James C. McCroskey and R. Samuel Mehrley

Rhetorical theorists commonly assert that message disorganization and nonfluent delivery reduce persuasive effectiveness. But empirical data provide only ambiguous support for this generally accepted view. The purpose of the present study was to test hypothesized effects of organization and fluency on attitude change and source credibility. A review of the literature clarifies the relationship between these variables and provides a theoretical schema for the hypotheses that we tested.



The research literature provides scant support for the belief that delivery contributes significantly to attitude change.1 Bettinghaus conducted the only study that found that effective delivery contributes to an immediate positive effect.2 Conversely, McCroskey and Arnold found no main effect on immediate attitude change attributable to delivery quality for live, video-taped, or audio-taped speakers.3 These researchers did find, however, a significant main effect on delayed (four weeks) attitude change attributable to delivery quality. Moreover, McCroskey and Arnold found that when good delivery quality was coupled with good message quality, it produced immediate positive attitude change; but when coupled with poor message quality, it did not. The delivery manipulation in these studies included gesture, movement, facial expression, eye contact, vocal rate, inflection, and nonfluency. In two studies specifically concerned with the effect of nonfluency, Miller and Hewgill4 and Sereno and Hawkins5 found no effect on immediate attitude change attributable to degree of nonfluency.

The above studies provide only minimal support for the belief that the quality of the delivery, particularly fluency, has an effect on attitude change. Nevertheless, these very same studies provide very strong support for the belief that the quality of the delivery has a significant effect on source credibility. Bettinghaus,6 McCroskey,7 McCroskey and Arnold,8 Miller and Hewgill,9 and Sereno and Hawkins10 all found that poor delivery detracts from perceived credibility.


It is difficult to cull any generalizations from a few studies that have investigated the effect of the quality of organization on immediate attitude change. Smith found that minor disorganization had no significant effect on attitude change but major disorganization significantly reduced the amount of attitude change produced.11 Thistlethwaite, de Haan, and Kamenetzky manipulated the quality of transitional material (a variable thought to be related to organization) but observed no significant effects on attitude change.12 Thompson observed no significant differences on attitude change between a well-organized speech an done in which the sentences in the speech were randomly ordered.13 Thompson's result, however, is difficult to interpret because his dependent variable measure was only a single seven-point scale.

Only one study of the effects of organizational quality on perceived source credibility has been reported. Sharp and McClung14 exposed subjects to the speeches developed by Thompson,15 measuring perceived credibility before and after exposure. They found that the organized message did not produce significantly lower credibility ratings, but the disorganized message did. It is important to note that both groups of subjects perceived the source to be moderately high in credibility prior to message exposure. Thus, one interpretation of their results is this: a disorganized message will lower the credibility of an initially highly credible source. The effect of a disorganized message on the credibility of an initially less credible source is not clearly indicated by these results.

One additional study should be reviewed, but it is difficult to classify it as either "delivery" or "organization" research. Baker16 manipulated "disorganization cues" and observed their effect on attitude change and perceived credibility. This study appears to be concerned with organization (subjects did perceive differences in "organization" as measured by a semantic differential type scale), but the following operational definition of "disorganization cues" is similar to that employed by other researchers studying effects of delivery:

These cues, or verbal/nonverbal suggestions of the speaker's difficulty in presenting his ideas systematically, included apologies for not remembering the next point, apologies for including an idea out of sequence, explanations for not being able to place materials in proper sequence, and silent pauses.

These disorganization cues did produce differentially perceived organization, but they also may have produced differentially perceived delivery quality. This, of course, is speculative because perceived delivery quality was not measured.

This manipulation, regardless of how it is defined ("disorganization or a combination of "poor delivery" and "disorganization"), did not have a significant effect on attitude change. Both experimental groups shifted their attitudes significantly in the desired direction, but there was no significant difference between the groups on attitude change.

The effects of the experimental manipulation upon source credibility are difficult to interpret because only scores summed across three dimensions of credibility (competence, trustworthiness, and dynamism) are reported. The results of the analysis of these summed scores indicated that the "organized" speaker's credibility significantly increased while no significant increase was observed for the "disorganized" speaker. Pretest data indicated that the speaker in each condition was perceived as moderately credible.

