PRIOR ATTITUDE AND LANGUAGE INTENSITY AS PREDICTORS OF MESSAGE STYLE AND ATTITUDE CHANGE FOLLOWING COUNTERATTITUDINAL COMMUNICATION BEHAVIOR Thesis for the Degree of Ph. D. MICHIGAN STATE UNIVERSITY MICHAEL BURGOON 1970 II’IIIII’IIIIII I III L, THtsw- This is to'certifg that the thesis entitled PRIOR ATTITUDE AND LANGUAGE INTENSITY AS PREDICTORS OF MESSAGE STYLE AND ATTITUDE CHANGE FOLLOWING COUNTERATI'ITUDINAL COMMUNICATION BEHAVIOR presented by Michael Burgpon has been accepted towards fulfillment of the requirements for Ph.D. degree in Communication M»? A m 6%» Major professor Date April 1141 1970 0-169 LIBRARY Michigan State University “I '” BINDING Iv I; am a some I em mm me. ‘. we ‘ {IE—3.7.3 “3‘55 ABSTRACT PRIOR A'ITI'IUDE AND LANGJAGE INTENSITY AS PREDICTORS OF MESSAGE STYLE AND ATI'I'IUIIJ CHANGE PULLDWING CGJN'I'ERA'ITITUDDIAL COM‘UNICATION BEHAVIOR By Michael Burgoon Research on counterattitudinal advocacy has generally denun- streted that a person who encodes a belief-discrepant message will shift his attitude to nore closely conform to the advocated position. As a result of this enphasis placed on attitude change as the primary dependent variable, little attention has been given to the encoded mes- sage, nor has prior research manipulated the possible messages to pre- dict attitude change . The purpose of this investigation was twofold: (l) to emtine the effects of coxmterattitudimul encoding on message style, and (2) to investiate ways in which encoding situations can be constructed to predict the nagnitude of attitude change resulting from comterattitudiml advocacy. The studies were designed to test the relationship between prior attitude and message intensity. The methodology required subjects to couplete partially constructed messages by choosing words of varied in- tensity. In the first emerinent, half of the subjects created belief- congruent massages and half created belief-discrepant messages. Subjects in both treatnent conditions chose from word lists of carpareble overall intensity. It was predicted that persons who encoded a belief-discrepant Michael Burgoon message would use language of significantly lower intensity than per— sons who encoded a belief-congruent nessage. In the second experi- ment, subjects prepared a comterettitudinal nessage using either high, noderate, or low levels of language intensity. It was hypothesized that attitude change would be directly related to message intensity: that the met attitude change would occur in the high intensity condi- tim, the least in the low intensity condition , and the moderate in- tensity condition would be somewhere between these extremes. The findings support both theoretical hypotheses. Although both groups in Experiment I encoded noderetely intense messages, belief- congruent subjects as predicted, did encode significantly more intense messages than did the subjects in the belief-discrepant condition. Also, the results of Experiment II indicated that relatively intense encoding is a necessary condition for attitude change following counter- attitudinal conmmioation behavior. Both the high and noderete inten- sity groups denonstreted significantly nore attitude change than did a non-encoding control group, the low intensity and control groups did not differ. The attitude change was greatest in the high intensity con- dition, least in the low intensity condition, and the moderate condition fell between. Findings were discussed in terms of prior research in counter- attitudinal advocacy. A nunber of research extensions, suggested by the findings of this study, were discussed. PRIORLATTTTUDE AND LANGUAGE INTENSITY.AS PREDICTORS OF'MESSAGB STYLE ANTIATTITUDE CHANGE FOLLOWING COUNTERAETTTUDINAL COMMUNICATION BEHAVIOR By MichaelBur-goon A.THESIS Submitted to Michipn State University in partial fulfillment of the requiremnts for the degree of DOCTOR OF PHILOSOPHY Department of Connmioation 1970 Accepted by the faculty of the Department of Connmnication, College of Communication Arts, Michigan State University, in partial fulfillment of the requirements for the Doctor of Philosophy degree. 44/2/79 17/212 57993 7'77 :r' Director of Thesis 0 o , _ fl 6‘ A [”7 . Guldance Comrnlttee: VIM/L4 if? K 1,? / / 31,, Chairman (“E 'H ((4 Cf @fll I'lllhflu ACKNOW LEIISMENTS It is of curse impossible to adequately express the appreci- ation I feel toward the many people who helped me to reach this point in my academic career. It seems appropriate to begin such an attempt by acknowledging the assistance of Dr. Gerald Miller who directed this thesis. Whatever capabilities I possess as a scholar of human commnicaticn are in large part due to the demands, challenges , and rewards that are the result of working with such a man. My guidance committee members, Dr. Randall Harrison, Dr. Vincent Farace, Dr. Dan Wackman, and Dr. Eugene Jacobson, were always willing and able to give needed assistance throughout my graduate program. Dr. Hal Helper and Dr. Verling 'Iroldahl were of help many times in the last few months. A special thanks is extended to Dr. Jay Weston who introduced me to the study of human camunicaticn. The entire Department of Cannunication provides an atmosphere that makes intellectual inquiry an exciting way of life. T910 of my colleagues. deserve special attention. First, my office-mate Roger Haney allowed me the luxury of a private office. Second, Edward Bodaken contributed so nuch that I must thank him. He helped me and allowed me to help him; such is my definitim of a colleague. To thank my wife, Joan, would be a denial of what we think is a sym- metrical relationship. Whatever accomplishnent this represents is as nuch a credit to her as anyone. What will we ever do with our weekends now. ii TABLE OF CONTENTS I. ImwcrlON OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO0.0. II. WTI OIOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO0.00...... mama OOOOOOOOOOOOOOOOO0.0.0.000...COCOOOOOOOOOOO Results 00.000.00.000.0.00.0000...OOOOOOOOOOOOIOOCOOOO III. WW II C0.0...O...O...O...OOOOOOOOOOOOOOOOOOOOOOOO mms 0.0.0.000...OOOOOOOOOOOOOOOOOOOOOO0.0.0.... ksults 0.0...0.0.000000000000000.0000000000000000.... Iv. DIwJSSION OOOOOOOOOOOOOCOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO. WMliations 0.00.0.0...OOOOOOOIOOOOOOOOOOOOO SW 0......OOOOOOOOOOOOOOOOOOOOOOO0.00.00.00.00... REFERENCES ....................................................... APPENDIXA: PRETESTATTITUIIZ QUESTIONNAIRE .... APPENDIX B: WIP‘IATERIAIS ..... APPENDIX C: DCPERIMENT II MATERIALS .... APPENDIX D: POST'IEST ATTITUDE QUESTICNNAIRE, IMPORTANCE RATINGS, AUDIENCE ATTITUDE RATINGS, TASK Dlmwm WGS OOOOOOOOOOOOOCCIOOOOOOOO0.0.0.0... iii 16 17 2M 29 30 33 ITl us In I49 53 58 63 72 10. ll. l2. 13. LIST OF TABLES Weigited scale values of the experimental words Means, standard deviations and t—test of the differences in language intensity scores in the experimental conditions. . Means and _t_-test of differences in pretest through post- encoding attitude change in the experimental conditions . . . . . T-test of differences in pretest through post—encoding Iatitude of acceptance change in the experimental con- ditions OOOOOOOOOOOOOOOOOOOOOOOOOOCOOOOCOCOOOOOOOOOOOOO0.00.0 Word values , neans, standard deviatims, and sumnary of the analysis of variance of differences of words selected for use in the experimental conditions Percentage, frequency, and Chi-square test of subjects correctly perceiving audience attitude . Pretest attitude and latitude of acceptance measures . . . . . . . . Means and standard deviations for experimental and cmtml mp8 O...OOOOOOIOOOOOOOOOOOCOO0.0.0.0....O....00... Attitude scores, ammt of change, and analysis of variance of attitude change scores of experimental mdcmtml ms 0......00.0.0000...0.00.00.00.00.0..00.... Analysis of differences of attitude change scores between each experinental and control group Attitude scores, anomt of change, and analysis of vari- ance of attitude change scores of experinental groups . . . . . . . Analysis of difference of attitude change scores “mahemrimtal m‘lp OOOOOOOOOOOOOOOOOOOOOOO00.... Latitude of acceptance means , standard deviations , and analysis of variance of pretest latitude of acceptance scores for experimental and control groups iv 21 26 27 28 31 33 3'4 35 36 36 38 38 39 1U. Latitude of acceptance scores , amount of change and anal— ysis of variance of latitude of acceptance change scores of experimental and control grwps H0 LIST OF FIGURES 1. mwtal “Sig: EmmtI OOOOOOCOCOOOOOOOOOOOOOO 16 2. mtmdemS-WAttimde me OOOOOOOOOOOOOOOOOOOOOO 18 3. TheExperimental Design: Experiment II 29 vi CHAPTER I INTRODJCTI ON The Problem For the past two decades researchers have been investigating the effects of certain encoding situations on source attitude change (Moreno, 19%; Hovland, Janis and Kelly, 1953; Festinger, 1957). The typical experiment has required persons to produce messages advocating a positim cattrary to their private Opinions. Upon completion of this counterattitudinal task, same attitude assessment measure is obtained. Research has emphasized specification of variables that increase the magnitude of attitude change resulting from counterattitudinal encoding behavior. The research consistently demonstrates that a person who encodes a belief-discrepant message will shift his attitude to cmform more closely to the advocated positia't. The effects of specific antecedent conditions such as justification (Festinger and Carlsmith, 1959; Elm and Janis, 1955; Rosenberg, 1965), type of encoding (Collins, 1968), and effort (Zimbardo, 1965) have produced much theoretical debate. This effort has beat aimed at specifying the conditims that lead to attitudinal shifts. As a result of the emphasis placed on attitude change as the primary dependent variable, little attenticn has been given to the encoded mes- sage, nor have there been many attempts to determine the effects of counterattitudinal encoding upm message production. Finally , prior research has not manipulated the possible messages produced in order to predict attitude change. The purpose of this investigaticn was twofold: (l) to examine the effects of counterattitudinal encoding on message production, and (2) to investigate ways in which encoding sit- uations can be constructed so that message analyses can be used to pre- dict magnitude of attitude change. Analysis 9_f_ Counterattitudinal Messages In early studies by Janis and King (19510 and King and Janis (1956) the explanation of counterattitudinal encoding effects was based upon an assumed relationship between the kind of counterattitudinal message encoded and the resulting magnitude of attitude change. Their research led to the conclusion that reformulating and elaborating at a belief- discrepant topic is a critical determinant of attitude change. Only very gross experimenter evaluations of message producticn were reported. Janis and Gilmore (1965) and Elms and Janis (1965) argue that when a person accepts the task of encoding a belief-discrepant message, he becomes motivated to think up all of the good belief-discrepant arguments he can and simultaneously repress belief-congruent arguments. Such "biased scanning" should produce higher quality belief-discrepant mes- sages. In order to test for quality differences in the messages, judges' blind ratings of counterattitudinal assays were obtained. In effect, the quality ratings mly measured the number of explicit arguments supporting the belief-discrepant positim. Janis and Gilmore fcnnd a direct relation- ship between message quality and attitude change, but Elms and Janis failed to replicate this finding. Festinger and Carlsmith (1959) used concealed audio-recording equipment to record the verbal belief-discrepant messages. These re- cordings were transcribed and then rated, by two independent judges , on several dimensions: (1) the strength of the positive statements about the dull task, (2) a rating of the overall content of what the subject said, (3) a rating of how persuaive and convincing the message was, and (II) a rating of the anoint of time the subject discussed the task as opposed to discussing irrelevant things. These researchers found no significant differences in message production as a result of counter» attitudinal encoding behavior under various levels of justification. Carlsmith, Collins, and Helmreich (1966) also used judges' ratings to analyze counterattitudinal messages. Transcripticns of verbal mes- sages were rated by three judges on five dimensions: (1) persuasiveness and emphais, (2) overall. positiveness, (3) overall positiveness and con- viction, ('4) percent of time discussing topic, and (5) dissociation of self from content. In addition to judges' ratings, the researchers aked for the experimental accomplice's evaluations of the oral presen— tation of each subject. The first three of the above scales were used in addition to ratings of apparent conflict and signs of discomfort. Carlsmith, Collins , and Helmreidt also analyzed written essays rated on four scales: (1) emphasis used in making points, (2) the extent to whidm the subject created reasms in support of the belief—discrepant position, (3) overall quality and persuasiveness, (1+) apparent effort expended with an attempt to control for writing ability. None of the judges' retirgs of the messages yielded any significant differences among treatment grmps. Also, no evidence was found that any message measure was correlated with posttsst attitudes. Similarly, Linder, Cooper, and Jones (1967) examined the messages produced in 1:40 self-persuasion studies. One objective measure was ob— tained by comting the number of words per essay. In addition, two judges rated each essay on scales measuring: (1) degree of organization, (2) overall persuasiveness, and (3) extremity of the position advocated. No significant message differences were observed. Again, only comter- attitudinal messages in different experimental conditions were analyzed. Rosenberg (1965) relied on two judges' evaluations of basic persua- sivensss. Word-comm were also taken on the essays. Rosenberg concluded that those who wrote comparatively unpersuasive essays for a small reward shoved significantly more negativism toward the comterattitudinal posi- tion. In the high reward treatment conditions, there was no correlation between essay persuasiveness and posttest attitude. In general, Rosenberg concluded that the subjects in the lot reward condition were insufficiently motivated to encode belief—discrepant messages. Thus, of the six self-persuasion studies that analyzed messages, all used some sort of overall evaluation of persuasiveness or quality. All evaluated oily belief-discrepant messages; none compared belief-congruent messages with comterettitudinal messages . Foir of the six studies used essay length as the mly objective measure; the other objective measures reported were time Spent encoding and number of comterattitudinal argu- ments produced. The prior research neglects to analyze messages on content or style at a level of abstraction less than overall judgments of a total message. No attempt has been made to manipulate kinds of message production to pre- dict attitude change. Greenberg (1960) and McBwen (1969) are the only researchers who have attempted to structure encoding situations so as to predict differences in message production and style variables. The present research solgtt to devise a methodology for dealing with two questions: (1) How do belief-discrepant messages differ from belief- congruent messages? (2) Is there a relatio'tship between how strmgly a persot argues against the positim he privately holds and the magnitude of his attitudinal shift toward the advocated pceition? Message Production Under Psychological Stress While attempting to formulate hypotheses abort message production in comterattitudinal encoding situations, McEwen (1969) summarized a number of studies that related psychological stress to verbal behavior. Greenberg and Tannenbaum (1962) found that subjects who produced mes- sages under induced cognitive stress took longer to encode the message, made more writing errors, and created less readable messages that did subjects in a low stress condition. The authors conclude that "Clearly atd strikingly, a communicatiot performance can be hindered by the degree of cogtitive stress in the encoding situation" (p. 176). Bettinghaus and Preston (l96u) also investigated the effects of cognitive stress on an encoder. Subjects encoded single sentences on a number of topics that were either belief-discrepant or belief-congruent. Each subject encoded both messages that were discrepant and congruent with his private Opinion. Subjects took significantly more time to encode belief—discrepant than belief-cotgruent messages. Osgood and Walker (1959) compared suicide notes with personal letters and fomd stylistic differences that they attributed to the psychological stress of the writers of the suicide notes. lazarus, Deese, and Osler (1952) reported that the induction of cognitive stress leads to longer encoding times, less learning, less recall, and more errors in the final message. To the extent that counterattitudinal advocacy results in psycho- logical stress (Festinger, 1957), changes in message output and style are to be expected. McBwen (1969) summarizes the expected effect of belief-discrepant encoding on message production: . . .messages produced by people under conditions of heightened motivation or tens ion (due to the introduction of variables tending to increase psychological imbalance) should exhibit certain measurable tendencies which are indicative of an overall performance decrement . The encoding rate should take longer (or proceed at a slower rate) and the message output should contain more errors . (p. 7) There is also evidence that encoding under cognitive stress , such as in a situation of counterattitudinal advocacy, will lead to the use of less intense language. Osgood and Walker (1959) found a higher per- centage of ambivalent constructions (e.g. "may " and "possibly") and ambivalent assertions used by encoders under stress. Taken as a whole, the research suggests that counterattitudinal encoding leads to more hesitancy and the use of less intense language. LITE. Relevance g M Intensifl £9 Attitude Research Language intensity is an important variable to communication re- searchers , for knowledge of intensity may permit inferences about the attitudinal state of a communicator and his effects on both receiver attitudes and his own attitudes. Nevertheless , many contemporary commu- nication researchers carry on studies in which they ignore intensity and most otter message variables in tleir search for predictors of attitude change. This study assumes that analyses of the messages produced in situations of cointerattitudinal advocacy can be useful for predicting the attitudinal state of the encoder. A number of terms in literary criticism and general semantics specify classes of words that express the attitudinal state of the com- municator. Thomas DeQuincey (1890) wrote of "language power;" Hayakawa (191+9) used the term "affective connotation;" and Ogden and Richards (1952) coined the term "emotive meaning." Each of these terms denotes a class of words that have an evaluative dimension expressing the com- municator's values of goodness and badness. These works suggest the following definition of language intensity: that quality of language which indicates the directim aid the strength of a communicator's attitude toward an attitude object. However, this definition offered by the semanticists is of limited use in theory building, since they contend that the intensity of a term is relative to the person judging it. Fortunately, there is empirical evidence to show that intensity is not completely relative. Osgood, Saporta, and Nunnally (1956) were among the first to devise a system for judging the intensity of a communication with high reliability. Use of their technique of evaluative assertion analysis revealed some commonality in the way pe0ple responded to various word units in their language. The study provided a basis to test the proposition that intensity is an indicatim of strength and direction of an attitude. 1mg Intensity: in Explicatiot The explicatiot of language intensity has led to three different patterns of Operetimalization: linguists have attempted to find values of intensity at the word level; others have sought to specify the intensity of phrases and sentences; investigators of counteratti- tudinal messages have used gross measures of entire messaggs. Word level. Lilly (1968a) defines language intensity as the amount of modification that adverbs have upon the meaning of object words . He suggests that the scale values of adverbs on an intensity continuum would be an equation in the form: x =AiSi+K, ii where xij is the empirical scale value of the i Eh adverb combined with the j E adjective; Ai is tle multiplying value of the i gt adverb; Si is the theoretical scale value of the j 311 unqualified adjective; and wlere K is the arbitrary zero point on the scale. In one study Lilly constructed three different questionnaires. Questionnaire l contaired intensive adverbs that were derived from sub- ject rankings . Nine adverbs were combined with seven adjectives con— noting strength, tre reutral word "average," and six adjectives con— noting weakness to produce 126 combinations. In addition, each adjec- tive was presented unqualified. Questionnaire 2 had ten probabilistic adverbs (e.g. "possibly"), while the Questionnaire 3 contained ten fre- quency adverbs (e.g. "always") combined with the adjectives. The method of smoessive intervals was used to transform the categorical ratings to an interval metric. The results indicated that the linear formula was a good predictor of the modification of meaning that certain adverbs will have on object words. In another study, Lilly (1968b) again tested this linear formu- lation by combining the same set of adverbs with sixteen adjectives to form a pool of 160 items. In order to check the reliability of the scale values obtained with the method of successive intervals, sixteen stimuli from a questionnaire with only positive-frequency adverbs (e . g. "always") and sixteen stimuli from a questionnaire with negative-fre— quency adverbs (e.g. "never") were sealed with pair-comparisons. Six- teen stimuli were placed in sixteen blocks of four items each and the subjects ranked each set of the four stimuli from most favorable to least favorable. Gulliksen and Tucker (1961) discuss the method of using multiple rank—orders to obtain pair-comparisons . The obtained correlation between successive interval and pair—comparison scaling was .97. In additiot the study reported that the sixteen adverb modifi- ers were found to fit the linear equation presented above . Howe (1966) defined intensity in terms of adverbial modification, verb tense, and negatives. Fourteen adjectives were rated on an eleven point scale of (un)favorableness singly and when preceded by twenty- one quantifiers. The quantifiers consisted of eight adverbs denoting frequency (e.g. "often"), seven adverbs denoting temporal frequency (e.g. "soon"), three verb tenses (e.g. "is," "was," "will be") and three negatives (e.g. "not," "un-," "not un-"). Data from eighty-eight sub- jects were scaled by successive intervals methods. Empirical scale values for all quantifier-adjective pairs were tested against a linear model. All four classes of quantifiers were found to exert decremental effect on the degree of evaluative polarization of the adjectives. This study adds to earlier research by including various forms of adverbs, verb tenses, and single and double negatives. There have been similar studies by Cliff (1959) and Hove (1962) which demonstrate that language can be scaled to create equal interval scales on an intensity 10 continuum. All of these studies have stopped after completion of an intensity scale. As a result, there has been little effort by linguists to use these scales to find correlates of language intensity. (lne reason for the infrequent use of these word scales is the dif- ficulty in contriving research situations in which such a small pool of words has predictive or explanatory value . It would seem that either research must be structured so that tight control restricts possible word use to the word pool with scale values , or the pOpulation of scaled words must be increased so that freely encoded messages can be analyzed. Phrase - Sentence level. Bowers (1963) was among the first to investi- gate the effects of language intensity and other variables upon attitude change. He defines language intensity éfi the quality 9_f_ the lme that indicates the degree E which the speaker's attitude toward a @- cept deviates from neutrality. Judges rated words and phrases according to intensity, and from tlese ratings high and low intensity messages were created. A predicted interaction between language intensity and social introversion on attitude change was not confirmed but low inten- sity messages against a topic were more persuasive on a separate audi- ence. The highly intense language caused a "boomerang" effect that lowered the credibility of the source and inhibited attitude change . In a later study (Bowers, 1969), a list of correlates of language intensity was presented. Again, language intensity was defined in terms of judges' ratings of l$82 sets of items. The results were: (1) a low but significant correlation between intensity and word length, (biseral 55 .10), (2) a moderate correlation between intensity and obscurity of words (tetrachoric r_ = .59), (3) a high correlation between intensity 11 and the presence of qualifiers (tetrachoric 3 = .89), and (A) a high correlation between intensity and metaphorical quality (tetrachoric _r_'_ = .83). Although the method of using judges' ratings of intensity leads to less precision than the scales reported earlier, the two stu— dies by Bowers suggest relationships between language intensity, person— ality variables, and attitude change . Message level. If language intensity is defined as the quality that indicates the degree to which the speaker' s attitude deviates from neutrality, the self-persuasion studies previously reviewed report gross measures of overall message intensity. Overall persuasiveness and ex- tremity of position advocated are message level measures of deviation from neutrality. The comterattitudinal encoding studies consistently found no dif- ferences in counterattitudinal messages across treatment cmditions . There are several possible explanations for this lack of findings. First, no one compared belief-discrepant and belief-congruent essays. Burgoon (1969a) found that judges rated belief-discrepant messages neu- tral in 31% of the judgments; only I+996 of the essays advocating the belief—discrepant side were correctly identified as belief-discrepant, and 20% of the belief-discrepant messages were rated as advocating the belief-congruent position:L By contrast, 96% of the belief-congruent messages were correctly identified. 