social Psychology,
35, 485-494.
Settle, Robert B. and Linda L. Golden (1974), “Attribution Theory and Advertiser Credibility,”
Journal of Marketing Research, 11, 181-185.
Shimp, Terence (1981), “Attitude toward the Ad as a Mediator of Consumer Brand Choice,”
Journal of Advertising, 10(2), 9-15.
_____________ (2002), Advertising, Promotion and Supplemental Aspects of Integrated
Marketing Communications, 6th ed., Cincinnati, OH: South-Western College Pub.
Smith, Robert E. and Shelby D. Hunt (1978), “Attributional Processes and Effects in
Promotional Situations,” Journal of Consumer Research, 3(December), 149-158.
Sorum, Kim A., Kjetil M. Grape, and David Silvera. (2003), “Do Dispositional Attributions
Regarding Peer Endorsers Influence Product Evaluations?,” Scandinavian Journal of
Psychology, 44, 39-46.
Sparkman, Richard M. Jr. and William B. Locander (1980), “Attribution Theory and Advertising
Effectiveness,” Journal of Consumer Research, 7(December), 219-224.
Stephens, Nancy and William T. Faranda (1993), “Using Employees as Advertising
Spokespersons,” Journal of Services Marketing, 7(2), 36-46.
Sternthal, Brian, Ruby Dholakia, and Clark Leavitt (1978), “The Persuasive Effect of Source
Credibility: Tests of Cognitive Response,” Journal of Consumer Research, 4(March),
252-260.
Storms, Michael D. (1973), “Videotape and the Attribution Process: Reversing Actors’ and
Observers’ Points of View,” Journal of Personality and Social Psychology, 27, 165-175.
Tavassoli, Nader T., Clifford J. Shultz, II, and Gavan J. Fitzsimons (1995), “Program
Involvement: Are Moderate Levels Best for Ad Memory and Attitude Toward the Ad?,”
Journal of Advertising Research, 35(September/October), 61-72.
Till, Brian D. and Michael Busler (2000), “The Match-Up Hypothesis: Physical Attractiveness,
Expertise, and the Role of Fit on Brand Attitude, Purchase Intent and Brand Beliefs,”
Journal of Advertising, 29(3), 1-13.
Tripp, Carolyn, Thomas D. Jensen, and Les Carlson (1994), “The Effects of Multiple Product
Endorsements by Celebrities on Consumers’ Attitudes and Intentions,” Journal of
Consumer Research, 20(March), 535-547.
106
Weiner, Bernard (1986), An Attribution Theory of Motivation and Emotion, New York:
Springer-Verlag.
_____________ (2000), “Attributional Thoughts about Consumer Behavior,” Journal of
Consumer Research, 27(December), 382-387.
Wells, William, John Burnett, and Sandra Moriarty (2002), Advertising: Principles and Practice,
6th ed., Upper Saddle River, NJ: Prentice Hall, Inc.
Yalch, Richard F. and Suzanne Yoshida (1983), “Consumer Use of Information and Confidence
in Making Judgments about a New Food Store: An Attribution Theory Experiment,” in
Advances in Consumer Research, Vol. 10, Richard P. Bagozzi and Alice M. Tybout, eds.,
Ann Arbor, MI: Association for Consumer Research, 40-44.
Zaichkowsky, Judith L. (1985), “Measuring the Involvement Construct,” Journal of Consumer
Research, 12(December), 341-352.
___________________ (1986), “Conceptualizing Involvement,” Journal of Advertising, 15(2),
4-14.
___________________ (1994), “The Personal Involvement Inventory: Reduction, Revision,
Application to Advertising,” Journal of Advertising, 23(4), 59-70.
107
APPENDICIES
APPENDIX A. TABLES
APPENDIX B. QUESTIONNAIRES
APPENDIX C. STIMULUS ADVERTISEMENTS
108
APPENDIX A
TABLES
109
Table 1. Independent Samples T-Tests: Manipulation Checks
(Experiment I)
N
Mean
(max.=7)
Std. Dev.
t-value
Sig.
100
100
4.630
3.876
1.105
1.027
4.999
.000
100
100
5.069
3.645
.888
1.302
9.039
.000
Variable
Source Credibility
High
Low
Product Involvement
High
Low
Table 2. Factor Analysis: Dispositional vs. Situational Attributions
(Experiment I)
Item
Belief in the brand
Experience with the brand
Knowledge about the product
Money
Self-publicity
Non-monetary compensation
Factor 1
Dispositional Attributions
.787
.877
.866
(-.314)
(.004)
(.275)
Factor 2
Situational Attributions
(-.122)
(.047)
(.085)
.761
.706
.571
* Rotation Method: Varimax with Kaiser Normalization
Table 3. Independent Samples T-Tests: Source Credibility (SC) Effects on Attributions
(Experiment I)
Variable
Dispositional Attributions
High SC
Low SC
Situational Attributions
High SC
Low SC
N
Mean
(max.=7)
Std. Dev.
t-value
Sig.
100
100
3.617
3.230
1.004
2.183
2.373
.019
100
100
4.760
4.977
.958
.967
-1.592
.113
110
Table 4. Independent Samples T-Tests: Source Credibility (SC) Effects
on Ad/Brand Evaluations (Experiment I)
N
Mean
(max.=7)
Std. Dev.
t-value
Sig.
High SC
Low SC
100
100
4.140
3.654
.891
1.118
3.397
.001
High SC
Low SC
Behavioral Intention
High SC
Low SC
100
100
4.222
4.023
.948
1.001
1.444
.150
100
100
4.202
4.055
1.137
1.206
.885
.377
Variable
Ad Attitude
Brand Attitude
Table 5. MANCOVA: Effects of Source Credibility and Product Involvement
on Attributions (Experiment I)
• Independent Variables: Source Credibility, Product Involvement
• Covariate: Gender
• Dependent Variables: Dispositional Attributions, Situational Attributions
Variable
Source Credibility
Dispositional
Situational
Product Involvement
Dispositional
Situational
Source Credibility
x Product Involvement
Dispositional
Situational
Gender
Dispositional
Situational
Corrected Model
Dispositional
Situational
SS
df
MS
F-value
Sig.
61.208
19.241
1
1
61.208
19.241
5.131
2.337
.025
.128
.389
20.317
1
1
.389
20.317
.033
2.468
.857
.118
2.679
17.285
1
1
2.679
17.285
.225
2.100
.636
.149
37.818
11.576
1
1
37.818
11.576
3.170
1.406
.077
.237
107.358
66.431
4
4
26.839
16.608
2.250
2.017
.065
.094
* Dispositional Attribution: R2=.044 (Adjusted R2=.025)
* Situational Attribution: R2=.040 (Adjusted R2=.020)
111
Table 6. Multiple Regression Analyses: Attribution Effects on Ad/Brand Evaluations
(Experiment I)
Variable
Ad Attitude
Dispositional
Situational
Brand Attitude
Dispositional
Situational
Behavioral Intention
Dispositional
Situational
B
SE
Beta
Sig.
R2
Adjusted R2
.922
.482
.176
.212
.345
-.150
.000
.024
.144
.135
.621
-.298
.193
.233
.222
-.088
.002
.203
.058
.048
.253
.158
.142
.171
.126
.065
.076
.358
.020
.010
Table 7. Hierarchical Regression Analyses (Step 4): Mediating Effects of Attributions
(Experiment I)
* Comparison of standardized coefficients (β) between “before” and “after” including dispositional or
situational attribution as a covariate in the regression model of source credibility and each evaluative factor.
