Rokeach, in The Open and Closed Mind, discusses two types of language that are commonly found in messages: opinionated and non-opinionated.¹ Non-opinionated statements convey information relating solely to the communicator's attitude toward a particular idea or belief (e.g., "I believe that the United States should withdraw its troops from Viet Nam"). By contrast, opinionated statements convey two kinds of information; they indicate the communicator's attitude toward an idea or belief and his attitude toward those who agree or disagree with him (e.g., "Only a warmonger would oppose the withdrawal of United States troops from Viet Nam" {opinionated-rejection statement}, or "any intelligent person knows that the United States should withdraw its troops from Viet Nam" [opinionated-acceptance statement]).

There have been two recent studies of the effects of this language variable in persuasive communication.² The researchers conceptualized opinionated and non-opinionated statements as a particular type of fear-arousing appeal which emphasizes the undesirable social consequences of failure to conform with message recommendations. They reasoned that in terms of social reinforcement, messages containing only non-opinionated statements are mildly fear-arousing. Conversely, messages containing opinionated statements are strongly fear-arousing because they explicitly indicate that the communicator's disapproval or approval of the receiver is contingent upon the latter's conformity to message recommendations.

The Miller and Lobe study examined the responses of open- and closed minded receivers to messages containing opinionated and non-opinionated statements attributed to a highly credible source.³ They found that the opinionated messages were more persuasive than the non-opinionated ones for both open and closed-minded receivers. The Miller and Baseheart study attempted to extend these results. They predicted the following interaction:

If a source is initially trustworthy (HT), a message containing opinionated statements will result in greater favorable attitude change that a non-opinionated communication, but if the source is initially untrustworthy (LT), a message containing only non-opinionated statements will result in greater favorable attitude change than an opinionated communication.⁴

As hypothesized, an interaction between type of message and initial communicator trustworthiness was observed. Moreover, the earlier findings of Miller and Lobe were replicated. These two studies suggest that no simple generalization can be made about the effects of opinionated and non-opinionated statements. Rather, the impact of this particular type of fear-arousing message may be dependent upon the interaction of a variety of source, message, and receiver variables.

The present study examined one such potentially mediating variable, the relationship between the intensity of a receiver's initial attitude toward a message topic and the relative persuasiveness of messages containing opinionated-rejection statements and non-opinionated statements. Whether cues denoting social disapproval increase persuasiveness is likely to be partially dependent upon the intensity or strength of a receiver's initial attitude toward a message topic. If a receiver initially holds a relatively neutral attitude toward a topic, and is exposed to a message in which the communicator expresses a favorable attitude toward the topic (non-opinionated statements), only mild cognitive inconsistency is likely to occur. But if the communicator expresses a favorable attitude toward the topic and also expresses his attitude toward those who disagree with him (opinionated-rejection statements), much greater inconsistency is likely to occur and thus greater attitude change. This reasoning is formally expressed in the following hypothesis:

1. A message containing opinionated-rejection statements will result in greater favorable attitude change than a message containing non-opinionated statements when subjects are initially relatively neutral toward the topic.

A different effect is likely to obtain if a receiver initially hold an intense attitude toward a topic, and is exposed to a discrepant message (a message that expresses a view contrary from the one held by the receiver). In this situation the receiver's attitude and the communicator's attitude toward the topic are quite discrepant and, thus, considerable cognitive inconsistency is likely to exist In addition, if the receiver is exposed to a discrepant message which also shows the communicator's attitude toward those who disagree with him (opinionated-rejection statements), a grossly discrepant attitudinal situation exists; in fact, such social disapproval cues might be expected to have a boomerang effect. However, the easiest mode of reducing dissonance is to reject or discredit the communicator. This line of reasoning is consistent with Rosenberg and Abelson's balance theory, which states that the smaller proportion of elements, not the larger, is most likely to be changed.⁵ This reasoning is formally expressed in the following hypotheses:

2. A discrepant message containing non-opinionated statements will result in greater favorable attitude change than a discrepant message containing opinionated-rejection statements if the subjects initially hold an intense attitude toward the topic.

3. A discrepant message containing non-opinionated statements will result in higher postcommunication credibility ratings than a message containing opinionated-rejection statements if the subjects initially hold an intense attitude toward the topic.

General Procedures

Subjects were 180 undergraduate students enrolled in an introductory public speaking course at Michigan State University. Subjects were given a packet which contained instructions, the manipulations for the two topics (high and low credibility and low-opinionated and opinionated-rejection language), and posttest measures of attitude and credibility. Each subject received the same conditions for the two topics. The order in which each subject received the topics and the conditions was randomly determined. Two weeks prior to the experiment subjects completed the pretest attitude measures on seven topics, two of which were the experimental topics. A control group completed pretest and posttest measures on both topics but read no message.

