If?" swam/me ‘INTERPERSGNAL samsmvm; AN EVALUATION OF A STEREOTYPE ACCURACY. mmma PROGRAM Thesis for the Degree of Ph. D. MICHIGAN STATE UNIVERSITY Morris S. Spier 1969 LIBRARY Michigan State University “W“ This is to certify that the thesis entitled IMPROVING II‘ITERPERSONAL SENSITIVITY: AN EVALUATION OF A STEREOTYPE ACCURACY TRAINING PROGRAM" presented by Morris S. Spier has been accepted towards fulfillment of the requirements for _P_h.D,_ degree in Jeanne—Leg}! ABSTRACT IMPROVING INTERPERSONAL SENSITIVITY: AN EVALUATION OF A STEREOTYPE ACCURACY TRAINING PROGRAM by Morris 5. Spier The word ”stereotype” has taken on a negative connotation, and has becono synonymous with ”bias” and "prejudice". But there is alple evidence in the literature that stereotypes do not auto- matically lead to poor understanding of others. Rather, it is the rigid adherence to an inaccurate stereotype that causes poor understanding. Accurate, flexible stereotypes may actually enhance interpersonal sensitivity. The primary purpose of the present research was to determine whether stereotype accuracy could be ilproved through participation in a formal training progran. Two experimental training prograns used a pretest-training- posttest paradigm to improve stereotype accuracy (SA): the ability to predict group norns. In Study One, a pilot program, training improved predictions about narital stereotypes evon though training was provided on executive, psychologist, college student, and professor stereotypes. But the criterion instrunents were unwieldy and the single criterion design gave little control over what occurred between pretest and posttest, while a great deal of sample shrinkage took place from the first to the last Morris 5. Spier SA training session. Study Two's revised and shortened criterion-training instruments made each training session a complete unit in itself. Significant SA improvement occurred in the second study, but improvement on one stereotype did not assure improvement on others, and vice versa. Improvement in Study One suggested generality of the under- lying principles and training designs. The specificity of SA and its improvement, research design requirements, and ease of administration, however, favor the single session unit design of Study Two. Moreover, combinations of training activities within a program emphasizing both content and process hold the greatest promise for maximization of training impact. Practice and feedback formed a core for the additional discussions, diagnosis, base-rate, noncommitment, pooling, and panel procedures and conditions of the present studies. An additional finding was that Js who did not make written pretraining predictions improved about as much as Js who had made the written judgments. Additional data showed that the personality and intellectual correlates were also stereotype specific, and that initially inaccurate Js benefited more from training than initially accurate Js. Both findings had relevance for the selection of trainees. Finally, moderately difficult criterion items best reflected Morris S. Spier training outcome from a statistical point-of-view. From a practical standpoint, however, the differences were not worth- while. SA measures should continue to reflect judging reality and include the total range of item difficulties. Subjective feelings of improvement and liking for training were unrelated to actual outcome which precludes their use as criteria for training effectiveness. Implications of the findings for future training and research were discussed. Approved: Date: IMPROVING INTERPERSONAL SENSITIVITY: AN EVALUATION OF A STEREOTYPE ACCURACY TRAINING PROGRAM By Morris 84 Spier A THESIS Submitted to Michigan State university in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Psychology 1969 .\ 1") cs] , '2, r (l " To Alice ACKNOWLEDGMENTS The completion of a dissertation is an opportunity for reflection on the preceeding years of preparation and on the people that made it possible. The author wishes to express special appreciation to the members of his committee, Drs. T. Allen, J. Wakeley, and C. Williams. Their patience and support were important not only in the completion of this manuscript, but throughout the author's graduate career. Deepest appreciation is due Dr. H.C. Smith, chairmmn of the author's committee, who provided guidance, support, and motivation in the present research, and who taught by his personal example what it means to be a professional psychologist. Special thanks are also extended to Drs. N. Abeles and H. Stern for their considerable contributions. Finally the author thanks his wife, Alice T. Spier, who seems to instinctively know what it means to be sensitive. She more than earned her Ph.T.. iii TABLE OF CONTENTS Page ACKNOWLEDGPENTS O O O 0 O O O O O O O O O O O O O O O O O O O 111 LIST OF TABLES, O O O O O O O O O O O O O O O O O O 0 O O 0 O 0 Vi LIST OF APPENDICESe e e e e e e e e e e e e e e e e e e e e e e Vii INTRODUCTION. . . . . . . . . . . . . . . . . . . . . . . . . . 1 CHAPTER I - HISTORY OF THE PROBLEM, . . . . . . . . . . . . . . 2 The Nature of the Ability. . . . . . . . . . . . . . . . . 2 The Measurement of the Ability . . . . . . . . . . . . . . a Training to Improve Sensitivity. . . . . . . . . . . . . . 6 Principles of Training . . . . . . . . . . . . . . . . 9 Practice and feedback . . . . . . . . . . . . . . . . 9 Participation-discussion, . . . . . . . . . . . . . . ll Diagnosis of error tendencies . . . . . . . . . . . . 13 An explicit personality theory. . . . . . . . . . . . 14 An empathic technique . . . . . . . . . . . . . . . . 1s Pooling . . . . . . . . . . . . . . . . . . . . . . . 16 Delay of impression formation . . . . . . . . . . . . 17 Use of base-rate data . . . . . . . . . . . . . . . . 17 Stetement of the Problem . . . . . . . . . . . . . . . . . 19 CHAPTER 11 - STUDY ONE. . . . . . . . . . . . . . . . . . . . . 21 General Design . . . . . . . . . . . . . . . . . . . . . . 21 Subjects . . . . . . . . . . . . . . . . . . . . . . . . . 21 Measures of Stereotype AccurCCY. . . . . . . . . . . . . 21 Trainlns Procedures. . . . . . . . . . . . . . . . . . . . 26 Contents of the Training Sessions, . . . . . . . . . . . . 29 Addition‘lvuiableseeeeeeeeeeeeeeeeeee 33 Methodofmlysis.................... 35 Results................... eeeeee 37 Summary of the Results of Study One, , . . . . . . . . . . 43 CHAPTER 111 - STUDY TWO . . . . . . . . . . . . . . . . . . 46 Problem. . . . . . . . . . . . . . . . . . . . . . . . . . 46 General Design . . . . . . . . . . . . . . . . . . . . . . 46 Subjecte . . . . . . . . . . . . . . . . . . . . . . . 47 Measures of Stereotype Accuracy, , . . . . . . . . . . . . £7 Training ProcedureS. . . . . . . . . . . . . . . . . . . . 50 Contents of the Training Sessions . . . . . . . . . . . . . 53 Additionel Verieb1e%.. . . . . . . . . . . . . . . . . . . 55 Method of Analysis . . . . . . . . . . . . . . . . . . . . 57 Results . . . . . . . . . . . . . . . . . . . . . . . . . . 58 Summary of the Results of Study Two . . . . . . . . . . . . 66 iv TABLE or CONTENTS--(Continued) CHAPTER IV - DISCUSSION. . . . . . . . . . . . . . . . Results of the Training Studies . . . . . . . . . Implications for Training to Improve Sensitivity. How can training be improved?. . . . who should be trained? . . . . . . . How should training be evaluated?. . Implications for Future Research. . . . . CHAPTER V - SUMMARY AND CONCLUSIONS. . . . . . BIBLIOGRAPW O O O O O O O O O O 0 O O O O O O APPENDICES . . . . . . . . . . . . . . . . . . Page 70 7O 75 75 78 79 8O 82 84 88 Table 10 ll 12 13 lb 15 16 17 LIST OF TABLES Summary of training activities in Study One. . . . . . The number of subjects who completed stereotype accuracy training . . . . . . . . . . . . . . . . Internal consistency of the stereotype accuracy predicmrseeeeeeeeeeeeeeeeeeee The results of training. . . . . . . . . . . . . . . . The relationship between initial accuracy and amount ofimprovcment.................. Summary of the relation of item difficulty to the measurement of improvement due to training. . . . The generality-specificity of stereotype accuracy. . . Summary of the relationship of intellectual ability to training outcome for men. . . . . . . . . . . . . Summary of item analyses for original and revised stereotype accuracy predictors (based on the Fall 1967 data) 0 0 O O O O O O O O O O O O O O O O O 0 Summary of training activities in Study Two. . . . . . The number of subjects who completed the revised stereotype accuracy training program. . . . . . . Spearman-Brown reliabilities of the shortened stereotype accuracy predictors. . . . . . . . . . The results of the revised training program for the total sample. . . . . . . . . . . . . . . . . . . The relationship between initial accuracy and amount of improvement for each revised criterion . . . . Summary of the results of delayed impression training. Summary of the relation of training outcome and intellectual ability. . . . . . . . . . . . . . . Summary of the results of the tests of the hypotheses. vi Page 36 37 38 38 40 bl 43 48 53 57 58 59 59 61 64 7O Appendix LIST OF APPENDICES Page STUDY ONE STEREOTYPE ACCURACY MEASURES. . . . . . . . 88 RESULTS OF ANALYSES FOR STUDY ONE: ADDITIONAL DATA. .103 STUDY TWO STEREOTYPE ACCURACY MEASURES. . . . . . . .109 RESULTS OF ANALYSES FDR STUDY TWO: ADDITIONAL DATA. .120 vii INTRODUCTION The primary aim of the present research was the improvement of the ability to understand others through participation in a formal training program. Two groups, each containing nearly 200 Michigan State University students, took part in respective 10 week experimental training programs designed to increase their interpersonal sensitivity (Mietus, I969; Price, 1969; Smith, 1968a). The present studies were concerned primarily with the training sessions aimed at improving stereotype accuracy: the ability to accurately predict group norms. Since an important practical consideration of the research was the development of general training principles, the focus could as well have been any component. Chapter I of the following report briefly reviews the recent literature with regard to the definition, assessment, and attempts to improve sensitivity. The data collected in the experimental training programs are reported in Chapters II and III. Chapter IV contains a discussion of the results including their relation to previous research, and Chapter V is a summary of the present research and its findings. CMHHI HISTORY OF THE PROBLEM The majority of men are subjective towards themselves and objective towards all others, terribly objective sometimes .. but the real task is to be objective towards oneself and subjective towards all others. Soren Kierkegaard (The J 1 , Oxford University Press, 18475 Interest in the ability to understand people existed long before experimental attempts to define, measure, or develop it. Indeed, interpersonal sensitivity has long been an elusive goal of the psychologist, manager, educator, and salesman alike. This pervasive interest appears in each man's occupational role: the psychologist wishes to understand his client; the manager, his employee; the educator, his student; the salesman, his customer; and as social beings, all want also to understand their families, friends, and peers. The Nature of the Ability Everyone wants to be sensitive, but not everyone can con- vincingly demonstrate his understanding of others, nor even communicate what it means to be sensitive. Smith's definition of sensitivity as "the ability to predict what an individual will feel, say, and do about you, himself, and others" (1966, P. 3), places the assessment of the ability in a central position. Prediction is the best, if not the only, test of understanding people. To be sensitive is to be able to accurately predict both private data (what a person will feel) and public data (what a person will say and do) about others. Gough's (1968) grid approach is reproduced in Figure l as an aid to conceptualizing the tasks and goals of interpersonal sensitivity. The key question seems to be “What do you want to understand about whom?" Figgge 1: Suggested domain sampling for appraising social acuity. (Reproduced from Cough, 1968, p. 3) perception of individuals 11 I Outer events, Inner events, actions, etc. feelings, attitudes, etc. III IV perception of peOple-in-general Gough's schema retains an emphasis on predictive accuracy while isolating the ”what” and "whom” as the two major dimensions of interpersonal perception. As in Smith's definition, the ”what" of understanding is contained on a dimension from inner to outer events (private to public). In the manner of Bronfenbrenner, Harding, and Gallwey (1961) who dichotomized the task of the perceiver into ”Interpersonal Sensitivity" and ”Sensitivity to the Generalized Other", Gough's "whom" ranges along a dimension from the understanding of a specific individual to an understanding of people-in-general. while the "good judge of others" should be able to demonstrate his predictive accuracy in each of the quadrants defined by the interaction of the dimensions, the present investi- gation was concerned only with predictions in quadrants III and IV: the prediction of behavioral, attitudinal, etc. norms of groups. The Measurement of the Ability Concern with ”who is sensitive” led to the development of an assessment paradigm with predictive accuracy at its core. Typically, a person (0) fills out a personality, interest, attitude, or other inventory. A judge (J) is introduced to 0 via filmed or tape interviews, face-to-face contact, personality sketches, case histories and/or other descriptive data. J then completes the personality, interest, or attitude questionnaire as he (J) feels 0 filled it out. The measure of J's understanding of 0 then, is the number of items correctly predicted. In 1955, Cronbach analyzed such accuracy scores and concluded that they represented an oversimplification of what was, in fact, a factorially complex ability. Cronbach pointed out that the judge with the highest score was not necessarily the most under- standing of others. Accuracy in judging 0 might result also from J's response tendencies, his understanding of groups, or even the degree of similarity to O assmned by J. Thus, for example, the J who habitually assumes a great deal of similarity between himself and others would achieve spurious accuracy in cases of a high degree of actual similarity. Smith (1966, 1968a, 1968b, 1968c) labeled response sets, evaluative accuracy (level); knowledge of groups, stereotype accuracy; and assumed similarity, empathic accuracy. The level component is based on the assumption that our judgments of others invariably contain an evaluative element. Indeed, there is a tendency to implicitly or explicitly assign goodness or badness to our descriptions of others. Campbell (1967), for example, pointed to the evaluative aspects of stereotypes. The English describe themselves as reserved and respecting the privacy of others, while viewing Americans as intrusive, forward, and pushing. Americans, on the other hand, saw the English as snobbish, cold, and unfriendly, and thought themselves friendly, outgoing, and open-hearted. Our judgments of people are also inevitably influenced by the groups to which they belong. To the extent that knowledge of the groups is incomplete and inaccurate, as it necessarily must be, we will err in our judgments. A final assumption postulates a strong motivation to deter- mine ways in which people around us are similar or dissimilar to ourselves. Thus, Smith (1966, p. 19) defined empathy as ”the tendency of a perceiver to assume that another person's feelings, thoughts, and behavior are simdlar to his own”. In trying to understand others, we usually assume too little or too much simi- larity between ourselves and them. The goal of sensitivity training, then, should be to introduce accuracy into the evaluative, stereotyping, and empathic processes. Training to Improve Sensitivity Evaluations of training program effectiveness either equate training with amount of clinical experience or psychological course-work of the judge, or administer a pretest and posttest with interpolated training. Evidence for the effectiveness of either type of program is difficult to obtain, and to complicate matters, all too often the actual training procedures remain unspecified (Wakeley, 1961). Some other programs depend on anecdotal evidence and the participants' "liking for training" or confidence in their improved judgment as "proof" of training value. In 1955, Taft reviewed the previous three decades of research into the effectiveness of training at increasing understanding of others. Most of the studies followed the educational model. Taft concluded, after controlling for similarity of Js and Os in an academic setting, that "physical scientists, and possibly other nonpsychologists, e.g., personnel workers, appear to be more capable of judging others accurately than either psychology students or clinical psychologists.... There is also evidence that suggests that courses in psychology do not improve ability to judge others and there is considerable doubt whether professional psychologists show better ability to judge than do graduate students in psychology" (p. 12). Eleven years later, in 1966, Smith's review included five studies carried out since Taft's review. The picture remained gloomy: training did not increase sensitivity. Fancher (1967), as part of a larger study, speculated that those in his sample of student Js who were most similar to clinical psychologists should also make the most accurate predictions. Results: experience (number of psychology courses) and competence (grade in an abnormal psychology course) correlated -.l7 and .05, respectively, with accuracy. His findings further suggested that the professional preparation of clinical psychologists makes its greatest impact not on the accuracy of judgments, but on the validity of conceptualizations of others. Goldberg's (1968) review found little success in training attempts to date. Similarly, Campbell and Dunnette (1968) reviewed the literature covering the effectiveness of T-group experiences. Conclusion: participants emerge with a larger vocabulary of interpersonal terms, but with no demonstrable increase in inter- personal accuracy. The findings are also generally negative with regard to the usefulness of ”confidence in accuracy or improvement" as a criterion of training effectiveness. Oskemp's (1965) judges, for example, estimated the percentage of 25 items they would predict correctly, and then made predictions on the basis of minimal information about an 0. While initially only slightly overconfident, Js (including experienced clinical psychologists) became more and more certain of their accuracy as the amount of information about 0 was increased. Finally, having read all of the data about 0, the average J estimated 53% correct. 