(For only the three relations that have passed the first three steps)
Variable
Before Inclusion
After Inclusion
Beta
Sig.
R2
Beta
Sig.
R2
.439
.328
.000
.000
.193
.108
.360
.284
.000
.000
.233
.120
.439
.000
.193
.426
.149
.202
Ad Attitude
Dispositional
Situational
Brand Attitude
Dispositional
112
Table 8. Demographic Profiles of Sample
(Experiment II)
a. Gender
Gender
Male
Female
Total
Frequency
168
167
335
Percent (%)
50.1
49.9
100.0
Age Group
18 – 24
25 – 34
35 – 44
45 – 54
55 – 64
Over 65
Total
Frequency
26
109
124
31
31
14
335
Percent (%)
7.8
32.5
37.0
9.3
9.3
4.2
100.0
Frequency
290
9
14
12
10
335
Percent (%)
86.6
2.7
4.2
3.6
3.0
100.0
Frequency
11
39
22
18
101
49
33
13
49
335
Percent (%)
3.3
11.6
6.6
5.4
30.1
14.6
9.9
3.9
14.6
100.0
b. Age
c. Ethnic Origin
Ethnic Origin
Caucasian
African-American
Asian
Latino (Hispanic)
Other
Total
d. Occupation
Occupation
College Student
Homemaker
Skilled Worker
Sales
Professional
Management
Self-employed
Unemployed
Other
Total
113
Table 9. Chi-Square Tests: Group Equivalency Check
(Experiment II)
a. Gender
Group
N
High (Source Credibility)
116
Low (Source Credibility)
112
Control
107
Male
57
(49.1%)
56
(50.0%)
55
(51.4%)
Female
59
(50.9%)
56
(50.0%)
52
(48.6%)
χ2
Sig.
.116
(df=2)
.944
b. Age
Group
High
Low
Control
Sig.
18-24
25-34
35-44
45-54
55-64 Over65
χ2
7
42
39
11
11
6
116
(6.0%) (36.2%) (33.6%) (10.7%) (10.7%) (4.8%)
10
35
40
11
11
5
3.647
.962
112
(8.9%) (31.3%) (35.7%) (9.8%) (9.8%) (4.5%) (df=10)
9
32
45
9
9
3
107
(8.4%) (29.9%) (42.1%) (8.4%) (8.4%) (2.8%)
N
c. Product Involvement
Group
N
High
116
Low
112
Control
107
High
34
(29.3%)
36
(32.1%)
41
(38.3%)
Moderate
37
(31.9%)
31
(27.7%)
34
(32.6%)
Low
45
(38.8%)
45
(40.2%)
32
(29.9%)
High
40
(34.5%)
29
(25.9%)
42
(39.3%)
Moderate
43
(37.1%)
40
(35.7%)
33
(30.8%)
Low
33
(28.4%)
43
(38.4%)
32
(29.9%)
χ2
Sig.
3.697
(df=4)
.449
χ2
Sig.
5.739
(df=4)
.220
d. Ad Involvement
Group
N
High
116
Low
112
Control
107
114
Table 10. One-Way ANOVA: Manipulation Check
(Experiment II)
Group
(by Source Credibility)
High
Low
Control (Moderate)
Mean
(max.=7)
4.599
3.702
4.117
N
116
112
107
Std. Dev.
F-value
Sig.
.842
.950
1.094
23.566
.000
* Between groups df=2; Within groups df=332
Table 11. Factor Analysis: Dispositional vs. Situational Attributions
(Experiment II)
Factor 1
Dispositional Attributions
.930
.923
.909
(.031)
(-.315)
(.259)
Item
Belief in the brand
Experience with the brand
Knowledge about the product
Money
Self-publicity
Non-monetary compensation
Factor 2
Situational Attributions
(.023)
(-.016)
(.034)
.822
.787
.522
* Rotation Method: Varimax with Kaiser Normalization
Table 12. Paired Samples T-Test: Dispositional vs. Situational Attributions
(Experiment II)
Variable
Dispositional Attributions
Situational Attributions
N
335
Mean
(max.=7)
3.966
4.879
* Pearson Correlation: r=.006 (p=.911)
115
Std. Dev.
t-value
Sig.
1.355
.943
-10.146
.000
Table 13. Independent Samples T-Tests: Gender Effects (Experiment II)
Variable
Source Credibility (measured)
Male
Female
Product Involvement
Male
Female
Ad Involvement
Male
Female
Dispositional Attributions
Male
Female
Situational Attributions
Male
Female
Ad Attitude
Male
Female
Brand Attitude
Male
Female
Behavioral Intention
Male
Female
N
Mean
(max.=7)
Std. Dev.
t-value
Sig.
168
167
4.040
4.266
1.052
.992
-2.022
.044
168
167
4.358
4.422
1.191
1.175
-.494
.622
168
167
3.885
4.287
1.046
1.095
-3.432
.001
168
167
3.651
4.283
1.358
1.280
-4.387
.000
168
167
4.911
4.846
.948
.939
.625
.533
168
167
4.246
4.413
1.069
1.275
-1.301
.194
168
167
4.098
4.362
.907
1.072
-2.439
.015
168
167
3.619
4.274
1.394
1.213
-4.585
.000
Table 14. One-Way ANOVAs: Age Effects (Experiment II)
Variable
Source Credibility (measured)
Product Involvement
Ad Involvement
Dispositional Attributions
Situational Attributions
Ad Attitude
Brand Attitude
Behavioral Intention
F-value
.661
3.463
1.144
1.585
.673
.908
1.113
.938
* Between groups df=5; Within groups df=329
116
Sig.
.653
.005
.337
.164
.644
.476
.353
.456
Table 15. One-Way ANOVAs: Source Credibility (SC) Effects on Attributions
(Experiment II)
N
Mean
(max.=7)
Std. Dev.
Dispositional Attributions
High SC
Low SC
Control
116
112
107
4.333
3.548
4.006
1.198
1.346
1.414
Situational Attributions
High SC
Low SC
Control
116
112
107
4.842
4.738
5.065
.903
.982
.922
Variable
F-value
Sig.
10.176
.000
3.484
.032
Table 16. One-Way ANOVAs: Source Credibility (SC) Effects on Ad/Brand Evaluations
(Experiment II)
N
Mean
(max.=7)
Std. Dev.
High SC
Low SC
Control
116
112
107
4.588
4.157
4.229
1.188
1.063
1.242
High SC
Low SC
Control
116
112
107
4.400
4.054
4.229
.959
.894
1.119
116
112
107
4.086
3.698
4.051
1.257
1.324
1.434
Variable
F-value
Sig.
4.474
.012
3.470
.032
Ad Attitude
Brand Attitude
Behavioral Intention
High SC
Low SC
Control
117
2.895
.057
Table 17. MANCOVA: Effects of Three Independent Variables on Attributions
(Experiment II)
• Independent Variables: Source Credibility, Product Involvement, Ad Involvement
• Covariate: Gender, Age
• Dependent Variables: Dispositional Attributions, Situational Attributions
Variable
Source Credibility (SC)
Dispositional
Situational
Product Involvement (PI)
Dispositional
Situational
Ad Involvement (AI)
Dispositional
Situational
SC x PI
Dispositional
Situational
SC x AI
Dispositional
Situational
PI x AI
Dispositional
Situational
SC x PI x AI
Dispositional
Situational
Gender
Dispositional
Situational
Age
Dispositional
Situational
Corrected Model
Dispositional
Situational
SS
df
MS
F-value
Sig.
