MEDIAYED AAWYUDES AS INDICESx OI! NON~C0=NGRUITY FACTORS IN ATIITUDE CHANGE mesh: for Hm 009m of M. A. MICWGAN STATE UNNER’SPTY fivan L Presi‘on WM ABSTRACT MEDIATED.ATTITUDES.AS INDICES OF‘NON-OONGRUITY FACTORS IN ATTITUDE CHANGE by Ivan 1. Preston The purpose of this research was to investigate the effects of non- congruity factors in attitude change. The research centered around the use of the congruity principle to predict changes occurring in attitudes when the attitude holder acquires new relevant information. The congrup ity principle's predictions are based on several change factors which are called congruity factors. .All other relevant change factors thus may be called non-congruity factors. The research was concerned with reported attitude changes which differed from those predicted by the congruity principle. The hypothesis was made that non-congruity factors were determinants of change in these cases. The specific problem, then, was to test this hypothesis by ob- taining an empirical demonstration of the effects of non-congruity fac- tors. A new attitude change predictor, called the mediated attitude prin- ciple, was develOped for this purpose. This principle makes predictions based on all relevant change factors, including congruity and non-con- gruity factors. It was assumed that such predictions would be more ac- curate than the congruity principle's predictions whenever non-congruity factors were determinants of change. Ivan L Preston To distinguish between situations in which non-congruity factors would and would not determine change, the research made use of a dis- tinction betweeen immediate and mediate attitude objects. These are ob- jects of whom a person's knowledge is principally the result of his own direct personal contact with them (immediate), or someone else's personal contact with them (mediate). Results of previous research suggested that congruitypredictions were accurate for changes involving mediate objects, but inaccurate for changes involving immediate objects. Accord- ingly, it was assumed that non-congruity factors were significant determ- inants of changes in attitude toward immediate objects, but were not so when mediate objects were concerned. The hypothesis of the research was: For situations involving im- mediate attitude objects, predictions of change based on the mediated at- titude principle will be significantly more accurate than will predic- tions based on the congruity principle. A questionnaire was written and administered to 126 subjects. The first part obtained measurements of subjects' attitudes toward imediate attitude objects. The second part, issued several days later, informed subjects that one attitude object had made a positive or negative asser- tion about another attitude object. The questionnaire then re-measured subjects' attitudes toward the objects. It also obtained subjects' med- iated attitudes toward the objects, and subj ects' indications of the ”real-life” situation. Measurements of subjects' openmindedness and closedmindedness were obtained with another measuring instrument. Congruity predictions, mediated. attitude predictions, and actual changes were computed for 50h change situations (four per subject). Ivan L. Preston Four statistical tests (two 3 tests, chi square test, and sign test) were used to determine differences in accuracy betwaen the two predic- tors. Chi square tests were used to determine whether relative accuracy of the predictors was independent of nine experimental variables. ‘ The principal conclusion was that mediated attitude predictions were significantly more accurate than congruity predictions (all tests significant). Accordingly, it was concluded that non-congruity factors were important determinants of changes in attitudes held toward inmediate attitude objects. Relative superiority of mediated attitude predictions occurred un- der all experimental conditions, but extent of superiority was not inde- pendent of the value of the attitude object (positive, negative, or neu- tral), or the type of change situation (congruent, incongruent, or other). Subjects treated incongruity as a relevant change factor when the experimental situation seemed conceivable in real life. But they ap- parently discounted, incongruity when the situation seemed inconceivable. No differences between openminded and closedminded subjects were found. MEDIATFD ATTITUDES AS INDICES 0F NON-CONGRUITY FACTORS IN ATTITUDE CHANGE by Ivan I. Preston A THESIS Submitted to the College of Communication Arts Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of General Commmication Arts 1961 W%% TABLE OF CONTENTS ACICNWGPENTS O O O O O O O O O O O O O 0 O O O O O O. O O O 0 iii HST OF TABIIES O O O O O O O O O O O O O O O O O O O O O O O O 0 iv HST OF MSTMTIONS O O O O O O O O O O O O O O O O O O O O 0 Vi INTRODUCTIC'Noooeooeeo00000000000000...1 CWTEIOOOOOOOOOOOOOOOOO0......OOOOh CHAPTER II 0 O O O O O O O O O O O O O O O O O O O O O O O O O O 32 CHAPTERIII...oo000.00.00.00'0000000000’42 CHAPTER”. 0 O O O O O O O O O O O O O O O O O O O O O O O O .61 BIBHOGWIH O O O O O O O O O O O O O C O O O O O O O O O O O 0 7h mmDIx O O O O O O O O O O C O O O O O O O O O O O O O O O O O 76 ii ACKNWLEDM TS I wish to express my appreciation to my advisor, Dr. David K. Berlo, who guided this program of research and. made my suggestions which led to substantial improvements. Appreciation also is expressed to members of my committee, Dr. Hideya Kumata and Dr. Erwin P. Bettinghaus, who made helpful suggestions and criticisms. In addition, I would like to acknowledge the help given by other faculty members ofthe College of Communication Arts, particularly Dr. Malcolm MacLean Jr. and Dr. William Stellwagen, along with many of the graduate students with whom I have been associated. Special thanks are due to my wife, family, and friends for their advice and encouragement. LIST OF TABLES Tables Page 1. Predicted and Obtained Mean Attitude Changes Withincellseeoeeeeeeeeoeeeeeeoee1&3 2. Expected and Observed Numbers of Sets of Four Responses Having Each of Five Outcomes as Pre- dicted for Independent Responses . . . . . . . . . . . 1:14 3. Comparison Among Predicted Changes and Obtained ChangeSeeoeeoooeoeeeeeeoeeeeoeohé 14. Relative Accuracy of Predictors for Values of theMediatedAttitUdeeeeeeeeeeoeooee00,49 5. Accuracy of Congruity Predictor for Values of theMediatedAttitude.................h9 6. Relative Accuracy of Predictors for Types of AttitudeObject....................50 7. Relative Accuracy of Predictors for Types of Assertion O O O O O O C O O O O O O O O O O O O O O O C so 8. _ Relative Accuracy of Predictors for Values of Source m for Values 0f Concept 9 o e e o e e o e o o 51 9. Relative Accuracy of Predictors for Values of AttitudeObject......o.............52 10. Directional Accuracy of Congruity Predictor £01." Values Of Attitllde OijCt e e e e e e o e e o o e e 53 ll. Directional Accuracy of Mediated Attitude Predictor for Values of Attitude Object . . . . . . . . 53 12. Relative Accuracy of Predictors for Values of Attitude Object When Mediated Attitude is Irrelevant......................5h 13. Relative Accuracy of Predictors for Values of Attitude Object When Mediated Attitude Has Value otherthanlrrelevan‘b......o..........51]. iv Tables Page 1h. Relative Accuracy of Predictors for Types of Sit‘lations . . O . . . C . 0 O O C O O C C C O O 0 O O S6 15. Directional Accuracy of Congruity Predictor for TYpes Of SituatiOHB e e o e e o e e e o e o o e o o 56 16. Directional Accuracy of Mediated Attitude Pre- dictor for Types of Situations . . . . . . . . . . . . 56 17. Relative Accuracy of Predictors for Types of Real-life Situations e e e e e e e o e e e e e e e e 0 S7 18. Relative Accuracy of Predictors for Types of Situation for Conceivable--same Situations, for Conceivable--not same Situations, and for In- conceivable Situations e e e e e e e e e e e e e o e e 58 19. Relative Accuracy of Predictors for Values of Attitude Object for Conceivable--same Situations, for Conceivable--not same Situations, and for Inconceivable Situations e e o o e o e e e e s e o e o 59 20. Relative Accuracy of Predictors for Open and Closed SUbjCCtS e o e o e o e e e e o e e o e e o o e e 60 Figure 2. 3. LIST OF ILIUSTRATIONS Page Predicted changes in attitude as corrected .forincredulity....................13 'Predicted changes in attitude for mediated attitudeprinciple..................25 Assignment of eighteen combinations of congruity conditions to nine groups of subjects . . . . . . . . . 33 IKTRODUCTION The purpose of this research is to investigate the effects of non- congruity factors in attitude change research. Congruity factors are those which are used by Osgood and Tannenbaum for predicting attitude change. Their method of deriving predictions from these factors is called the congruity principle. Osgood and Tannenbaun believe that it. accounts for the most significant factors bearing upon attitude change. They do not state that there are no other factors relevant to attitude change, but their research suggests that no other factors have signif- icant effects upon change. Findings of other research imply that non-congruity factors do af- fect change significantly under certain conditions. Attitude changes will occur which are different from those predicted by the congruity principle. If congruity factors did not determine these attitudes, we may sumise that non-congruity factors were significant determinants. The existmce of non-congruity factors, however, is not demonstrated nec- essarily by a negative relationship between change and congruity predic- tions. What is required is some direct evidence that non-congruity fac- tors are related to change. Another requirement is some means of dis- tinguishing betwem situations in which non-congruity factors are and are not important determinants of change. Il'he present research attempts to meet these requirements. To demon- strate that non-congruity factors are associated with attitude change, a new change predictor is presented. This predictor, called the mediated attitude principle, is designed to make predictions on the basis of all relevant change factors. Its predictions will be compared with those of the congruity principle. The congruity principle' 8 predictions are made on the basis of congruity factors only. Therefore, if the mediated atti- tude principle's predictions are different, we assume that they are made on the basis of non-congruity as well as congruity factors. Furthermore, if the mediated attitude predictions are more accurate, we will conclude that the non-congruity factors are importantly related to attitude change. To distinguish between situations in which non-congruity factors are and are not relevant to change, we will make a distinction betwem two kinds of attitude objects, immediate and mediate. Immediate attitude objects are those of whom a person's knowledge is principally the result of his own direct personal contact with them. Mediate attitude objects are those of whom a person's knowledge is principally the result of some- one else's personal contact. It has already been demonstrated in other research that changes in attitudes toward mediate attitude objects are highly correlated with congruity factors and not significantly asso- ciated with non-congruity factors. In contrast, we will attempt to show here that non-congruity factors are significant determinants in changing attitudes toward immediate attitude objects. The significance of this research lies in the opportunity to in- crease knowledge of the behavior of the receiver in a communication sit- uatiOn. The receiver's attitude toward something constitutes part of his predisposition for behaving in certain ways toward it. Given additional information about an object for which he already has an attitude, the receiver may or may not change the attitude as a result. The congruity principle's value lies in the Opportunity it provides to determine in ad- vance what changes the receiver will make as a result of receiving cer- tain messages. Thus it is possible to increase the conmnmicator's ef- ficiency by enabling him to know when the congruity principle will pro- vide accurate advance information and whm it will not. In situations for which the congruity principle will not be useful, the communicator may obtain useful advance infomation by means of the mediated attitude principle. The study is to be reported in four chapters. Chapter I includes a background for the research, and a rationale for the experimmtal hy- pothesis. Chapter II describes the experiment, including descriptions of the subjects, measures, design, and administration. Chapter III re- ports the results of the experimmt, the statistical analysis used to test the hypothesis, and some additional analyses of the experimental data. Chapter IV includes the conclusions, discussion of the conclu- sions, and implications of the thesis for further research and for pre- dictions of attitude change. CHAPTER I This chapter reports on research on consistency and attitude change as background for the present study, and presents a rationale for the experimental hypothesis. THE CONCEPT OF CONSISTENC! IN ATTITUDE RESEARCH In typical studies of attitude change, there is first an individual who holds a known attitude toward a lmown attitude object. Next, there is the occurrence of some event which the individual perceives, and which is presumably relevant to the attitude. Finally, there is the individ- ual's post-event attitude. It is the task of theory to characterize the perceived event so as to predict post-event attitude. Much of the recent attitude change research has involved. the phenom- ena of consistency and inconsistency. These general concepts have been incorporated into various theoretical models of human behavior. Osgood and Tannenbaum call the phenomena "congruity" and "incongruity. '1 Heider refers to “balance" and "imbalance."2 Festinger uses the terms "conson- ance' and "dissonance."3 Newcomb's I'symme‘lu'y" and "co-orientation" are lCharles E. Osgood and Percy H. Tannenbaum, "The Principle of Con- ty in the Prediction of Attitude Change," Psychological Review, LIII January, 1955), h2-55- 2 Fritz Heider, The Psycholog 9_f_ Interpersonal Relations (New York: 3 Leon Festinger, A Theo 23 Cofltive Dissonance (Evanston: Row, Peterson and Company, I937}. 5 related concepts.h Hovland. and his associates at Yale have instituted a program of research on attitude change in which the basic thesis is that the separate components of an attitude tend toward consistent organiza- tion with each other.5 Osgood and Tamenbaum, Raider, and Newcomb orient their formulations around a similar three-element attitude change model. For Heider, the elements involved include either three persons or two persons and one. inanimate attitude object. Additional elements include the attitudes of one person toward the other two attitude objects, plus the attitude of one of the latter toward the other. Newcomb's A-B—X model involves two persons, A and B, and one addi- tional attitude object, I. Also involved are the attitudes of A and B toward each other, and the attitudes of both toward 1. A model involving similar elements, though they are labeled differ- ently, is that of Osgood and Tannenbaum. They are interested in predict- ing a subject's behavior toward a communication source and a concept. Also involved are the attitudes toward the concept held both by the sub- ject and the communication source, as well as the subject's attitude toward the source. The concept may be any kind of attitude object, but the source has been restricted to persons or human organizations. The consistency-inconsistency characterization of the intervening event in attitude change situations has been useful in predicting l"Theodore Newcomb, "An Approach to the Study of Communicative Acts,” Psychological Review, 1.x (November, 1953), 393-1101.. 5 Milton J. Rosenberg, Carl I. Hovland, William J. McGuire, Robert P. Abelson, and Jack W. Brehm, Attitude Organization _a_n_d_ Change (New Ham: Yale University Press, 1935,. post-event attitudes. Among the various systems, Osgood and Tannenbaum's congruity principle permits the most precise predictions. Given the subject's prior attitudes toward both source and concept, and the source's own attitude toward the concept (as indexed by the source's statements about the concept), the congruity principle predicts the dir- ection and amount of change of the subj ect's attitudes toward both source and concept. I M Principle-n-Theoretic Riga-The congruity principle has been developed from a theoretical background in a theory of signal.6 Any theory of signs must specify conditions for the use of the term “sign," i.e., a meaning for ”meaning" or "significance." Osgood defines meaning as the representational mediation process which the sign evokes within the organism. When the sign is perceived, the organism produces an in- ternal response which is some fraction of the total response elicited by the thing signified (significate). The internal response in turn pro- duces an internal stimulus. This stimulus may produce overt instrumental responses which would not occur in the absence of the sign-significate contiguity. Meanings, than, are internal responses which constitute predisposi- tions toward external responses. 'lhey cannot be measured directly, but Osgood suggests that measurements of overt instrumental responses can be used as indices of internal states. 