11
Journal of Consumer Research, Inc. Highly Credible Sources: Persuasive Facilitators or Persuasive Liabilities? Author(s): Ruby Roy Dholakia and Brian Sternthal Source: Journal of Consumer Research, Vol. 3, No. 4 (Mar., 1977), pp. 223-232 Published by: The University of Chicago Press Stable URL: http://www.jstor.org/stable/2489608 . Accessed: 20/09/2013 15:22 Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp . 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 [email protected]. . The University of Chicago Press and Journal of Consumer Research, Inc. are collaborating with JSTOR to digitize, preserve and extend access to Journal of Consumer Research. http://www.jstor.org This content downloaded from 192.236.36.29 on Fri, 20 Sep 2013 15:22:08 PM All use subject to JSTOR Terms and Conditions

Highly Credible Sources: Persuasive Facilitators or Persuasive Liabilities?

Embed Size (px)

Citation preview

Page 1: Highly Credible Sources: Persuasive Facilitators or Persuasive Liabilities?

Journal of Consumer Research, Inc.

Highly Credible Sources: Persuasive Facilitators or Persuasive Liabilities?Author(s): Ruby Roy Dholakia and Brian SternthalSource: Journal of Consumer Research, Vol. 3, No. 4 (Mar., 1977), pp. 223-232Published by: The University of Chicago PressStable URL: http://www.jstor.org/stable/2489608 .

Accessed: 20/09/2013 15:22

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

.

The University of Chicago Press and Journal of Consumer Research, Inc. are collaborating with JSTOR todigitize, preserve and extend access to Journal of Consumer Research.

http://www.jstor.org

This content downloaded from 192.236.36.29 on Fri, 20 Sep 2013 15:22:08 PMAll use subject to JSTOR Terms and Conditions

Page 2: Highly Credible Sources: Persuasive Facilitators or Persuasive Liabilities?

Highly Credible Sources: Persuasive

Facilitators or Persuasive Liabilities?

RUBY ROY DHOLAKIA BRIAN STERNTHAL*

A low credibility source induced a more positive attitude toward his ad- vocacy than did a highly credible source when message recipients' own behavior served as a cue for determining their attitudes. In contrast, when the behavioral cue was absent, a highly credible source did not have an adverse effect on individuals' attitudes or behavior. These findings are interpreted in terms of self-perception theory and cognitive response analysis.

Communication practitioners appear to share the belief that a communicator's attributes of character have a significant impact on the persuasiveness of an appeal. Thus, advertisers select consumers who typify the target audience to provide testimonials, manufac- turers seek seals of approval from independent testing agencies, and politicians marshal the support of nationally prominent dignitaries to endorse their platforms.

Support for this premise derives from experimental research dealing with the persuasive effect of source credibility. Characteristic of this research is the determination of the attitude change induced by systematic manipulation of one or more of the dimensions of credibility-trustworthiness, expertise, and attractiveness. Using this paradigm it has been found that highly trustworthy and/or expert sources induce more immediate attitude change than do sources having less of these attributes (Hovland and Weiss, 1951; Johnson, Torcivia, and Poprick, 1968; Kelman and Hovland, 1953; Miller and Baseheart, 1969; Schulman and Worrall, 1970; Warren, 1969; Watts and McGuire, 1964; Whittaker and Meade, 1968; but not Hovland and Mandell, 1952). Furthermore, when the attractiveness dimension of source has been varied, either by manipulating similarities between the source and message recipient in attitudes or group member- ship (e.g., ethnic, racial, religious membership), it has been observed that increasing the communicator's

attractiveness enhances positive attitude change, pro- vided that the similarities have some bearing on the issue in question (see Simons, Berkowitz, and Moyer [1970] for a review of this literature).

In contrast to the main effect studies, which strongly suggest the efficacy of using highly credible sources to deliver persuasive appeals, ?nvestigations examining the interaction between source credibility and other variables indicate that under certain circum- stances credibility does not have a significant per- suasive effect. More specifically, it has been reported that highly credible communicators induce about the same amount of attitude change as low credibility sources when (1) the message presented is incon- gruous with the source's best interest (Eagly and Chaiken, 1975; Walster, Aronson, and Abrahams, 1966), or the source justifies the position advocated with arguments that are unfamiliar to message re- cipients (McCroskey, 1969); (2) the audience is highly authoritarian (Johnson, Torcivia, and Poprick, 1968), is highly involved in the communication issue (John- son and Izzett, 1972; Johnson and Scileppi, 1969; Rhine and Severance, 1970), or has an internal locus of control (Ritchie and Phares, 1969); (3) the situation is threatening (Sigall and Helmreich, 1969); (4) the time delay between the presentation of the communication and the administration of the posttest is relatively long (see Capon and Hulbert, 1973).

Thus, in main effect studies it has been found that highly credible sources are more persuasive than ones of low credibility. On the other hand, in investigations of the joint effects of source credibility and certain message, audience, and situational characteristics as well as the passage of time, it has been observed that low credibility communicators do not constitute a

* Ruby Roy Dholakia is Assistant Professor of Marketing, Indian Institute of Management, Calcutta. Brian Stemthal is Assistant Professor of Marketing, Northwestern University, Evanston, Illinois 60201. The authors thank Bobby J. Calder and the JCR reviewer for helpful comments on an earlier draft of this paper.

