Vous êtes sur la page 1sur 15

Construction and Validation of a Scale to Measure Celebrity Endorsers' Perceived Expertise,

Trustworthiness, and Attractiveness


Author(s): Roobina Ohanian
Source: Journal of Advertising, Vol. 19, No. 3 (1990), pp. 39-52
Published by: M.E. Sharpe, Inc.
Stable URL: http://www.jstor.org/stable/4188769
Accessed: 04/10/2010 06:35

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=mes.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

M.E. Sharpe, Inc. is collaborating with JSTOR to digitize, preserve and extend access to Journal of
Advertising.

http://www.jstor.org
Constructionand
Validationof a Scaleto Measure
CelebrityEndorsers'PerceivedExpertise,
Trustworthiness, and Attractiveness

RoobinaOhanian The purposeof this study was to developa scale for measuring celebrityendorsers'perceived
expertise, trustworthiness,and attractiveness.Accepted psychometricscale-developmentpro-
cedures were followed which rigorouslytested a large pool of items for their reliability and
validity. Using two exploratoryand two confirmatorysamples, the current researchdeveloped
a 15-item semantic differentialscale to measure perceivedexpertise,trustworthiness,and at-
tractiveness.The scale was validated using respondents'self-reportedmeasuresof intention to
purchase and perceptionof quality for the products being tested. The resultingscale demon-
strated high reliabilityand validity.

Roobina Ohanian (Ph.D., University of Texas Marketing and advertising practitioners share the belief that a communicator's
at Austin) is associate professor of marketing, character has a significant effect on the persuasiveness of the message. In testi-
School of Business, Emory University.
monial advertising,consumers traditionallyhave been chosen as product endorsers
The author wishes to thank Janet Cox, Armen because of their similarity to target audiences. Although this practice continues,
Tashchian, and three anonymous reviewers for a more noticeable trend appears to be endorsements by actors/actresses, athletes,
their helpful comments on previous drafts of and other celebrities and well-known athletes, who are closely associated with
this article. Funding for this research was both the product and the target audience (BusinessWeek 1987; Miller 1989; Mor-
provided by a research grant from the School
of Business, Emory University.
rison 1980; Slinker 1984).
The selection of an appropriate spokesperson for a product or a service is an
important, yet difficult, decision. Is an effective and credible spokesperson some-
one who is attractive,trustworthy,or an expert, or even a combination of all three
traits? Is a credible spokesperson an individual who is dynamic, qualified, au-
thoritative, sociable, or safe?Since Aristotle's time (or before), politicians, orators,
and public speakers have attempted to identify the determinant qualities of ef-
fective speakers (Giffin 1967).
A number of empirical investigations have examined the effectiveness of using
credible spokespersons to enhance the persuasiveness of messages. Studies have
measured the process by which a communicator'sperceived attractiveness, trust-
worthiness, and expertise mediate immediate and delayed attitude change and
persuasion (Anderson and Clevenger 1963;Baker and Churchill, Jr.1977; Hovland
and Weiss 1951;Johnson, Torcivia,and Poprick 1968; Kelman and Hovland 1953;
Patzer 1983; Simon, Berkowitz, and Moyer 1970; Whittaker and Meade 1968).
Several researchers in the field of speech communication have utilized factor
analytic techniques to uncover the perceptual structure of source credibility (Ap-
plbaum and Anatol 1972; Berlo, Lemert, and Mertz 1969; Bowers and Phillips
1967; McCroskey 1966; Whitehead 1968). Their attempts have resulted in the
development of scales, each of which includes a different set of dimensions for
the measurement of source credibility. For example, Berlo, Lemert, and Mertz
(1969) define source credibility as encompassing the dimensions of safety, quali-
fication, and dynamism. On the other hand, McCroskey (1966) identifies au-
thoritativeness and character as other dimensions of source credibility; while

?journal of Advertising
Volume 19, Number 3, 1990, Page 39-52
Whitehead (1968) identifies objectivity Table 1 presents a summary of major and validity of the resulting scales. As
as another dimension of source credi- researchstudiesthat have addressedthe should be apparent, most attempts to
bility.In the process of developingscales scaling of source credibility. assess the impact of source credibility
to measure the effectiveness of celeb- Although all the studies were de- have been based on instruments of un-
rity endorsers, a number of researchers signed to measure the same construct, known reliability.This fact partiallyex-
in advertising and marketing have ex- there is no consistency among the au- plains the inconsistencies in the
panded the number of dimensions en- thors as to the number and types of literatureregardingthe impact of com-
compassing the source-credibility dimensions that source credibility municatorcredibility as it relates to at-
construct (DeSarbo and Harshman comprises. Furthermore, with the ex- titude formation and attitude change.
1985; Simpson and Kahler 1980-81; ception of McCroskey (1966), none of Given the accumulative nature of re-
Wilding and Bauer 1968; Wynn 1987). the authors have assessedthe reliability search, and the fact that researchers

TABLE 1
Source Credibility Scales

Dimensions Number of Reliability Validity Scale Method of


Author(s) Measured Items Checks Checks Type Analysis
Applbaum and Anatol Trustworthiness 13 No No SD
(1972) Expertness 10
Dynamism 5
Objectivity 3
Berlo, Lemert, and Safety 5 No No SD Factor
Mertz (1969) Qualification 5 Analysis
Dynamism 5
Bowers and Phillips Trustworthiness 7 SD Factor
(1967) Competence 5 Analysis
DeSarbo and Expertness 4 No No SD
Harshman (1985) Attractiveness 2
Trustworthiness 1
Likability 2

Additional Dimensions
Evaluative
Potency
Activity
McCroskey (1966) Authoritativeness 6 Yes Yes SD Factor
Character 6 Analysis
Authoritativeness 23 Yes Yes LIK Factor
Character 20 Analysis
Simpson and Kahler Believability 8 No Limited SD Factor
(1980-81) Dynamism 6 Analysis
Expertness 7
Sociability 3
Whitehead (1968) Trustworthiness 18 No No SD Factor
Competence 4 Analysis
Dynamism 3
Objectivity 3
Wynn (1987) Expertness 12 No No SD Factor
Dynamism 6 Analysis
Believability 3
Sociability 3

SD = Semantic Differential Scale, LIK = Likert Scale

40
base and build the findings of their (1953) defined expertise as "the extent opment of the scale but were employed
studies on that of others, there must to which a communicator is perceived as measures for nomological validity.
be a consistent measurement approach to be a source of valid assertions,"and Following is a discussion of the three
for source credibility. This measure- trustworthiness as "the degree of con- dimensions of expertise, trustworthi-
ment approachfirstmust provide a the- fidence in the communicator'sintent to ness, and physical attractivenessas the
oretical basis for the selection of communicate the assertions he consid- hypothesized dimensions of celebrity
constructs to represent the hypothe- ers most valid." endorsers' credibility.
sized dimensions of source credibility,
and second, must produce a valid, re- Trustworthiness. The trust paradigm
liable measurement scale. In view of in communication is the listener's de-
the widespread theoretical and empir- Understandingand gree of confidence in, and level of ac-
ical interest in the concept of source
credibility,the purpose of the present definingsource ceptance of, the speaker and the
message. Giffin (1967) reviewed the
research is to advance and then to as- crediblity .. is often concept of trust, in a tour of the cen-
sess a tri-component construct using
psychometrically accepted procedures confusingbecause of turies from Aristotle to King, and con-
cluded that what Aristotle called
to produce a reliable and valid scale. the many "ethos,"and what Hovland, Janis, and

Definitions of
operationalizations Kelley (1953)called "sourcecredibility,"
are the same concept: a listener's trust
Source Credibility that appearin the in a speaker. Furthermore, such terms