28% were actually correct. Oskamp concluded that ”the judges' confidence ratings show that they become convinced of their own increasing under- standing of the case... while... their certainty about their decisions became entirely out of proportion to the actual correctness of those decisions” (p. 264). Opsahl (Dunnette, 1968) used a computer technique to control for 3 possible sources of variance in interpersonal prediction: assumed similarity, base rate of item endorsement, and social desirability. Assessment instruments developed by the computer used all combinations of the above variables. In the 5 resulting treatments, target 0s appeared in filmed interviews. In a sixth condition, however, Js made predictions based only on stereotypic data (0's age, sex, education, and occupation) with all 3 controls applied. All Js participated in all treatments including a feedback or no feedback condition. For each response, Js indicated their degree of confidence in the correctness of their prediction as either Not Confident, Somewhat Confident, or Very Confident. Among the findings: under the film case conditions with all controls either present or absent, accuracy and con- fidence were strongly related. When controls were only partially relaxed, Opsahl's Js, like those of Oskamp (1965), were con- fident of the correctness of more of their answers without actual increases in accuracy. Finally, where Js predicted only on the basis of stereotypic data, confidence was unrelated to actual accuracy. In a cross-cultural study (Cough, 1968), 231 American Js completed a 66 item questionnaire as they thought male and female Italian 0s would answer. The Js viewed films of the Italian 0s and completed the criterion instrument a second time. Js reported, both in groups and individually, that they found the task ”meaning- ful”, ”interesting”, and ”stimulating”. Result: ”After viewing the film, and in spite of subjective feelings of greater con- fidence and insight, the American viewers declined in accuracy" (p. 8). Mean accuracy before viewing the film was 36.87 items correct: after viewing the film, 32.90 items correct, a statisti- cally significant reduction. Principles of Training As Gough's experimental design suggests, some researchers have gone beyond the traditional educational training grounds in attempts to develop sensitivity training programs. Many of these attempts have carefully specified the procedures, materials, etc.. used in training. Thus, some generally effective training procedures have emerged from the programs, among them: practice at making judgments, feedback about accuracy, discussion, diagnosis, use of an explicit empirical personality theory, an empathic technique, pooling, delay of impression formation, and the use of base-rate data. gzgcgjce and geegbggk, Their facilitative effects on learning in general (Cline, 1968; Dunnette, 1968; Dunnette and Hakel, 1968; Locke, 1967; 10 Smith, 1966) have made practice and fggggggg Sgggwledge gf results‘ central components of attempts to improve understanding. Cline's (1968) ”feedback in judging accuracy training research“ provided experimental evidence for the usefulness of feedback in increasing sensitivity. Eight groups of Js took tests of 3rd person accuracy (i.e., inferred how O's behavior and personality were rated by some significant others; Bronfenbrenner, gtflgl‘, 1961) after being exposed to one of 8 possible combinations of 3 dichotomous variables: feedback after each response to each item versus no feedback; a brief personality summary versus no summary; filmed interviews versus no film. A ninth group (control) made its predictions only on the basis of 0's sex and age. Results: feed- back groups made significantly more accurate 3rd person predictions of behavior and personality than the no feedback groups. Further- more, feedback was the most powerful discriminant on both the behavioral and personality trait predictors. Goldberg (1965) administered massive doses of practice and feedback to experienced clinicians, psychology graduate students, and naive Js as part of a program to improve diagnostic accuracy. Goldberg established both training and testing sets of tasks from 1530 MMPI profiles with criterion diagnoses of neurosis and psychosis. In 9 weeks of training, Js received intensive practice in predicting neurosis or psychosis on over 4,000 MMPI profiles followed by immediate feedback after each guess. In all, when both training and testing sessions are combined, each J viewed 11 over l9‘222,profiles. Results: expert and middle Js showed little change from their initially similar levels of accuracy, and while naive Js showed substantial improvement, their accuracy was still well below that of the more experienced Ja. Practice and feedback, it appears, may be facilitative, but are not in themselves, sufficient for optimal learning. W Jecker, Maccoby, and Breitrose (1965) added a Eggsigiggtigg; gigggggigg,component within the practice and feedback framework to improve teachers' use of non-verbal cues to determine compre- hension. Sound film clips collected in classroom settings were edited to show a single student during the instruction of one item and the time allotted to answer the question on that item. Js guessed whether the 0 answered the question in each clip correctly or incorrectly. Twenty-five such clips comprised a pretest. An experimental group received training: i.e., viewed 15 to 30 clips; identified the nonpverbal cues; made judgments; discussed the correct scoring, interpretation, and meaning of the cues; and were given feedback regarding the accuracy of their predictions. A control group viewed and discussed a film on interpersonal communication. Both groups were then posttested. Results: when scores were equated for initial accuracy, experimental Js showed a mean gain in accuracy of 7.2% while the control Js declined .51 in accuracy. The difference was statistically sig- nificant. 12 A programmed learning approach toward increasing understanding (Dailey, 1966a, l966b) also stressed practice and feedback. Dailey edited published biographies and personally collected case histories to identify key events in a target 0's life. A completed programmed case contained 10 to 15 such key episodes arranged chronologically and each paired with two alternative, but incorrect, events. Each episode required J, initially presented only with 0's occupation, to predict what would happen in the next segment. Immediate feedback followed the prediction. Thus, a formalized program of practice and feedback could be combined with role-slaving exercises, group discussions, etc., to form a complete laboratory (1966a). In one such application, Dailey presented his Js with 12 cases, 2 at a time. Mean predictive accuracy increased 17% from the first 2 cases to the last 2 cases, a statistically significant improvement (1966b). Kepes' (1965) aim was to improve individual accuracy (Gough's (1968) quadrants I and II) through a program of practice, participation-discussion, and feedback. Js viewed a sound film interview of 3 men and 3 women and completed an inventory based on the film calling for 2nd person predictions in the form of differentiating between the film Os (i.e., inferences about how an 0 rates himself; Bronfenbrenner, 25,5;‘, 1961) and recall of specific mannerisms, dress, speech, etc. (observational accuracy) of each 0. Since the focus of training was individual accuracy, the 120 item criterion instrument was constructed to be free of 13 evaluation (level) and stereotype influences. Following pretest administration of the film based inventory, an experimental group of 58 Js went through 8 weekly training sessions. In a typical one hour meeting, Js answered a series of case study items, discussed reasons for responding as they did, received feedback about accuracy, proceeded to the next series of items, and so on, until the case was completed. A control group re- ceived no training, but both groups were posttested with the film based inventory. when Js were matched for initial accuracy, there was no significant differences between pretest and posttest for either group. A trend of gains in 2nd person accuracy was found for the experimental group, but no such trend occurred for the control Js. In a second experiment, Js were pretested using only the 3 men interviews on the film based inventory. The training involved discriminating among the 3 women interviewees who were presented in film, on tape, live, and in written cases to 4 groups, respectively. The men film based inventory was readninistered after training with the result that training on the women interviews did not generalize to accuracy about men. There was no difference in accuracy among the treatment groups and even the general trends of differential gain noted in the first study failed to replicate. Diagnggis of eggor tendenciesI In his first study noted above, Kepes found good gains in observational accuracy among experimental Js. Similar improvement 14 occurred among the control Js with no training. After pretesting, both groups were given their observational accuracy pretest scores and told that they would be tested again at some later date. Hhile seemingly only a special case of feedback, Kepes felt that such diagngsis of present proficiency motivated both groups to improve (Locke, 1967). Grossman (1967) made diagnosis of error tendencies a formal part of his training program. Attempts were made with three experimental training groups to improve evaluative accuracy, empathic accuracy, and individual accuracy, respectively. Js were pretested and posttested on each of the components regard- less of the focus of training to follow. Grossman's training focussed primarily on individual accuracy and stressed practice, participation-discussion, and feedback. Improvement due to training was statistically significant. Comparable gains in individual accuracy, however, were recorded for members of the experimental groups which received specific training on the other components. Grossman, like Kepes, noted that each J had been apprised of his evaluative and empathic error tendencies and felt that this knowledge acted both as a motivator and a basis for change. An gaglicit personality theory, Grossman (1967) was somewhat less enthusiastic about the effectiveness of an empirical personality trait theory in increasing accuracy. Smith (1966, 1968a) noted that we all use 15 implicit, untested theories in our attempts to understand others. He proposed (1966) that Js be taught an explicit empirical personality trait theory to provide a framework within which they could change their own theories and thus improve their under- standing. Grossman (1967) instructed Js in the nature and intercorrelation of the 5 trait dimensions (Linden, 1965): impulsive-controlled, rational-empirical, introverted-extroverted, cautious-bold, and emotional-calm. Next, he introduced Js to the use of the traits in differentiating between 0s and gave opportunity for practice and feedback in making differential predictions. He concluded that the theory contributed little to improvement. WWW. we often use ourselves as a basis for prediction about others. Such tendencies may provide valuable input when we are accurate in our assumptions of similarity or dissimilarity. Diag- nosis, practice, discussion, and feedback are useful tools also in increasing sensitivity through an grgrgnrg_£rg§rlggg_(Mietus, 1969; Silkiner, 1962; Smith, 1966, 1968a). Such a program should emphasize both the content and process of empathic accuracy. First Js would be instructed in the nature of empathy and its relation to sensitivity. Second, they would‘complete, for example, an interest inventory answering for their gwn likes and dislikes. Then they answer the same inventory as they felt O, a typical group member, would respond. Suppose that a typical 0 16 had actually answered ”like" to the first item. Given feedback about Os actual answers, J would find that if he had answered ”like” for himself and for 0, he would be correctly assuming similarity. If J answered "like" for himself but "dislike” for 0, he would be incorrectly assuming dissimilarity. If he answered ”dislike” for himself and for 0, he would be incorrectly assuming similarity. If he answered "dislike" for himself and "like" for 0, he would be correctly assuming dissimilarity. Such an exercise provides both a diagnosis of empathic tendencies and a demonstration of the influence of empathic assumptions on accuracy. mum. Hakeley (1961) focussed on prediction as a process of informa- tion gathering and interpretation culminating in the actual recording of judgments. Experimental training programs dealt with each of these tasks. Two programs aimed at improving J's observation (information gathering) by instructing J to focus on his own reaction to O, or to focus on 0 when observing. Two other programs (information processing) instructed Js to either concentrate on the uniqueness of each 0, or on 0's simi- larity to other people known by J Sig£grr£rg=ggglrrgl. Another program focussed on the use of rating scales, and a final program combined the techniques of the others. Js were pretested and posttested with a test of the ability to differentiate between people regardless of the training Js received. Only the inferring-pooling training, which advocated a kind of stereotyping, 17 and the program combining all the techniques increased accuracy. A second study replicated the results. Del f i i We almost invariably form instantaneous impressions of others (Smith, 1966; 1968c). Springbett (1958), for example, demonstrated that employment interviewers made their decisions to hire or reject within the first 4 minutes of the interview. We do not seek out a person's specific qualities and then attempt to form a general picture of him, rather, we move from the immediate overall impression to the search for specific traits. Our search, however, is indelibly influenced by our early impression. Accuracy should be improved, therefore, by slowing down the impression formation process. Dunnette and Hakel (1968) were unable to specify a procedure, but felt that sensitivity training should aim to develop a de f i sion £2;se£len.tnd instill in J ”an awareness of the necessity to gather all information prior to making or 'fixing‘ early impressions or judgments” (p. 38). 0 e b e- d t Opsahl (Dunnette, 1968), it was noted earlier, controlled for response set strategies in his computer generated sensitivity measure by correcting for base-rate of item endorsement. Items were matched according to the relative proportion of people in general who endorse (or reject) a particular item. Such know- ledge, however, when accurate, could aid in understanding. 18 Large and Diamond (1954) asked Js to predict whether or not student Os would pass ability test items. After pretesting, Js were classified as poorest, mediocre, or best and ”trained” by being told the difficulty level of some specimen items. When Js responded only to the items, the average correlation between predictions and actual outcomes for the best Js was .73, and for the poorest Js, .56. After base-rate information was given, the same Js had mean correlations of .77 and .73. Conclusion: ”Apparently the difference between 'poorest', 'mediocre', and 'best‘ judges is that the 'best' judges have some experiential referent for the percentage of the population that can pass an item. Giving such referents to the 'poorest' and 'mediocre' judges... leads to a significant reorientation of such judgments" (p. 33). Stelmachers and McHugh (1964) used MMPI norms to develop a scoring key around the base-rate of item endorsement of 3 broad groups: college females, adult females, and adult males. Experi- enced psychologists, psychiatrists, and psychiatric social workers predicted the responses made by two women and two men to 171 MMPI items. The female 0s were a normal college sophomore, and an elderly woman with a long medical and psychiatric history. The male Os were a teenage homosexual and an adult depressive. The authors made similar predictions using only the special base- rate key. Results: the special base-rate keys were more accurate than the expert Js for 3 out of the 4 criterion 0s. 19 Statgrent of she Prgblem The primary purpose of the present research was to evaluate the effectiveness of a pilot and a revised training program designed to improve stereotype accuracy (Gough's (1968) quadrants III and IV). For decades the social science literature has supported the view that stereotypes are simplistic, virtually impossible to alter, and 252:; accurate (Harding, 1968). Stereotypes have become synonymous with prejudice and insensitivity despite a growing body of empirical evidence which suggests that the converse is true (Cline, 1968; Tagiuri, 1968). Attenpts have been made to increase sensitivity, but attempts to increase stereotype accuracy, i.e., the ability to predict group norms, have been largely ignored. While a number of investigators have developed measures of the ability and begun to explore its nature (Harris, 1962; Johnson, 1963; Shears, 1967; Silkiner, 1962; levels, 1960), training activities have attempted to eliminate stereotyping influences on judgments. Study One, a pilot study, proposed to: (1) form a stereotype accuracy training program based on principles established in the research literature; (2) test the hypotheses and suggest new ones; and (3) generate hypotheses about training effectiveness, means for measuring effectiveness, and the relation cf training impact to personality, intellectual, and attitudinal factors. Study Two was a further test of hypotheses derived from the pilot 20 program as well as an opportunity to apply the findings to the development of a revised program. The following hypotheses guided Study One. Hypothesis 1: A training program will be effective in improving stereotype accuracy. Hypothesis 2: The effectiveness of training varies inversely with initial accuracy. Hypothesis 3: The effectiveness of training will be reflected more clearly on stereotype accuracy predictor items of moderate difficulty, than on those of either high or low difficulty. Hyngthesis A: Stereotype accuracy is a situation specific ability. Hypothesis 5: The effectiveness of training varies with personality, intellectual ability, and attitudes toward training. Chapter II discusses the methods and results of Study One. CHAPTEP II STUDY ONE General Design Study One used a pretest-posttest design with interpolated training. Subjects were administered a single criterion, received training in making predictions about specific groups or group members, and were posttested with the criterion measure used in the pretesting. Training took place in A successive sessions of about 1 hour and 10 minutes each. Subjects Judges (Js) in this study were 200 Michigan State University students, representing a wide range of college majors and levels, who were enrolled in a Fall 1967 section of a Psychology of Personality course. While training took place during the twice weekly class meetings and as a part of the regular class-work, attendance was not mandatory. Me s es 0 e e Ac u S The paper-and-pencil measures of stereotype accuracy included tests of accuracy in judging the difference between: (1) happily married, unhappily married and divorced men and women; (2) executives and unskilled workers; (3) a professor and college men in general; 21 22 (6) psychologists and men-in-general; and (5) a particular man (Morgan Johnson) and men-in-general. Copies of the instruments appear in Appendix A. An IBM scoring sheet, provided for use in recording answers, allowed machine scoring of all tests. 1. The criterion instrument used in the present study was based on the research of Johnson and Terman (1935) concerning the relation of marital happiness to similarity (in interests and personalities) of husbands and wives. The development of the criterion measure is of interest in that it illustrates the construction of all of the dependent variables. Johnson and Terman compared the resemblance of partners in three groups: happily marrieds, unhappily marrieds, and divorced couples. One hundred couples in each category were matched to control for other relevant group memberships (e.g., occupational level, socio-economic status, age, religion, national origin, etc.), responded to lengthy personality and interest inventories, and participated in in-depth clinical interviews. Results: no relationship between marital happiness and husband-wife resemblance, but many of the individual inventory items correlated with marital happiness and thus distinguished between the members of the three groups. Close (1963) used Johnson and Terman's findings to develop the gaggigge Tgst, the criterion measure of the present study. The instrument measures a J's ability to differentiate between 23 typical happily married, unhappily married, and divorced people. Part I deals with men (28 items); Part 11 deals with women (28 items). The two-part construction served to partial out another significant group membership: sex. The instructions read: For each of the following statements, rrrk the one of the three grggrs xou think 1; bert gesrribed by it, The correct answers are based on the answers that the groups made on lengthy questionnaires. For example, 100 happily married women, 100 unhappily married women, and 100 divorced women answered the question: "Do you prefer a play to a dance?". Results: 81% of the happily married women answered ”yes". 