Source Credibility: on the Independent Effects of Trust and Expertise

Joshua L. Wiener, Oklahoma State University
John C. Mowen, Oklahoma State University
ABSTRACT - Marketing researchers and practitioners have tended to assume that a more trustworthy source is more credible than a less trustworthy source. However, previous studies have either confounded trustworthiness with expertise and/or attractiveness, or found that trustworthiness did not influence persuasion. In the 2 x 3 between subjects factorial experiment three levels of source expertise were crossed by two levels of trustworthiness. Careful pretesting assured that the manipulations did not influence ratings of source attractiveness. The results supported the hypotheses developed from attribution theory that sources perceived to have external reasons for making a persuasive argument will be discounted. In addition, the results revealed that expert sources influenced perceptions of the product's qualities.
[ to cite ]:
Joshua L. Wiener and John C. Mowen (1986) ,"Source Credibility: on the Independent Effects of Trust and Expertise", in NA - Advances in Consumer Research Volume 13, eds. Richard J. Lutz, Provo, UT : Association for Consumer Research, Pages: 306-310.

Advances in Consumer Research Volume 13, 1986      Pages 306-310

SOURCE CREDIBILITY: ON THE INDEPENDENT EFFECTS OF TRUST AND EXPERTISE

Joshua L. Wiener, Oklahoma State University

John C. Mowen, Oklahoma State University

ABSTRACT -

Marketing researchers and practitioners have tended to assume that a more trustworthy source is more credible than a less trustworthy source. However, previous studies have either confounded trustworthiness with expertise and/or attractiveness, or found that trustworthiness did not influence persuasion. In the 2 x 3 between subjects factorial experiment three levels of source expertise were crossed by two levels of trustworthiness. Careful pretesting assured that the manipulations did not influence ratings of source attractiveness. The results supported the hypotheses developed from attribution theory that sources perceived to have external reasons for making a persuasive argument will be discounted. In addition, the results revealed that expert sources influenced perceptions of the product's qualities.

INTRODUCTION

Marketers have a long held belief that a source with superior intrinsic characteristics is more persuasive. Examples abound of the desire of companies to use intrinsically positive sources and to avoid sources which are perceived as biased abound. National advertisers pay vast amounts to famous endorsers, such as sports stars, while local advertisers cite favorable evaluations from local customers. Estimates are that one in three television commercials use endorsements (Business Week 1978). In fact, companies will go to great lengths to have their products associated with positive sources. For example, the Regina Co. argued in a Federal Appeals court that it had the right to cite in its advertisements a favorable Consumer Union evaluation of its Electrikbroom Powerteam. In other instances companies have tried to avoid having sources, which might be perceived as biased, associated with a product or idea. Thus, when dietary manufacturers sponsored advertisements opposing the banning of saccharin, the advertisements were attributed to a third party, The Calorie Council [Sturdivant 1981] Similarly, when the natural gas lobby initiated its direct mail campaign to create grass-roots support for deregulation, the pamphlets were attributed to "The Alliance for Energy Security.

Early research supported the intuitive belief that positive sources enhance persuasion. Generally, the methodology of the studies assessed the immediate attitude change, or the extent of agreement with a message, induced by an intrinsically superior source in comparison to that induced by an intrinsically inferior one. With one critical exception, these studies found that greater immediate agreement or attitude change was induced by the superior source (McGuire 1968). From these studies emerged the construct of source credibility. A credible source was one who had such intrinsic attributes as trustworthiness, expertness, and attractiveness [e.g., Dholakia and Sternthal 1977, Harmon and Coney 1982].

