- “1'- Vfi— - " .9. ':C".“‘ I)”. .' ,A l ‘ ‘-D' e ‘ ‘ w w.» ' -. “ .....‘. ..‘ yifi‘l‘ifi'wnbx‘lfl‘f “.33.: -'I‘E.A“Y'T‘I¥rv.l.€1u if-i.:.;-.t.:.:,.\1Ee; l ‘ ' OPEN - ANU CLOSED-MiNDED‘NESS AND THE SELF-PERSUASION PH’ENOMENON Thesis for the Degree of Ph. D. MICHIGAN STATE UNIVERSITY HOWARD REBACH .1968 ihhblb This is to certify that the thesis entitled Open- and Closed-Mindedness and The Self-Persuasion Phenomenon presented by Howard Rebach has been accepted towards fulfillment of the requirements for Ph.D. Communication degree in gem/o? ff 7% // Major professor Date November 15, 1968 0-169 ABSTRACT OPEN- AND CLOSED-MINDBDNBSS AND THE SELFBPBRSUASION PHBNOMENON by Howard Rebach This study~hypothesized that simple commitment to encode. a counterattitudinal message would lead to more self-persuasion among high dogmatics while with actual encoding of the counterattitudinal messages there would be greater self-persuasion among low dogmatics than among high dogmatics. With assignment of subjects on the basis of a median split of the dogmatism scores, a non-significant trend in the predicted directionwas observed for the self-persuasion scores. Elimination of the middle third of the distribution of dogmatism scores indicated that _ greater self-persuasion occurred for the high dogmatics in the simple coumitment condition and that greater self-persuasion occurred for the low dogmatics in the encoding condition. Internal analysis also suggested that self-persuasion may be depressed when persons are highly ego-involved with the topic or when there are other messages in the environment that support their initial stand. OPEN- AND CLOSED-MINDEDNESS AND THE SELF-PERSUASION PHENOMENON By rMC-I f ,3, Howard Rebach A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Communication 1968 Accepted by the faculty of the Department of Communication, College of Communication Arts, Michigan State University, in partial fulfillment of the requirements for the Doctor of Philosophy degree. ,4 / /7 / ‘ I gig/2A5! (77.47 .X XXI/4711 Director of Thesis Guidance Committee jflmgfl? A; 2711 g/{ ,Chairman . fimfig‘ M111 /( €4.4ch ACKN OWLE DGMBN TS My thanks go, first and foremost, to Dr. Gerald Miller,‘Who served as chairman of my advisory and dissertation committee and to the members of my committee: Dr. Randall Harrison, Dr. Hideya Kumata, Dr. Bradley Lashbrodk, and Dr. James L. Phillips. Thanks also to: Jack Baseheart, Walter Combs, Brenda Dervin, and Win Rowe for allowing me to use their students as subjects; to Mrs. Shirley Sherman for typing the final manuscript; and, last but not least, to Dr. David Berlo, chairman of the Department of Communication, for his support and interest throughout: my graduate years. ii TABLE OF CONTENTS CHAPTER I RATIONALE AND HYPOTHESIS . . . . . . II METHOD . . . . . . . . . . . . . . III RESULTS . . . . . . . . . . . . . . IV DISCUSSION . . . . . . . . . . . . . BIBLIOGRAPHY O O O O O O O O O O O O O O O 13 21 36 H8 TABLE LIST OF TABLES Item intercorrelations of Opinion items . . . . . . . . . 1% Distribution of Ss' dogmatism scores (N = 123) . . . . . lS Attitude change score for subjects in exPerimental and contrOI groups 0 o o o o c o o o o o o o c o o o o o 22 Comparison of mean attitude change scores of treatment groups to all controls: Dunnett's test . . . . 22 Comparison of mean attitude change scores of treatment , groups to control groups at each level of Dogmatism: 10 ll 13 Dunnett's test 0 o o o o o o o o o o o o o o o o o o o o 23 Schematic of predicted differences in mean attitude change SCOreS o o o c c o o o o o o o c o o o o 24 Analysis of Variance Summary: The effects of dogmatism level and treatment condition on attitude change scores 0 o o o c o o o o c c c o o o o o c o o o 25 Analysis of Variance Summary: The effects of dogmatism level and treatment condition on attitude change--upper and lower 3rds of dogmatism range . . . . . . . . .' . . . 26 Frequencies of positive opinion shift (+), negative shift (-), and no change within each cell, All.§s . . . . . . . 28 Frequencies of positive opinion shift (+), negative shift (-), and no change within each cell, upper and lower thirds Of dogmatism range only 0 o o o o o o o o o o o o 29 Analysis of Variance Summary: The effects on attitude change of treatment conditions, level of dogmatism, and IGVEl Of ego-involvement. o o o c o o c o o c c c o o o c 31 Analysis of Variance Summary: Analysis of the effects of Ss' prior position on attitude change scores . . . . . . 33 Mean attitude change scores of all groups when divided with reSPeCt to initial pOSitiOn o o o o o o o o o o c o o o o 35 iv LIST OF APPENDICES Page APPENDIXA PRBTEST.............. 50 APPENDIXB POSTTEST ............o 52 CHAPTER I RATI ONALE AND HYPOTHESIS Communication researchers have devoted much attention to the phenomenon of selfepersuasion. Considerable prior research (e.g., Festinger and Carlsmith, 1959; Janis and King, 1954, Janis and Gilmore, 1965, Kelman, 1953) reveals that persons who agree to encode counter- attitudinal messages shift their attitudes in the direction of the role-played position. Obviously, however, this generalization represents an oversimplification of the self-persuasion phenomenon, for a number of relevant variables will determine the magnitude of observed change. The present study deals with one potentially relevant personality variable: the relative cpen- or closed-mindedness of the counter- attitudinal communicators. Moreover, the study examines possible differences between open- and closedwminded counterattitudinal communicators at two points of time in the selfepersuasion process: first, following simple commitment to engage in counterattitudinal communication, and second, following actual encoding of a counter- attitudinal message. Specifically, it is posited that open- and closedrminded persons will differ as to the point in time that they will manifest selfhpersuasion effects. The grounds for this interaction prediction are developed in the remainder of this chapter. Dissonance and Incentive Explanations of the Self-Persuasion Phenomenon: One of two eXplanatory principles are generally invoked to account for the selfepersuasion effect. Dissonance theorists (e.g., Pestinger, 1957; Brehm and Cohen, 1962) hold that a state of cognitive imbalance or dissonance exists if a person is aware of two prepositions or cognitive elements, one of which implies the obverse of the other. In the counterattitudinal role-playing situation a person is aware that "I believe X," and he is also aware of the conflicting cognition that "I advocated (or am committed to advocate) not-X." Since the counterattitudinal encoding task presents the person with a £225 accgmpli whose existence cannot be denied and since the existence of dissonance results in motivation for dissonance reduction, the prediction is that subjects will alter their beliefs to be more consistent with the role-played position. Altering one cognitive element toward greater consonance with another--i.e., changing opinion-- is but one mode of dissonance reduction. Another mode is to add cognitive elements that bring the original two cognitions into a consonant relationship. When peeple feel they have no choice but to perform the belief discrepant behaviors, they may reduce the inconsistency by adding the cognition that they were forced to comply. Research indicates that attitude change is more likely to occur under conditions of high rather than low or non-existant choice of participation (Brehm, 1956; Brehm, 1959; Rabbie, Brehm and Cohen, 1959; Penner, Pitch, and Weick, 1966). Similarly, a person who is offered a large reward for counter- attitudinal advocacy will experience little motivation to change opinion. Rather, his dissonance is reduced by the consonant cognition that the reward was the reason for his behavior.