"Source credibility" is a term com-


literature. as "favorable disposition," "accep-
tance," "psychological safety," and
monly used to imply a communicator's "perceived supportive climate" are
positive characteristics that affect the often mentioned as favorable conse-
receiver'sacceptance of a message. Un- quences of trust (Giffin 1967).
derstanding and defining source cred- The source-attractivenessmodel has Numerous studies support the effect
ibility in the advertising and speech its origins in the social psychological of trustworthiness on attitude change.
communication context is often con- research and is a component of the For example, in the context of fear-
fusing because of the many different "source valence" model of McGuire arousing communications, Miller and
operationalizations that appear in the (McGuire 1985). The attractiveness Baseheart (1969) investigated the im-
literature. For example, in experimen- model contends that the effectiveness pact of source trustworthiness on the
tal studies, source credibility is often of a message depends on source's "fa- persuasibility of the communication.
considered a categorical variable, such miliarity,""likability,""similarity,"and The results indicated that when the
that individualsare presented as having "attractiveness"to the respondent. communicator was perceived to be
high or low credibility (e.g., Anderson For the present research, both the highly trustworthy, an opinionated
and Clevenger 1963; Griffitt 1966; source-credibilitymodel of Hovland, et messagewas more effective than a non-
Maddux and Rogers 1980). Other ap- al. (1953), and the attractivenessmodel opinionated communication in pro-
proaches commonly used to describe of McGuire (1958) were used in defin- ducing attitudechange. However, when
this phenomenon include the use of ing the dimensions of source valence. trustworthiness was low, this relation-
such labels as: ethos, prestige, reputa- Expertise and trustworthness as sug- ship was not significant.McGinnies and
tion, status, authority,competence, etc. gested by Hovland, et al. (1953) (also as Ward (1980) manipulateda source's ex-
(e.g., Applbaum and Anatol 1972; Gif- the credibility dimension of the pertise and trustworthiness to assess
fin 1967; McCroskey 1966). McGuire source-valence model), and the impact of each of these compo-
Research and reflection on the topic attractiveness were used as hypothe- nents on the communicator's per-
of celebrity endorsement rest on two sized dimensions of source attributes. suasiveness. Their findings indicated
general models: the source-credibility The decision to use attractivenesswas that a source who was perceived to be
model and the source-attractiveness further motivated by the fact that at- both an expert and trustworthy gen-
model. The source-credibility model tractiveness has become an important eratedthe most opinion change. In fact,
resulted from a landmark study by factor through the increasing use of ce- the trustworthy communicator was
Hovland and his associates(1953).They lebrities as endorsers for products, ser- persuasive, whether an expert or not.
analyzedthe factors leading to the per- vices and/or social causes (Baker and Further, Friedman and Friedman
ceived credibility of the communicator Churchill 1977; Caballero, Lumpkin, (1976), and Friedman, Santeramo, and
and concluded that two factors and Madden 1989; Caballero and Sol- Traina (1979) investigated several cor-
namely, expertness and trustworthi- omon 1984; DeSarbo and Harshman relatesof trustworthiness and conclud-
ness-underscore the concept of source 1985;Patzer1983).Source likabilityand ed that celebrities who are liked will
credibility. Hovland, Janis, and Kelley similarity were not used in the devel- also be trusted. In addition, celebrity

41
trustworthiness was highly correlated pert salesperson (Woodside and researchhas shown that highly credible
with a respondent's perceived similar- Davenport, Jr. 1974). sources induce more behavioralcompli-
ity to the source, the level of source's ance than do less-crediblesources (Ross
expertise, and the source's attractive- Attractiveness. A considerable body 1973;Woodsideand Davenport,Jr.1974,
ness. of research in advertising and com- 1976). However, it is important to rec-
In summary,trustworthiness of the munication suggests that physical at- ognize that highly credible sources are
communicator (celebrity) is an impor- tractiveness is an important cue in an not alwaysmore effectivethan less-cred-
tant construct in persuasion and atti- individual'sinitial judgment of another ible ones. In particular,when the audi-
tude-change research. Therefore, a person (Baker and Churchill 1977; ence is alreadyfavorablypredisposedto
reliable measurement of this construct Chaiken 1979;Joseph 1982; Kahle and the message, a less-crediblesource can
requires a series of items, rather than Homer 1985; Mills and Aronson 1965; induce greater persuasion than can a
the typical single item commonly used Widgery and Ruch 1981). Despite the highly credible source (Sternthal,Dho-
to measure the variable as a trustwor- vast quantity of literature addressing lakia, and Leavitt 1978). Furthermore,
thy-untrustworthy dichotomy. physical attractiveness, the issue is far researchdealingwith the interactionof
from clear. A review of the area indi- sourceand audiencecharacteristics(such
Expertise. Expertise is the second di- cates that the construct of attractive- as level of authoritarianismor issue in-
mension of source credibilityas defined ness is not uni-dimensional and that volvement) do not always report the
by Hovland, Janis, and Kelley (1953). there are myriaddefinitions used to op- greatereffectivenesssof higher-credibil-
This dimension is also referred to as erationalize attractiveness. For exam- ity sources (Johnson and Izzett 1972;
"authoritativeness"(McCroskey 1966), ple, the construct has been definedboth Johnson, Torcivia,and Poprick 1968).
"competence" (Whitehead 1968), "ex- in terms of facial and physical attrac-
pertness"(Applbaumand Anatol 1972), tiveness (Baker and Churchill 1977;
or "qualification"(Berlo, Lemert, and Caballero and Solomon 1984; Patzer
Mertz 1969). Adjectives such as 1983), with physical attractiveness op- it is important to
"trained-untrained,' "informed-unin- erationalizedin terms of model attrac-
formed," and "educated-uneducated" tiveness (attractive-unattractive)(Baker recognize that highly
commonly have been used to measure and Churchill 1977; Kahle and Homer credible sources are
this dimension. 1985), chicness (Mills and Aronson
Research investigatingsource exper- 1965), sexiness (Steadman 1969), or not always more
tise in persuasive communication gen- sexualness and likability (Maddux and effective than less,
erally indicates that the source's Rogers 1980).
perceived expertise has a positive im- In an exhaustivereview,Joseph(1982) credible ones.
pact on attitude change (Horai, Nac- summarizedthe experimental evidence
cari, and Fatoullah 1974; Maddux and in advertisingand relateddisciplinesre-
Rogers 1980; Mills and Harvey 1972; garding physicallyattractive communi- With the increased use of celebrities
Ross 1973). For example, Crisci and cators' impact on opinion change, in advertising,a valid instrument meas-
Kassinove(1973) investigatedthe effect product evaluation, and other depend- uring a celebrity endorser's credibility
of the perceived level of communicator ent measures.He concluded that attrac- is essential for understanding the im-
expertise ("Dr."versus "Mr.")and the tive (versusunattractive)communicators pact of using such individuals in ad-
strength of advice (positive versus neu- are consistently liked more and have a vertising. As in other forms of
tral)on behavioralcompliance. The re- positive impacton productswith which persuasivecommunication,-advertisers'
sults of this study indicated that they are associated. Except for a few primarygoals are to persuade their au-
respondents' compliance with the studies (Mills and Aronson 1965; Mad- dience and to induce an attitudechange
source'srecommendationsdirectly var- dux and Rogers 1980),Joseph'sfindings toward their offerings (Walley 1987).
ied with the perceived level of exper- are consistent with others that report The following discussion presents
tise and the strengthof advice.Similarly, that increasing the communicator'sat- the steps for the development of a tri-
Crano (1970) experimentally manipu- tractiveness enhances positive attitude component celebrity endorser's credi-
lated the dimensions of expertise and change (Simon, Berkowitz, and Moyer bility scale. It includes item generation
found that subjects exposed to an ex- 1970;Kahle and Homer 1985). and reduction, exploratory and confir-
pert source exhibited more agreement matory studies, and reliability and va-
with the advocated position than did Summary lidity of the final subscales.
those exposed to a low-expertisesource.
Finally, in a selling context, an expert The review of source-credibility liter- Research Methodology
salesperson induced a significantly ature provides evidence that credible
higher number of customers to pur- sources are more persuasive than are Development of Items for the Source-
chase a product than did the nonex- sources of low credibility Additionally, Credibility Scale In the initial phase