01' the possible overt responses to the representational mediation process which might be measured, Osgood 6 . Charles E. Osgood, George J. Suci, and Percy H. Tannenbaun, The Measurement 2; Meaning (Urbana: University of Illinois Press, 1957'):— chap. ii. chose linguistic behavior as being easiest to record, to quantify, and to interpret. He decided to elicit linguistic responses chosen by subjects from a sample of alternate responses selected by the researchers as rep- resentative of the principal variations in meanings. The alternatives used were pairs of terms representing verbal Opposites, the members of each pair being reciprocally antagonistic. The choice of one alternative meant rejection of the other, and thus the meaning of a sign was delim- ited by each response. In order to determine the strength of the choice of one alternative, a scale was inserted bettceen the two. The direction of the choice from the center of the scale indicated the direction of the meaning, and the distance from the center indicated the intensity of the meaning. The scale appears as follows, using various pairs of verbal opposites such as GOOD-BAD: GOOD: 8 : : z z : :BAD ‘75 '75 TI ’0" Ti" 2'2" T The center or zero scale position indicates no choice between alterna- tives. A set of such scales comprises the measuring instrument, which Osgood calls the Semantic Differential. A subject's meaning for a concept is determined by his selection from pairs of alternatives. In order to combine the pieces of meaning provided by each choice, Osgood postulates a multi-dimensional semantic 32a}: in which each scale is represented as a straight-line function passing through the point-of-origin. The center position of the measure- ment scale is represented in the semantic space as the point-of-origin. All other scale positions are represmted in the semantic space as posi- tions at corresponding directions and distances away from the point-of- origin. Each Semantic Differential scale represents a specific function passing through semantic space. Only one position may be chosen on each scale. There may be as many positions chosen as there are scalesu“ Mean- ing, then, is defined as the position in the semantic space which is chosen by successive specification of positions along each individual scale. In cases where the scale positions chosen by subjects for two or more scales are identical or highly similar, the semantic differential's developers assume that the scales are associated with the same dimension. Through factor analysis, they have compared the responses made by sub- jects to a large number of scales. The scales fell into at least three groups showing relatively high within-group correlation and relatively low between-group correlation. These groups were interpreted as three dimensions of meaning, which were labeled I'e'craluative," I'activity," and ”potency." The dimensionality of the semantic space is far from exhaust- ed by these three dimensions, but the factor analysis did not provide sufficient data to justify the specification of additional dimensions. ’In practice, meaning is defined by Osgood and associates as the po- sition in the semantic space chosen by specification of positions along the evaluative, activity, and potency dimensionsf The congruity principle involves attitudes, and attitude is defined by Osgood as one dimension of meaning: Despite a plethora of definitions of "attitude" in contemporary social science, some consensus and agreement is evident, par- ticularly with respect to the major prOperties that attitudes are assumed to possess. Most authorities are agreed that atti- tudes are...predispositions to respond, but are distinguished from other such states of readiness in that they predispose toward an evaluative response. Thus, attitudes are referred to as "tendencies of approach or avoidance,“ or as "favorable or unfavorable," and so on. This notion is related to another shared view-«that attitudes can be ascribed to some basic bi- polar continuum with a neutral or zero reference pgint, imply- ing that they have both direction and intensity... Osgood thus equates attitude with evaluative meaning-~that is, with the evaluative dimension of meaning which is determined through responses to evaluative scales: In terms of the operations of measurement with the semantic differential, we have defined the meaning of a concept as its allocation to a point in the multidimensional semantic space. We then define attitude toward a concept as the projec on of this point onto—Wluative dimension of that space. The two characteristics of an attitude, direction from point-of- origin and distance from point-of-origin, are obtained by judgmelts made on scales representing the evaluative dimension. thermally, an individual will have only one attitude toward a given object. Two or more attitudes toward the same object are mutually incom- patible. There are circumstances, however, in which an individual, to be logically consistalt, will invoke more than one attitude toward an ob- ject. Suppose he has attended to an object and has invoked an attitude of +3. The object, let us say, is a person whom we will call A. The in- dividual has also invoked an attitude of -2 toward another attitude ob- ject, a person called B, and we will imagine that this implies for him a similar attitude of -2 toward anyone who is a friend of B. Sometime later, the following event occurs. Our individual hears A describing B 7Ib1de , 189-190 a 8 Ibid., 190. 10 as his close personal friend.9 This event legically commits the individ- ual to invoke an attitude of -2 toward A. Consequently, if the logic of consistalcy is talable, he now has attitudes of both +3 and -2 toward A. The congruity principle assmnes that the individual will find it unsatisfactory to retain both attitudes toward A. Such logical uincon- gruity' will be uncomfortable, and the individual will experience a stress toward reducing the two attitudes to a single attitude. The con- gruity principle is designed to predict the value of the resulting attitude. ( The congruity principle states a set of conditions under which two attitudeslo are related congruently: Whenever two signs [i.e., attitude objects] are related by an assertion, they are congruent to the extent that their mediating reactions [i.e., the attitudes held toward them] are equally intense, either in the same (compatible) dir- ection of excitation in the case of associative assertions or in opposite (reciprocally antagonistic) directions in the case of dissociative assertionseu Whenever the attitudes do not meet these conditions, they are related incongruently. Congruity, or a congruent relationship, is thus a state in which the attitudes are harmonious and do not tend toward change. Incongruity is a state in which the attitudes are not harmonious and tend toward change in the direction of congruence. Predictions of the direction and intensity of change are determined 9In other words, A makes an associative assertion toward B. A nega- tive statement by A toward B would constitute a dissociative assertion. 10 The congruity principle does not apply only to attitudes. How- ever, it will be presented here with emphasis on such application. llOsgood, Suci, and Tannenbaum, 32. 22.12., 203. 1.1 in a two-step procedure. First, the locations o_f_ congruence are determ- ined. The location of congruence of an attitude is the direction and in- tensity which it must have in order to be congruent with other relevant information. This information consists of (1) an attitude toward some other attitude object, and (2) an assertion which relates the two ob- Jects. The location of congruence is determined as follows: Whenever one object of judgment is associated with another by an assertion, its congruent position along the evaluative di- mension is always equal in degree of polarization [i.e. , in- tensity] to the other object of judgment and in either the same (positive assertion) or opposite gnegative assertion) evaluative direction [as the latter].1 This can be expressed by the following equations: pal .- 92 and p02 . p1 (for associative assertions) p°1 - -p2 and pc2 - -p1 (for dissociative assertions) where p1 and p2 refer to the direction and intensity of attitudes for ob- jects l and 2, respectively, and p°1 and pc,2 refer to the direction and intensity of the location of congruence for these attitudes. For exam- ple, if a +3 attitude'is associated with object 1, and a-2 attitude with object 2, and the two objects are related by an associative assertion, then the location of congruence for the +3 attitude is -2. If the ob- Jects are related by dissociative assertion, then the location of congru- ence for the +3 attitude is +2. For the -2 attitude, the location of congruence would be either +3 for an associative assertion or -3 for a dissociative assertion. When the locations of congruence are determined, the second step is to predict the magnitude of the change. The principle on which such pre- 12 Osgood and Tannenbaum, 22. £133., 16. 12 dictions are based can be stated as follows: The original attitude shifts toward the location of congru- ence, the magnitude of the change (0) being a proportion of the distance from the original attitude (p1) to the loca- tion of congruence (p ), that prOportion being the ratio of the intensity of the location of congruence to the sum of the intensities of the location of congruence and the orig- inal attitude. This principle is expressed by Osgood and Tannenbaum through the following equation: 'pz‘ 0 '73 H j “'6' N A ’"C N 'U H V For example, if the original attitude is +2, and the location of congruence is -2, then the predicted change is ~2. An additional problem in making predictions is the ”believability" of the message, i.e., the subject's belief in the credulity of the rela- tionship. Osgood and Tannenbaum suggest an incredulity factor which tends to diminish the amount of change predicted. Although they do not specify a qmntitative formula for incredulity, they do state a general rule: The amount of incredulity produced when one object of judgment is associated with another by an assertion is a positively ac- celerated function of the amount of incongruity which exists and Operates to decrease attitude change, completely elimin- ating change when maximal.]~3 Figure 1 states predictions of direction and intensity of change for various combinations of original attitude and location of congruence. Corrections for credulity are included. These corrections, although fol- lowing the general rule, are empirical determinations arrived at arbi- trarily by Osgood and Tannenbaum: Research 93 Congggity.--Enq3erimental evidence for the congruity 131mm, 1.9. LOCATION ORIGINAL “mtg +3 0 .6 1.5 3 2.6 1.5 0 +2 -. o .7 2 1.8 1.3 .5 +1 -.5 -.3 O l .9 .7 .h o o o o o o o o -1 -.h -.7 -.9 -1 o .3 .5 -2 -.5 -1.3 -1.8 -2 -.7 o .h -3 o -1.5 -2.6 -3 -1.5 -.6 0 .Fig. l.-4Predicted changes in atti ude as corrected for incredulity. principle has been obtained by examination of several types of change situations. Most typical is a situation in which the source is a public news medium, or a public figure being quoted in a public news medium, and the concept is an event or concern of public interest. In an experiment by Tannenbaum,15 subjects saw newspaper articles in which a source made either a positive or negative assertion toward a concept. The sources 'were "labor leaders," "Chicago Tribune," and "Senator Robert.A. Taft." They made assertions about, respectively, the concepts of "legalized gam- bling," "abstract art," and "accelerated college programs." Subjects' attitudes toward source and concept were measured in advance, then again after exposure to the newspaper article. The resulting changes in atti- tudes correlated .91 withcongruity principle predictions. 1hIbid. 151m. , so-su. lb Tannenbaum varied all possible combinations of attitudes and asser- tions. Attitudes toward both source and concept could be either positive, negative, or neutral; the assertion could be positive or negative. Pre- dictions were made for both source and concept for each of the 18 combin- ations. In every cell the direction of change corresponded. to predic- tions. Another test of the principle was made by Kerrick.16 According to Osgood's summary of this experiment: Kerrick has tested the congruity hypothesis under more real- istic conditions. The effects of an editorial (message) in a weekly newspaper (source) urging election of certain peo- ple to the school board were predicted from pro-exposure attitude scores in a sample of the readers of the weekly. The average effect of the editorial was nil, but when the sample was broken into two groups--those relatively favor- able toward this weekly and those relatively unfavorable-- it was found that attitudes toward the concept went up for the group favorably disposed toward the source and went down for the other group. In an experiment of a different sort, the congruity principle was applied not to sources, concepts, and assertions, but to adjectives, nouns, and adj active-noun combinations}8 Subjects recorded meanings on Semantic Differential scales for two types of signs, adjectives ("sin- care,n "breezy," "listless,’l etc.) and nouns “husband," ”prostitute," ”nurse,‘I etc.). Then all possible combinations of the two ("sincere hus- band," etc.) were prepared and submitted to the subjects for rating on l6Jean S. Kerrick, "The Weekly Newspaper as a Source: Prediction of an Editorial's Effectivenessu (unpublished research, 1959). 17 Charles E. Osgood, "Cognitive Dynamics in the Conduct of Human Affairs," Public Opinion Quarterly, XXIV (Summer, 1960), 360. 18Osgood, Suci, and Tannenbaum, 33. 23., 275-281;. 15 the same scales. Before such ratings were obtained, the meanings of the combinations were predicted from the available information. The event of combining adjective and noun constituted an associative assertion. If the adjective was rated +1 and the noun +3, it was predicted that the ad- jective-noun combination would be rated at +2.5. In Figure 1 this pre- diction may be read by noting that the change predicted for the adjective is +1.5 and that for the noun is -.5. This brings the two together at +2.S.- Correlations between predicted and reported meanings were found to be .86, .86, and .90 for the evaluative, activity, and potency dimen~ sions’ respectively. In another experiment, Osgood and Tannenbaum examined the effects of color on the meanings of advertised products and on the meanings of abstract three-dimensional forms.19 The meanings for the products changed in the direction of congruity in 77% of cases. Meanings for the toms changed in the direction of congruity in 75% of cases. Both these results were significant. In other experiments, Dodge examined the development of meanings of nonsense assign and found that results correlated highly with congruity predictions, 2° and Kerrick emnined the relevance to each other of var- ious attitude objects and found that congruity effects occurred more strongly when relevant rather than non-relevant objects were paired.21 mm” 299-302. 20.1 can S. Dodge, "A Quantitative Investigation of the Relation Be- Ween Meaning Development and Context" (unpublished doctoral disser- tation, University of Illinois, 1955). 21 Jean S. Kerrick (unpublished research, University of Califor- nu, 1955). 