James C. McCroskey

Assistant Professor of Speech, Michigan State University

Many of our popular textbooks in public speaking and argumentation suggest that good evidence is a nearly indispensable element in successful persuasion. An equation is sometimes used to define proof:

Evidence + Reasoning = Proof

Not all writers, of course, advance such a rigid view. Most indicate that good evidence is one of a variety of means open to a speaker seeking acceptance of his ideas. Few, however, fail to emphasize the constituents of good evidence. These constituents vary somewhat as textbook writers explain them; however, several are commonly identified: the evidence should be objectively verifiable; it should be clearly related to the point in question; the source of the evidence and the qualifications of the source should be made clear to the audience.

Several experimental researchers have attempted to verify the importance of good evidence in persuasive communication. Their efforts have been relatively unsuccessful.1 After reviewing the results of this research, Gregg concluded that:

. . . the audience reaction to an argument may have little or nothing to do with whether the argument includes fully documented or completely undocumented evidence, relevant or irrelevant evidence, weak or strong evidence or any evidence at all.2

Several previous writers have attempted to explain the experimental findings within the context of contemporary rhetorical theory.3 It will be the purpose of this paper to expand their observations and to suggest how further research findings make necessary a modification of theoretical views concerning the role of evidence in persuasive communication.


To begin with, let us set forth a model of a persuasive unit. Persuasion is the process of relating new beliefs to beliefs already held by an audience in such a manner as to gain audience acceptance of the new beliefs. A single persuasive unit, frequently referred to as an "argument," consists of two elements accepted by an audience which, when related to each other, produce audience acceptance of a new element. The new element, of course, is the speaker's claim--a new belief he hopes his audience will accept.

The other two elements are data and warrant.4 Data consist of one or more specific beliefs held by an audience. There are three types of data which will be discussed below. A warrant is a general belief held by an audience which relates the data to the speaker's claim. Such general beliefs may be concerned with relationship of things in the external world, values held by the audience, or the audience's conceptualization (ethos) of the source of the argument.

All persuasive units include data, warrant, and claim--either stated or implied. Their relationship is exemplified in the two diagramed units below:


Demonstrations against the Viet Nam war should be permitted in this country.


People demonstrating against the Viet Nam war are exercising the freedom of speech.


Since we believe that

freedom of speech is

essential to a

democratic society.


The price of fruit will increase soon.


California fruit pickers have received a substantial wage increase.


Since in a capitalistic economy wage increase usually force employers to increase prices.

Data and warrant should be considered coordinate and indispensable parts of the process of gaining acceptance of claims. They are the support upon which the claim rests. If either is not believed by the audience or if they are unrelated to each other, the persuasive process will be disrupted. The acceptance of the claim for which that persuasive unit was created will not occur; and any subsequent claims which are dependent on the completion of this persuasive unit will also be prevented from gaining acceptance.

A clear understanding of the nature of the three elements of a persuasive unit is vital to the potential persuader, the teacher of rhetoric, and the scientist concerned with the rhetorical process. The natures of claims and warrants have been set forth and discussed at length by Ehninger and Brockriede.5 Their discussion is most satisfactory. Those who are not familiar with this work should become so. For our purposes, however, the nature of data is the crucial concern.

In their textbook Ehninger and Brockriede substitute the term "evidence" for "data."6 While such a substitution is acceptable if one is concerned with the persuasive unit only as it appears in the setting of academic debating, it is not acceptable if one is concerned with the process of persuasion as it appears in other contexts. Unfortunately, the narrow conception of data expressed by Ehninger and Brockriede is the same, or very nearly the same, conception within which experimental researchers have investigated the importance of good evidence usage in persuasive communication.

Since data is defined as evidence in this narrow view, evidence becomes one of the three indispensable elements of a persuasive unit as we have described it above. However, the results of experimental investigations have clearly indicated that in some circumstances, at least, evidence is not indispensable. In fact, it would appear from this body of research that frequently evidence contributes little or nothing to the persuasiveness of a communicative effort.

We shall look at data in a broader context to see if we can develop a better understanding of the role of good evidence usage in persuasive communication.


There are three distinct types of data. The first of the highest order, for it is ultimately the only type upon which a meaningful argument may be developed. One example of this type of data is audience opinion. If the audience believes that Negroes are inferior to whites, this opinion may be used by the persuader as data for an argument. He may not use the opinion that Negroes and whites are equal as data unless he first instills that opinion in his audience's mind. Thus, in any given persuasive circumstance a speaker is restricted in his choice of arguments by data which he can find or implant in the beliefs of his audience.