42
of this research, the literature in the based on the frequency of mention, ca- Using the above results as a guide,
areas of psychology, mass communica- tegorizedby gender, and classifiedas to it was decided to include frequently
tion, and advertising was reviewed to whether they had ever participated in purchased products used by a wide
identify words, phrases, or adjectives a paid commercial.The most frequent- cross-section of the population. Thus,
used in measuring the traits associated ly mentioned names for each gender, the final product/celebrity list includ-
with credible sources. In addition, a among those who had previously en- ed Linda Evans promoting a new per-
large pool of adjectives describing per- dorsed a product, were John McEnroe fume; Madonna, a new line of designer
sonality traits was developed by con- and LindaEvans.Celebritiesmost often jeans; John McEnroe, a line of tennis
sulting previously available sources mentioned, who had not been involved rackets; and Tom Selleck, a new line
(Allport and Odbert 1936; Anderson in advertisements, were Tom Selleck of men's cologne.
1968; Osgood, Suci, and Tannenbaum and Madonna. The next phase of the study was to
1957). An effort was made to extract purify and validate the celebrity en-
all entries that were likely to be useful dorser's credibility scale. This task was
in the development of a celebrity en-
dorser's credibility scale. This process
The most frequently accomplished in two stages. In the ex-
ploratory phase, a convenient student
resulted in 182 adjectivesor descriptor mentioned names for samplewas employed to further reduce
words. The 182 adjectives were
screened and reduced to a list of 139,
each gender, among the number of items and to refine the
structure of the scale. In the confir-
using the following criteria: Extreme those who had matory phase, two adult samples were
words, such as "ferocious" and "boast-
ful," and words denoting temporary
previouslyendorsed a used to finalize the list of items in the
scale and to establish its reliability and
states,such as "aghast"and "hurt" were product,were John validity.
eliminated,since they were not suitable
for the impression-formation task.
McEnroeand Linda Study 1-The Exploratory Phase. In
It was believed that some of the 139 Evans. the exploratory phase of this research,
words retained could be unfamiliar to questionnaires for Madonna and John
survey respondents. To cull the unfa- McEnroe were developed, each of
miliar adjectives or descriptor words, In the next phase, 38 college stu- which contained two parts. The first
the list was rated by 38 college stu- dents were asked to indicate their level section asked respondents to indicate
dents. Based on their knowledge of the of familiaritywith each of the four ce- whether or not they were familiarwith
meaning of each word, the students lebrities mentioned above. In addition, a particular celebrity and could iden-
were instructed to rate each word as they were asked to specify the most- tify the person as being associatedwith
either familiaror unfamiliar.Words that and the least-appropriateproducts that a show or a specific profession. Addi-
were rated unfamiliarby more than 25 these individualscould sponsor. All the tionally,the respondents were asked to
percent of the subjects were eliminat- students in the sample showed famil- supply some standarddemographic in-
ed. This process reduced the list to a iarity with each celebrity. Further- formation. At this stage, if a respond-
smaller set of 104 words. more, the results suggested that Linda ent were to fail to recognize the
In the final editing phase, 52 college Evans would be most suitable for pro- celebrity, s/he would be removed from
students were presented with defini- moting a line of female cosmetics or a the study. In the next section, the re-
tions of trust, expertise, and attrac- perfume and least appropriatefor pro- spondents were presented with a scen-
tiveness. They were then instructed to moting items such as cigarettes, alco- ario in which a celebrity endorsed a
carefully study each adjective or de- hol, or "fatty foods." Tom Selleck was product; they were then asked to eval-
scriptorword and to indicateif the item perceived as appropriatefor promoting uate the celebrity, given 72 semantic
belonged to any of the three dimen- sports cars and men's cologne and as differentialitems discussed in the Item
sions defined earlier.Items with 75 per- incompatible with promoting credit Developmentsection of this paper. Two
cent or more agreement as belonging cards or long-distance telephone ser- hundred fifty students in a southern
to a certain construct were thus re- vices. The sample indicated tennis university completed the first version
tained for further analysis. This pro- rackets and other sporting equipment of the questionnaire, in which Madon-
cedure reduced the list to 72 words. as items that Jonn McEnroe should na was the celebrity promoting a new
promote. Cigarettes,clothing, and drug brand of designer jeans. In the second
Celebrity/Source Selection Proce- products were not considered appro- version of the instrument, respondents
dure. To identify an appropriatelist of priate for him. Finally, for Madonna, evaluated John McEnroe promoting a
personalities, 40 college students were the sample indicated that she should line of tennis rackets.A differentgroup
instructed to list all the celebrity names promote such items as designer jeans of 240 students completed the second
they could remember in three minutes. and other modern clothing, but should version. Except for changing the name
The celebrity names were later ranked not promote cars or breakfast cereals. of the celebrity, the product endorsed,

43
and the gender adjectives (such as The same steps were repeated for the Whitehead 1968).The third dimension
"beautiful"to "handsome"),the ques- sample evaluating John McEnroe. Ta- consists of eight items describing at-
tionnaires remained identical. ble 2 presents the results of the final tractiveness. Adjectives such as "at-
factor analysis for both the Madonna tractive,""beautiful,""charismatic," and
Exploratory Factor Analysis. To as- and McEnroe data. "sophisticated"are used to operation-
sess the structure of the source-credi- As can be seen from Table2, the first alize this dimension (Baker and
bility scale, all the items in the 11 adjectivesof the first factor identify Churchill 1977; DeSarbo and Harsh-
questionnairewere factor analyzed,us- the expertise dimension. Adjectives man 1985;Patzer1983).These patterns
ing the principal components analysis such as "expert,""knowledgeable "ex- appear fairly consistent for both the
followed by a varimaxrotation. The in- perienced," and "qualified"-all of Madonna and McEnroe samples, giv-
itial factor solution for the sample eval- which have been found to be clear in- ing further credence to the stability of
uatingMadonnaresulted in four factors dications of expertise (Applbaum and these factors.
with eigenvaluesgreaterthan one. The Anatol 1972;Simpson and Kahler 1980- The final items derived from each
four-factorsolution accounted for 68.4 81; Wynn 1987)-have loadings of 0.6 factor analysiswere tested for their re-
percent of the variance. In order to or higher on that factor. The second liability by submitting them to item
purify the list, items with loadings of factor, consisting of eight items, meas- analysis using item-to-total correla-
0.3 or greater on more than one of the ures the trustworthiness dimension. tions. The items for each subscale were
factors were eliminated. The reduced Again, items such as "trustworthy," analyzed separately.To obtain a prac-
list was factor analyzed a second time. "honest,""dependable,""reliable,"and tical size scale (five items per factor),
This resulted in three factors with ei- "ethical,"which have previously been items with the lowest item-to-total cor-
genvalues greater than one, while ac- used to represent this factor,have high relations were deleted while maintain-
counting for 61 percent of the variance. loadings (Bowers and Phillips 1967; ing an acceptable level of reliability as