JOURNAL OF CONSUMER RESEARCH . Vol. 3 . March 1977 223

This content downloaded from 192.236.36.29 on Fri, 20 Sep 2013 15:22:08 PMAll use subject to JSTOR Terms and Conditions

Page 3: Highly Credible Sources: Persuasive Facilitators or Persuasive Liabilities?

224 THE JOURNAL OF CONSUMER RESEARCH

persuasive liability. Yet, despite the discrepancy between main effect and interactive source credibility findings, they can both be ordered by cognitive response analysis (Greenwald, 1968). According to this theory, persuasion entails the rehearsal of one's own attitude-relevant thoughts as well as those con- tained in a communication. Message rejection occurs when individuals who are opposed to an issue re- hearse counterarguments in response to a persuasive appeal. Further, it is hypothesized that a highly credible source serves as a cue that inhibits counter- argumentation, while a low credibility source facilitates these responses. Despite the lack of direct evidence for the contention that source credibility mediates cognitive responses, it is congenial to the repeated observation of a source credibility main effect. In addition, it accounts for the nonsignificant source credibility effect reported in interactive studies. For example, when people are highly involved with an issue, they are unlikely to be sensitive to the cues regarding the credibility of the communicator. Thus, no source credibility effect is predicted.

In essence, the cognitive response formulation accounts for the persuasive effects of source credibility in situations where individuals determine their atti- tudinal response to a communication by assessing its causal antecedents. It does not account for the effect of source credibility in situations where individuals' own experience as well as the communication source constitute the bases for their attitudinal judgments. In such situations it has generally been found that low credibility sources are more persuasive than ones of high credibility. Powell (1965) reported that a low credibility source was more persuasive than one of high credibility when compliance with the communica- tor's request was perceived to be voluntary. On the other hand, when compliance was involuntary no source effect emerged. A similar source x choice interaction was observed in experiments where mes- sage recipients' choice of whether or not to expose themselves to a counterattitudinal appeal was system- atically manipulated (Himmelfarb and Arazi, 1974; Jones and Brehm, 1967). Those subjects given a choice regarding exposure were more persuaded by a low than high credibility source, whereas those given no choice were more persuaded by highly credible sources. Eagly and Chaiken (1975) also observed that the perception of choice enhanced the persuasive- ness of a low credibility source, though the source x choice interaction was not significant.

The finding that low credibility sources cause greater persuasion than highly credible sources after individ- uals have voluntarily complied with a request can be explained in terms of self-perception theory (Bem, 1972; Kelley, 1973). According to self-perception theory, making it salient to message recipients that their behavior was not voluntary may cause them to focus on the causal antecedents of the message. As a result, highly credible sources are expected to be

more persuasive than low credibility sources. On the other hand, making their volition salient causes individuals to focus on the causes of their behavior. When source credibility is high, subjects could at- tribute their voluntary compliance either to a positive attitude toward the behavior or to the fact that they were solicited by a highly credible source. Thus, rival explanations for their behavior would be available to the subjects, resulting in a less positive attitude toward the behavior than when the source's credibility is low and therefore not a plausible explanation.

In sum, the source credibility literature indicates that in main effect studies highly credible sources are more persuasive than ones of low credibility, in interactive studies other variables may obviate the credibility effect, and in studies where people's behavior is available as a cue regarding their attitudes, low credibility sources are more persuasive than sources of high credibility. Given these data, it is hypothesized that the influence context is an important determinant of the effect of source credibility. Further, we contend that different models are required to ac- count for the situation-specific source effect. Spe- cifically, when only the source and communication- related cues are available, cognitive response analysis may be employed to predict that highly credible sources will be more persuasive or as persuasive as low credibility communicators. On the other hand, when individuals' own behavior as well as source and message cues are available, self-perception theory can be employed to predict that low credibility sources will be more persuasive than highly credible ones.

The purpose of the investigation reported in this paper is to test the above hypotheses within the context of a single experiment. Although the hypoth- eses appear to order existing data, they are based on the findings of different studies, allowing for the possibility that variables other than the nature of the influence situation account for the disparate source effects findings. To this end, subjects in the present study were asked to read voluntarily a persuasive appeal attributed to either a high or low credibility communicator, and to sign a petition supporting the communication issue. A series of scaled questions were administered to determine subjects' attitudes toward the issue either before or after they had made a decision regarding whether or not they would sign the- petition.

If the hypotheses stated above are correct then the interaction between source credibility and the timing of the behavioral request should be significant (see Figure 1). In the treatment where subjects assess their attitudes prior to responding to the behavioral request, it is expected in accord with cognitive response analysis that highly credible sources will facilitate persuasion and low credibility sources will inhibit it. In contrast, in those treatments where subjects' behavioral response to the request occurs

This content downloaded from 192.236.36.29 on Fri, 20 Sep 2013 15:22:08 PMAll use subject to JSTOR Terms and Conditions

Page 4: Highly Credible Sources: Persuasive Facilitators or Persuasive Liabilities?

HIGHLY CREDIBLE SOURCES 225

32

3 1 0

MEAN 3 X o = LOW CREDIBILITY

ATTITUDE 2 x = HIGH CREDIBILITY 28

27 X

26

TIMING OF ATTITUDE MEASURE RELATIVE TO BEHAVIORAL REQUEST

FIGURE 1 Mean attitude as a function of the timing of

the behavioral request.

before they assess their attitudes, self-perception theory predicts that low credibility sources will be more persuasive than highly credible communicators. For subjects who comply with the request, observing that they complied despite the fact that compliance was advocated by a low credibility source should augment the attribution of behavior to internal causes. Con- versely, if the appeal is attributed to a highly credible source, subjects will be uncertain about the underlying causes of their compliance. It may be attributable either to some personal feeling about the issue (i.e., an internal cause) or to the fact that the message was presented by a highly credible source.