However, a fundamental problem exists in the way in which both marketing practitioners and marketing researchers conceptualize the effects of source credibility. The presumption that a source is more credible if he or she is more "trustworthy and/or expert" [e.g., Dholakia and Sternthal 1977] is not supported by experimental studies. In their reviews of the source credibility literature, both Anderson & Clevenger (1963) and McGuire (1968) concluded that strong support exists for attractiveness and expertise as components of source credibility, but none for trustworthiness. McGuire argued that the belief that a more trustworthy source is more persuasive is an example of "bubbapsychology" (i.e., a compendium of psychological principles that one's grandmother knew but which aren't so., p. 183)

The issue of whether or not trustworthiness influences persuasion is of theoretical importance. Sternthal, Phillips, and Dholakia (1978) proposed that source effects could be understood by a framework which integrates cognitive response ideas with attribution theory. Attribution theory predicts that a claim made by a biased source will be discounted (Kelly 1973). In justifying their framework, Sternthal, Phillips, and Dholakia (1978) cited the prototypical example of a biased source's claim being discounted. If McGuire is right, Sternthal et al. (1978) are wrong.

The issue of whether trustworthiness contributes to persuasion also has pragmatic significance. The creation of a seemingly trustworthy source can be costly, as well as unethical. One can argue that it is unethical for a company to create in a self-serving manner a seemingLy "trustworthy" organization to act as a "false-front" for the presentation of ideas. Costs can arise if the company is caught in its duplicity. For example, the natural gas lobby created a false front, the citizens' alliance, to support deregulation of the industry. The major news story produced in response to the creation of the citizen alliance was that the organization was comprised of the industry's employees.

The primary purpose of this paper is to further investigate the relationship between the constructs of trust and credibility. Secondarily it investigates expertise as a component of the construct of source credibility. It first reviews the literature on source credibility for evidence of the relationship. Next, it presents an experiment in which source trust and expertise are orthogonally manipulated while holding the attractiveness of the source constant.

SOURCE CREDIBILITY AND ATTITUDE CHANGE

The numerous empirical studies of the relationship between source credibility and attitude change were reviewed by Anderson and Clevenger (1961), McGuire (1968), and Sternthal, et al. (1978). The first two reviews focused upon the issue of whether or not a source's intrinsic characteristics affect the persuasiveness of a message. The reviews reached a common conclusion - a source that was more expert and/or more attractive was more persuasive. The review by Sternthal et al. (1978) focused upon the issue of what situational factors interacted with source credibility to enhance, vitiate, or reverse the persuasiveness of a more credible source. They reached the conclusion that trustworthiness is one of the components of credibility. What is the evidence supportive of Sternthal et al.'s presumption that trustworthiness is a component of source credibility?

Table 1 lists the studies cited by Sternthal et al. (1978) and Dholakia and Sternthal (1977) to support their contention that more trustworthy and/or expert sources induce more immediate attitude change than do sources having less of these attributes. Note that in most of the studies source credibility was manipulated by creating a highly attractive source (e.g., President Eisenhower or genius doctor) versus a high negative source (e.g., Nazi leader or convicted quack). The use of such highly attractive or negative sources tends to confound the concepts of trustworthiness and expertise, which are purported to form the construct of credibility.

TABLE 1

EXAMPLES OF STUDIES MANIPULATING THE "SOURCE CREDIBILITY" CONSTRUCT

METHODOLOGY

The study employed a 2 x 3 between subjects factorial design. Thus, two levels of the trustworthiness independent variable were crossed by three levels of source expertise.

The trustworthiness manipulation was based upon Kelly's (1973) augmenting-discounting ideas. In the study subjects received messages from a mechanic concerning the mechanical condition and value of an automobile. In the high trust conditions, subjects learned that a car mechanic had no relation to a nearby automobile dealership. In the low trust conditions the mechanic was described as part owner of the automobile dealership. Hence, trustworthiness was equated in this study with intrinsic self-interest. Importantly, no normative statements, such as honest or dishonest, were used to describe the source. It was anticipated that in the tow trust case, the consumer should discount the message and place less trust in it.

Expertise was manipulated by altering the individual's training, experience, and certification. Certification was by the National Association of Mechanics. The salient differences are depicted by Table 2.

Two major dependent variables were used: the subjects' perceptions of the automobile's mechanical quality and of its value. Each dependent variable was measured via a seven-point semantic differential scale.