22.064
3.663
2
2
11.032
1.831
8.400
2.107
.000
.123
21.880
3.081
2
2
10.940
1.540
8.330
1.772
.000
.172
51.507
7.432
2
2
25.753
3.716
19.610
4.275
.000
.015
4.319
1.366
4
4
1.080
.341
.822
.393
.512
.814
4.119
2.151
4
4
1.030
.538
.784
.619
.536
.650
2.126
3.025
4
4
.532
.756
.405
.870
.805
.482
10.213
1.932
8
8
1.277
.241
.972
.278
.458
.973
15.324
1.236
1
1
15.324
1.236
11.668
1.422
.001
.234
1.875
.969
1
1
1.875
.969
1.428
1.115
.233
.292
211.531
30.891
28
28
7.555
1.103
5.753
1.269
.000
.169
* Dispositional Attribution: R2=.345 (Adjusted R2=.285)
* Situational Attribution: R2=.104 (Adjusted R2=.022)
118
Table 18. One-Way ANOVAs: Product Involvement (PI) Effects on Attributions
(Experiment II)
Variable
Dispositional Attributions
High PI
Moderate PI
Low PI
N
Mean
(max.=7)
Std. Dev.
F-value
Sig.
116
112
107
4.538
3.931
3.475
1.280
1.231
1.331
19.928
.000
116
112
107
5.075
4.814
4.754
.903
.982
.922
3.778
.024
Situational Attributions
High PI
Moderate PI
Low PI
Table 19. One-Way ANOVAs: Product Involvement (PI) Effects on Ad/Brand Evaluations
(Experiment II)
Variable
N
Mean
(max.=7)
Std. Dev.
F-value
Sig.
116
112
107
4.515
4.538
3.985
1.275
.964
1.178
8.568
.000
116
112
107
4.510
4.256
3.953
1.188
.759
.920
9.546
.000
116
112
107
4.478
3.904
3.496
1.360
1.186
1.294
17.033
.000
Ad Attitude
High PI
Moderate PI
Low PI
Brand Attitude
High PI
Moderate PI
Low PI
Behavioral Intention
High PI
Moderate PI
Low PI
119
Table 20. One-Way ANOVAs: Ad Involvement (AI) Effects on Attributions
(Experiment II)
Variable
Dispositional Attributions
High AI
Moderate AI
Low AI
N
Mean
(max.=7)
Std. Dev.
F-value
Sig.
116
112
107
4.763
3.848
3.275
1.190
1.189
1.265
41.934
.000
116
112
107
5.153
4.744
4.741
.912
.844
1.018
7.306
.001
Situational Attributions
High AI
Moderate AI
Low AI
Table 21. One-Way ANOVAs: Ad Involvement (AI) Effects on Ad/Brand Evaluations
(Experiment II)
Variable
N
Mean
(max.=7)
Std. Dev.
F-value
Sig.
116
112
107
4.995
4.080
3.913
1.162
1.012
1.067
32.097
.000
116
112
107
4.783
4.051
3.853
1.066
.850
.826
31.325
.000
116
112
107
4.839
3.843
3.136
1.037
1.187
1.236
60.108
.000
Ad Attitude
High AI
Moderate AI
Low AI
Brand Attitude
High AI
Moderate AI
Low AI
Behavioral Intention
High AI
Moderate AI
Low AI
120
Table 22. Multiple Regression Analyses: Attribution Effects on Ad/Brand Evaluations
(Experiment II)
Variable
Ad Attitude
Dispositional
Situational
Brand Attitude
Dispositional
Situational
Behavioral Intention
Dispositional
Situational
B
SE
Beta
Sig.
R2
Adjusted R2
.393
.162
.042
.060
.452
.130
.000
.008
.222
.217
.367
.195
.034
.049
.497
.183
.000
.000
.282
.277
.532
.143
.046
.066
.536
.100
.000
.030
.298
.294
Table 23. Hierarchical Regression Analyses (Step 4): Mediating Effects of Attributions
(Experiment II)
* Comparison of standardized coefficients (β) between “before” and “after” including dispositional or
situational attributions as a covariate in the regression model of each of three independent variables
(source credibility, product involvement, ad involvement) and each of three dependent variables (ad
attitude, brand attitude, behavioral intention)
a. Independent Variable: Source Credibility
Variable
Before Inclusion
After Inclusion
Beta
Sig.
R2
Beta
Sig.
R2
.481
.481
.000
.000
.232
.232
.322
.485
.000
.000
.260
.252
.545
.545
.000
.000
.297
.297
.384
.550
.000
.000
.326
.337
.495
.495
.000
.000
.245
.245
.240
.498
.000
.000
.318
.258
Ad Attitude
Dispositional
Situational
Brand Attitude
Dispositional
Situational
Behavioral Intention
Dispositional
Situational
121
(Table 23 Continued)
b. Independent Variable: Product Involvement
Variable
Before Inclusion
After Inclusion
Beta
Sig.
R2
Beta
Sig.
R2
.253
.253
.000
.000
.064
.064
.064
.242
.239
.000
.208
.076
.288
.288
.000
.000
.083
.083
.083
.272
.117
.000
.253
.108
.408
.408
.000
.000
.167
.167
.212
.402
.000
.000
.324
.171
Ad Attitude
Dispositional
Situational
Brand Attitude
Dispositional
Situational
Behavioral Intention
Dispositional
Situational
c. Independent Variable: Ad Involvement
Variable
Before Inclusion
After Inclusion
Beta
Sig.
R2
Beta
Sig.
R2
.407
.407
.000
.000
.166
.166
.241
.396
.000
.000
.249
.171
.454
.454
.000
.000
.206
.206
.274
.436
.000
.000
.304
.220
.595
.595
.000
.000
.354
.354
.436
.593
.000
.000
.431
.354
Ad Attitude
Dispositional
Situational
Brand Attitude
Dispositional
Situational
Behavioral Intention
Dispositional
Situational
122
Table 24. Chi-Square Tests: Effects of Three Independent Variables
on Correspondence Bias (Experiment II)
a. Source Credibility (manipulated)
Level
N
High
116
Low
112
Control (Moderate)
107
Correspondence Bias
Yes
No
28
88
(24.1%)
(75.9%)
15
97
(13.4%)
(86.6%)
27
80
(25.2%)
(74.8%)
χ2
Sig.
5.770
(df=2)
.056
χ2
Sig.
21.270
(df=2)
.000
χ2
Sig.
33.721
(df=2)
.000
b. Product Involvement
Level
N
High
116
Moderate
112
Low
107
Correspondence Bias
Yes
No
39
72
(35.1%)
(64.9%)
17
85
(16.7%)
(83.3%)
14
108
(11.5%)
(88.5%)
c. Ad Involvement
Level
N
High
111
Moderate
116
Low
108
Correspondence Bias
Yes
No
43
68
(38.7%)
(61.3%)
18
98
(15.5%)
(84.5%)
9
99
(8.3%)
(91.7%)
123
Table 25. Three-Way ANOVA: Effects of Three Independent Variables
on Correspondence Bias (Experiment II)
• Independent Variables: Source Credibility, Product Involvement, Ad Involvement
• Covariate: Gender, Age
• Dependent Variables: Correspondence Bias
Variable
Source Credibility (SC)
Product Involvement (PI)
Ad Involvement (AI)
SC x PI
SC x AI
PI x AI
SC x PI x AI
Gender
Age
Corrected Model
SS
df
MS
F-value
Sig.