The second example of this type of data is audience knowledge. Anything that the audience knows can serve as data for an argument. If they are aware of wage increases for fruit pickers, this knowledge may be used by the persuader to obtain acceptance of the claim that fruit price increases in the near future are likely. If, however, the audience is unaware of such wage increases, the speaker is precluded from this data option until such time as he informs his audience of the wage increases and secures their belief in the fact that they occurred.

There is a narrow line between "knowledge" data and "opinion" data. What is knowledge to one person may be opinion to another. We need not be concerned with this, however, because knowledge and opinion operate in almost exactly the same manner. If the audience "believes" or "knows" something, it can be used as data. If they do not "believe" or "know" something, it can not be used as data.

The next type of data is of a lower order, though essential to a persuader in nearly every circumstance. This type of data consists of speaker opinion and asserted information. This type of data is dependent on a secondary, usually implied, argument in every case. This secondary argument has its data the asserted opinion or information. The warrant is based on the credibility of the source. An example of this would be: "I say X's are usually Y" (data), "I am a credible source" (warrant), therefore "X's are usually Y" (claim). The data in this example meets the test of "first order" data, for if the audience hears me say it, their opinion that I said it is immediately assured.7 The crucial determinant of whether the claim is accepted or not has to do with whether my ethos is high enough for the warrant to be acceptable to the audience. This is the case whenever a speaker makes an assertion in a speech. So long as his credibility is high enough there is a warrant that will permit his assertion to become audience opinion or audience knowledge. Thus, assertions of high credibility sources can serve as data for further arguments while assertions of sources with lower credibility serve no persuasive purpose. When a persuader makes an assertion the audience immediately (thought usually not consciously) completes the secondary argument. If the persuader's ethos is high enough the assertion is accepted by the audience and becomes either audience knowledge or audience opinion. At this point data of the first order is present and the persuader can continue to develop further argument.

When appropriate audience opinion or knowledge is not available and the persuader's ethos is not sufficient to establish his assertions as audience opinion or knowledge, the persuader must resort to "third-order" data. Third-order data consists of opinions of others and facts attested to by others. This type of data will be recognized as what we have traditionally called "evidence." As in the case of "second-order" data discussed above, the introduction of opinions of others or facts attested to by others immediately causes the audience to complete a secondary argument. This argument has as data the belief that the outside source made the statement attributed to it and the warrant is based on the ethos of the outside source.

Actually a third-order argument is also produced. It goes something like this: The speaker says that so-and-so said X (data), the speaker is a credible source (warrant), therefore probably so-and-so did say X (claim). Obviously the speaker must have a certain minimal amount of credibility for even this argument to be accepted. But if it is, the data for the secondary argument is established. Then the credibility of the outside source (warrant) becomes crucial. If the outside source is credible enough, the secondary argument is established and new audience opinion or knowledge has been created. This can then be used as data by the persuader to develop further argument.

The establishment of first-order data by means of evidence (third-order data), then, is dependent upon the credibility of the speaker. Is he at least honest enough to tell the truth about what others say? Thus, for the very low ethos persuader, evidence would serve no persuasive purpose. The audience would reject the evidence because of the person presenting it. But if the speaker has the minimal ethos necessary to overcome this obstacle, the credibility of the outside source can become crucial. If the audience is unfamiliar with the outside source, it is quite likely that they will not accept the authoritative warrant. Thus, the advice given in the public speaking and debate textbooks concerning citing and qualifying sources of evidence appears to be theoretically and practically justified.

It is important to note here that either second- or third-order data may be rejected by an audience if the acceptance of it would force acceptance of an unacceptable claim. There are those among us, for example, who would not accept the claim that communism is a better form of government than democracy no matter who asserted it and no matter what evidence he brought forth to support his assertion. We find this an unacceptable claim. No first-order data are available that can persuade us of this claim, and no speaker could establish any by means of second- or third-order data. Some people just can not be persuaded to accept some claims.

What we have suggested, then, is that there are three orders of data available to a communicator when persuasion is possible. First-order data consists of existing audience opinion and audience knowledge. If this type of data is available, it is to be preferred on rhetorical grounds over data of a lower order. It is the data most likely to enable the speaker to achieve his intended persuasive goal. Second-order data consists of asserted opinions and information. This type of data is dependent on the credibility of the persuader. If the speaker is a highly credible source, this type of data is rhetorically preferable to data of a lower order. However, this type of data is available only to a moderately high- to high-ethos communicator. Third-order data consists of opinions of others and facts attested to by others. This type of data is what has traditionally been called "evidence." Its persuasiveness is dependent on both the credibility of the persuader and the credibility of the outside source. Since there are two supplementary arguments introduced each time this type of data is used, there are more chances for this type of data to be rejected by the audience than when first- or second-order data are used. Thus, on rhetorical grounds, we should consider opinions and facts attested to by others the least desirable of the data options open to the persuader. It is, however, a useful option to all but the sources of lowest credibility. Very low ethos sources may have no promising data options available to them and, therefore, they will probably be unsuccessful persuaders whether they include evidence in their speeches or not.