1The analysis was performed on messages produced in a study by Miller and Bodaken (1969). Five messages were randomly drawn from the four treatment conditions yielding ten belief-congruent and ten belief - discrepant essays for analysis . Thirteen faculty and graduate students rated the messages as advocating "pro- or anti- mandatory on-campus living" or "neutral as to topic side." This yielded 260 judgments (l3 judges x 20 messages) that were analyzed by a 3 x 2 Chi-square 12 The conclusions drawn from these data suggest that the subjects did not fulfill the requirements of counterattitudinal advocacv in a majority of the essays. More precise measures are needed to establish the differences in deviation from neutrality of belief-discrepant versus belief-congruent essays. Moreover, research should be undertaken to determine the correlation between the intensity of the counteratti- tudinal position and posttest attitude. The Relationship Between Intensity gr: Assertion and Source Attitude M. There have been few attempts to specify the relationship be- tween the intensity of an assertion and attitude change following counterattitudinal advocacy. Festinger (1957) posits "believing x" and "saying not-x" leads to psychological inconsistency. It has been con- sistently demonstrated (Festinger and Carlsmith, 1959; Carlsmith, Collins and Helmreich, 1966; Miller and MoGraw, 1969) that one way of reducing this inconsistency is to change private opinion to more closely conform to the advocated position. None of the research demonstrates that "believing x" but "strongly advocating not —x" leads to more atti- tude change than "moderately advocating not-x. " design. Chi-square = 56.06, p (.001. Cmdition Congruent Discrepant congruent Correct 96% 1+9% treatment- JUDGMEN'I‘ Neutral His 71% *Correct = con ent Error 0% 20% ° BI“ V judgment discrepant _ treatment- ‘ discrepant judgment 13 Nevertheless, dissonance theory offers several reasons for pre- dicting a direct relationship between message intensity and attitude change. First, dissonance theory postulates that increased commitment to belief-discrepant encoding increases the magnitude of dissonance and resulting attitude change (Ashmore and Collins, 1968; Helmreich and Collins, 1968). Forcing a person to "strongly say not-x" is a method of inducing commitment to the counterattitudinal message. If a person encodes a very intense message, the probability of the receiver perceiving what position is being advocated is high. (m the other hand, Burgoon (1969a) demonstrated that some counterattitudinal messages are so ambiguous that an audience can not determine the advocated position. This ambiguity stemmed largely from the low intensity of the counter- attitudinal message; in fact, as indicated above, numerous messages were perceived as advocating the other side of the issue. A person encoding a highly intense counterattitudinal message satisfies McGuire's (1961+) conditions for high commitment: (l) a pri- vate decision to encode the message, (2) a public pronouncement of the position, (3) active participation, and (A) external commitment (a per- son telling another that he holds a particular belief). If a message is of low intensity or is neutral the fourth condition is apparently not satisfied. This might account for the lack of attitude change in the Miller and Bodaken (1969) study, since a majority of the messages judged could not be identified as advocating the counteratt itudinal position. In summary, highly intense counterattitudinal messages should increase commitment to the belief-discrepant task , and one possible way of avoiding dissotance is to write messages that do not convey a mes- sage that is interpretable . In Another of Festinger's (1957) postulates leads to the prediction of a direct relationship between assertion intensity and attitude change. Festinger states that tlne greater the difficulty in reversing a decision the greater the dissonance produced by the decision. If one advocates a position that deviates markedly from neutrality, the posi- tion of the encoder should be apparent to the receiver. This would make the possibility of denying the counterattitudinal nature of the message more difficult. For enample, when a source constructs a message with low intensity modifiers such as "doubtfully" and "conceivably" (Lilly, 1968), he should find it easier to deny that he ever actually took a belief-discrepant position. A probabilistic assertion containing such modifiers allows the encoder the opportunity to concede that the opposite of the assertion might in fact be the correct position. If the encoder is forced to state a position with a high degree of certainty or definiteness , tle opportunities to deny the correct— ness of his assertion are reduced. Thus, the more definite the source makes his assertion the more difficult it is to accept any other posi- tion or assertion and, in effect, reverse his decision. The irrevo- cable nature of hign intensity assertions should lead to an increased magnitude of dissonance and therefore more attitude change should re- sult when the counterattitudinal encoding behavior is of high intensity. Rationale and Hypotheses The research evidence summarized above leads to the following con- clusions: (1) counterattitudinal advocacy has a measureable effect upon message production, and (2) there is a direct relationship between mes- sage intensity and the magnitude of attitude change resulting from 15 counterattitudinal encoding. This investigation tested the following hypotheses: Prior attitude as a )redictor of language intensity. Based upon the work of McBwen (1969), Burgoon (1969a) and Osgood and Walker (1959) it was predicted that the encoding style of persons writing counteratti- tudinal essays will differ from others writing belief-congruent essays, Specifically: H1: Persms who encode a counterattitudinal message will use language of significantly lower intensity than will persons who encode a belief-congruent message. Language intensity as a predictor of attitude. It was predicted based upon the Theory of Cognitive Dissonance (1957) and subsequent research that the greater the difference between prior attitude and the intensity of the counterattitudinal advocacy, the greater the magnitude of the attitude change, specifically: H2: Given counterattitudinal advocacy by people with similar prior attitudes , the magnitude of attitude change varies directly with the intensity of the counterattitudinal assertion . CHAPTER II EXPERIMENT I Overview All experimental subjects wrote either a belief-discrepant or belief-congruent message. The intensity of the messages created was compared for groups of subjects arguing counterattitudirally and atti- tudinally-consistent. Each subject's attitude and latitude of accep- tance toward the experimental issue was assessed imnediately after completion of the required encoding tasks . §_s (11 = tt5) were members of an introductory business writing class at Michigan State University. Subjects were told that the project re- quired help in creating written messages and thus business writing classes had been selected. Emoding condition was the independent variable used . Each sub- ject wrote either a belief—discrepant or belief-congruent message. The overall design is presented in Figure l. Encoding Conditions Belief —Congruent Belief-Discrepant language IntenSity Figure 1. The Experimental Design: Experiment I 16 17 Language intensity measures were obtained for each message in both cells. Comparisons were then made between the two encoding con- ditions . Procedures Pretest Two weeks before the actual experiment, §_s were administered a pretest questionnaire, ostensibly to solicit student opinion on possi- ble issues to be used in a project for another class. (See Appendix A) The instrument contained several issues believed to be salient for undergraduate students. Each issue was followed by a seven-interval scale, with the intervals labeled excellent, good, fair, neutral, poor, Ed, and terrible ('Ihurstone and Jones, 1955). The _S_s were instructed to mark an "A" above the word that best represented their attitude to- ward the issue. The Thurstone and Jones-type measure was developed by computing scale values for word anchors. Successive interval scaling techniques yielded values expressed in standard score deviations from a neutral point. This lawn-interval instrument allowed precise estimates of the width of each unit on the scale. Prior correlational analysis yielded an r of .86 between this scale and a standard four-item semantic differ- ential-type scale using polar adjectives loading on the evaluative di- mensicn. The Thurstone and Jones-type scales were scored by using the stan- dard score weight assigned to each anchor. Figure 2 presents the scale and the weighted values of the anchors. For analysis purposes, u.1 was added to each value to yield a scale with a low value of 1.0 and a high of 7.8. 18 7.8 6.0 U.9 “.0 2.9 2.1 1.0 g man‘s-s m S'fig‘la #1 d E: H W ‘3 "’ '6‘ d’ Figure 2. Thmstone and Jones-type Attitude Scale None of the issues tested produced a bimodal distribution that would have allowed the use of the same set of words for each experimental condition. Thus, it was necessary to select the issue that had the most skewed distribution so that two separate encoding conditions could be created. The issue producing the most skewed distribution advocated mandatory on-campus living during college attendance . Independent Variable One independent variable was manipulated in this study. This variable was dichotomized as Belief-Congruent Encoding and Belief- Discrepant Encoding. Since the attitude issue yielded a skewed dis- tribution, all §_s who encoded a message supporting mandatory on-campus living comprised the Belief-Discrepant condition. Those S5 who argued against mandatory on-campus living comprised the Belief-Congruent con- dition. Inductims and Posttest mo weeks after the pretest had been administered and the experi- mental issue had been chosen, §s were induced to perform the encoding task. _E_I gave the following instructions to _S_: 19 Dear Student: For the past several months, the Center for Opinion Research at Michigan State University has been conducting research aimed at finding out what it takes to strengthen or change people's attitudes. What we do is get some indication of the attitudes people have on certain issues and then, at some later date , we present certain arguments in the form of speeches, essays, debates, -- various methods - which are designed to get them to change or strengthen their attitudes. Let me explain briefly what we are attempting to do in the research that you will be helping us with. Sometime ago we administered questionnaires to students enrolled in the Lansing-East Lansing Area High Schools asking them their attitudes toward the policy of required cn—campus living during their attendance at college. Analysis of the ques- tionnaires indicate that these students are generally UN- IIICIDED m the issue. we then asked over 500 Michigan State University students to write persuasive messages to support (reject) a policy that would require students to live on campus while attending the University. You will soon be given the arguments that stu- dents like you created. It is even possible that some of you helped in the original research effort. We found that the arguments produced were similar in content. HOWEVER, (DR ANALYSIS OF THE PERSUASIVE EFFECT OF THESE MESSAGES INDICATES THAT CERTAIN WORDS ARE THE POST LORI‘ANT IN CHANGING THESE MODERN W _- Thus, in preparing your persuasive messages it is very im- portant for you to choose what you feel are the most per— suasive words. It is important that you select the words you think will be successful for we feel that all of our effort to date depends on your choice of the persuasive words . Think aboutthechoices andtrytodecidewmrichwordineachargrmrent might be best used to change attitudes. Also the order of the arguments is importantandwewantyoutothinkaboutwhat order of presentation might be most effective. This part of the research is very important to the successful completion of the project. We want to thank all of you for helping us create messages that we will show to high school students in an attempt to change their attitudes toward on- campus living. Remember we want you to persuade the students to support (reject) mandatory cn-campus living. _Ss were then given a message with ten blank spaces and were told to use only words on an attached list to create the most persuasive message they could to change high school students' attitudes. (See 20 Appendix B) After 15 minutes, the messages were collected and the post-encoding questionnaires were distributed. (See Appendix C) The instrument contained the same Thurstone and Jones-type scale em- ployed in the pretest to measure attitudes toward compulsory on-campus living. The post-encoding instrument presented tle anchors in complete sentences to avoid sensitization resulting from using an instrument identical to the pretest questimnaire . Also, the instrument contained items designed to measure the Ss' perceptions of the importance of the encoding task, audience attitude , and task difficulty. To ensure randomization of S3 to treatment conditions , two dif- ferent persuasive word lists were distributed as the SS entered the room. (he list of words allowed the subject to encode a message favor- ing compulsory on-campus living; the other list contained only words that would allow the encoding of an essay opposing the issue. Preparation of EQerimental Material The dependent variable in this study was language intensity: the quality that indicates the degree to which the writer's attitude deviates from neutrality. Thurstone and Jones (1955) developed scale values for 51 words that measured deviation from a neutral point. Through the method of successive interval scaling based upon judgments of a like-pqaulation of students, standard score weights were assigned to the pool of words. Table 1 presents the population of words and the standard score for each word or phrase. In the Belief-Comment condition (opposition to mandatory on-campus living), _S_s were given a choice of words with scale values ranging from -O.3O to -6.1m: in the Belief -Discrepant condition (support of mandatory on-oampus living) the scale values ranged from +0.69 to +6.15. Table l. Weighted scale values of the experimental words Best of all 6.15 Fair .78 Favorite '4 . 