The Baker study (if it is interpreted as a manipulation of organization) and the Sharp and McClung17 study, taken together, suggest the following low level generalizations: (1) A disorganized message will owner the credibility of a highly credible source. (2) An organized message will increase the credibility of a moderately credible source (possibly also a less credible source).


A tempting conclusion from previous research is that organization of a persuasive message and fluent delivery of that message have minimal impact on receiver's attitudes. This conclusion should be avoided; none of the cited studies has manipulated these two variables conjunctively.

The major purpose of this study was to test the assumption that message organization and fluent delivery of that message can interact in the persuasive equation. It is posited that message organization and fluency of delivery are two species of the same genus. That genus may be called "aids to clarity." Many teachers and some theorists posit a relationship between clarity of message and attitude change. The intervening variable in this relationship is comprehension. While numerous studies have found that comprehension is not sufficient to produce attitude change without the presence of other motivating factors, these studies do not permit the conclusion that there is no relationship between comprehension and attitude change. Comprehension can be viewed from at least two positions. The first position views comprehension as a necessary but not a sufficient antecedent condition to attitude change. Consequently, any variable that prevents the comprehension of message content will also reduce attitude change. This position views comprehension as an either/or proposition. That is, either the message is comprehended or it is not. Thus, a variable which reduces the clarity of a message will prevent comprehension, and the result will be little or no attitude change. Hence, if two such clarity reduction variables are simultaneously present there will be no greater reduction in attitude change than if only one such variable is present.

The second position views comprehension on an extremely unclear to extremely clear continuum. On this continuum one variable can cause the message to be extremely unclear. However, the introduction of a second variable which would normally in and of itself reduce message clarity will not significantly reduce attitude change beyond that already singularly produced by the first variable. The first variable will already have insured that the audience will respond at or near the unclear end of the comprehension continuum.

From both of these viewpoints the following conclusion is reached: If a message is either very badly organized or presented in a very nonfluent manner, comprehension should be reduced sharply with an accompanying reduction in attitude change; but, if a message is both very badly organized and presented in a very nonfluent manner the degree of comprehension should not be reduced substantially more than if only one of these undesirable elements is present.

Based on the above discussion the following interaction between message type (organized and disorganized) and form of delivery (fluent and nonfluent was hypothesized.

1. An organized, fluently presented persuasive message will produce significantly more attitude change than a disorganized-fluently presented message, and organized-nonfluently presented message, or a disorganized-nonfluently presented message; but the latter three conditions will not differ significantly from each other in the production of attitude change.

A secondary purpose of the study was to confirm and extend the generalizations that were culled from the Baker and the Sharp and McClung studies. Specifically, the following main effects for fluency and organization on terminal credibility were hypothesized:

2. A source who has presented an organized version of a message will be perceived as more credible than will a source who has presented a disorganized version of the same message.

3. A source who has presented a fluent version of a message will be perceived as more credible than will a source who has presented a nonfluent version of the same message.

In addition to testing the above hypotheses, we also explored possible interactions between initial credibility, fluency, and organization on attitude change and post-communication credibility.


Independent Variable: Organization

The experimental stimuli included two versions of a speech developed by Thompson advocating a guaranteed annual wage for all industrial employees.18 The speech included seven points and their supporting materials: two points in the introduction, three in the body, and two in the conclusion. Nine of ten judges employed by Thompson rated the "organized" version as adequately or very adequately structured. Thompson created the "disorganized" version by randomizing the sentences within teach of the three main divisions of the speech (introduction, body, conclusion). All ten of Thompson's judges rated this version either inadequately or very inadequately structured.

Independent Variable: Fluency

Each version of the speech was recorded on two audio-tapes. On one tape the speech was presented with no nonfluencies or other serious delivery flaws. On the second tape the speech was presented in a similar manner except that a nonfluency was introduced approximately every fourteenth word. This procedure was comparable to one employed by Miller and Hewgill.19 Both vocalized pauses and repetitions were included, in alternating order. A total of one hundred nonfluencies were included, fifty vocalized pauses and fifty repetitions, in the "nonfluent" version of the "organized" speech and the same number in the "nonfluent" version of the "disorganized' speech.

