Reluctant testimony is theoretically superior to testimony emanating from a biased source. Some writers suggest that it is also superior to other unbiased testimony. Two experimental studies indicated that reluctant testimony produced higher credibility ratings than biased testimony, but unbiased testimony produced higher credibility ratings than either biased or reluctant testimony.

Much has been written in both public speaking and debate textbooks concerning the use of authoritative opinion as evidence in persuasive argument. Because authoritative opinion is generally considered a useful tool for the advocate, authors typically suggest a list of criteria for the advocate's use in selecting the authoritative opinion which he will use in his speech. A review of leading texts in public speaking and argumentation and debate indicated that two of the most frequent criteria included in such lists are concerned with the source's expertness and trustworthiness on the topic in question.¹ Is the source an expert in the field? Is the source unbiased?

Most speech teachers will agree that quoting from a source whom the audience believes to be biased is likely to lessen the value of that source's testimony. To quote George Meany's opinion that anti-trust laws should not be applied to labor unions is likely to contribute little to the advancement of an advocate's case. On the other hand, if George Meany indicated that he favored the application of anti-trust laws to labor unions his testimony could be viewed in a different light. He would then be testifying against what would generally be thought to be his, "best interests." This type of testimony is commonly referred to as "reluctant testimony."²

Reluctant testimony is considered excellent evidence by most writers. Some recommend it above all other types of testimony. For example, Miller poses a question as the primary test of evidence based on authoritative testimony. "What else would you expect him to say?" He continues by stating, "As a beginning step, then, one should seek testimony composed of authority-based assertion in which the assertions made are not what would be routinely expected."³ The basis for this theoretical superiority of reluctant testimony is explained by Bettinghaus:

Most people are inclined to place considerable weight on evidence coming from a reluctant witness, although it may be no more true than evidence that is not reluctant. But the credibility of the witness seems to most of us to be better established when he is testifying against what seem to be his best interests.⁴

It would seem, then, that testimony to be useful as evidence in argument should come from a source that is highly credible in terms of his authoritativeness on the topic and his trustworthiness in the given case. If he is reluctant to give the testimony, so much the better.

There is one problem, however, which is not apparent from a cursory examination of this theory. If the audience shares the quoted source's bias, we should not expect the fact that the source is biased on the point in question to affect adversely the impact of his testimony. In addition, if the source is quoted as expressing a view contrary to what would be expected to be his best interests (reluctant testimony) and his audience shares those interests (holds the same bias as the quoted source), we should not expect an improved effect for his testimony. The effects of bias or reluctance, therefore, should be expected to vary with the bias of the audience. With a randomly selected audience, of course, we would expect a considerable range of bias on a given question. Thus the lack of bias or presence of reluctance on the part of a quoted source should increase the impact of a given piece of testimony as evidence with some audience members even though these factors would have no effect on other audience members.

There has been no experimental research reported in the literature pertinent to the question of what type of authoritative testimony has the greatest impact as evidence in persuasion. No studies have compared the effectiveness of biased versus reluctant testimony or of reluctant versus unbiased testimony with any audience. However, considerable research has been conducted concerning the force of credibility or ethos in persuasion. The results of these studies indicate that, in general, the higher the credibility of a source, the greater the persuasive effect of the source's message.⁵

Since quoting an authority as evidence is a direct attempt to use the authority's ethos to increase persuasion, we may conclude on the basis of the credibility research that the authority-based evidence which is best is that which is considered by the audience as emanating from the most credible source.⁶ Thus, the question posed for the studies reported below was, what source of authority-based evidence is considered the most credible; the biased source; the reluctant source; or the unbiased source?

Two studies were conducted. The procedures for the two studies were similar. Two opinion statements of approximately fifty words in length were constructed, one statement pro-labor in orientation, the other anti-labor. Each stated a definite position as to whether labor unions make a significant contribution to inflation. In order to assure that the opinion statements were correctly labeled as pro- or anti-labor, students in two sections of Speech 200 at the Pennsylvania State University were asked to read the statements and label them as either "pro-labor" or anti-labor."⁷ All but two of the thirty-eight students indicated that the statements were interpreted as intended.

Introductions of three sources were developed. One introduction was for a labor leader, a second was for a management leader, and a third was for a professor of economics. Each introduction was then attached to each of the opinion statements. This provided six different conditions for the experiments.