6 8 Acceptable . 7 3 Like extremely l4.16 Only fair .71 p Like intensely l4.05 Like slightly .69 Excellent 3 . 71 Neutral . 02 Wonderful 3.31 Like not so well - .30 Strmgly like 2.96 Like not so much - .91 Like very much 2.90 Dislike slightly - .59 Mighty fine 2.88 Mildly dislike - .7u Especially good 2.86 Not pleasing - .83 Mighty favorable 2.81 Don't care for it -l.10 Like very well 2.60 Dislike moderately -1.20 Very good 2.36 Poor -l.35 Lile quite a bit 2.31 Dislike -l.58 Enjoy 2.21 Don't like -l.8l Preferred l. 96 Bad -2 .0 2 Good 1.91 Highly unfavorable —2.16 Welcome 1.77 Strongly dislike -2.37 Tasty 1.76 Dislike very much -2.'49 Pleasing 1.58 Very bad -2.53 Like fairly well 1.51 Terrible -3.09 Like 1.38 Dislike intensely -3.3l Like moderately 1.12 Loath -3.76 OK .87 Dislike extremely -6.22 Average . 86 Despise -6 . nu Mildly like .85 Messages were constructed with ten blank spaces left for insertion of the word choices. (he blanked message was used for both treatment cmditions. language intensity scores were obtained by summing the scale values of the words written in the ten blanks . To ensure that the two treatment conditions did not differ in language intensity prior to the experiment, a E-test was used to compare the mean scale values of the words used in both treatment conditions. This analysis yielded a mean of 2.3 in the Belief-Discrepant condition and a mean of 2.2 in the Belieth conditim; these scores were not statistically different . 22 In addition, attitude change and latitude of acceptance were ana- lyzed to measure differences resulting from the encoding condition. Correlaticnal analyses determined the relationship between attitude change and language intensity. Cmtrol Variables 93<_>i_c§_. Central to Festinger's (1957) theory of cognitive dis- smance is choice to commit belief-discrepant behavior. It is only when the individual makes a choice that he experiences dissonance. Since this study sought to create conditions conducive to dissonance arousal, all _S_s were given a choice to participate in the experiment. Audience Commitment. Nel, Helmreich, and Aronson (1969) and Bodaken (1970) offer SLpport for the position that possible aversive consequences increase the amount of dissonance experienced. _S_s in this study were told that the target audience was uncommitted on the issue of mandatory on-oampus living . This was intended to increase the aver- sive consequences of the encoding behavior by not only placing the _S_ in the positim of creating a persuasive message but also increasing the probability of that mes sage persuading the intended audience . Publicness. Collins (1968) and others contend that public com- mitment to a position increases the dissonance associated with counter- attitudinal advocacy. All S_s were instructed to place their names on the messages so that they could be shown to high school students. Manipulation Checks The importance of the contributim to the persuasion project was étermined by having _S_s respmd to the following question: "Did you 23 feel that your contribution to the persuasion campaign was important?" _S_s responded on a seven-interval semantic differential-type scale bounded by the polar adjectives m - unimportant. The effectiveness of the audience attitude manipulation was deter— mined by having SS respond to the following question: "Before you wrote your essay, how did you think the high school students felt toward the issue of compulsory on-campus residency?" SS responded by marking either Favorable, Undecided, Opposed, or Don't Know. The amount of perceived difficulty in completing the experimental task was measured by having S_s respond to the following question: "I-bw difficult was it for you to write an essay on this issue?" _S_s responded on a seven-interval, semantic differential-type scale bounded by the polar adjectives difficult - e_a_s_y_. To ensure that subjects were arguing either counterattitudinally or attitudinally-consistent , latitude of acceptance scores were obtained. In addition to marking personal attitude on the Thurstone and J ones-type scale, _S_s were instructed to place an "X" by each word they could accept concerning the attitude statement . Latitude of acceptance was computed by subtracting the lowest valued acceptable statement from the highest . In the Belief-Discrepant condition, the encoded statements included words outside the Ss' latitudes of acceptance. The message encoded in the Belief-Congruent condition contained words within the _S_s' latitudes of acceptance . 24 Results Manipulation Checks Choice. No direct written measure of perceived choice was ob- tained. However, in the Belief—Discrepant condition four subjects re— fused to participate, while two subjects chose not to encode the mes- sages in the Belief-Congruent condition. Audience Attitude. The effectiveness of the audience attitude man- ipulatim was determined by computing the number of individuals in each treatment condition who correctly perceived the experimental induct ion. In the Belief-Congruent condition 70% of the subjects perceived the amiience to be uncommitted on the issue while 59% of the subjects in the Belief-Discrepant condition correctly perceived the experimental in- duction. A Chi-square test indicated that these percentages were not significantly different. Although the induction was successful for a majority of the subjects in each cell, the Belief—Discrepant condition had a rather large number of subjects who incorrectly perceived audience attitude 041%). Difficulty. To determine the amount of difference in perception of task difficulty in the two experimental conditions , a t—test on mean differences was computed. On a seven-point scale, with difficult being scored as one, the Belief-Congruent condition had a mean of 3.13 while the Belief-Discrepant condition had a mean of 3.09. These means were not significantly different . Subjects' ratings indicated that both groups rated the task as moderately difficult. Importance. On a seven-point scale, with important being scored as seven, the Belief-Congruent condition had a mean rating of l4.63 25 while the Belief-Discrepant condition had a mean rating of 3.90; these means were not significantly different. Apparently, subjects perceived the task as moderately important . Latitude of Acceptance. The pretest attitude ratings in the Be- lief-Congruent condition yielded a mean of l. 35 with a latitude of ac- ceptance of l. 35 scale units. The Belief-Discrepant condition had a mean of 1.82 with a latitude of acceptance of 1.56 scale units. The pretest measures indicated that the attitude issue was very skewed and all subjects had a latitude of acceptance on the negative side of neu- trality. Thus, the results indicated that any subject who encoded statements on the positive side of neutrality was encoding a belief- discrepant message, i.e. , the statements encoded were not included in the range of acceptable statements. Any subject who encoded a message a: the negative side of the neutral point was arguing attitudinally- consistent . Test of the Hypothesis The first hypothesis of this study was tested by means of a _t_-test. Data were the summated scale values for the ten words chosen to complete the message. The significance level employed for all analyses was .05. The first hypothesis predicted that belief-discrepant messages would differ in language intensity from belief-congruent essays, speci- finally: Persons who encode a comterattitudinal message will use language of significantly lower intensity than will persons who encode a belief-congruent message. . Table 2 indicates the level of language intensity and presents a smuary of the test of differences between the two experimental groups. 26 The results of this analysis indicate that Hypothesis 1 is supported and that subjects encoding counterattitudinal messages used less in- tense language than subjects encoding belief-congruent messages. Had the subjects in the Belief-Congruent condition selected the ten most intense words, the mean intensity rating would have been 310.39; the mean intensity rating of the ten most intense words in the Belief—Discrepant condition was 37.66. This difference is a potential cmservative bias as the hypothesis predicted that the Belief-Discrepant cmdition would be less intense. The results indicated that while both groups encoded moderately intense messages, the intensity of the Belief- Discrepant messages were significantly less intense than those produced in the Belief-Congruent condition (Table 2). Table 2. Means, standard deviations and t-test of the difference in language intensity scores in the experimental con- ditions. Belief-Congruent Belief-Discrepant t P Y = 25.28 Y = 22.20 2.33 < .05 S.D.= 2.72 S.D.= 2.98 3.05, 93 = to, = 2°02 Supplementagg Analyses In addition to the analysis of the language intensity data, anal- yses were performed on measures related to the attitude and latitude of acceptance changes of the subjects in both experimental conditions. 27 Attitude Change Pretest. After random assignment of subjects to experimental con- ditions, a E—test was performed on the attitude scores of the two exper- imental groups to ensure that the pretest scores were not significantly different. The mean pretest attitude of the Belief—Congruent condition was 1.35 while the Belief-Discrepant condition had a mean of 1.82. The results of the analysis indicated that the mean ratings did not differ significantly. Pom. Immediately after the encoding task, attitude measures were again obtained. Table 3 indicates the amount of pretest through post-encoding attitude change and presents a summary of the test of mean differences between the two groups . The Belief-Congruent condition showed more attitude change than did the Belief-Discrepant condition although the differences were not statistically significant. Although no specific predictions were made concerning attitude change in this experiment, it was expected that the Belief-Discrepant condition would show more attitude change. The results are inconsistent with such an expectation . Table 3. Means and _t-test of differences in pretest through post- encoding attitude change in the experimental conditions . Condition Pretest Posttest Change t P Belief Congment 1.35 2.80 1A5 < l N.S. Belief Discrepant 1.82 2.82 1.00 E. .05, g; = to, = 2-02 28 A product-moment correlational analysis of the relationship be— tween attitude change and language intensity in the Belief-Discrepant condition yielded an r of .28 which was not significant (r 05 df _ 21 .. *0 a __ " 9 = .36). Latitude of Acceptance Pretest. After random assignment of subjects to experimental con- ditions , a t-test was performed on the latitude of acceptance pretest scores to ensure that the two experimental conditions were not signif- icantly different. The mean latitude of acceptance in the Belief- Congruent condition was 1. 35 while the Belief-Discrepant condition had a mean of 1.56. The results of the analysis indicated that the group means did not differ significantly. Posttest. In conjunction with the posttest attitude measure , the subjects completed latitude of acceptance measures . Table It indicates the amount of pretest through post-encoding latitude of acceptance changes and presents a summary of the mean differences between the two groups . The analysis indicated that encoding condition had no effect an the width of the latitudes of acceptance of the individuals perform— ing the task. Table 1+ . T—test of differences in pretest through post-encoding Iatitude of acceptance change in the experimental con- ditions. Condition Pretest Posttest Change t P Belief Congruent 1.35 1.08 .13 <1 N.S. Belief Discrepant 1.56 1.77 .21 3.05, g_f_ .-. 1:0, = 2.02 CHAPTER III EXPERIMENT II Overview All experimental _S_s encoded one belief-discrepant message that was either high, moderate, or low in language intensity. Attitude change scores for groups of SS arguing counterattitudinally with dif- ferent levels of language intensity were compared. SS (11 = 106) were obtained from three introductory business writing classes at Michign State University. The subjects were told that the project required help in creating written messages and thus business writing classes had been closen. Language intensity was the independent variable used in this study. The variable language intensity took three values: high, moderate, and low. The overall design is presented in Figure 3. Language Intensity Hign T‘bderate Um Attitude Change Figure 3. The Experimental Design: Experiment II Attitude change scores were computed for each _S_ in all cells. Com- parisons in terms of the dependent measure were made among the three encoding conditions. 