A similar line of reasoning can be used to predict the attitudes of subjects who did not comply with the behavioral request. -Noncomplying subjects exposed to a low credibility communicator are likely to be un- certain whether their failure to comply was due to the low credibility of the source or the fact that they have a negative attitude toward the issue. Conversely, if source credibility is high, noncompliance is more likely to be attributed to subjects' internal disposi- tions; the requested behavior was not performed de- spite the fact that source credibility was high.

This investigation addresses two issues that are of particular concern to those interested in consumer behavior. First, it examines the effect of source

credibility in a situation where individuals have their own behavior as a basis for assessing their attitudes. Although consumers often have had experience with a product or service before being exposed to a persuasive appeal, this situation has not been subject to rigorous investigation in source effect studies. Second, the im- pact of the sources credibility on individuals' be- havior as well as their attitudinal responses are determined. Relatively few investigations have identified the effect of source credibility on behavior (an exception is Brock, 1965), despite the fact that this criterion is of great concern in understanding consumer behavior.

METHOD

Subjects

The 110 participants in the study were recruited from three sections of an introductory graduate course in management. During one of the regular class hours, the students were asked by their class in- structor (who was not the experimenter) to participate in a study on federal legislation. Participation was made voluntary. The sponsors of the study were de- scribed as an external group who had sought the as- sistance of the course instructor to administer the

This content downloaded from 192.236.36.29 on Fri, 20 Sep 2013 15:22:08 PMAll use subject to JSTOR Terms and Conditions

Page 5: Highly Credible Sources: Persuasive Facilitators or Persuasive Liabilities?

226 THE JOURNAL OF CONSUMER RESEARCH

study. All the students agreed to participate. An in- formed consent procedure (Berscheid, Baron, Dernier, and Libman, 1973), administered during the debriefling sessions, revealed that less than five percent of the participants had any objections to the experimental task.

Procedure

Experimental subjects were given booklets con- taining the independent and dependent variables by their class instructor. The experimental task involved reading a one page message dealing with the Consumer Protection Agency Bill (S.707) that was pending before the United States Senate. The message pre- sented arguments favoring the bill and stated the functions of the proposed agency. Subjects were randomly assigned to experimental treatments so that half the subjects received a message attributed to the highly credible source, while the other half received a message attributed to the less credible source. The second independent variable involved manipulating the timing of the behavioral request relative to other posttest measures.1 After subjects had completed the dependent measures, the booklets were collected. Subjects were debriefed and dismissed.

Independent Variables

Two independent variables were manipulated in the study: source credibility and the timing of the behavioral request. The source was introduced at the beginning of the written communication as a supporter of the bill and the reasons for his advocacy were given. The source description was manipulated to represent high and low credibility. In the high credi- bility condition, the source was described as a Harvard-trained lawyer with extensive experience in the area of consumer issues and a recognized expert whose advice was widely sought. His support of this particular bill was made distinctive by describing him as a person who generally did not favor government controls. The less credible source, on the other hand, was portrayed as an individual with no special ex- pertise but who was interested in consumer protection because of a job opportunity as consumer lobbyist. He was described as holding socialistic views with strong opinions in favor of all government controls.

The second independent variable involved manip- ulating the timing of the dependent measures. For half the subjects, the verbal scales were adminis- tered after they had the opportunity to comply with or reject the request to support the bill. The remaining subjects completed the verbal scales first and then responded to the behavioral request.

Measuring Instruments

Measures of subjects' attitudes toward the Con- sumer Protection Agency Bill, their evaluation of the source's credibility, as well as behavioral compliance with the source's request served as criterion variables. These measures were administered in an order such that the source credibility measures always followed the attitudinal items. Further, depending upon the treatment to which subjects were randomly assigned, the behavioral request either preceded the attitude measures or followed the source credibility measures.

Individuals' attitudes toward the bill were assessed using a seven point Likert-type scale composed of six items. These items included two measures of affect (e.g., "Passage of the Consumer Protection Agency Bill would make me feel good."), two measures of cognition (e.g., "It is my belief that consumers do not need protection.") and two measures of behavioral intention (e.g., "I would personally support estab- lishing the Consumer Protection Agency."). Since the evidence regarding the uniqueness of these at- titude components is equivocal (Insko and Schopler, 1967; Kothandapani, 1971; Ostrom, 1969), the at- titudinal data were analyzed in two ways. One pro- cedure entailed examining subjects' responses on each attitudinal component separately, as is advocated by proponents of the three component distinction. In addition, an overall measure of attitude was obtained by summing each individual's responsos on the affect, cognitive, and behavioral intention scales. This ap- proach conforms to the procedure suggested by those who contend that the three components of attitude are highly convergent.

Two types of source evaluation measures were ad- ministered to determine the efficacy of the source credibility manipulation. One set of dependent vari- ables measured subjects' perception of the trust- worthiness and expertise of the source that were systematically varied in constructing the source credibility independent variable. Six items (three re- lated to expertise and three related to trust), each rated on a seven point semantic differential scale, were used for this purpose. Items included: expert- not expert, experienced-not experienced, trained- untrained, trustworthy-not trustworthy, moral-im- moral, good-bad. A second set of dependent variables were related to the attractiveness aspect of source credibility, which was not manipulated in the present study. These nonequivalent source credibility items included: attractive-unattractive, dynamic-not dy- namic, aggressive-not aggressive. They were admin- istered to determine whether demand characteristics accounted for subjects' perception of the source's credibility (Orne, 1969). If the source credibility induction was effective and not attributable to demand characteristics, then subjects should perceive the highly credible source to be more trustworthy and

1 To avoid confounding classes with treatments, all four exper- imental treatments were run in each of the three classes used.