In both studies the experimental subjects were exposed to one message-introduction combination and asked to complete measures of source credibility. Separate measures of perceived authoritativeness and character of the source were taken in each study. These represent the two experimentally verified components of the credibility or ethos construct.⁸

The experimental hypotheses for both studies were that there would be no difference in perceived authoritativeness or character between biased sources and reluctant sources or between reluctant sources and unbiased sources. To test these hypotheses the credibility scores of the labor leader--pro-labor treatment and the management leader--anti-labor treatment were combined to represent the "biased source" condition; the labor leader--anti-labor treatment and the management leader--pro-labor treatment were combined to represent the "reluctant source" condition. This procedure provided for reduction of the possibility that audience bias on the labor question would influence the results of the study. While random audience bias was assumed, it was felt wise to reduce the potential effects of such bias as much as possible within the limitations of the design of the studies. The professor of economics--pro-labor condition and the professor of economics--anti-labor condition were combined to represent the "unbiased source" condition.

Statistical tests of the hypotheses were provided by subjecting the credibility scores to analysis of variance and computing t-tests of the differences between mean scores of the three conditions.⁹

Six sections of students enrolled in Speech 200 at the Pennsylvania State University were selected to serve as experimental subjects in the first study. Each section received a different introduction-message treatment. The course instructor read both the source introduction and the opinion statement to the subjects. After hearing the experimental stimuli, the subjects completed Likert-type scales designed to measure the perceived authoritativeness and character of the stimulus source.¹⁰

Results

Table 1 reports the obtained mean authoritativeness and character scores for the three experimental conditions, the difference in mean scores between conditions, and the results of the t-tests of those differences. The hypothesis of no difference in perceived credibility between the biased and reluctant source conditions and between the biased and unbiased source conditions were rejected. The reluctant and unbiased source conditions produced considerably higher perceived credibility than the biased source condition on both the authoritativeness and character dimensions. The hypothesis of no difference in perceived credibility between the reluctant and unbiased sources, however, could be rejected for only one dimension of source credibility. The unbiased source condition produced considerably higher perceived character than the reluctant source condition, but the difference between the two conditions on the authoritativeness dimension was not statistically significant.

*The hypothetical neutral point on this measure is 66.0. Scores can range from 22 to 110. The lower the score the higher the perceived authoritativeness.

**The hypothetical neutral point on this measure is 60.0. Scores can range from 20 to 100. The lower the score the higher the perceived character.

These results provide support for the theory expressed in our public speaking and argumentation textbooks that reluctant and unbiased testimony should be more helpful to the advocate than biased testimony. However, the assumed superiority of reluctant testimony over all other forms of authority-based evidence was not supported by this study. The unbiased source--the professor of economics--was perceived as considerably higher on the character dimension than the reluctant source and the difference between the two conditions on perceived authoritativeness, though not statistically significant, was in the direction opposite to that implied by the above assumption.

From the results of this study we may be tempted to conclude that the best evidence is that which comes from an unbiased source, not that which comes from a reluctant source. However, the procedures used in this study call for caution. The experimental subjects were not sampled randomly but were taken intact in the available sections. The size of the sample was small, only 29 subjects in each of the three conditions. In addition, the experiment was conducted in the classroom with a professor reading the stimulus messages to the subjects. Because of this latter condition we may suspect that the "economics professor" to whom the messages were attributed may have received artificially high ratings of perceived credibility. Because of these problems, a second study was conducted outside of the classroom with a much larger sample of subjects selected on a random basis.

The experimental stimuli used in the second study were the same as those used in the first study described above. Six audiences were selected to read the messages and source introductions, each audience reading a different message-introduction combination. Each audience was composed of twenty-five male and twenty-five female students enrolled in Speech 200 at the Pennsylvania State University. The experimental subjects were among students in forty-eight sections of Speech 200 randomly assigned to six audiences for another experiment. Because of the nature of the experimental stimuli in Study 2 (written) it was possible to distribute the subjects in this study across the cells of the other experiment in random order so as to preclude carry-over confounding effects attributable to the other experiment.

After the unrelated experiment the subjects were asked to turn to the next page in the booklets they had previously received. Instructions for the present study were included on that page. No oral instructions were given. The written instructions informed the subjects that "a booklet of examples of speech units was being developed by the Speech Department" and that it was necessary to "determine the reactions of Speech 200 students to the sources of these materials." The subjects were asked to read the material about the source and his statement and then complete two brief semantic differentials. These semantic differentials were measures of perceived source authoritativeness and character.¹¹

The experimental hypotheses and statistical analyses for this study were the same as for Study 1.

Results

Table 2 reports the obtained mean authoritativeness and character scores for the three experimental conditions, the differences in mean scores between conditions, and the results of the t-tests of those differences. The hypothesis of no difference in perceived character between the biased and reluctant source conditions and between the reluctant and unbiased source conditions could not be rejected. There was no statistically significant difference between the conditions. The difference between the unbiased and biased source conditions approached significance. This difference can be considered significant if we use the results of Study 1 as justification for a directional prediction. The hypotheses of no difference in perceived authoritativeness between the various source conditions, however, were rejected. The results indicated that the unbiased source (professor) was perceived to be more authoritative than the reluctant source, which in turn was perceived to be more authoritative than the biased source.