16 In summary, then, there is considerable evidence to support the con- gruity principle as a predictor of attitude change. RATIONALE FOR THE PRESENT STUDY It is the purpose of this study to demonstrate the effects of non- congruity factors in attitude change. Osgood and associates have pro- duced considerable evidence which supports the predictive ability of the congruity principle. From these results, we might conclude that congru- ity factors alone are related to attitude change. Harburg and Price, 22 however, report an experiment with differing results. Harburg and Price asked subjects whether they would want a disliked acquaintance to like, dislike, or be neutral to their friend. The die- liked acquaintance was seen by the subjects as the source of an assertion about the liked friend (who was the concept). The source was negative, the concept was positive, and the value of the assertion was to be elic- ited from the subject. If the assertion were positive, the situation would be incongruent according to congruity theory. If the assertion were negative, the situation would be congruent. Congruity theory states that persons will tend to eliminate incongruity, but will not tend to change congruent situations. Consequently, the congruity principle would predict that a negative assertion would be preferred by subjects in order to avoid incongruity. Harburg and Price found that 39% of subjects wanted their fried to be liked by the disliked person. Thirty-seven per cent chose a neutral relationship. Only 214% chose the negative relationship predicted by 2Ernest Harburg and Kenneth Price (unpublished research, University of Michigan, 1960) . l7 congruity theory. The Harburg and Price study, as a test of the congruity principle, was limited in at least three respects: (1) The study did not examine all 18 possible combinations of the three congruity factors of attitude toward source, attitude toward. concept, and value of assertion. (2) The data obtained were, in effect, predictions about whether or not change would occur under given circumstances. No data were obtained to determ- ine whether change actually occurred. Thus Harburg and Price obtained predictions concerning change, but did not test the predictions. (3) The study considered only predictions of "change“ and "no change, " whereas the congruity principle permits predictions of amount of change as well. Although it was limited in these ways, the Harburg and Price study constituted an adequate test of congruity theory for the situation which it examined. Moreover, these results suggest other types of situations in which subjects might differ in their reactions. For example, suppose the source of an assertion is positive, the concept is positive, and the assertion is negative. Congruity theory would predict, on the basis of the incongruity involved, that the subject will reduce his positive atti- tude tantard both source and concept. The important point to be noted about congruity theory is that the same prediction will be made for all possible occurrences of this situation involving positive source, posi- tive concept, and negative assertion. The substitution of one positive source for another will not change the congruity prediction. Nor will the substitution of different subjects, different positive concepts, or different negative assertions. Congruity theory makes predictions on the basis of three factors only: (1) the attitude whose change is to be 18 predicted, (2) the subject's attitude toward some other object, and (3) the subject’s knowledge of an assertion made by one of the objects about the other. No other factors are relevant to change according to the con- gruity principle. Therefore, one‘prediction is sufficient for each com- bination of congruity factors. A possible interpretation of the Harburg and Price experiment is that congruity factors are not the only factors relevant to attitude change. Instead, the set of all relevant change factors might include non-congruity as well as congruity factors. The congruity principle does not account for these factors, and congruity predictions consequently are weakened. It is the purpose of this study to demonstrate such non-congruity effects. One possible method of doing this would be to obtain'data for which congruity predictions had a low correlation with observed change. The Harburg and Price study reports such data, but it is incomplete for reasons reported above. Alternately, we might expand. the Harburg and Price design into a replication of the Tannenbaum design, using "friends" and "disliked acquaintances" as sources and. concepts. If congruity pre- dictions had a low correlation with, change, we might attribute such find- ing to the effects of non-congruity factors. However, a low correlation of congruity predictions with change would not necessarily permit an interpretation that non-congruity factors were related to change. The result could be attributed to error. Fur- thermore, the correct interpretation of "low correlation" is a problal. Non-congruity factors could affect change considerably even in cases where the correlation of congruity predictions with change is high enough 19 to be statistically significant. A better method of demonstrating the effects of non-congruity fac- tors would be to have an attitude change predictor which incorporated their effects. A suitable predictor, for example, would be one which ac- counted for all relevant change factors, including both congruity and non-congruity factors. If such a predictor were developed, we could compare its correlation with change to that of the congruity predictor. If we found the new predictor to be more accurate than the congruity pre- dictor, we could confirm our assumption that non-congruity factors are present and are related to change. In order to mks such an experimental comparison, we must develop a predictor which is based on all relevant change factors. For this pur- pose, we will present a predictive principle called the mediated attitude rinci 1e, which is derived from the congruity principle. It is based. on the mediated attitude, a concept which is derived from the congruity principls's location of congruence. The location of congruence is a variable which the subject presum- ably perceives as relevant to the attitude whose change is to be predict- ed. In a typical congruity situation, the subject has an original atti- tude toward some attitude object. We want to predict the change which will occur in this attitude when the subject considers some additional information concerning the object. This additional information consists of the subject's attitude toward a second attitude object and his know- ledge of an assertion relating the two objects. As a result of exposure to this new information, the subject chooses an attitude toward which he thinks his original attitude should change. Osgood and Tannenbaum call 20 this attitude the location'of congruence. Its value is the direction and intensity which the original attitude should have in order to be congru- ent with the two additional pieces of information (which are called con- gruity factors). Osgood and Tannenbaum assume that the value of the location of con- gruence is fixed by these two congruity factors (according to the formu- lae reported earlier on page 11), and that it will vary only as they vary. Because of this, there is no need to measure the value directly. The mediated attitude principle stems from the assmnption that the correct value of the location of congruence cannot necessarily be derived from the two congruity factors, as the congruity principle assumes. In- stead, we assume that the value must be derived from all factors which are relevant to the subject in making such a decision. Given the two congruity factors, we assume that the direction and intensity are not fixed precisely by these two factors but may vary in accordance with ad- ditional non-congruity factors which may occur. Therefore, Osgood and Tannenbaum's method of determining the value of the location of congru- ence may be inaccurate in some cases. To correct for possible inaccuracies, we will present here an al- ternate method for determining the value of the location of congruence. This method involves a somewhat different conception of the location of congruence. we will give this new conception a new name: mediated git;- tude. Mediated attitude and location 2; congruence represent slightly different ways of describing the intervening variable, and involve dif- ferent methods of determining its value. The distinction between the mediated attitude and the location of congruence is the basis for the 21 difference between the mediated attitude principle and the congruity principle. There are no other differences between these two predictive principles. The principles predict attitude change in the same manner except that the mediated’attitude principle uses the value of the med- iated attitude instead of the value of the location of congruence. The location of congruence has a value which is dependent upon the two con- gruity factors. The mediated attitude has a value which is not necessar- ily dependent upon them. The mediated attitude is an attitude chain made up of two direct attitudes. A M attitude involves an assertion such as "I like (or dislike) so-and-sod' It may be illustrated in this way: Subject -——-) Concept In addition to'a direct attitude toward the concept, the subject may have an attitude which is mediated through another person who acts as the source of his own direct attitude toward the concept: Subject ---—) Source --—-9 Concept This mediated attitude toward the concept is made up of two atti- tudes. The first is the subject's direct attitude toward the source. The second is the source's direct attitude toward the cancept. When a source's direct attitude toward a concept is part of the subj ect's med- iated attitude toward the concept, we will call the source's direct at- titude a relay attitude. This term is equivalent to the term 5mg!- ___tiog used by Osgood and Tannenbaum. A mediated attitude toward a concept thus is an attitude chain made up of a direct attitude toward some other attitude object plus a relating attitude which associates the two objects. Whenever there is both a 22 mediated attitude and. a direct attitude toward a concept, illustrated as follows: Subject 5 Source 1 Concept there is also present a mediated. attitude toward the source as well. It is made up of the direct attitude toward the concept, plus the relating attitude of the source toward the concept. It is illustrated in the way following: ' Subject " ' ’Cmcept‘ ) Source The two illustrations show the same structure. They are drawn dif- ferently to emphasi as, in the first drawing, the mediated attitude toward. the concept, and, in the second drawing, the mediated attitude toward the source. Both are present when two direct attitudes are combined by a relating attitude. The value of a mediated attitude is assmned to be a product of all factors which are considered relevant by the subject. It is not nec- essarily dependent on the values of its two component attitudes (although the location of congruence is depmdent on these values). Nor are the two mediated attitudes in a given situation necessarily dependent upon each other-«the value of one cannot be derived from knowledge of the other. If the mediated attitude is to be used, we need a means for determ- ining its value. There is at present no basis for deciding which are relevant factors and which are irrelevant. Accordingly, the mediated attitude' 3 value is not computed from a postulated algebraic process but is elicited directly from the subject. 23 Suppose we are concerned with a mediated attitude toward a concept C. A subject's direct attitudes toward C and toward another object S are known. The subject is then confronted with the fact that S has an atti- tude of such-and-such toward C. The subject is asked whether or not this fact has any relevance to his own evaluation of C. The following is a way of asking this question: Please indicate, by making a check-mark next to one of the fol- lowing statements, whether S's attitude toward C has some ef- fect on what your own attitude toward C should be---or has no such effect. Please think very carefully about this. Try to think of any possible way in which S's attitude may have some effect on your own. Do not specify "no effect" unless you are confident that none exists: ‘ S's attitude toward C: 1. Has some effect on what your own attitude toward C should be. 2. Has no effect on what your own attitude toward C should be. If the subject checks the second statement, we conclude that S's attitude is "irrelevant" and that no mediated attitude toward 0 exists. If he checks the first statanent, we conclude that a mediated attitude exists, and we ask the subject what evaluation of C the situation im- plies. He is not asked to state what evaluation of C he would conse- quently make, but what evaluation of c is implied by this single fact alone. The following is a way of asking this question: If you checked statement 1, please record below the ratings of C which you feel would be in perfect accord with your know- ledge of 3's attitude. Do not state what your ratings of C actually have been in the past, or are now, or will be in the future. Instead, state the ratings of C which you would make if you were rating C on the basis of this one fact only, with all other considerations eliminated. Following this a set of attitude measurement scales could be com- pleted. The direction and intensity recorded by the subject on these 2h scales would be used. as the value of the mediated attitude. Not counting variations in intensity, the mediated attitude may oc- cur in four states: positive, negative, neutral, and irrelevant. The mediated attitude principle obtains a change prediction by sub- stituting the value of the mediated attitude in place of the value of the location of congruence and then proceeding in the manner of the congru- ity principle. Except for this, the two principles are similar. The mediated attitude principle thus may be stated as follows: The direct attitude shifts toward the mediated. attitude, the magnitude of the change being a proportion of the distance between the'two, that proportion being the ratio of the in- tensity of the mediated attitude to the sum of the intensi- ties of the mediated attitude and the direct attitude. The congruity principle, as noted earlier, includes corrections for incredulity. The mediated attitude predictions are made without the in- credulity corrections. It is assumed that if incredulity is a relevant variable to the subject, the mediated attitude will account for it and no adjustment need be made. Figure 2 presents a set of predictions for change in attitude based on the mediated attitude principle. Other than the elimination of in- credulity corrections, the values of the predictions correspond. to those of the congruity principle. The independent variables of Figure 2 sub- stitute "direct attitude" for "original attitude " and substitute "med- iated attitude" for "location of congruence." If the mediated attitude is irrelevant, the prediction is for zero change. From the above description of the mediated attitude principle, we see that mediated attitude predictions will differ from congruity predic- tions only when the mediated attitude and location of congruence have 25 DIRECT ATTITUDE MEDIATED w *3 +2 *1 .2. .2. .22.. .1. +3 0 .6 1.5 3 3 3 3 +2 -. O .7 2 2 2 2 +1 -.5 -.3 o l 1 1 1 o o o o o o o o -1 -1 -1 -1 -1 o .3 .S -2 -2 -2 -2 -2 -.7 o .h -3 -3 -3 -3 -3 -1.5 -.6 0 Fig. 2.--Predicted changes in attitude for mediated attitude principle different values. We assume that differences in values will occur only when the two values are obtained on the basis of two different sets of factors. We know that the value of the location of congruence is determ- ined on the basis of two congruity factors. Therefore, whenever the med- iated attitude has some different value, we will assume that it has been determined by a set of factors which include non-congruity as well as congruity factors. Whenever the mediated attitude and the location of congruence differ in value, the two predictive principles will produce different predic- tions. The mediated attitude prediction will account for non-congruity factors, and the congruity prediction will not. Therefore, if the med- iated attitude predictions are more accurate than congruity predictions, we will assume that the non-congruity factors were determinants of change and important in predicting change. 26 The present study was designed to compare the accuracies of the two principles in predicting change. To make this comparison, an experiment was designed which utilizes certain characteristics of the Tannenbaum and Harburg and Price experiments. The design used by Tannenbaum was replicated in order to include all possible combinations of congruity factors. However, an 33293 replication of the Tannenbaum experimental sit- uation was not used. Tannenbaum's reported correlation of congruity pre- dictions with change was extremely high. There is little chance that mediated attitude predictions would be significantly higher. From this, we conclude that non-congruity factors had no relationship to change in the Tannenbaum experimean situation. However, the Harburg and Price experiment suggests that the Tannenbaum situation might be altered in a way that would make non-congruity factors become important determinants of change. The attitude objects used by Harburg and Price were personal ac- quaintances of the subjects, any persons the subjects might choose as "friends" and "disliked acquaintances." The attitude objects used by Tannenbaum were known to the subject principally because they were widely known to the public. Thus the Tannenbaum and Harburg and Price attitude objects may be characterized, respectively, as mediate and immediate. An immediate attitude M is one of whom the subject's knowledge is principally the result of his own direct personal contact. A mediate 35‘." m M is one of whom the subject's knowledge is principally the result of someone else's contact. The "someone else" may be a single individual, a group, or the entire public. 27 The type of attitude object appears to be the only difference be- tween the Tannenbaum and Harburg and Price experiments which can be re- lated to the differences in their correlations with congruity predic- tions. Since attitude change for immediate attitude objects was weakly related to congruity predictions, it may be that such change was signif- icantly related to non-congruity factors. If this were the case, then we could distinguish between immediate and mediate attitude objects on the basis of their relationship to non-congruity factors. Both congruity and non-congruity factors would. be determinants of attitudes toward im- mediate attitude’ objects. Determinants of attitudes toward mediate ob- jects would be congruity factors only. Accordingly, the experiment reported here has utilized immediate at- titude objects. The design replicates that of Tannenbaum, with the sub- stitution of immediate attitude objects. HTPOTHESIS The assumptions underlying the mediated attitude principle suggest the following expectation: In cases where relevant factors include a large proportion of non-congruity factors, the mediated attitude predic- tions of change will be significantly more accurate than will congruity predictions. It is expected that situations involving imediate attitude objects will involve a large proportion of non-cOngruity factors. For these reasons, given an immediate attitude object situation, it is hypothesized that: 2.21: _w_i_l_._l 23 significantly more accurate M Lill predictions based an .t_hg congruity principle. 