The Validity of Using Consumer Input to Choose Advertising Spokesmen

David W. Finn, Texas Tech University
ABSTRACT - A commonly mentioned trait of advertising spokesmen is "credibility", and consumer views about the relative credibility of alternatives are often sought before endorser choices are made. This paper casts strong doubt on the ability of consumers to rate potential spokesmen on credibility dimensions and questions whether or not "credibility" is an important attribute of product spokesmen.
[ to cite ]:
David W. Finn (1980) ,"The Validity of Using Consumer Input to Choose Advertising Spokesmen", in NA - Advances in Consumer Research Volume 07, eds. Jerry C. Olson, Ann Abor, MI : Association for Consumer Research, Pages: 776-789.

Advances in Consumer Research Volume 7, 1980     Pages 776-789

THE VALIDITY OF USING CONSUMER INPUT TO CHOOSE ADVERTISING SPOKESMEN

David W. Finn, Texas Tech University

ABSTRACT -

A commonly mentioned trait of advertising spokesmen is "credibility", and consumer views about the relative credibility of alternatives are often sought before endorser choices are made. This paper casts strong doubt on the ability of consumers to rate potential spokesmen on credibility dimensions and questions whether or not "credibility" is an important attribute of product spokesmen.

INTRODUCTION

The situation explored in this paper concerns using consumer input to choose an advertising spokesman. Specifically, it addresses the issue of whether or not consumers can tell us who good advertising spokesmen might be. If a decision maker is faced with a decision about choosing an appropriate advertising spokesman for his product, he must have some criteria on which to compare the potential spokesmen. Some commonly mentioned endorser attributes include; attractiveness, likeability, similarity to the audience, recognizability, and "credibility'' -- the perceived trustworthiness and expertise with regard to the product. Each of these source variables is expected to have a unique effect on audience acceptance of the message.

According to Kelman (1961) increased source "credibility" (higher trust and higher competence) leads to internalization of the message arguments and more lasting favorable attitudes, while attractiveness and likeability lead to identification with the source and imitation. In terms of the "hierarchy of effects" model of the consumer adoption process (Palda 1966), it can be argued that likable, well known spokesmen should have their strongest impact on the "awareness" and "purchase" stages through 1) their ability to attract attention, and 2) the audience's tendency to identify with and imitate them, respectively. Trustworthy, expert spokesmen, on the other hand, will most likely have their strongest impact on the "liking", "preference", and "conviction" stages through the internalization process whereby consumers generate attitudes toward the topic of the message (the advertised brand) rather than toward the source. These arguments make it clear that there are apparent trade-offs to be made in selecting between well-known celebrity endorsers and less well-known, but "credible" spokesmen. It is probably for these and related reasons that endorser "credibility" is such a common phrase in advertising. A trustworthy, competent spokesman is expected to have a bigger impact on liking and preference and to be more believable than one who is less trustworthy and less competent. Good decisions require that the decision maker know the perceived trustworthiness and expertness of his alternative endorsers.

There are two broad methods for the decision maker to acquire this information. First, he can request completed advertisements with several alternative spokesmen, run them in different markets, question the consumers in each market about the perceived "credibility" of a particular spokesman and compare. Second, he can perform some pretest among consumers to determine the relative credibility of potential spokesmen. The latter method is far less expensive and requires no prior contractual commitment with any of the spokesmen. In fact, an article in a popular magazine a few years ago (Time, 1973) hinted that this procedure was common. In this article it was reported that A. R. Nelson Research, Inc. had conducted a study to determine what sports personalities are more likely to be trusted by consumers for product endorsements. The research company had polled potential buyers of various products and offered them a list of sports personalities. The personalities were somehow rated (apparently on at least a trust dimension). The conclusions of the study included the statement "an athlete's potential success as an endorser depends not on his skill or fame but on his 'likeability' by the public. And what the public appears to like is the quiet, comfortable, old shoe personalities -- not the abrasive or swinging types" (p.8). The conclusions were based on the findings that less flamboyant personalities (e.g., Stan Musial, Mickey Mantle, and Yogi Berra) received the highest trust ratings as product spokesmen by potential consumers. The actual purpose of A. R. Nelson's research is unknown by this author, but it is interesting to note that not long after this, a likable, comfortable, quiet, old shoe sports personality became a spokesman for coffee makers.