TABLE 2
Factor Loadings for the Three Dimensions of Source Credibility

Factor Loadings
Madonna Sample (n = 249) John McEnroe Sample (n 237)
1 2 3 1 2 3
Authoritative .617 -.037 .051 .654 -.156 .009
Compatiblewith the product .795 .103 .070 .706 - .147 .022
Expert .765 .107 .045 .701 - .037 .056
Informative .812 .170 .183 .747 .108 .027
Experienced .725 .186 .021 .531 .077 .034
Intelligent .697 .262 .119 .699 .133 .054
Informedaboutthe product .799 .212 .055 .631 -.025 - .195
Knowledgeable .798 .246 .174 .695 .121 .079
Qualified .869 .069 .131 .536 .116 - .141
Familiarwith the product .696 .071 .132 .606 -.020 -.233
Skilled .699 .042 .186 .607 -.061 .054
Dependable .159 .742 .150 .042 .683 - .013
Fair .005 .655 .198 .108 .709 .188
Reliable .234 .780 .185 .109 .739 .006
Sincere .152 .720 .206 .002 .609 .223
Trustworthy .128 .768 .215 .169 .772 .158
Truthful .211 .704 .185 .108 .744 .146
Honest .139 .664 .280 .142 .639 .251
Ethical .241 .587 .227 .178 .646 .153
Attractive .046 .234 .831 -.051 .158 .721
Classy .206 .162 .646 .049 .196 .500
Sophisticated .189 .275 .596 .052 .175 .496
Handsome/Beautiful .154 .252 .773 .054 .090 .692
Glamourous .149 .202 .761 .079 .187 .541
Elegant .212 .185 .661 -.184 .059 .704
Sexy .107 .248 .744 - .065 .093 .761
Charming .170 .283 .691 -.112 .261 .597

44
measured by Cronbach's alpha (Peter with the product, the celebrity's role tal of 360 questionnaires (180 for each
1979). To determine if the subscales in helping both the image and the sales celebrity) were delivered, 289 collect-
were equally reliable for different ce- of the product, and the perceived com- ed, and 265 found suitable for analysis.
lebrities and genders, Cronbach'salpha parativedistinctiveness of that product The attrition rate was due either to ex-
was computed for both male and fe- as compared to other brands in the cessive missing data or to obvious re-
male respondents for Madonna and market. sponse bias. The final sample included
John McEnroe. The results indicated a 138 usable questionnairesfor Linda Ev-
highly reliable scale. Both male and fe- ans and 127 for Tom Selleck.
male respondents had equally reliable
response patterns, and the total sample ... questions were Confirmatory Factor Analysis. It is
for each subscale and celebrity had a widely recognizedthat exploratory fac-
reliability coefficient of 0.8 or higher.
included to assess tor analysis can be quite useful in the
respondents' early stages of scale development.
Study 2-The Confirmatory Anal- However, as more knowledge is ac-
ysis. In the confirmatory phase of the
perceived similarity to quired about the nature of the scale,
study,the final scale'sreliabilityand va- and likability of the new data and more rigorous statistical
lidity were assessed by using the 15 techniques should be applied to con-
items (five items per subscale) obtained
celebrity. firm or disprove the results obtained in
from the exploratory phase, along with the exploratory stage. Thus, in the sec-
several other validation items. These ond phase of this research, a confir-
additional items were included to matory factor analysis model was
The 15 adjectives developed in the
measure respondents' likelihood to in- specifiedin order to verify the tri-com-
exploratoryphaseof this research,along
quiTe about, consider purchasing, and ponent structure of the scale. The con-
with the validationbehaviors discussed
actually purchasefor personal use the cept of confirmatory factor analysis is
above, were administered to an adult
product sponsored by the particular as follows: Given a set of observable
sample. Furthermore, in the confir-
celebrity. Inquiry, consideration, and response variables(the 15 items for at-
matory phase, differentcelebrities, spe-
purchase represent increasing levels of tractiveness, expertise, and trustwor-
cifically,Linda Evans and Tom Selleck,
commitment toward the product. thiness), this process attempts to
were used to support the generaliza-
These criterion variablesare common- determine a smaller set of underlying
bility of the scale. Linda Evans was en-
ly used to measure the effectiveness of latent factors (attractiveness,expertise,
dorsing a new brand of perfume, and
source credibility in marketing re- and trustworthiness dimensions). Us-
a new brand of men's cologne was en-
search (Baker and Churchill 1977; ing the procedure suggested by Jores-
dorsed by Tom Selleck.
Kahle and Homer 1985; Rubin, Mager, kog (1979), the confirmatory factor-
and Friedman 1982). For further vali- Data Collection Procedure. The sub- analysis model was defined as follows:
dation of the scale, respondents were jects for the confirmatoryanalysiswere
x=A4+6
asked to indicate the likelihood that selected through a systematicarea-sam-
they would inquireabout, considerpur- pling technique (Churchill 1987)which where:
chasing, or actually purchasethe prod- x is a (15 X 1) column vector of ob-
has been widely used and accepted in
uct as a gift. Because of a gift's served variables (the 15 items for
survey research (Lovelock et al. 1976;
conspicuous nature, gift-giving deci- Survey Research Center 1976). In the source-credibility scale)
sions, as compared to purchase for self, current study,all census tractsin a small A is a (15 X 3) column pattern coef-
are often perceived as being more im- southeastern city were chosen for sam- ficient matrix of x on 4
portant and more involving (Belk 1982; pling. Within each tract a number of t is a (3 X 1) column vector of di-
Clarke and Belk 1979;Kassarjian1981). blocks (depending on the population) mensions (attractiveness, expertise,
Additionally, questions were includ- were chosen for sampling, excluding and trustworthiness) derived from
ed to assess respondents' perceived commercialblocks and blocks contain- the observed variables (x)
similarity to and likability of the ce- ing parks, churches, or schools. Each a is a (15 X 1) column vector of er-
lebrity. Similarity and likability have interviewer was given a map of the area rors of measurement of x
been used extensively in the literature to be sampled. For each block the in-
as determinants of identification and terviewer was instructed to randomly (D is a (3 X 3) symmetric covariance
interpersonal attraction between the select a house and then to conduct in- matrixof 4
source and the message recipient terviews at every other house until the The confirmatoryfactoranalysis
(Aronson and Worchel 1966; Ber- quota for that block was filled. The in- model for the 15 source-credibility
scheid 1966; Griffitt 1966; Kelman terviewing procedure involved the per- items is presented in Figure 1. Table 3
1961). Finally,respondents were asked sonal delivery and collection of self- presents the input correlation matrices
to evaluatethe celebrity's compatibility administered questionnaires. A to- for the 15 source-credibility items of

45
the Linda Evans and Tom Selleck data p =.168; X2EVANS=109.71, df = 87, Sorbom 1988). The root mean square
sets. p = .051). Further, the plot of the nor- residual was .048 and .046 for the Sel-
Using the LISRELmethodology (Jo- malized residuals approximated a leck and Evans models, respectively.
reskog and Sorbom 1988) to verify the straight line, indicating that there were The reliability estimates of each item
relationship between observable vari- no specification errors or departures are shown in Table 4. As can be seen,
ables and latent constructs, two con- from normality in the data.In addition, the individual items appear to be reli-
firmatoryfactor-analysismodels (Linda examination of the Q-plots indicated able. Further, the confirmatory factor
Evans and Tom Selleck) were tested that the normalized residuals had a model for each celebrity explains about
separately.The x2statistic was nonsig- slope larger than one as compared to 90 percent of the variationfor the three
nificant for each model, indicating an the 45-degree line, which is an addi- dimensions of source credibility, indi-
adequate fit of the confirmatory model tional confirmation of the fit of the data cating highly reliable dimensions. Fi-
to the data. (X2SELLECK= 99.60, df = 87, to the specified model (Joreskog and nally, the values of pc(4), which are the