28 VARIATIONS IN THE RELATIVE ACCURACY OF THE PREDICTORS The data used to test the hypothesis will be collected under varied experimental conditions. Therefore, if a significant difference in ac- curacy betzveen the two attitude change predictors is obtained we will want to ask whether or not this result may be expected to occur under all of these conditions. It is possible that the data will be independent of certain variables and not independent of others. This could mean that there are variations in the extent to which non-congruity factors are determinants of changes in attitudes toward immediate attitude objects. Accordingly, for each of nine experimental variables the null hy- pothesis will be tested thatthe relative accuracy of the two predictors is independent of the variable. No theoretical hypotheses will be stated concerning independence or dependence. The nine variables are: 1. Value of the mediated attitude: positive, negative, neutral, or irrelevant. 2. Type of attitude object: source or concept. 3. Value of the assertion: positive or negative. 14. Attitude toward source: positive, negative, or neutral. 5. Attitude toward concept: positive, negative, or neutral. 6. Attitude toward attitude object: positive, negative, or neu- tral. 7. Type of situation. The 18 combinations of attitude toward source, attitude toward concept, and value of assertion (considering di- rection but ignoring intensity) may be described as congruent, incongru- ent, or ”other." Four of the combinations are congruent: a. Source: positive; Concept: positive 3 Assertion: positive 29 b. Source: negative; Concept: negative; Assertion: positive c. Source: positive; Concept: negative; Assertion: negative d. Source: negative; Concept: positive; Assertion: negative Four other combinations are incongruent: e. Source: positive; Concept: positive; Assertion: negative 1‘. Source: positive; Concept: negative; Assertion: positive g. Source: negative; Concept: positive; Assertion: positive h. Source: negative; Concept: negative; Assertion: negative 0f the ten other combinations, those involving one neutral attitude object are partially incongruent, and those involving two neutral atti- tude objects are neither congruent nor incongruent. These ten combina- tions are included in an "other" category. 8. State of the real-life situation. When an experimental subject is asked to consider an imagined situation involving real persons, the situation may or may not be at variance with what is actually the case. There may not have been an assertion by the source about the concept in real life. If there was an assertion it may have been positive, nega- tive, or neutral. The subject may or may not know what is the actual case. A parallel question is that of the conceivability of the suggested situation. If the experimental assertion is the same as the real-life assertion there can be no question of its conceivability. But in all other cases the subject may feel that the assertion either could or could not have happened in real life. Or, he may not know which is the case. Accordingly, subjects may be divided into four categories: a. Those who find the situation conceivable and report that the attitude actually held by the source is the same as is suggested in the experiment. b. Those who find the situation conceivable although they re- port that the attitude actually held is not the same as that suggested. 30 c. Those who find the situation inconceivable and report that the source's attitude is not the same as that suggested. d. All others. This includes those who make the "don't know" responses plus any persons who may make the illogical response that the situation is inconceivable but the source's attitude nevertheless is the same as that suggested. 9. A personality characteristic of subjects: Openmindedness and closedmindedness. This distinction is based. on Rokeach's work with the Dogmatism Scale.23 Persons with high and low scores on this scale are called, respectively, openminded and closedminded. The description of open and closed minds suggests the possibility that closedminded subjects will make attitude changes as predicted by the congruity principle to a simificantly greater extent than will openminded subjects. .A recent study by Powell supports this implication from Rokeachis theory.2h Powell obtained empirical data showing that Open minds have a relatively higher ability than closed. minds to differentiate between, and independently evaluate, a message and a message source. The results of this study, which uses mediate attitude objects, suggest that a hypoth- esis of greater congruity effects by closed minds would have been sup- ported in Tannenbaum's experiment if his subjects had. responded to the Dogmatism Scale. Accordingly, confirmation of the same hypothesis might be expected 23Milton Rokeach, .1113 Open and Closed Mind (New York: Basic Books, 1960). at Frederic A. Powell, "Open- and Closedmindedness and the Ability to Differentiate Between Source and Message" (unpublished Master's thesis, Michigan State University, 1961). 31 in the present study; On the other hand, we have also suggested that congruity effects will be less evident in a study'involving immediate attitude objects. If this expectation is confirmed, it seems less likely that a hypothesis of greater congruity effects by closed minds can be confirmed. Consequently, no theoretical hypothesis will be offered here concerning the difference between Open and closed minds. CHAPTER II This chapter is concerned with the experiment used to test the hy- pothesis, including descriptions of the subjects, the experimental de- sign, the measuring instruments, and the procedures used to obtain the data. SUBJECTS One hundred and twentybsix.undergraduate college students were chosen as subjects. All were participating in a class in communication arts at Michigan State University during the spring quarter of 1961. They carried out their task during portions of two class meetings. EXPERIMENTAI.DESIGN (A typical pre-post design was used. In the pre-test phase, subjects rated attitude objects. In the intermediate phase, the subjects were told that one attitude object (a positive, neutral, or negative source) had expressed an attitude (a positive or negative assertion) toward another attitude object (a positive, neutral, or negative concept). In the post-test phase, subjects made final ratings of attitude objects. For the intermediate phase, fourteen subjects were randomly assigned to each of nine groups, as seen in Figure 3. Two separate attitude change situations, each using different attitude objects, were assigned to each group. These 18 situations presented all combinations of value of attitude toward source, value of attitude toward concept, and value 32 33 POSITIVE ASSERTION POSITIVE A NEUTRAL NEGATIVE SOURCE CONCEPT CCNCI‘PT CONCEPT POSITIVE Group 1 Group 5 Group h (first)3 (first) (first) NEUTRAL Group 9 Group 2 Group 8 (first) (second) (second) NEGATIVE‘ Group 7 Group 6 Group 3 (first) (second) (second) NEGATIVE ASSERTION POSITIVE NEUTRAL NEGATIVE SOURCE CON CEPT CONCEPT CONCEPT POSITIVE Group 2 Group h Group 5 (first) (second) (second) NEUTRAI. Group 8 Group 3 Group 9 (first) (first) (second) NEGATIVE Group 6 Group 7 Group 1 (first) (second) (second) a "(first)' and "(second)" indicate the first and second.situations encountered by each group Fig. 3.-«Assignment of Eighteen.Combinations of Congruity Conditions to Nine Groups of Subjects of assertion. Since each situation included two kinds of attitude ob- jects, a source and a concept, ratings of attitude objects were made un- der 36 combinations of conditions. Each subject responded under four of the 36 possible conditions, rating both source and concept in both situations assigned to his group. 3h Ideally, the design would have had each subject respond under one condition only. Instead, the design permitted subjects to respond under all four conditions assigned to a given group. This could have allowed undue homogeneity of responses among conditions within groups. This was undesirable, but was permitted because of the desire to obtain as large a sample as possible. .A statistical test to determine whether greater than chance dependence occurred within groups of four responses will be reported in Chapter III. An additional independent variable in the experiment was the med- iated attitude, varying in four ways: positive, negative, neutral, ir- relevant. The use of this variable posed a problem for the experimenter, since it was impossible to control the assignment of its values. Ideal- ly, the design would involve assignment of an equal number of subjects to each of the values of this variable for each of the 18 change sit- uations. Maximum generalizability of a theoretical principle cannot be obtained unless the experimental situation exposes subjects to all pos- sible states and combinations of states of the independent variables. However, the values of the mediated attitude were selected by the subject, not by the experimenter. There was no means of determining g ‘priggi‘what they would be. The experimenter had to acceptwwhat the sub- ject provided. For example, for the combination of positive source, pos- itive concept, and positive assertion, the 28 obtained mediated attitudes in the experiment took only the positive and irrelevant values. They were never negative or neutral. If every logically possible event is represented in the population of empirical events, then any set of data not covering all such events 35 is limited in generalizability. In the present experiment, however, those events not covered by obtained data quite possibly did not have em- pirical counterparts. Consequently, it is the opinion of the experiment- er that the absence of data covering certain values of the mediated at- titude does not unduly restrict generalizability of the data. MEASUREMENT The measures required were (1) pre-experimental and post-experiment- al attitudes toward the attitude object, (2) mediated attitude toward the attitude object, (3) an index of the "real-life" variable, and (h) openmindedness and Closedmindedness of subjects. In addition, computatiore were needed. to determine change of atti- tude, the congruity predictions for change of attitude, and. the mediated attitude predictions for change of attitude. Pre-experimental and post-merimental attitudes.--These scores were obtained as responses to a set of six measuranent scales chosen accord- ing to two criteria. The first and major criterion was that responses should differentiate among positive, neutral, and negative attitude ob- jects. The second criterion was that the scales have high factor load- ings on the evaluative dimension of Osgood, Suci, and Tannenbaum. 25 Performance on these criteria was examined in a small pilot study. Twanty-four scales were tested, including 23 chosen from those most ac- ceptably fulfilling the second criterion. An additional scale, LIKE- DISLIKE, was included on the basis of its apparent association with eval- uative meaning. Although the two criteria appeared. to be equivalent, 2 5Osgood, Suci, and Tannenbaum, 92. 21.3., chap. ii. 36 they were not for all scales. Because the choice of six scales out of 2h was carried out largely on the basis of the major criterion, the chosen scales did not have the highest possible association with evaluative meaning. The six scales were LIKE-DISLIKE, FAIR-UNFAIR, PLEASINGqANNOY- ING, GRATEFULPUNGRATEFUL,.ALTRUISTIC-EGOTISTIC, and HARMONIOUS-DIS- SONANT. .A set of scales appeared as fellows: IIKE: : : : : : : :DISLIKE FAIR: : : : : : : :UNFAIR PLEASING: : : : : : : :ANNOYENG GRATEFUL: : : : : : : :UNGRATEFUL .ALTRUISTIC: : : : : : : :EGOTISTIC HARMONIOUS: : : : : : : :DISSONANT .Attitudes were defined as the algebraic mean of responses to six scales. Data involving responses to individual scales (e.g., LIKEADIS- LIKE) were not desired for any purpose and were not tabulated. Mediated Attitude.--This measure was obtained by asking the subject the following question: Please indicate, by making a check-mark next to one of the following statements, whether.A's attitude toward B has some effect on what your own attitude toward B should be---or has no such effect. Please think very carefully about this. Try to think of any possible way in which A's attitude may have some effect on your own. Do not specify "no effect” unless you are confident that none exists: A's attitude toward B: 1. Has some effect on whatlyour own attitude toward B Should bee 2. Has no effect on.what your own attitude toward B should be. If you have checked statement 1, please record below the 37 ratings of B which you.feel.would.be in perfect accord with your knowledge of A's attitude. Do not state what your ratings of B actually'have been in the past, or are now, or Will be in the future. Instead, state the rat- ings of B which you would.make if yoquere rating B on the basis of this one fact only,'with all other consid- erations eliminated: (If you have checked statement 2, do not do these ratings) This was followed by the set of six scales described in the previous section. If the subject checked statement 2, the value of the mediated attitude was taken to be "irrelevant." If he checked statement 1 and filled in the scales, the value of the mediated attitude was the alge- braic mean of responses to the six scales. The "real-life" situation.--This measure was obtained through re- sponses to two questions: 1. Please indicate whether the situation (A's attitude toward B) that we have suggested: 1. Could conceivably have happened. 2. Could not conceivably have happened. 3e Don't knOWe 2. Please indicate what attitude actually is held by’A toward B: 1. Positive attitude 2. Neutral or indifferent attitude 3. Negative attitude be NO attitUd e 5. Don't know According to their responses to the two questions, subjects were assigned to four categories as described earlier on pages 29-30. Qpenmindedness and closedmindedness.--The subject's score on the Rokeach.Dogmatism Scale26 determined this measure. Subjects scoring 175 or higher were called openminded. Those scoring 130 or lower were called closedminded. The remaining subjects were not important to the 26 ROkOQCh, go 2%" Chap. 1Ve 38 experimenter's consideration of this variable. Cormutationsw-Three pieces of data were derived from the basic measures. The amount of change of attitude was computed by subtracting the pro-experimental attitude from the post-experimental attitude. The congruity predictions were made according to the Osgood and Tannenbaum formulation presented in Chapter I. The mediated attitude predictions were made by substituting the mediated attitude for the location of congruence and proceeding accord- ing to the congruity principle. Incredulity corrections were not used. When the mediated attitude was irrelevant, the mediated attitude pre- diction was for zero change. PROCEDURES The Dogmatism Scale had been administered to subjects earlier in the school tem by the class instructor, and subjects' scores were made available. All other data were collected by questionnaire (see Appen- dix A). The questionnaire was issued in two parts, both of which were admin- istered to all subjects. Part One (pages 1-5) and Part Two (pages 6-11) obtained pre-experimental and post-experimental data, respectively. Part One was administered during a class lecture session. The questionnaires were distributed and the experimenter made the following announcement: The questionnaire you are receiving is concerned with a study being done in the department which involves the way in which we maintain and form our attitudes toward other persons. It is the kind of job which requires contacting and obtaining 39 data from many persons, and I hope you will be willing to give it your cooperation so that we may cover as many per- sons as possible in a short time. The questionnaire is not a requirement of your course. There aren't any right answers, and it has nothing to do with your grades. If you have any objections to participating, you certainly do not need to. However, it does involve cer- tain principles of interpersonal communication which you should find to be of relevance to your work in the course. The questionnaire comes in two parts. The first part only will be issued today. The second part will be issued at a later date. Because we need to match the two parts, it is very important that you place your name on the questionnaire. Please make sure you understand all instructions. Before you ask questions, please read the entire first page and. the sec- ond page as far as the statement concerning asking questions before you continue. Hold your questions at least until you reach that point. If you have no questions, keep right on going. Following this there were no questions, and the questionnaires were completed and turned in without incident. Two students turned in blank questionnaires. In Part One, subjects reported their original ratings of six persons of their own choice. They were requested. to choose two persons for whom they had a high regard, two of low regard, and two of neutral or indif- ferent regard. They also provided brief descriptions of these persons according to