‘ Low justification does not result in such a consonant cognition and dissonance is likely to be reduced by altering one's attitude; i.e.,by manifesting self; persuasion (Festinger and Carlsmith, 1959; Carlsmith, Collins, and Helmreich, 1966; Greenbaum, 1966). Thus, the generalization offered by the dissonance theorists is that when a.person commits himself to expend effort in counter- attitudinal behavior, particularly if there is little justification for such behavior, dissonance is aroused. In turn, this dissonance produces pressures toward consonance which are manifested in a change of attitude toward the belief discrepant position (Cohen, 1959; Brehm and Cohen, 1962, pp. 73-78). Set_against the dissonance theory rationale is that of the so- called incentive theorists. The basic idea underlying incentive theory is far simpler and has the added feature of squaring with common sense predictions: incentive theory states that greater rather than lesser reward is more likely to result in selfepersuasion. Under conditions of _ greater rewards the subject is more likely to rehearse the new position more fully, and, as some have held (e.g., Hovland, Janis, and Kelley, 1953) suCh rehearsal festers the learning of the counterattitudinal position which is required for attitude change. Therefore, with resPect to the self persuasion situation, incentive theory argues that the more effectively the person encodes the counterattitudinal communication, the more likely he is to adopt the new position. In addition, a large reward will.motivate him to do a more adequate job of encoding, and more self persuasion is likely to occur. According to this 'incentive theory' when a person accepts the task of improvising arguments in favor of a view at variance with his own personal convic- tions, he becomes temporarily motivated to think up all the good positive arguments he can, and at the same time suppresses thoughts about the negative arguments which are supposedly irrelevant to the assigned task. This 'biased scanning' increases the salience of the positive arguments and therefore in- creases the chances of acceptance of the new attitude position. A gain in attitude change would not be expected, however, if resentment or other interfering affective reactions were aroused by negative incentives in the role-playing situation. (Janis and Gilmore, 1965) This biased scanning argument is supported by the work of Janis and King (195“), Janis and Gilmore (1965), Rosenberg (1965), and.Elms and Janis (1965). The issue, then, is whether the selfepersuasion effect is based on rehearsal of counterattitudinal arguments or on a desire to maintain cognitive balance. While the two viewpoints have been presented as a dichotomy, some recent work hints that it may not be an either/or matter (Carlsmith, Collins and Helmreich, 1966; Greenbaum, 1966). In addition, some researchers have attempted to add specificity to the :findings by investigating ways that differences in an individual's belief structure may affect the outcomes of counterattitudinal roledplaying (Hunt and Miller, 1968; Greenbaum, 1966). The present study considers that these two approaches are related. The suggestion is that. given the organization of the individual's belief system, some situations allow him the opportunity to rehearse the belief-discrepant position free of the threat of imbalance, while other situations lead to closure and rejection of the new position in favor of cognitive balance. Single Commitment LPost-encodingr, and Dissonance and Incentive Bylanations-u-Some Possible Relationships: As has been noted, dissonance theory eXplains the effect of counterattitudinal encoding in terms of motivation toward cognitive consistency. Brehm and Cohen (1962, p. 255) assert that mere commitment to the role-playing task is dissonance producing and that subsequent pressures to restore consonance lead to Opinion change (Rabbie, Brehm, and Cohen, 1959). If, however, as incentive theory posits, self-persuasion results from the rehearsal of the belief-discrepant position, actual encoding should result in greater self-persuasion than simple commitment. Studies by Janis and Gilmore (1965) and Elms and Janis (1965) found that actual encoding produced more self persuasion than simple commitment. Another of Janis and Gilmore's findings may assist in reconciling the two positions. These researchers manipulated level of justification as well as simple commitment is. actual encoding. Though Janis and Gilmore found the overt role-playing superior, they found that dissonance predictions were supported in the simple commitment condition. That is, in the simple commitment condition, self-persuasion was negatively related to justification, while in the overt role-playing condition, the two variables were positively related. Thus the relationship posited in this study--one consistent with the reconciliation presented by McGuire (1965)--is that dissonance interpretations are most applicable to simple commitment situations, while incentive theory predictions are more closely associated with the post encoding effects of counterattitudinal advocacy. But once again, this is rather a sweeping generalization. It assumes that commitment to a counterattitudinal task is dissonance- producing for everyone, and that actual encoding does indeed lead to biased scanning of the part of all counterattitudinal communicators. While these situational inducements apparently have their effects, they may be mediated by the characteristics of the individual's belief system: the individual's characteristic mode of processing information may interact with situational demands to determine the outcome. This possibility will be considered at greater length in the next section which will culminate in the presentation of hypotheses. The Mediating;Effects of Open- and ClosedpMindedness on the Self- Persuasion Paradigg: According to deeach (1960, p. 68) an individual's belief system is in the service of two very important and potentially conflicting values: A person will be open to information if the need to know is paramount, and closed to information if he is primarily concerned with warding off threat. Striving for cognitive clarity represents movement in the direction of uncertainty reduction. To reduce uncertainty, the person must be able to experiment with new courses of action, pursuing them as long as they seem to be working and abandoning them if they prove to be unworkable. If the person controls his own outcomes, he can choose to continue or to reject the new course of action as he sees fit. There is no threat to the stability of his belief system. Opposed cognitive elements are imbalancing only if they raise uncertainty to a level that makes coping behavior difficult. Extant balance theories apparently fail to make a distinction between imbalance and the threat of imbalance. They remove from the individual choice and control of behavior. For Pestinger, an individual is said to be in a dissonant state-unbalancede-when faced with mutually contradictory propositions. The position taken here is that imbalance occurs when the individual is unable to cope with a situation. Two contradictory propositions create imbalance only if the individual cannot cope with the existing state of affairs. 