.706
1.100
2.707
.870
1.015
.197
1.905
.008
.079
10.966
2
2
2
4
4
4
8
1
1
28
.353
.550
1.353
.218
.254
.049
.238
.008
.079
.392
2.431
3.789
9.326
1.499
1.748
.339
1.641
.054
.544
2.699
.090
.024
.000
.202
.139
.852
.113
.816
.461
.000
* R2=.198 (Adjusted R2=.125)
Table 26. Independent Samples T-Tests: Effects of Correspondence Bias (CB)
on Ad/Brand Evaluations (Experiment II)
N
Mean
(max.=7)
Std. Dev.
t-value
Sig.
CB - Yes
CB - No
70
265
4.841
4.194
1.281
1.113
4.192
.000
CB - Yes
CB - No
Behavioral Intention
CB - Yes
CB - No
70
265
4.723
4.099
1.130
.922
4.258
.000
70
265
4.698
3.747
1.105
1.334
6.117
.000
Variable
Ad Attitude
Brand Attitude
124
APPENDIX B
QUESTIONNAIRES
APPENDIX B-1. QUESTIONNAIRE FOR PRETEST I
APPENDIX B-2. QUESTIONNAIRE FOR PRETEST II
APPENDIX B-3. QUESTIONNAIRE FOR PRETEST III
APPENDIX B-4. QUESTIONNAIRE FOR EXPERIMENT I
APPENDIX B-5. QUESTIONNAIRE FOR EXPERIMENT II
125
Appendix B-1. Questionnaire for Pretest I (For Product and Occupation Selections)
Section I
Instructions: The following questions are designed to estimate your general involvement with the
listed product categories. Please circle the number closest to the tendency (likelihood) that best
reflects your attitudinal or behavioral reactions to each product category after carefully reading the
italicized descriptions.
1. I am a current or frequent user of this product category. (Please mark in the blank)
Product Category
Auto Insurance
Athletic Shoes
Credit Card
Whitening Toothpaste
Sun Cream
Mobile Phone Service
MP3 Player
Weight-loss Program
Low-fat Dairy Product
Pain Reliever
Internet Access Service
Online Travel Service
Yes
No
Not Sure
2. I would be interested in reading information about this product category.
Product Category
Auto Insurance
Athletic Shoes
Credit Card
Whitening Toothpaste
Sun Cream
Mobile Phone Service
MP3 Player
Weight-loss Program
Low-fat Dairy Product
Pain Reliever
Internet Access Service
Online Travel Service
Strongly Disagree
1
1
1
1
1
1
1
1
1
1
1
1
Neutral
2
2
2
2
2
2
2
2
2
2
2
2
126
3
3
3
3
3
3
3
3
3
3
3
3
Strongly Agree
4
4
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
5
5
3. I have compared product characteristics (e.g., price, quality) among brands in this product
category.
Product Category
Auto Insurance
Athletic Shoes
Credit Card
Whitening Toothpaste
Sun Cream
Mobile Phone Service
MP3 Player
Weight-loss Program
Low-fat Dairy Product
Pain Reliever
Internet Access Service
Online Travel Service
Strongly Disagree
1
1
1
1
1
1
1
1
1
1
1
1
Neutral
2
2
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
3
3
Strongly Agree
4
4
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
5
5
4. I have a most preferred brand in this product category.
Product Category
Auto Insurance
Athletic Shoes
Credit Card
Whitening Toothpaste
Sun Cream
Mobile Phone Service
MP3 Player
Weight-loss Program
Low-fat Dairy Product
Pain Reliever
Internet Access Service
Online Travel Service
Strongly Disagree
1
1
1
1
1
1
1
1
1
1
1
1
Neutral
2
2
2
2
2
2
2
2
2
2
2
2
127
3
3
3
3
3
3
3
3
3
3
3
3
Strongly Agree
4
4
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
5
5
5. In general, this product category is relevant to me.
Product Category
Auto Insurance
Athletic Shoes
Credit Card
Whitening Toothpaste
Sun Cream
Mobile Phone Service
MP3 Player
Weight-loss Program
Low-fat Dairy Product
Pain Reliever
Internet Access Service
Online Travel Service
Strongly Disagree
1
1
1
1
1
1
1
1
1
1
1
1
Neutral
2
2
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
3
3
Strongly Agree
4
4
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
5
5
6. In general, this product category is important to me.
Product Category
Auto Insurance
Athletic Shoes
Credit Card
Whitening Toothpaste
Sun Cream
Mobile Phone Service
MP3 Player
Weight-loss Program
Low-fat Dairy Product
Pain Reliever
Internet Access Service
Online Travel Service
Strongly Disagree
1
1
1
1
1
1
1
1
1
1
1
1
Neutral
2
2
2
2
2
2
2
2
2
2
2
2
128
3
3
3
3
3
3
3
3
3
3
3
3
Strongly Agree
4
4
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
5
5
Section II
Instructions: The following questions are designed to estimate your perceived trust-worthiness
of people in the listed occupations. Trustworthiness here is specifically defined as “the
perception that a person with the occupation is unbiased when his/her experience or opinion
about some product is spoken in advertising”. Please circle the number that best represents your
trustworthiness toward each type of person.
Very Untrustworthy
The Person’s Occupation
Medical Doctor
Pop Singer
Insurance Agent
Journalist
Sports Player
Professor
Actor/Actress
Pharmacist
Salesperson
College Student
Banker
Computer Programmer
Anchorman
Lawyer
Politician
Dietitian (Nutritionist)
Dentist
Fashion Model
C.E.O.
Travel Agent
Aerobics Instructor
Engineer
Research Worker
Housewife
Businessman
Elementary School
Teacher
Your Gender: Female (
)
Neutral
Very Trustworthy
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
1
2
3
4
5
Male (
)
*** THANK YOU VERY MUCH ***
129
Appendix B-2. Questionnaire for Pretest II (For Source Credibility Manipulation)
Instructions: Suppose that an advertiser is considering which type (occupation here) of endorser
will be most credible for a new advertising campaign. In this situation, credibility is defined as
“the perception that a person with the occupation is knowledgeable and unbiased when the
person speaks about his/her experience or opinion regarding a certain product (or service) in
advertising”. However, since an endorser is supposed to tell about his/her experience or opinion
as a general user of the product (or service) in many testimonial ads, the endorser’s particular
occupation does not necessarily have to deal specifically with the product being endorsed.
After carefully looking over the occupation list on the next page, select two occupations that you
perceive to be “most credible” (i.e., best choices) and two occupations that you perceive to be
“least credible” (i.e., worst choices) if a person with those occupations were to endorse a
particular brand of the product category in testimonial advertising; then mark the numbers
corresponding to each occupation you selected in the blanks for the seven product categories
provided. If you want to select an occupation that is not given in the list, please mark “30” and
specify the occupation in parentheses. (You are not required to rank the two occupations you
selected for each item, i.e., no difference between 1 and 2 in order. See the example of simple
response for your reference.)