This content downloaded from 192.236.36.29 on Fri, 20 Sep 2013 15:22:08 PMAll use subject to JSTOR Terms and Conditions

Page 6: Highly Credible Sources: Persuasive Facilitators or Persuasive Liabilities?

HIGHLY CREDIBLE SOURCES 227

TABLE 1

MEAN (X), STANDARD DEVIATION (S.D.), AND CELL SIZE (N) FOR ATTITUDE CATEGORIZED BY INDEPENDENT VARIABLES

Source Credibility

Timing of Response to High Low Behavioral the Behavioral Dependent

Request Request Measure X S.D. n X S.D. n

Attitude Compliance Attitude (Overall) 35.79 3.38 14 33.50 4.65 10 Measure Affect 12.50 1.02 11.60 2.12 Administered Cognition 12.64 1.60 12.00 1.94 Before Behavioral Intention 10.65 1.78 9.90 2.38 Behavioral Request Noncompliance Attitude 25.13 7.27 15 24.95 7.97 19 (A -e B) Affect 9.47 2.17 8.58 3.10

Cognition 9.26 3.22 9.21 3.38 Behavioral Intention 6.40 3.25 7.16 2.89

Behavioral Compliance Attitude 31.56 4.46 18 34.43 5.26 14 Request Affect 11.17 1.92 12.29 1.38 Made Cognition 11.83 1.54 12.64 1.78 Before Behavioral Intention 8.56 1.79 9.50 2.93 Attitude Measure Noncompliance Attitude 19.75 1-0.29 8 27.25 8.56 12 Administered Affect 6.00 3.74 10.00 3.30 (B --A) Cognition 7.75 4.03 10.00 3.80

Behavioral Intention 6.00 3.51 7.25 2.90

Note: The higher the score the more favorable the evaluation.

expert than the low credibility source, but there should be no systematic source credibility effect on the attractiveness items.

Finally, behavioral compliance was determined by whether or not subjects signed a petition. All subjects were asked to sign a petition that was to be sent to their Senators in Washington urging these representatives to vote for the bill.

RESULTS

Manipulation Check

A manipulation check was performed to determine whether or not the source credibility induction was effective. After receiving the persuasive appeal from either the high or low credibility source, subjects evaluated the communicator's credibility on nine measures. A Cronbach a test (1951) for internal consistency indicated that these items were highly reliable (a = .84). Further, it was found that subjects who received the message from the high credibility source perceived the communicator to be significantly more trustworthy and expert than did subjects who received the message from the low credibility source (High Credibility Source: X = 27.82, S.D. = 8.49, n = 55; Low Credibility Source: X = 22.78, S.D. = 7.24, n = 55. F = 11.14, df = 1/108, p < .001). However, there was no systematic difference in sub- jects' perception of the attractiveness of the high and low credibility sources (High Credibility Source: X = 12.56, S.D. = 3.82; Low Credibility Source: X = 12.68, S.D. = 3.48, F < 1). Since the source

credibility manipulation involved varying trustworthi- ness and expertise but not attractiveness, these data indicate that the source credibility induction was ef- fective and not attributable to demand characteristics.

Attitudes

The subjects' attitudinal responses categorized by independent variables are reported in Table 1. The data include subjects' responses on each of the 3 attitudinal components-affective, cognitive, and be- havioral intention as well as on the sum of these components. A test for internal consistency indicated that these items were highly reliable (Cronbach a = .88).

To determine the effects of the experimental vari- ables on subjects' attitudes toward the Consumer Protection Agency Bill, an analysis of variance was performed employing the sum of the attitudinal responses as the dependent measure. It was found that neither source credibility nor timing of the be- havioral request had a significant effect on individuals' attitudes (F < 1). However, the source credibility x timing of the behavioral request interaction was marginally significant (F = 3.28, df = /I106,p = .07). This effect is attributable to the fact that the highly credible source induced a more positive attitude toward the bill when the attitude measure was administered before the behavioral request, whereas the low credibility source was more persuasive if the behavioral request preceded the administration of the attitudinal posttest (Figure 1).

This content downloaded from 192.236.36.29 on Fri, 20 Sep 2013 15:22:08 PMAll use subject to JSTOR Terms and Conditions

Page 7: Highly Credible Sources: Persuasive Facilitators or Persuasive Liabilities?

228 THE JOURNAL OF CONSUMER RESEARCH

TABLE 2

ANALYSIS OF VARIANCE OF ATTITUDE FOR COMPLIERS AND NONCOMPLIERS

Source df MS F

Compliers Source Credibility (A) 1 6.30 <1 Timing of Behavioral

Request (B) 1 56.77 2.84 A xB 1 89.18 4.46*

Noncompliers Source Credibility (A) 1 89.98 1.31 Timing of Behavioral

Request (B) 1 9.89 <1 A x B 1 180.31 2.63

*p < .05.

Of particular interest in this experiment is the effect of the independent variables on the attitudinal re- sponses of compliers and noncompliers. A separate analysis of variance for compliers and noncompliers yielded a significant source x timing interaction for those who complied with the request to sign the petition (Table 2). For noncompliers, the joint effect of source credibility and timing was similar to that for compliers, but it did not reach conventional levels of statistical significance (F = 2.628, df = 1150, p = .10).