instructions. In order that the widest possible variety of types of persons might be chosen, subjects were restricted to a single selection from any one of a number of descriptive categories listed on page 2. After collecting Part One, the experimenter selected for each sub- ject four of the attitude objects chosen and recorded their symbols and descriptions on page 6 of Part Two. The selection was made according to the requirements of the experimental design. The experimenter was able hO to select attitude objects of positive, negative, or neutral attitude by choosing appropriately from the six attitude objects specified in Part One. The two unchosen attitude objects from Part One were subsequently eliminated from experimental considerations. The experimenter also chose, for each subject, the values of the assertions as required by the design. This was done by entering one of two words, "high" or "low," in the four blank spaces found on pages 6, 7, and 9. The experimenter then entered the subject's name at the begin- ning of’Part Two. ‘With these additions, Part Two was ready for use. Part Two was completed later in the same week during another session of the same course. It was administered during the discussion sessions, for which the class met in 11: small groups at different times and places. It was issued by the experimenter to several of the groups. Since cer- tain of the groups met at the same hour, the questionnaire was issued in a number of cases by the regular instructor. Part Two was not accompan- ied by instructions. It was distributed with a statement such as ("This is the second part of the questionnaire which you began last Monday. Please fill it out." There were a number of questions asked concerning the manner of responding to Part Two. Apparently, all were answered to the subjects' satisfaction. Twenty-six persons were absent from the dis- cussion sessions and all were administered Part Two individually by the experimenter during'the following week. Part Two obtained the subject's report of his final attitudes and mediated attitudes toward the four attitude objects. Part.Two also obq tained written comments by the subject concerning the experimental sit- uations, and his responses about the conceivability of the situation and ’41 whether it was actually the case. The written comments were requested with the hope of obtaining additional non-quantitative information which might help the experimenter to analyze the attitude change process. No systematic analysis of the comments was planned or undertaken. The requests for written comments were ignored by several subjects, but all other information in Part Two was provided by all subjects. CHAPTER III This chapter reports the results of the experiment, the statistical analyses used to test the hypothesis, and some data analyses in addition to those which test the hypothesis. RESUDTS The experiment exposed each of 126 subjects to four separate atti- tude change situations. The experimental.data consisted of five items obtained from each situation: (1) congruity prediction of change, (2) mediated attitude prediction of change, (3) observed change, (A) "real- life" data, and (5) Dogmatism Scale score. Congruity predictions, mediated attitude predictions, and observed changes were computed for each of the 50h individual cases. In addition, mean predictions and mean changes were compiled for the group of 1h sub- jects in each of the 36 experimental cells (see Table 1). Response Independence.-Itvwas pointed out in Chapter II that the experimental design permitted each subject to make responses in four of the 36 cells, and that these responses might have a consequent dependence on each other. A test for independence of the responses was needed and devised as follows. Out of 50h cases, the mediated attitude prediction was closer to observed change than was the congruity prediction in 308 instances. The reverse was true in 161 instances. The predictions were tied in 35 instances. Excluding tied cases, the mediated attitude pre- diction was more accurate in 65.7% of cases. From this .65? probability, 1:2 V h3 TABLE l.--Predicted and Obtained Mean Attitude Changes Within Cells POSITIVE ASSERTION NEGATIVE ASSERTION VALUE VALUE OF CONCEPT VALUE OF (DNCEPT OF SOURCE + O - + O - -.h -1e5 "he6 "he2 -201 "1e6 4' 1e]. "' e1 “1.0 "heé "' e8 -l.0 e1 "' e9 "2e1 -502 -2e7 e2 SOURCE 10.2 1e3 -909 “ll-e]. ‘5e9 8e? CMGES 0 09 3.0 1.0 '44.? -1401 “101'. la“. 2.9 09 44.0 -201 2011 10e9 2e2 -09 '2e0 2e]. Se6 - 2e9 1.5 ’eh " e2 e2 e2 700 Se6 2e2 " e6 20? 5.0 08 1006 807 -1406 -1140]- "Bel + .5 301‘ 5.0 -307 - 1e2 e0 e0 3.0 7e2 '1e6 e1 3e? CONCM "leh ‘1e3 1.5 '107 - OS 2.]. 0mm 0 - .6 100 2e0 -106 -2es 102 'le7 Zeb 5e]- “he9 Zeh he? 44.0 -10e2 e0 -1.0 10e2 6e8 " -1e9 lel 2e8 - e8 e? la; -1.0 Zeb» 6e9 “2.3 leh " 703 ' Upper value is mean congruity prediction for 1h subjects Middle value is mean mediated attitude prediction for 11: subjects lower value is mean obtained change for 11: subjects one can derive a set of expectations concerning the set of four responses made by each subject. Within each of these sets, the mediated attitude prediction could have been more accurate than the congruity prediction in either 14, 3, 2, l, or 0 cases. Table 2 shows the expected probability of each of these outcomes. Since there were 98 sets of four responses hh TABLE 2.--Expected and Observed Numbers of Sets of Fbur Responses Having Each of Five Outcomes as Predicted for Independent Responses POSSIBLE OUTCOMES (NUMBER OF MEDIATED n- EXPECTED NUMBER TITUDE PREDICTIONS OUT OF SETS (EKPEC- OBSERVED or I: WHICH WERE was EXPECTED TED PROBABILITY NUMBER ACCURATE) , __ PROBABILITY runs 98) OF SETS h .18 17.19 16 3 .38 37.68 he 2 .31 3o.h3 31 l .11 10.93 10 o .02 1.1;? 1 (not counting an additional 28 sets containing ties), Table 2 also indi- cates the number Of sets out of 98 expected to have each Of the possible outcomes. If responses within sets were made independently, then the 98 sets should reflect the expectations. If they were not independent, they should be significantly different from the expectations. In the final column of Table 2, the observed. figures are reported. A chi square test of the differences between Observed and expected figures was made and produced a chi square value of .5. The probability of chance occurrence of a value equal to or greater than .5 was greater than .95 (with h (11'). From these results, it seems reasonable to assume that each response was made independently of all others. Testing the Hypothesis.--The hypothesis to be tested was: Predictions of attitude change based on the mediated attitude prin- ciple will be significantly more accurate than will predictions based on the congruity principle. Basic data used in testing the hypothesis were derived by performing 145 the following operations: 1. Pearson product-moment correlations were computed between ob- served change and (a) congruity predictions and (b) mediated attitude predictions. This was done twice, once using data from the 50h individ- ual cases and once using means from the 36 cells. The experimenter chose the data involving individual cases as the most appropriate data for a statistical analysis. However, it was decided. to use the cell means in the analysis, also, in order to compare the results with a similar anal- ysis made for Tannenbaum's data. 2. Frequency of correct and incorrect predictions of direction of change were tabulated under each of the predictor systems. This was a less effective means of measurement because directional accuracy in many cases could not be computed. 3. Relative accuracy of the two systems was determined by tabulat- ing the number of times each predictor was more accurate than the other. This measm'e accounted for h69 out of 501; cases, excluding 35 ties. Table 3 presents these derived. data. STATISTICAL ANAHSIS In order to investigate the theoretic hypothesis, three null hypoth- eses were tested: (1) the correlations of the two predictors with change were not significantly different; (2) the two predictors were in the right direction in the same proportion of cases; (3) each predictor was more accurate than the other in an equal number of cases. All tests as- sumed a probability level of .05. The first test compared the .21 and .55 correlations for individual cases, and also compared the .55 and .77 correlations for cell means. 16 TABLE 3.--Comparison Among Predicted Changes and Obtained Changes CON GRUITY MEDIATED ATTITUDE TYPE OF COMPARISON PREDICTOR PREDICTOR Correlation between predictions and .21 .55 changes for 50).; individual cases Correlation between predictions and .55 .77 changes for 36 cell means Times prediction in right direction" 258 179 Times prediction in wrong direction“ 1118 h]. (Times directional accuracy of pre- 98 28h dictor could not be computed)‘ Times predictor more accurate than 161 308 other predictorb aDirectional accuracy of predictions was possible only whm both change and prediction had values other than zero. Because of this, di- rectional accuracy could not be computed for 98 out of 5014 congruity predictions nor for 28h out of 501; mediated attitude predictions. bThis measure could not obtained when the two predictions had iden- tical values. This occurred in 35 out of 50).; cases. Correlational analysis generally involves the assumptions of parametric statistics. Its use thus might be questioned here, since the values of the 5011 mdiated attitude predictions were not normally distributed. They included 261 values of zero, due largely to the 23!; cases in which the sub;) ect specified that the mediated attitude was irrelevant. Hotelling,27 however, has developed a t test to compare two predic- 27Harold Hotelling, "The Selection of Variates for Use in Predic- tion, with some Commts on the General Problem of Nuisance Parameters," Anna___1_s_ 2; Mathematical Statistics, XI (Fall, 1910), 271-283. ’47 tors. This test uses correlations but does not make assumtions about the distribution of the values of the predictors. Its use requires that the correlation between the two predictors be known. This correlation was .23 for individual cases and .56 for cell means. For individual cases, the value of 1: was 8.07. Thus the probability of the difference between .21 and .55 (or a greater difference) occurring by chance under the null hypothesis was less than .001 for a two-alternative test with 501 df. For cell means, the value of _t_ was 2.11. The probability of the difference between .55 and .77 (or greater) occurring by chance under the null hypothesis was less than .05 for a two-alternative test with 33 df.28 The second test compared the congruity predictor's figures of 258 right (63.5%) and 11.8 wrong (36.5%) with the mediated attitude predic- tor's figures of 179 right (81.11%) and 111 wrong (18.6%). A chi square test was used to show the probability of these two pairs of results oc- curring by chance under the null hypothesis that the proportions of right and wrong cases would be the same for both. The value of chi square was 21.5, the probability of chance occurrence thus being less than .001 28Hotell'Lng's test involves a qualification. The test's "advan- tages, he says in the article cited, "are attained at the expense of sac- rificing the precise applicability of the results to other sets of values of the predictors." In other words, the results reported here may be generalized to subsequent samples only when the values of the predictors in those samples are the same. The values of the congruity predictions reported herein were in line with the values expected by theory, and thus reoccurrence of the same values in subsequent samples seems probable. For the values of the mediated attitude predictor there were no theoreti- ' cal expectations. It cannot be asserted whether subsequent samples will or will not deviate from the present sample. The sample size suggests that some measure of confidence in the sample's reliability is justi- fied. In my event, the Hotelling test is only one of three statistical tests used to test the hypothesis. h8 ‘with 1 df. The third test compared the number of times each predictor was closer to observed change than the other. Thirty-five tied cases were excluded from this test. Of the remaining h69 cases, the mediated atti- tude predictions were closer to change in 308 cases (65.7%). Congruity predictions were closer to change in 161 cases (3h.3%). A sign test was used to show the probability of this result occurring by chance under the null hypothesis that each predictor will be closer to the observed change in equal numbers of cases. The probability was less than .01 for a two- tailed test. In all tests, the experiment's hypothesis was confirmed. First, for both individual cases and cell means, it was confirmed that the mediated attitude predictions had a higher correlation with change than did the congruity predictions. Next, it was confirmed that the mediated attitude predictions were in the right direction in a greater proportion of cases. Finally, it was confirmed that the mediated attitude predictions were closer to observed change in a greater proportion of cases. ANALYSIS OF REIATIVE ACCURACY ACROSS ECPERmmTAL CONDITIONS This section will examine the following question: ‘Was the differ- ence between the two change predictors independent of changes in experi- mental conditions? Chi square tests (two-alternative, .05 level) of in- dependence will be reported for each of nine variables. The null hypoth- esis under test is that the data relevant to the experiment's hypothesis ‘was independent of changing conditions. Mediated Attitude.-4was the relative accuracy of the two predictors independent of the value of the mediated attitude? The data of Table h 119 TABLE h.--Relative Accuracy of Predictors for Values of the Mediated Attitude VALUE OF MED- TIMES CCNGRUITY TIMES MED. ATT. IATED ATTITUDE MORE ACCURATE MORE ACCURATE ETA} Positive 25 (27.2%) 67 (72.8%) 92 Negative 2l (32.3) hh (67.7) 65 Neutral 28 (29.2) 68 (70.8) 96 Irrelevant 87 (110.3) 129 (59.7) 216 X2 - 6.8, df - 3; p>.O7 show a tendency (p> .07) toward rejecting a hypothesis of independence. When the mediated attitude was irrelevant, either (1) congruity predic- tions were relatively more accurate, or (2) mediated attitude predictions were relatively less accurate, or (3) both. The data in Table 5 were used to test whether the accuracy of the congruity predictions alone was independent of the value of the mediated attitude. The assumption of in- dependence cannot be rejected. Accordingly, it may be tentatively TABLE 5.--Accuracy of Congruity Predictor for Values of the Mediated Attitude VALUE OF MED- TIMES CCNCRUITI TIMES CCNCRUITI IATED ATTITUDE RIGHT DIRECTIoN WRONG DIRECTION fl Positive 52 (65.8%) 27 (311.2%) 79 Negative 36 (59.0) 25 (h1.o) 61 Neutral he (62.3) 29 (37.7) 77 Irrelevant 122 (6h.5) 67 (35.5) 189 X“ .8, df '33 p>o804 50 assumed that variation in the accuracy of mediated attitude predictions must have been responsible for the variation found in Table )4. However, this cannot be tested directly by examining directional accuracy of med- iated attitude predictions, because irrelevant mediated attitudes led to predictions of no change. m gf Attitude M.—Relative accuracy of the two predictors appears to be independent of the fact that some attitude objects were sources and others were concepts (see Table 6). $123 if Assertion.--As can be seen from Table 7, there were no dif- ferences in relative accuracy produced by positive and negative asser- tions. Source.--Relative accuracy was no different for change situations in TABLE 6.--Relative Accuracy of Predictors for Types of Attitude Object TIPE 0F TIMES CONGRUIT! TIMES MED. ATT. OBJECT MORE ACCURATE MORE ACCURATE TOTAL Source 81 (311.2%) 156 (65.8%) 237 Concept 80 (317.5) 152 (65.5) 232 X2 " .005, df - 1; p) .90 TABIE 7.--Relative Accuracy of Predictors for Types of Assertion ms 0F TIMES CONCRUITI TIMES MED. ATT. ASSERTION MORE ACCURATE MORE ACCURATE TOTAL Positive 81 (317.6%) 153 (65.1.58) 23!. Negative 80 (3h.0) 155 (66.0) 235 x2 - .018, df - 1; p>.90 TABLE 8.--Relative Accuracy of Predictors for Values of Source VALUE OF TIMES CONCRUITI TIMES MED. ATT. SOURCE MORE ACCURATE MORE ACCURATE 2913; Positive us (28.8%) 111 (71.2%) 156 Negative 61 (38.11) 98 (61.6) 159 Neutral 55 (35.7) 99 (68.3) 15h X2 - 3.11,, df - 2; P> .18 VALUE OF CONCEPT Positive 147 (30.3) 108 (69.7) 155 Negative 63 (110.1) 9h (59.9) 157 Neutral 51 (32.5) 106 (67.5) 157 and for Values of Concept x2 - 3.6, df - 2; p>.16 which the source was positive, negative, or neutral (see Table 8). Concgtw-Table 8 also indicates that relative accuracy was no dif- ferent for change situations in which the concept was positive, negative, or neutral. Attitude Mw-Though no differences were shown in Table 8, it may be noted that the pattern of differences was the same, with a rela- tively greater strength of congruity predictions and/or weakness of med- iated attitude predictions occurring both when sources were negative and when concepts were negative. Accordingly, we may ask if it makes any difference whether attitude objects, considered individually, were posi- tive, negative, or neutral. In each of the invo parts of Table 8, the value of only one of the two objects in each change situation was 52 TABLE 9.--Relative Accuracy of Predictors for Values of Attitude Object VALUE OF TIMES CONGRUITY TEES MED. ATT. OBJECT MORE ACCURATE MORE ACCURATE 3035 Positive ’48 (31.0%) 107 (69.0%) 155 Negative 71 05-8) 8’4 (51k?) 155 Neutral A2 (26.1.) 