Occupation List
1
Medical Doctor
11
Businessman
21
Computer Programmer
2
Actor/Actress
12
Engineer
22
Aerobics Instructor
3
Pop Singer
13
Housewife
23
College Student
4
Sports Player
14
Professor
24
Dietitian (Nutritionist)
5
Insurance Agent
15
Journalist
25
Elementary School Teacher
6
Fashion Model
16
Lawyer
26
Music Critic
7
Cosmetic Dentist
17
Pharmacist
27
Broadcast Reporter
8
Salesperson
18
Composer
28
Film Director
9
Travel Agent
19
Banker
29
Public Official
10 Graphic Designer
20
Anchorman
30
Other (Please specify)
[Example] DVD Player
Most Credible
Least Credible
1: ___12____
1: ___ 8____
130
2: ___28____
2: ___13____
1. Athletic Shoes (brand examples: Nike, Reebok, Adidas)
Most Credible 1: _________
Least Credible 1: _________
2: _________
2: _________
2. Whitening Toothpaste (brand examples: Colgate, Crest, Rembrandt)
Most Credible 1: _________ 2: _________
Least Credible 1: _________ 2: _________
3. Wireless Service (brand examples: AT&T, Verizon, Cingular)
Most Credible 1: _________
Least Credible 1: _________
2: _________
2: _________
4. Weight-loss Program (meal/exercise planning and management)
Most Credible 1: _________
Least Credible 1: _________
2: _________
2: _________
5. MP3 Player (portable)
Most Credible
Least Credible
1: _________
1: _________
2: _________
2: _________
6. Online Travel Service (brand examples: Expedia, Orbitz, Priceline)
Most Credible
Least Credible
1: _________
1: _________
2: _________
2: _________
1: _________
1: _________
2: _________
2: _________
7. Low-fat Dairy Food (e.g., milk, yogurt, cheese)
Most Credible
Least Credible
*** THANK YOU VERY MUCH ***
131
Appendix B-3. Questionnaire for Pretest III (For Brand Name Selections)
Vote for your favorite brand names
Please select your favorites among the following choices of newly created brand names
for two product categories: athletic shoes and yogurt. The results from your vote will be
used to determine fictitious brand names for Kyoo-Hoon Han’s (a PhD student of Grady
College) experimental study. This pretest is important because the research is concerned
about consumer attitude toward the brand being advertised. Thank you for your help. *
PLEASE MARK IN THE BOX
Athletic Shoes
Most Favorite (pick one)
2nd Favorite (pick one)
Wing
F
F
Merica
F
F
Kaos
F
F
RunAway
F
F
Stellar
F
F
Freezon
F
F
RoadRunner
F
F
Yogurt (Main lines include low-fat and fat-free yogurts)
Most Favorite (pick one)
2nd Favorite (pick one)
Yoslim
F
F
Dyan
F
F
Haidi
F
F
Kelly’s Farm
F
F
Yodel
F
F
Fruitty
F
F
Viki Valley
F
F
*** THANK YOU VERY MUCH ***
132
Appendix B-4. Questionnaires for Experiment I
a. High Product Involvement Treatment
Section I
Please mark in the space closest to the adjective that best reflects your perception or relevance of
athletic shoes.
To me athletic shoes are
Important
____ : ____ : ____ : ____ : ____ : ____ : ____
Unimportant
Boring
____ : ____ : ____ : ____ : ____ : ____ : ____
Interesting
Relevant
____ : ____ : ____ : ____ : ____ : ____ : ____
Irrelevant
Exciting
____ : ____ : ____ : ____ : ____ : ____ : ____
Unexciting
Means nothing
____ : ____ : ____ : ____ : ____ : ____ : ____
Means a lot to me
Appealing
____ : ____ : ____ : ____ : ____ : ____ : ____
Unappealing
Fascinating
____ : ____ : ____ : ____ : ____ : ____ : ____
Mundane
Worthless
____ : ____ : ____ : ____ : ____ : ____ : ____
Valuable
Involving
____ : ____ : ____ : ____ : ____ : ____ : ____
Uninvolving
Not needed
____ : ____ : ____ : ____ : ____ : ____ : ____
Needed
How much do you agree or disagree with each of the following statements about your perception
or behavior regarding athletic shoes? Please mark in the space that best reflect the degree of
your agreement or disagreement with each statement.
Strongly
Disagree
Neutral
Strongly
Agree
1. I would be interested in reading
information about athletic shoes.
___ : ___ : ___ : ___ : ___ : ___ : ___
2. I have compared or would compare
product characteristics among brands
of athletic shoes.
___ : ___ : ___ : ___ : ___ : ___ : ___
3. I have a most preferred brand of
athletic shoes.
___ : ___ : ___ : ___ : ___ : ___ : ___
4. In general, athletic shoes is
important to me.
___ : ___ : ___ : ___ : ___ : ___ : ___
133
A stimulus advertisement
is placed in this page
(See APPENDIX C)
134
Section II
Please mark in the space closest to the adjective that best reflects your perception of the model
(Lisa Jones) who is endorsing the brand (Stellar) in the ad you’ve just seen.
Expert
____ : ____ : ____ : ____ : ____ : ____: ____
Not an expert
Trustworthy
____ : ____ : ____ : ____ : ____ : ____: ____
Untrustworthy
Experienced
____ : ____ : ____ : ____ : ____ : ____: ____
Inexperienced
Honest
____ : ____ : ____ : ____ : ____ : ____: ____
Dishonest
Knowledgeable ____ : ____ : ____ : ____ : ____ : ____: ____
Unknowledgeable
Reliable
____ : ____ : ____ : ____ : ____ : ____: ____
Unreliable
Qualified
____ : ____ : ____ : ____ : ____ : ____: ____
Unqualified
Sincere
____ : ____ : ____ : ____ : ____ : ____: ____
Insincere
Skilled
____ : ____ : ____ : ____ : ____ : ____: ____
Unskilled
Dependable
____ : ____ : ____ : ____ : ____ : ____: ____
Undependable
How much do you agree or disagree with each of the following statements about the reasons why
the model (Lisa Jones) in the ad endorsed the brand (Stellar)? Please mark in the space that best
reflect your agreement or disagreement with each statement.
Strongly
Disagree
Neutral
Strongly
Agree
Lisa Jones endorsed the brand,
1. To convey her real belief in the brand
___ : ___ : ___ : ___ : ___ : ___ : ___
2. To earn money
___ : ___ : ___ : ___ : ___ : ___ : ___
3. To present her image to the public
___ : ___ : ___ : ___ : ___ : ___ : ___
4. To talk about the brand based on her actual
experience with the brand
___ : ___ : ___ : ___ : ___ : ___ : ___
5. To receive non-monetary compensations
(e.g., free products)
___ : ___ : ___ : ___ : ___ : ___ : ___
6. To tell about brand benefits based on her
knowledge about the product
___ : ___ : ___ : ___ : ___ : ___ : ___
135
Section III
Please mark in the space closest to the adjective that best reflects your attitude toward the
(Stellar’s) advertisement you’ve just seen.