0n the other hand, some people appear to relish and to seek out ambiguous situations and are usually able to cepe with them. For these people, such situations carry no threat to equilibrium, and offer opportunities for growth and development. It is when the situation threatens to, or has in fact exceeded the in- dividual's ability to cope that other adaptive, threat escaping mechanisms--withdrawal, denial, etc.-- are invoked. Thus, as Ackerman (1958, p. 76-78) has suggested, it is not SO‘mUCh a question of whether or not a balanced state is achieved; it gill be achieved. The question is how this balance will.be achieved. Thus, apparent inconsistencies of behavior are not ruled out. An individual may deviate from his normal path to do something different and to try out new ideas. The Opportunity to invoke a host of psychological adaptive mechanisms enables the individual to maintain balance. Lest this view be taken to imply that this process is always conscious and rational, no such suggestion is intended. The rehearsal, consideration, and reasoned acceptance or rejection of arguments, the altering of beliefs to retain cognitive consistency, the processes of avoidance, distortion, and denial are all potentially functional cognitive processes. For the individual, they all serve his need to achieve . greater cognitive clarity while at the same time warding off threats to his stability. At various times, cognitive consistency may be a function of both drives. Sometimes it may help the individual reach ‘ greater clarity, at other times it may be the only avenue open to him to maintain his equilibrium. This latter point is especially crucial to the present study. If the person is unable to extricate himself from a selfbpersuasion situation and if he is unable to invoke other psychological mechanisms, attitude change in the direction of the discrepant position may be the only means at his diaposal fer warding Off threat to the stability of his belief system. In the extreme, ROkeach's definition of an ideological conversiono-the party line change-- is an example of this kind of change. It is assumed that Open- and closed~minded.persons differ in the extent to which they perceive the world as generally hostile or friendly. This belief as to the friendliness of the world is a very primitive one. Emerging from such a belief is an orientation that the person develOps toward authority and the self, as well as toward various other beliefs and disbeliefs. If an individual perceives the world as hostile and threatening, he will have a lower threshold for assigning a threatening meaning to a.particular situation, and consequently a significantly greater number of situations will be perceived as threatening to the belief system. This circumstance will enable the more open person to handle greater amounts of information, since fewer inputs will threaten imbalance. The open person is thus more likely to be able to tolerate greater ambiguity or inconsistency. Thus, the more Open a person's belief system, the more receptive he will be to infOrmation seemingly incon- sistent with a particular belief. When compared to the closed person, Open individuals will more frequently be motivated by the drive toward cognitive clarity. What seems to be involved in open- or closedsmindedness is a cluster of traits descriptive of the belief system, the person's capacity to make certain kinds of connections among beliefs, and his ability to consider alternatives free of the threat of imbalance. These traits are related to the individual's threshold for perception of threatening ambiguity, to his desire for certainty as opposed to his desire fer further expansion of his belief system. Such differences 10 will in turn influence his behavior in the face of belief-discrepant materials. In general these differences suggest that the closed- minded individual will be more likely to avoid materials not consistent with his attitudes and to be unfamiliar with materials related to positions in his disbelief system. Thus, when asked to engage in counterattitudinal communication, his rehearsal will be less adequate due to his lack of information concerning the new position and the rehearsal itself will be highly threatening to him. Culbertson (1957) and Elms (1966) found that role players with traits suggestive of closedsmindedness showed less attitude change than did those with traits suggestive of open-mindedness. In a study directly relevant to the present research, Hunt and Miller (1968) found that closed-minded persons showed greater selfi persuasion after_agreeing to engage in counterattitudinal communication. However, the dependent measures were obtained prior to any actual counterattitudinal encoding by the subjects. Hunt and Miller theorized that closed-minded persons would be relatively unfamiliar with the discrepant beliefs required of their role-playing assignment and that this unfamiliarity would lead to tension. Also, the magnitude of dissonance would.be increased by the closedeminded subjects' perceptions that the task would require substantial effort on their part. It was suggested that this would not be true for the open-minded persons. The present study accepts the rationale presented by Hunt and Miller for situations involving only commitment to engage in counter- attitudinal communication. The closed-minded person should be moderately ll threatened by the inconsistency. This in turn should produce pressures culminating in selfepersuasion. By contrast, the request to encode a counterattitudinal message should have no threat to the balance of the open-minded person's belief system. In the commitment situation, he has_agreed to perform a specific task requested by a relatively benign sponsor, but he has not actually come into contact with the material. Neither has he rehearsed the new position. Thus, the absense of tension and the lack of having processed or explored the new position offer the open-minded individual no inducements for selfepersuasion. This interpretation is enhanced by Hunt and Miller's finding that the mean change scores for open-minded subjects who agreed to encode belief- discrepant messages was virtually zero. But what would have happened if self-persuasion measures had been Obtained for the open- and closedpminded subjects after they had actually encoded beliefediscrepant communications? It is suggested that closed-minded individuals who actually come face-to-face with the beliefediscrepant material will be unable to tolerate it. Instead of dealing with the cognition of engaging in the role-playing task, they must now confront a set of more highly disbeliefhrelevant cognitions, those in the message that they encode. Not only should closed-minded persons do a poorer job of encoding the counterattitudinal messages, they should also be more upset by them than the open-minded persons. The beliefediscrepant material should create high uncertainty, and attitude change toward consistency should no longer be the easiest means of tension reduction. 