29 30 Procedures Pretest . Two weeks before the actual experiment, the identical pretest procedure described in Experiment I was completed on §_s in this investigation. (See Appendix A) Indgpendent Variable (be independent variable , language intensity , was manipulated in this study. This wariable is defined as’the quality that indicates the degree to which the writer's attitude deviates from neutrality . The Thurstone and Jones (1955) word values were used as measures of language intensity . The treatment canditions were created to meet the following criteria: (1) that the mean value of the three groups differ from each other by at least one _z_ score, and (2) that the groups be statistically different from each other. Table 5 presents a summary of values of the three language intensity treatment conditions . Induct ions and Posttest Two weeks after the pretest had been administered and the experi— mental issue had been chosen, §_s were induced to perform the encoding task. The same induction used in Experiment I to urge S5 to encode a message supporting mandatory on-campus living was used in this study. _S_s were then given an envelope containing ten strips of paper, each with an argument typed on it. Each sentence had at least one blank space in it. The _S_s were given the cloice of two words or phrases to insert in the blank to create the most persuasive message they could. In all treatment conditions §_s were given only words that matched the level of intensity they had been assigned; e.g., SS in the hign intensity condition had a cl'oice of two highly intense words for each sentence . 31' Table 5 . Word values , means , standard deviations , and summary of the analysis of variance of differences of words selected for use in the experimental conditions. Like extremely 0.16 Very good 2.36 Poor Like intensely 0 .05 Like quite Like moderately Excellent 3.71 a bit 2.31 OK Wonderful 3 . 31 Enjoy 2 . 21 Average Terrible 3 . 09 Highly Mildly like Strongly like 2.96 unfavorable 2.16 Not pleasing Mignty fine 2.88 Bad 2.02 Fair Like very much 2.90 Preferred 1.96 Acceptable Especially Good 1.91 Only fair good 2.86 Welcome 1.77 Like slightly Mighty Pleasing 1.58 favorable 2 . 81 Like fairly Very bad 2.53 well 1.51 _ Like _ 1.38 _ x = 3.21 x = 1.92 x = .88 SOD. = .5“ SOD. : .33 SOD. = .21 Analysis of Variance Summary Source 21: Variance §_S_ gt; 1~_1§_ _F_ P Between treatments 28.57 2 10.28 90.78 (.05 Within treatments 0.37 2_9_ .15 Total 32.90 31 .505, g; = 2,29, = 3°33 1.35 1.12 .87 .85 .83 .73 .71 .69 32 The _Ss were told to write the most persuasive word in the blank and to order the arguments to create a total message. After 15 min- utes, the SS were given a blank piece of paper and told to write the message they had created on this page. They were instructed to place their name on this page so that they could receive proper credit for their ideas when the messages were slown to high school students. After another 20 mninutes, the messages were collected and the post-encoding questionnaires were distributed. (See Appendix C) This instrument was the same as the post-encoding questionnaire described in Experiment I. To ensure randomization to the three treatment conditions , the envelopes containing the words with varying levels of language inten— sity were randomly distributed. The control group was an intact group (fl = 15) which only filled out the pretest and post-encoding question- naire . Dependent Variable The dependent variable in this study was pretest through post- enccding attitude change. The Thurstone and Jones-type scale, pre- sented in Figure 2 (Onapter 2), was used to measure subject attitudes. The instrument also contained items designed to measure Ss' perk captions of the importance of the task, the attitude of the target audience, and the difficulty of the task. The same control variables and manipulation checks that were described in detail in Experiment I were used in this study 33 Results Manipulation Checks 93393. No written measure of perceived choice was obtained. However, in the high intensity condition four subjects refused to par- ticipate, one refused in the moderate intensity condition, and no one refused in the low intensity condition. Audience Attitude . The effectiveness of the audience attitude manipulation was determined by computing the number of individuals in each treatment condition who correctly perceived the experimental in- duction. A Chi-square test was performed to determine if there were differences in the three experimental conditions. Tlnere were no sig- nificant differences in the perceptions of the three groups; however, the manipulation was successful as all goups had a majority perceiving . the audience attitude as intended (Table 6). Table 6. Percentage, frequency, and Chi-square test of subjects correctly perceiving audience attitude . Condition Correct Incorrect X2 P High Intensity 65% (23) 35% (12) Pbderate Intensity 60% (16) 36% ( 9) Low Intensity 79% (23) 21% < 6) 1.90 N.S. x2 = 5.99 —.05, 9:; = 2, Difficulty. To determine the amount of difference in perception of task difficulty in the three experimental conditions, a simple anal- ysis of variance was performed. No differences were found in the 30 experimental conditions. Subjects' mean ratings indicated the task was moderately difficult. The ratings produced a mean of 0.00 in the Hign Intensity condition, 0.11 in the Moderate and 0.05 in the Low. Importance . To determine the perception of the importance of the encoding, a simple analysis of variance was performed. No differences were found in the experimental conditions. All experimental groups had mean ratings near the midpoint of the seven-interval scale. The ratings produced a mean of 3.60 in the High Intensity condition, 3.78 in the Moderate and 3.72 in the Low. Latitude of Acceptance . Mean attitude and latitude of acceptance ratings indicated that all subjects' latitude of acceptance was on the negtine side of neutrality. This indicated that the subjects in all three experimental conditions were encoding statements that had been labeled in the pretest as unacceptable . (Table 7) Table 7 . Pretest attitude and latitude of acceptance measures . Latitude of Condition Attitude bean Acceptance Mean High Intensity 1.00 1.58 Moderate Intensity 1.07 1.33 Low Intensity 1.27 1.50 Pretest. After random assignment of subjects to experimental and control groups, a simple analysis of variance was performed on the pre- test attitude scores to ensure that they were not significantly differ— ent. The results of this analysis indicated that the group means did not differ significantly (2 = 1.01, N.S.). 35 Table 8 . loans and standard deviations for experimental and control SIOUPS- Condition Mean S.D. Hign Intensity 1.00 .65 Moderate Intensity 1.07 .59 Low Intensity 1.27 .56 Control 1.63 .99 Experimental and Control Differences Attitude change scores of the experimental and control groups were submnitted to a simple analysis of variance. The overall amount of change and analysis of variance results are found in Table 9. The results cbmonstrated that the four groups differed significantly on amount of posttest attitude change. The specific differences between each experimental group and the control group were computed by means of Scheffe's test. The results of the test indicated that both the High and boderate Intensity conditions differed significantly from the con- trol group. No significant differences were found between the Low In- tensity and Control conditions . (Table 10) Test of the Hymthesis The second hypothesis of this study was tested by both a simple analysis of variance and Product-Monent/Eta correlation coefficients. Data were the mean pretest to post—encoding attitude change scores of the experimental subjects using the seven-interval Thurstone and Jones- type scale. The significance level employed for all analyses was .05. 36 Table 9. Attitude scores, amount of change , and analysis of vari- ance of attitude change scores of experimental and con- trol groups. Condition Retest Posttest Change Hign Intensity 1.00 3.90 2.50 Moderate Intensity 1.07 3.58 2.11 Low Intensity 1.27 2.60 1.37 Control 1.63 1.91 .28 Source of Variance SS df MS F P Between Treatments 58.85 3 19.62 7.75 (.05 Within Treatments 2% _1_0_2_ 2 . 5 3 Total 316.58 105 F " 2.76 Table 10. Analysis of differences of attitude change scores between each experimental and control group . Condition A Mean Control High Intensity 2.50* .28 Moderate Intensity 2 . 11* Low Intensity l. 37 *P (.05 37 The second hypothesis predicted a direct relationship between attitude change and language intensity; specifically: Given counterattitudinal advocacy by people with similar prior attitudes, the magnitude of attitude change varies directly with the intensity of the counteratt itudinal assertion. Table 11 indicates the amount of pretest through post-encoding attitude change and presents a summary of the analysis of variance for the experimental groups. The results indicate that Hypothesis 2 is supported. Table 12 presents the results of specific comparisons of experimental groups using Scheffe's test. The High and Low Intensity experimental groups are significantly different and the analysis of variance results show support for Hypothesis 2. As a further test of Hypothesis 2 , a Product-Moment correlation coefficient was conputed on the relationship between language intensity and attitude change. This analysis yielded an 3 = .27 which is signif— icant (2.05, df = 90,: .17). To check the linearity of the- relationship between language intensity and attitude change, an Eta coefficient was computed. Since the Eta coefficient (.28) was not significantly greater than the Product-Moment coefficient, this established the linearity of the relationship. The correlational analyses are further support for Hypothesis 2 . Supplementagy Analyses In addition to the analyses of attitude change scores, analyses were performed on latitude of acceptance measures. 38 Table 11. Attitude scores, amount of change, and analysis of vari- ance of attitude change scores of experimental groups . Condition Pretest Posttest Change High Intensity 1.00 3.90 2.50 Moderate Intensity 1.07 3.58 2.11 Low Intensity 1.27 2.60 1.37 Source of Variance SS df MS F P Between Treatments 20.92 2 10.06 3.71 < .05 within Treatments 207.75 §_§_ 2.82 Total 268.68 90 F = 3.15 _005’ g = 2, 60, Table 12. Analysis of differences of attitude change scores between each experimental group. High Moderate Low Condition Intensity Intensity Intensity Means 2_.5_0 fl 1. 37 High Intensity ---— N.S. 2.70* Moderate Intensity ---- N.S. Low Intensity -...... *P <.05 Critical K = 2 .51 -.05, g = 2, 60, 39 Latitude of Acceptance Pretest. After random assignment of subjects to experimental and control conditions , a simple analysis of variance was performed to ensure that the groups were not significantly different on pretest lat- itude of acceptance scores. The analysis indicated no significant dif- ferences between groups. (Table 13) Posttest . Pretest through posttest latitude of acceptance change scores were analyzed by a simple analysis of variance. The results in- dicated no differences in latitude of acceptance between groups . (Table 10) Table 13 . Latitude of acceptance means , standard deviations , and analysis of variance of pretest latitude of acceptance scores for experimental and control groups . Condition Mean S.D. Hign Intensity 1.58 1.00 Pbderate Intensity 1.33 .79 Low Intensity 1.50 .59 Control 1.01 .90 Source of Variance SS df . PB F P Between Treatments 1.10 3 .38 <1 N.S. Within Treatments 72.00 13; .71 Total 73.50 105 F = 2.76 -005, ,9;- = 3, 60, 00 Table 10. Latitude of acceptance scores, amount of change , and analysis of variance of latitude of acceptance change scores of experimental and control groups . Condition Pretest Posttest Change High Intensity 1.58 1.95 .37 Moderate Intensity 1.33 1.61 .28 Low Intensity 1.50 1.60 .06 Control 1.01 1.82 .01 Source of Variance SS df MS F P Between Treatments 3.28 3 1.10 1.56 N.S. Within Treatments 71.52 E .70 Total 70.80 105 F = 2.76 ‘7005, df = 3, 60, CHAPTER IV DISCUSSION Research on counterattitudinal advocacv has generally demonstrated that a person who encodes a belief-discrepant message will shift his attitude to more closely conform to the advocated position. As a re- sult of the emphasis placed on attitude change as the primarv dependent variable, little attention has been given to the encoded message, nor has prior research manipulated the possible messages to predict atti- tude change. The purpose of this investigation was twofold: (l) to examine the effects of counterattitudinal encoding on message style, and (2) to investigate ways in which encoding situations can be con— structed to predict the magnitude of attitude change resulting from counterattitudinal advocacv. Prior research by Festinger and Carlsmith (1959); Rosenberg (1965); Carlsmith, Collins, and Helmreich (1966); and Linder, COOper, and Jones (1967) all analyzed supposedly belief-discrepant messages. All used some overall evaluation of persuasiveness or quality. The researchers consistently failed to find any significant message differences among belief-discrepant treatment groups. Also, no evidence was found that any nessage measure was correlated with posttest attitudes. The sup- port of the two theoretical hypotheses in this study provide suggestions for another approach to the analysis of counterattitudinal messages. 