These findings were obtained by using the sum of subjects' affective, cognitive, and behavioral intention responses as the dependent measure. To assess the contribution of these attitudinal components in- dividually, the above analysis was repeated using sub- jects' affective, cognitive, and behavioral intention responses as the dependent variables. The main effects of the independent variables on the individual atti- tudinal components were essentially the same as the effects on the sum of the attitude components. As Table 3 indicates, the source credibility and timing main effects were not significant on any of the atti- tudinal components, whether the sample was com- prised of all subjects, compliers, or noncompliers. (The one exception to this conclusion is a significant

timing effect for compliers on the behavioral inten- tion measure.)

Table 3 also reveals that it is the subjects' affective responses which contributed to the significant source credibility x timing of the behavior request interaction. Indeed, the source credibility x timing interaction is not statistically significant when cog- nitive or behavioral intention responses serve as the measure of attitude. Moreover, the nonsignificant source credibility x timing interaction observed for noncompliers when the sum of attitude components is considered, reaches conventional levels of sta- tistical significance when subjects' affective responses are considered alone.

The foregoing analysis indicates that the source credibility x timing interaction is significant and that it is attributable to subjects' affective responses. While this finding is necessary to demonstrate the efficacy of cognitive response and self-perception theories, it is not sufficient. It must also be demon- strated that (1) when attitudes were measured before the behavioral response the highly credible source is more persuasive than the low credibility source; and, (2) when behavioral response preceded subjects' as- sessment of their attitudes, the low credibility source induced more persuasion than the high credibility source.

Contrary to expectation, it was found that when the attitude measure preceded the behavioral request, the highly credible source was not significantly more persuasive than the low credibility copmunicator. As Table 4 indicates, this result emerged whether the sample was comprised of all subjects, compliers, or noncompliers, or whether a sum scale of all atti- tude components or only the affective measure was used.2 The one exception was the finding that the highly credible source induced a more positive attitude than the low credibility source when the affective responses

TABLE 3

F RATIOS ON AFFECTIVE (A), COGNITIVE (C), AND BEHAVIORAL INTENTIONS (B) DIMENSIONS OF ATTITUDES

All Subjects Compliers Noncompliers (df = 1/102) (df = 1/52) (df = 1/50)

Source A B C A B C A B C

Source Credibility 1.49 1.27 .89 .34 .14 .18 1.15 .71 1.22 Timing of Behavioral

Request 1.28 2.14 .12 1.14 5.18* .18 .51 .03 .02 Behavior 47.60** 32.23** 37.33** - - - -

Source x Timing 12.53** 1.11 3.15 5.03* 1.92 2.47 7.99* 1.43 .08

** p < .01. *p < .05.

2 One tailed tests were performed in Table 4 because the direction of treatment differences were predicted. Further, since the source credibility x timing of the behavior interaction was significant, the contrasts reported in Table 4 involved protected t's. There- fore, no correction for error rate was needed (Winer, 1971).

This content downloaded from 192.236.36.29 on Fri, 20 Sep 2013 15:22:08 PMAll use subject to JSTOR Terms and Conditions

Page 8: Highly Credible Sources: Persuasive Facilitators or Persuasive Liabilities?

HIGHLY CREDIBLE SOURCES 229

TABLE 4

CONTRAST OF EFFECT OF HIGH AND LOW CREDIBILITY SOURCES ON SUM SCALE AND AFFECTIVE ATTITUDINAL MEASURES WHEN ATTITUDE MEASUREa PRECEDES AND FOLLOWS BEHAVIORAL REQUEST

Attitude Measure Attitude Measure Precedes Behavior Follows Behavior

Sample Attitude Measure (A -* B) (B -* A)

All subjects Sum Scale t = 1.14 (df = 56) t = 1.41 (df = 50) n = 110 Affect t = 1.82* (df = 56) t = 1.92* (df = 50)

Compliers Sum Scale t = 1.33 (df = 22) t = 1.99* (df = 30) n = 56 Affect t = 1.25 (df = 22) t = 1.93* (df = 30)

Noncompliers Sum Scale t = 1 (df = 32) t = 1.74* (df = 18) n = 54 Affect t = 1 (df = 32) t = 2.46** (df = 18)

a One tailed test. ** p < .025. *p < .05.

of all subjects were examined. In contrast, and con- sistent with the outcome postulated on the basis of self-perception theory, when the behavioral request preceded the administration of the attitudinal measure, the source credibility effect was generally significant, regardless of whether all subjects, compliers, or non- compliers were analyzed and regardless of whether the sum scale or affective measure of attitude was em- ployed. In nearly every instance the low credibility source induced a significantly more favorable attitudinal response than did the highly credible source. (The one exception is when all subjects' responses were deter- mined using the sum scale. Here t = 1.41, df = 50, p < .07.)

Behavioral Compliance

The independent variables had a significant effect on behavioral compliance with the request to sign the petition. Of the subjects randomly assigned to the high credibility source condition, 58 percent complied, whereas 44 percent of those assigned to the low credibility condition signed the petition (Table 1). This difference is marginally significant (z = 1.49, p = .07). Moreover, 62 percent of the subjects com- plied with the request to sign the petition when it followed the communication, while 41 percent com- plied when the request followed the administration of the attitude and source evaluation measures, a difference that is highly significant (z = 2.12, p = .03). The interaction between source credibility and petition was not significant (p > .80).