12 Conversely, Open-minded subjects should be less likely to perceive the situation as threatening, and the rehearsal effect is more likely to produce substantial selfpersuasion. It was stated earlier that when conditions are relatively free of threat, the individual's desire for cognitive clarity will cause him to eXplore and to experiment with new courses of action and new beliefS. Thus, when counterattitudinal role- playing actually occurs, it seems likely that the Open-minded person will manifest greater selfepersuasion. Taken as a whole, these considerations lead to the following hypotheses: H 0 When persons are commited to but do not actually encode counterattitudinal messages, closedeminded persons will exhibit more change in the direction of the role-played position than will Open-minded persons. H2: After the messages are actually encoded, Open-minded persons will show more change in the direction of the role-played position than will closed-minded persons. The first hypothesis suggests a replication Of the findings of the Hunt and Miller study, and the second predicts a reversal of that finding when the subjects actually encode belief-discrepant messages. Taken together, the two hypotheses stipulate an interaction between relative open- or closedpmindedness and the point in time at which measures of selfepersuasion are obtained. CHAPTER II METHOD 292325 Of the five topics tested for use, the issue Of whether or not 18 year Olds should be given the vote was chosen. Feur state- ments were used to measure attitudes toward the issue. §_s indicated their position by marking their agreement or disagreement with each statement on a seven point scale ranging from Strongly Agree to Strongly Disagree. These items were coded so that a score of 28 indicated that the S was maximally favorable toward giving the vote to 18 year Olds, while a score of four indicated strong opposition. A random sample of 50 questionnaires was used to estimate the item intercorrelations. Table 1 presents these correlations, which seem sufficiently high to treat the items as a scale. Since the control group received no intervening treatment their pretest and posttest scores may be considered as a test-retest. The test-retest reliability coefficient was .90. Also included was a measure of Ss' latitudes of acceptance and rejection on the issue of 18 year Old voting developed according to methods outlined by Sherif and Hovland (1961) and Sherif, Sherif, and Nebergall (1965). 13 19 Table 1. Item intercorrelations of opinion items item 1 2 3 . n l -- .96 .89 .66 2 --- .91 .71 3 --- .72 n --_ Subjects: §s were undergraduates enrolled in communication and business letter writing courses at Michigan State University. Their participation was secured with the cOOperation of their in- structors. During the first week of the term 213 §§ completed a pretest which included a 20 item short-fern dogmatism scale (Troldahl and Powell, 1965), items fer measuring opinions on the issue Of lowering the voting age to 18, and a number of filler items. §s were told that t:he questionnaire was part of a general survey Of how college students feel about themselves and about a number of public issues. Two weeks later Se in the various classes were requested by tfleir instructors to participate in some research projects requiring about an hour of their time one of three evenings. Of the 213 §s taking pretests 123 chose to participate in the remainder of the study. Table 2 shows the distribution of dogmatism scores for these 123 Se, 120 of whom returned usable data. Three SS were eliminated because they did not follow instructions. 15 Table 2. Distribution of 83' dogmatism scores (N = 123) Score Freq Cum. Fr. Score Freq. Cum Fr. Score Freq. Cum Fr. 113 1 123 88 1 115 63 5 42 112 0 122 87 1 114 62 3 37 111 0 122 86 2 113 61 2 34 110 0 122 85 2 111 60 9 32 109 0 122 84 3 109 59 3 23 108 0 122 83 1 106 58 0 20 107 0 122 82 1 105 57 1 20 106 l 122 81 2 104 56 O 19 105 0 121 80 3 102 55 2 19 .104 0 121 79 3 99 54 3 17 103 O 121 78 1 96 53 2 14 102 0 121 77 8 95 52 0 12 101 0 121 76 3 87 51 4 12 100 1 121 75 l 84 50 2 8 99 1 120 74 1 83 49 1 6 98 l 119 73 4 82 48 l 5 97 1 118 72 6 78 47 0 4 96 0 117 71 4 72 46 0 4 95 0 117 70 1 68 45 1 4 94 0 117 69 1 67 44 0 3 93 0 117 68 6 66 43 0 3 92 1 117 67 7 60 42 l 3 91 0 116 66 4 53 41 1 2 90 l 116 65 3 49 40 0 1 89 O 115 64 4 46 39 1 l 16 Comparison of the dogmatism scores Of pretest Se who participated in the study with those who did not revealed no significant differences. The mean score for those participating was 69.3, while fer those not participating the mean was 67.1 (:_= 1.21, df = 212, p) .05 two tailed test). Thus, it would appear that there were no selective factors Operating which would result in a disproportionate number of high or low dogmatic Se volunteering for the study. Moreover, a comparison of the pretest attitudes of §§ who participated and those who did not yielded no significant differences (3?, participatOrs = 17.6; 3?, non-participators = 17.7, 5(1). Thus the available evidence indicated that those §§ who actually participated were a representative subset of the original sample. Procedures: As a result of the pretest measures Se were classi- fied as high or low dogmatic on the basis of whether they were above or below the median dogmatism score of 67. Within high and low dogmatic groups, §s were randomly assigned to the "Write" condition, the "Not Write" condition, or to the control group. §§ scoring below 16 on the attitude items--i.e., §s who were more or less against extending the vote to 18 year Olds--were assigned to write essays supporting the proposition that 18 year olds should be given the vote. §s scoring above 16 were assigned to write an essay arguing that 18 year olds should not be given the vote. 17 About two weeks after administration of the pretest, Ss' instructors announced that volunteers were needed to participate in some research projects, and promised extra credit for those who were willing to participate. Instructors were cautioned not to mention the pretest when soliciting the cooperation Ochs. When Ss arrived at the experimental sessions, they found their sign-up sheets posted with assignment to rooms entered beside their names. They were told that they had been randomly assigned to the various rooms, although, as mentioned above, actual assignment was made on the basis of pretest scores. Two rooms were used for the "Write? group and two for the "Not Write? group. Within each condition, one room was for those assigned to the pro 18 year old voting essay, and the other for those assigned to the anti 18 year Old voting essay. A fifth room contained all controls. At the outset, gs were told that there were two experiments, each taking about one-half hour, that they should participate in both to get the credit promised by their instructors, and that some would do one first and some the other. The two ostensible experiments were the actual essay writing task and a task called a test of creative imagination which involved writing stories about slides. Actually the slide task was a subterfuge