FIGURE 1
Confirmatory Factor Analysis Model for the Three Dimensions of the Source-Credibility Scale

x, xI,,
ATTRACTIVE

62~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 12* LS

B22 EXPERIENCED 0 62

A TRATIVNESTWRTIEXPRISSNO)DGAL

3
x

FIGURE 1
Confirmatory Factor Analysis Model for the Three Dimensions of the Source-Credibility Scale

Xl X2 X3 x11 X4
xE4 X5 X6 X7 X8 X9 X10 X12 X13 X15

Xl ~~.595 .787 .636 .715 .421 .374 .411 .526 .468 .326 .333 .420 .377 .327
X2 .621 .620 .523 .498 .400 .330 .402 .491 .476 .247 .336 .367 .332 .300
X3 .721 .678 .579 .706 .335 .375 .366 .485 .425 .354 .320 .395 .354 .366
X4 .582 .666 .605 .530 .368 .166 .312 .380 .226 .290 .331 .396 .293 .276
X5 .653 .561 .712 .618 .362 .367 .426 .480 .414 .449 .383 .447 .423 .415
X6 .281 .357 .301 .371 .288 .590 .694 .598 .604 .333 .392 .427 .354 .324
X7 .277 .295 .219 .294 .243 .603 .589 .592 .664 .295 .338 .329 .360 .376
X8 .306 .314 .311 .352 .304 .672 .536 .598 .643 .286 .431 .384 .314 .293
Xg .392 .395 .343 .370 .371 .609 .576 .647 .762 .375 .440 .467 .461 .399
X10 .299 .282 .236 .280 .261 .632 .649 .602 .758 .307 .403 .365 .336 .375
X,1 .188 .167 .197 .147 .183 .123 .058 .115 .139 .167 .583 .660 .623 .600
X12 .286 .278 .289 .259 .292 .217 .220 .317 .333 .342 .636 .701 .563 .543
X13 .204 .152 .232 .250 .189 .191 .199 .239 .201 .266 .570 .585 .684 .640
X,4 . 165 .158 .200 .228 .177 .116 .109 .153 .154 .129 .512 .636 .715 .608
X15 .221 .270 .259 .282 .281 .204 .150 .283 .250 .152 .488 .639 .630 .650
'Above-diagonal entries represent Tom Selleck data. Below-diagonal entries represent Linda Evans data.

46
averagevariances extracted by each di- TABLE 4
mension, are well above the 0.5 cut-off Item Reliability, Construct Reliability, Interconstruct
point. Thus, the variance captured by Correlation, and Average Variance Extracted for the
each dimension is significantly higher Three Dimensions of Celebrity Endorser-CredibilityScale
than the variance due to measurement
error, indicating adequate convergent
validityfor each dimension (Fornelland Item Reliability
Larcker1981).The final tri-component Linda Evans Tom Selleck
scale is presented in the Appendix. Attractiveness Dimension
Attractive .669 .799
Nomological Validity. Nomological va- Classy .637 .476
lidity investigates the relationship be- Handsome/Beautiful .748 .764
tween the scoresof a scaleand how these Elegant .548 .468
scores relate to the measures of other Sexy .661 .638
constructsor behaviors.If the suggested Construct Reliability .904 .893
relationships between constructs are Avg. Var. Extracted .653 .629
empirically supported, then it is as- Construct Correlation Attract/Trust .477 .621
sumed that the measures of those con- Trustworthiness Dimension
structs have a certain degree of Dependable .674 .575
nomologicalvalidity(Peter 1981). Honest .524 .558
The nomologicalvalidityin this study Reliable .604 .596
was tested by relatingscores on each di- Sincere .696 .704
mension of expertise, trustworthiness, Trustworthy .653 .734
and attractivenessto severalself-report- Construct Reliability .895 .896
ed behaviors. Specifically,intention to Avg. Var. Extracted .630 .633
purchasethe product, the role of the ce- Construct Correlation Trust/Expert .319 .579
lebrity in helping the imageand the sale Expertise Dimension
of the product, and respondents' liking Expert .564 .590
of and perceivedsimilarityto the source Experienced .702 .587
were used as validationbehaviors. Giv- Knowledgeable .567 .767
en that the validation measures were Qualified .647 .616
single-itemscales,PearsonProduct Mo- Skilled .556 .557
ment Correlations were determined to Construct Reliability .885 .892
be more appropriatefor use than the Avg. Var. Extracted .607 .623
more rigorousstructuralequation mod- Construct Correlation Expert/Attract .350 .553
eling approach for the assessment of
nomologicalvalidity.Correlation coeffi-
cients were tested for significance at
a =.05 for the one-tail test. Table 5 Self-report measures of intention to correlatedwith celebrity endorser's ex-
shows the correlation coefficients for purchase the product for a gift also pertise, trustworthiness, and attrac-
these analyses. produced significantcorrelationsacross tiveness. Similar results of a smaller
All three intention-to-purchaseitems the three source-credibility dimen- magnitude were also found between
were significantly correlated with the sions. This pattern was true for both respondents' perceived similiarity to
expertise, trustworthiness, and attrac- the Selleck and the Evans sample. Re- the source and the tri-component con-
tiveness dimensions for Tom Selleck. spondents considered the brands en- struct.
For the Linda Evans sample, inquiring dorsedby Tom Selleck and LindaEvans
about the brand and considering pur- as distinctive,comparedto other brands Convergent and Discriminant Valid-
chase of the brand were significantly on the market, and perceived that the ity. The final step in validatingthe scale
correlated with the three dimensions. celebrity would help the image and the was the determination of its conver-
The magnitude of the relationship be- sale of the product. The relationships gent and discriminant validity by way
tween source trustworthiness and in- between these variables and the at- of a multitrait-multimethod matrix
tention to purchase the product for tractiveness, trustworthiness, and ex- (MTMM). For the development of the
individual consumption was smaller pertise dimensions were significantand MTMM, at least three different meth-
than the other two purchase measures; in the positive direction. ods are required to measure each di-
however, it was still significant at p < Finally,as suggested in the literature, mension. For this research, Likert and
.05 level. likability of the endorser was highly Stapel verions of the source-credibility

47
TABLE5
CorrelationsAmong Three Source-Credibility
Factorsand SeveralSelf-ReportValidityMeasures'

LindaEvans TomSelleck
Attractiveness TrustworthinessExpertise Attractiveness TrustworthinessExpertise
Self-Consumption
Inquireaboutthe brand .507 .311 .474 .473 .584 .520
Considerpurchasingthe brand .433 .210 .455 .474 .568 .598
Purchasethe brand .374 .145 .485 .467 .586 .554
Gift-Giving
Inquireaboutthe brand .320 .197 .315 .501 .451 .583
Considerpurchasingthe brand .290 .176 .367 .489 .428 .575
Purchasethe brand .248 .190 .404 .447 .450 .561
Other ValidityMeasures
Distinctivecomparedto otherbrands .319 .286 .364 .445 .437 .526
Celebritywill help the product'ssale .386 .361 .492 .495 .464 .584
Celebritywill help the product'simage .301 .302 .418 .427 .417 .551
Celebrityis likable .582 .598 .307 .655 .661 .597
Perceivedself and sourcesimilarity .257 .249 .253 .364 .325 .220
'All correlationcoefficientsare significantat p < .05 level.

