THE IMPORTANCE OF THE SOCIAL CONTEXT ON VERBAL EXPRESSIONS 0F RACIAL DISCRIMINATION AND RELATED STUDIES Thesis for the Degree of M. A. MICHIGAN STATE UNIVERSITY BERNIE I. SILVERMAN 1969 .......... L [HR .4 R '- Michigan Siarc , a“: University I' w‘ i THESIS ’ 1 .I i {ll li[[ (.{r ¢[ Ill ( [[ l‘lll‘ THE IMPORTANCE OF THE SOCIAL CONTEXT ON VERBAL EXPRESSIONS OF RACIAL DISCRIMINATION AND RELATED STUDIES BY Bernie I. Silverman A Thesis Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Psychology 1969 ' 5152337 /o/zz/c.;? ABSTRACT THE IMPORTANCE OF THE SOCIAL CONTEXT ON VERBAL EXPRESSIONS OF RACIAL DISCRIMINATION AND RELATED STUDIES By Bernie I. Silverman This investigation was designed with several purposes in mind. The first was to get some idea of the validity of the several referents of the subjects' attitudes towards Negroes. All subjects either signed or refused to sign a petition supporting the principles of open housing and indicated on a questionnaire that they either would or would not sell their homes to a financially qualified Negro. Also, subjects who were middle class Caucasian homeowners living in an all white area of Lansing evaluated hypothetical persons, some of whom were black, some of whom were white, in various social situations. From these evaluations racial discrimination scores were calculated for each subject. It was found that refusing to endorse the petition, indicating that one would not sell one's home to a Negro, and generating large racial discrimination scores when the hypothetical persons were evaluated as home purchasers, were all significantly related. Thus, through the process of construct validity, the probability that each instrument measures what it purports to is enhanced. A second aim of the study was to ascertain the importance of the social situation in which hypothetical persons were evaluated in deter- mining expressions of racial discrimination on the part of the subjects. It was found that situational differences accounted for at least twice as much of the variance in racial discrimination scores as individual differences. Further there were insignificant correlations between the amount an individual discriminated in one situation and the amount he discriminated in another. Thus knowing that an individual will either accept or reject a Negro as a friend is of little value in predicting the individuals' reactions to Negroes, compared to his peers, when the hypothetical persons are rated as neighbors or members of the family. A third aspect of the study concerned the effect of subjects' per- ceptions of their neighbors' attitudes towards Negroes on their own willingness to sell their homes to a Negro, as inferred from petition signing behavior. Subjects did not object to Negroes as neighbors but were reluctant to sell their homes to a Negro. To the extent that this reluctance was caused by perceived social pressure, informing the subjects that their neighbors had no objections to Negroes as neighbors should affect the subjects' attitude towards selling their homes in a nondis- criminatory manner and its concomitant, signing a petition supporting open housing. This proved not to be the case. A fourth facet of exploration dealt with the theory of Belief Con- guence. It was found, in accord with the theory, that the factor of attitude similarity was the most important determinant of ratings assigned the hypothetical persons. However, as the social situation in which the hypothetical persons were evaluated changed from friend to home purchaseer to son-in-law, the factor of attitude similarity accounted for less and less of the variance in ratings. Evidence was presented which showed that societal constraints in the form of perceived social pressure may have compelled subjects to discriminate on other factors in the latter situations, thus reducing the potency of the attitude similarity variable. A fifth function of the thesis was to highlight some variables that influence the relationship between verbal expressions of behavioral in- tention and overt behavior, often thought of as the relationship between attitudes and behavior. Data showed that there was a significantly stronger relationship between indicating on a questionnaire that one would sign a petition supporting open housing and actually doing so than between indicating on a questionnaire that one would sell his home to any qualified buyer and signing the petition. Thus correspondence between the situation depicted in the questionnaire, to which the subject verbally responds, and the situation in which the subject performs the observed overt behavior, is a factor determining the strength of agreement between overt and verbal behavior. Approved: Rah/MW) CAM Date° "\ 6 .46“ 3:9; I°Ij Thesis Committee: Dr. Raymond Cochrane, Chairman Dr. Milton Rokeach Dr. Charles Hanley .ACKNOWLEDGEMENTS In spite of the Freudian implications, I would first like to express thanks to my mother for typing not only the thesis itself, but also many of the materials used in the study. Much gratitude is also felt towards Dr. Raymond Cochrane for both his cogent advice and the many hours spent in thoroughly reading the thesis. Thanks are also due Drs. Charles Hanley and Terrance Allan for sharing their statistical acumen, and Dr. Milton Rokeach for the theoretical underpinnings upon which this thesis is based. ii Acknowledgements List of Tables List of Appendices Introduction . Method . . Results Discussion . Summary References Appendices . TABLE OF CONTENTS Page ii vii 45 105 . . . 