Pretests of this type rely on the assumption that consumer perceptions of the "credibility" of spokesmen are the same in a pretest condition as in an actual advertising situation. If consumers for some reason rate a potential spokesman differently in a pretest than in an actual advertisement, we would wonder whether or not the best spokesman choice had been made. That is, if "credibility'' differences between various spokesmen are only present in a pre-test, choosing endorsers on this basis would be an unwise strategy. In this case, a spokesman choice based on factors such as recognition, flamboyance, or similarity to the audience would be strategically sounder. This paper explores differences in consumer perceptions of endorser credibility (perceived expertness and trustworthiness) in pre- and post-advertising conditions.

METHOD AND RESULTS

Pretest

One week prior to the pretest, forty seven bipolar adjectives were elicited from the subject pool (students). These bipolar adjectives were offered by them as being non-visual characteristics upon which advertising spokesmen differed. These bipolar adjectives were collected for two purposes: first, to develop scales for measuring spokesmen trust and competence; second, to search for spokesmen who differed in ratings on these discovered scales.

As a first step in developing scales to measure spokesmen trust and competence, a list of sixteen potential advertising spokesmen was generated. Four hundred and ninety-four different subjects from the same subject pool rated these potential endorsers on the forty-seven bipolar scales. Each subject rated four different sources as spokesmen in an advertisement for toothpaste. Toothpaste was chosen because of considerations concerning the subject pool. The larger study, of which this is a small part, required the use of a product category with which the subjects had use experience and for which they were the principle decision makers. Toothpaste satisfied these requirements. An advertising context was used in the pretest to define a level of specificity for the ratings, thereby avoiding generalized ratings of the endorsers.

The dimensions of source credibility have been studied by many authors. McCroskey (1966), Berlo, Lemert, and Mertz (1969-1970), and Markham (1968), all used factor analytical techniques in searching for these dimensions and in developing scales to tap them. The most consistent factors to appear in these studies were Trustworthiness (sometimes labeled "character" or "safety") and Expertness (also called "competence", "authoritativeness'', or "qualification"). The "dynamism" or "theatricalness'' of the source is also a recognized factor, but this dimension appears to be much less stable.

In keeping with this standard methodology, the subjects' ratings were subjected to a factor analysis using the principle factors technique followed by an orthogonal rotation to simple structure with the varimax criterion. Eigenvalues of 1.0 or greater was the determining cutoff point for the interpretation of rotated factors. The results of this factor analysis are summarized in Table 1.

TABLE 1

FACTOR LOADINGS ON 16 PRE-TEST SPOKESMEN

The strongest factor that emerged could be easily labeled "trust" (as expected). The adjective pairs loading most heavily and purely on this factor were trustworthy-untrustworthy, honest-dishonest, reliable-unreliable, fair-unfair, truthful-deceiving, and helpful-unhelpful. These six adjective pairs were chosen as a measure of "trust". The second factor that emerged is obviously the expected expertness dimension. The adjective pairs loading most heavily and purely on this factor were skilled-unskilled, trained-untrained, experienced-inexperienced, and professional-unprofessional. These four adjective pairs were chosen as a measure of "competence". "likeability" and "dynamism" type dimensions also emerged and are not included in Table 1 because of this study's focus on trustworthiness and competence. A "trust" score and a "competence'' score were then computed for each of the sixteen potential spokesmen by summing his/her scores on the associated adjective pairs. The scales were coded in such a way that low numbers corresponded to high levels of trust and competence. For example, the "highest" possible trust rating was 6, while the "lowest" possible trust rating was 42. Competence scores had a potential range from 4 to 28.