items were developed and used, along for determining the convergent and Given that the test for congenerity
with the original semantic differential disciminant validity was to determine was rejected, the next step was to test
version to determine if the three di- if the three traits as measured by the for discriminantvalidity.This involves
mensions (trait factors)-attractive- three methods were congeneric. This including a method factorfor each trait
ness, trustworthiness, and expertise test assumes that the correlation coef- in H1: The solution for this model
had convergent and discriminantvalid- ficient among the three traits is equal yielded a goodness-of-fit of X2= 12.06,
ity. The celebrity used for this analysis to unity-that is, the three dimensions d.f. = 13 (p = 0.523), which represents
was Linda Evans, promoting a new of attractiveness,trustworthiness, and a good fit of the data to the hypothe-
brand of perfume. expertise are the same. Congenerity is sized model. Table 7 presents the par-
One hundred eight undergraduate determined by computing the differ- titioning of variance due to the trait
students were recruited for this phase ence between the x2of the following method and the error factor for each
of the analysis, with data collected in two hypotheses (Joreskog 1971). dimension of the source-credibility
three stages. In the first stage, each re- scale. As can be seen, each of the three
spondent was randomly given one of H1: X1,... Xg*6 . .. 6g PI, P2, and P3 methods adequately captured the di-
the three versions (Likert, Stapel, and are unconstrained. mensions of source-credibility.In each
Semantic Differential)of the scale,along H2: pI = P2 = p3 = 1 case, the variance due to the trait is
with filler questions. After completing significantly larger than the variance
the first version of the questionnaire, The LISREL model for testing H1 due to either the method or the error
the students were given a five-minute X2(X2= 33.07,
produceda nonsignificant factor.The above analyses provide evi-
break and then given a second version, d.f. = 24, p <0.103). However, the dence that the source-credibility scale
which also contained filler questions. model for H2 produced a highly signifi- developed in this study has acceptable
After a second break, the students cant x2value (X2
= 374.87, d.f. = 27, convergent and discriminant validity.
completed the third version of the scale. p < 0.001). Based on this evidence, we
To minimizedemandcharacteristicsand can reject the assumptionof congeneri- Discussion
response bias, the order of presentation ty (X2= 341.80, d.f. = 3, p < 0.001) and
of the items and the versions of the conclude that the attractiveness,trus- Since the original contributions of
scale were randomly varied among re- tworthiness, and expertise dimensions Hovland and his colleages (Hovland,
spondents. do not measurethe same trait. Further- Janis, and Kelly 1953), the concept of
The assessment of convergent and more, the nonsignificantx2 for H1 pro- source credibility has been extensively
discriminant validity of the scale fol- vides evidence for convergent validity. studied in psychology (Berscheid 1966;
lowed the steps outlined by Joreskog Table 6 provides the MTMM correla- Chaiken 1979; Johnson, Torcivia, and
(1971) and Bagozzi(1980). The first step tions matrixfor testing H1 and H2. Popprick 1968; McGinnies and Ward

48
1980; Mills and Harvey 1972; Ross were the manipulatedfactor,a post-test ent demographic and psychographic
1973; Wu and Shaffer 1987), in com- administrationof the scale should pro- consumer groupscan be evaluated.This
munication (Applbaum and Anatol duce a statisticallysignificantdifference concept also has applications beyond
1972; Berlo, Lemert and Mertz 1969; for low- and high-treatment groups. the use of celebrity spokespersons: the
McCroskey 1966; Miller and Baseheart Given the large sums of money spent scale can be applied effectively in
1969; Whitehead 1968), in marketing, on celebrity advertising, advertisers choosing the most appropriate "aver-
and in advertising(Baker and Church- should use the scale as an integral part age consumer" as a spokesperson. Fi-
ill 1977; Caballero and Solomon 1984; of their effectivness testing and track- nally, the dimensions of a celebrity
DeSarbo and Harshman 1985; Kahle ing. The scale is simple to use and con- endorser's credibility, along with con-
and Homer 1985; Mowen and Brown venient for large-sample administra- sumer demographics and psycho-
1981; Wynn 1987). Despite its wide- tions. For segmentation strategies, the graphics, can be used as potential
spread use, source credibility has not wisdom of using a celebrity and the predictors of attitude toward and in-
been properlyoperationalizedby means spokesperson's effectiveness for differ- tention to purchase a specific product.
of a reliableand valid scale. In addition,
the experimentalstudies that have used
various dimensions of source credibil-
ity have not been consistent in their
TABLE 6
manipulationchecks of the experimen- Input Correlations for the Multitrait-Multimethod Matrix' (n = 108)
tal variables. The current research has
defined the domain of the source-cred- Semantic
ibility construct and has developed a Differential
Scale LikertScale StapelScale
reliable and valid scale for its measure- Attract Trust Expert Atract Trust Expert Attract Trust Expert
ment. SemanticDifferentialScale
From a theoretical perspective, the Attract
present scale should replace the ple- Trust .397
thora of single-item measures of at- Expert .402 .454
tractiveness, expertise, and trust- Likert Scale
worthiness. By identifying and meas- Attract .787 .410 .431
uring this tri-component construct, re- Trust .405 .717 .393 .428
searchers can validly assess the impact Expert .383 .419 .755 .339 .357
of each component of a celebrity en- Stapel Scale
dorser'spersuasiveness.The consistent Attract .832 .440 .413 .807 .393 .461
Trust .399 .856 .501 .472 .783 .442 .485
use of the same instrument can illu-
Expert .404 .437 .793 .461 .443 .779 .456 .477
minate the comparison of findings
across several studies and can contrib- 'Attract= Attractiveness; Expert= Expertise
Trust= Trustworthiness;
ute to the source-credibility literature.
The present scale can be adapted to
a variety of situations. Researchers in
political science can use the scale to TABLE 7
investigate the credibility of political
Partitioning of Variance Due to Trait, Method, and Error
candidates. In political campaigns, a
candidate'ssuccess depends on his/her
ability to acquire the voter's trust, ap- Trait PercentDue to Method Error
proval, and confidence in his/her SemanticDifferentialScale
knowledge and ability.Periodic checks Attractiveness 0.756 0.037 0.204
could be performed at various stages Trustworthiness 0.790 0.027 0.183
in the campaign to evaluate the level Expertise 0.778 0.005 0.217
of the candidate's credibility. In in- likert Scale
structional settings, the scale can be Attractiveness 0.912 0.072 0.016
used to evaluate the influence of the Trustworthiness 0.649 0.001 0.341
Expertise 0.733 0.071 0.197
instructor's characteristics on student
Stapel Scale
evaluations of the teacher. Finally, in Attractiveness 0.876 0.134 0.000
experimental studies of source credi- Trustworthiness 0.943 0.008 0.050
bility, the scale can be used to assess Expertise 0.828 0.003 0.169
the effectiveness of the experimental
'Denotesa parameterfixedat zeroto eliminatethe HaywoodEffect
manipulation.For example, if expertise