01 02 Hypothesis I predicted that persons encoding a counterattitud inal message would use significantly less intense language than persons encod- ing a belief-congruent message. All of the prior research has neglected to compare belief-congruent messages with counterattitudinal messages. It was reasoned that even though prior gross evaluations of belief-dis- crepant messages have been disappointing, the stress associated with counterattitudinal advocacy might produce effects on message style. Consistent with this reasoning, the results of the study indicate that people do encode less intense messages when committing belief-discrepant behavior. Given this support for Hypothesis 1, it would be useful to compare the Festinger and Carlsmith (1959) rating of persuasiveness of belief- discrepant messages with belief-congruent messages on the same issue. Similar congruent-discrepant carpariscns could be made on Carlsmith, Collins, and Helmreich's (1966) measure of positiveness. Since none of the prior research manipulated language intensity, this study also sought to determine the relationship between how stronglv a person argues against a position he privately holds and the magnitude of his attitudinal shift toward the advocated position. Hypothesis 2 predicted a direct relationship between the intensity of a counteratti- tudinal assertion and the magnitude of subsequent attitude change. The support of this hypothesis is encouraging. The results indicated that forcing a person to strongly advocate a belief-discrepant position re- sulted in significant attitude change. Moreover, although the widths of the latitude of acceptance showed no significant post-encoding changes among the experimental groups, the nature of the acceptable statements '43 changed. For example , prior to the experiment the average individual in the High Intensity encoding condition held the attitude that man- datory cum-campus living was terrible and accepted that it was bad and poor. After encoding a belief-discrepant message he claimed to be neutral toward on-campus residency , but could accept that it was poor and fair. Thus with the large attitudinal shifts and the maintenance of latitudes of acceptance nearly two scale units wide, the nature of acceptable statements following comterattitudinal advocacy was di f fer- ent. In Experiment I , there was no difference in pretest through post- encoding attitude change between the belief-congruent and belief-dis- crepant encoding conditions . In Experiment II , the low language in- tensity condition did not differ significantly from a control group. Taken together this evidence suggests that there may be a threshold of assertion intensity below which attitude change does not occur. These findings suggest possible reasons why sate cormterattitudinal advocacv studies (Siegel, 1969; Janis and Gilmore, 1965; Miller and Bodaken, 1969) have failed to report significant attitude change following be- lief-discrepant encoding. The results suggest that if persons encode messages below a certain threshold of intensity, attitude change may not occur. Unfortunately, prior research has not systematically in- vestigated the messages produced to specify the level of language in- tensity used. Also, the clear establishment of such a threshold is not produced by the data of these two experiments. In the first experiment, subiects used moderately intense language and did not demonstrate significant nu attitudinal shifts; however, the moderate intensity group in the second experiment showed a significant attitude change. A planned replication of these experiments will make a methodological change to attempt to resolve these conflicting outcomes. The second experiment in this study required more effort of the subjects, as each subject had to write out the entire message from an assortment of arguments. In the first experi- ment the subjects only had to write the words in the blanks. Thus, ef- fort required to commit camterettitudinal behavior might have con- founded the results when comparing attitude change across the two exper- iments. The replication will have one experimental situation with more categories of language intensity. This should allow the establishment of a threshold of language intensity above which attitude change would be expected to occur. ‘ As discussed by Burgccn (1969a), Miller (1969), and Tate (1970), the research on countfiettitudinal advocacy has often reported con- flicting results. R'I'his study suggests a possible reason for the con- flicting findings. me can speculate that the differences in message intensity across studies might have resulted in differential magnitudes of attitude change . For example , subjects in the low reward condition of Festinger and Carlsmith (1959) might have encoded more highly in- tense messages than did the low reward subjects in the study by Elms and Janis (1965). No objective measure has been used to evaluate messages produced in different research situations. Using the procedures em— ployed in this study, message intensity can be controlled to allow a more precise estimate of the variance in attitude change accounted for by variables of interest to counterattitudinal advocacy researchers . H5 This procedure could ensure that the assertions were comparable in in- tensity and prevent this message variable from confounding the effects of other mnanipulations , thus reducing the apparent theoretical conflict reported in the literature. At the present time, seven studies are in progress to investigate variables that have led to conflicting findings. Since this procedure worked successfully in these two experiments, it is being used with some modification in the seven studies. In order to prevent subjects from being sensitized by having to use the sameewords in the encoding task that anchored the attitude scale, the attitude scale anchors have been deleted as possible choices in the pre-constructed sentences. All of the seven studies have high and low intensity encoding con- ditions while manipulating justification, effort, commitment, audience attitude, choice, type of encoding, and publicness of the counteratti- tudinal encoding behavior. Such a compendium of studies could help re- solve sone of the present theoretical conflict. These studies ensure that the messages are at a.prescribed level of intensity in a variety of research situations. Thus, all subjects in these studies will encode a comparable message under different experimental conditions. Hopefully, this will alIOW'estimates of the variance in attitude change accounted for by a number of potentially relevant variables. The data seem to be generally supportive of dissonance theory pre- dicticns while providing evidence that questions the propositions ad- vanced by assimilaticnpcontrast theory. Sherif and Sherif (1967) pos- it a curvilinear'relaticnShip between the amount of change advocated and the magnitude of demonstrated attitude change. Sherif , Sherif , and 1+6 Nebergall (1965) argue that maximal attitude change occurs when the argument is near the person's boundary of latitude of acceptance. The findings of this study indicate that maximal attitude change occurred when the distance between prior attitude and the intensity of the coun- terattitudinal assertion is greatest . These findings question the ex- tent to which assimilation-contrast propositions about the standard per- suasion process apply to the counterattitudinal advocacy paradigm. Research Implications Why study the counterattitudinal encoding situation? First, in many everyday situations , people are induced to say and do things that are not in accord with their private beliefs. A manager is required to discipline an employee for violating a policy that the manager does not believe is fair, under White House pressm'e a Senator endorses a Southern jurist, a college professor who opposes grading publicly defends grading on the curve. The common element in all of these situations is that pe0ple are playing social roles that would not be predicted based only on knofledge of their private beliefs . It would seem logical that such situations should (1) result in differences from an individual's more normal encoding behavior, and (2) should affect a change in his private beliefs. The present study begins to specify some of the probable dif- ferences. It appears that an encoder' 3 prior attitudes affect how and what he writes . From the results of this study , one can conclude that to the extent that decreased language intensity produces a less desirable message, prior attitude should be considered when inducing people to encode specific messages. PbEwen and Greenberg (1969) reported lower I47 evaluations of the message source when receivers rated messages em- ploying low intensity language. Thus it seems that decreased language intensity could reduce the probability of a message having a desired impact. Future research should compare the extent to which professional encoders (speech writers , reporters , editorial writers) behave similarly to business writing students. A future study should investigate the relative persuasive impact of messages produced under counterattitudinal and belief-congruent encoding conditions . My: Two separate experiments were conducted to test the following hy- potheses: (1) Persons who encode a counterattitudinal message will use language of significantly lower intensity than persons who encode a belief-congruent message, and (2) Given counterattitudinal advocacy by people with similar prior attitudes , the magnitude of attitude change varies directly with the intensity of the counterattitudinal assertion. The same general procedures were used in both experiments. Lan- guage intensity scores were conputed by summing tl'e values (Thurstone and Jones, 1955) of words selected to complete partially pro—constructed sentences. Attitudes were measured by a seven-interval scale anchored by words assigned weights by Thurstone and Jones (1955). The findings support both theoretical hypotheses . Although both groups in Experiment I encoded moderately intense messages, the Belief- Congruent condition had a mean intensity score significantly greater than the Belief-Discrepant corndition. In Experiment II the results in- dicated a significant direct relationship between the intensity of a counterattitudinal assertion and subsequent attitude change. '48 Findings were discussed in terms of prior research in counteratti— tudinal advocacy. A number of research extensions, suggested by the findings of this study, were discussed. REFERENCES References Ashmore, R.D. and 8.1-3. Collins Studies in forced compliance: X. Attitude change and cormitment to maintain publicly a counter— attitudinal position. Psychological Reports, 1968, 33, 1229-1230. Bettinghaus, E.P. and I. Preston Dogmatism and performance of the communicator under stress. Journalism Quarterly, 19614, 31, 399-1402. Bodaken, ELM. Choice and perceived audience attitude as determinants of cognitive dissonance and subsequent attitude change following counteratt itudinal advocacy . Unpublished Dissertation , Michigan State University, 1970. Bowers, J .W. Language intensity, social introversion, and attitude change. Speech Monog'aphs, 1963, _32, 3145-352. Bowers, J .w. Some correlates of language intensity. Quarterly Journal of SEech, 196%, 62, 1115-1420. Burgcon, M. The analysis of counterattitudinal messages. Unpublished paper, Michigan State University, 1969a. Burgoon, M. 'I're effects of selected encoding situations on source attitudes. Unpublished paper, Michigan State University, 1969b. Carlsmith, J .M., 8.5. Collins and R.L. Helmreich Studies in forced compliance: I. The effect of pressure for compliance on atti- tude change produced by face-to-face role playing and anonymous essay writing. Journal of Personality and Social Psychology, 1966, 1+, 1-13. Cliff, N. Adverbs as modifiers. Psychological Review, 1959, Q, 27—00. Collins, B.E. Studies in forced compliance: III and VIII. The effect of true-persuasive instructions, public-private essays and finan— cial inducement on attitude change produced by forced conpliance. Unpublished mnimeographed paper, U.C.L.A. , 1968. DeQuincey, Thomas Collected Works. (David Masscn, ed.) Edinburgh, 1890, Q, p. 220. 149 50 Diedrich, G.W., S.J. Messick and L.R. Tucker A general least squares solution for successive intervals. P_sychometrika, 1957, 22, 159-173. Elms, A. and I. Janis Counter-norm attitudes induced by consonant versus dissonant conditions of role-playing. Journal of linger- imental Research in Personaliq, 1965, _1_, 50-60. Pestinger, L. A Theorg of Cognitive Dissonance. Evanston, 111.: Row, Peterson, 1957. Festinger, L. and J .M. Carlsmith Cognitive consequences of forced compliance. Journal of Abnormal and Social Psychology, 1959, 58, 203-210. Greenberg, 8.8. Performance and message consequences of encoding be- havior under cognitive stress . Unpublished Doctoral Dissertation , University of Wisconsin, 1960. Greenberg, 8.8. and P.H. Tannenbaum Comnunicator performance urnder cognitive stress. Journalism Merly, 1962, §_9_, 169-175. Gulliksen, H. and L.R. Tucker A general procedure for obtaining paired ' s from multiple rank orders. Psychometrika, 196l, _2_6_, 173-183. Hayakawa, 8.1. Language and Thought In Action. New York: Harcourt, Brace and Conpanyflglng. Helmreich, R.L. and 8.8. Collins Studies in forced compliance: Com- mitment and magnitude of inducement to comply as determinants of opinion change . Journal of Personality and Social Psyclnology, 1968, 19, 75-81. Hovland, C., I. Janis and H. Kelley Communication and Persuasion. New Haven: Yale University Press, 1953. Howe, E.S. Probabilistic adverbial qualification of adjectives. Journal of Verbal Learning and Verbal Behavior, 1962, 1, 225-201. Ibwe, B.S. Verb tense, negatives, and other determinants of tie inten- sity of evaluative meaning. Journal of Verbal Learning and Verbal Behavior, 1966, 5, 107-155. Janis, I. and B. Gilmore The influence of incentive conditions on the success of role playing in modifying attitudes. Journal of Person- ality and Social Psychology, 1965, 3_9_, 211-218. Janis, I. and B. King The influence of role playing on opinion change. Journal of Abnormal and Social Psychology, 1959, 39, 211-218. 