58% of the unhappily married women answered "yes". 44% of the divorced women answered "yes". Therefore, the correct answer to the statement, "Most apt to prefer a play to a dance” is ”happily married women." While these measures have no demonstrated empirical validity, their construction suggests that they do have content validity. The instruments purport to measure stereotype accuracy: the ability to predict group norms. The correct answers throughout, are based on the actual responses of the Os :nd are not the opinions of the trainer. Thus, a J's stereotype accuracy is the number of correct predictions he makes. 2. The sixty item test of The lntergsts gf Errgurivrs vsI Unskglled Workers required the J to differentiate between professional and nonprofessional menixntheir answers to an interest inventory (“ilkiner, 1962). J was informed that: A large number of executives and professional men (lawyers, managers, etc.) and a large number of unskilled workers (laborers, porters, etc.) checked whether they liked or disliked each of the interests or activities below. A larger percentage of the unskilled workers like 24 half of the interests; a larger percentage of the executive and professional group liked the other half. Err; ”1” if you think more unskilled workers liked a particular interest. Err; "2" if you think more professional workers liked it. 3. The C se 0 the P es provided separate exercises for the development of level, stereotype, and empathic accuracy. Js were told the professor's position on five personality trait dimensions and then made predictions about how the professor answered the items on a personality inventory. The Sterrgrxre Trrining exercise is a two-part test of the ability to predict the ways in which the professor conforms to the stereotype of college men in general (part 1) and the ways in which he differs from the group norm (part 2). There are twenty true-false items on each part of the test. The instructions to part 1 read: The statements below were in the Inventory completed by the professor. On these particular statements, as it turned out, he gave the sage answers as those given by more than two-thirds of the college men who comleted the inventory. The question here, therefore, is not just how well you understand the professor. It is also how well you understand the typical college man. Circle the answer "true” if you think both the college professor and the typical college man answered it ”true". Circle the answer ”false” if you think both the college professor and the typical college man answered it "false". The instructions to part 2 read: To the following statements, the professor gave a differenr answer to the one given by two-thirds or more of the college men. The question here, therefore, is, How does the professor differ from the typical college man? Circle "true” if you think the professor answered "true" but the typical student answered ”false". 25 Circle "false” if you think the professor answered ”false” but the typical student answered "true". 4. ID! Spggial 19222955! gf Paxcholggists (60 items) required J to differentiate between psychologists and men-in-general on the basis of their responses to a series of vocational interest blank items (Silkiner, 1962). The directions for the test read: How do the interests of psychologists differ from those of other men? To answer the question, several hundred male psychologists and several thousand other business and professional men checked whether they would "like” each of many different occupations, amuse- ments, activities, and kinds of people. A higher percentage of the psychologists liked same interests. For example, 41 percent of the psychologists said they would like to be the "author of a novel"; only 32 percent of men in general expressed such a liking. A lgggr percentage of the psychologists liked some interests. For example, only 29 percent of the psychologists said they would like to be a ”sales manager" whereas 37 percent of men in general expressed a liking for this occupation. Mark for each of the interests below whether you think mgre or less psychologists liked the interest. Mark "1" if you think a higher percentage of psychologists than men in general liked the interest; mark "2" if you think a lgwer percentage of psychologists liked the interest. 5. The Case of Morgan Jghnson asked J to differentiate between the interests of a college senior majoring in psychology and those of men-in-general. The 66 item test begins with a brief biographical sketch of "Morgan Johnson" (see Appendix A) and contains 2 sets of 32 items each. The first set of items ("Morgan is like the typical man") notes that: Morgan filled out the Strong Vocational Interest Blank that requires the respondent to answer "Like," "Indifferent," 26 or "Dislike" to a long list of intersts. The directions of this test ask the respondent to "disregard consider- ations of salary, social standing, future advancement, etc. ... consider only whether or not you would enjoy the interest regardless of any necessary skills, abilities, or training which you may or may not possess.” In the first group of interests below Morgan gave the reg: answer as more than half of several thousand representative American men. Mark the answer that you think was given by both Morgan and the typical men. Use "1" for "Like"; "2" for "Indifferent"; and ”3" for "Dislike”. In the second series of 32 items J considered the ways in which "Morgan is unlike the typical man." In the group of interests below, the answer given by Morgan is the one in small letters (l,i,d). His answers to these interests, however, were different from those given by the majority of men. From the two possible answers not chosen by Morgan that are indicated by capitals (L,l, or D) mark the one that you think was chosen by the typical man. T ini ce e Study One used three basic training procedures: I. A program of diagnosis of error tendencies, practice in making judgments, feedback regarding accuracy, and discussion: 2. A pooling technique; and 3. An empathic technique. Each procedure emphasized various aspects of the SA training program. The first group of procedures (diagnosis, practice, feedback, and discussion) stressed the SA principles. -Our impressions of a group and its members are necessarily 27 based on inaccurate and incomplete information, but Stereotypes are not, in themselves, the.cause of poor prediction (understanding). -Rather, it is the rigid adherence to an inaccurate, incomplete stereotype that interferes with prediction. oAn accurate, flexible stereotype can aid in prediction. The process of diagnosing errors involved the administration of a pretest which required prediction about a specific group. Formally, a 15 item, true-false Test f th ' an H ke (see Appendix A) demonstrated that an inaccurate stereotype results in poor (inaccurate) prediction. Js predicted workers' responses to statements about their jobs. A low score was indicative of an inaccurate stereotype of workers. On a less formal level too, Js got feedback about the accuracy of their stereotypes as they practiced applying them. They also received knowledge of results and discussed their predictions for a wide range of groups. Practice in applying a stereotype to a single group member demonstrated the need for maintaining flexibility. A particular 0 may fit the group norm in some ways, but will almost certainly differ from it in other ways. Hence, the accurate stereotype that is a "best bet" about a person when we first meet him will have to be modified. Each J recorded his accuracy scores on a Personal Profile Sheet (see Appendix A) which contained norms for the class as a whole. He then compared his accuracy with that of the other members 28 of the class. Within the program of diagnosis, practice, feedback, and discussion, the pooling and empathic procedures stressed the process of acquiring information about group norms,and the psycho- logical organisation of the information in order to make more accurate judgments. Wakeley': (1961) pooling method had J pick from among people he already knew well, a representative "pool" of people that most closely matched O. J was instructed to assume that the O judged thought; felt, and behaved as the average 0 in the matched pool. The following instructions (Smith, 1966, p. 146) were read to the Js. In the course of your living you have obtained a great deal of information about many people. The pooling principle simply suggests that you use this information when making inferences about a person with whom you have had little contact. When you are attempting to make inferences about a person whom you do not know well, one of the things which you can do is to form a pool of people whom you do know well who are like the unknown person. You take what you do know about the person, forms pool of people you know well, and then make your predictions or judgments based on the pool. The important things to remember in making these pools are to use people you know well and to use all of the information you have about the person you are trying to judge. You may form a pool that leads to wrong predictions if you use just one piece of informa- tion about the person, such as, his skin color, his religious preferences or any other single piece of information. You may also form some pools that lead to wrong predictions if you use people whom you do not know well. Js listed, at the top of their answer sheets or test booklets, the names of as many individuals that they knew well who matched 29 the characteristics of the O. Thus, the pooling technique en- couraged conscious utilization of information and experience gleaned from past encounters with similar people. Smith defines ”empathy" as "the tendency of a perceiver to assume that another person's feelings, thoughts, and behavior are similar to his own" (1966, p. 19). In using the empathic technique as a part of the SA training program, Js used their own responses to the tests as a base of similarity-dissimilarity to the Os being judged. For example, The Specirl Igterests of Psychglogigts required J to give his own responses directly on the test booklet. The instructions were: Mark ”1" if 122,would say "like" or "yes" to the interest or statement: Mark "2" if ygg_wou1d say "dislike" or “no” to the interest or statement. J completed the test by predicting the interests of the psycholo- gists. The correct answers were read to the class. J compared the number of items he answered correctly with the number of items he answered (or expected to answer) in the same way as psychologists. J was thus encouraged to focus on both the benefit and hindrance to understanding of assuming similarity or dissimilarity between himself and the person being judged. Contents 2f the Trainigg Sessions . One or more of the training procedures were combined with each of the SA measures to comprise a single training session. Table 1 summarizes the content of the 4 meetings devoted to SA training within a larger program of training to improve sensitivity to others. SA training took place in the 6th, 7th, 8th, and 15th sessions of the overall program. The criterion was readministered (posttestfi in the 17th meeting. Table 1. Summary of training activities in Study One. Session Stereotype Pretest Training Procedure* Posttest (I) (2) (2) 6 Marital (Criterion) X 6 American Worker Diagnosis Only 7 Morgan Johnson X X 8 Executives-Workers X X 8 Psychologists X X X X 15 Professor X X X 17 Marital (Criterion) X *Note: (I) - diagnosis, practice, feedback, discussion; (2) - pooling; (3) - empathic technique. Training session 1. -- Js were given the criterion instrument (the Marriage Test), the Tgst 2f the grerican Wgrker, a scoring pencil, and an IBM answer sheet as they entered the classroom. Following a lecture on the nature of stereotypes and SA, and an introduction to the training principles, Js completed the test of marital stereotypes on the separate answer sheet. Test book- lets and answer sheets were then collected. The Tgst 2f the Amerigrg Wgrker was completed directly on the test booklet. Correct answers were given and a distribution of scores was written on the blackboard and discussed by the class as a whole. 31 Trainigg sessign 2. -- h C se Johns n was handed to each J as he entered the classroom. The principles of train- ing were reiterated and the Js completed the test. The separate answer sheets were collected, but the test booklets were retained. Js completed the 64 item test, 10 items at a time, directly on the test booklet. The correct answers were given immediately after each series of 10 items and responses were dis- cussed. Training session 3. -- Session 3 focussed on The Interests 2f Erecutives vg, Uhrkilleg Wgrkers and The Sgrrial Inrerests of Pszchglogists. The training materials were distributed as in sessions 1 and 2. Js completed the 60 item Executives vs. Unskilled Workers test and returned the separate scoring sheet. Js then wrote the numbers 1 through 9 on the back of the test booklet and completed the following exercise which was read to them by the trainer. 1. In the following frames one or more words is missing. You will be required to write in the missing word(s) before turning to the next (gags. frame) where you find the correct (rggggn§_). 2. Properly used, this manual will teach you to accurately predict the difference in the interests of executives and unskilled workers. Illustrations or examples will teach you the basic principles involved in the (differences) in interests of the two groups. 3. It will be helpful for you to assume the imaginary role of both the typical executive and the typical unskilled worker. Then, deciding that an executive would he more likely to (2£2£££) a musical comedy is not at all difficult. a. As you imagine the typical individual in both groups, note that going to a musical comedy would be more agreeable to (grgcutives) than to (uns ed kers). 32 5. In the same way, note that unskilled workers would be more likely to (prefer) the sport of hunting. 6. Hunting is a less intellectual and more rugged type of interest than seeing a musical comedy. These differences are typical of the differences between (pppglllpg_gpr§prp) and executives. 7. For similar reasons, the usually better educated (erecutive) would prefer a sport like golf more than would the (ppskilleg pprker). 8. Imagine which would (preger) to be a corporation lawyer as opposed to being a criminal lawyer. The executive type would probably (prefer) to be a corporation lawyer. 9. The unskilled worker type would prefer the more adven- turous job: being a (criminal) lawyer. 10. In your temporary role as an executive you should (prefer) the interest of ”scientific research worker". In your other role as an (unskilled workpr) you would be (less) likely to prefer such an interest. (Smith, 1966, p. 148) The programmed exercise introduced the Js to the technique of training to understand others via a linear program for developing SA. Following the exercise the Js completed the 60 test items, 15 at a time, directly on the test booklet. Feedback and dis- cussion followed each series of 15 items. Next, the Js were taught, and encouraged to use, the pooling and empathic procedures in judging psychologists' interests. They completed the test, handed in the separate answer sheet, and were given feedback followed by discussion. ai in n 4. -- Tpe Case 2; the gsxppplpgx :rpfessgr contains separate exercises for each of the evaluative, stereotype, and empathic components of understanding allowing simultaneous training on all these components. Js were again presented the principles and goals of SA training. They then completed the Stereotype Training exercise and returned the separate answer 33 sheet before receiving feedback and discussing the case. Training sessions 1 - 3 were consecutive, but session 4 took place 3 weeks after the third session. The criterion post- test was administered approximately 5 weeks after the first SA training session and one week after the fourth session. Addirional Variables In addition to the SA scores, data were collected to explore the effects of: (l) J's personality; (2) his intellectual ability; and (3) his attitude towards training, on training outcome. The personality variables are those measured by the Protepgb Perrgpplirz Inventory developed by Linden (1965). The test contains 200 items, with 40 items measuring each of 5 basic traits: cautious-bold, emotional-calm, introverted-extroverted, impulsive- controlled, rational-empirical. A sixth score, acquiescence, measured the J's tendency to mark "true" to a statement. The measures of intellectual performance include: 1. Scores on a 56 item, multiple choice Midterm examination. Items 1 - 28 are a measure of the student's knowledge of the text material. Items 29 - 56 measure his com- prehension of the lecture (training) concepts and principles. 2. Scores similar to those for the Midterm are also available for the 86 item, multiple choice Final examination. Forty three items tested the text material since the midterm, and forty three items tested knowledge of the lecture (training) material. 3b 3. Grade Point Average: based on the trainee‘s overall academic record to the beginning of the term of training, i.e., his GPA as of the end of Summer term l967. 6. The Cgllege Qpplifipprign Tert (Form B, 1956) is a measure of general academic aptitude. Test scores are reported in the form of percentile ranks which specify a student's position on the test relative to all new students in his entering class. The present study used scores for the verbal (V) and information (I) subtests, and the total score (T). ”T” included a subtest measuring numerical proficiency. "V" is a measure of (recognition) vocabulary and reflects verbal abilities associated with success in social science, literature, and other similar fields. ”I” measures general information in the social and natural science. An additional, demographic, variable was the sex of each J. Finally, three variables apply directly or indirectly to an assessment of J's attitudes toward the training experience. 1. Confidence in improvement. When the posttest criterion for the overall training program, of which this study was a part, was administered (Test 9; the ébiliry rp Ungerrtrpd Eggple, Part II), Js estimated the number of items (out of 60) they expected to answer correctly. They were instructed to evaluate the impact they felt the entire training program had had on their ability to understand others. 35 2. Liking for training. In the same session in which the Js made their estimates, they were asked to rank, in terms of bene- fit and interest, their 5 most recent courses. The 5 courses were to be ranked from 1 to 5 (with 1 being highest and 5 lowest). 3. The number of training sessions attended. The type of sensitivity instruments used in the present study were used in previous research (Kepes, 1965; Grossman, 1967). Control groups that did not receive training did not improve. In as much as this finding was consistent over a number of studies, it seems to be a well established phenomenon that no improvement can be expected without training. Hence, the control groups in the previous literature may serve as the controls for the present research. Method 2: (mlxsit 200 subjects completed the training program. 