Good
____ : ____ : ____ : ____ : ____ : ____: ____
Bad
Unpleasant
____ : ____ : ____ : ____ : ____ : ____: ____
Pleasant
Favorable
____ : ____ : ____ : ____ : ____ : ____: ____
Unfavorable
Boring
____ : ____ : ____ : ____ : ____ : ____: ____
Interesting
Like
____ : ____ : ____ : ____ : ____ : ____: ____
Dislike
Uninformative
____ : ____ : ____ : ____ : ____ : ____: ____
Informative
Believable
____ : ____ : ____ : ____ : ____ : ____: ____
Unbelievable
Unconvincing
____ : ____ : ____ : ____ : ____ : ____: ____
Convincing
Credible
____ : ____ : ____ : ____ : ____ : ____: ____
Not credible
Untrustworthy
____ : ____ : ____ : ____ : ____ : ____: ____
Trustworthy
Please mark in the space closest to the adjective that best reflects your attitude toward the brand
(Stellar) being advertised.
Good
____ : ____ : ____ : ____ : ____ : ____: ____
Bad
Unpleasant
____ : ____ : ____ : ____ : ____ : ____: ____
Pleasant
Favorable
____ : ____ : ____ : ____ : ____ : ____: ____
Unfavorable
Boring
____ : ____ : ____ : ____ : ____ : ____: ____
Interesting
Like
____ : ____ : ____ : ____ : ____ : ____: ____
Dislike
Unconvincing
____ : ____ : ____ : ____ : ____ : ____: ____
Convincing
Credible
____ : ____ : ____ : ____ : ____ : ____: ____
Not credible
Low quality
____ : ____ : ____ : ____ : ____ : ____: ____
High quality
Useful
____ : ____ : ____ : ____ : ____ : ____: ____
Useless
Worthless
____ : ____ : ____ : ____ : ____ : ____: ____
Valuable
136
How much do you agree or disagree with each of the following statements about your intention
or expectation regarding the use of the brand (Stellar athletic shoes)? Please mark in the space
that best reflect your agreement or disagreement with each statement.
Strongly
Disagree
Neutral
Strongly
Agree
1. The next time I purchase an athletic
shoes, I would buy Stellar.
___ : ___ : ___ : ___ : ___ : ___ : ___
2. I would consider using Stellar.
___ : ___ : ___ : ___ : ___ : ___ : ___
3. If a free trial (e.g., 30 days money-back
guarantee) is offered, I would try Stellar.
___ : ___ : ___ : ___ : ___ : ___ : ___
4. If a discount coupon is offered, I would
buy Stellar.
___ : ___ : ___ : ___ : ___ : ___ : ___
5. I would search for more information
about Stellar. (e.g., visit Web site(s),
ask to current users)
___ : ___ : ___ : ___ : ___ : ___ : ___
6. I would compare Stellar with other
athletic shoes brands to make a purchase
decision.
___ : ___ : ___ : ___ : ___ : ___ : ___
Section IV
Please mark in the space that best describe your personal information. The information you
provide below will not be associated with your identity in any way.
1. What is your gender?
Male ____
2. What is your age?
______ years
Female ____
3. What is your year in college?
Freshman ____
Sophomore ____
Junior ____
Senior ____
Graduate student ____
Other (please specify) _______________
4. What is your ethnic origin?
Caucasian ____
African-American ____
Asian ____
Latino (Hispanic) ____
Other (please specify) _______________
*** THANK YOU VERY MUCH ***
137
b. Low Product Involvement Treatment
Section I
Please mark in the space closest to the adjective that best reflects your perception or relevance of
low-fat yogurt (or fat-free yogurt hereafter).
To me low-fat yogurt is
Important
____ : ____ : ____ : ____ : ____ : ____ : ____
Unimportant
Boring
____ : ____ : ____ : ____ : ____ : ____ : ____
Interesting
Relevant
____ : ____ : ____ : ____ : ____ : ____ : ____
Irrelevant
Exciting
____ : ____ : ____ : ____ : ____ : ____ : ____
Unexciting
Means nothing
____ : ____ : ____ : ____ : ____ : ____ : ____
Means a lot to me
Appealing
____ : ____ : ____ : ____ : ____ : ____ : ____
Unappealing
Fascinating
____ : ____ : ____ : ____ : ____ : ____ : ____
Mundane
Worthless
____ : ____ : ____ : ____ : ____ : ____ : ____
Valuable
Involving
____ : ____ : ____ : ____ : ____ : ____ : ____
Uninvolving
Not needed
____ : ____ : ____ : ____ : ____ : ____ : ____
Needed
How much do you agree or disagree with each of the following statements about your perception
or behavior regarding low-fat yogurt? Please mark in the space that best reflect the degree of
your agreement or disagreement with each statement.
Strongly
Disagree
Neutral
Strongly
Agree
1. I would be interested in reading
information about low-fat yogurt.
____ : ____ : ____ : ____ : ____ : ____ : ____
2. I have compared or would compare
product characteristics among brands
of low-fat yogurt.
____ : ____ : ____ : ____ : ____ : ____ : ____
3. I have a most preferred brand of
low-fat yogurt.
____ : ____ : ____ : ____ : ____ : ____ : ____
4. In general, low-fat yogurt is important ____ : ____ : ____ : ____ : ____ : ____ : ____
to me.
138
A stimulus advertisement
is placed in this page
(See APPENDIX C)
139
Section II
Please mark in the space closest to the adjective that best reflects your perception of the model
(Lisa Jones) who is endorsing the brand (YoSlim) in the ad you’ve just seen.
Expert
____ : ____ : ____ : ____ : ____ : ____: ____
Not an expert
Trustworthy
____ : ____ : ____ : ____ : ____ : ____: ____
Untrustworthy
Experienced
____ : ____ : ____ : ____ : ____ : ____: ____
Inexperienced
Honest
____ : ____ : ____ : ____ : ____ : ____: ____
Dishonest
Knowledgeable ____ : ____ : ____ : ____ : ____ : ____: ____
Unknowledgeable
Reliable
____ : ____ : ____ : ____ : ____ : ____: ____
Unreliable
Qualified
____ : ____ : ____ : ____ : ____ : ____: ____
Unqualified
Sincere
____ : ____ : ____ : ____ : ____ : ____: ____
Insincere
Skilled
____ : ____ : ____ : ____ : ____ : ____: ____
Unskilled
Dependable
____ : ____ : ____ : ____ : ____ : ____: ____
Undependable
How much do you agree or disagree with each of the following statements about the reasons why
the model (Lisa Jones) in the ad endorsed the brand (YoSlim)? Please mark in the space that best
reflect your agreement or disagreement with each statement.
Strongly
Disagree
Neutral
Strongly
Agree
Lisa Jones endorsed the brand,
7. To convey her real belief in the brand
___ : ___ : ___ : ___ : ___ : ___ : ___
8. To earn money
___ : ___ : ___ : ___ : ___ : ___ : ___
9. To present her image to the public
___ : ___ : ___ : ___ : ___ : ___ : ___
10. To talk about the brand based on her actual
experience with the brand
___ : ___ : ___ : ___ : ___ : ___ : ___
11. To receive non-monetary compensations
(e.g., free products)
___ : ___ : ___ : ___ : ___ : ___ : ___
12. To tell about brand benefits based on her
knowledge about the product
___ : ___ : ___ : ___ : ___ : ___ : ___
140
Section III
Please mark in the space closest to the adjective that best reflects your attitude toward the
(YoSlim’s) advertisement you’ve just seen.