109 114 iii Table 10. LIST OF TABLES The Various Experimental Groups Perceived Social Pressure to Discriminate Against Negroes in Various Situations -- Inferred from Attitudes Attributed to Friends, Neighbors, and Relatives Amount of Variance Accounted for by the Traits of the Hypothetical Persons in Four Situations Discrimination on the Race Factor as a Function of the Situation in which the Hypothetical Persons are Evaluated Negroes as Neighbors and Negroes as Home Purchasers -- The Differential Significance of Race as a Function of the Situation Discrimination on the Attitude Factor as a Function of the Situation in which the Hypothetical Persons are Evaluated Racial Discrimination Scores of 21 Subjects Who Refused to Sign a Petition for Open Housing Compared to 22 Subjects Who did Sign Such a Petition When the Hypothet- ical Persons were Evaluated as a Home Purchaser Racial Discrimination Scores of 21 Subjects Who Refused to Sign a Petition for Open Housing Compared to 22 Subjects Who Did Sign Such a Petition When the Hypothet- ical Persons were Evaluated as Friends Racial Discrimination Scores of 21 Subjects who Refused to Sign a Petition for Open Housing Compared to 22 Subjects who did Sign Such a Petition when the Hypothet- ical Persons were Evaluated as Neighbors Racial Discrimination Scores of 21 Subjects who Refused to Sign a Petition for Open Housing Compared to 22 Subjects who did Sign Such a Petition when the Hypothet- ical Persons were Evaluated as Sons-in-Law iv Page 43 50 51 53 54 55 56 57 57 58 Table 11. 12. 13. 14. 15. 16. 17. l8. 19. 20. Relationship Between Stating "I would sell my home to any financially qualified buyer, regardless of race" and Racial Discrimination Scores Given the Hypothetical Persons as Potential Home Purchasers Proportion of the Total Variance in Racial Discrimina- tion Scores Accounted for by the Situation in which the Hypothetical Person was Placed and Accounted for by Individual Differences Subjects Consistency in Discrimination Against Negroes in Four Situations as Expressed by Correlations Between Racial Discrimination Scores Effects of the Traits of Hypothetical Stimulus Persons on Ratings Assigned Them by 43 Caucasian Homeowners when the Former were Placed in the Situation of Friend Effects of the Traits of Hypothetical Stimulus Persons on Ratings ASsigned Them by 43 Caucasian Homeowners when the Former were Placed in the Situation of Neighbor Effects of the Traits of Hypothetical Stimulus Persons on Ratings Assigned Them by 43 Caucasian Homeowners when the Former were Placed in the Situation of Home Purchasers Effects of the Traits of Hypothetical Stimulus Persons on Ratings Assigned Them by 43 Caucasian Homeowners when the Former were Placed in the Situation of Son-in- law The Reliability of Petition Signing Behavior Relationship Between Racial Discrimination Scores and Perceived Social Pressure to Behave in a Discriminatory Manner in the Situation of Friendship Relationship Between Racial Discrimination Scores and Perceived Social Pressure to Behave in a Discriminatory Manner in the Situation of Selling One's Home to a Negro Page 58 6O 61 63 64 66 67 69 69 7O Table 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. Page Relationship Between Racial Discrimination Scores and 71 Perceived Social Pressure to Behave in a Discriminatory Manner in the Situation of Having a Negro for a Son-in- law A Comparison of Perceived Social Pressure for Selling 71 One's Home to a Negro Between those Subjects who Signed a Petition Supporting Open Housing and those Subjects who Refused to Sign Such a Petition The Effect of the Letter on Signing the Second Petition 72 for those Subjects who Refused to Sign the First Petition The Effect of Participation in a Psychological Study on 72 Petition Signing Among Subuects who Refused to Sign the First Petition The Combined Effect of Participating in a Psychological 73 Study and Receiving a Letter on Signing the Second Petition for Subjects who Refused to Sign the First Petition The Effect of the Letter on Signing the Second Petition 74 for those Subjects who Signed the First Petition The Effect of Participating in a Psychological Study 74 on Petition Signing Among Subjects who Signed the First Petition The Combined Effect of Participating in a Psychological 75 Study and Receiving a Letter on Signing the Second Petition for Subjects who had Signed the First Petition Relationship Between Signing a Petition Supporting Open 75 Housing and Indicating on a Questionnaire One would Sign Such a Petition Relationship Between Signing a Petition Supporting Open 76 Housing and Indicating on a Questionnaire One would sell his Home to Any Qualified Individual, Regardless of Race Perceived Social Pressure of Subjects Who Say They Will 90 Sell Home to Any Financially Qualified Buyer Compared to Subjects Who Say They Will Not Sell Home to Any Financially Qualified Buyer vi LIST OF APPENDICES Appendix A. Petition Purportedly Sponsored by the Urban League Supporting Open Housing Speech Used by Coeds when Presenting the Open Housing Petition Letter of Introduction Sent to all Subjects Adult Attitude and Interest Survey Descriptions of the Hypothetical Persons Measure of Perceived Social Pressure Measure of Verbal Expressions of Behavioral Intentions Persuasive Communications sent to Subjects Petition Purportedly