5c.5% of the sample were females; 45.5% males. 100 Js, 50 males and 50 females, were randomly assigned to a validation group (V). Analysis of the V data aimed at generating hypotheses to be tested on the remain- ing 100 subjects, of whom 39 were males and 61, females. The latter sample provided a cross-validation (C-V). Since class attendance was not mandatory, not all subjects completed each and every phase of the overall training program. Table 2 shows how many trainees took part in the SA program. The V group was consistently superior to the C-V group on measures of intellectual ability. The groups did not differ on 36 measures of attitude toward training. Personality trait patterns are similar for the total group and for females, but C-V males were more extroverted than V group males. Table 2. The number of subjects who completed stereotype accuracy training. A, Validation Srrple Total Erica Ffilllié Session 1 ' 9O 48 42 Session 2 89 47 42 Session 3 89 46 43 Session 4 82 44 38 Criterion Posttest 81 42 39 Criterion Pre and Bosttest 75 4O 35 C ovalidation S le T tal Males F ales Session 1 84 34 50 Session 2 84 35 49 Session 3 83 34 49 Session 4 76 32 44 Criterion Posttest 72 31 41 Criterion Pre and Posttest 62 28 34 The basic measure of SA improvement due to training was the difference between pretest and posttest criterion scores. Such scores were available for 75 Js in the V sample and 62 in the C-V group. Tests of differences between means utilized t-tests, and matched t-tests where appropriate. Reliabilities (internal consistency) were computed for each SA test. Criterion scores were correlated to each other and to improvement, and analyzed according to level of item difficulty. A secondary analysis concerned itself with the differential 37 impact of training. These analyses involved a general correlational study using the Missing Data Routine (MDSTAT) on Michigan State University's CDC 3600 computer. Results Table 3 presents Kuder-Richardson #20 reliabilities for each of the SA measures. The coefficients are based on the total sample which varied, according to class attendance, for each predictor. Table 3. Internal consistency of the stereotype accuracy predictors. P4291212; N 3912211111.... Marriage Test (Criterion) 188 .68 The Case of Morgan Johnson 184 .66 The Interests of Executives vs. Unskilled Workers 181 .66 The Special Interests of Psychologists 180 .77 The Case of the Psychology Professor 160 .68 Below are results in light of the respective hypotheses. Wherever possible, the large volume of data has been summarized and the original data source placed in an appendix. A brief discussion section appears after the presentation to summarize the findings. he 1: A in am vi 1 b e ve in irproving stergprzpe acguracx, Comparison of pretest and posttest scores for the total V and C-V samples indicates agreement on the effectiveness of training in improving SA. Evidence regarding sex differences in 38 improvement was less clear. Table 4 summarizes the findings. Table 4. The results of training. Criterion Variance Criterion Mean Subjects N Pre st Pre D ff t Valigation Sgrple Total 75 35.97 47.86 34.95 37.39 02.44 2.32* Males 4O 31.12 47.61 34.90 36.33 01.43 1.02 Females 35 42.59 46.72 35.00 38.60 03.60 2.26* Crgss-validatipn Sgrple Total 62 33.91 32.69 35.11 37.66 02.55 2.48* Males 28 28.84 27.73 33.89 37.21 03.32 2.34* Females 34 36.77 37.42 36.12 38.03 01.91 1.31 *p<.os While gain for males in the V group and females in the C-V group did not reach statistical significance, there was a trend shown toward improvement which was consistent for all groups. flngthesis 2: The egfectivrpess 2g rrriping varies inversely with initial accuracrI Table 5 shows the relation between initial accuracy and improvement. In all comparisons, pretest scores were negatively Table 5. The relationship between initial accuracy and amount of improvement. Valigption Sgpple Crprr-valigptign Srrple Grin Gain A. Total Pretest -.48** -.54* 3. Men Pretest - . 55**' - . 63‘” C. Women Pretest -. 52** -.40* “p<.01 * p<.05 39 related to improvement scores supporting the hypothesis. The lower a judge's accuracy score before training, the more improve- ment he is likely to show, and vice versa. The results are similar for separate male-female analyses. gxppthesis 3: The effectivgpess pf trrinipg will be reflected more clearly on ptgreorxpe rpppracx predictor items of moder te difficul than on t se 9: either high or low difficulrr, The results of training were analyzed in light of a J's degree of initial accuracy (low, moderate or high) and the level of criterion pretest item difficulty1 (easy, moderate, or hard). Js whose criterion pretest raw scores ranged from O to 31 (N - 38) were classified as having low pretraining accuracy. Pretest scores of initially moderately accurate subjects ranged from 32 to 37 (N - 58), while pretest scores of 38 or higher (N - 41) placed J in the group considered to have started with high SA. Similarly, indices of item difficulty allowed categori- zation of criterion items as easy, those missed by 0-291 of the Js (N - l9); moderate, those missed by 30-46% of the Js (N - 19); or hard, those missed by 470% of the Js (N - 18). Table 6 presents a summary of the analysis. The original data appear in a more cosplete form in Appendix 8 (p. 104). Criterion pretest scores on the easy items declined after training. Scores increased on the moderately difficult items, and on the hardest items as well. The trend was consistent in that all three groups of Js showed greatest improvement on the l a The index of difficulty is the percentage of the total group marking a wrong answer or omitting the item. moderate items and between easy and hard items were statistically significant (p‘<.01), whereas the moderate-hard item difference was not. It would seem that more SA improvement can be expected on difficult items than on easy items. It may be that most ”difficult" items did not tap the extremes of difficulty, as well as the easy items tapped the extremes of ease. The items were initially grouped so that item categories contained approximately equal numbers of items. The data lend qualified support for the hypothesis. Table 6. Summary of the relation of item difficulty to the measurement of improvement due to training. C t n I s 1 fi Initirl Accuracy E erate d Me Total Low .5 3.2 2.4 2.0 Moderate - .3 1.7 1.2 .9 High -l.2 O - .1 - .4 Mean Total -4.3 1,6 1,2 Note:--These data are mean criterion improvement scores. Tests of the significance of the difference between differences for the Tptpl marginals yielded the following results: Easy item score vs. Moderate item score: t - 5.14, df - 272,p<.01; Easy item score vs. Hard item score: t - 4.55, df - 272, p (.01; Moderate item score vs. Hard item score: t - 1.00, df - 272. firpgthesis 4: Stereorype accuragp is a rlrprripn gpecific rpiliry, If SA is situation specific, accuracy in judging one group should be independent of accuracy in judging another. SA scores for 5 situations were intercorrelated to test the hypothesis. The data appear in Table 7. Table 7 shows little consistency in the comparison of V and C-V findings. The presence of relationships statistically different from zero suggested generality, but were not reproduced in cross-validation. The failure to replicate, coupled with the generally small correlation coefficients, offered conflicting evidence with regard to the situational specificity of the ability. Of particular interest was the pattern of relationships observed: statistically significant values occured mainly between scores on instruments requiring predictions about similar groups. Thus, observed generality may be due to the similarity of the tasks involved rather than the generality of the ability. The number of pairs of observations ranged from 67 to 89. Results for men and women were similar to those for the total group. Table 7. The generality-specificity of stereotype accuracy. 4. Vrlidation Srrple - Total Stereotype Marital Morgan Executive Psychologist Professor Pretest Jphnppn Marital Pretest - .23* .21* .12 .21 Morgan Johnson - .13 .31** .24** Executive Worker - .07 -.05 Psychologist - .41** Professor - 8I Cross-valigrtion Bragg: - Tpral Marital Pretest - .20 .03 .34** .10 Morgan Johnson - .03 .32** .18 Executive-Worker - .30* .02 Psychologist - .08 Professor - **p (.01 * p<.05 42 e na i intel e tual b 1 d prtitudrr reward rrpining, Analyses relating personality trait scores with SA criterion posttest and improvement scores found no relation between any of the 5 personality traits measured and improvement due to training. (See Appendix E, p. 105). Correlation coefficients were computed between training outcome measures for the total groups and measures of intellectual ability (Appendix b,p.106). Only the midterm examination grade cover- ing the text material was related to criterion posttest scores in both V and C-V samples. Other correlations statistically different from zero (but not replicated) occured in the V sample when comparing criterion posttest scores with GPA and CQT verbal, information, and total test scores. Improvement was statistically related to GPA and CQT information scores without replication. The results with regard to the relation of intel- ligence and training outcome are inconclusive. Analyses performed separately on males and females yielded a clearer result. Table 8 summarizes the results of the male analysis. Gain for men was positively related to intellectual ability as measured by GPA: (correlated with both criterion posttest and improvement scores in both V and C-V); CQT-Verbal score (criterion posttest); COT-Information score (improvement); COT-Total (criterion posttest). Both measures of training outcome were statistically related to lecture and total final 43 Table 8. Summary of the relationship of intellectual ability to training outcome for men. figst-training Measures Crirerion Posttert lrprovement A. Validation Sample - Males Final exam grade: lecture questions ** ** Final exam grade: total * e0 Grade Point Average COT-Verbal score COT-Information score ** CQT-Total score ** IA“ 1 B. Cross-validation Sample - Males Final exam grade: text questions * Grade Point Average ** COT-Verbal score COT-Information score COT-Total score * I...» ”p < .01 * p (.05 examination scores in the V sample but not in C-V. Inspection of the data for women showed no relationship between training out- come and any of the intelligence measures. Finally, neither confidence in improvement, liking for train- ing, or number of class sessions attended was related to training outcome. Data for these analyses appear in Appendix B (p. 108), S ts S One Study One was largely exploratory, designed to generate hypotheses for further testing, and to serve as the basis for a revised training program. The results of the analyses are sum- marized below and will be more fully discussed in a later chapter. 44 Participation in a training program improved SA. There existed, however, an inverse relation between initial accuracy and training outcome: the lower a judge's pretraining accuracy, the more likely he was to benefit from training, and vice versa. SA predictors had good internal consistency, but moderately difficult and hard criterion items best reflected improvement due to training. Subjects showed either no change or a decline in accuracy on the easy items. The findings about the generality-specificity of the ability were ambiguous. Overall, however, SA tended to be situation specific: judges who were accurate in their predictions of one group, were not necessarily accurate in predictions of other groups. The intercorrelations observed among the predictor pretest scores may have reflected the similarity of tasks involved (i.e., the similarity of the groups to be judged) rather than any underlying continuum of generality. SA improvement scores were unrelated to J's personality, sex, or attitude toward training. For men, gain was positively related to some intellectual factors. For women, improvement was unrelated to intellectual ability. Several limitations of Study One should be noted. First, the frequent failures to reliably replicate results in cross- validation may have been due to wide fluctuations in sample sizes. C-V results were difficult to interpret in light of generally shrinking N's from the V to the C-V analyses. A 45 second, related problem, concerned the degree of sample shrink- age due to the relative inefficiency (e.g., length, lack of posttest) of the SA measures. Study Two was an extension of Study One. Training measures and procedures were modified in light of experience with the pilot program. The next chapter describes the methods and results of Study Two. CHAPTER III STUDY TWO Prgblem Study Two was an extension of Study One. The pilot research was mainly exploratory with the aim of suggesting revised training procedures and hypotheses. The second study tested the revised hypotheses and applied the findings of Study One to the development of a revised training program. The following hypotheses guided the present research. flrpprprrrr_lx Participation in a training program will improve stereotype accuracy. Improvement will be stable over time. 5222£22!l!.33 The effectiveness of training varies inversely with initial accuracy. gppgrprrrr_z: Stereotype accuracy is situation specific, and improvement is similarly specific. flpppthepip 4: Specific training with stereotypes of female Os will generalize to similar stereotypes of male Os. £222£fl££1!.23 Training will result in greater improvement in accuracy when written judgments are not made before training, than when written judgments are made before training. yrpgrrrrrr_g: The effectiveness of training is unrelated to personality and attitude towards training. .5!22£B2£l£_13 Improvement in stereotype accuracy is posi- tively related to intellectual ability for men, but unrelated for women. Gggrrgl Derigp Study Two used a pretest-posttest design with interpolated training for mp stereotype. Subjects were administered a pretest, received training in making predictions about groups or particular group members, and were posttested with the pretest 46 47 criterion measure. Each session formed a complete pretest. training-posttest sequence. Training took place in 3 successive sessions of about one hour and ten minutes each. Subject; The trainees were 182 M50 students, representing a wide range of college majors and levels, who were enrolled in a Winter 1968 section of a Psychology of Personality course. Training, as in the pilot study, was a part of the course work and took place during the regularly scheduled class meetings. Attendance was not mandatory. W The SA predictors used in the present study were greatly shortened from the original forms used in Study One. The revisions were more in keeping with the relative specificity of the ability and, by making possible a pretest-posttest design for each training session, avoided the wide variation in sample sizes. Item analyses of the Study One measures identified a core of items covering the whole range of item difficulty levels for each instrument. Table 9 presents the original and revised item difficulty data. While items were chosen to represent the continuum from ”easy” to ”hard”, the mean item difficulties (in terms of the average percentage of trainees answering an item correctly) was relatively similar for the original and shortened test forms. The revised paper-and-pencil tests of SA appear in Appendix Table 9. Summary of item analyses for original and revised stereotype accuracy predictors (based on the Fall, 1967 data). P e t 2221.22: to tud ... The Case of the Psychology Professor (N - 160) 76.9 40 74.4 16 The Special Interests of Psychologists (N - 180) 65.4 60 58.5 16 The Case of Morgan Johnson (N - 184) 57.8 64 53.2 20 Marriage Test (N - 188) 61.6 56 61.2 28 Exercise A. Men 58.6 28 56.5 16 Exercise 8. Women 64.4 28 66.0 12 C. An 18M scoring sheet, provided for recording judgments, allowed machine scoring of all instruments. The following revisions were made. 1. The 2-part Case 2; pp; Psrgpplogp grpgesrpr became two separate exercises in the revised form. The first exercise, 2h: I:plgp1_§pllpgp;flpp,(CM-I), asked J to predict the group norm of typical college men in responding to a personality inventory. Both a pretest and a posttest were administered on 16 items which required J to: Mprk_:12,on the separate answer sheet if you think the typical college man answered "£222? to the statement; flprk_2g:,if you think the typical college man answered ”grippr. 16 training items, ordered from the least to the most difficult, were provided for practice and feedback to be interpolated between the pretest and posttest. 2. The second exercise called for the W W (CM-11> on the was 16 its- 49 used in CM-I. The same training items are also used between the pretest and posttest as were administered in CMpl. J was informed that most people conform to a stereotype in some respects and differ in other respects. J's task was to consider how a particular professor conformed to and was different from the group norm of college man in general. For each item J was to: m on the separate answer sheet if you think the p;g£gggg;_answered ”true” to that particular statement; Mark ”2" if you think he answered ”false”. In this manner, J demonstrated his ability to predict group norms and to apply his ability to predictions about a unique individual. 3. W 0-1) cam-cm .... the shortened and revised form of T e S s f P st test. The instructions were similar to those for the original test form, except that J took both a pretest and a posttest on 16 items designed to differentiate between psychologists and men-in-general. 16 interpolated training items ordered from least to most difficult gsve opportunity for practice. 4. The revised form of the Morgan Johnson test was an gpglication f the Ste e P Ps c 1 st (P-ll). Instructions on the revised form were similar to those for the original test. While the P-I measure described above required the trainee to differentiate between psychologists as a group and the group of men-in-general, P-ll required J to apply the stereotype to a particular psychologist both as a group member and as a unique 50 individual. 20 items were provided for use as a pretest and posttest, and 16 additional items ordered by difficulty level, served as training items. 5. The M!;1§§1_§£gggg§xggg exercise was the revised (28 items) form of the pilot criterion instrument. The instructions for the revised measure were similar to those for the original predictor. The shortened version, however, allowed a pretest and a posttest, in a single session, on 16 items aimed at differentiation between happily married, unhappily married, and divorced m (MM). and 12 items aimed at similarly differentiating between woman (MW) in these merital groups. 12 training items were included on the test form to provide practice in, and feedback about, applying the stereotypes to the women marital groups. The training items were arranged so that J proceeded from the easiest to the most difficult items. 1W Six training procedures or conditions were used in Study 1. a program involving diagnosis of error tendencies, practice in making judgments, feedback, and discussion; 2. a pooling technique; 3. a panel technique; 4. a trait method; 5. a noncommitment training condition; and 6. a base-rate condition. 