117 (73.6) 159 x2 - 1h.3, df - 2; p< .001 manipulated, but data was reported for both objects. In Table 9 each attitude object is considered individually according to value. Table 9 indicates that relative accuracies of the two predictors were not inde- pendent of this variable. Relative accuracy in the case of negative at- titude objects was considerably different from that for positive and men- tral objects. This rejection of independence may be examined further by asking whether it occurred because of a variation in the predictive abil- ity of either or both predictors. Tables 10 and ll were used to examine such variation. Table 10 shows that the variation shown in Table 9 is not attributable to varia- tion in the predictive accm'acy of the congruity predictor. Table 11 shows a nearly sipificant variation (p < .052) in the predictive ability of the mediated attitude predictions. However, this tendmcy was in the, opposite direction to that shown in Table 9. In Table 9 the mediated at- titude predictions were shown to have a relative weakness in handling negative attitude objects, while in Table 11 they are seen to be rela- tively strong for negative attitude objects. Table 11, however, did not encompass the effects of mediated attitude predictions made when the 53 TABLE lO.--Directiona1 Accuracy of Congruity Predictor for Values of Attitude Object VALUE OF TIMES CONGRUITI TD’ES CONGRUITI OBJECT RIGHT DIRECTION WRONG DIRECTION TOTAL P08115179 82 (67.8%) 39 (32.2%) 121 Negative 9? (66.11) 11.9 (33.6) 1116 Neutral 79 (56.8) 60 (83.0) 139 x2 - n.2, dr - 2; p>.lO TABLE ll.--Directional Accuracy of Mediated Attitude Predictor for Values of Attitude Object VALUE OF TIMES MED. ATT. TIMES MED. ATT. OBJECT RIGHT DIRECTION WRONG DIRECTION TELL Positive 57 (83.85) 11 (16.2%) 68 Negative 65 (87.8) 9 (12.2) 7h Neutral 57 (73.1) 21 (26.9) 78 X2 . 509, d! - 2; p< e052 mediatedattitude was irrelevant. It is thus possible that the results of Table 9 may be attributed to the effects of irrelevant mediated atti- tudes. To explore this possibility, Table 12 examines the relative ac- curacies of the two predictors across the three values of the attitude object for cases when the mediated attitude is irrelevant In other words, Table 12 takes these cases out of the data in Table 9 and examines them separately. By finding a significant difference, Table 12 suggests that the variation of Table 9 is attributable largely to cases in which the mediated attitude is irrelevant. To help confirm this, Table 13 Sh TABLE l2.--Relative Accuracy of Predictors for Values of Attitude Object ‘When Mediated Attitude is Irrelevant VALUE OF TEES CDNGRUITY TIMES MED. ATT. OBJECT MORE ACCURATE MORE ACCURATE MAE Positive 22 (31.9%) h? (68.1%) 69 Negative 1:3 (57.3) 32 (112.7) 75 Neutral 22 (30.6) 50 (69.11) 72 x2 - 13.9, df - 2; p< .001 TABLE 13.--Relative Accuracy of Predictors for Values of Attitude Object When Mediated Attitude has Value other than Irrelevant VALUE OF TIMES CONGRUITY TIMES MED. ATT. OBJECT MORE ACCURATE MORE ACCURATE 93g. Positive 26 (30.2%) 60 (69.8%) 86 Negative 28 (35.0) 52 (65.0) 80 Neutral 20 (23.0) 67 (77.0) 87 x2 - 2.9, df - 2; p >.20 presents the data remaining in Table 9---in other words, Table 13 pre- sents the data of Table 9minus the data examined in Table 12. The data of Tables 12 and 13 suggest that the variation found in Table 9 came largely from effects occurring when mediated attitudes were irrelevant. Table 12 also suggests that the weakness which occurred when mediated at- titudes were irrelevant (see Table 14), occurred only when the attitude object was negative. In other words, the existence of cases in which mediated attitude predictions were relatively weak was confined large}; SS attitude was irrelevant. A breakdown of the mean number of scale units of change in cases having irrelevant mediated attitudes supports this statement. The mean change fOr all such cases was .71. This figure may be broken down by value Of attitude Object to a mean change of 1.00 units, .73 units, and .37 units, when the attitude object was, respec- tively, negative, neutral, and positive. In summary, then, the significant difference reported in Table 9 and the near-significant difference reported in Table h may be attributed to the relatively higher incidence of change for negative attitude objects than fOr positive or neutral.objects in cases where the mediated attitude was irrelevant. 2222M2£ Situation.-JWas relative accuracy of the predictors inde- pendent of whether the experimental situation was congruent, incongruent, or "other"? .According to Table 1h, relative accuracy was not.independent of this variable. The mediated attitude predictions were relatively strongest for congruent and "other" situations, and relatively weakest fer incongruent situations. These results may be pursued further by ask- ing whether they occurred because of variations in one or the other or both predictors. Tables 15 and 16 were used to examine for such varia- tion. Table 15 shows that the accuracy of congruity predictions was not independent of the three types of situations. Congruity predictions ‘were extremely strong for incongruent situations, less accurate for "other" situations, and no better than chance for congruent situations. Table 16 suggests that the accuracy of mediated attitude predictions was independent of the three types of situations. Real-life Situation.-AWas the relative accuracy of predictors 56 TABLE lh.--Relative Accuracy of Predictors for Types of Situations TIPE OF TIMES CONGRUITY TIMES MED. ATT. SITUATION MORE ACCURATE MORE ACCURATE w Congruent 28 (27.7%) 73 (72.3%) 101 Incongruent 1&7 (MAJ) 59 (55.7) 106 Other 86 (32.8) 176 (67.2) 262 x2 - 6.9, Of - 2; p< .0; TABLE 15 .--Directional Accuracy of Congruity Predictor for Types of Situations TIPE 0F TIMES GONGRUITY TIMES GONGRUITY SITUATION RIGHT DIRECTION WRONG DIRECTION MAE Congruent ’43 (51.2%) hl (h8.8) 8h Incongruent 73 (76.0) 23 (211.0) 96 Other 1A2 (62.8) 81; (37.2) 226 x2 - 12.5, df - 2; p< .005 TABLE l6.--Directional Accuracy of Mediated Attitude Predictor for Types of Situations TYPE OF TIMES MED. ATT. TDIES MED. ATT. SITUATION RIGHT DIRECTICN WRONG DIRECTION _T9_’I.'_A_]_2_. Congruent m; (89.8%) s (10.2%) h9 Incongruent ’49 (811.5) 9 (15.5) 58 Other 36 (76.1) 27 (23.9) 113 x2 . 1.107, df - 23 p) .09 57 independent of whether the situation was conceivable and the same in real life, conceivable but not the same in real life, or inconceivable in real life? On the basis of Table 17, a hypothesis of independence can- not be rejected. TABLE l7.--Re1ative Accuracy'of’Predictors for Types of Realvlife Situations REAL-LIFE TIMES CONGRUIT! TIMES MED. ATT. SITUATION MORE ACCURATE MORE ACCURATE TOTAL Conceivable-u- 18 (28.6%) 1:5 (71.14%) 63 same Conceivablen- 5h (36.2) 95 (63.8) 11.9 not same Inconceivabls 25 (32.1) 53 (67.9) 78 x?- - 1.1, df - 2; p) .50 Tables 18 and 19 test whether or not the "type of situation" var- iable and the "attitude object" variable varied across the values or the "real-life" variable. Table 18 tests the null hypothesis that the rela- tive accuracy of the two predictors is independent of the ”type of sit- uation" variable for, in turn, conceivable--same, conceivable—not same, and inconceivable situations. Table 18 shows that the dependence found in Table 11; may be confirmed for both types of conceivable situations, but not for inconceivable situations. Table 19 tests the null hypothesis that relative accuracy of the predictors is independent of the "atti- tude obj ect" variable for, in turn, conceivable-mama, conceivable-mot same, and inconceivable situations. Table 19 shows that the dependence team in Table 9 may not be continued on an individual basis for 58 TABLE 18.-«Relative Accuracy of Predictors for Types of Situation for Conceivable--same Situations, for Conceivable--not same Situations, and for Inconceivable Situations TYPE OF TIMES CDNGRUITY TIMES TED. ATT. SITUATION MORE ACCURATE MORE ACCURATE TOTAL Congruent 9 (37.5%) 15 (62.5%) 214 CONCEIVABIE --SAME Incongruent 6 (50.0) 6 (50.0) 12 SITUATIONS Other 3 (11.1) 21; (88.9) 27 X2 - 7.7, or - 2; p< .05 Congruent h (17.h) 19 (82.6) 23 CONCEIVABIE -NOT SAME Incongruent 19 (51.14) 18 (148.6) 37 SITUATIONS Other 31 (311.8) 58 (65.2) 89 X2 - 7.2, d! - 2; p< .05 Congruent 6 (33.3) 12 (66.7) 18 INCONCEIVABIE Incongrth 3 (21.10 11 (78.6) 1).; SITUATIONS Other 16 (3h.8) 30 (65.2) N6 X2 - .9, df - 2; p) .50 conceivable-mama, conceivable-mot same, and inconceivable situations. @en and Closed Minds.--Table 20 shows that relative accuracy of predictors was independent of openmindedness and Closedmindedness. 59 TABLE 19.-Re]ative Accuracy of Predictors for Values of Attitude Object for Conceivabls--same Situations, for Conceivable-—not same Situations, and for Inconceivable Situations VALUE OF TD’EES CONGRUITY TD’EES MED. ATT. OBJECT MORE ACCURATE MORE ACCURATE TOTAL Positive 5 (26.3%) 111 (73.7%) 19 CONCENABIE -SAME Negative 9 (33.3) 18 (66.7) 27 SITIIATIONS Neutral h (23.5) 13 (76.5) 17 )(2 8:- .6, d1 - 2; p > .70 Positive 20 (37.0) 31. (63.0) 51. commune «410T SAME Negative 20 ((47.6) 22 (52.11) h? SITUATIWS Neutral 11; (26.14) 39 (73.6) 53 X2 ‘ (4.6, d: - 2; p > 009 Positive 7 (28.0) 18 (72.0) 25 mooncmvmm Negative 10 (38.5) 16 (61.5) 26 SITUATIONS ‘ Neutral 8 (29.6) 19 (70.10 27 x2 .102, “.23 p>050 60 TABLE 20.--Relative Accuracy of Predictors for Open and Closed Subjects TIPE 0F TIMES CONORUITI TIMES MED. ATT. W MORE ACCURATE MORE ACCURATE m3 Open 30 (81:77) 1.2 (58.3%) 72 Closed 25 (31.6) 51. (68.11) 79 X2 - 1.7, df - l; p>.15 TIMES CONCRUITI TIMES OONORUITT RIGHT DIRECTION WRONG DIRECTION Egg. Open “4 (69.8) 19 (30.2) 63 Closed h5 (63.10 26 (36.6) 71 x2 - .6, or - 1; p >.h0 TIMES MED. ATT. TIMES MED. ATT. RIGHT DIRECTION WRONG DIRECTION 939;. Open 26 (72.2) 10 (27.8) 36 Closed 25 (86.2) h (13.8) 29 X2 '- 1.9, d! - 1; p >.10 CHAPTER IV This chapter includes the conclusions of the research, discussion, and implications fOr further research. CONCLUSIONS Mediated attitude predictions of change were significantly more ac- curate than congruity predictions in situations involving immediate atti- tude objects. Mediated attitude predictions were more accurate under all experi- mental conditions. However, the extent of their superiority was related to the Operation of certain situational variables. In situations described. by subjects as conceivable in real life, the relative superiority of mediated attitude predictions was greatest under congruent experimental conditions, less so for "other" conditions, and least under incongruent conditiOns. In situations described. as incon- ceivable in real life, the relative superiority of mediated attitude pre- dictions didwnot vary under congruent, incongruent, and "other" con- 'ditions. The relative superiority of mediated attitude predictions was great»- est when attitude objects were positive and neutral, and least when at- titude objects were negative. DISCUSSION We have examined the notion that non-congruity factors are relevant 61 62 to change under certain conditions. The present research was designed to test this hypothesis empirically and to specify a type of condition un- der which non-congruity factors may be expected to affect change. The comparison of mediated attitude predictions and congruity pre- dictions was intended to demonstrate the existence of non-congruity fac- tors among those factors relevant to attitude change. Differences be- tween the two predictions occurred only when the location.of congruence and the mediated attitude had different values. location of congruence and mediated attitude are names for alternate conceptions of an inter- vening variable which mediates attitude change. The value of this var- iable is the directionand intensity which the subject believes his at- titude should have in order that it may be congruent with certain new (or newlyhrelevant) information. The congruity principle calls the variable the location of congruence, and states that its value is established by the values of two congruity factors. The mediated attitude principle calls the variable the mediated attitude, and states that its value is not necessarily dependent on the two congruity factors and should be measured by asking the subdect what it is. If the obtained value of the mediated attitude was no different than that of the location of congruence, then we would conclude that the sub- Ject determined the mediated attitude on the basis of the two congruity factors only. If the mediated attitude and location of congruence varied in value, we would conclude that the mediated attitude was determined on the basis of nonpcongruity as well as congruity factors. Furthermore, if mediated attitude predictions were more accurate than congruity pre- dictions, then we would conclude that the non-congruity factors were 63 relevant to determining attitude change. Finally, if non-congruity factors were relevant to attitude change, then we would conclude that the distinction between immediate and mediate attitude objects is significant in attitude research. In the Tannenbaum experiment, involving mediate attitude objects, the accuracy of congru- ity predictions was extremely high. It is unlikely that mediated atti- tude predictions could have been more accurate. From this, we conclude that changes in mediate attitude objects were independent of non-congru- ity factors. In the present study, however, if mediated attitude predic- tions were more accurate than congruity predictions, we would conclude that immediate attitude objects were related to non-congruity factors. Immediate and mediate objects would then be contrasted in toms of their association or non-association with non-congruity factors. The experimental data in the present study show that the values of the mediated attitude and location of congruence differed in 1169 out of 50!; cases. They also show that the mediated attitude principle was sig- nificantly more accurate in predicting change than was the congruity principle. hon these results, we conclude that non-congruity factors were rel- evant to attitude change in situations involving immediate attitude ob- jects. We also conclude that situations using immediate attitude objects may be contrasted with situations using mediate attitude objects because non-congruity factors apparently were not relevant to change for the latter. Although the mediated attitude predictions were more accurate than congruity predictions under all experimental conditions, the extent of 6h this superiority appeared to be related to the operation of certain var- iables. This meant that in situations involving immediate attitude ob— jects, knowledge of non-congruity factors was more important to predic- tive accuracy under certain conditions than under others. Non-congruity factors were more important under congruent than under incongruent conditions. The relative superiority of mediated attitude predictions was greatest under congruent conditions, less so for "other" conditions, and least under incongruent conditions. This happened be- cause congruity predictions were most accurate under incongruent condi- tions, were less so for "other" conditions, and were not more accurate than chance under congruent conditions. The greater the incongruity, the stronger the congruity predictions. This effect seems reasonable under the assumption that what is comonly called the I'congruity" phe- nomenon is actually a characteristic of incongruity. An incongruent situation is one in which the characteristic is present; a congruent situation is one in which it is absent. In a congruent situation, any non-congruity factors which are present may dominate as change agents; in incongruent situations this is less likely to happen. The congruity predictor thus operated variably for the different kinds of situations.” The mediated attitude predictor, intended to account for all rele- vant variables, operated at the same strength in congruent, incongruent, and "other" situations. This indicated that the presence or absence of 291t might be asked how the congruity principle predicted accurate- ly for congruent situations in the Tannenbaum experiment. The answer probably involves the fact that most so-called congruent situations have a small amount of incongruity due to differences in intensities. This, plus the absence of any other relevant factors, would enable the con- gruity principle to predict accurately. 6S incongruity did not affect predictions based on all relevant change factors. The variable operation of the congruity predictor across congruent, incongruent, and “other" situations applied only for situations which subjects reported as conceivable in real life. Apparently subjects went through a process of discounting the factor of incongruity when the sit- uation was inconceivable. They convinced themselves that the incongruity was too unimaginable to be considered as a relevant factor. Consequent- ly, under "inconceivable" conditions the Osgood and Tannenbaum predic- tions were no stronger when incongruity was present than when it was not. Mediated attitude predictions showed considerably less superiority whal attitude objects were negative. This result is attributed to an effect which occurred only when the attitude obj ect was negative and the mediated attitude was "irrelevant" (the effect may be called the "irrel- evant effect"). This effect was produced by subjects who failed. to re- flect their evaluation of a situation as "irrelevant." If a mediated attitude was irrelevant, then no change should have occurred in the orig- inal attitude. Instead, considerable change occurred, amounting to a mean of 1.00 scale units (absolute change) per case. When mediated at- titudes toward positive and neutral objects were called "irrelevant, " the change was only .37 and .73 units, respectively. The airrelevant effect" occurred only for mediated attitude predic- tions for negative objects when the mediated attitude was "irrelevant. " It did not occur for mediated attitude predictions for negative objects when the mediated attitude held any other value, and it did not occur for congruity predictions for negative objects. 