If the above analysis of the structure of units of persuasion and the types of data is accepted, what are the implications for rhetorical education and experimental research? I would suggest that prospective speakers should be given a more realistic view of their persuasive potentialities. Our current stress on the need for evidence in persuasive speeches may rest on a sound ethical base, but its rhetorical legitimacy is highly questionable. We should certainly continue to stress the need for seeking substantial evidence in the investigation of problems. Here social and pedagogical ethics and goals cohere. But we also need to teach students the difference between the processes of investigation and persuasion. Nothing can substitute for solid evidence in investigation, but third-order data (evidence) may be the least useful data option available to speakers in many persuasive situations.

Student speakers should certainly be taught how evidence needs to be used for fullest effect. And they should be encouraged to use third-order data when the other options are not open to them. However, they should be equally aware that to use third-order data in many circumstances will contribute little or nothing to their persuasive goals. That many leading speakers in our society have intuitively reached this awareness is indicated by empirical studies of their speaking.8

The implications for empirical and experimental research are clear. If an investigator wishes to determine whether use of evidence affects persuasive communication, he must design his study so that he is investigating a persuasive situation where first- and second-order data are not available to the persuader. This condition has certainly not been established in most experimental studies.9 Neither has it been established in two of the most frequently cited empirical studies of evidence. In these studies the speakers were a Secretary of State and U.S. Senators.10 For these speakers second-order data were readily available. The finding that these speakers used very little "evidence" should have been expected.

The importance of evidence in the persuasive process, then, can be said to vary with the audience and the source of the communication. It should be our goal as researchers to determine more precisely the circumstances where evidence will have a significant impact on the persuasive process and where it will make no contribution. As educators it is our responsibility to make students aware of the data options open to them and provide them with the understanding of how each of those options may be used for fullest rhetorical effect.


1. For a summary of this research, see William R. Dresser, "Effects of 'Satisfactory' and 'Unsatisfactory' Evidence in a Speech of Advocacy," Speech Monographs, XXX (August, 1963), 302-306; and "Studies of the Effects of Evidence: Implications for Forensics," AFA Register, X (No. 3, 1962), 14-19.

2. Richard B. Gregg, "Some Hypotheses for a Study of the Psychology of Evidence," paper read at the SAA convention, Chicago, Illinois (December 28, 1964).

3. See, for example, Gregg, Ibid. and William B. Dresser, "The Impact f Evidence on Decision Making," paper read at the SAA Convention, New York, New York (December 28, 1965).

4. The terminology for the model of a persuasive unit herein described is taken from Stephen A. Toulmin, The Uses of Argument (London, 1958). The reader is cautioned not to interpret what follows in terms of Toulmin's logical model. While the terminology is the same, the process being described is quite different. While Toulmin was concerned with the logical development of an argument, we shall be concerned with the rhetorical development of an argument. For a discussion utilizing this latter approach, see James C. McCroskey, "Toulmin and the Basic Course," Speech Teacher, XIV (March, 1965), 91-100.

5. Douglas Ehninger and Wayne Brockriede, Decision by Debate (New York, 1963), Chs. 8, 10, and 11.

6. Ibid., 99-101.

7. While this is usually the case, it should be recognized that under some circumstances "communication distortion" will occur. An audience member may hold a very strong attitude relevant to the statement and a favorable attitude toward the speaker but find the statement to be inconsistent with both of these other attitudes. In order to keep his attitude universe in harmony, he may unconsciously misperceive the statement of the speaker. The explanation of the "communication distortion" phenomenon is based on dissonance theory. See Leon Festinger, A Theory of Cognitive Dissonance (Evanston, 1957).

8. See, for example, Charles S. Goetzinger, Jr., "An Analysis of the 'Validity' of Reasoning and Evidence in Four Major Foreign Policy Speeches, 1950-51" (Unpublished M.S. thesis, Purdue University, 1952) and Paul D. Brandes, "Evidence and Its Use by Selected United States Senators" (Unpublished Ph.D. dissertation, University of Wisconsin, 1953).

9. For a discussion of this defect in the experimental studies of evidence, see James C. McCroskey, "Experimental Studies of the Effects of Ethos and Evidence in Persuasive Communication" (Unpublished Ed.D. dissertation, Pennsylvania State University, 1966), Ch. 1.

10. Goetzinger, Op. Cit., and Brandes, Op. Cit.

Click Here To Go Back To PERIODICALS