Independent Variable: Source Credibility

Initial source credibility was manipulated by means of the following introductions, given in writing on the pretest booklet:

Highly Credible. The first speaker that you will hear is Thomas A. Wilcox, Vice-President in charge of labor relations for the Ford Motor Company. Mr. Wilcox is a graduate of the University of Michigan in Business Administration. He has been employed by Ford Motor Company in their Division of Labor Relations since his graduation from Michigan in 1935. He has been called the "Father of the Guaranteed Annual Wage" because of his work with the program adopted by the Ford Motor Company in 1955. The speech you will hear was delivered at the national convention of the National Association of Manufacturers in 1963. The speech concerns the desirability of the Guaranteed Annual Wage.

Less Credible. The first speaker that you will hear is Thomas A. Wilcox. When this speech was recorded Mr. Wilcox was an undergraduate student in Speech 101. He was requested to withdraw from the University during the term in which this speech was presented by the office of student affairs, but we do not know why. We do not know what has become of him since. This speech was given as a regular Speech 101 assignment. The topic of the speech is the Guaranteed Annual Wage. It is presumed that Mr. Wilcox knew little more about this subject than most students enrolled in Speech 101. It should be stressed that the Speech Communication Research Laboratory neither endorses nor condemns the views expressed in the speech, it is being used only because it is appropriate for this research project.


These three manipulations produced a 2 X 2 X 2 design. A pretest for attitude toward Guaranteed Annual Wage was administered along with several other concepts at the beginning of the period in which the experiment was administered to the subjects. Pretests for the three dimensions of credibility--authoritativeness, character, dynamism--were administered immediately after the subjects had been exposed to the introductions of the speaker. Attitude and credibility posttests were administered immediately following the speech.


Subjects were students enrolled in the public speaking course (Speech 101) at Michigan State University during the fall term 1967. A total of 352 subjects distributed across the eight experimental conditions (n in the conditions ranged from 43 to 46) were included in the final analysis of the data. The experiment was administered the second day of the term under the guise of a study concerned with "style" in order to avoid sensitization to either delivery or organization. The experimental treatments were administered at the beginning of the class period by graduate assistants in Speech at MSU.

Measurement and Administration

Each subject was presented a packet of materials. The first page of the packet included a statement indicating that the project was being conducted under the auspices of the Speech Communication Research Laboratory, briefly discussed the need for the importance of research in speech, and informed the subject that the current project concerned "style" in speech. The following pages included semantic differential type scales for several concepts, one of which was "There Should Be A Guaranteed Annual Wage For All Industrial Employees." Scales employed were I Agree-I Disagree, Good Idea-Bad Idea, Yes-No, True-False, Right-Wrong, Correct-Incorrect.

After all subjects in a given class had completed these differentials, they were asked to turn to the next page. That page included the initial credibility induction and pretest semantic differential scales for the alleged source. Scales developed by McCroskey20 for "authoritativeness" and "character" were included along with six scales for "dynamism" which were found to have high loadings on this dimension by Berlo, Lemert, and Mertz.21

The appropriate tape-recorded speech was played after all of the subjects in the class had completed the credibility pretest. When the speech was finished, the subjects were directed to turn to the final pages in their packets which included the posttests for attitude and credibility and six speech-evaluation scales. These scales were Clear-Confused, Good Delivery-Poor Delivery, Good Content-Poor Content, Organized-Disorganized, Objective-Subjective, and Well Supported-Poorly Supported.

A control group (N = 31) completed the pretest and posttest measures of attitude and credibility but were not exposed to an experimental stimulus. Half of the control group was exposed tot he highly credible initial introduction, the other half was exposed to the less credible initial introduction.

Statistical Analysis

Analysis of the pretest attitude measures indicated the groups were not all equal; therefore, the posttest attitude measure was analyzed by means of three-factor analysis of covariance with the pretest attitude measure serving as the covariate. The differences between the pretest and posttest measures on each of the three credibility dimensions were analyzed by means of a three-factor analysis of variance. The scores obtained from each of the speech evaluation scales and the total for the six scales were also analyzed by means of three-factor analyses of variance. The .05 criterion was set for significance in all statistical tests.


Credibility Inductions

Examination of the results of the pretest measures of credibility indicated that the high and low credibility indications were successful and that there was no significant difference on any of the credibility dimensions among groups exposed to the same induction.22 While the differences between inductions were significant for all three dimensions, the manipulation produced the largest difference on perceived authoritativeness and smallest difference on perceived character. This finding is consistent with what was expected on the basis of the content of the inductions.