Good
____ : ____ : ____ : ____ : ____ : ____: ____
Bad
Unpleasant
____ : ____ : ____ : ____ : ____ : ____: ____
Pleasant
Favorable
____ : ____ : ____ : ____ : ____ : ____: ____
Unfavorable
Boring
____ : ____ : ____ : ____ : ____ : ____: ____
Interesting
Like
____ : ____ : ____ : ____ : ____ : ____: ____
Dislike
Uninformative
____ : ____ : ____ : ____ : ____ : ____: ____
Informative
Believable
____ : ____ : ____ : ____ : ____ : ____: ____
Unbelievable
Unconvincing
____ : ____ : ____ : ____ : ____ : ____: ____
Convincing
Credible
____ : ____ : ____ : ____ : ____ : ____: ____
Not credible
Untrustworthy
____ : ____ : ____ : ____ : ____ : ____: ____
Trustworthy
Please mark in the space closest to the adjective that best reflects your attitude toward the brand
(YoSlim) being advertised.
Good
____ : ____ : ____ : ____ : ____ : ____: ____
Bad
Unpleasant
____ : ____ : ____ : ____ : ____ : ____: ____
Pleasant
Favorable
____ : ____ : ____ : ____ : ____ : ____: ____
Unfavorable
Boring
____ : ____ : ____ : ____ : ____ : ____: ____
Interesting
Like
____ : ____ : ____ : ____ : ____ : ____: ____
Dislike
Unconvincing
____ : ____ : ____ : ____ : ____ : ____: ____
Convincing
Credible
____ : ____ : ____ : ____ : ____ : ____: ____
Not credible
Low quality
____ : ____ : ____ : ____ : ____ : ____: ____
High quality
Useful
____ : ____ : ____ : ____ : ____ : ____: ____
Useless
Worthless
____ : ____ : ____ : ____ : ____ : ____: ____
Valuable
141
How much do you agree or disagree with each of the following statements about your intention
or expectation regarding the use of YoSlim low-fat yogurt? Please mark in the space that best
reflect your agreement or disagreement with each statement.
Strongly
Disagree
Neutral
Strongly
Agree
1. The next time I purchase a low-fat yogurt,
I would buy YoSlim.
___ : ___ : ___ : ___ : ___ : ___ : ___
2. I would consider trying YoSlim low-fat
yogurt.
___ : ___ : ___ : ___ : ___ : ___ : ___
3. If a free trial (e.g., 30 days money-back
guarantee) is offered, I would try a YoSlim
low-fat yogurt.
___ : ___ : ___ : ___ : ___ : ___ : ___
4. If a discount coupon is offered, I would
buy YoSlim low-fat yogurt.
___ : ___ : ___ : ___ : ___ : ___ : ___
5. I would search for more information
about YoSlim low-fat yogurt. (e.g., visit
Web site(s), ask to current users)
___ : ___ : ___ : ___ : ___ : ___ : ___
6. I would compare YoSlim with other
low-fat yogurt brands to make a purchase
decision.
___ : ___ : ___ : ___ : ___ : ___ : ___
Section IV
Please mark in the space that best describe your personal information. The information you
provide below will not be associated with your identity in any way.
1. What is your gender?
Male ____
2. What is your age?
______ years
Female ____
3. What is your year in college?
Freshman ____
Sophomore ____
Junior ____
Senior ____
Graduate student ____
Other (please specify) _______________
4. What is your ethnic origin?
Caucasian ____
African-American ____
Asian ____
Latino (Hispanic) ____
Other (please specify) _______________
*** THANK YOU VERY MUCH ***
142
Appendix B-5. Questionnaire for Experiment II
General Instructions
This study attempts to examine how consumers respond to an advertisement (hereafter "ad") of
Stellar, a new athletic shoe brand tentatively named for this research. Over the next pages, you
will see a magazine-style testimonial ad for Stellar athletic shoes, followed by questions divided
into several sections. Before you look at the ad and fill out the questionnaire, please carefully
read the instructions in each section and follow them when answering. Thank you for your
participation in this study.
Please turn to the next page now and begin…
143
Before you see the ad in this page…
Below is a magazine-style ad for Stellar athletic shoes. Please look at the ad for 20-30 seconds,
or longer if you wish. When you see the ad, please assume that you find this ad while you are
reading a magazine.
A stimulus advertisement
is placed in this page
(See APPENDIX C)
If you've finished looking at the ad, now you can turn to the next page.
144
Section I
Please click on the space that best reflects your feelings about the Stellar ad you just saw.
1. To me the ad is …
Good (in general)
F
F
F
F
F
F
F
Bad (in general)
Unpleasant
F
F
F
F
F
F
F
Pleasant
Favorable
F
F
F
F
F
F
F
Unfavorable
Boring
F
F
F
F
F
F
F
Interesting
Likable
F
F
F
F
F
F
F
Dislikable
Useless
F
F
F
F
F
F
F
Useful
Believable
F
F
F
F
F
F
F
Unbelievable
Unconvincing
F
F
F
F
F
F
F
Convincing
Credible
F
F
F
F
F
F
F
Not credible
Untrustworthy
F
F
F
F
F
F
F
Trustworthy
Please click on the space that best reflects your attitude toward the brand (Stellar) being
advertised.
2. To me the brand (Stellar) is …
Good (in general)
F
F
F
F
F
F
F
Bad (in general)
Unpleasant
F
F
F
F
F
F
F
Pleasant
Favorable
F
F
F
F
F
F
F
Unfavorable
Boring
F
F
F
F
F
F
F
Interesting
Likable
F
F
F
F
F
F
F
Dislikable
Unconvincing
F
F
F
F
F
F
F
Convincing
Credible
F
F
F
F
F
F
F
Not credible
Low quality
F
F
F
F
F
F
F
High quality
Useful
F
F
F
F
F
F
F
Useless
Worthless
F
F
F
F
F
F
F
Valuable
145
For questions 4-9, please click on the space that best reflects the degree of your agreement or
disagreement with each statement regarding the use of Stellar athletic shoes.
Strongly
Disagree
Strongly
Agree
Neutral
4. The next time I purchase athletic shoes,
I will buy Stellar.
F
F
F
F
F
F
F
5. I will consider using Stellar.
F
F
F
F
F
F
F
6. If a free trial (e.g., 30 days money-back
guarantee) is offered, I will try Stellar.
F
F
F
F
F
F
F
7. If a special sale is offered, I will buy
Stellar.
F
F
F
F
F
F
F
8. I will search for more information about
Stellar. (e.g., visiting Web site(s), asking
to current users)
F
F
F
F
F
F
F
9. I will recommend Stellar to others.
F
F
F
F
F
F
F
Section II
For questions 10-15, please click on the space that best reflects your agreement or disagreement
with each statement about why Lisa Jones endorsed the brand (Stellar) in the ad.
Strongly
Disagree
Strongly
Agree
Neutral
Lisa Jones endorsed the brand …
10. To convey her real belief in the brand
F
F
F
F
F
F
F
11. To earn money
F
F
F
F
F
F
F
12. To become better known
F
F
F
F
F
F
F
13. To express her feeling about the brand
based on her actual experience
F
F
F
F
F
F
F
14. To receive non-monetary compensation
(e.g., free products)
F
F
F
F
F
F
F
15. To speak about brand benefits based on
her knowledge about the product
F
F
F
F
F
F
F
146
For questions 16-22, please click on the space that best reflects your opinion about Lisa Jones'
product endorsement and her use or feeling about the brand.
Strongly
Disagree
Strongly
Agree
Neutral
16. Lisa Jones received money for her
endorsement.