51 It will be useful to view the techniques as stressing allied, but different aspects of training. First, aiagaaaia, agaatice, feeg- WM stressed the basic training principles: St n t e ve bad a n c n f e b e. In the present study, diagnosis was less formalized than in the pilot program. Diagnosis involved demonstration to J, on the training items, that inaccurate stereotypes result in poor prediction of group characteristics, and that inflexible stereotypes result in inaccurate predictions about individual group members. In addition, at the start of each session, Js were instructed to recall their performance on the previous session's exercises as a way of assessing their progress in SA improvement. Practice, feedback, and discussion were further system- atized. Training items were completed in series of, for example, 4 or 5 at-a-time followed by feedback and discussion, completion of the next 4 or 5 items, and so on until all training items were answered. Second, the methods described below emphasised the process of acquiring information about groups and the psychological organization of the information to make for more accurate judgments. The W described in the Training Procedures section of the Study One chapter, stressed the training principle: Uae 102; kaagledge a; similar gaggle (Smith, 1968a). The aanel method applied the principle: Liagaa Ea what ataeas 52 ve t be e ti n . A panel of 4 to 6 student volunteers sat in front of the classroom. The panel members completed a series of training items and one member read and explained his predictions. Panel members then discussed their predictions, particularly disagreements. The class members as a whole then made their judgments on the series of training items and discussed the responses. The discussions were led, at all times, by the panel. The role of the trainer was the purely nondirective one of reflecting the proceedings. The correct answers were read and the panel proceeded to the next series of training items. The W encouraged the J to: S e ici 5W. Js were taught and encouraged to use a trait theory involving the dimensions: emotional-calm, introverted- extroverted, impulsive-controlled, rational-empirical, and cautious- bold. As a part of training, each J took a personality test and received a graphic profile of his own trait scores. Further, Js were told the relative position of the group being judged on the continuum of, for example, introversion to extroversion. The J then used the information and the theory to gaacribe the group being judged and to gaaaaga the judged group with other groups. Under the n t ndit n, Js were instructed to: De f t n e ns. Js whose last names began with the letters "A” through "C" were given the some training materials as the "H" to "Z" Js, but did not answer the pretest 53 items. ”H“ to ”2” did answer the items. All Js participated fully, however, in the training. The noncommitment condition was used only with the P-1 and P-II stereotypes. Lastly, the W, used only with P-II, told J the percent of men-in-general who expressed a liking for a particular interest vs. the percent of thezmembers of a group (to which the 0 belongs) who expressed a liking for the particular interest. Again, the Js were instructed in the naaa £2; glaaibiligy in applying a stereotype to an individual group member. While * psychologists in general say they like a particular interest, any single psychologist might dislike it. WW One or more of the training procedures were combined with each of the SA measures to comprise a single training session. Table 10 summarizes the content of the 3 meetings devoted to SA. Table 10. Summary of training activities in Study Two. W e e 2 4 6 5 College Men I x x, x x x x 5 College Man II x x x x x x 6 Psychologist I x x x x x x x 6 Psychologist II x x x x x x x x 7 Marital-Men x x 7 Marital-women x x x x x x *Note: (1) - diagnosis, practice, feedback, discussion; (2) - pooling; (3) - panel; (4) . trait; (5) - noncommitment con- dition; (6) - base rate condition. As in Study One, the revised program was a part of an overall program designed to improve the understanding of others. SA 54 13111111118 occurred in the 5th, 6th, and 7th sessions of the broader program. In the 20th session, a random sample of 68 33 again completed the CM-II, P-I, M-M, and M-W tests. W. -- Js took test booklets (CM-I and CM-ll) when they entered the room. Following a lecture on the nature of stereotypes and SA and an introduction to the basic training principles and methods, they completed the 16 item CMeI pretest on a separate answer sheet and were introduced to the pooling and trait methods. The norms for college men and women on the personality trait dimensions were graphically depicted. (The norms for the personality inventory appear in Smith, 1968b.) The volunteer panel came to the front of the classroom and the train- ing items were completed in series of 4 items each. Following training, the Js completed the 16 item CM-I posttest on the separate answer sheets. CM-II was now taken up with a procedure similar to that for CM-I. Iaainiag aessian 2. -- Js took a P-I and P-II test booklet. Students with last names beginning with the letters ”A" through ”G” sat on the left side of the classroom and dig gat take the P-I pretest (noncommitment condition). The remaining Js com? pleted the P-I pretest. The class members as a whole were given instructions for pooling and the empiricism of psycholo- gists was emphasized. Martin Luther King and David Hume were contrasted (Smith, 1968b) as representing opposite ends of the rational-empirical continuum. The panel came to the front of the classroom and answered the training items 8 at-a-time. 55 All Js completed the training items and the P-I posttest. Procedures for P-II were similar to those for P-l. Training session 3. -- Training materials were distributed as Js entered the classroom. The principles of SA and of training were reviewed. Protests of flggital Stereoszggg of men and women were completed on the separate answer sheets and the panel assembled to begin training only on the items pertaining to stereotypes of women. Js used the pooling technique again. The relative positions of the marital groups on the emotional-calm and introversion-extroversion trait dimensions were graphically presented on the blackboard. The training items for women were completed in 2 series of 6 items each. After training the subjects completed both the MM and MN posttests. Additional Vagigblgs In addition to the SA measures, data were collected to explore the relationship of: (l) J's personality; (2) his intellectual ability; (3) his attitude towards training; and (4) his sex, age, and scores on a human relations inventory, to training outcome. The personality variables were those measured by the Prote- bob Personality Inventory described in Study One. As in Study One, measures of intellectual performance included scores on multiple choice midterm (56 items) and final (88 items) examinations. Half of the items on each test were devoted to the text material and half to the lectures (training). 56 A total text and lecture score was available as the sumnof the appropriate midterm and final exam.scores. A total score com- prised the sum‘of the whole midterm and final exam scores. Simi- larly, Js' CQT Verbal, Information, and Total scores were recorded, as were cumulative grade point averages (GPA) up to, but not including the Winter 1968 quarter. Attitudinsl factors were measured by: (l) J's ranking of the Psychology 225 course as described in.Study One, and (2) a count of the number of SA training sessions attended. The demographic variables included the Js' sex.and age. Finally, Js filled out W (Dore, 1960). It contains 60, 2-alternative, forced-choice items designed to measure attitudes toward 2 leadership methods: consideration (C) and responsibility (R). The inventory yielded 3 scores: C, R, and a total score. Items 1-30 required a choice between a less and a more considerate statement. Persons choosing a greater number of considerate items tend to think it more important to be employee-oriented than workporiented; subjects who mark more inconsiderate items, work-oriented rather than employee-oriented. Items 31-60 required a choice between items that are high or low in reponsibility. Persons scoring high on R generally feel that a superior should play a role different from that of his subordinates, spending more time in activities his followers are not able to do: e.g., organizing, planning, providing information, etc. (Smith, 1968b). 57 Method of Analysis 182 subjects took part in some phase of training. Since attendance was not mandatory, the number of Js varied from instrument to instrument. Table 11 shows the number of Js who attended each training session. In all, 116 Js attended all 3 training sessions. 61 were men; 73, women. Table 11. The number of subjects who completed the revised stereotype accuracy training program. Sub s S St an e ot F les 1 College Man I 157 56 101 College Man II 157 56 101 2 Psychologist I 107 61 66 noncommitment 51 20 31 Psychologist II 101 37 6A noncommitment 55 23 32 3 Marital-Men 138 52 86 Marital-Women 138 52 86 Retest CH-II,P-I,H-H,H-W 68 24 44 Kuder-Richardson #20 reliabilities were computed for the shortened criterion instruments. The basic measures of improvement were differences between pretest and posttest criterion scores available for 3:29 stereotype. Tests of differences between means used t-tests and matched t-tests where appropriate. Correlations were computed on Michigan State University's CDC 3600 computer using a prepared routine (MDSTAT). Criterion scores were intercorrelated and related to the additional variables. 58 Results Table 12 presents the reliabilities of the revised SA ‘measures. The revised reliabilities are lower than those for the pilot instruments. The tests used in the second study were greatly shortened from the originals, and contained items chosen to represent the entire available range of content and difficulty and discrimination levels. Table 12. Reliabilities of the shortened stereotype accuracy predictors. Reliability N KR£20 The Typical College Man (College Men I) 157 .33 The Application of the Stereotype to a Particular Man (College Man II) 157 .14 The Typical Psychologist (Psychologist I) 107 .56 Application of the Stereotype to a Particular Psychologist (Psychologist 11) 101 .15 Marital Stereotypes - Men Exercise 138 .44 Marital Stereotypes - “omen Exercise 138 .26 As with the pilot program, the data for Study Two are pre- sented in relation to the relevant hypotheses formulated to guide the study. Similarly, wherever possible, the data have been summarised and the original data placed in Appendix D (p. l20ff). : c i w v v t will be stable gvg; gig, Table 13 summarises the results of training for the trainees as a group. Separate analyses of male and female data yielded similar results (see Appendix D, p. 121). Training did not affect scores on the CM-I or CM-Il measures. Improvement due to training was statistically significant, however, for the stereotypes of psychologists and of happily married, unhappily married, and divorced 59 men and women. Additional analyses (Appendix D. P. 122), yielded no reliable differences between men and women in either accuracy before training or the impact of training. Moreover, when Js were retested on 4 stereotypes (CM-II, P-l, M-M, M-W), 7 weeks after training, there was no decline in accuracy (Appendix D, p. 137). Training impact was stable. The results endured over time. Table 13. The results of the revised training program for the total sample. Stereotype N Pretest Posttest Difference t Mg M029 College Man I 157 13.69 13.45 - .24 1.09 College Man II 157 10.19 10.23 .04 .20 Psychologist I 107 10.20 11.42 1.22 3.49** Psychologist II 101 10.88 11.99 1.11 3.70** Marital-Men 138 9.50 10.84 1.34 5.36** Marital-Women 138 7.45 9.58 2.13 3.74** the 2: Th e fe s v“ e e W Table 14 shows the relation between pretraining accuracy and improvement. Table 14. The relationship between initial accuracy and amount of improvement for each revised criterion. §gb1e§§§ gsgrgggypg__ T t G G H G in College Men I Pretest -.31** -.40** -.27** College Man II Pretest -.40** -.29* -.45** Psychologist I Pretest -.59** -.60** -.58** Psychologist 11 Pretest -.53** -.49** -.57** Marital-Men Pretest -.61** -.70** -.56** MaritaldWomen Pretest -.17 -.59** -.13 “p< .01 * P( .05 60 It was found that initial accuracy was negatively related to gains in all comparisons. The results offered support for the hypothesis. : 5 er t e accurac is i uat n ci ic d i ovemen i s milarl s ecific The interrelation of pretraining, posttraining, and improve- ment scores between all six criteria (Appendix D, p. 123) yielded few statistically reliable relationships. Out of 135 correlation coefficients only 7 were statistically significant at the .05 level. Seven statistically significant relationships out of such a large number could certainly be expected by chance alone. These data support the hypothesis that SA and SA improve- ment are specific. The average correlation for 135 comparisons for the total sample was .09. The average intercorrelations were .14 and .10 for the male and female data respectively (Appendix D.. a 123ff). flyggthesis 4: Specific training with stereotypes of female 9; will generalige to similar stereosyges of gale 0sI The data in Table 13 (page 59) are relevant to the present hypothesis. Js were pretested on MM and MW stereotypes, trained on MW stereosypes gnly, and posttested on both MM and MW stereotypes. As shown in Table 13 gains on both MM and MW stereotypes were equally statistically significant. Had there not been generalisation, these results could not have occurred. Additional support for the hypothesis was derived from separate 61 analyses of the male and female data (Appendix D, P° 122) which obtained similar results. Hyggthesig 5: Tyaining will result in greater improvement in gggugggy when written jugglents are not zgde before training. than when written judg- ments age ggde before training, For P-1 and P-II only, an experimental group of Js did not formally commit themselves to judgments before training, but took a written posttest. Control Js wrote their P-1 and P-Il predictions on both a pretest and a posttest. Table 13 (page 59) showed that the control group improved significantly on the P-1 and P-II measures. DIF accuracy on P-I and P-II was not, however, reliably different from the posttest accuracy scores of the total, male and female control groups. Table 15 summarises the results of delayed impression formation (DIP). Additional data appear in Appendix D (p. 126). Table 15. Summary of the results of delayed impression training. Stergggype Coggarisons of Means Diff. df t Psychologist I Total group posttest vs. DIF .07 156 .17 Male posttest vs. 01? -.49 59 -.77 Female Posttest vs. DIF .42 95 .78 Psychologist 11 Total group posttest vs. DIF .06 154 .15 Male posttest vs. DIF .68 62 1.10 Female posttest vs. DIF -.30 94 -.60 ”N .01 * p( .05 The data in Table 15 lend no support for Hypothesis 5. Additional information about the relationships of DIF accuracy and the additional experimental variables are contained in Appendix D ~.126f). 62 n1 dat tude w t ni The present analyses separately correlated SA posttest and gain scores with measures of personality, age, and training and leadership attitudes (Appendix 0, pp. 131 - 136). Analysis of the combined male-female data revealed that more emotional Js show greater gains on the P-I test («524. P< .05), while the calmer Js score higher on the P-11 posttest (.21, p(.05). The more cautious Js do better on the m posttest (-.18, p (.05), but ispulsive Js show more imrovement (-.l9, p<.05) due to training. Among men, emotionality was related to gain on P-I (-.39. P( .05); and caution with iaprovement on P-II (-.43, p (.01). MW posttest scores were higher for controlled and emirical the impulsive (.38, p<.01) and rational (.31, p<.05). Among women, introversion was related to posttest scores on CM-II (c.22, p (.05), while calmness was related to P-II posttest scores (.33, p ( .01). Bolder women scored lower on the MM post- test than the more cautious (-.22, p (.05). Gain on the m test was greatest for imulsive (-.27, p(.05), cautious (-.35, p (.01) women as cornered to controlled, bold women. Furthermore, male and female correlation coefficients differed statistically for the following comparisons: -P-II posttest vs. cautious-bold scores: for men, -.43; for mm, .11. (Z I- 2.59, p (.05); em improvement vs. cautious-bold scores: for men, .20; for women, -.35. (Z :- 3.35, p (.01); 63 -MW posttest vs. impulsive-controlled scores: for men, .38; for women, -.09. (Z - 2.76, p(.01). With regard to the age and attitude variables for the total samle, age was negatively related to gain (-.20, p(.05) on the MU test. Liking for training (course rank) was unrelated to training outcome. Attendance at SA sessions was reliably related to P-II improvement (.23. P‘(.05). Finally, Responsibility (R) leadership scores were negatively correlated with gain on 04-1 (-.28, p(.01) and MM (-.25, 9<.01) tests. Both MM posttest (-.34) and gain (-.35) scores correlated negatively (p (.05) with the ages of male Js. MW posttest (-.36) and gain (-.36) scores correlated negatively (p( .05) with R scores. Attendance was positively related to CM-II improvement (.27, p<.05), but inversely related to P-I gain (-.45, p (.01). For women, CM-I posttest (-.25) and gain (-.29) scores were negatively related (p( .05) to R scores, and CM-II posttest accuracy correlated positively with the total leadership inventory scores (.29, p(.05). P-l improvement was related to R (.28, p(.05), and MW improvement (.26, p(.05) to attendance. Personality, age, and attitude, it appears, may be related to gain, but in such a specific manner that the initial hypothesis may not be relevant. thesi 7: I vem in te e ccu s- tively {elated ta intellectual ability for aanl but aaaelated for yamen, Table 16 sumarizes the relation between training outcome and measures of intellectual performance. Total group CM-II 64 Table 16. Summary of the relation of training outcome and intellectual ability. T College Psychologist Psychologist Marital Marital Man II I 11 Women Women t Gain G est Gain Midterm: text ** * ** Midterm: lecture ** * * Midterm: total ** ** ** Final: text ** Final: lecture * * Final: total ** * Text: total * ** * Lecture: total * ** * Total: total * * ** ** CPA * * COT-Verbal * CQT-Information 29219.22 * * * no College Psychologist Psychologist Marital Marital Man II I 1 Women Women P e est G n Midterm: text * ** Midterm: lecture * ** * Midterm: total * ** Final: text * * Final: lecture * * ** Final: total * * ** Text: total * ** Lecture: total * ** ** Total: total ** ** GPA * CQT-Verbal * COT-Information ** ** CQI-Tota; ** ** wees: College Psychologist Marital Marital Man II I Men Women P tt s G n G st e t Midterm: text ** Midterm: lecture Midterm: total ** Final: text ** Final: lecture * Final: total * * Text: total * Lecture: total * Total: total * GPA * QQI-Veybal * ”p < .01 *p < .05 65 posttest scores correlated positively with the 3 midterm exam scores, the 3 composite exam measures, GPA, and CQT-V and Total scores. P-I gain was inversely related to Total-total and GPA. P-II gain was inversely related to CQT-Total. MW posttest scores were related to all intellectual measures except GPA and CQT-V and 1 scores. MW gain was related to all intellectual measures except final text subscores, GPA, and the CQT measures. For men, CM-II posttrained scores correlated positively with final exam lecture subscore and total grade, and with the composite lecture scores. P-I posttest was related to midterm lecture subscores, while improvement on the test was negatively related to CPA. MW posttest scores were correlated with all measures except CPA; gain scores were related to all intellectual measures except GPA and CQT-V. For women, CM-II posttest scores correlated positively with midterm text and total grades, and with CQT-V scores. P-I gain was inversely associated with final exam lecture and total scores and with composite lecture subscores. MM gain was similarly negatively related to GPA. MW posttest scores varied positively with final exam text and total grades as well as with the composite measures of text and total exam performance. Furthermore, tests of the significance of the difference between correlations (male vs. female Js) of MW posttest scores and CQT-V (Z - 2.12), CQT-I (Z - 2.24), and CQT-T (Z - 2.53) scores were statistically significant at the .05 level. Similar tests 66 compared male and female correlations of MW gain scores with final exam test (Z - 2.24), lecture (Z - 2.24), and total grades (2 - 2.06), and composite text (Z - 2.24), lecture (Z - 2.18), and total grades (2 - 2.06). Each of the comparisons differed at the .05 level. MW-CQT-I (Z - 2.76) and MW-CQT-T (Z - 2.88) correlations for males and females differed at the .01 level. In all of these analyses, the correlations between intellectual factors and improvement were larger for males than for females. SEIBIEZ af tha Resulgs af Stuay TwoI The revised instruments of Study Two were greatly shortened from those in the pilot project. Fewer items allowed a pretest- training-posttest design in each training session. The relia- bilities of the new measures declined, however, (except for P-I which increased) even when corrected for length by the Spearman- Brown formula. Participation in a training program did improve predictive accuracy with regard to the stereotypes of psychologists and married and divorced persons. Training did not affect accuracy in making predictions about college men. Male and female Js did not differ in either pretraining accuracy or in the benefits derived from training. Moreover, improvement due to training was stable. The lower a J's pretraining accuracy, the greater the benefit from training, and vice versa. This inverse relationship between initial accuracy and gain may account for the failure of training to improve the already high CM accuracy scores. 