In simulating a management decision, if we are hoping to choose the most expert and/or trustworthy spokesman, then the spokesman with the highest score on the desired dimension would be chosen. However, in this research it was desirable to choose spokesmen who differed in perceived trust and/or competence in the pretest and compare them to an advertisement. In the pretests, two potential spokesmen were identified as being perceived as relatively high in trust. They were "a practicing dentist in Cleveland, Ohio" and Robert Young. (It should be noted that this data was collected when Robert Young was starring in the television series "Marcus Welby"). Two other potential spokesmen, Johnny Carson and "a Lever Brothers chemist" were identified as being perceived relatively low in trust. See Table 2 for the Pretest Trust Scores.

TABLE 2

TRUST RATINGS--A PRE- AND POST-ADVERTISING COMPARISON

Additionally, the Cleveland dentist and Lever Brothers chemist were perceived in the pretests to be of relatively high competence for a toothpaste advertisement, while Johnny Carson and Robert Young were perceived as having relatively little expertise in this context. (See Table 3).

At this point in our spokesman choice strategy it would seem that Robert Young would be a wise choice if we wished to hire a celebrity spokesman who was perceived as high in trust. On the other hand, if competence is the dimension we wish to emphasize, the Cleveland dentist or the Lever Brothers chemist would be wise choices. The next part of this paper explores the question of whether or not these would indeed be wise choices.

TABLE 3

COMPETENCE RATINGS -- A PRE- AND POST-ADVERTISING COMPARISON

Post Tests

To simulate an actual advertising situation, 260 other subjects from the same subject pool were told that the experimenter had advertising information available from a current test market for a new fluoride toothpaste by Lever Brothers. The verbal portion of that advertisement was said to be available. They were to read it and answer some questions about it. Table 4 is a representation of the advertisement.

TABLE 4

THE ADVERTISING MESSAGE

The advertisement was attributed to one of the four spokesmen. Each subject read only one ad attributed to only one spokesman. The distribution of spokesmen to subjects was random. Among the questions to be answered were fourteen bipolar seven interval scales for rating the spokesmen. Six of these scales were the same six as the trust scale developed earlier, four of the scales were the same as the competence scale developed earlier. Four other bipolar scales were included as fillers. The measures of interest were changes in trust and competence ratings for each spokesman.

To detect changes in perceived trust and competence, the scores from the pretests were compared to the scores from the advertisement situation for each of the spokesmen. The results are displayed in Tables 2 and 3. It is clear from the Tables that the perceived trust of the "high" trust spokesmen was significantly and meaningfully different in the two conditions--the low trust spokesmen remained low. While one of the low trust spokesmen had significant change, the magnitude of the change does not compare with the magnitude of the change of the high trust spokesmen. A similar phenomenon is evident on the competence scores. Those spokesmen who were perceived in the pretests as having high competence for a toothpaste advertisement (the Cleveland dentist and the Lever Brothers chemist) lost their competence ratings when they were actually put into an ad.

DISCUSSION

Explanations for the observed results can be broken down into three general categories; 1) demand characteristics, 2) the impact of advertising on source credibility, and 3) differences in consumer role playing between the pretest and the advertisement.

If subjects became bored or felt somehow coerced into answering the post-advertising questionnaire, they would tend either to put marks at random places on the bipolar scales or to answer on one side of the scales. Demand characteristics of this type would lead to a finding of no difference between ratings of trust and competence. In fact, we would find that the measuring instruments were not actually tapping trust and competence dimensions. To test for this possibility a factor analysis was performed on the post-advertisement data. If demand characteristics of this type were operating we would expect to see no clearly defined factors. The results of this factor analysis are reproduced as Table 5. It is clear from Table 5 that the trust factor corresponding to the scales trustworthy/untrustworthy, helpful/unhelpful, reliable/unreliable, honest/dishonest, fair/unfair,

TABLE 5

FACTOR ANALYSIS OF POST-TEST DATA

and truthful/deceiving did emerge. The second factor is again the competence dimension discovered earlier. The fact that the same dimensions emerged increases our confidence that the subjects were treating the questionnaire in a serious vein.