used to mask the connection between the experimental treatments and the post-test. Controls were sent directly to the slide task on arrival. Se in the Not Write condition--the treatment calling for only simple 18 commitment--went directly to their treatment rooms. When everyone had arrived, the §_read the following instructions: The president has recently suggested that l8year olds should be given the right to vote. We are interested in how well college students know the issues involved and whether or net they are able to verbalize the arguments pro and con. We have randomly assigned.you to the various positions that can be taken on the issue and we ask you to write an essay with as many arguments as you can think up taking the position indicated by the topic at the head of the sheet to be passed out. Try to think up as many good arguments as you can fer the topic assigned. You have a half hour and please remain here for the whole time. After you're finished there is another study you are also to participate in in room 202. Please put your name, student number, and the class and teacher to be notified for your extra credit. Your papers will, of course, be anonymous and no names will be attached to papers in the analysis or at any other time. The task assignment sheets were then distributed, and the E asked that everyone go over the instructions as they were read aloud. When the instructions were finished, §_asked if there were any questions. When there were no further questions, a confederate entered and said that due to scheduling prOblems with the projector, the §s should participate in the other study first. The Se were then told to leave everything and to go across the hall where the projector was set up. When the Se in the Not Write condition arrived at the next room, the §_announced that the task was a test of creative imagination and passed out materials which included a questionnaire to be filled out befOre beginning with the slides. Included with a number of demographic questions and other attitude items were the critical posttest items. 19 When a11.§s had completed the questionnaire, the slides were presented as if it were an actual research project. These procedures were used to obtain posttests from both the control group and the Not Write treatment. When the slide task was completed, NOt Write §s were sent back to the original rooms to write the essay for the "other" study. The introduction and instructions for the Write treatment were the same as for the Not Write condition. §s were told that they had a half-hour to write as many arguments as they could, and that they should use the whole time. By the time the Write group had finished, the control and Not Write conditions had completed the slide task and had gone back to the other rooms. The Write group was then told to report to the projection room to participate in the other task. _Again, materials were passed out for the test of creative imagination, the first part of which was the questionnaire containing the critical attitude items. Although the various subterfuges called for complex timing and maneuvering of peOple, the procedures functioned without any problems. It was critical that Not Write §s understood the task they had to perform and they believed the deception associated with the administration of the posttest. It was also important that n°.§§ see the relationship between the two tasks. It can be reported that the movement °f.§§ was quite successful. In general, less than a minute elapsed between the time Se in both 20 the Write and Not Write conditions left their original rooms and hegan work on the posttest questionnaire. Post experimental dis- cussions with Se also indicated that the various subterfuges were apparently accepted at face value. In no case was anyone able to state accurately the purpose of the study, and no §s reported a connection between pretest and experiment or experiment and posttest. Most expressed surprise when the nature of the experiment was disclosed and when told that pretest, manipulation, and posttesting were all part of the same study. CHAPTER III RESULTS Data for the present study consisted of pretest to posttest attitude change scores. Changes were calculated so that positive change indicates change in the direction of the belief discrepant position advocated, while negative change indicates a contrast effect where the subject shifted to a stand even more strongly in favor of his initial position. Comparison of Treatment Groups With Controls: The mean pretest to posttest attitude change scores for subjects in the experimental and control groups are found in Table 3. Dunnett's test (Winer, 1962) was used to compare the amount of attitude change in each of the treatment groups with the change occurring in the combined All Controls group. As Table 3 indicates, only low dogmatic subjects in the encoding condition reported significantly greater self persuasicm than the control subjects; i.e., when compared with All Controls, only the Low Dogmatic; Write subjects demonstrated a significant amount of self persuasion. 21 22 Table 3. Attitude change scores for subjects in experimental and control groups. Not Write Write Control High 3? = 2.65 E = 2.43 3? = 0.52 All Dogmatic n = 20 n = 23 n = 19 Controls X = 0.89 _ _ _ n = 39 Low X = 0.52 X = 4.24 X = 1.25 Dogmatic n = 21 n = 17 n = 20 Table 4. Comparison of mean attitude change scores Of treatment _ groups to All Controls: Dunnett's test. Comparison Difference t p High Dog.: Not Write 33 All Controls 1.76 1.23 ns High Dog.: Write 33 A11 Controls 1.59 1.19 ns Low Dog.: Not Write is All Controls -0.37 (1 ns Low Dog.: Write 33 All Controls 3.35 2.20 (.05 In addition to comparing each treatment group with the combined controls, Dunnett's test was also used to compare treatment and control groups at each of the twO levels of dogmatism (Table 5). The results of these comparisons were identical to the first analysis: only low dogmatic subjects in the writing condition reported significantly . greater self-persuasion than their counterpart control subjects. 23 Table 5. Comparison of mean attitude change scores Of treatment groups to control groups at each.1eve1 of Dogmatism: Dunnett's test. Comparison Difference t p High Dog.: Not Write 3§_High Dog. Controls 2.13 1.22 ns High Dog.: Write 2§_High Dog. Controls 1.96 1.18 ns Low Dog.: Not Write _v_s Low Dog. Controls -0.73 4 1 ns Low Dog.: Write 3§_Low Dog. Controls 2.99 1.93 «(.05 These treatment and control group comparisons should be con- sidered in view of the interaction hypothesis posited in the present study. Ideally, it was hOped that the high dogmatic subjects in the simple commitment condition and the low dogmatics in the encoding condition would report significantly more selfepersuasion than the control subjects, while the rest of the subjects in the other conditions would not. Although the actual comparisons did not conform exactly to this ideal situation, it is encouraging to note that the amount of selfepersuasion in the writing condition by low dogmatic persons was significantly greater than thatoccurring in the control group, and that the amount of self-persuasion observed among high dogmatics in the simple commitment condition, while not significantly different from the control condition, was somewhat