67 The overall results point to the relative specificity of the stereotype accuracy ability and its improvement. This finding may make the original hypotheses no longer relevant, or at best, difficult to interpret. A J's accuracy in predicting about one group does not necessarily mean that he will be accurate in his predictions about other groups. Inaccurate prediction about one group, on the other hand, did not preclude accuracy in judging other groups. Similarly, improvement in prediction about one group did not necessarily improve prediction about other groups, nor did failure to improve in accuracy about one group mean that a J will not benefit from training on some other stereotype. Training on stereotypes of women generalized to stereotypes of men on the Marital Stereotypes exercise. There was no difference in improvement between judges who make written predictions before training vs. those who delay the formation of impressions. An additional analysis explored the correlates of the delayed impression formation (DIF) accuracy scores. Turning to the correlates of posttraining accuracy and gain, specificity is once again an important factor. Total group P-II and MM posttest scores were related to calmness and caution, respectively. Gain on P-1 and MM were associated with emotionality and impulsiveness, respectively. Controlled, empirical men scored higher on the MW posttest, emotional men 68 showed greater P-l gain and cautious men improved more on P-II. Introverted women had greater CM-Il posttraining accuracy; calm women, on P-11; and bold women on the MM exercise. Impulsive, cautious women showed greater MM improvement. The following sex differences were statistically significant: caution was related to P-Il posttest scores for men, but unrelated for women; control was related to men's MW posttest performance, but unrelated for women; and gain in MM accuracy tended to by higher for gala men, but was higher for cautlaus women. Age for the total sample was negatively related to MW gain; for male Js, negatively related to MM posttest and gain; and unrelated to women's performance. Liking for training was unrelated to training outcome. Attendance at SA training sessions was reliably related to P-II gain for the total group; to CM-II and P-II (negatively), for men; and to MW gain for women. Leadership inventory R scores correlated negatlvely with total group CM-I and MM gain, with male sample MW posttest scores and gain, and with female sample CM-I posttest scores and gain. R was positively correlated with female P-I gain. Total leadership inventory scores were positively related to CM-II posttest scores for women Js. With regard to statistically reliable sex differences, younger men scored higher on post- training measures of MW stereotypes; age was unrelated to women's performance. Attendance was unrelated to P-I gain for females, but negatively related to male performance. 69 As with other measures, intellectual ability also appeared to be related in a specific manner to SA. Scores of both male and female Js were related to some measures of their intellectual performance. Intelligence was a more salient aspect of male accuracy than for that of female Js. Male Js' overall college academic ability varied positively with MW posttest accuracy, but female CQT scores were unrelated. The difference between male and female correlations was statistically reliable. MW improve- ment was greatest for men with highest course achievement as measured by three final exam scores and three composite examination scores. The male-female differences between correlations were again statistically significant. CHAPTER IV DISCUSSION Results of the Training Studies Table 17 summarizes the results of the tests of the hypotheses. Table 17. Summary of the results of the tests of the hypotheses. Hypothesis Outcome ud One Stud T Training will improve SA. ee The amount of improvement varies inversely with initial accuracy. 40 ee Moderately difficult criterion items best 0. reflect improvement. 9 NA SA is a situation specific ability. e 99 SA improvement is situation specific, but will generalize to opposite sex predictions. NA 94 The delay of impression formation will enhance improvement. NA 0 The amount of improvement varies with Js personality and intellectual attributes and with his attitude toward training. 1 ? ee - strongly confirmed; e - confirmed; 0 - rejected; 7 - inconclusive; NA - not applicable. The most significant finding in light of the number of training programs reporting little or no success in improving accuracy, was that both of the present studies improved the SA ability. Explanations for the success of the present program lie in the nature of stereotypes and the training activities, and in the interaction of the two. Neutrally regarded, stereotypes are concept systems which 70 71 organize our experiences much as do other concepts (Vinacke, 1957). But while in the course of normal development we are given con- tinual and disciplined guidance in the application of many concepts, the inevitable categorization of people, which we call stereotyping, is relatively de-emphasized or becomes subtle and more complex. The present program provided some belated concept training. But why was SA training successful while attempts in other areas of sensitivity have failed? Little success was recorded by the programs that were vague as to both the content and process of understanding: these have mainly been the studies equating training and ”education” (Fancher, 1967; Goldberg, 1968; Smith, 1966; Taft, 1955). Similarly, little success was realized when the arocess of understanding was emphasised to the exclusion of content (e.g., Goldberg, 1965). T-group training has consistently stressed process with little demon- strable improvement in interpersonal sensitivity (Campbell and Dunnette, 1968). Programs reporting some success, however, stressed both the content aaa process of understanding and have combinaa training procedures and conditions (Dailey, 1966a; Grossman, 1967; Jecker, a£_aly, 1965; Wakeley, 1961). The present studies, too, combined specific, relevant training procedures and conditions, and to a greater extent than in previous programs, emphasized both the content (the nature of stereotypes, SA, and understanding) and process (practice 72 and feedback, etc.) of sensitivity. All of this was done, also, with the most effective criterion and training instruments devised to date. Thus, practice and feedback in the present study provided the chance to test the appropriateness of present concepts and progress towards improvement, while diagnosis gave information and motivation. Wakeley's (1961) pooling technique encouraged the development of stereotypes and introduced Js to the usefulness of base-rates. The base-rate condition was formalised in Study Two to emphasize the need to move from the instance to the class and vice versa. Modification of the pilot predictors and training prompted the dropping of the empathic technique in Study Two with no apparent loss in the power of the program. The explicit, empirical personality trait theory criticized by both Grossman (1967) and Smith (1968a) was appropriate and useful in SA training since personality and intellectual trait names form the basis of stereotypes. The discussion technique of "Why did you answer like you did?” proved unwieldy with the larger group in Study One and gave way to the panel method in Study Two which was a good compromise and at the same time provided information regarding the social reality. The success of the present program lies, then, in an interaction of the nature of stereotypes and the training procedures. The data supported previous findings of an inverse relation- ship between pretraining accuracy and improvement (Goldberg, 1965; 73 Kepes, 1965; Large and Diamond, 1954). The phenomenon, possibly regression to the mean or a ceiling effect, occurred on the College Man (Study Two) tests: the already high pretest scores virtually precluded significant gain. The generality-specificity controversy in psychology extends also to the area of sensitivity. Empirical evidence has been presented by both the generalists (Cline and Richards, 1960; 1961) and the specificists (Crow and Hammond, 1957) to support their respective positions. The results of the present research support the latter group. The Study One data were cautiously interpreted as supportive of the specificity of SA (there were no consistent findings from the V to the C-V groups). But the power of the test was suspect owing to the degree of over- lap of the stereotypes tested. Consider, for example, the tasks involved in each of the pilot training instruments: Morgan Johnson - differentiate between a psychology student and men in general; Executive Test - differentiate between professional and unskilled men; Psychologist Test - differentiate between psychologists and other professional men; Professor Test - differentiate between a psychology professor and college men in general. Thus, the evidence for generality may simply reflect similarity of the tasks. The only relatively "pure" stereotype was that of psychologists and this predictor also had the greatest internal consistency. The stereotype for Morgan Johnson and the professor were obscured by other group memberships. It is not clear, for example, whether Js' predictions about Morgan were based on a 74 "student” or a ”psychologist” stereotype. It is not surprising to find statistically reliable but irreproducible relations between a task requiring predictions about ”average men" and “unskilled workers”, or ”professionals", "executives”, ”psycholo- gists”, and ”college men”. The revised SA measures (Study Two) produced less stereotype overlap (the Marital Stereotype test, for example, further separated out males and females) and provided strong support for the specificity hypothesis. SA improvement was also found to be specific. The hypothesis that delay of impression formation will enhance improvement (Dunnette and Metal, 1968; Smith, 1966; 1968a) was not supported, but with some doubt about the conclusions to be drawn. The nature of the experimental condition (no written judgments before training) may not have slowed down the judgment process, but may simply have hindered the extent to which a prediction was ”fixed”. While such findings are certainly relevant, it is more appropriate to limit generaliza- tions to the noncommitment condition itself. In this present study, noncommitment neither enhanced nor detracted from improve- ment: posttraining accuracy reached the same levels for both experimental and control Js. Cough (1968), Opsahl (Dunnette, 1968), and Oskamp (1965) found no relation between Js' confidence in gain and actual training outcome. The data from the present studies are consistent with their findings. 75 Moreover, the students in the present studies consistently ranked the subjective value of the Psychology of Personality course about 3 on a scale from 1 to 5: that is, the average student felt that the course was about as valuable to him as the average course he was taking. In another program, adult employment agency administrators unanimously rated SA training as valuable, relevant, job related, and so on (Spier, in prepara- tion). Results: no relation of training outcome to liking for training. Implications for Training to Improve Sensitivity The results of the present studies had implications for 3 broad aspects of sensitivity training: the improvement of training design, selection of trainees, and evaluation of training outcome. Haw aan training he iaaraved? While the optimal weighting of each remains to be determined, future sensitivity training programs should include both lectures and participative exerercises. It is not enough to confront an individual with his shortcomings as a judge of people (process). Some cognitive input is required (content) to enable J to assess and integrate the relevancy of his training experience and to generalize from the training setting to other arenas of inter- personal interaction. Similarly, the design of Study Two, in which each training session was a complete unit in itself proved valuable in terms of research methodology and helped to maintain 76 Js' interest by providing immediate feedback and closure. Such a future design is also relevant in light of the relative specificity of the ability and its improvement. In Study One, the SA principles were sufficiently general so that marital stereotypes became more accurate even though Js were trained on other stereotypes. Study Two demonstrated that SA principles and procedures may be applied to a wide variety of stereotypes, but that not only the SA ability, but SA improvement, was specific: Js who improved in accuracy on one stereotype, did not necessarily gain in accuracy on other stereotypes. Thus, future programs should apply the general principles to the development of understanding of specific groups. The most relevant question for future programs to ask is "Who do we want to know what about whom?” (Smith, 1968a). The answer may well be that the psychologist wishes to understand his client; the manager, his employee; the college administrator, his restless students; the salesman, his customer; and so on. The empathic technique proved of little value in SA training and may be dropped in future SA programs as a separate condition. But the trait method should be included even though it has not proven successful in other areas of sensitivity training. The panel method, while seemingly a good compromise for discussion considering the size of the training groups, should not supplant the need for smaller training groups. Most importantly, combin- ations of training procedures were most effective. Additional 77 techniques may prove useful. Goldberg (1968), for example, likened the acquisition of understanding to the research process: we should constantly be forming our impressions of others as hypotheses to be tested and revised and rejected. A trial train- ing session using an hypothesis approach showed promise. Thus, SA was improved in the present studies, and the task of the future may well be the search for new techniques and technique combinations that will further maximise outcome. Finally, except for the diagnosis, little attention was given the trainees' motivation to improve. Some findings sug- gested that this might be a serious omission. While a ceiling effect might explain the failure of moderately and highly accurate Js to gain on initially easy items, low accuracy Js showed no improvement on the low difficulty items even though they showed the greatest overall improvement as a result of training. Consider, however, that each J enters training with some ”picture" of himself as a judge of people, and that Js have varying degrees of confidence in their numerous predictions. Thus, judgments perceived as easiest (i.e., those having the highest subjective probabilities of correctness) should be the least amonable to change. What, then, was the effect of diag- nostic feedback which informed J of his actual prowess? The Marital Stereotype; test was revised into 4 series (matched for item difficulty) of la items each. As a part of SA training, adult employment agency administrators estimated 78 the number of items they expected to answer correctly before training and received feedback about actual performance after each series. Js unanimously overestimated their actual performance for the first lb items, and as a group, drastically reduced their estimates for the second series of items. Thereafter, two trends emerged: some Js consistently set higher goals for them- selves than their just previous performance; the remaining Js consistently set their goals lower than their just previous performance. The former were superior in SA throughout, and showed signficantly greater gain in accuracy. A second study replicated the results (Spier, in preparation). Failure to improve, then, may result from Js' focus on predicting their own ability when the task at hand is actually the predictions about others. Conclusion: sensitivity training programs should attempt to measure (perhaps through level of aspiration), interpret, and deal openly with Js' feelings about their ability as judges of others. Some perspective needs to be supplied to motivate J to predict for O and to prevent a possible defensive resistance to training. W The findings that SA and its improvement were specific suggested that virtually anyone could benefit from training on relevant stereotypes. Similarly, both men and women showed comparable accuracy gains. But additional data suggested greater complexity. A study of the correlates of SA improvement showed them to be similarly stereotype specific. That is, 79 personality and intellectual factors were not related to SA improvement in a general way. Rather, J's specific personality and intellectual traits were related to improvement on specific stereotypes as the programs were presently conducted. The overall implications were not entirely clear, but further research is certainly necessary. Interesting implications exist should personality and intellectual factors turn out (on cross-validation) to be stereotype specific. Perhaps, to obtain optimal results from an SA training effort, tailor-made training designs may be indicated for groups that differ in personality and/or intellectual make-up. The specificity issue was also relevant to the finding that training had its greatest impact on initially inaccurate Js. Since SA and gain were specific, diagnosis of pretraining accuracy may serve as a screening device in future programs. Motivation may be a greater problem, for example, among initially high scoring Js. Also. the costs of interfering with an already effective judging system may outweigh the slight benefits that accrued to the accurate Js. w’ ul t u b ev l e 7 The criterion problem is an issue in evaluating the effective- ness of sensitivity training as it is in all areas of psychological research. but, perhaps are than in most areas, there has been a tendency to accept anecdotal evidence and subjective feelings about sensitivity as indications of training success. Based 80 on the present findings, Js' self reported confidence in improve- ment and liking for training should not be used as criteria of