DISCUSSION

The data reported in this paper indicate that in- dividuals' attitudinal judgments are causally related to the interactive effects of source credibility and the timing of the behavioral request, whereas their be- havioral responses are determined primarily by the timing of the behavior request. The efficacy of self-

perception theory and cognitive response analysis in ordering these findings is discussed in this section.

Attitudinal Effects

When message recipients' attitudinal responses served as the criterion measure, the interactive effect of source credibility and the timing of the behavioral request was significant, whether the sample was com- prised of those who complied with the behavioral request, noncompliers, or both of these groups. More specifically, when individuals had only source and message cues on which to make an attitudinal judg- ment, the effect of source credibility was not sig- nificant. On the other hand, when people had their own behavioral response as well as source and mes- sage cues as a basis for assessing their attitudes, the low credibility source was significantly more persuasive than the highly credible communicator. Furthermore, the significant source credibility x timing of the behavioral request interaction is attributable to subjects' affective attitudinal responses. Indeed, for all subjects and compliers, cognitive and be- havioral intention responses did not contribute to the significant interaction, while for noncompliers the interaction was significant only when the affective responses were considered.

These findings provide support for the self-percep- tion theory. Consistent with the self-perception predic- tion, when behavior served as a cue, and an external justification or pressure for that behavior was not present, subjects inferred stronger internal dispositions than when behavior could readily be attributed to an external cause. More specifically, for those people who complied with the high credibility source's re- quest to sign the petition favoring the Consumer Protec- tion Agency Bill, the source's credibility served as a discounting cue. Since compliance could be attributed both to internal dispositions and to an external cause, (i.e., source credibility), individuals in this treatment exhibited relatively unfavorable attitudes toward the

This content downloaded from 192.236.36.29 on Fri, 20 Sep 2013 15:22:08 PMAll use subject to JSTOR Terms and Conditions

Page 9: Highly Credible Sources: Persuasive Facilitators or Persuasive Liabilities?

230 THE JOURNAL OF CONSUMER RESEARCH

bill. However, when the communicator was of low credibility, the source cue augmented the attribution of complaint behavior to internal or personal causes. Hence, subjects evinced a relatively favorable attitude toward the Consumer Protection Agency Bill.

Self-perception theory also orders the attitudinal responses of noncompliers. Individuals' refusal to comply with the behavioral request despite its advocacy by a highly credible source made them certain that their attitude toward the bill was negative. On the other hand, those noncompliers receiving the communication from a low credibility source were uncertain whether their failure to comply was at- tributable to the source or some negative disposition toward the bill. Thus noncompliers receiving the ap- peal to support the bill from the low credibility source exhibited more favorable attitudes toward it than did those receiving the message from a highly credible communicator.

Despite the fact that self-perception theory orders the observed findings, it does not explain why the source credibility x timing interaction for noncom- pliers was significant only when their affective responses constituted the dependent measure, whereas this interaction was significant for compliers regardless of whether affect alone or all three components of attitude served as the dependent measure. In essence, the question is why the disparity between compliers and noncompliers? To answer this question requires examination of the experimental task. Individuals who complied to the behavioral request may have had few alternative explanations for their behavior; either their own attitudes were favorable (when source credibility was low), or the perceived external justifi- cation was high (when source credibility was high). Noncompliers, on the other hand, may have had a wider range of reasons for their behavior than did those who signed the petition. For example, they may have had a favorable affect toward the bill, but not toward signing the petition; or they may not have wanted to get involved; or they may have had un- favorable attitudes toward the bill. Consistent with this line of reasoning, significantly greater variability in attitudinal responses was observed among non- compliers than among compliers (F = 3.227, df = 53/ 55,p < .01).

The data provide little support for cognitive re- sponse analysis. In addition, they are at odds with the source credibility main effect consistently reported in the literature. Indeed, the highly credible source was not significantly more persuasive than the low credibility communicator when subjects' attitudes were determined before they had an opportunity to respond to the behavioral request, even though the source credibility manipulation was successful.

The failure to confirm the source main effect in the present investigation may be attributable to the paradigm used. In contrast to previous research, where

focus centered on the effects of source credibility on attitude change, in the present experiment the source was seeking behavioral compliance with his request. In pretesting the source credibility in- duction, it was observed that presenting the low credibility communicator as someone with virtually no expertise or trustworthiness, as was typical in previous research, severely affected subjects' percep- tion of the experiment's reality. Message recipients did not believe that a communicator who was seeking compliance with a request would present himself in a highly negative light. In addition, providing an ex- tremely negative profile of the low credibility source would, in all probability, have resulted in very low levels of compliance with the behavioral request, obviating an adequate test of self-perception theory.

Thus, in the effort to insure the reality of the experimental task and gain substantial compliance with the behavioral request even when the com- municator was not highly credible, the low credibility source was not portrayed in a totally negative manner. Although this resulted in the finding that the source credibility effect on attitudinal responses determined immediately after the communication was not signif- icant, this finding is not necessarily damaging to the cognitive response analysis formulation. Rather, it indicates that in the present experimental context, though subjects perceived the high and low credibility sources to have different levels of credibility, this difference was not sufficient to cause a difference in their attitudinal responses.