F
F
F
F
F
F
F
17. Lisa Jones received “big” money for
her endorsement.
F
F
F
F
F
F
F
18. Whether or not she was paid, Lisa Jones F
received something other than money
(e.g., free products) for her endorsement.
F
F
F
F
F
F
19. Lisa Jones took the opportunity to get
publicity by her endorsement.
F
F
F
F
F
F
F
20. Lisa Jones frequently uses the brand.
F
F
F
F
F
F
F
21. Lisa Jones likes the brand.
F
F
F
F
F
F
F
22. Lisa Jones regards the brand as good.
F
F
F
F
F
F
F
Section III
Please click on the space that best reflects your evaluation of the endorser (Lisa Jones) in the ad,
considering her image, occupation, and the endorsed product (athletic shoes).
23. The endorser (Lisa Jones) in the ad is …
Expert*
F
F
F
F
F
F
F
Not an expert
Trustworthy
F
F
F
F
F
F
F
Untrustworthy
Experienced* F
F
F
F
F
F
F
Inexperienced
Honest
F
F
F
F
F
F
F
Dishonest
Knowledgeable
F
F
F
F
F
F
F
Unknowledgeable
Reliable
F
F
F
F
F
F
F
Unreliable
Qualified
F
F
F
F
F
F
F
Unqualified
Sincere
F
F
F
F
F
F
F
Insincere
Skilled
F
F
F
F
F
F
F
Unskilled
Dependable
F
F
F
F
F
F
F
Undependable
Section
* about IV
or with athletic shoes
147
24. Please click on the space that best reflects your level of attention to each of the following
parts of the Stellar ad when you looked at the ad.
Not At All
1
2
Moderately
4
5
3
Very Much
7
6
The endorser’s visual image
F
F
F
F
F
F
F
The endorser’s name
F
F
F
F
F
F
F
The endorser’s occupation
F
F
F
F
F
F
F
Headline copy
F
F
(testimonial message in big-size letters)
F
F
F
F
F
Body copy
F
(product message in small-size letters)
F
F
F
F
F
F
Product picture
F
F
F
F
F
F
F
Brand slogan
F
(“Feel your best. Be your best.”)
F
F
F
F
F
F
Brand name and symbol
F
F
F
F
F
F
F
25. Overall, how much did you pay attention to the Stellar ad?
Not At All
F
Moderately
F
F
F
Very Much
F
F
F
For questions 26-28, please click on the space that best reflects your reactions when you were
looking at the Stellar ad.
Strongly
Disagree
26. When I saw the Stellar ad, I felt the
information in it was important to me.
Strongly
Agree
Neutral
F
F
F
F
F
F
F
27. When I saw the Stellar ad, I felt the
F
information in it was relevant to my needs.
F
F
F
F
F
F
28. I paid attention to the Stellar ad as if I
were considering buying athletic shoes.
F
F
F
F
F
F
F
148
Section V
Please click on the number that best reflects your opinion about athletic shoes.
29. To me athletic shoes are
Important
F
F
F
F
F
F
F
Unimportant
Boring
F
F
F
F
F
F
F
Interesting
Relevant
F
F
F
F
F
F
F
Irrelevant
Exciting
F
F
F
F
F
F
F
Unexciting
Meaningless
F
F
F
F
F
F
F
Meaningful
Appealing
F
F
F
F
F
F
F
Unappealing
Fascinating
F
F
F
F
F
F
F
Mundane
Worthless
F
F
F
F
F
F
F
Valuable
Involving
F
F
F
F
F
F
F
Uninvolving
Not needed
F
F
F
F
F
F
F
Needed
Section VI
Finally, please provide your personal information (for statistical purposes only).
30. Gender
F Male
F Female
31. Age
F
F
F
F
F
F
F
Under 18
18-24
25-34
35-44
45-54
55-64
Over 65
32. Ethnic origin
F
F
F
F
F
Caucasian
African-American
Asian
Latino
Other
149
33. Occupation
F
F
F
F
F
F
F
F
F
College student
Homemaker
Skilled worker
Sales
Professional
Management
Self-employed
Unemployed
Other
*** THANK YOU VERY MUCH ***
150
APPENDIX C
STIMULUS ADVERTISEMENTS
APPENDIX C-1. STIMULUS ADVERTISEMENTS FOR EXPERIMENT I
APPENDIX C-2. STIMULUS ADVERTISEMENTS FOR EXPERIMENT II
151
Appendix C-1. Stimulus Advertisements for Experiment I
#1. High Source Credibility / High Product Involvement
“Stellar shoes help me
feel great every day.”
Lisa Jones
Aerobics instructor
I mprove your health, your looks, and how you feel with Stellar
running shoes. They’ve been designed using the latest technical
know-how, so you know they work. But what really matters is
the proof you see in the mirror - and that you feel - every day.
Feel your best. Be your best.
Ê Stellar
152

#2. Low Source Credibility / High Product Involvement
“Stellar shoes help me
feel great every day.”
Lisa Jones
Sales Representative
I mprove your health, your looks, and how you feel with Stellar
running shoes. They’ve been designed using the latest technical
know-how, so you know they work. But what really matters is
the proof you see in the mirror - and that you feel - every day.
Feel your best. Be your best.
Ê Stellar
153

#3. High Source Credibility / Low Product Involvement
“YoSlim helps me feel
great every day.”
Lisa Jones
Dietician
I mprove your health, your looks, and how you feel with YoSlim.
It’s been formulated using the latest scientific know-how,
so you know it works. But what really matters is the proof you see
in the mirror - and that you feel - every day.
LOWFAT YOGURT
Feel your best. Be your best.
154
YoSlim 
YoSlim
#4. Low Source Credibility / Low Product Involvement
“YoSlim helps me feel
great every day.”
Lisa Jones
Sales Representative
I mprove your health, your looks, and how you feel with YoSlim.
It’s been formulated using the latest scientific know-how,
so you know it works. But what really matters is the proof you see
LOWFAT YOGURT
in the mirror - and that you feel - every day.
Feel your best. Be your best.
155
YoSlim 
YoSlim
Appendix C-2. Stimulus Advertisements for Experiment II
#1. High Source Credibility
“Incredible fit, comfortable.
I feel great when I’m in Stellar.”
Lisa Jones
Aerobics instructor
Stellar ’s state-of-the-art technology provides an ideal balance of
light weight, ultra-cushioning, and maximum stability.
Comfort and fit, a scientific way, for men - and for women.
Begin now to improve your health, your looks, and how you feel.
Feel your best. Be your best.
Ê Stellar
156

#2. Low Source Credibility
“Incredible fit, comfortable.
I feel great when I’m in Stellar.”
Lisa Jones
Sales Representative
Stellar ’s state-of-the-art technology provides an ideal balance of
light weight, ultra-cushioning, and maximum stability.
Comfort and fit, a scientific way, for men - and for women.
Begin now to improve your health, your looks, and how you feel.
Feel your best. Be your best.
Ê Stellar 
157
#3. Control of Source Credibility
“Incredible fit, comfortable.
I feel great when I’m in Stellar.”
Lisa Jones
Stellar ’s state-of-the-art technology provides an ideal balance of
light weight, ultra-cushioning, and maximum stability.
Comfort and fit, a scientific way, for men - and for women.
Begin now to improve your health, your looks, and how you feel.
Feel your best. Be your best.
Ê Stellar 
158