Topics

On the basis of a study of 156 possible topics, two were chosen that met the required initial attitudes of the subjects: "United States Policy toward Brazil" (initially relatively neutral attitude) and "The Ku Klux Klan" (initially intensely unfavorable attitude).⁶ Subjects' responses on the pretest verified the results of the previous study. They are reported in Tables 1 and 2.

Independent Variables

Credibility. Credibility was manipulated in order to obtain greater generalizability of results. Several studies indicate that credibility interacts with message variables.⁷ Sources on the topic "United States Policy Toward Brazil" were Charles L. Wilson, former ambassador to Brazil (highly credible) and Lin Tai, Ambassador to Brazil from Red China (low in credibility). Sources on the topic "The Ku Klux Klan" were John F. Kennedy (highly credible) and George C. Wallace, former Governor of Alabama (low in credibility).

Opinionatedness. The basic content of the message was identical for all four conditions on the "Brazil" topic. The experimental message argued against present United States policy in Brazil. Subjects in two of the conditions read a message composed entirely of non-opinionated statements; that is, reference was made only to the communicator's beliefs about the issue, with no information provided about his attitude toward those who would disagree with him. In the two opinionated conditions, subjects read the same basic message with the additional of two opinionated-rejection statements. The specific sentences used were as follows:

Non-opinionated. I can't abide that policy.

Opinionated. Only ignorant people fail to agree that the U. S. policy toward Brazil is working directly counter to the interests of the people f Brazil.

Non-opinionated. I can't abide that policy.

Opinionated. No self-respecting person can abide that policy.

The basic content of the message was also identical for all four conditions on the "Ku Klux Klan" topic. The experimental message argued that some good can be said about the Ku Klux Klan. Subjects in two of the conditions read a message composed entirely of non-opinionated statements and in two other conditions subjects read the same basic message with the addition of two opinionated-rejection statements. The specific sentences used were as follows:

Non-opinionated. Not all KKK members are racial bigots.

Opinionated. People that assert that all KKK members are racial bigots are either uninformed or are bigots themselves.

Non-opinionated. It [the KKK] is one of the strongest forces supporting law and order in many areas.

Opinionated. It [the KKK] is one of the strongest forces supporting law and order in many areas. Those who do not support the KKK are mainly people who do not support law and order.

Both the non-opinionated and the opinionated messages on the two topics were approximately 100 words long. Each message was preceded by an introductory paragraph identifying it as a quotation from the appropriate source which appeared in a New York Times interview. Recent dates of the case of the John F. Kennedy condition which was dated 1960.

Dependent Measures

After reading each message, subjects indicated their attitudes toward the topic. Posttest attitudes were measured with the same four scales used in the pretest. For the Brazil topic the scales used were beneficial-harmful, good-bad, valuable-worthless, and wise-foolish. For the Ku Klux Klan topic the scales used were beneficial-harmful, good-bad, fair-unfair, and wise-foolish. Since attitude was obtained by summing across the scales, a score of 28 indicated the most favorable attitude, a score of 4 of the most unfavorable.

In addition, posttest ratings of credibility using three scales for character and three scales for authoritativeness were obtained. Scales employed for character were friendly-unfriendly, honest-dishonest, and unselfish-selfish. Those employed for authoritativeness were expert-inexpert, qualified-unqualified, and informed-uninformed.⁸ Since the same summing procedure was used, a score of 21 indicated highest credibility on each dimension, a score of 3 lowest on each dimension.

Subjects With Neutral Attitudes Toward Topic

Table 1 summarizes the mean pretest and posttest attitudes scores and the mean attitude change for subjects in the four conditions. An analysis of variance of pretest attitude scores indicated no differences in initial attitudes toward United States Policy Toward Brazil (F < 1).

A two-factor analysis of variance of the attitude change scores showed a significant main effect for initial credibility (F = 18.55, p < .001) and opinionatedness (F = 3.12, p < .05, one-tailed). The credibility by opinionatedness interaction was not significant.

The observed main effect for credibility resulted from the greater persuasiveness of Wilson, the highly-credible source, when compared with Lin Tai the less credible source. This is consistent with the results of many prior studies.

Examination of Table 1 reveals that the significant main effect for opinionatedness supports the first major hypothesis of the study. When the subjects were initially relatively neutral toward the message topic a message containing opinionated-rejection statements produced greater favorable attitude change than a message containing non-opinionated statements.

Table 1 also summarizes the mean posttest credibility ratings on both dimensions for the subjects in the four conditions. Analysis of variance revealed significant main effects for initial credibility on both dimensions: character (F = 86.67, p < .001) and authoritativeness (F = 27.47, p < .001). No other effects approached significance.

The observed post-communication main effect of credibility is consistent with the findings of other studies. That is, an initially highly credible source (Wilson) produced higher postcommunication credibility ratings than an initially less credible source (Lin Tai).

Subjects With Intense Attitudes Toward Topic

Table 2 summarizes the mean pretest and posttest attitude scores and the mean attitude change for the subjects in the four conditions. Analysis of variance of pretest attitude scores indicated not differences in initial attitudes toward the Ku Klux Klan (F < 1).