An alternative explanation for the failure to observe a significant source credibility effect when message recipients assessed their attitudes prior to behavior is also viable. The cognitive response prediction that a highly credible source will be more persuasive than a low credibility source is predicated on the as- sumption that message recipients are unfavorably predisposed to the message issue. Under these circum- stances, it is expected that a highly credible source will depress counterargumentation and a low credibility source will facilitate it. If the subjects in the present study were not unfavorably predisposed to the Con- sumer Protection Agency Bill, this would account for the finding that source credibility had no systematic attitudinal effect. Since measures of message re- cipients' pre-communication attitudes were not ad- ministered, the veracity of this explanation is problem- atic. To address this issue requires determining subjects' attitudes toward the communication issue prior to message exposure and analyzing the responses of those who are positively and negatively predis- posed separately.

Behavioral Effects The source credibility and timing of the behavioral

request independent variables had a somewhat differ- ent effect on message recipients' behavioral response

This content downloaded from 192.236.36.29 on Fri, 20 Sep 2013 15:22:08 PMAll use subject to JSTOR Terms and Conditions

Page 10: Highly Credible Sources: Persuasive Facilitators or Persuasive Liabilities?

HIGHLY CREDIBLE SOURCES 231

than they did on subjects' attitudinal responses. The timing of the behavioral request had a significant effect; more people signed the petition when the re- quest preceded rather than followed the attitude questionnaire. Further, the high credibility source in- duced greater behavioral compliance than the low credibility source, though the magnitude of the differ- ence was only marginally significant.

These data provide support for the contention that as dependent variables attitudes and behavior con- stitute different methods for measuring the same under- lying trait (Cook and Selltiz, 1964). Indeed, when behavior was measured immediately after the com- munication, it yields the same inference as is derived from measuring attitudes immediately after the mes- sage; the high credibility source is slightly (though not significantly) more effective in inducing support for his advocacy than a low credibility source.

Although attitudes and behavior constitute alterna- tive dependent measures of the same disposition, they mediate influence in distinctly different ways. As is predicted by self-perception theory, when message recipients' own behavior serves as a mediator (i.e., in the behavior-attitude condition), individuals examine the causal antecedents of that behavior in assessing their attitudes. As a result, the low credibility source is more persuasive than the highly credible source. In contrast, when individuals' attitudes mediate their behavioral response (i.e., in the attitude- behavior condition), the causal antecedents of the message guide their actions. Thus, the highly credible source is somewhat more persuasive than the low credibility source. Moreover, the timing of the be- havioral request has a significant effect on behavior. In effect, making message recipients' attitudes salient to them reduces their likelihood of subsequent com- pliance with the communicator's advocacy.

This latter finding may be interpreted in terms of cognitive response analysis. Specifically, the statement of attitudes may have stimulated message recipients to rehearse thoughts that were in opposition to the Consumer Protection Agency Bill, thus reducing their willingness to sign the petition. On the other hand, when the request to sign the petition followed im- mediately after the communication, message recipients had little opportunity to rehearse arguments opposing the communicator's advocacy, and therefore their compliance was relatively high.

CONCLUSIONS

From a pragmatic perspective, the present findings suggest the need for caution in deciding whether or not to pursue strategies to enhance source credibility. If the strategy is to influence behavior by inducing a positive change in attitudinal precursors, then it is appropriate to consider strategies for enhancing source credibility. However, if the objective is to induce be-

havior directly (e.g., product trial), advocacy by a highly credible source is likely to undermine the acquisition of a positive attitude toward performing the behavior. In turn, this outcome will reduce the likelihood of behavioral persistence (e.g., repeat purchase).

Although the present study only provides evidence for the fact that a highly credible source may under- mine the acquisition of positive attitudes, there is in- direct evidence from other studies supporting the as- sertion that highly credible sources may also undermine behavioral persistence. Scott (1975) found that very substantial discounts to motivate trial of a newspaper subsequently resulted in fewer subscriptions to the newspaper than did a lower discount. Uranowitz (1975) reported that women asked to watch the ex- perimenter's packages when justification was low (experimenter had to retrieve a dollar) exhibited greater willingness to help a second experimenter than when justification for watching his packages was high (retrieve his wallet). It seems reasonable to contend that the substantial incentive and the retrieval of a wallet are similar to the high credibility source in the present study; all are cues indicating strong external justification for behavior. If this argument is correct, then it suggests that like other cues in- dicating external justification, a highly credible source will undermine behavioral persistence.

From a theoretical perspective, the present research provides evidence for the complementarity of two attitude formulations. Cognitive response analysis may be used to explain the effects of source credibility manipulations in contexts where individuals' own behavior is not a salient cue in making judgments. In these settings, individuals' attitudinal and be- havioral judgments are predicated on the causal ante- cedents of the message. Thus it was observed that a highly credible source was not a liability. On the other hand, in situations where individuals' actions are an important determinant of their attitudinal judgments, self-perception theory accounts for the finding that a low credibility source induces a more positive attitude toward the issue than a high credibility communicator. Moreover, the present experiment sug- gests that the attribution of behavior affects people's feelings about an issue (i.e., affective responses) rather than their cognitions or behavioral intentions regarding it.

Further research is needed to confirm the finding that attributions made on the basis of one's own behavior have their predominant influence on in- dividuals' affective responses. In addition, investiga- tions are needed to determine the effect of individuals' initial disposition toward an issue on the persuasive effect of source credibility. Finally, future research should identify the relationship between message recipients' cognitive responses and their subsequent actions.

This content downloaded from 192.236.36.29 on Fri, 20 Sep 2013 15:22:08 PMAll use subject to JSTOR Terms and Conditions

Page 11: Highly Credible Sources: Persuasive Facilitators or Persuasive Liabilities?