greater than the change for high dogmatic subjects in the writing condition and low dogmatic subjects 24 in the simple commitment condition. As expected, this latter group reported practically no pretest to posttest attitude change in the direction of the belief discrepant position. Test of the Hypotheses Using the Original Median Split on Dogmatism: As indicated above, the following hypotheses were tested in this study: H1: When persons are commited to but do not actually encode a counterattitudinal message, closed minded persons will exhibit more change in the direction of the role-played position than will Open-minded persons. H2: After the messages are actually encoded, open-minded persons will show more change in the direction of the role-played position than will closedsminded persons. Taken together, these hypotheses predict that whether or not subjects encode counterattitudinal communications will interact with level of dogmatism in producing a selfepersuasion effect. The predicted effect is schematized in Table 6: It was predicted that in the Not Write czondition the mean attitude change score of the high dogmatic persons would be greater than the mean change score fer the Low Dogmatic group, and that in the Write condition, the mean attitude change score of the low dogmatic subjects would be greater than the mean change score of the High Dogmatic group. Table 6. Schematic of predicted differences in mean attitude change S coma O Not Write Write High _ _ Dogmatic X11 > X12 1... _v A Dogmatic X21 ‘L 222 25 Though differences were in the predicted direction, the hypothesized interaction was not statistically significant. Table 7 presents the analysis of variance summary of the effects of the level of dogmatism and the treatment condition on the self-persuasion scores of subjects, using a median Split of the dogmatism scores. It can be seen that the largest F-ratio was obtained for the interaction effect; however, the eEect is not statistically significant. Table 7. Analysis of Variance Summary: The effects of dogmatism level and treatment condition on attitude change scores. _Not Write __ Write High Dog. 5 = 2.65 (3:20) X = 2.48 (n = 23) Low Dog. X = 0.52 (nf2l) X = 4.24 (n = 17) _SOURCE S_5 93 11.5. 2. Dogmatism (A) 0.03 1 0.03 4 1 ns Treatment (B) 3.13 1 3.13 1.84 us A x B 3.77 l 3.77 2.22 ns Error 77 1.70 Test of the Hypotheses Using Upper and Lower Thirds of the Dogmatism sage: Dogmatism is a complex set of traits. Persons who score high or low on the dOgmatism scale must be high or low on a majority of the factors that constitute the scale. On the other hand, people in the middle of the distribution constitute a heterogeneous group--at least 26 more so than those scoring high or low. In order to determine if the upper and lower thirds of the dogmatism score distribution would function more discriminatively, the opinion change scores of subjects between the 33rd and the 67th percentiles of the distribution were dropped out and the attitude change scores were reanalyzed. Table 8 presents a summary of this analysis: Table 8. Analysis of Variance Summary: The effects of dogmatism level and treatment condition on attitude change--upper and lower 3rds of dogmatism range. Not Write Write High Dog. Z = 3.35 (n = 14) Z = 1.14 (3 = 14) Low Dog. X = 1.67 (n_= 15) X = 5.70 (n_= 10) _SOURCE es 2: m r p. Dogmatism (A) 2.05 1 2.05 4:1. ns Treatment (B) 0.82 1 0.82 (1 us A x B 9.77 1 9.77 4.07 (.05 Error 49 2.40 i As the table shows, the interaction effect originally hypothesized fer the entire group was found to be significant when the subjects in the middle of the dogmatism distribution were eliminated; i.e., when subjects were commited to but did not write counterattitudinal messages, the high dogmatic group showed greater shift of Opinion than did the 27 low dogmatic grow. When counterattitudinal messages were written subjects in the low dogmatic grow showed greater self-persuasion than did high dogmatics. A Possible Side Effect of Treatments: Differences in the Distributions of Attitude Change Scores Between Write and Not Write Conditions: An interesting unanticipated result was revealed when the within cell variances of the four treatment grows were studied. The variance of the Not Write grow was approximately half that of the Write grow and the ratio of these variances was significant (p (.05). Apparently the act Of writing the counterattitudinal message exercised an effect on the distribution of attitude change scores. The rationale for this study suggested that high dogmatic subjects in the Write condition would experience greater threat to their belief system and that such threat should result in a contrast effect. The finding concerning the differential variance of attitude change scores for the Not Write and Write groups suggested further inspection of these distributions. For purposes of this analysis, a shift of +3 or more was classified as a positive shift, 129 11, and 0 were classified as no change, and anegative shift of -3 or less was classified as a negative shift. Table 9 presents the frequencies for these results within each cell. 28 Table 9. Frequencies of positive Opinion shift (+), negative shift (-), and no change within each cell, All _S_s Not Write Write . Control + 8 10 6 High 0 11 6 12 Dogmatic - l 7 1 + 4 7 4 Low 0 14 9 15 Dogmatic - 3 1 1 Inspection of the distribution of opinion change scores shows that a greater number of high dogmatic subjects in the Write condition demonstrated a contrast eEect. The distribution of positive, negative, and no change scores in this cell was compared to that of high dog- matic Control subjects (‘12 = 7.5, p<.05, one tail test), to that of High Dogmatic Not Write subjects (‘X2 = 5.9, p (.05, one tail test), to that of Low Dogmatic Not Write subjects (12 = 4.8, p¢.05, one tail test), and to that of Low Dogmatic Not Write subjects (12 = 7.3, p ‘,05, one tail test). Thus the distribution of self-persuasion scores of the high dogmatic subjects in the Write grow appears to differ from that of each of the other grows, a difference due to the greater number of High Dogmatic Write subjects who demonstrated a contrast effect. 29 After eliminating attitude change scores for the middle third of the dogmatism range, the distributions were again studied in terms Of positive, negative, and no change, using the same criterion for classifying change scores. Table 10 summarizes (these frequencies: Table 10. Fraquencies of positive Opinion shift (+), negative shift (-), and no change within each cell, wper and lower thirds of dogmatism range only. Not Write Write + 6 4 High Dogmatic 0 7 5 .. 