A two-factor analysis of variance of the attitude change scores showed a significant main effect for initial credibility (F = 10.64, p < .001) and opinionatedness (F = 4.27, p < .01). The credibility by opinionatedness interaction was not significant.

The observed main effect for credibility results from the greater persuasiveness of Kennedy, the highly credible source, when compared with Wallace, the less credible source. Again this is a commonplace finding.

Examination of Table 2 reveals that the significant main effect for opinionatedness supports the second major hypothesis of the study. When the subjects were initially intense toward the message topic a discrepant communication containing non-opinionated statements produced greater favorable attitude change than a message containing opinionated-rejection statements.

Table 2 also summarizes the mean posttest credibility ratings on both dimensions for the subjects in the four conditions. Analysis of variance revealed significant main effects for initial credibility (character, F = 199.42, p < .001; authoritativeness, F = 10.64, p < .001) and opinionatedness (character, F = 6.17, p < .01; authoritativeness, F = 4.09, p < .05). The credibility by opinionatedness interaction was not significant.

Examination of Table 2 reveals that the significant main effect for opinionatedness supports the third major hypothesis of the study. When the subjects were initially intense toward the topic a discrepant message containing non-opinionated statements produced higher postcommunication credibility ratings than a discrepant message containing opinionated-rejection statements.

The findings of this study confirm the hypothesis that the persuasive effectiveness of a message containing opinionated statements depends partly upon the intensity of a receiver's initial attitude toward the topic of the message. When the receiver holds an initially relatively neutral attitude toward the topic, a message containing opinionated-rejection statements results in greater favorable attitude change than a message containing non-opinionated statements. By contrast, when the receiver holds an initially intense attitude toward the topic, a discrepant message containing non-opinionated statements results in greater favorable attitude change and higher postcommunication credibility ratings than a message containing opinionated rejection statements.

This differential effect can be at least partially explained by the "minimum-maximum" hypothesis derived from dissonance theory: the greater the pressure used to elicit behavior contrary to one's attitude (beyond the minimum needed to elicit it), the less his attitude will change. Thus, when the receiver initially holds a relatively neutral attitude, non-opinionated statements are not potent enough to elicit the desired attitude change. Social disapproval cues, however, do produce this minimum level of cognitive inconsistency, and therefore more attitude change. When a receiver initially holds an intense attitude toward the topic, a contrary situation holds: opinionated-rejection statements cross the threshold of tolerable cognitive inconsistency which is manifested in less favorable attitude change and more derogation of the source of the message.

The present findings do not conform to those obtained by Miller and Baseheart.⁹ Those researchers observed a significant credibility by opinionatedness interaction. No such interaction was observed on either of our message topics. The rationale provided by Miller and Baseheart for expecting such an interaction seems plausible. Further, the present investigation differed from theirs in at least two important respects. First our messages were much shorter (approximately 100 words vs. approximately 250 words). Perhaps length of message is a significant intervening variable. Also each of our messages contained fewer opinionated-rejection statements (2 vs. 4). This difference indicates that indeed opinionated is a potent message variable in persuasive communication. These two differences taken together also suggest to the communicator that even when short messages are needed opinionated can play an important role.

1. Milton Rokeach, The Open and Closed Mind (New York, 1960).

2. Gerald R. Miller and Jon Lobe, "Opinionated Language, Open- and Closed-Mindedness and Response to Persuasive Communications," Journal of Communication, XVII (1967), 333-341; Gerald R. Miller and John Baseheart, "Source Trustworthiness, Opinionated Statements, and Response to Persuasive Communication," SM XXXVI (1969), 1-7.

3. Miller and Lobe.

4. Miller and Baseheart, p. 2.

5. Milton J. Rosenberg and Robert P. Abelson, "An Analysis of Cognitive Balancing," in Carl I. Hovland and Milton J. Rosenberg (eds.), Attitude Organization and Change (New Haven, CT: Yale University Press 1960), pp. 112-163.

6. The ideal design for the present study would have been to find one topic on which the subjects could be stratified on the basis of their initial attitude into two groups: (1) relatively neutral and (2) intense. This design would have allowed us to hypothesize an interaction between the subjects' initial attitude and the opinionatedness of the language. The study of 156 possibilities failed to show the existence of such a topic. That is, we could not find a topic on which enough subjects for meaningful statistical analysis were located in the appropriate points in the distribution. Therefore, two topics were needed to test the hypotheses. This allowed us to predict only main effects and required us to analyze the data as two different experiments.

7. See, for example, James C. McCroskey and Robert E. Dunham, "Ethos: A Confounding Element in Communication Research," SM, XXXIII (1966), 456-463.

8. These scales were selected on the basis of prior factor analytic studies. See James C. McCroskey, "Scales for the Measurement of Ethos," SM, XXXIII (1966), 65-72.

9. Miller and Baseheart.

Click Here To Go Back To PERIODICALS