232 THE JOURNAL OF CONSUMER RESEARCH

REFERENCES

Bem, D. "Self-Perception Theory," in L. Berkowitz (Ed.), Advances in Experimental Social Psychology. New York: Academic Press, 1972.

Berscheid, E., R. Baron, M. Dernier, and M. Libman. "Anticipating Informed Consent," American Psycholo- gist 28 (1973), 913-925.

Brock, Timothy. "Communicator-Recipient Similarity and Decision Change," Journal of Personality and Social Psychology, 1 (1965), 650-654.

Capon, N. and J. Hulbert. "The Sleeper Effect, an Awaken- ing," Public Opinion Quarterly, 37 (1973), 333-358.

Cook, Stuart and Claire Selltiz. "A Multiple-Indicator Ap- proach to Attitude Measurement," Psychological Bul- letin, 62 (1964), 36-55.

Cronbach, L. "Coefficient Alpha and the Internal Structure of Tests," Psychometrika, 16 (1951), 297-334.

Eagly, A. and S. Chaiken, "An Attribution Analysis of the Effect of Communicator Characteristics on Opinion Change: the Case of Communicator Attractiveness," Journal of Personality and Social Psychology, 32 (1975), 136-144.

Greenwald, A. "Cognitive Learning, Cognitive Response to Persuasion, and Attitude Change," in A. Greenwald, T. Brock, and T. Ostrom (Eds.), Psychological Founda- tions of Attitudes. New York: Academic Press, 1968, 147-170.

Himmelfarb, S. and D. Arazi, "Choice and Source Attrac- tiveness in Exposure to Discrepant Messages," Journal of Experimental Social Psychology, 10, (1974), 516- 527.

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

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

Insko, Chester and John Schopler. "Triadic Consistency: A Statement of Affective-Cognitive-Conative Con- sistency," Psychological Review, 74 (1967), 361-376.

Johnson, Homer and Richard Izzett. "The Effects of Source Identification on Attitude Change as a Function of the Type of Communication," Journal of Social Psychology, 86 (1972), 81-87.

Johnson, Homer and John Scileppi. "Effects of Ego-In- volvement Conditions on Attitude Change to High and Low Credibility Communicators," Journal of Person- ality and Social Psychology, 13, (1969), 31-36.

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

Jones, R. and J. Brehm, "Attitudinal Effects of Communi- cator Attractiveness When One Chooses to Listen," Journal of Personality and Social Psychology, 6, (1967), 64-70.

Kelley, H. "The Processes of Causal Attribution," Ameri- can Psychologist, 28, (1973), 107-128.

Kelman, H. and C. Hovland, " 'Reinstatement' of the Com- municator in Delayed Measurement of Opinion Change," Journal of Abnormal and Social Psychol- ogy, 48, (1953), 327-335.

Kothandapani, V. "Validation of Feeling, Belief and In- tention to Act as Three Components of Attitude and Their Contribution to Prediction of Contraceptive Be- havior," Journal of Personality and Social Psychology, 19, (1971), 321-333.

McCroskey, James. "A Summary of Experimental Research on the Effects of Evidence in Persuasive Communica- tion," Quarterly Journal of Speech, 55, (1969), 169- 176.

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

Orne, Martin. "Demand Characteristics and the Concept of Quasi-Controls," in R. Rosenthal and R. Rosnow (eds.), Artifact in Behavioral Research. New York: Academic Press, 1969.

Ostrom, T. "The Relationship between Affective, Be- havioral, and Cognitive Components of Attitude," Journal of Experimental Social Psychology, 5, (1969), 376-379.

Powell, F. "Source Credibility and Behavioral Compliance as Determinants of Attitude Change,",Journal of Per- sonality and Social Psychology, 2, (1965), 669-676.

Rhine, Ramon and Laurence Severance. "Ego-Involvement, Discrepancy, Source Credibility, and Attitude Change," Journal of Personality and Social Psychology, 16, (1970), 175-190.

Ritchie, E. and E. Phares, "Attitude Change as a Func- tion of Internal-External Control and Communicator Status," Journal of Personality, 37, (1969), 429-443.

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

Scott, C. "Past Behavior, Incentives, and Maintained Behavior Change: A Field Study of Self-Attribution Theory," unpublished doctoral dissertation, North- western University, 1975.

Sigall, H. and R. Helmreich, "Opinion Change as a Func- tion of Stress and Communicator Credibility," Journal of Experimental Social Psychology, 5, (1969), 70-78.

Simons, H., N. Berkowitz, and R. Moyer, "Similarity, Credibility, and Attitude Change: A Review and a Theory," Psychological Bulletin, 73, (1970), 1-16.

Uranowitz, S. "Helping and Self-Attributions: A Field Experiment," Journal of Personality and Social Psy- chology, 31, (1975), 852-854.

Walster, E., E. Aronson, and D. Abrahams, "On In- creasing the Persuasiveness of a Low Prestige Com- municator," Journal of Experimental Social Psychol- ogy, 2, (1966), 325-342.

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

Watts, W. and W. McGuire, "Persistence of Induced Opinion Change and Retention of the Inducing Message Contents ," Journal ofAbnormal and Social Psychology, 68, (1964), 233-241.

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

Winer, B. Statistical Principles in Experimental Design. New York: McGraw-Hill, 1971.

This content downloaded from 192.236.36.29 on Fri, 20 Sep 2013 15:22:08 PMAll use subject to JSTOR Terms and Conditions