51 King, B. and I. Janis Comparison of the effectiveness of improvised versus non-improvised role-playing in producing opinion change . I-Inmnan Relations, 1956, _9_, 177-186. Lazarus, R.S., J. Deese and S.F. Osler The effects of psychological stress upon performance. Psychological Bulletin, 1952, :49, 293-317. Lilly, R.S. Multiplying values of intensive, probabilistic, and frequency adverbs when comnbined with potency adjectives. Jomnal of Verbal Learning and Verbal Behavior, 1968a, 1, 850-858. Lilly, R.S. The qualification of evaluative adjectives by frequency adverbs. Journal of Verbal Learning and Verbal Behavior, 1968b, 7, 333-336. hinder, D.E., J. Cooper and E.E. Jones Decision freedom as a determinant of the role of incentive magnitude in attitude change. Journal of Personality and Social Psychology, 1967, 6, 2u5-25u. McBwen, W.J. Counter-attitudinal encoding effects on message style and performance . Unpublished Doctoral Dissertation , Michigan State University, 1969 . McBwen, W.J. and B.S. Greenberg Effects of comunioation assertion intensity. Journal of Communication, 1969, 1_9_, 257-265. McGuire, W.J. Inducing resistance to persuasion. In L. Berkowitz (Ed.), Advances in 3%. tal Social Pgfllfl, Vol. 1, New York: 88, , o - 0 Miller, G.R. Some problems in prior research on justification and self- persuasion. An unpublished paper, Michigan State University, 1969 . Miller, G.R. and E.M. Bodaken Justification and self-persuasion follow- ing commnitnent to encode, and actual encoding of counterattitudinal communication: A replication and extension. Paper presented to Speech Association of America, New York, December, 1969. Miller, G.R. and R.L. McGraw Justification and self-persuasion following commitment to encode and actual encoding of counterattitudinal com- munication. Speech Manogrgphs, 1969, 36, uu3-u51. fibreno, J.L.. Psyghodrnama, Vol. 1. New York: Beacon House, 19%. Ogden, C.K. and LA. Richards The Me ' of Pbanin , 10th Edition, New York: Harcourt, Brace , ., , p. 239. Osgood, C.E., S. Saporta and J.C. Nunnally Evaluative assertion analysis. Literna, 1956, 3, 07-102. 52 Osgood, C.E. and P.H. Tannenbaum The principle of congruity and the prediction of attitude change. Psychological Review, 1955, _6_2_, “2‘55. Osgood, C.E. and B.S. Walker Motivation and language behavior: A content analysis of suicide notes. Journal of Abnormal and Social Psychology, 1959, EB, 58-67. Rosenberg, M.J. When dissonance fails: On eliminating evaluation apprehension from attitude measurement . Journal of Personality and Social Psycholog, 1965, 1, 28-92. Sherif, M. and C.W. Sherif Attitude, Ego-Involvement, and Gang, New Sherif, M., C.W. Sherif and R.E. Nebergall Attitude and Attitude Change, Philadelphia: Saunders, 1965. Tate , E . Deception and role play in counterattitudinal advocacy research , Unpublished paper, Michigan State University, 1970. Thurstone, L.L. and L.V. Jones The psychophysics of semantics: An experimental investigation. Journal of Applied Psycholo , 1955, 39, 31-390 Zimbardo, P.G. The effect of effort and improvisation on self-persuasion produced by role-playing . Journal of Engerimental Social Psychology , 1965, _1_, 103-120. APPENDICES APPENDIX A PREI'EST ATTITUDE QUESTIONNAIRE 53 MICHIGAN STATE UNIVERSITY STUDENT OPINION'PROFILE NAME STUDENT NUMBER COURSE SECI'ION INSTRUCTOR Communication 101 at Michigan State University is attemtping to solicit the opinions of students in many colleges on a variety of current is- sues--campus and nationa1--which may serve as the foci of a problem analysis. On each of the following pages you will find a number of is- sues followed by a series of scales. PLEASE READ THE INSTRUCTIONS ON CDMPLEI'ING THESE SCALES VERY CAREFULLY. For enample, if you think tlat it would be an MELIENT‘ idea for all universities to establish Black Studies programs , you _‘Tzould mark the following scale by placing an "A" above the word Excellent. "A11 universities and colleges should establish Black Studies programs" "“5? gm P009 are; amp-181 taxman; meniscus > If you thought it was only a GOOD idea you would place an "A" above the wordGoodandsoon foreach wo'r'achoice. Wewantyoutomarktheword that BEST represents your feelings toward the statement. Now go back to the scale above and consider ALL of the word choices. Please mark every word tlat you can agree with by placing an "X" above that word. For enample, if you felt the Black Studies statement was EXCELLENT, you might also agree tIat it was GOOD and PAIR. If this were Hicaseyouwouldmarkthe soaleas follows?— A:X:X: 83 are; as gs {when GIQ'W-Ial. Manama 5'4 Again, please place an "A" above the word that BEST indicates your feeling about the statement. Then place an "XV-5.15% every other word choice you can agree with: The use and possession of marijuana should be legalized. HEN” nuattaaxa atqnddan. Draft deferments should be eliminated for college students . "U2"! Eggnog aIQ‘FJ-Ial. nuattaaxa The required University College "Basic Courses" should be eliminated. H n atqnaaan taxman «119.: luaIIaa‘H All students should be required to live on campus during their college attendance . nuatnaoxa P009 are; ImlnaN .. aood P98 annual, Psychological testing should be used to determine a student '5 academic major. P009 are; Immam . aood P98 1uanaoxg BMW-131. 55 Now, here is annotlner set of scales on the SAME issues. THE INSTRUCTIONS gimme mass scams ARE DIFFERENT SOPLEASE READ _ng FOLLOWING VERY CAREFULLY: You will find several statements followed by several scales . Please mark each scale in the blank that BEST represents how you feel. For example, here isanitemlikethoseyouwill see: The United States should withdraw from the United Nations . Bad : : : : : : Good Your job is to place a check-mark (X) above the line that best indicates your judgnent about the proposition . For example, if you feel that 0.8. withdrawal would be very bad, you would check as follovs: Bad X : : : : : : Good If you feel that such a move (withdrawal) would be quite bad, yon should check as follows: Bad :X: : : : : Good If yon feel neutral or indifferent about the proposition, or if you feel that the scale is irrelevant to the proposition, you would check as follovs: Bad: : :X: : : Good Remember: Never put more than one check-mark on a single scale and be sure that each check is in the middle of the line , not on the boundaries. WORK RAPIDLY --- RECORD FIRST IMPRESSIONS --- DO NOT CHANGE MARKS PLACE 931; "x" or EACH sous. 56 The use and possession of marijuana should be legalized. Good Bad Worthless Valuable Pleasant Unpleasant Unfair Fair Draft deferments should be eliminated for college students. Good Bad Unfair Worthless Pleasant Pair Valuable Unpleasant All universities and colleges should establish Black Studies programs . Unpleasant Pleasant Bad Good Fair Valuable Unfair Worthless All students should be required to live on campus during their college attendance . Valuable : Good Worthless Bad Unfair Pleasant Fair Unpleasant 57 Psychological testing shonld be used to determine a student' 5 academic major. Good : : : : . . Bad Worthless : : : ' : : Valuable Unpleasant : : : : : : Pleasant APPENDIX B WIMKI‘ERIAIS 58 MICHIGAN STATE UNIVERSITY um LANSING - HICHIGAN 48323 Dear Student: For the past several months , the Center for Opinion Research at Michigan State University has been conducting research ainned at find- ing out what it takes to strengthen or change people's attitudes. What we do is get sore indication of the attitudes people have on cer- tain issues and then, at some later date, we present certain arguments in the form of speeches, essays, debates - various methods -- which are designed to get them to change or strengthen their attitudes. Let me explain briefly what we are attempting to do in the research that you will be helping us with. Sonetime ago we administered question- naires to students enrolled in the Lansing-East Lansing Area High Schools asking them their attitudes toward the policy of required on-campus liv- ing for college students. Trat is, we were interested in their Opinions on the issue of whether or not college students should be required to live on campus during their attendance at college. Analysis of the questionnaires indicate that these students are generally UNECIDED on the issue . We then asked over 500 Michign State University students to write per- suasive messages to reject a policy that would require students to live on campus while attending the University. You will soon be given the arguments that students like you created. It is even possible that some of you helped in this original research effort. We founnd that the arguments produced were similar in content. HOWEVER, (UR ANALYSIS OF THE PERSUASIVE EFFECT OF THESE MESSAGES INDICATES THAT CERTAIN WORDS A_RE__T:£_;_MOST IMPORTANT IN (HANGING THESE STUIINT'SAT'I'IW Thus, in preparing your persuasive messages it is very important for you to choose what you feel are the most persuasive words. It is important thatyouselectthewordsyouthinkwillbe successful forwe feel that all of our effort to date depends on yonr choice of the persuasive words. Thinkabontthe choicesandtrytodecidewhichmrdineachargnment mignt be best used to crange attitudes. Also the order of the arguments is important and we want you to think abort what order of presentation might be most effective. This part of the research is very important to the successful completion oftheproject. Wewannttothankall ofyouforhelpinguscreate mes- sages that we will show to high school students in an attempt to change their attitudes toward on-campus living. Remember we want you to per- stade the students to reject mandatory on-canqnus living. Thannksagain, Gerald R. Miller, Ph.D. Professor and Research Director 59 PERSUASIVE WORIB PLEASE USE WHAT YOU FEEL ARE THE MOST APPROPRIATE WORDS TO FILL IN THE BIANKS IN THIS MESSAGE INTENDED TO OPPOSE THE POLICY 01" MANDA- TORY (IV-CAMPUS LIVING: TERRIBLE DISLIKE DISLIKE SLIGHTLY POOR STROVGLY DISLIKE DOV'T CARE FOR IT NOT PLEASING DESPISE LIKE NO]? 80 WCH DISLIKE EXTREMELY DISLIKE VERY PUG-I VERY BAD BAD LIKE NOT SO WELL DISLIKE MODERATELX MILDLX DISLIKE HIGHLX UNPAVORABLE DOV'T LIKE DISLIIGI INTENSELY LDATH 60 MICHIGAN STATE UNIVERSITY usn' wsmc - MICHIGAN 4882} Dear Student: For the past several months, the Center for Opinion Research at Michigan State University has been conducting research aimed at finding out what it takes to strengthen or change people's attitudes. Wrat we do is get sore inndication of the attitudes people have on certain issues and then , at some later date, we present certain argnmnents in the form of speeches , essays, debates -- various methods -- which are designed to get them to orange or strenngthen their attitudes . Let me explain briefly what we are attempting to do in the research that you will be helping us with. Soretime ago we adnministered questionnaires to students enrolled in the Lansing-East Lansing Area High Schools asking tlnem their attitudes toard the policy of required on-campus living for college students . That is , we were interested in their Opinions on the issue of whether or not college students should be required to live on campus during their attendance at college. Analysis of the question- naires indicate that these students are generally UNIIZCIDED on the 1ssue . We then asked over 500 Michigan State University students to write per- suasive messages to support a policy that would require students to live on campus while attending the University. Yon will soon be given the arguments trat students like you created. It is even possible that some of you helped in this original research effort. We found that the argu- ments produced were similar in content. HOWEVER, OUR ANALYSIS OF THE PERSUASIVE EFFECT OF THESE MESSAGES INDICATES THAT CERTAIN mRDS ARE THE MOST IMPORTANT IN CHANGING THESE STUENT'S AW"— Thus, in preparing yonr persuasive messages it is very important for you to choose what yon feel are the most persuasive words. It is impor- tant that you select the words you think will be successful for we feel that all of onr effort to date depends on your choice of the persuasive words. Thinkaboutthechoices anndtrytodecidewhichwordineach argument might be best used to change attitudes. Also the order of the arguments is important and we want you to think about what order of presentation might be most effective . This part of the research is very important to the successful conpletion of the project. We want to tharnk all of you for helping us create mes- sages that we will