Hypotheses

Two major hypotheses were formulated.

H1: Subjects will reveal higher agreement with the claims of the mechanic regarding the car's mechanical condition and value in high trust than in low trust conditions.

H2: Subjects will reveal higher agreement with the claims of the mechanic regarding the car's mechanical condition and value in high expertise than in the moderate expertise conditions and more agreement in the moderate than the low expertise conditions.

Hypothesis 1 was based upon Kelly's (1973) discounting principle and is consistent with the arguments of Sternthal, Phillips, and Dholakia (1978). Hypothesis 2 is also derived from the literature on source effects. As various reviewers have noted, highly expert sources have proven to be more persuasive than less expert sources. The contribution of the present study is that the two variables (trust and expertise) are manipulated independently.

TABLE 2

THE EXPERTISE MANIPULATION

Procedure

In order to insure the construct validity of the manipulations, the study was divided into two phases--a pretest and a main test. In both the pretest and main test, all subjects were presented with a common situation. The situation was described to the subjects as follows:

You have found a 1979 mit-size American sedan, with am/fm radio but no air-conditioning, at an auto dealership. You are thinking of buying it. You decide that a mechanic should inspect it before you buy it. You take it to a large auto repair center in a nearby mall so that you can eat while they inspect it.

You are especially interested in learning how long you can expect major components, such as the transmission, to last before they will need to be replaced or repaired .

The man who inspects the car is in his mid-twenties, polite, friendly, and well-spoken. He says that he is married, has a daughter, and that his family is going to Yellowstone for their vacation next month.

After high school he joined the Army, received an award for bravery, and several promotions. After being discharged he went to work for his uncle who owns the garage. The mechanic is part owner of the garage. Since the guy who normally does the inspections is sick, he is doing it.

The brief common biography was included in order to create a generally favorable view of the source of information. The goal was to create a source who would be equally attractive co respondents regardless of the trustworthiness or expertise condition. The use of a common biographical sketch is a distinguishing feature of the methodology. For example, Wart and McGinnies (1980) explicitly described the trustworthy source as "honest, sincere, and trustworthy; whereas the untrustworthy source was described as 'devious, calculating, and sympathetic to the Nazi party.

In all cases the mechanic made the assertion that:

Because it is in excellent mechanical condition I would say it is worth a lot more than the average automobile of its year and make. In fact I'd say it's worth 10% more than the blue book value.

The automobile example was used because most undergraduates are sufficiently familiar with automobiles to understand questions, but are not so familiar as to hold strong prior beliefs. To check the latter point, data on sex, experience, (buying) selling used cars, and experience with mechanics was collected from respondents.

The manipulations were pretested on a class of 40 undergraduate business students. The manipulations of trust and expertise were pretested independently. The trustworthiness manipulation was assessed via three, seven point semantic differential scales: trustworthy/not trustworthy, partial/not partial and biased/not biased. The expertise manipulation was measured by the scales used previously by Harmon and Coney (1982) and Dholakia and Sternthal (1977), i.e., expert/not expert, trained/not trained, and experienced/not experienced. The two manipulations were successful, i.e., at the .05 a level there was a significant difference between the scale scores of the treatment groups. The scales revealed good internal reliability; Cronbach 3 scores were .73 for trustworthiness and .89 for expertise. The pretest revealed that the manipulations of inherent self-interest did strongly influence perceptions of trustworthiness. In addition, the variations in source training, experience and certification did influence PercePtions of expertise.

RESULTS

In the main study, students were randomly assigned to one of the 6 (2 trustworthiness x 3 expertise conditions). In all conditions the common situation and biography were used. After reading the material they indicated their level of agreement with the source's claim (dependent variables), and then evaluated the source (the manipulation checks).