Attitude Change

Analysis of covariance of the posttest attitude scores produced two significant effects, those for source credibility (F = 6.13, p < .05) and for the interaction of fluency and organization (F = 5.62, p < .05). The effects attributable to initial credibility were consistent with those found by numerous previous researchers: the highly credible source produced more favorable attitude change than the less credible source.

Table 1 reports the covariance adjusted mean posttest attitude scores. (Because no interactions including the initial credibility variable were observed, the scores for the two credibility levels are combined.) Examination of these results indicates that the observed significant interaction was consistent with the results predicted by hypothesis 1. The well-organized, fluently-presented message produced significantly more attitude change than any other condition, and the other three conditions were not significantly different from each other.


Covariance Adjusted Posttest Attitude Scores

Experimental Condition







Attitude Score*





*Range of possible scores from 6 (lease favorable) to 42 (most favorable). Mean pretest score was 20.8.

Analysis of variance of the change scores on authoritativeness produced two significant main effects, those for initial credibility (F = 103.05, p < .05), fluency (F = 30.94, p < .05), and one effect approaching significance, that for organization (F = 3.78, p < .10). Since the observed difference is in the direction predicted by hypothesis 2, a one-tailed test is appropriate. With this test, the organization effect is significant at the .05 level. Table 2 reports the observed shifts on the authoritativeness measure.

Analysis of variance of the change scores on the character and dynamism measures produced in each case two significant main effects, those for initial credibility and fluency, those for initial credibility and fluency. (Character, credibility--F = 29.64, p < .05, fluency--F = 20.16, p < .05; dynamism, credibility--F = 133.44, p < .05, fluency--F = 50.88, p < .05). Observed mean shifts on credibility dimensions are reported in Table 2.


Change Scores on Credibility Dimensions

Credibility Dimension

Independent Variable Authoritativeness Character Dynamism
Initial Credibility
High -3.0 -.5 -8.5
Low 4.1 2.5 -2.1
Difference 7.1* 3.0* 6.4*
Organized 1.2 .9 -5.4
Disorganized -.5 1.5 -5.3
Difference 1.7* .6 .1
Fluent 2.5 2.2 .1
Nonfluent -1.4 -1.8 -10.5
Difference 3.9* 4.0* 10.6*


On each of the credibility dimensions the observed differences attributable to fluency and organization were in the direction predicted by hypotheses 2 and 3. Hypothesis 3 (fluency) was confirmed for all three dimension; however, the effects of organization predicted by hypothesis 2 were significant only for the authoritativeness dimension.

In each case the significant effect attributable to initial credibility was produced by the tendency of the highly credible source to lose (or gain less) credibility and the less credible source to gain (or lose less) credibility. While this would appear on the surface to be the normal measurement regression effect, the fact that the control group did not experience a similar effect suggests that this result was produced by the experimental message.

Speech Ratings

Analyses of the results from the speech-evaluation scales showed significant main effects for organization and fluency on four individual scales and for the total score. No effects for initial credibility were observed on any scale, and effects of all variables were not significant for two scales--Well Supported/Poorly Supported and Objective/Subjective. Table 3 reports the mean scores observed for the various fluency-organization cell combinations on the scales which produced significant results. These results indicate that subjects perceived the organization and fluency inductions as intended.


Speech-Evaluation Scale Ratings

Experimental Condition








Clear-Confused* 5.9 4.0 4.8 2.7
Good Delivery-Poor Delivery** 5.9 2.7 5.3 2.1
Good Content-Poor Content*** 5.8 4.7 5.3 4.0
Organized-Disorganized**** 5.9 4.5 5.0 3.2
Total of 6 Scales**** 32.6 23.8 29.6 20.0

All F's are significant.

*Organization F = 41.84, Fluency F = 102.73.

**Organization F = 11.01, Fluency F = 372.03.

***Organization F = 6.70, Fluency F = 43.71.

****Organization F = 41.82, Fluency F = 83.36.

*****Organization F = 25.14, Fluency F = 155.58.


The results of this study confirmed the hypotheses. The presence of either serious disorganization or extensive nonfluencies was sufficient to significantly reduce the amount of attitude change produced by a speaker; but the presence of both of these detracting element produced no greater reduction of attitude change than did the presence of either one alone. However, subjects exposed to the disorganized nonfluent message rated the message significantly more "confused" than the subjects exposed to either the organized, fluent message. Apparently "clarity" is not linearly correlated with attitude change but rather it may operate as a "permissive" variable. That is, if "clarity" reaches a certain, at present unknown, level, sufficient comprehension can occur so that maximum attitude change can occur; but, if "clarity" is below that level, reducing it still further may not have a direct effect on attitude change. This possibility seems worthy of systematic investigation, especially in view of the importance attached to clarity by many teachers of communication.