training effectiveness. Clearly, the problem has been greatest for the T-group training efforts, but the search for adequate criterion and training measures is no less acute in studies similar to the present one. An exploration was made of the relation of training impact and criterion item difficulty in Study One. The data supported the hypothesis that moderately difficult items best reflected training outcomes. But items at each level of difficulty measured a unique aspect of the total training impact. Moreover, there was no difference between outcomes on the moderately difficult and hard items. Only the easy items showed no change. Predictions about others, however, range from the easy to the difficult in reality. Conclusion: future sensitivity predictors and training instruments should contain items at all levels of difficulty. Implications for Future Research Emphasis on both the content and process of understanding used combinations of training techniques and conditions and improved SA. Future studies need to explore the relative con- tribution made to the total outcome by each of the procedures. Training has improved SA, but future programs should seek to maximise the impact through the design of new programs and evaluation and redesign of existing paradigms (e.g., additional concern for J's motivation to improve); development and use of 81 new training procedures (e.g., the ”hypothesis technique”); and continued evaluation and revision of criterion and training measures. An additional question of interest concerns the extent to which training results endure over time. The most serious shortcoming of the present research was the lack of adequate control groups. The absence of control groups limited the generality of the present results despite the fact that previous studies using similar techniques have shown no improvement in control groups. Future studies should include adequate control groups. An exploration of the correlates of improvement showed the stereotype specific nature of the personality and intellectual factors measured. The importance of the specificity issue suggested a need to cross-validate the findings. ”Purer" (i.e., less stereo- type overlap) SA measures may help resolve the question of whether the observed specific relationships are spurious. Additionally, an exploration of the nature, correlates, and measurement of the SA ability itself has relevance to the issue at hand and to the question of ”Who do we want to know about whom?". While the present studies used a shot-gun approach, the proposed research should provide the rifle needed to aim programs more precisely and derive maximum outcome. CHAPTER V SUMMARY AND CONCLUSIONS Two experimental training programs used a pretest-training- posttest paradigm to improve stereotype accuracy (SA): the ability to predict group norms. In Study One, a pilot program, training improved predictions about marital stereotypes even though training was provided on executive, psychologist, college student, and professor stereotypes. But the criterion instruments were unwieldy and the single criterion design gave little control over what occurred between pretest and posttest, while a great deal of sample shrinkage took place from the first to the last SA training session. Study Two's revised and shortened criterion-training instru- ments made each training session a complete unit in itself. Sig- nificant SA improvement occurred in the second study, but improve- ment on one stereotype did not assure improvement on others, and vice versa. improvement showed no erosion in Js retested 7 weeks after training. Improvement in Study One suggested generality of the under- lying principles and training designs. The specificity of SA and its improvement, research design requirements, and ease of administration, however, favor the single session unit design of Study Two. Moreover, combinations of training activities within a program emphasizing both content and process hold 82 83 the greatest promise for maximization of training impact. Practice and feedback formed a core for the additional discussions, diagnosis, base-rate, noncomitment, pooling, and panel procedures and conditions of the present studies. An additional finding was that Js who did not make written pretraining predictions improved about as much as Js who had made the written judgment. Additional data showed that the personality and intellectual correlates were also stereotype specific, and that initially inaccurate Js benefited more from training than initially I accurate Js. Both findings had relevance for the selection of trainees. Finally, moderately difficult criterion items best reflected training outcome from a statistical point-of-view. From a practical standpoint, however, the differences were not worthwhile. SA measures should continue to reflect judging reality and include the total range of item difficulties. Subjective feelings of improvement and liking for training were unrelated to actual outcome which precludes their use as criteria for training effectiveness. Implications of the findings for future training and research were discussed. \? ~+ BIBLIOGRAPHY Bronfenbrenner, U., Harding, J., and Callwey, Mary The measure- ment of skill in social perception. In D.C. McClelland (Ed.), T lent and Socie . New York: Van Nostrand, 1958, 29 - 108. Bruni, E. A film test of accuracy in observing people and its correlates. Unpublished master's thesis, Michigan State University, 1963. Campbell, D.T. Stereotypes and the perception of group differences. American Psychologist, 1967, 22, 817 - 829. Campbell, J.P., and Dunnette, M.D. Effectiveness of T-group experiences in managerial training and development. W. 1968. 70. 73 - 102. Cline, v.3. Person perception from the perspective of an empiricist. Paper delivered at the 76th annual meeting of the American Psychological Association, San Francisco, California, 1968. Cline, V.B. and Richards, J.M., Jr. The generality of accuracy of interpersonal perception. Journal of aaaarmal and §acial Psycholagz, 1961, 62, 446 - 449. Cline, v.3. and Richards, J.M., Jr. Accuracy of interpersonal perception: a general trait? Jaurnal af Abnoraal and Close, Patricia A test of stereotype accuracy in judging differences in marital couples. Unpublished manuscript, 1963. Cronbach, L.J. Processes affecting scores on ”understanding of others" and "assumed similarity". Paycaalogical Bulletin, 1955, 52, 177 - 193. Crow, W.J. and Hammond, K.R. The generality of accuracy and response sets in interpersonal perception. Jaurnal of Abnormal aad §gcial Paycgalgax, 1957, 54, 384 - 390. Dailey, C.A. The experimental study of clinical guessing. Journal of Individu 1 Ps cholo , May, 1966, 22, 65 - 79. Dailey, C.A. Prejudice and decision making. Personnel Administration. September - October, 1966, 6 - 13. 84 85 Dore, R. The development and validation of forced choice scales measuring attitudes toward leadership methods. Unpublished master's thesis, Michian State University, 1960. Dunnette, M.D. Forms of interpersonal accommodation: processes, problems, and research avenues. Paper delivered at the 76th annual meeting of the American Psychological Associa- tion, San Francisco, California, 1968. Dunnette, M.D. and Hakel, M.D. Interpersonal perception and behavior prediction. A proposal for research submitted to the National Science Foundation under NSF Grant GS-108l, 1968. Fancher, R.E., Jr. Accuracy vs. validity in person perception. Jaurnal of Qansulting Paychologg, 1967, 31, 26h - 269. Goldberg, L.R. Learning clinical inference: the results of intensive training on clinicians ability to diagnose psychosis vs. neurosis from the MMPI. Paper delivered at the annual meeting of the Western Psychological Association, Honolulu, Hawaii, June, 1965. Goldberg, L.R. Simple models or simple processes? Some research on clinical judgments. agaricaa Paxcaalagist, 1968, 23, 483 - 496. Cough, H.G. An exploratory cross-cultural study of interpersonal perception. Paper delivered at the 76th annual meeting of the American Psychological Association, San Francisco, California, 1968. Grossman, 8.5. Evaluation of a training program to improve the ability to differentiate between people. Unpublished doctoral dissertation, Michigan State university, 1967. Harding, J. Stereotypes. In David L. Sills (Ed.) International Enc c edi of e c a1 Sciences. New York: Macmillan and Free Press, 1968, 15, 259 - 262. ' Harris, W. The relation of observational to inferential accuracy in judging people. Unpublished master's thesis, Michigan State University, 1962. ' Jecker, J.D., Maccoby, N., and Breitrose, H.S. Improving accuracy in interpreting non-verbal cues of comprehension, gayghalagy in the Schools, July, 1965, 2(3), 239 - 264. Johnson, R.L. Correlates of a test of group sensitivity. Unpublished master's thesis, Michigan State University, 1963. 86 Johnson, Winifred B., and Terman, L.M. Personality characteristics of happily married, unhappily married, and divorced Persons. W151. 1935. 3. 29° - 311- Kepes, S.Y. Experimental evaluations of sensitivity training. Unpublished doctoral dissertation. Michigan State University, 1965. Linden, J. The self-centered orientation in interpersonal rela- tionships. Unpublished master's thesis. Michigan State University, 1965. Locke, E.A. Motivational effects of knowledge of results: knowledge or soul «tuna? W. 1967, 51. 32a . 329. Lorge, I., and Diamond, Lorraine K. The value of information to good and poor judges of item difficulty. Egggaslgaa; MW. 1954. 16. 29 - 33. Mietus, J. Development of a multiple component criterion for the evaluation of sensitivity training. Unfinished master's thesis, Michigan State University, 1969. Oskamp, S.W. Overconfidence in case-study judgments. Jaurnal af 223!!l£1£l.§£!£§2128!a 1965. 29s 261 r 265. Price, R. Is the favorableness of interpersonal evaluations a general trait? unfinished master's thesis, Michigan State University, 1969. Shears, A.D. The development and analysis of a criterion of the ability to understand people. Unpublished master's thesis, Michigan State University, 1967. Silkiner, D.S. A cross-cultural study of the measurement, determinants, and effects of stereotype accuracy. Unpublished master's thesis, Michigan State University, 1962. Smith, H.C. Bill Wilkins as a model for sensitivity training. Paper delivered at the 76th annual meeting of the American Psychological Association, San Francisco, California, 1968a. Smith, H.C. £32223!ll£¥.222112222259 New York: McGraw-Hill, 1968b. 87 Smith, H.C. Sensitivity to People. In H. Tech and H.C. Smith (Eds.) WW IIBISIIlSBIp Princeton, N.J.: D. Van Nostrand, 1968c, 10 - 19. Smith, H.C. §a9a1§11153_5a_£aaala, New York: McCraw-hill, 1966. Springbett, B.M. Factors affecting the final decision in the «Ploy-mt int-"M. WW. 1953. 12. 13 - 22. Stelmachers, Z.T. and McHugh, R.B. Contribution of stereotyped and individualised information to predictive accuracy. WW. 1964. 28. 23¢ - 242. Taft, R. The ability to judge people. P Bul ti , 1955, 52, l - 23. Tagiuri, R. Person perception. In David L. Sills (Ed.) IEESE: nagiaaal Bngzclaaedia a; tag fiaaia} Salaaaea. New York: Macmillan and Free Press, 1968, 11, 560 - 567. Vinacke, W.E. Stereotypes as social concepts. £22I2Il.2£_§22l£l gaxaaalagx, 1957, A6, 229 - 243. Wakeley, J.H. The effects of specific training on accuracy in judging others. Unpublished doctoral dissertation, Michigan State University, 1961. Zavala, A. A test of stereotype accuracy. Unpublished master's thesis, Michigan State University, 1960. APPENDIX A STUDY ONE STEREOTYPE ACCURACY MEASURES Page Attitudes of American Workers 89 Marriage Test 90 The Interests of Executives vs. Unskilled Workers 93 The Case of the Psychology Professor (Stereotype Training) 94 The Special Interests of Psychologists 97 The Case of Morgan Johnson 99 Stereotype Accuracy Personal Profile Sheet 102 89 Attitudes at American Workers Several thousand workers in 150 companies in an anonymous questionnaire were asked to indicate whether they thought the statements below were "true” or "false”. how well can you predict what their answers were? If you think a majority agreed with the statement, answer ”true”. If you think a majority disagreed with the statement answer "false”. Remember to answer them not as 122. would but as the typical worker would. T .T. IO-l I"! '8 [H IF! I"! IF! IO-l OH) .8 Iv-l IH IF! F F l. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. lb. 15. The average American worker today really enjoys his work. Most of the time the average worker is worried about being laid off. The average worker takes pride in what he does on the job. He thinks that wages are increased primarily because employees are able to demand an increase. He does aa§,feel that his Company's investment in new labor-saving equipment makes it possible for him to earn more money. He thinks he should be advanced by seniority rather than by Chi 11 12’. He should turn out as much work as he can. He feels that his work is important to the Company's customers. He feels that his management thinks his work is important to the Company. He feels that the Company will prosper whether he does his work well or not. He believes that everyone on the job would benefit if each worker did the best he could. He feels he receives fair treatment from his supervisor. He is more concerned about his job security than about h‘ an pay. He is more interested in getting advancement than in having an easy job. He feels that he gets impartial treatment from the management of his Company. 9O gaaaiage Test Directions: The following test measures your ability to identify differences between the typical happily married, unhappily married, and divorced person. Part I deals with men; Part II, with women. For each of the following statements, aaak tap one of the t as s ou think i be s bed b it. The correct answers are based on the answers that the groups made on lengthy questionnaires. For example, 100 happily married women, 100 unhappily married women, and 100 divorced women answered the question: ”Do you prefer a play to a dance?”. Results: 81% of the happily married women answered "yes”. 58% of the unhappily married woman answered ”yes”. 44% of the divorced women answered "yes”. Therefore, the correct answer to the statement "Most apt to prefer a play to a dance" is ”happily married women”. PART I . MEN Mark: (1) if you think the correct answer is ”happily married men"; (2) "unhappily married men”; or (3) ”divorced men". Qiffiaaltz Level 1. Most tolerant of sick people. a 2. Most prefers to spend a night at home. 6 3. Most apt to like religious people. 22 a. Slowest in making decisions. 37 5. Most likely to enjoy taking risks. 23 6. Most prefers to make plans with others. 54 7. Most apt to make bets. 48 8. Most likely to organise a club or team. 66 9. Most dislikes modern languages. as 10. Most apt to be critical of others. 29 11. Most touchy on the most subjects. 26 12. Most prefers fashionably dressed people. 19 13. Most often experiences feelings of loneliness. 50 14. Most dislikes foreigners. 40 15. Most apt to like the occupation of novelist. b6 16. Least often takes the lead to enliven a dull party. 49 nuns: uNNN Hut-0w upon-smog n Correct 2 17. 3 18. 2 l9. 3 20. 1 21. 1 22. 2 23. 2 24. 1 25. 2 26. 1 27. 3 28. Least interested in artistic activities. Most Least likes symphony concerts. Most Most Most Least often organises teams or clubs. Most actually being conservative. Most Least interest in the occupation of teaching. Most Most 91 ff cult Level 51 likely to like the occupation of stock broker. 43 likes the occupation of crimdnal lawyer. 6O 56 likely to stress quality in his work. 23 apt to like talkative people. 53 58 likely to think of himself as a radical while 63 likes cautious people. 46 57 meticulous and methodological in work. 45 likely to enjoy competition. 60 PART II. WOMEN Mark: (1) if you think the correct answer is ”happily married women"; (2) ”unhappily married women"; or (3) ”divorced women”. Correct F'N’hih’ r‘b1h1h3 F‘U’U’U’ row-har- NMNU 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. Most Most Most Most Most Most Most Most Most Most Most Most Most Most Least effective in emergencies. Most Least likely to blush. Least methodological. Most Least willing to work things out for themselves. D ffi lt Level apt to like old people. 18 often has spells of dizziness. 6 apt to like religious people. 26 apt to like music. 28 prefers working on commission to definite salary. 31 willing to be unconventional. 26 prefers taking chances to playing safe. tolerant of minority groups. 31 3O apt to arrive late for work. 27 often troubled by feelings of inferiority. likely to consider themselves as nervous. likes picnics and excursions. ll 31 13 interested in change and travel. 26 prefers work that makes heavy demands. 31 conservative in social and political opinions. 43 34 41 49 self-assertive and self-reliant. 33 29 Correct wwun Mt-‘wH 49. SO. 51. 52. 53. 54. 55. 56. Most Most Most Most Most Most Most Most 92 Difficult Level interested in avoiding technical responsibilities.51 interested in being an inventor. 41 dislikes quick-tempered people. 55 ambitious 45 apt to dislike working in isolation. 50 apt to like playing chess. 67 apt to like people who never drink. 35 apt to like psychology. 84 A large number of executives and professional men (lawyers, managers, etc.) and a large number of unskilled workers (laborers, porters, etc.) checked whether they liked or disliked each of the interests or activities below. A larger percentage of the unskilled workers liked half of the interests; a larger percentage of the executive and professional group liked the other half. Mag; ”1" if you think more unskilled workers liked a particular interest. Mag; ”2" if you think more professional workers and executives liked it. D f l L el gigglaalgx Level ;_1. Typist 4O 1_36. People Who Talk Very Slowly l,2. Bank Teller 41 15 1_3. Civil Service Employee 39 1,37. People With Gold Teeth 20 1_4. Interior Decorator 81 1,38. Nervous People 40 2_5. Sales Manager 32 1_39. People Who Always Agree 1_6. Secret Service Man 43 With You 25 1.7. Office Clark 19 2,40. Thrifty People 50 g_8. Scientific Research Worker 2,41. Religious People 74 7 g 42. Prepare the Advertising for ;.9. Draftsman 43 a New’Machine 23 $.10. Electrical Engineer 81 3,43. Courteous Treatment from 2.11. Editor 15 Superiors 23 $.12. Marine Engineer 69 2 44. Opportunity to make use of all 2_l3. Magazine Writer 29 one's knowledge and experience 2 14. Advertiser 20 12 2,15. Lawyer, Corporation 6 $.45. J. J. Pershing. Soldier 18 16. Lawyer, Criminal 72 2.46. William H. Taft, Jurist 18 17. Manufacturer 46 2,47. John Wanamaker, Merchant 53 18. Physics 8 2.48. President of a Club 15 19. Physical Training 13 g 49. Do a job yourself 67 20. Mathematics 14 ;_50. Definite Salary 34 21. History 18 2_51. Work for Yourself 33 22. Agriculture 12 2.52. Great Variety of Work 31 23. Golf 5 g_53. Emphasis on Quality of Work 15 24. Hunting 21 1_54. Opportunity to understand just ININP‘IPINIP It‘ifllr‘ifliNir‘lNlNlt‘ 25. Boxing 5 how one's superior expects 26. Musical Comedy 23 work to be done 34 27. Pet Monkeys 29 2 55. Freedom in working out one's 28. Detective Stories 15 own methods of doing the work 29. ”New Republic” 13 27 30. Conservative People 32 l_56. Repairing a Clock 16 1 57. Repairing Electrical Wiring 21 g 31. People Who Chew Gum 97 ;_58. Giving "first aid” assistance l_32. Socialists 38 36 Z 33. Energetic People 23 g 59. Adjusting difficulties of 2,34. Optimists 38 others 17 l 35. People Who Assume Leadership1_60. Climbing along edge of 29 precipice 38 94 The C e of the Ps lo es (Stereotype Training The statements below were in the Inventory completed by the pro- fessor. On these particular statements, as it turned out, he gave the aaaa_answers as those given by more than two-thirds of the college men who completed the inventory. The question here, there- fore, is not only just how well you understand the professor. It is also how well you understand the typical college man. Circle the answer "true" if you think both the college professor and the typical college man answered it ”true”. Circle the answer "false" if you think both the college professor and the typical college man answered it ”false”. if L vel T F l. I prefer quiet games to extremely active ones. 44 T g, 2. I am considered rather emotional by my friends. 