1 5 + 3 5 Low Dogmatic 0 l2 5 - o o Because of the zeroes in two of the conditions, statistical tests of the differences between the distributions was not appropriate. Even so, comparison of Table 10 with Table 9 reveals the influence that the middle third of the dogmatism distribution had on the overall result. The High Dogmatic Write grow still shows a relatively strong contrast effect, while the relative frequency of positive Opinion change 30 sharply drops. Examination of the cells suggests that drOpping out the subjects scoring in the middle third of the dogmatism distribution eliminates many of the cases that deviate from the hypothesized effects of the treatments. Ego-Involvement and SelfePersuasion: Another analysis sought to examine the effects of subjects' .ego-involvement on subsequent selfepersuasion, using the experimental issue of extending the vote to 18 year olds. Five issues were pre- tested: draft deferrments for college students, 18 year old voting (the one used), gun control laws, the election, and radical or hippie . groups (the SDS) on campus. Of these, the first two resulted in the most extreme polarization Of opinion scores. Dissonance theory posits 'that the more involved the person is with the issue, the greater the dissonance that is aromed by counterattitudinal material. The p greater the magnitude of dissonance, the greater should be the pressures to change. Conversely, Sherif, Sherif, and Nebergall (1965, p. 198) report a study by Elbing, who found that higher ego-involvement tended to depress the selfhpersuasion effects of counter attitudinal role-playing. According to Sherif and his co-workers, ego-involvement is reflected in: (l) the extremeness of the position the person adopts as his own, (2) the size of his latitude of rejection, and (3) the size of his latitude of non-commitment. Thus, highly_ego-involvement persons will mark the most extreme positions as most acceptable, will reject 31 relatively more positions, and will consequently remain uncommited on relatively fewer positions. Taking those predicted behaviors into account, an index was constructed to rank order subjects on ego— involvement on the basis of their pretest scores. The rank ordered positions were assigned a value according to the method of rank order (Guilford , 1954) . The product of this value, times the size of the latitude of rejection, plus 9 minus the latitude of non-commit- ment provided a rough index on which to order the subjects. A median split was then made on the basis of this index of ego-involvement, and a 2 x 2 x 2 analysis of variance was performed, using Opinion change scores as the dependent variable. an alysis: Table 11 presents a summary of the Table 11. Analysis of Variance Summary: The effects on attitude change of treatment conditions, level of dogmatism, and level of ego-involvement. Not Write Write High B-I _Low E-I High E-I how E-I High X = 1.88 X = 3.27 X = 0.58 X = 4.56 Dogmatic n=9 n=11 _n_=l2 n_=11 Low 76 = -o.4o E = 1.35 3? = 7.14 3? = 2.20 Dogmatic n: 10 n=11 _n_=7 n: 10 SOURCE SS df MS P p Dogmatism (A) 0.20 1 0.20 41 ns Ego-involvement (B) 0.14 l 0.14 r 1 ns Treatment (C) 9.94 1 9.94 3.02 us A x B 11.56 1 11.56 3.70 ns A x C 10.09 1 10.09 3.23 ns B x C 3.52 l 3.52 1.12 us A x B x C 13.67 1 13.67 4.68 (.05 Error 73 3.12 32 As Table 11 indicates, the double interaction was statistically significant. Examination of the mean attitude change scores in each cell shows that in three of four cases low ego-involvement leads to _ greater self-persuasion than high ego-involvement. The interaction, however, arises from the fact that differences in attitude change scores between high and low ego-involvement and high and low dogmatics are greater in the Write condition than in the Not Write condition. This would suggest that ego-involvement does not simply add a constant to the self-persuasion effect, but interacts with other factors in a non-additive fashion. Application of the conservative Scheffe's test indicated that none of the Observed means differed significantly from each other. One possible conclusion is that neither the dissonance explanation nor the Sherif position will entirely account for the Observed results, _ given that it is a reliable finding. The findings suggest that further attention should be given the variable of ego-involvement in future study of the self-persuasion phenomenon. A Note on Some Possible Contaminatingrgnvironmental Factors: Prior to the experimental sessions, but after the administration of the pretest, the 18 year Old voting issue selected for the present study received a considerable amount of public attention, all in one direction. There were articles in the student newsPaper, a public rally that received local television news coverage, and posters on campus, all urging swport for giving 18 year olds the right to vote. 33 Given unidirectional inputs on a current issue in which many subjects are generally ego-involved, a depressing effect on the attitude change scores of those initially favoring 18 year old voting would be expected. On the other hand, the attitude change scores of. those initially Opposing 18 year Old voting should be magnified. As Table 12 indicates, the subject's prior stand was indeed a highly significant source of variance in the attitude change scores. Table 12. Analysis of Variance Smmnary: Analysis of the effects of Ss' prior position on attitude change scores. Not Write Write _ Pro _Con _ Pro _ Con High X = “1.10 X = 6.” X = 1016 X = 3.91 Dogmatic n = 10 n = 10 n = 12 n = 11 Low '2' = -0.31 If = 1.88 Y = 3.43 SE = 8.00 Dogmatic n_=13 n_=8 _n_=l4 2:3 SOURCE a if. _Mé. 2: 2 Dogmatism (A) 0.86 l 0.86 (1 ns Side (B) 36.11 1 36.11 9.84 (.01- Treatment (C) 11.61 1 11.61 3.16 us A x B 1.52 1 1.52 (1 ns A x C 12.71 1 12.71 3.46 ns B x C 0.70 l 0.70 (1 us A x B x C 6.38 1 6.38 ns Error 73 3.67 34 It can also be seen from the table that subjects who were initially Opposed to giving the vote to 18 year olds (Con group) showed greater average shift of opinion than subjects who were initially in favor of lowering the voting age (Pro group). Moreover, among the groups initially opposed to changing the voting _age, mean attitude change scores fOllowed the pattern predicted by the hypotheses: high dogmatic Not Write subjects showed greater change than Low Dogmatic Not Write subjects, but Low DOgmatic Write subjects showed greater change than High Dogmatic Write subjects. Further evidence on the effects of outside influences may be seen when the mean attitude change scores of experimental groups are compared to those of the control group, also divided with reapect to initial stand. If, as has been held, the outside influence boosted or depressed the attitude change due to experimental treatments, and if the control group reflects changes of Opinion due to Outside input only, then perhaps the differences in net gains in self-persuasion Observed between Pro and Con groups are due to such influences. Specifically, among the Cons only, net gains in self persuasion occurred among the High Dogmatic Not Write and Low Dogmatic Write groups. These . gains are consistent with the hypotheses. Among those initially in favor of giving the vote to 18 year Olds, the net gains in selfe persuasion that did occur seemed to be associated with the Write groups, with greater net gains for the low dogmatics. Table 13 presents this information. 