The results of the present study tend to indicate that both fluency and organization are important variables affecting credibility. However, this study's design was such that it provides useful information about the effects of extremes in organization and fluency, but says nothing about the myriad gradations in between. It is not justifiable to posit a linear relationship between these variables and attitude change or credibility on the basis of one study. Further studies which examine gradations of these variables are needed. However, on the basis of the findings of this study, as well as others cited earlier, an appropriate conclusion is: serious disorganization and extensive nonfluencies seriously restrict the amount of attitude change a communicator can produce and substantially reduce the communicator's credibility. Future research needs to determine if lesser degrees of disorganization or nonfluency result in comparable effects on attitude change and credibility.


1. Most of the previous research on delivery and organization have employed comprehension as the dependent variable, and therefore are not directly relevant here. In addition, numerous studies have been reported on the primacy-recency problem, but these studies have been concerned with "type" of organization rather than "quality" of organization and, thus, rather than "quality" of organization and, thus, are not directly relevant to the present study.

2. Erwin P. Bettinghaus, "The Operation of Congruity in an Oral Communication Situation," SM, XXVIII (1961), 131-142.

3. These studies are discussed in James C. McCroskey, "Studies of the Effects of Evidence in Persuasive Communication," Report SCRL 4-67 (Speech Communication Research Laboratory, Michigan State University, 1967). (Multilithed.)

4. Gerald R. Miller and Murray A. Hewgill, "The Effect of Variations in Nonfluency on Audience Ratings of Source Credibility," QJS, L (1964), 36-44. The attitude change results were not reported in this paper. The results were obtained from the authors.

5. Kenneth K. Sereno and Gary J. Hawkins, "The Effects of Variations in Speakers' Nonfluency upon Audience Ratings of Attitude toward the Speech Topic and Speakers' Credibility," SM, XXXIV (1967), 58-64.

6. Op. cit.

7. Op. cit.

8. Op. cit.

9. Op. cit.

10. Op. cit.

11. Raymond G. Smith, "An Experimental Study of the Effects of Speech Organization upon Attitudes of College Students," SM, XVIII (1951), 292-301.

12. Donald L. Thistlethwaite, Henry de Haan, and Joseph Kamenetzky, "The Effects of 'Directive' and 'Non-Directive' Communication Procedures on Attitudes," The Journal of Abnormal and Social Psychology, LI (1955), 107-113.

13. Ernest C. Thompson, "An Experimental Investigation of the Relative Effectiveness of Organizational Structure in Oral Communication" (unpublished Ph.D. dissertation, University of Minnesota, 1960).

14. Harry Sharp, Jr., and Thomas McClung, "Effects of Organization on the Speaker's Ethos," SM, XXXIII (1966), 182-183.

15. Op. cit.

16. Eldon E. Baker, "The Immediate Effects of Perceived Speaker Disorganization on Speaker Credibility and Audience Attitude Change in Persuasive Speaking," Western Speech, XXIX (1965), 148-161.

17. Op. cit.

18. The writers wish to express their appreciation to Dr. Thompson for permitting them to use the speeches he developed in this study. Copies of the speeches are available in Thompson, op. cit.

19. Op. cit. pp. 37-38.

20. James C. McCroskey, "Scales for the Measurement of Ethos," SM, XXXIII (1966), 65-72.

21. David K. Berlo, James B. Lemert, and Robert J. Mertz, "Dimensions for Evaluating the Acceptability of Message Sources" (Department of Communication, Michigan State University, 1966). (Multilithed.) Scales employed were Aggressive-Meek, Energetic-Tired, Bold-Timid, Emphatic-Hesitant, Active-Passive, and Forceful-Forceless.

22. With a possible range of scores on each credibility dimension of 6-42, the mean perceived initial authoritativeness of the highly credible source was 36.8, mean perceived dynamism was 35.2. For the less credible source, the means were 16.3, 23.1, and 18.7 respectively.

Click Here To Go Back To PERIODICALS