66 The cause of the "irrelevant effect" may not be traced to relevant change factors, at least not those of which subjects were conscious. A check of written comments by subjects shows that they had no idea that the effect was occurring. For exaxnple, a subject who recorded 1.? mean units of attitude change per scale indicated that the new information in- troduced in the experiment was "irrelevant," and wrote that it "...would make very little difference..." to his attitude. The "irrelevant effect" may be traced to two characteristics in subjects' handling of attitude objects: 1. (They responded unconsciously to relevant factors. Many sub- jects, upon recording mediated attitudes, probably did not bring to con- sciousness (either through laziness or through some more profound inabil- ity) all of the factors affecting their post-experimental attitude toward the subject. Consequently, some subjects probably stated that the mediated attitude was ”irrelevant" when it was not. If, however, a subject recorded a not-"irrelevant" mediated attitude, then the uncon- scious factors probably affected both the mediated attitude and the final attitude in parallel fashion. Because of this, the mediated attitude predictions were weakened if the subject recorded an "irrelevanfl'med- iated attitude, but not if he recorded a positive, negative, or neutral mediated attitude. 2. They made deliberately erroneous responses. The questionnaire instructed the subject who recorded an "irrelevant“ mediated attitude to skip the additional task of responding to six semantic differential scales. Probably some deliberately inaccurate responses of "irrelevant" were made for this purpose. 67 Both these kinds of mishandling produced erroneous responses of "irrelevant." ‘We have seen, however, that these errors weakened mediated attitude predictions only when the attitude object was negative. The reason for this is that these errors were more crucial to predictive ac- curacy when they occurred for negative attitude objects because the neg- ative objects underwent greater change than other objects, as reported above. The extreme amounts of change by negative objects were accounted for by congruity'predictions. The negative objects in the present experi- ment had lower average intensities than the positive objects. 'When they ‘were matched with the positive objects congruity theory'predicted.more pressure for the negative objects to change because magnitude of change is inversely proportional to intensities. Thus the greater charges by negative objects were paralleled by congruity predictions for greater change. In summary, then, the "irrelevant effect" was produced by subjects' errors. These errors did not affect congruity predictions. NOr did they affect mediated attitude predictions except when the mediated attitude was "irrelevant." In the latter case, they affected predictive accuracy to an observable extent only for negative attitude objects because of the considerably higher average change of negative objects. Errors in mediated attitude predictions did not occur only in the case of "irrelevant" mediated attitudes. In the cases of positive, neg- ative, and neutral.mediated attitudes, the resulting predictions were in the wrong direction in hl.out of 220 cases. "written comments by the subjects involved suggest that: 68 1. .At least ten subjects read the instructions for the mediated attitude incorrectly; The use of the instrument, or else the instrument itself, was faulty. These errors were relevant to the data collection process rather than to the change situation. A check of written comments of all subjects suggests that such errors were not numerous in compar- ison with those cases in which the instrument was handled acceptably. 2. At least 21 subjects showed faulty recall of the original atti- tude. ‘While handling Part Two they were making incorrect assumptions about the attitude which they had recorded in Part One. Consequently, the responses they'made in Part Two did not accurately reflect their in- tended treatment of the attitude object. .Again, written comments of all subjects indicated that in.most cases the responses in Part Two were made on the basis of adequate recall of the original attitude, and were in accord with subjects' intentions. As a result of the various factors discussed above, mediated atti- tude predictions were limited in their predictive ability. Confirmation of the hypothesis in spite of these limitations, however, stresses more strongly the greater predictive ability of all relevant factors as com- pared to congruity factors only. Relative accuracy of the predictors apparently was independent of differences between openminded and.closedminded persons. Directional accuracy of each predictor, taken individually, also was independent of. these differences. The failure to reject these hypotheses of independence is attributed to the fact that congruity effects are less evident in situa- tions involving immediate rather than mediate attitude objects. It was suggested earlier that the two types of persons should be expected to react 69 differently to congruity factors. Greater reduction of inconsistency was predicted for closedminded than openminded persons. This would mean that the mediated attitude predictor would have a greater relative superiority for Openminded than for closedminded subjeCts. However, it was also predicted that congruity effects would be less evident in a study which replaced mediated attitude objects with.immediate objects. With con- gruity effects less evident, the predicted difference between Openminded and closedminded persons would be less likely to occur. IMPLICATIONS AND SUGGESTIONS FOR FURTHER RESEARCH The findings suggest the congruity predictions must be supplemented by information concerning non-congruity factors for at least one importalt kind of attitude change situation. The laboratory situation produced by Tannenbaum has significant parallels in everyday attempts to communicate, influence, and change at- titudes. This is also the case for the situation produced in the present study. The two are complementary. The Tannenbaum experiment suggests the indirect public approach of the public news media. The present experiment suggests the kind of direct interpersonal contacts which Katz and Lazarsfeld have designated as the second step of the "two-step flow" of communication.30 Perhaps the distinction already established between these two steps can be enriched by an understanding of the relative dominance of congruity effects within each. It might be suggested, fer exanple, that a certain congruity 30ElihuKatz and Paul F. Lazarsfled, Personal Influence (Glencoe, Illinois: Free Press, 1956). 70 effect may be expected to "flow" with great effectiveness through the public media portion of a communication network, but will get less effective dissemination through the lines of personal communication. Problems of "source credibility" may be subject to interpretation, also, in terms of the relative dominance of congruity factors. In a typical study,31 speakers presenting identical messages to each of three experimental groups were introduced in three different ways which identified them as respected, neutral, and disrespected sources. The three groups altered their attitudes differentially toward the object of the message, the "positive introduction" group making the greatest change in the direction advocated by the message. This result is easily interpretable as a.congruity effect. If the congruity effect was less dominant, however, the significant differences among the groups may not have occurred. The mediate-immediate distinction may provide an Opportunity for re-interpretation of source credibility studies in which the hypothesized differences in attitude change did not occur. The problem of identifying the non-congruity factors which affected the present experiment has not been pursued here, but the experiment has ob- tained some data which are relevant to the isolation of such factors. Subjects"written comments suggest specific factors as the cause of ob- served changes. Two of the comments will be cited here. The first one stated: I gainacertain admiration of B and pride because she has been recognized by'A who is a highly respected and well known individual even though I don't care for him. The questionnaire had suggested, in this case, that.A, originally rated negatively, had a high regard for B, originally rated positively. :The sub- 31l-Ierbert C. Kelman and Carl I. Hovland, "'Reinstatement' of the communicator in.delayed measurement of opinion change," Journal of.Ab- nomal ing Social ngchology, XLVIII (February, 1953), 357-335. "" '— 71 ject indicated that A was "teacher" and B was "closest friend-- same sex." Congruity theory would predict a lowering of attitude toward B due to the incongruity involved. Actual.change was in a favorable direction. A second comment was: Both people are so close that I know there was probably just 'some small understanding that could be ironed out easily. In this case, the attitude objects were both positive and the sug- gested assertion was negative. The subject indicated that the persons were "Mother" and ”Father." Congruity theory would predict a lowering of attitude toward both due to the incongruity involved. Actual change was a mean of zero for both. In the first case, the non-congruity factor involved was that of partitioning an attitude into categories of "respect" and "personal lik- ing." In the second case, the factor was rationalization of the assertion into an event of low and insignificant intensity. A follow-up to the present study, then, would be the isolation of factors such as these. Also, since all such factors might conceivably occur in both the Tannenbaum contextland the context of the present research, it would be meaningful to study the extent to which given factors will occur in each context, and to specify reasons for differences in likelihood of occurrence. The present research has produced unfortunate consequences insofar as it implies that attitude researchers need more complex models. If congruity factors were dominant under all types of situations, explanation and pre- diction could maintain the relative simplicity of the Osgood and Tannenbaum model. If they are not dominant, the researcher must take one of two steps: (1) incorporate other factors into the model; (2) retain the model by re- evaluating the factors on which congruity predictions are based. 72 The mediated attitude principle has been based on the first step. It is a suggested method for incorporating non-congruity factors. How- ever, the writer would like to conclude this report with an.argument im- plying that the second step is also promising. The following remarks are made in support of the conclusion that congruity predictions could be improved, not by altering the congruity principle, but by altering the means of obtaining the basic information from whiCh congruity predictions are derived. Suppose that a subject's attitude toward person.A is +3, generally, and that his attitude toward B is also +3. However, when it comes to con- sidering remarks by'A about B, then his attitude toward A is -3. In other words, he likes A very much, generally, but when A talks about B, the subject feels that A doesn't know what he's talking about. Consequently, when the subject sees A deriding B he finds it perfectly congruent. In the light of this situation, the following question becomes per- tinent: ‘Which of the subject's two attitudes toward A would the research instrument be likely to measure? In the present experiment, the instrument measured the general attitude. Hewever, the other attitude, the one specific to the cases of.A's interaction with B, would have been the proper one to measure, since it would have led to a congruity prediction which assumed a congruent situation. The problem facing the researcher, then, is that of determining the correct item to measure. Another unfortunate aspect of the present research is that the mediated attitude, as presented herein, involvesa method unavailable for prediction 'when the researcher needs it. The mediated attitude was obtained no earlier 73 than the time at which the post-experimental attitude was available. It was thus obtained only when it was no longer needed for prediction. How- ever, this does not preclude the possibility of obtaining a mediated attitude measurement before the fact. The subject could have been asked the following: Please indicate, by making a check-mark next to one of the following statements, whether an expression of a positive attitude toward B bw'A would have any effect on what your own attitude toward B should be: A positive attitude toward B by'A would: 1. Have some effect on what your own attitude toward B should be. 2. Have no effect on what your own attitude toward B should be. The prospect of obtaining a mediated attitude in advance, and the prospect of measuring an original attitude which is specific rather than general, have parallel consequences. In both cases the researcher would be asking the subject to consider the attitude object within a specific rather than a general context. This would mean that the researcher, in order to make a prediction.within a specific context, is obtaining in- formation drawn from that same context. This seems to be a better method than one which employs only general information to predict within specific contexts. If congruity research has made habitual use of the Latter method, then the implication is clear: Attitude change research must be designed in such a way that assures that information upon which predictions are based is no less specific than are the required predictions. BIBLIO GRAPHY Dodge, Joan S. , "A Quantitative Investigation of the Relation Between Meaning Development and Context" (unpublished doctoral disser- tation, University of Illinois, 1955). Festinger, Leon, _A_ Theory 91 Cognitive Dissonance (Evanston: Row, Peterson and Company, 1957?. Harburg, Ernest, and Price, Kenneth, (unpublished. research, University of Michigan, 1960). Heider, Fritz, The Psychology of Interpersonal Relations (New York, Wiley, 19537? Hotelling, Harold, "The Selection of Variates for Use in Prediction, with Some Comments on the General Problem of Nuisance Parameters," Annals of Mathematical Statistics, XI (Fall, 1910), 271-283. Kata, Elihu, and Lazarsfeld, Paul F. , Personal Influence (Glencoe, Illinois: Free Press, 1956). Kellnan, Herbert C., and Hovland, Carl I. , ”'Reinstatement' of the com- municator in delayed measurement of opinion change," Journal of _A_b_- normal and Social P holo , XLVIII (February, 19535, 327-33. Kerrick, Jean 5., (unpublished research, University of California, 1955). , "The Weekly Newspaper as a Source: Prediction of an Editorial's fectiveness" (unpublished research, 1959). Newcomb, Theodore, "An Approach to the Study of Communicative Acts," Psychological Review, 1.1 (November, 1953), 393-1014. Osgood, Charles E. , and Tannenbaum, Percy H., "The Principle of Congruity in the Prediction of Attitude Change," Psychological Review, LXII (January: 1955), (42'55- , Suci, George J., and Tannenbaum, Percy H., The Measurement .93.. Meaning (Urbana: University of Illinois Press, T937). "Cognitive Dynamics in the Conduct of Human Affairs," Public Qpinion Quartery, XXIV (Summer, 1960), 360-365. 