show to high school students in an attempt to change their attitudes toward on-campus living. Remember we want you to per- suade the students to support mandatory on-campus living. Thannks again, Gerald R. Miller, Ph.D. Professor and Research Director 61 PERSUASIVE WORDS PLEASE USE WHAT YOU FEEL ARE THE POST APPROPRIATE WORDS TO FILL IN THE BLANIG IN THIS MESSAGE INTENDED TO SUPPORT THE POLICY OF MANDA— TORY Uni-CAMPUS LIVING: GOOD OK EXCELLENT PLEASING MILDLY LIKE LIKE EXTREMELY BEST OF ALL STRONGLX LIKE WOVDEREUL LIKE INTENSELY ESPECIALLY GOOD TASTY PREFERRED LIKE FAIRLY WELL LIKE MODERATELX LIKE QUITE A BIT LIKE LIKE VEW WELL WELCO’IE MIGHT FINE AVERAGE LIKE SLIG-ITLX FAIR ACCEPTABLE ENJOY ONLX PAIR MIGHTY PAVORABLE FAVORITE LIKE VERY MUCH VERY GJOD 62 I the idea of mandatory on—campus living during college attendance. Forcing students to live on campus can be a (an) experierce for them in many cases. It can be a social experience for a person to take housing in which organized social functions are a part of the living program. Snch organized social functions are usually experi- ences. The living facilities offered on-cannpus are usually . I think any high school student would be demonstrating judgnent by living on campus. Mandatory on-campus living would be a idea for the University as a whole and most peOple such an idea. People who do live on-campus during their college attendance usually the activities and furctions offered. All in all, it is a (an) experience to live on campus while attending the University. APPENDIX C EDG’ERIMENTIIMATERIALS 63 MICHIGAN STATE UNIVERSITY usn' LANSING - MICHIGAN 48823 Dear Student: For the past several months, the Oenter for Opinion Research at Michigan State University has been conducting research aimed at find- ing out what it takes to strengthen or change people' 3 attitudes. mat we do is get some indication of the attitudes people rave on certain is- sues annd then, at some later date, we present certain arguments in the form of speeches, essays, debates -- various methods -- which are de- signed to get them to change or strengthen their attitudes. Let me explain briefly what we are attempting to do in the research that you will be helping us with. Sometime ago we administered questionnaires to students ennrolled in the Lancing-East Lansing Area High Schools ask- ing them their attitudes todard the policy of required on-campus living for college students. That is, we were interested in their opinions on the issue of whether or not college students stculd be required to live on campus during their attendance at college. Analysis of the question- naires indicate that these students are generally UNIECIED on the issue. We then asked over 500 Michigan State University students to write per- suasive messages to support a policy that would require students to live on campus while attending the University. Yon will soon be given the arguments that students like you created. It is even possible that sore of you helped in this original research effort. We found ttat the arguments produced were similar in content. HOWEVER, (11R ANALYSIS OF THE PERSJASIVE EFFECT OF THESE MESSAGES INDICATES THAT CERTAIN WORDS ARE mamsr IMPORTANT‘INCHANGINGTHESE SIUENT'SW__ TIM, inpreparingyourpersuasive messages it is very important foryon todncosevanatyonfeelarethemostpersuasivemrds. Itisimportant thatyonselectthewordsyouthinkwillbesuccessful forwe feel tnat all of onr effort to date depends on your choice of the persuasive words. Wabonttrechoicesanndtrytodecidewhichwordineachargnment migntbebest usedtochanngeattitudes. Alsotheorderofthearguments isinnportantandmvantyoutothinRabonthatorderofpresentation mightbemsteffective. Thispartoftheresearchisveryimportanttothesnccessfulconple- tion of the project. We want to thannk all of you for helping us create messages that we will show to high school students in an attempt to change their attitnndes toward on-campus living. Remember we want you to persuade the students to support mandatory on-campus living. Thanksagin, Gerald R. Miller, Ph.D. Professor annd Research Director 1. 61+ INSTRUCTIONS Yon are being given a set of arguments finat studennts on this campus have created favoring the policy of mandatory on-campus living. Each argumennt has one or more sets of words that tend to be critical determinants of attitude channge. Please go throughALLof the arguments andselect finewords THATYOU FEEL wwm BE MOST PERSUASIVE. From the word choices , select fine word yon feel is most appropriate and WRI'E it in the blank space. Think abont yonr choices annd how high sdnool students mignt react. We want yon to create fine most persuasive message yon possibly can. Non place fine arguments in fine ORDER OF PRESENTATIOJ THAT YOU THINK WOULD BE MOST PERSUASIVE. When yon have made all fine word selections and ordered the arguments, read fine entire message that yon have created. Whon YOJ ARE SATISFIED WIm YWR WORK, PLEASE MSB YwR HMDOOOOOOOOOOOOOOOOOOOO 65 PRE-CONSTRUCTED SENTENCES High Intensity Condition Ithinkanyhighschcolstudentwouldbemakinga mistake if he did not expose himself to the many ideas, attitudes, annd values onne finds in university housing. PERSUASIVE mRD CHOICES: VERY BAD TERRIBLE It would be for students to miss the experience of living with many different lands of people. PERSJASIVE WORD CHOICES: VERY BAD TERRIBLE All in all, it is a(an) experience to live on campus while attending the unIversity . PERSUASIVE mRD CHOICES: EXCELLENT VDNDERE'UL I the idea of mandatory on-camnpus living during coIIege attendance . PERSUASIVE map CHOICES: LIKE EXTREMELY LIKE INTENSELY People who live on campus during their college attendance usually the opportunity to meet new people arnd do different things . PERSUASIVE mRD CHOICES: LIKE VERY MUCH STRONGLY LIKE 66 To be in an environnment where a person will be subject to new ideas constantly is . PERSUASIVE WORD CHOICES: EXCELLENT WONDERFUL It can also be a(an) social experience for a person to take housing in which he has opportunities to engage in organized functions. PERSUASIVE mRD CHOICES: mNDERFUL EXCELLENT Also, on-campus living offers (B1) progranms such as intramurals that can be (B2) learning expe- riences for students. PERSUASIVE WORD CHOICES: (an) mam FAVORABLE, MIGHTY FINE (32) ocmm, WONDERFUL It is generally agreed trat on—campus housing offers recreation facilities that are . PERSJASIVE WORD CHOICES: EXCELLENT ESPECIALLY GOOD I personally the concept of living on campus annd becoming involved with the many events a university has to offer. PERSUASIVE WORD CHOICES: LIKE INTENSELY LIKE EXTREMELY I on-campus living because it makes the person a more well-rounnded individual . PERSUASIVE mRD CHOICES: STRONGLY LIKE LIKE VERY NEH 67 PRE-CONS'TRUCTED SENTENCES Moderate Intensity Condition I think any high school student would be making a mistake if he did not expose himself to the many idéas, attfiudes, and values one finds in university housing. PERSUASIVE WORD CHOICES: BAD HIGHLY UNFAVORABLE It would be for students to miss the experience of living with many different kinds of people. PERSIASIVE mm) CHOICES: BAD HIGHLY UNFAVORABLE All in all, it is a(an) experience to live on campus while attennding the unniversity . PERSJASIVE WORD CHOICES: GmD PLEASING I the idea of mandatory on-campus living during college attendance. PERSUASIVE WORD CHOICES: LIKE LIKE FAIRLY WELL People who live on campus during their college attendance usually the opportunity to meet new people and do different things . PERSUASIVE WORD CHOICES: ENJOY WELCOME 68 Tobeinanenvironmentwhereapersonwillbesubjecttonewideas constantly is . PERSUASIVE WRD CHOICES: GOOD WELCOME It can also be a(an) social experience for a person to take housing in finich he has opportunities to engage in organized functions . PERSUASIVE WORD CHOICES: GmD VERY GOOD Also, on-campnc living offers (Bl) programs such as intramurals that can be (B2) learning expe- riences for students. PERSUASIVE WORD CHOICES: (Bl) GOOD, WELCOME (82) VERY GOOD, PREFERRED It is gennerally agreed fiat on-campus housing offers recreation facilities that are . PERSUASIVE WORD CHOICES: GOOD PREFERRED I personally the cocept of living on campus annd becoming ifivoIved with the many events a university has to offer. PERSUASIVE mRD CHOICES: LIKE FAIRLY WELL LIKE QUITE A BIT I on-campus living because it makes the person a more well-rounded individual. PERSUASIVE WORD QiOICES: LIKE QUITE A BIT ENJOY 69 PRE-CONSTRUCTED SENTENCES Low Intensity Condition I think any high school student would be making a mistake if he did not expose himself to the many ideas, attitudes, and values one finds in university housing. PERSUASIVE WORD CHOICES: NOT PLEASING POOR It would be for students to miss the experience of living with many difgrent Rinds of people . PERSUASIVE WORD CHOICES: POOR NOT PLEASING All in all, it is a(an) experience to live on campus while attending the university. PERSUASIVE WORD CHOICES: OK ACCEPTABLE I the idea of mandatory on-campus living during college attendance . PERSUASIVE WRD CHOICES: LIKE MODERATELY LIKE SLIGHTLY People who live on campus during their college attendance usually the opportunity to meet new people and do aifferent things . PERSUASIVE mRD CHOICES: MILDLY LIKE LIKE SLIGHTLY 70 To be in an environment where a person will be subject to new ideas constantly is . PERSUASIVE WORD CHOICES: OK HAIR It can also be a(an) social experience for a person to take housing in whichfihe has opportunities to engage in organized functions . PERSUASIVE WORD CHOICES: FAIR ONLY FAIR Also, on-campus living offers (Bl) programs such as intramurals that can be (B2) learning experiences for students . PERSUASIVE WORD CHOICES: (Bl) ACCEPTABLE, AVERAGE (32) PAIR, ONLY FAIR It is gennerally agreed that on-campus housing offers recreation facilities that are . PERSUASIVE WORD CHOICES: ACCEPTABLE AVERAGE I personally the concept of living on campus and becomning involved with the many events a university has to offer. PERSUASIVE WORD CHOICES: MILDLY LIKE LIKE SLIGHTLY I on-campus living because it makes the person a more well-rounded ihdividual. PERSUASIVE WORD CHOICES: LIKE SLIGHTLY LIKE MODERATELY 71 NAME Now please refer to the cards yon have. We want yon to write the total message AS 392 HAVE CREATED IT. Remember this is yonr message and we will want to present 1t to fie students who are UNDECIDED on fine issue of mandatory on-campus living. We hope onr efforts to change fineir attitudes to SUPPORT such a policy will be successful with yonr help. Please be sure yonr name is on finis sheet so that we give yon credit for yonr ideas. APPENDIX D POSTTEST ATTITUIE QUESTIO‘INAIRE IMPORTANCE RATINGS AUDIENCE ATTITUE RATINGS TASK DIFFICULTY RATINGS 72 NAME STUDENT NUMBER SECTION NUMBER Now, we wonld like to get yonr opinion on fine issue of whefiner or not students shonld be expected to live on campus during their attenndannce at fine University . Please read finese instructions very carefully as this is the same type of questionnaire that we gave to the high school students and we want to be able to score it in the same way. Please consider fine folloIing enample: All colleges shonld establish BLACK STUDIES PROGRAMS. ____This is an excellent idea. This is a gofl idea. This is a fair— idea. This is a neutral idea. This is a pier: idea. This is a bad idea. This is a terrible idea. Insouctions : 1. PLACE A LARGE ”A" in the BLANK YOU MOST AGREE WITH. 2. NON GO BACK AND PLACE AN "X" IN EACH BLANK THAT YOJ ALSO AGREE WITH. Fon instance yon mignt finink the establishment of a Black Studies Program in Colleges is an EXCELLENT idea but yon mnight also agree finat it is GOOD and EUR. 73 Again, please place an "A" by the blank that BEST indicates yonr feeling abont on-campus living. Then place an "X" by the blanks that you also agree with. ALL STUDENTS SHOULD BE REQUIRED TO LIVE (N CAMPUS HIRING THEIR COLLEGE ATTENDANCE This is an excellent idea. This is a good idea. w This is a fair idea. This is a neutral idea. This is a peg; idea. This is a b_a_c_1_ idea. This is a terrible idea. NCM PUT AN "X" ON EACH SCALE NEXT TO THE BLANK THAT BEST INDICATES YOUR FEELINGS: ALLSTUDENTS SHOULD BEREQUIREDTOLIVEG‘ICAI‘IPUS HIRING THEIR COLLEGE ATTENDANCE ____Very Pleasant ___Quite Pleasant _____Slightly Pleasant _______Neutral _______Slightly Unpleasant _Quite Unpleasant Very Unpleasant 7n ALL STUDENTS SHOULD BE REQUIRED TO LIVE ON CAMPUS HIRING THEIR COLLEGE ATTENDANCE ___Very Good ____Quite Good ______Slightly Good _____Neutral ____Slightly Bad ____Quite Bad ___Very Bad ALL STUDENTS SHOULD BE REQUIRED TO LIVE ON CAMPUS DURING THEIR COLLEGE ATTENDANCE ______Very Valuable _____Quite Valuable ______Slightly Valuable _Neutral ___Slightly Worthless ____Quite Worthless _______Very Worthless ALL STUDENTS SHOULD BE REQUIRED TO LIVE ON CAMPUS DURING THEIR COLLEGE ATTENDANCE ____Very Unfair _____Quite Unfair ____Slightly Unfair _____Neutral ______Slightly Fair _____Quite Fair Very Fair 75 Did you feel fiat your contribution to the persuasion campaign was important? Important : : : : : : Unimportant Before you wrote your essay, how did you think the high school students felt toward the issue of conpulsory on-campus residency? They were favorable toward conpulsory on-campus residency. ey were W toward compulsory on-campus residency. y were BREE—:g to compulsory on-campus residency. I don't know w t their attitude was toward conpulsory on-campus residency. How difficult was it for you to write the essay on this issue? prime é ' Néither. ' Easy