49
Limitations and For example, how does source credi- References
bility influence purchase intentions
Research Extensions with high-involvementproducts, as op-
Allport, G. and H. Odbert (1936), "Trait-
Names: A Psycho-LexicalStudy,"Psycho-
The present study has a number of lim- posed to low-involvement products? logicalMonographs,211.
itations;the recognition of these should Should celebrities and other credible Anderson,N.H. (1968),"LikablenessRatings
sources be used with high-involvement of 555 Personality-Trait Words,"Journalof
help refinefuture researchefforts.With Personalityand SocialPsychology, 9 (2),272-
regard to the three dimensions of the or low-involvementproducts?And how
279.
scale, the selection of expertise, trust- does the level of consumer confidence Anderson, K. and T. Clevenger (1963), "A
worthiness, and attractivenesswas mo- and knowledge about the product me- Summary of Experimental Research in
tivatedby previoustheoreticalwork and diate the impact of source credibility Ethos:' SpeechMonographs,30 (June),59-
on intentions to purchase the product? 78.
empiricalobservations, especially in se- Applbaum,Ronald F and Karl WE. Anatol
lecting the attractiveness construct. Additionally, given that the con-
(1972), "The Factor Structure of Source
Therefore, one should be cautioned structs of expertise, trustworthiness, Credibilityas a Function of the Speaking
that the quantitativeanalysis establish- and attractivenessare correlated,an in- Situation,'SpeechMonographs, 39 (August),
es the reliabilityand validityof the scale teresting line of researchwould involve 216-222.
the study and determination of the Aronson, Elliot and Philip Worchel (1966),
rather than discovers their existence.
causal order among these constructs. "Similarityversus Liking as Determinants
As research findings continue in this of InterpersonalAttractiveness,"Psychon-
area,the existing scale can be expanded For example, does a celebrity endor- omicScience,5 (4), 157-158.
or modified. ser'sattractivenessaffecthis or her per- Bagozzi,RichardP. (1980), Causal Modelsin
ceived expertise, which in turn, affects Marketing, New York:JohnWileyand Sons,
the level of trustworthiness? What Inc.
conditions moderate the order and im- Baker,MichaelJ. and Gilbert A. Churchill,
Given the large sums pact of these variables? Answers to
Jr. (1977), "The Impact of PhysicallyAt-
tractive Models on Advertising Evalua-
of money spent on these and other questions should help tions,"Journalof MarketingResearch,14
advertising practitioners identify the (November),538-555.
celebrity advertising, most appropriate sources for their Belk, RussellW. (1982),"Effectsof Gift Giv-
ing Involvementon Gift Selection Strate-
advertisers should use clients' products or services. Further,
gies,"in Advancesin ConsumerResearch,
use of this scale can improve the un- Vol. 9, Andrew Mitchell, ed., Ann Arbor,
the scale as an derstanding of how consumers in dif- MI: Association for Consumer Research,
integral part of their ferent situations react to different 408-411.
Berlo,DavidK.,JamesB. Lemert,and Robert
sources and how source credibility in-
effectiveness testing fluences purchase intentions. J. Mertz(1969),"Dimensionsfor Evaluating
the Acceptabilityof MessageSources,"Pub-
and tracking. The lic OpinionQuarterly,33 (Winter),563-576.
Berscheid,Ellen(1966),"OpinionChangeand
scale is. . . APPENDIX Communicator-CommunicateeSimilarity
andDissimilarity,"
convenient for large- Source-Credibility Scale SocialPsychology,
Journalof Personality
4 (6), 670-680.
and

sample Bowers, John W and William A. Phillips


(1967),"ANote on the Generalityof Source
Attractiveness
administrations. Attractive-Unattractive
CredibilityScales,"SpeechMonographs,34
(August),185-186.
Classy-Not Classy BusinessWeek(1987), "Nothing Sells Like
Beautiful-Ugly Sports,"(August31), 48-52.
With regard to generalizability,the Elegant-Plain Caballero,Marjorie,JamesR. Lumpkin,and
Sexy-Not sexy CharlesS. Madden(1989),"Using Physical
findings of this study are limited to the Attractivenessas an AdvertisingTool: An
celebrities and product endorsements Trustworthiness EmpiricalTest of the Attraction Phenom-
tested in this research. An interesting Dependable-Undependable enon,"Journalof AdvertisingResearch,29
retrospection involves the use of celeb- Honest-Dishonest (August/September),16-22.
rities endorsing products that are not Reliable-Unreliable Caballero,Marjorieand PaulSolomon (1984),
Sincere-Insincere "Effectsof Model Attractivenesson Sales
considered appropriatefor their image. Response,"Journal of Advertising,13 (1),
Trustworthy-Untrustworthy
This approachwould provide addition- 17-23, 33.
al credence to the reliabilityand valid- Expertise Chaiken,Shelly(1979),"Communicator Phys-
ity of the scale. Expert-Not an expert icalAttractivenessand Persuasion," Journal
Future research should examine the Experienced-Inexperienced of Personalityand SocialPsychology,37 (2),
Knowledgeable-Unknowledgeable 1387-1397.
impact of consumer involvement and Churchill, Gilbert A., Jr. (1987), Marketing
Qualified-Unqualified
confidence in the product as mediating Research:MethodologicalFoundations,4th
Skilled-Unskilled
variablesin source-credibilityresearch. ed., Chicago,IL: The Dryden Press.