71: 75 Powell, Fredric A., "Open- and Closedmindedness and the Ability to Dif- ferentiate Between Source and Message" (unpublished Master's thesis, Michigan State University, 1961). Rokeach, Milton, The Qpen and Closed Mind (New York: Basic Books, 1960). Rosenberg, Milton J .3 Hovland, Carl 1.; McGuire, William J .; Abelson, Robert P.; and Brehm, Jack W., Attitude Organization and Change (New Haven: Yale University Press, 1960). APPBIDIX QUESTIONNAIRE ON ATTITUDES NAME Please write your name in the space above. The questionnaire comes in two portions, to be issued separately, and your name is needed in order to match them. / . y \ . vi” .) (‘2. \ \Igyfi‘qj . “u” . .L“ f“ This Questionnaire will ask you to make ratings ofn'peoplen on sets of scales such as the one shown below. Here is how you are to use the scales. Suppose the person in question is PRESIDENT KENNEDY, and the scale is: FAIR: : : : x : : :UNFAIR T T T T T 32' :3" The spaces between FAIR and UNFAIR have the following meanings: VERY FAIR QUITE SLIGHTLY NEITHER FAIR NOR SLIGHTII QUITE VERY FAIR FAIR UNFAIR--m SCALE UNFAIR UNFAIR UNFAIR DOES NOT APPLY Therefore, if you feel that the person (PRESIDENT KENNEDY) is vs? closey related to one end of the scale, you should place your check-mark as o ows: FAIR:_}_:__:__:__:___:___:__:UI~IFAIR 01‘ FAIR:__: : : : :___:__1_£__: UNFAIR If you feel the concept is uite close related to one or the other end of the scale, you should place Wcmie‘xam: FAIR:___: __1_{_ :____:___:____:___:__:UNFAIR or FAIR: : : : : :__X__:____:UNFAIR If the concept seems only slightly related to one side or the other: FAIR: : : X : : :____:_____:UNFAIR or ' FAIR:__:___:__:___:_§£_:__:__:UWAIR Finally, if you consider the concept to be neutral on the scales, or if the scale is completely irrelevant, then place your check-mark in the middle space. D’iPORTANI‘: Place check-mrks in the middle of the space, not on the lines. Place one (and only one) check-mark on each scale. -2- 75? .. . The first step of the questionnaire is to choose the names of some persons. We are going to ask you to rate these persons on scales such as those described above. We will use the ratings in our research, but the actual names of the persons are not important to us; You may disguise the persons by using only their initials to refer to them. If you like, you may reverse the initials or use other code symbols in order to maintain complete secrecy. However, please be careful to choose symbols which you will be able to identify at a later time. The persons you choose must be real persons and you must know uite a lot about than. You must know enough about them to have definite attitu es or feelings toward theme It is necessary that you know them through your own personal contact with them. Your knowledge of them should be based principally upon your personal observations. The persons you choose must be selected from the following descriptions: Father or mother Brother or sister Spouse or closest friend of opposite sex Closest friend of same sex Roommate or neighbor i Employer ' Co-worker Teacher Doctor Clergyman Administrator (indicate which kind) Cbach Officer of organization to which you belong Attendant or clerk with whom.you come in frequent contact Authority on some subject in which you are interested (indicate the subject) .....or any other descriptions which in a similar way show the relationship the other personshas toward you. You will be asked to choose six persons. We will call them Persons 1, 2, 3, h, 5, and 6. Fill in the blanks with the symbols you choose to represent than. Also fill in their descriptions. For example: Description A ”:2“ ("1 Ii) * Person 1 ' L/- ”\QQW‘J‘ .3 3, No individual may be chosen more than once, and none of the above descriptions my be used more than once. If you have any questions about making your choices, please ask the instructor before you continue. Your first two choices will be called Person 1 and Person 2. They should be per- sons for whom you have a high regard. The nature of your regard need not be of amr specific kind. It may be "liking" or "respect" or "admiration" or "enjoyment“ or amr other kind of‘positive attitude, depending on the way in which you happen to know the persons. Please designate Persons 1 and 2 by filling in the blanks below with symbols which you choose to represent them. Also fill in their descriptions: Description Person 1 -3- 78 Now indicate how you feel about these two persons by rating them on the following scales according to instructions. . Rate Person 1 on the following set of scales. Work rapidly. Record your first impressions only: LIKE: : : : : : :__:DISLIKE ~“u__~ PLEASIM}: : : : : :___:___:ANNOYING FAIR:____:___: : : : ___:UNFAIR ALTRUISTIC :___: : : : : :EGOTI STIC HARIIDNICIJS: : : : : : : :DISSONANT GRATEFUL: : : : : : : :UNGRATEFUL LIKE: : :fi:__:__:___: :DISLIKE PLEASING: : : °___:__:___:___:ANNOYING FAIR: : : : : :__:____:UNFAIR ALTRUISTIC: : : : : : : : EGOTISTIG MARI-IONIOUS: : : : : :___:___:DISSONANT GRATEFUL:___: : : : :__:__:UNGRATEFUL Persons 3 and 1; should be persons for whom you have a low regard. Again, the nature of your regard need not be of a specific kind. It may, for instance, be "disliking" or "disrespect" or "disapproval" or "displeasure" or amr other kind of negative attimdee ’ . . Persons 3 and h matte chosen with reference to the descriptions listed above. No individual nor-description may be used more than one time. Please designate Persons 3 and h by filling in the blanks with the symbols you choose to represent them. Also fill in their descriptions: Description Person 3 Person 1: Rate Person 3 on the following set of scales. Work rapidly. Record your first inpressions only: PLEASING:_:__: : : :___:__:ANNOYING FAIR:___: : : : : :___: UNFAIR ALTRUISTIO:____:___: : : : :____:EGOTISTIC HARMONIOUS: : : : : : : :DISSONANT GRATEFUL: : : : : : : :UNGRATEFUL LIKE: : : : : : : :DISLIKE PLEASING:__:__:__:__:__:___:___:ANNCYING FAIR:___:___: : : : : :UNFAIR ALTRUISTIC:___: ____: : : :_____:____: EGOI‘ISTIC HARMONIOUS :____:_____: : : :____:____: DI SSONANT GRATEFUL:___:____: : : : . :UNGRATEFUL Persons 5 and 6 should be persons for whom you have a neutral regard. This, too, may be defined in several ways. It may be someone toward whom you are almost com- pletely indifferent, even though you know a good bit about them. It may be someone toward whom.you have both high and low regard at the same.thme. It may be someone who has very little importance for you. It may be someone toward whom you don't really know how you feel, or else toward whom you don't care how you feel. Persons 5 and 6 must be chosen with reference to the descriptions listed above. No individual nor description may be used more than one time. Please designate Persons 5 and 6 by filling in the blanks with the symbols you choose to represent them. Also fill in their descriptions: Deacription .‘ Person 5 Person 6 Rate PerSOn 5 on the following set of scales. Work rapidly. Record your first impressions only: LIKE:____:___: : : :__:____: DISLIKE PLEASING:___: : : : :___:___:ANNOYING FAIR:___:___: : : : :___:UNFAIR ALTRUISTIC :___: : : : : :____: EGOTISTI C HARMONIOUS: : : : : : : :DISSONANT LIE: : : : : : : :DISLIKE PIEASING: : : : : : : :ANNOYING 80 FAIR:___§ : : : : : :UNFAIR You are now finished with the first part of the questionnaire. The second part will be issued to you at a later time. Before you turn in this part, please look carefully at the initials and descrip- tions of the six persons you have chosen. Make sure you will remember correctly the persons referred to when you see their initials and descriptions again. If you think you will have trouble remembering, record the information in your class notes. 81 QUESTIOt-UIAIRE ow ATTITUDES PART TWO MAME In the second part of the Questionnaire, you will consider four of the persons chosen in the first part. The symbols and descriptions for these persons are recorded below. In the meterial that follows, these persons will be referred to as Persons A, B, C, and D. Whenever you see "Person A" being discussed, think of the real person whose symbol and description are listed after "Person A" in the following list: mpg]; Description Person A Person B Person c Person D In the followin mg we are going to talk about Persons A and B. We are going to sug- gest to you a s tuation which involves them. Please think of it as a real situa- tion, because we are going to ask you to consider the way in which it would affect your feelings if it actually happened. We want you to imagine that Person A has expressed a regard for Person B. You are very certain that A has this feeling toward B. You have been present when he has personally elqaressed this feeling, and you have seen him showing it in the way he acts toward B. We are studying the problem of whether or not one person's feelings about another have anything to do with what your own feelings toward both should be. There have been some suggestions that people differ in the way they react to such information. Some peOple would say that one person's feelings about another should influence the way in which they themselves should feel toward both persons. Other peeple would say that such feelings should not matter as far as their own feelings are concerned. Still others would say it works differently in different cases, depending on the individuals being considered. In order to help us study this problem, please make ratings of A and B based on all possible considerations that have ary reference to them. Use all your old in- formation about them plus the new information we have given you. Please treat the new information as fact even though it may conflict with what you believe is actually the case. Rate Person A on the following scales: Work rapidly. Record first inpressions only: PLEASING: : : : : : : :ANNOYING ALTRUISTIC: : : : : :___:____:EX}OTISTIC GRATEF‘UL:__: : : : : :___: UNGRATEFUL Rate Person B on the following scales: PLEASING:___:__: : : : ALTRUISTIC:__:__: : : :__:___:EGOTISTIC ___:ANNOYING GRATEFUL: : : : : :___:___:UNGRATEFUL LIKE: : : : : : : : DI SLIKE HARMONIOUS: : : : : : : :DISSONANT E 5’ ____:UNFAIR Now we want to ask you something more about the event we have suggested: the regard of A toward B. Remenber that we want you to treat this as a real event: Please indicate, by making a check-mark next to one of the following statements, whether A's attitude toward B has some effect on what your own attitude toward B should be---or has no such effect. Please think very carefully about this. Try to think of any possible way in which A's attitude may have some effect on your own. Do not specify "no effect" unless you are confident that none exists: A's attitude toward B: 1. Has some effect on what your own attitude toward B should be. 2. Has no effect on wlnt your own attitude toward B should be. If you have checked statement 1, please record below the ratings of B which you feel would be in perfect accord with your knowledge of A's attitude. Do not state what your ratings of B actually have been in the past, or are now, or will be in the future. Instead, state the ratings of B which you would make if you were rating B on the basis of this one fact only, with all other considerations elim- inated: (If you have checked statement 2, do not do these ratings) PLEASDJG: : : : : : :___:AI‘ENOYING ALTRUISTI C: : : : : : : : EGOTISTIC GRATEFUL: : : : : :___:___: UNGRATEFUL IIKE: : : : : : :__:DISLIKE HARICNIOUS:_'__:__: : : :___:___:DISSONANT FAIR:___:__:___:__:_:__:___:UNFAIR Regardless of which statement you checked, please write a sentence or two explain- ing why the attitude in question had the effect it did concerning B: 83 -8- In reference now to Person A, please indicate, by making a check-mark next to one of the following statements, whether A's attitude toward B has some effect on what your own attitude toward A should be---or has no such effect. Please think very carefully about this. Try to think of any possible way in which A's attitude may have some effect on.your own. Do not specify "no effect" unless you are confident that none exists: A's attitude toward B: 1. Has some effect on what your own attitude toward A should be. 2. Has no effect on what your own attitude toward A should be. If you have checked statement 1, please record below the ratings of A which you feel would be in perfect accord with your knowledge of A's attitude. Do not state what your ratings of A actually have been in the past, or are now, or will be in the future. Instead, state the ratings of A which you would make if you were rating A on the basis of this one fact only, with all other considerations elim- inated: (If you have checked statement 2, do not do these ratings) PLEASING: : : : : : : :ANNOYING ALTRUISTIC : : : : : :____: ____: MOTISTIC QATEFUL:_:__: : : :___:__:UNGRATEFUL LIKE : : : : : : : :DISLIKE HARMONIOUS: : : : : : :__:DISSONANT FAIR: : : : : : : :UNFAIR Regardless of which statement you checked, please write a sentence or two explain- ing why the attitude in question had the effect it did concerning A: Please indicate whether the situation (A's attitude toward B) that we have suggested: 1. Could conceivably have happened 2. Could not conceivably have happened 3. Don' t know Please indicate what attitude actually is held by A toward B: 1. Positive attitude 2. Neutral or indifferent attitude 3. Negative attitude 1:. No attitude 5. Don't know ~ 8h '9‘ we have now finished with our consideration of Persons A and B. we are now going to discuss Persons 0 and D in the same manner, suggesting a situation involving them and asking you the same kinds of questions about that situation. Please think of it as a real situation, because we are going to ask you to consider the way in which it would affect your feelings if it actually happened. we want you to imagine that Person C has expressed a regard for Person D. You are very certain that C has this feeling toward D. You have been present when he has personally expressed this feeling, and you have seen him showing it in the way he acts toward D. we are studying the prOblem of whether or not one person's feelings about another have anything to do with what your own feelings toward both should be. There have been some suggestions that peeple differ in the way they react to such information. Some people would say that one person's feelings about another should influence the way in which they themselves should feel toward both persons. Other people would say that such feelings should not matter as far as their own feelings are concerned. Still others would say that it works differently in different cases, depending on the individuals being considered. In order to help us study this problem, please make ratings of C and D based on all possible considerations that have any reference to them. Use all your old in» formation about them plus the new information we have given you. Please treat the new information as fact even though it may conflict with what you believe is actually the case. Rate Person C on the following scales. Wbrk rapidly. Record first impressions only: PIEASING: : : : : : : :ANNOYIM} ALTRUISTIC: : : : : :__:___: EGOTISTIC GRATEFUL: : : : : : :UNGRATEFUL : : : : : :DISLIKE HARMONIOUS: : : : : : : :DISSONANT LIKE FAIR: : : : : : : :UNFAIR Rate Person D on the following scales: PlEASING: : : : : :___}___:ANNOYING ALTRUISTIC: : : : : : : :EGOTISTIO mmaoulous: : : : : :____:___:DISSONANI' FAIR:___: : : : :___:___:UNFAIR New we want to ask yOu something more about the event we have suggested: the regard of 0 toward D. Remember that we want you to treat this as a real event: 85 1° Please indicate, by making a check-mark next to one of the following statements, whether C's attitude toward D has some effect on what your own attitude toward D should ban-or has no such effect. Please think very carefully about this. Try to think of am possible way in which C's attitude may have some effect on your own. Do not specify "no effect" unless you are confident that none exists: C‘s attitude toward D: 1. Pas some effect on what your own attitude toward D should be. 2. Has no effect on what your own attitude toward D should be. If you have checked statement 1, please record below the ratings of D which you feel would be in perfect accord with your knowledge of C's attitude. Do not state what your ratings of D actually have been in the past, or are now, or will be in the future. Instead, state the ratings of D which you would make if you were rating D on the basis of this one fact only, with all other considerations elim- inated: (If you have checked statement 2, do not do these ratings) PLEASING:__:__: : : : ___:ANNOYING ALTRUISTI C :____: _: : : : :____: EGOTISTIC GRATEFUL: : : °____:__:___:___: UNGRATEFUL LIKE°___ ___:___:___ ____:___:___:DISLIKE HARD'DNICUS:__: : : : - :DISSONANT Regardless of which statement you checked, please write a sentence or two emlain- ing why the attitude in question had the effect it did concerning D: In reference now to Person C, please indicate, by making a check-mark next to one of the following statements, whether C's attitude toward D has some effect on what your own attitude toward C should be---or has no such effect. Please think very carefully about this. Try to think of any possible way in which C's attimde may have some effect on your own. Do not specify “no effect" unless you are confident . that none exists: C's attitude toward D: 1. Has some effect on what your own attitude toward C should be. 2. Has no effect on what your own attitude toward 0 should be. If you have checked statement 1, please record below the ratings of C which you feel would be in perfect accord with your knowledge of C's attitude. Do not state what your ratings of 0 actually have been in the past, or are now, or will be in the future. Instead, state the ratings of C which you would make if you were rating 0 on the basis of this one fact only, with all other considerations elim- inated: (If you have checked statement 2, do not do these ratings) (uvtr) -11- PLEASING: : : : : : : :ANNOYING ALTRUISTIC: : : : : : : :EGOTISTIC GRATEFUL:___: : : : : : :UNGRATEFUL LIKE: : : : : : :‘~ :DISLIKE 5 p; r: 8 . 23 :DISSONANT FAIR: : : : : r___r___:UNFAIR Regardless of which statement you checked, please write a sentence or two explain- ing why the attitude in question had the effect it did concerning 0: Please indicate whether the situation (G's attitude toward D) that we have suggested: 1. Could conceivably have happened 2. Could not conceivably have happened 3. Don't know Please indicate what attitude actually is held by C toward D: 1. Positive attitude 2. Neutral or indifferent attitude 3. Negative attitude Nb attitude Don't know Thank you for your help. Your assistance is much appreciated. MICHIGAN STATE UNIVERSITY LIBRARIE ES I! ll” MINI