*The hypothetical neutral point on both of these measures is 24.0. Scores can range from 6 to 42. The lower the score the higher the perceived credibility.

**On the basis of the results of the first study a directional hypothesis would be justified. A one-tailed t test based on this hypothesis would meet the usual .05 significance criterion.

The results of these two studies provide confirmation for the theory expressed in our textbooks that reluctant and unbiased testimony should be more helpful to the advocate than biased testimony. In both studies the reluctant and unbiased sources were perceived to be more authoritative than the biased source. Character scores were also higher for the reluctant and unbiased sources than for the biased source, although the difference in favor of the reluctant source in the second study was not statistically significant. Therefore, on the basis of these results and the previous research on the effect of source credibility on persuasion, we may conclude that reluctant and unbiased testimony are more potent persuasive tools than biased testimony.

The theory that reluctant testimony is superior to unbiased testimony was not supported by the results of these studies. In all cases the differences between the reluctant source and unbiased source conditions favored the unbiased source condition. It would, however, seem unwise to conclude from these limited studies that unbiased testimony is superior to reluctant testimony. In these studies the reluctant source condition was a composite of two treatments including source who were labor leaders and management leaders. The unbiased source was an economics professor. It would be reasonable to assume that college students have an inflated conception of the credibility of professors and are not overly impressed by either labor or management leaders. The design of these studies, therefore, with college students as experimental subjects may not have provided a good test of the hypotheses concerning the superiority of reluctant testimony over unbiased testimony.

A study in which subjects' bias were controlled rather than a random factor could be expected to provide useful information. If, as we suspect, a quoted source with a known bias held in common with an audience is derogated when he speaks in opposition to this bias, we need to revise some of the theory expressed in modern textbooks. Reluctant testimony may, indeed, be the least desirable form of authority-based evidence in some cases. Until this needed research is completed, we may cautiously conclude that unbiased testimony is superior to biased testimony and that the value of reluctant testimony is dependent on the audience to which the messages is addressed.

1. See for example, Donald C. Bryant and Karl R. Wallace, Fundamentals of Public Speaking (New York, 1960); Douglas Ehninger and Wayne Brockriede, Decision by Debate (New York, 1963); Henry Lee Ewbank and J. Jeffery Auer, Discussion and Debate: Tools of a Democracy (New York, 1951); William Foster, Argumentation and Debate (Boston, 1932); Austin Freeley, Argumentation and Debate: Rationale Decision Making (Belmont, 1961); Arthur Kruger, Modern Debate: Its Logic and Strategy (New York, 1960); James McBath, ed. Argumentation and Debate (New York, 1963); James McBurney and Glen Mills, Argumentation and Debate: Techniques of a Free Society (New York, 1964); Glen Mills, Reason and Controversy (Boston, 1964); Alan H. Monroe, Principles and Types of Speech (Chicago, 1962); Robert T. Oliver, Harold P. Zelko, and Paul D. Holtzman, Communicative Speech (New York, 1962).

2. Ewbank and Auer, p. 116.

3. Gerald R. Miller, "Evidence and Argument," in Perspectives on Argumentation, Gerald R. Miller and Thomas R. Nilsen, Eds. (Chicago, 1966), pp. 45-46.

4. Erwin P. Bettinghaus, Message Preparation: The Nature of Proof (Indianapolis, 1966), p. 48.

5. Kenneth Anderson and Theodore Clevenger, Jr., "A Summary of Experimental Research in Ethos," Speech Monographs, XXX (June, 1963), 59-78.

6. It is assumed by the writers, as it is assumed by the authors of the textbooks previously cited, that evidence itself has a positive impact in persuasion. Early experimental studies of the effect of evidence in persuasion produced equivocal results. Recent studies have more clearly specified the circumstances in which inclusion of evidence increases impact of a persuasive message. See James C. McCroskey, "Experimental Studies of the Effects of Ethos and Evidence in Persuasive Communication," unpublished dissertation (The Pennsylvania State University, 1966) and James C. McCroskey, "Toward An Understanding of the Importance of 'Evidence' in Persuasive Communication," The Pennsylvania Speech Annual, XXIII (September, 1966), 65-71.

7. Speech 200 is a required course in oral communication. Students from all colleges in the University are enrolled and the classes include students from all undergraduate levels.

8. See James C. McCroskey, "Scales for the Measurement of Ethos," Speech Monographs, XXXIII (March, 1966), 65-72.

9. Statistical procedures used in these studies were based on procedures described by B. J. Winter, Statistical Principles in Experimental Design (New York, 1962).

10. For a discussion of these measures see McCroskey, "Scales for the Measurement of Ethos."

11. Ibid.

Click Here To Go Back To PERIODICALS