The second and third categories of explanation cannot be as easily dismissed, and further research is needed to measure the relative impact of each. Without the pretest ratings, we could logically argue that anyone in an advertisement is getting paid and has persuasive intent, and this fact will influence the perceived trust ratings of any spokesman. However, the pretest ratings reduce the validity of this argument, since even then the sources were rated in an advertisement for toothpaste. Any advertisement can be thought of as consisting of a combination of sources: the company itself may vary along dimensions of trust or competence; the advertising vehicle may vary similarly; or the product spokesman may be perceived as high or low in these characteristics. Decisions made using the pretest method implicitly assume that these individual characteristics are additive. That is, by holding the company and the vehicle constant we can improve the overall trust of the communication by improving the trust of the spokesman. The actual advertising situation hints that the perceived trust and competence of these various sources influence each other in a more complex fashion.

Finally, differences in trust and competence ratings could be due to differences in role playing by the two groups of subjects. For example, in the pretest, respondents may have thought of themselves as being in a decision making situation. That is, they may have been trying to decide if Robert Young (or any of the other spokesmen) would be a trustworthy toothpaste spokesman for other people. In the actual advertising situation they were rating the spokesman's trust and competence as perceived by themselves. If this difference in role perception existed, it would explain the difference in ratings and cast doubt on the validity of other pretest results. We would have to ask the question, "do pretests of this type place the respondent in a role other than that of consumer?"

CONCLUSION

As alluded in the introduction of this paper, the actual process that advertisers go through in choosing product spokesmen is not common knowledge. Findings and beliefs of particular agencies and manufacturers are "trade secrets'' that give individual companies apparent competitive advantages. A hint at the procedures was revealed in the A. R. Nelson study. This paper used a similar pre-advertising method to choose trustworthy and competent spokesmen and went one step further by choosing contrasting sources. A comparison of the spokesmen in an advertisement revealed no meaningful differences in trust or competence among the chosen spokesmen. The advertising context may have influenced potential trust and competence differences. The A. R. Nelson pretest found large differences between Stan Musial and Mohammed Ali, but the results of this paper suggest a possibility that these differences may not have existed in an actual advertisement and that a less than optimal choice may have been made.

If flamboyant, well-known personalities and less well-known personalities are perceived as equally low in trust and competence, they will not differ in their influence on the attitudinal dimensions of the product adoption process nor be more believable, and the more flamboyant, well-known personalities (despite their low pre-advertisement levels of competence and/or trust) are clearly more valuable because of their expected impact on attention. The findings of this paper move us in the direction of an overall conclusion that "credibility" may not be an important attribute of product spokesmen. Further research is clearly needed before the results of this paper can be totally accepted.

REFERENCES

Berlo, D. K., Lemert, J. B. and Mertz, R. J. (1969-1970), "Dimensions for Evaluating the Acceptability of Message Sources," Public Opinion Quarterly, 33, (Winter), 563-76.

Kelman, H. C. (1961), "Process of Opinion Change," Public Opinion Quarterly, 25, (Spring), 57-78.

Markham, D. (1968), "Dimensions of Source Credibility of Television Newscasters," Journal of Communication, 18, 57-64.

McCroskey, J.C. (1966), "Scales for the Measurement of Ethos," Speech Monographs, 33, 65-72.

Palda, K. S. (1966), "The Hypothesis of a Hierarchy of Effects: A Partial Evaluation," Journal of Marketing Research, 3, 13-24.

"Who Do You Trust?" Time, (December 31, 1973), 8.

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