50
Clarke,KeithandRussellW Belk(1979),"The Authoritarianismand Attitude Change," (1), (March),1-7.
Effects of Product Involvement and Task Journalof Personalityand SocialPsychology, Mills, J. and E. Aronson (1965), "Opinion
Definitions on Anticipated Consumer Ef- 9 (2), 179-183. Change as a Function of Communicator's
fort," in Advances in ConsumerResearch, Joreskog,KarlG. (1971),"StatisticalAnalysis Attractiveness and Desire to Influence,"
Vol.6, WilliamL. Wilkie, ed., Ann Arbor, of Sets of CongenericTests,"Psychometrika, Journalof Personalityand SocialPsychology,
MI: Association for Consumer Research, 36 (1), 109-133. 1 (2), 173-177.
313-318. Joreskog,KarlG. (1979),"BasicIdeasof Factor Mills,JudsonandJohnHarvey(1972),"Opin-
Crano,WilliamD. (1970),"Effectsof Sex, Re- and Component Analysis,"in Advancesin ion Change as a Function of When Infor-
sponse Order, and Expertisein Conform- Factor Analysis and StructuralEquations mation About the Communicator is
ity:A DispositionalApproach,"Sociometry, Models,K.G.Joreskogand D. Sorbom,eds., Receivedand Whether He is Attractiveor
33 (September),239-252. Cambridge,MA: Abt Books, 5-20. Expert,"Journal of Personalityand Social
Crisci,Richardand HowardKassinove(1973), Joreskog,Karl G. and Dag Sorbom (1988), Psychology,21(1), 52-55.
"Effectsof PerceivedExpertise,Strengthof LISREL7:A Guideto the Programand Ap- Morrison,Ann M. (1980),"The Boss as Pitch-
Advice, and EnvironmentalSetting on Pa- plications,Chicago, IL: SPSS Inc. man,"Fortune,(August 25), 66-73.
rental Compliance,"The Journal of Social Joseph,W. Benoy (1982),"The Credibilityof Mowen, John C. and Stephen W Brown
Psychology,89 (2), 245-250. Physically Attractive Communicators: A (1981),"On Explainingand Predictingthe
DeSarbo, Wayne S. and Richard A. Harsh- Review,"Journalof Advertising,11 (3), 15- Effectivenessof Celebrity Endorsers,"Ad-
man (1985),"Celebrity-BrandCongruence 24. vancesin ConsumerResearch:Vol. 8, Kent
Analysis,"Current Issues and Researchin Kahle,Lynn R. and PamelaM. Homer (1985), M. Moore,ed., Ann Arbor,MI:Association
Advertising,].H. Leighand C.R. Martin,Jr., "PhysicalAttractivenessof the Celebrity for Consumer Research,437-441.
eds., Ann Arbor,MI:Division of Research, Endorser: A Social Adaptation Perspec- Osgood, C.E., G.A. Suci, and P.H. Tannen-
GraduateSchool of Business Administra- tive," Journal of Consumer Research, 11 baum(1957),The Measurementof Meaning,
tion, The University of Michigan, 17-52. (March),954-961. Urbana,IL: University of Illinois Press.
Fornell, Claes and David F Larcker(1981), Kassarjian,Harold H. (1981), "Low Involve- Patzer,Gordon L. (1983),"SourceCredibility
"EvaluatingStructural Equation Models ment-A Second Look," in Advances in as a Function of CommunicatorPhysical
with UnobservableVariablesand Measure- ConsumerResearch,Vol. 8, Kent B. Mon- Attractiveness," Journal of Business Re-
ment Error," Journalof MarketingResearch, roe, ed., Ann Arbor, MI: Association for search,11 (2), 229-241.
18 (February),39-50. ConsumerResearch,31-34. Peter,Paul J. (1979), "MeasurementReliabil-
Friedman,Hershey and I. Friedman(1976), Kelman, Herbert C. (1961), "Processes of ity: A Review of PsychometricBasicsand
"Whom Do Students Trust?"Journal of Opinion Change,"PublicOpinionQuarter- Recent Marketing Practices,"Journal of
Communication,26 (1), 48-49. ly, 33 (Spring),57-78. MarketingResearch,16 (February),18-25.
Friedman,HersheyH., MichaelJ.Santeramo, Kelman, H.C. and C.I. Hovland (1953), Peter, Paul J. (1981), "ConstructValidity:A
and Anthony Traina(1979), "Correlatesof "Reinstatementof the Communicatorin Review of BasicIssuesand MarketingPrac-
Trustworthinessfor Celebrities,"Journalof DelayedMeasurementof Opinion Change," tices," Journal of MarketingResearch,18
Academy of MarketingScience, 6 (4) 291- Journalof Abnormaland SocialPsychology, (May),133-145.
299. 48 (July),327-335. Ross, JoelA. (1973),"Influenceof Expertand
Giffin, Kim (1967), "The Contribution of Lovelock,ChristopherH., Ronald Stiff, Dav- Peer Upon Negro Mothers of Low Socio-
Studies of Source Credibilityto a Theory id Cullwick, and Ira M. Kaufman(1976), economicStatus,"TheJournalof SocialPsy-
of InterpersonalTrust in the Communi- "AnEvaluationof the Effectivenessof Drop- chology,89, 79-84.
cation Process,"PsychologicalBulletin, 68 Off Questionnaire Delivery," journal of Rubin, Vicki, Carol Mager,and Hershey H.
(2), 104-119. Marketing,13 (November),358-364. Friedman(1982),"CompanyPresidentver-
Griffitt,WilliamB. (1966),"InterpersonalAt- Maddux, James E. and Ronald W Rogers sus Spokesperson in Television Commer-
tractionsas a Functionof Self-Conceptand (1980),"Effectsof Source Expertness,Phys- cials,"Journalof AdvertisingResearch,22
Jour-
Personality Similarity-Dissimilarity," ical Attractivenessand Supporting Argu- (August/September),31-33.
nal of Personalityand Social Psychology,4 mentson Persuasion:A Caseof BrainsOver Simon,HerbertW, Nancy N. Berkowitz,and
(6), 581-584. Beauty,"journal of Personalityand Social R. John Moyer (1970), "Similarity,Credi-
Horai, J.M., N. Naccari , and E. Fatoullah, Psychology,39 (2), 235-244. bility,and Attitude Change:A Review and
(1974),"The Effectsof Expertiseand Phys- McCroskey,JamesC. (1966), "Scalesfor the a Theory," PsychologicalBulletin, 73 (1),
ical AttractivenessUpon Opinion Agree- Measurement of Ethos," Speech Mono- (January),1-16.
ment and Liking,"Sociometry,37 (4), 601- graphs,33, 65-72. Simpson,EdwinK. and Ruel C. Kahler(1980-
606. McGinnies, Elliott and Charles D. Ward 81), "A Scale for Source Credibility,Vali-
Hovland,Carl I., Irving K. Janis,and Harold (1980), "Better Liked Than Right: Trust- dated in the Selling Context,"The Journal
H., Kelley(1953),Communicationand Per- worthiness and Expertise as Factors in of PersonalSellingand Sales Management,
suasion,New Haven, CT: Yale University Credibility,"Personalityand Social Psychol- 12 (Fall/Winter),17-25.
Press. ogy Bulletin,6 (3), 467-472. Slinker,BarryH. (1984), "WouldYou Buy a
Hovland,CarlI. andWalterWeiss(1951),"The McGuire, William J. (1985), "Attitudesand Burger from This Man? A Car? Some
Influence of Source Credibility on Com- Attitude Change,"in Handbookof Social Stocks?"MadisonAvenue,(April),52-58.
munication Effectiveness,"Public Opinion Psychology,Vol.2, GardnerLindzeyand El- Steadman,M. (1969),"How Sexy Illustrations
Quarterly,15 (Winter),635-650. liot Aronson, eds., New York: Random AffectBrandRecall,"Journalof Advertising
Johnson,HomerH., and RichardIzzett(1972), House, 233-346. Research,9 (February),15-19.
"The Influenceof Source Identificationon Miller,Cyndee (1989), "Celebs' Sweet Smell Sternthal,Brian, Ruby Dholakia, and Clark
Attitude Change as a Function of the Type of Success GeneratesDollars and Scents," Leavitt (1978), "The PersuasiveEffect of
of Communication,"Journalof Social Psy- MarketingNews, (September25), 8. Source Credibility:Tests of Cognitive Re-
chology,86 (1), 8-87. Miller, Gerald P. and John Basehart (1969), sponse,"journal of ConsumerResearch,4
Johnson, Homer H., JamesM. Torcivia,and "Source Trustworthiness, Opinionated (March),252-260.
MaryAnn Poprick(1968),"Effectsof Source Statements, and Response to Persuasive Survey ResearchCenter (1976), Interviewer's
Credibility on the Relationship Between Communication,"Speech Monographs,36 Manual, RevisedEdition,Ann Arbor,MI:

51
The University of Michigan. "ConsumerGoalsand Reactionsto a Com- personsand SalesManagers," Developments
Walley,Wayne (1987), "ActorsSet Contract municationSource,"Journalof Markeing in MarketingScience,Vol.10,JonM. Hawes
Talks:'AdvertisingAge, (Dec. 21), 4. Research,5 (February),73-77. and George B. Glisan, eds., Bal Harbour,
Whitehead,JackL. (1968),"Factorsof Source Woodside,Arch G. andl. WilliamDavenport, FL: Academy of MarketingScience, 353-
QuarterlyJournalof Speech,54
Credibility," Jr.(1974),"The Effectof SalesmanSimilar- 358.
(1), 59-63. ity and Expertiseon ConsumerPurchasing Wu, Chenghuanand DavidR. Shaffer(1987),
Whittaker,James0. and Robert D. Meade Behavior,"Journalof MarketingResearch, "Susceptibilityto PersuasiveAppeals as a
(1968),"Retentionof Opinion Change as a 11 (May),198-202. Function of Source Credibilityand Prior
Functionof DifferentialSourceCredibility: Woodside,Arch G. andJ.WilliamDavenport, Experiencewith the AttitudeObject,"Jour-
A Cross-Cultural Study," International Jr. (1976), "Effectsof Price and Salesman nal of Personalityand Social Psychology,
52
Journalof Psychology,3 (2), 103-108. Expertiseon CustomerPurchasingBehav- (4), 677-688.
Widgery, Robin N. and Richard S. Ruch ior,"Journalof Business,49 (January),51-
(1981), "Beauty and the Machiavellian,' 59. ReceivedJune 9, 1989. Revisionacceptedfor
Communication Quarely, 29 (Fall),297-301. Wynn, George W (1987), "The Effectsof a publicationApril 13, 1990.
Wilding,John and RaymondA. Bauer(1968), Salespersons'Credibilityon Other Sales-

52

Vous aimerez peut-être aussi