19 T g_ 3. I have occasional difficulty getting the temperature of my bath the way I like it. 39 T g, 4. Divine inspiration is an infallible source of truth. 11 T, F 5. I sometimes think more about my ideas than about the routine demands of daily life. 15 I, F 6. I often think for a long time about an idea that has occurred to me. 13 I, F 7. I am generally active in my everyday life. 11 I, F 8. I am practically always tolerant even in dealing with people that I don't like. 18 2, F 9. I always keep control of myself in an emergency situation. 31 1, F 10. I occasionally neglect serious things in order to have a good time. 26 T E. 11. I always finish one task before taking on others. 32 I] F 12. I have occasionally doubted the reality of God. 6 I. F 13. I like to discuss abstract questions with my friends. 14 T E, 14. Artistic experiences are of great importance in my life 31 T §_ 15. Quite a few things make me emotional. 24 T §_ 16. The thought of God gives me a complete sense of security. 10 95 Difficulgx Level 2_ F 17. Radical agitators should be allowed to make public speeches 23 I. F 18. I believe that what a person does about a thing is more important than what he feels about it. 29 T §_ 19. I become emotional fairly easily. 17 T g_ 20. I would rather be a salesman than a scientific research worker. 15 To the following statements, the professor gave a di f ent answer to the one given by two-thirds or more of the college men. The question here, therefore, is: How does the professor differ from the typical college man? Circle ”true” if you think the professor answered ”true” but the typical student answered "false". Circle ”false" if you think the professor answered ”false" but the typical student answered "true". giggigulgy Level T g, 21. I believe that competitiveness is a necessary and desirable part of our economic life. 38 T E, 22. I am greatly influenced in minor decisions by how I happen to feel at the moment. 19 1’ F 23. I enjoy work more than play. 17 I, F 25. I like ballet performances. 12 1. F 25. I think cremation is the best method of burial. 27 I, F 26. Compared to your own self-respect, the respect of others means little. 69 T. F 27. I think I would like to decorate a room with flowers. 27 T F_ 28. l have never tried to collect pictures of paintings I have liked. 15 T g, 29. I am cautious about undertaking anything which may lead to humiliating experiences. 28 T 5' 30. I am a very adventurous person. 30 T g, 31. I am always taking on added social responsibilities. 54 T E, 32. I'm occasionally disorganized if I am called on suddenly to make a few remarks. 25 IF! '8 I'll P“ P” '1; I"! I"! 33. 34. 35. 36. 37. 38. 39. 60. 96 Difficult Level If I had the ability, I would enjoy teaching poetry at a University. 20 I prefer friends who have well developed artistic tastes. 19 I find it rather hard to keep to a rigid routine. 19 I believe in getting as much fun as I can out of life. 10 I would rather see a movie than read a book. 5 I never complain about my sufferings and hardships. 29 Sports generally interest me somewhat more than very intellectual affairs. 7 I have never been seasick, plane sick, or car sick. 67 97 e S ts DIRECTIONS: How do the interests of psychologists differ from those of other man? To answer the question, several hundred male psychologists and several thousand other business and professional ‘men checked whether they would ”like” each of many different occupations, amusements, activities, and kinds of people. A higher percentage of the psychologists liked some interests. For example, 41 percent of the psychologists said they would like to be the "author of a novel”; only 32 percent of men in general expressed such a liking. A 1222;,percentage of the psychologists liked some interests. For example, only 29 percent of the psychologists said they would like to be a ”sales manager" whereas 37 percent of men in general expressed a liking for this occupation. Mark for each of the interests below whether you think more or less psychologists liked the interest. Mark ”1” if you think a higher percentage of psychologists than men in general liked the interest; mark "2” if you think a love; percentage of psychologists liked the interest. m Di ficu L vol 1. Actor 33 2. Artist 18 3. Astronomer 54 4. Corporation lawyer 47 5. Manufacturer 14 6. Athletic director 27 7e ChenlSt 52 8. Cashier in bank 16 9. Editor 28 10. Foreign correspondent 21 11. Inventor 24 12. Magazine writer 28 13. Office manager 21 14. Orchestra conductor 54 15. Physician 23 16. ’0': 27 17. Rancher 17 18. Sculptor 37 19. Statistician 4O 20. Surgeon 36 21. Wholesaler 8 22. Geometry 48 23. Algebra 46 24. Physical training 32 25. Physiology 19 26. Literature 7 HHMHr—NHHHNHv-HNHHHHNr-‘NNNHHH NHNHHNNHh-HHHNNNNNNNNNHHNMHNNHHNNHME 0 n 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 45. 46. a7. 43. a9. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 98 Difficult: Level Hunting 24 Symphony concerts 22 Sporting pages 33 Golf 68 Chess 13 Solving mechanical puzzles 37 Travel movies 53 Fishing 39 Making a speech 22 Teaching adults 12 Taking responsibility 88 Doing research work 5 Writing reports 25 Regular hours of work 31 Developing business systems 31 Saving money 42 Conservative people 32 Energetic people 83 People who are natural leaders 76 People who make fortunes in business 38 Thrifty people 48 Religious people 42 Socialists 44 Independents in politics 13 People who talk about themselves 28 Carelessly dressed people 56 Absentdminded people 57 Outside work 59 Physical activity 42 Usually drive myself steadily 31 Have more than my share of novel ideas 17 My feelings are easily hurt 19 My advice is sought by many 9 Put drive into the organization 68 99 The Case of Morgan Johnson Morgan is a twenty-two-year-old unmarried college senior who is plan- ing to study psychology in graduate school. His parents died when he was four, and he and his younger brother were raised by permissive grandparents in Brooklyn. Of his childhood, Morgan said: "As I grew up, I always had the feeling that I was inferior to everybody else because I had no parents. In grade school, I was very loud and boisterous and made persistent attempts to dominate my peers and to excel in everything I did." Today he places emphasis on being a ”well-rounded scholar.” About his values, he now says: ”I do not believe there are any determining forces in the universe that make us what we are; everybody rules his own destiny. I can think of nothing more important than being a good friend or having good friends, but I don't think it is possible to have more than a few really close ones. I place little value on material things: cars, clothes, etc.” Morgan filled out the Strong Vocational Interest Blank that requires the respondent to answer "Like," "Indifferent," or "Dislike" to a long list of interests. The directions of this test ask the respondent to ”disregard considerations of salary, social standing, future advancement, etc... consider only whether or not you would enjoy the interest regardless of any necessary skills, abilities, or training which you may or may not possess." DIRECTIONS: In the first group of interests below Morgan gave the gag; answer as more than half of several thousand representative American men. Mark the answer that you think was given by both Morgan and the typical men. Use ”1“ for "Like"; ”2" for "Indifferent"; and ”3” for "Dislike”. Mo L ke the T Man Diff 1 evel l. Auctioneer 64 2. Auto salesman 75 3. Auto repairman 65 4. Typist 27 5. ”Atlantic Monthly" 55 6. Literature 34 7. Philosophy 39 8. Driving an automobile 24 9. Sporting pages 13 10. "Time” 19 11. Meeting and directing people 22 12. Meeting new situations 27 13. Adjusting difficulties of others 57 14. Contributing to charities 75 15. Progressive people 26 16. Foreigners 3O r‘lt‘ll‘lr‘lr‘urnmhur‘ww t" t" r- r- r. 'HHHHHHHHHHH'HHHHH cooocooooooolololuio 100 Diff ult Level 17. Sick People 5O 18. Cripples 40 19. People with protruding jaws 33 20. People with hooked noses 38 21. Blind people 41 22. Deaf Mutes 38 23. Men who chew tobacco 6O 24. People who chew gum 28 25. Civil Service Employee 86 26. Clergyman 74 27. Printer 76 28. Drilling in a company 46 29. Pet monkeys 61 30. Snakes 44 31. Acting as yell-leader 47 32. Quick-tempered people 13 an is Unl ke the T ic Man DIRECTIONS: In the group of interests below, the answer given by Morgan is the one in small letters (l,i,d). His answers to these interests, however, were ifferent from those given by the majority of men. From the two possible answers not chosen by Morgan that are indicated by capitals (L, I, or D) mark the one that you think was chosen by the typical man. Difficulty Level 33. Auto Racer 60 3‘00 Consul ‘50 35. Factory worker 21 36. Floorwalker 21 37. Poet 52 38. Politician 59 39. Real estate salesman 77 40. Specialty salesman 65 41. Algebra 60 42. Arithmetic 45 43. Economics 48 44. Physics 68 45. Physiology 67 46. Taking long walks 53 47. Performing sleight of hand tricks 56 48. Fortune tellers 37 49. Detective stories 7 50. Educational movies 66 51. Travel movies 39 52. Operating machinery 42 53. Pursuing bandits in sheriff's posse 57 54. Regular hours of work 14 55. Saving money 17 56. Conservative people 50 57. Energetic people 9 r-‘r-‘l-r‘r-‘t-t-‘r't-‘r'r'r'r‘t-‘Hr HHHHHHHH'H'H'H'H.H'H.H'H IUIUIUIUIOIUIUIO U U U U U U U U u-HHHHHHHsaH'HHHHHHHMHHHu-m's-OH 00.n-QIUGQGUIUQGQQQO-O-IUIUIU'UIOID DID IFIF‘Ir‘Ir‘ HIF‘IF‘UI‘II‘ H FIFII‘IFII‘IF‘IF‘ 1" l'" H 0- t‘ l" 0-! an I1" 1" 0- FIFIPII’" ”WHH'HWW calcium :5. o o 58. 59. 60. 61. 62. 63. 64. 101 Difficulty Level Optimists 10 People who are natural leaders ll Thrifty people 50 Spendthrifts 15 People who always agree with you 37 Bolshevists 9 Fashionably dressed people 15 102 mm we as as as on an on an an as nonsuoououm umswosoeosum .m scammem on as as we no we as as on an m ness—sue: .u> ao>.u:ooxu .m counaom Acomszoh :ewuozv as as Ne as on an on «n mm on mm monsuoououm use: on couemom as as as an on an «a an on em as nossuoouasm «we—sun: .o :o_emem cos co on o“ co on as on om om o «quqduuuuuuu Human mammomm A04 emu no couueueuweuca one 124 .mooe~o ~wa~oow 03» cu mumu~o~uuooo co~um~ouuoo one muooasc -< "ouoz n0.v as ~0.v.nae o s~e0 sewn 1 much seen: s~m u out co6o31~cu~umz 00 001 ~01 n c~m0 n0 o~ 00 sine u umom ~01 - n~ «son- tan 1 ohm sexu~ou~uez m~u 00 on ~01 w~ 0w 1 :~o0 00 00 00 nu- 00 um «son 1 upon an «0. mac 0N1 -n - skoeneano 1 sum - um~wo~oaozmm n01 o0 n~ 001 ~01 00 501 001 ~01 1 c~e0 «N ~n n0 n01 001 ~01 0~ eon o~ on 1 once «a o~ ~01 ~0 001 001 n~ eon m~ ee001s4~o . cum ~ ue~wo~ozommm N0 00 001 mm n~ 0~1 ~01 man on. semi n0 mm 1 c~m0 on - 0N1 - w~a sun 00- o~u N~u nan N0 h~ some 1 umom mm n~ «N. 501 somu c~u ~01 ~01 n0 N0 00 ~01 «mansion a cum - cm: owo-ou - w~ 001 n0 N0 «01 0~1 00a n~ 0N1 00: ~0 can on 001 1 ammo 0~ 0~ 00 -1 «N. 00. -u h~ on 00 mm s~ sew ow n0 es~m 1 umom 001 N0 - nun awn- n01 00 mm - mu ~n - N~n n01 n0 «some mm 1 ohm c~e0 uaom sum c~e0 upon some: ~qu.ucx 4.4.4.: so: sum =~ao seem ohm - ua~wo~oaommm a. u once was ~ ue~wo~oaozem ~ as: oue-ou c~e0 umom sum :~e0 umom mum - max H as: ewe-o0 ewe-o0 .e~niea e~efl can now sauce» ousasei hosusooe oen~>ou on» no co~ue~ouueus~ may 125 woos: ~oa~ooo 25 cu nus—303800 mo~um~ouuoo one oneness -< 3qu n0. voe ~0.vaee a c~e0 keno 1 umom m~1se0mu I cum ceaozn~ou~uez N0 00. no u c~e0 o~o n~n sou ea~e . umcm c~1 n01 n~ secnussun 1 sum cezn~eu~uex n0 nu m~c ~0 ~01 ~01 a c~e0 n0 0N won 001 q~ 0N sane u once 001 n~o h~ 0~1 0~ on stun- ow 1 0km - um~wo~ozohmm o0 o~ -n 501 «0 00 ~01 n0 n0 1 s~n0 N0 n~ 00 mo 00 ~01 001 00 n~ «see u once ~01 n01 00 0~ n~ 00- n~u ~01 n~ ssmnaeane 1 ohm ~ um~wo~onohnm 00 n0- o~ 00 n0 ~01 N0 n~ n~ no mo. 00. a c~m0 - 00 on no 00 «0 n~1 - ~n -c nu e~n same a once so 0~ 00 ~01 n0 n0 w~o m0- w~ n~u sou seem senoosa~o a cum ~H as: «wo-oo 00 n~ 00 - m~ o0- ecu 0~ «~1 «0 mm «N «0 00 do u c~e0 ~0 o~ 00 o~ n~ ~01 ecu s~n «01 n0- n~ h~ n0 n0 ~0 ea- 1 uaom no- ~0 so. no: 40 00 no 0N e~ 001 -u no. ~01 001 ~01 sum-sews a sum c~e0 anon sum c~e0 page sum c~eo upon a m u :62 owe-ou a. o swam mum came sacs mum coco: ~eu~ue2 ~ausuaz :02 - um~wo~ocoanm a - :dz uu~wo~oa0hem ewo-ou ~ on: mme-oo .nsoum o~ofiou emu new neuoon hoausooe oee~>eu on» no so~um~euuouc~ ozH 126 Results of delayed impression training for all judges. Stereotype/Subjects Delayed N Pretest N Posttest N 5322, M930; Mean Psychologist 1 Total 11.49 51 10.20 107 11.42 107 Males 11.05 20 10.07 41 11.54 41 Females 11.77 31 10.27 66 11.35 66 Psychologist 11 Total 12.05 55 10.88 101 11.99 101 Males 12.17 23 10.86 41 11.49 41 Females 11.97 32 10.89 64 12.27 64 Results of delayed impression training. Additional data. S.D.; N Variance Psychologist I Total 2.44 51 5 95 Men 2.31 20 5 34 Women 2.53 31 6 40 Psychologist 11 Total 2.36 55 5 57 Men 2.23 23 4 97 Women 2.48 32 6 15 127 Correlates of delayed impression scores. PM PM TW Impulsive-Controlled .05 -.28 .21 -.14 -.05 -.20 Rational-Empirical -.00 .05 -.03 .20 .26 .15 Introverted-Extroverted .03 .18 .01 -.12 -.11 -.18 Cautious-Bold -.05 -.02 -.15 .18 .24 .18 Emotional-Calm .16 .18 .29 .10 -.04 .20 Acquiescence -.18 -.21 -.17 -.34* -.43* -.27 Midterm Text .10 -.05 .24 .44** .55** .36* Midterm Class .35* .09 .50** .32* .47* .22 Midterm Total .25 .02 .42* .41** .54** .32 Final Text .14 .02 .25 .30* .53* .12 Final Class .11 -.16 .29 .36** .46* .29 Final Total .13 .07 .31 .36** .53* .23 Text Total .17 -.03 .43* .35* .54* .17 Class Total .25 -.09 .48** .36** .49* .26 Total Total .23 -.06 .51** .38** .54* .27 GPA .18 .13 .26 .38** .38 .38* CQT-Verbal .18 .27 .12 .27 .35 .21 CQT-Information .23 .16 .32 .17 .25 .10 CQT-Total .17 .24 .16 .24 .28 .20 Age .16 .14 .32 .06 .00 .07 Attendance -.03 .08 -.18 .29* .30 .35* Course Rank -.12 -.50 .07 .19 .29 .13 Consideration .28 .27 .20 .11 .18 .06 Responsibility .07 .13 .03 -.04 -.05 -.03 Leadership Total .27 .36 .17 .06 .12 .02 College Man I Pre .12 -.23 .24 -.02 .02 -.02 Post .31* .26 .33 .04 .04 .06 Gain .22 .34 . 12 .06 .03 . 11 College Man 11 Pre .21 .30 .17 .12 .26 -.12 Post .30* .29 .33 .05 .00 -.09 Gain .12 .08 .17 -.03 -.25 .04 Marital Men Pre -.17 -.01 -.24 .06 .05 .07 Post .00 .06 .02 -.08 .19 -.29 Gain .19 .08 .27 -.15 .14 -.33 Marital Women Pro .45** -.02 .67** -.01 .23 -.14 Post .15 -.22 .43* .09 .40 -.20 Gain -.33* -.20 -.46* .10 .20 -.01 Wav - - - -_..1.L__.19__.2L- * p(.05 128 #0. v nit m0.vos 0—. s0—. 00. 00.- «0N.- n0.- —0. ~—. 00.- an—. ——. ——. —~uoe-900 0—. —0. 00. ——. 0—.- 00.- 00.- 00. 00.- 0—. 00.- 00. so—uoELOu:—-hoo 0—. ~—. n0. 00.- 0—.- ~0.- 00. n—. n0.- «n—. 0—. n—. —ueuw>-eou e—. No. a—.- «0.- so.- 00.- «cu.- no. no.- -0—. so. ——. -aou nu. n—. no. n—.- «o. 00. «mm.- ~—. 00. «a. ——. ~—. <00 -nq. «tun. no.- n—.- —o. on. 0—.- n—. co.- «m. ~—. no. ~muou-—auoa «a—q. «c—o. q—.- —~.- 00.- 0—. n—.- 0—. —0. «on. n—. N—. —~uou-maa—0 «a—0. c—m. n0. n0.- no. 0—. u—.- ~—. ~—.- 0—. 00. N0. —uuou-uxoe «#00. «an. —0.- ——.- n0. —~. 0~.- ——. m0.- a—«. B0. 00. —au0u-—s:—m aa—e. can. 00.- 0—.- 00.- 0a. —~.- 00. n0. «an. ——. ~—. mun—o-—ec—m a0n. «mu. no. n0.- 0—. «N. 0—.- n—. 00.- —~. «0. —0.- uxou-—ec—m «ace. «an. e—.- e—.- —o. n—. ~—.- «N. ——.- —~. oo. co. ~uuou-auoue.z «on. «com. 0~.- —~.- 00. ~—. 00.- «en. m0.- 0a. 00. no. «me—o-auouv—z -ne. -m~. «o.- no.- mo.- ——. ——.- ——. n—.- m—. ——. no. axou.auuue.z c—eu unom s—ao uaom 0—00 Much c—no case: :02 —— — —— as: — mu: —au—uoz —eu—uox um—wo—ozohmm uu—wo—o50hum awo——oo owu——ou omfluumuwmw .mowvafi o—ufi on» 400 mu———ne —e:uoo——auc— o» oaoouso wc—c—euu no co—uo—om 130 so.v a- mo.v a 4 n0.- 00. n0. 00. ~—.- ~0.- ——.- —0.- <0.- 0—. 00. ~—. —ouoa-Hou 00.- n0. 0—. —m. a—.- —0.- —~.- 00.- 00.- m—. 00. 00. :o—aofiuous—-H00 00.- —0.- no. 00. ——.- n0.- —0.- ——. —0. sqm. n—. 0—. —mnuo>-Hou n0. —0.- «0~.- 00. 00.- h—. 0—.- 00. ~—.- 00. «0. m—. 0h8fi~ InQGOuUIm lye—@123; oozuoououm :o—ocoE—a u—uuH u——a a 0m .m—aeom —euOu on» new >u——ecomuon o» maoouso wc—c—ouu mo co—uo—wx 132 —0. v9: n0.v.o s 0—. 0—- 00.- NM.- sin0.- N0- 0—- —~.- n0.- —~.- 00. 00. 0—. n—. 0—.- 00.- nu. ——. 00. Q0.- s0n.- 00.- n—.- n—. n0. n—.- ——.- —0. 0m.- 00- 00. a—ou oo:ooao—:00< -—odo—uoau -n:o—usou co—mcoa—a u—uu -ao—muo>ouu:— mnnmqmummwuom ——- —N.- ”—- no. no.- ——. No.- -—aao.uax a—QU swam 6—00 0000 c—mu umom c—mu umom a—uo umom c—su umom co&o:-—eu—uoz :oz-—su—usz —— uw—wo—ozommm — um—wo—ocohmm —— an: oww——ou — as: owm——ou -o>—m—:ds— .mowvan o—ea onu no >u——e:Omuuq ou oaoouao wc—c—uuu mo co—uo—om mamuomuoum 3 3 1 —0.V.nst no.uv0 « ~0.- ~0.- no. 0—.- 00- 00.- c—au 00.- 00. n—.- —0- ~0.- 00.- uoom coaoz-—eu—usx n—.- s—- «inn-I NO. 00. #NN.- 0—00 00.- 00. «-.- n0. 00.- 00.- umom :oz-—ou—uuz no. 00. 00. 00. —0. 0—. a—ou 0—.- simm- ——. ——- n—. w—. umom —— um—wo—ocommm ~0- 0—.- 0—.- ~0- n0.- 00.- c—au 00- n0. 0—.- n0. N—. —0. umom — um—wo—ozozmm 00.- n0. N0. N0. n0.- 00.- a—eu 0—.- n0.- 00.- sum.- 00. N0. umom —— as: awo——oo ~0.- 00. ——.- —0.- 0—.- 00.- 6—00 ~0.- ~—. —0.- 00. 0—.- n0. umom — as: s“we—30 00——ouusou oo:oowe—:00< -—s:o—u06m -mso—u=o0 -co—muo>ouu:— -—sco—uom -o>—m—:oa— onzuowuoum .momnsn o—aaou 0:» no >u——ocomuom ou oeoouso wc—c—euu 0o :o—uo—om 134 —0.v ass wouswsoz —sco—u—00< n0.v a s —O.0 no.0 Go. nO-I na- «ON.I c—mc NO. mo-I no. NO-I no. 00.- Huck IGOEOSI—flu—uflz 0—.- «inN- no. 00-- m0. ——.- c—uo O~.I hO-I QO.I Nu-I NO. OO-I umom ICflZIuau—kmi NO. wO-I an. ucoi {MN- MO. ammo mO-u 00-0 #0-0 no- cu. m—- umomlun um—wo—OSOhmm «0. 0O. mo.- 09- m—-u no.0 :mGU ON. n—- 03- #0.. 0°. Mn. uwom 0— ummw0—O£0%mm no. m—. mO-I co. mc- no. ammo mm. d—- ho- QO-I 0O. mu. 809m Inn an: www—AOU oo.- {imN.I mu. $0.0 00. qo.o C—QU 00.0 h—-0 00. 00.0 ~0- o—- umom In an: mum—HOD —euOH u—zmumvawA mu———n—mcmmmom co—uuhwu—mcou xamm mausou oocmvcouu< om< unauowumum .mowcsn ——~ yew mo—ns—us> a—cmuoumo— 0cm .onsu—uus .o—nasuonmc o» mEoouao mc—c—suu no ocean—am 135 —0.v 0000s no-v.n s 00.- 0m.. 00. 00.- 00.- «00.- 0000 00.- 00.- 00. 00. 00.- «00.- 0000 -00003-0000002 00.. 000.- 00. 00. 00. 00.- 0000 0~.- 000.. 00. 00.- 00.- . 00.- 0000 -002-0000002 00. 00. 00. 00.- 00. 00. 0000 00.- 00. 00.- 00.- 00. «0. 0000-00 000000000000 00.- 00.. 0~.- 00. 0000.- 00.- 0000 00. 00. 00. 00.- 00. 00. 0000 -0 000000000000 00. «0. 00. 00.- 00w. 00. 0000 00. 00.- 00. 00.- «N. 00. 0000 -00 00: 0000000 00.- 0~.- 00. 00. N0. 00.- 0000 00.- 00.- 00.- 00. 00. 00. 0000 -0 :0: 0000000 filuOH «nuHU—vCO-M U m u «A map—O 000% COmUGhO—u «up—00 35m oak—000 flogficwuu< 0W< wen» owhwum nounmsux —sco—u—vv< .mowusa 0—05 000 mw—ns—ua> a—nmuouqu— can .uvau—uus .o—nnouwoamv 00 0500030 ws—c0ouu no co—um—mm 136 00.0.v00 s n0. 0—. ~0.- n0.- mo. 0—. 000- a—-- n—- h—- 00.- N—.- «0. 00. 00-- N—.- n0.- ——.- so. 00.- 0—- 00.- n0.- —0.- e—. swu- n0. 00. n—. h—. 0—. no. no. —N. ——- 0—- s0“. CN- n—- 0—- 0—.- s0~.- 0—- n—- 00.- sn~.- n—- <—- —suoa 0nmuovsou ———n—nso mom : —uauov—ocou adsa «00:00 sow. ad- No.- un- 0—. 0O- h—. n0.- «0.- no.- no.- oucsvcouu< N—.- c—su umom -:oao3-—su—uaz 00:0 umom -cwz-—mu—unz :000 0000-00 000000000000 c000 0000 -— um—wo—osuzmm c010 uuom -—— an: owo——o0 c—au anon -— as: ewo——ou wQMuoouwum .nowvafi e—ofiau uou ao—ns—ue> n—naususo— can .evau—uus .o—sneuwoaou ou oaoouao 00000000 No :o—uu—mm 13? Stability of SA training outcome. Mean Accuragy Stereotype Total Males Females College Man II Posttest 10.26 10.64 10.05 College Man 11 Retest 10.66 10.92 10.52 Psychologist I Posttest 11.42 11.54 11.35 Psychologist 1 Retest 11.37 12.13 10.96 Marital-Men Posttest 10.84 11.14 10.66 Marital-Men Retest 10.71 11.00 10.55 Marital-women Posttest 9.51 9.29 9.64 Marital-Women Retest 9.17 9.25 9.13 "Iclllm'cflifl‘ljflfl'flfi'ifllwflifl'fl'fliflmfl1T5