35 Table 13. Mean attitude change scores of all groups when divided with respect to initial position. CONS ONLY _ Not Write Write , Control High 3? = 6.4 3? = 3.91 if = 3.75 Dogmatic n: 10 3:11 3:8 Low Dogmatic Z = 1.88 Y = 8.00 3? = 3.43 Dogmatic 2:8 n_=3 n_=7 PROS ONLY _ Not Write Write Control High Y = -1.10 if = 1.15 '5? = -0.91 Dogmatic n = 10 n = 12 n = 11 Low '3? = -0.31 32' = 3.43 Y = 0.07 Dogmatic n = 13 n = 14 n = 13 In smary, the data, although strongly suggestive, do not provide clear cut support for the hypothesized interaction. Some implications and conclmions drawn from these data are considered in the next c hapter. CHAPTER IV DISCUSSION This study was based on the rationale that the openness or <:1osedness of a person's belief system interacts with situational factors in determining selfhpersuasion effects resulting from counterattitudinal communication. It was suggested that dissonance reduction would operate in the simple commitment situation, but only for the person with the more closed belief system. It was argued that the closedpminded person experiences tension because of the inconsistency between the assigned task and his own beliefs and because he lacks practice in considering discrepant positions. 0n the other hand, the relatively open person is less likely to experience tension resulting from the-apparent in- <:onsistency between task and beliefs and is also more practiced in considering positions other than his own. Thus, for the relatively closedpminded person, tension resulting from the inconsistency should motivate tension reduction, which may be manifested in selfhpersuasion. If the more open person does not perceive inconsistency and/or experiences no tension, he is much less likely to change his Opinion than is the closedsminded person, since the Open person is not motivated to change. 36 36a The rationale also posited that the situation may change when persons go beyond mere commitment to the task and actually encode counterattitudinal.messages. Open-minded persons should be more familiar with handling belief-discrepant ideas and should be able to deal with such ideas more comfortably. In contrast, closed-minded persons are less practiced at dealing with positions other than their own. Not only will the closedsminded.person be less likely to do an adequate job of rehearsing the counterattitudinal position, but the rehearsal itself may be a threat to his belief system. Rather than resulting in selfepersuasion, this situation is more likely to lead to a contrast effect. Thus, if the open person is more accustomed to the juxtaposition of ideas, the rehearsal effect should lead to opinion shift in the direction of the role-played position. Operationally, the above statement implies that high dogmatic persons should show greater positive change than low dogmatic individuals when there is only simple commitment to engage in counterattitudinal communication, but that a reversal should occur when counterattitudinal messages are actually written. Examination of the data, however, revealed the following: (1) Only the low dogmatic Write group showed attitude change significantly different from the no-treatment controls. (2) When data from all subjects were analyzed, the hypoth- esized interaction was not statistically significant, although differences in attitude change scores were in the predicted directions. 37 (3) Post hoc analysis of the upper and lower third of the distribution of dogmatism scores did reveal the interaction effect initially hypothesized fer the entire group. (4) The variance of the attitude change scores of the Write group was significantly greater than that of the Not Write group. (5) Though the means did not differ significantly, the diatribution of change scores in the high dogmatic Write ‘ group did differ significantly from the distribution of the other groups, apparently due to the greater number of large negative shifts in that group. This result was more pronounced when the middle third of the dogmatism distribution was eliminated. (6) When all subjects were classified in terms of high or low ego-involvement and the attitude change scores were .again analyzed, the ego-involvement by dogmatism by treatment interaction was significant. (7) When all subjects were classified according to initial position, the results indicated that subjects who were initially Opposed to lowering the voting age showed significantly greater selfepersuasion than did those initially in favor of lowering the voting age. Among the Cons only, the cell means were also in the predicted direction. While the overall analysis of the data did not fully support the hypotheses, the fact that the mean opinion change scores were _ generally as predicted and the fact that elimination of the middle third of the dogmatism distribution did produce the hypothesized reversal effect indicates that the original rationale has some merit. Subsequent analyses, however, suggest that some other factors need to be considered, which may in turn lead to some modifications of the rationale. as a .1 First, it sesms plausible to argue that for low dogmatic persons the rehearsal effect may be the main factor influencing selfhpersuasion. Fer all analyses performed, the Low Dogmatic Write group showed greater net positive shift of Opinion than the Low Dogmatic Not Write group-- regardless of whether they were initially Pro or Con or high or low ego-involved. In addition, the Low Dogmatic Not Write group did not differ significantly from the control group, while the Low Dogmatic Write group did. As expected, a rehearsal effect rather than reduction of tensions caused by apparent inconsistency seems to best explain the responses of low dogmatic individuals. This conclusion must be some- what tempered by the fact that conservative statistical procedures used for internal analysis (e.g. Sheffe's test) do not indicate that the larger mean attitude change scores of the Low Dogmatic Write group are significantly greater'than those of the Low Dogmatic Not Write group. The attitude change scores of high dogmatic persons show no such discernible trends. As previously reported, the mean selfepersuasion scores for the high dogmatic groups do not differ significantly from each other or from the controls. One plausible interpretation is that writing the counterattitudinal messages had no effect on the selfe persuasion of high dogmatic individuals; i.e., that for high dogmatic persons any attitude change that occurred took place after simple commitment. The observed similarities between the cell means of the two high dogmatic groups are somewhat deceptive. When the distributions of the 39 attitude change scores are compared across high dogmatics we find that the High Dogmatic Write group has a slight edge in positive changers (selfhpersuasion)‘and_a somewhat larger edge in negative changers (contrast effect). As previously noted, this result was more pronounced when the middle third of the dOgmatism scores were drOpped from the analysis. If, as posited earlier, there is some individual threshold for tension beyond which attitude change toward consistency is no longer possible, then it would be eXpected that observation of the relative frequency of contrast effects should become clearer when the middle dogmatics are eliminated. Elimination of the middle group primarily eliminates positive changers from the high dogmatic . Write grOUp. Perhaps fer this middle group the writing task was not sufficiently threatening to exceed threshold. When considering the interaction between the structure of an individual's belief system and situational determinants in the context of the selfepersuasion process, a dichotomous, high/low classification of dogmatism may be inadequate. Perhaps dogmatism as an index of open and closed belief systems needs to be classified as high, low, and ‘medium, or perhaps other? The present level of specificity of the