In the main study, tests were first run to identify any halo or attractiveness effects. Optimally, no significant effects for attractiveness should be found across any of the manipulations. Attractiveness was measured by using the 3 item scale used by Harmon and Coney (1982) and Dholakia and Sternthal (1977); the items are attractive/ not attractive, dynamic/not dynamic, aggressive/not aggressive. The Cronbach a levels were .71 for attractiveness, .76 for trustworthiness, and .84 for expertise. Using a .05 a level criterion, attractiveness was not found to be influenced by variations in expertise or trustworthiness. In addition, the expertise manipulations did not influence perceived trustworthiness and the trustworthiness manipulations did not influence perceived expertise.

Tables 3 and 4 present the means and the summary table for the two-way analyses of variance used to analyze the mechanical quality and automobile value claims. The semantic differential scales used to measure persuasions were anchored by excellent mechanical quality/not excellent mechanical quality, and worth a lot more/not worth a lot more. For the dependent variable of mechanical quality, significant main effects occurred for expertise (p < .01), and trust (p < .03). A Scheffe test of expertise found that (at the .05 level) only the high and low expertise manipulations were associated with different agreement levels. The size of the trustworthiness and expertise effects were measured by omega squares. The variance accounted for was in line with that typically found in social psychological research. It is interesting to note that the expertise measure accounts for more than twice the variance trustworthiness measure (.15 vs. .06). For the dependent variable of automobile value, significant effects were found only across levels of trust (p < .03). A significant main effect for value was not found across levels of the expertise independent variable.

DISCUSSION

The results for both the major dependent variables (mechanical quality and automobile value) supported the attribution theory component of Sternthal et al.'s (1978) source credibility framework. In conditions in which one would expect a source to have a bias (i.e., low trust conditions in which external pressures exist to elicit a particular message), the results revealed that subjects had lower perceptions of the automobile's value.

TABLE 3

KEAN AGREEMENT SCORES FOR DEPENDENT VARIABLES OF MECHANICAL QUALITY AND AUTOMOBILE VALUE

The finding that variations in trust influence consumer reactions to messages is important for the understanding of source effects. McGuire (1968) has argued against trustworthiness as a factor influencing persuasiveness. Indeed, the effects of trustworthiness had not been unambiguously demonstrated in the literature heretofore. Previous work tended to confound trust with the expertise of the source.

TABLE 4

ANOVA RESULTS

In two previous studies trustworthiness and expertise were investigated. In the Hovland and Mandell (1952) study, variations in trustworthiness were not found to affect persuasion. A possible explanation for their failure to find trustworthiness effects was the overwhelming effects of source expertise. In the present study, a comparison of the mean agreement scores in the highly expert case (see Table 3), supports such a conJecture. The effects of the trustworthiness variable showed signs of being mitigated in the high expertise cases. In other words, the situation may be similar to ones in which evidence is used to overwhelm the source effects (Hunt 1972, Hendrick and Schaffer 1970).

In the second study investigating trustworthiness and expertise, McGinnies and Ward (1980) confounded trust worthiness with attractiveness. In the present study, efforts were mate to hold attractiveness constant while manipulating trustworthiness. Results revealed that these efforts were successful. An attractiveness scale was not influenced by levels of trustworthiness. The significant effects found for trustworthiness, therefore, supported the arguments of Sternthal et al. (1978) and removed the construct of trustworthiness from the realm of "bubbapsychology" (McGuire 1968).

The study also found that the level of expertise of the source strongly influenced perceptions of the auto's mechanical quality. McGinnies and Ward (1980) previously found no evidence for the increased persuasiveness of expert sources. Possibly their extremely strong manipulation of attractiveness/trustworthiness overwhelmed the effects of variations in expertise so that subjects failed to attend to expertise differences. Note, however, that variations in expertise did not influence perceptions of the automobile's value. This outcome may have occurred because perceptions of value were not inferred directly from knowledge of the source's expertise. Instead, expertise may influence value through its influence on the mechanical value perception. Therefore, the likelihood of expertise affecting value perceptions is lowered because of the requirement of a two-step inference process. In this two-step process, observers may first evaluate expertise. Second, they may form beliefs about the mechanical quality of the car from the message and the expertise. After estimating the car's mechanical quality, they would then assign some value to the car. The additional cognitive effort required to engage in such complex reasoning lowers the likelihood that subjects would engage in such a two-step process resulting in variations in expertise failing to influence the value of the car.

From a managerial perspective, the results are straight forwardly applied. When the receiver is being influenced by peripheral cues, such as source characteristics, a message from a source of information, who is perceived to be biased, has less impact than a message from a source not perceived to be biased independent of the source's attractiveness and expertise. Similarly, the greater the perceived expertise of the source, the greater the impact of a message upon technical, but not product value, beliefs independent of the source's attractiveness and perceived level of bias. Based upon these findings, advertisers, public relations officers, and sales managers should care fully consider the extent which prospective target audiences perceive the level of bias and expertise of sources of information.

REFERENCES

Anderson, K. and T. Clevenger (1963), A Summary of Experimental Research in Ethos, Speech Monographs, 30, 59-78.

Business Week (1978), The Big New Celebrity Boom," May 22.

Dholakia, R. and B. Sternthal (1977), Highly Credible Sources: Persuasive Facilitators or Persuasive Liabilities?" Journal of Consumer Research, 3, 223-232.

Harmon, R. and H. Coney (1982), The Persuasive Effects of Source Credibility in Buy and Lease Situations, Journal of Marketing Research, 19, 255-260.

Hendrick, C. and D. Shaffer (1970), The Effects of Arousal and Credibility on Learning and Persuasion, Psychometric Science, 20, 241-243.

Hovland, C. and W. Mandell (1952), An Experimental Comparison of Conclusion-Drawing by the Communicator and by the Audience, Journal of Abnormal and Social Psychology, 47, 581-588.

Hovland, C. and W. Weiss (1951), The Influence of Source Credibility on Communication Effectiveness, Public Opinion Quarterly, 15, 635-650.

Hunt, K. (1972), Source Effects, Message Effects, and General Effects, Proceedings 3rd Annual Conference, Association for Consumer Research, Chicago.

Johnson, H., J. Torcivia, and M. Poprick (1968), "Effects of Source Credibility on the Relationship between Authoritarianism and Attitude Change," Journal of Personality and Social Psychology, 9, 179-183.

Kelly, H. (1973), "The Process of Causal Attribution," American Psychologist, 28, 107-128.

Kelman, H. and C. Hovland (1953), "Reinstatement of the Communicator in Delayed Measurement of Opinion Change, Journal of Abnormal and Social Psychology, 48, 327-335.

McGinnes, E. and C. Ward (1980), "Better Liked Than Right: Trustworthiness and Expertise in Credibility,-Personality and Social Psychology Bulletin, 6, 467-472.

McGuire, W. J. (1968), "The Nature of Attitudes and Attitude Change, In G. Lintzey and E. Aronson (Eds.) Handbook of Social Psychology. Reading, MA: Addison-Wesley.

Miller, G. and J. Baseheart (1969), Source Trustworthiness, Opinionated Statements, and Response to Persuasive Communication, Speech Monographs, 36, 1-7.

Schulman, G. and C. Worrall (1970), "Salience Patterns, Source Credibility, and the Sleeper Effect. Public Opinion Quarterly, 34, 371-382.

Sternthal, B., L. Phillips, and E. Dholakia (1978), "The Persuasive Effects of Source Credibility: A Situational Analysis," Public Opinion Quarterly, 285-314.

Sturdivant, F. (1981), Business and Society, Homewood, IL: Richard D. Irwin, Inc.

Warren, I. (1969), "The Effects of Credibility in Sources of Testimony on Audience Attitudes Toward Speaker and Message," Speech Monographs, 36, 456-458.

Whittaker, J. and R. Meade (1968), "Retention of Opinion Change as a Function of Differential Source Credibility," International Journal of Psychology, 3, 103-108.

----------------------------------------