Sponsored by the Organization for Equal Opportunity Supporting Open Housing Analyses of Variance of Ratings Given Hypothetical Persons in Four Situations vii Page 114 115 116 117 118 125 128 129 132 133 THE IMPORTANCE OF THE SOCIAL CONTEXT ON VERBAL EXPRESSIONS OF RACIAL DISCRIMINATION AND RELATED STUDIES INTRODUCTION It may be inferred from the principle of Belief Congruence, pro- pounded by Rokeach and Rothman (50), that individuals tend to like others whom they perceive as having beliefs and attitudes similar to their own, while disliking those whom they perceive as holding attitudes and beliefs different from their own. This same inference may also be drawn from the Balance Theories of Raider (29) and Newcomb (41). How- ever, one unique implication of the principle of Belief Congruence is that attitude similarity or dissimilarity is a more potent determinant of interpersonal attraction than is racial similarity. "We don't hate niggers. We just don't want them near us. That's why we moved from Chicago (27)." Though too crude to be uttered by most, this statement accurately reflects the sentiments of a large number of white Americans. Both Grodzins (24) and Farley and Taeuber (20) provide evidence that shows residential racial segregation is increasing. Caucasians are moving to the suburbs while Negroes are confined largely to the older areas of the cities. Neighborhoods are integrated only in the interim between the first Negro family moving in and the last Caucasian family moving out. It is evident that the principle of Belief Congruence and the reality of residential racial segregation are in sharp conflict. Rather than discard the theory as inaccurate, Rokeach and Mezzi (49) ask for a dispensation. They write "the focus of racial and ethnic discrimination is to be sought in society, not in the individual's psyche. If a society's constraints were altogether removed, man would not discrimin- ate in terms of race or ethnic grouping...but in terms of the principle of Belief Congruence." One might conclude that whenever there are strong societal sanctions supporting racially discriminatory behavior the theory is not to be applied. When these sanctions are absent or innocuous, the principle of Belief Congruence is thought to account for a large portion of the variance in interpersonal attraction. The remainder of this thesis will be based on three separate studies that largely were born from issues discussed in the previous paragraphs. The first study deals with differential discrimination on race as a function of the situation in which a Negro attitude object is encountered. It follows from the theory of Belief Congruence that in situations that are pregnant with sanctions against non-discriminatory behavior, the variable of race should account for the largest proportion of variance in a measure of interpersonal attraction. 0n the other hand, in situa- tions relatively free of sanctions against racially non-discriminatory behavior interpersonal attraction should be more strongly related to attitude similarity. The second study constitutes an attempt to ascertain the effect of perceived social pressure supporting racial discrimination on the attitudes of Caucasian homeowners towards selling their homes to financially qualified Negroes. It is consistent with the principle of Belief Congruence that individuals discriminate on race in the selling of their homes in part because they believe, either rightly or wrongly, that their neighbors expect or even demand of them such behavior. If this is the case, manipulating subjects' perceptions of their neighbors' feelings towards Negroes ought to result in a change in subjects' atti- tudes toward selling their homes to Negroes. The third study involves an investigation of the relationship be- tween verbal expressions of behavioral intentions and corresponding overt behavior. An attempt was made to determine the effect of the de- gree of correspondence between the situation depicted in the question, to which the subjects verbally respond, and the situation in which the subjects perform the overt behavior on agreement between verbal and overt behavior. FIRST STUDY Before engaging in a review of studies that pit attitude similarity against other factors such a race and socioeconomic class as determin- ants of interpersonal attraction, it might be wise to look at several papers that establish the relationship between attitude similarity and interpersonal attraction. Byrne (5), after having determined the atti- tudes of his subjects, presented them with a hypothetical person, through the medium of a checklist that expressed opinions on 26 different issues. Those hypothetical persons that agreed with the subjects on all 26 issues were seen as significantly more desirable as friends and work partners than those who disagreed. Those hypothetical persons that agreed with the subjects on the 13 most important issues were rated significantly higher than those who agreed with the subjects on the 13 least important issues. Thus the relevance or importance of attitudes held in common as well as their number appears to determine interpersonal attraction. Byrne and Nelson (10) attempted to discover whether the gross number of similar attitudes held in common or the proportion of similar attitudes accounted for ratings assigned to hypothetical persons as friends and work partners. If the latter is the case, subjects would evaluate a hypothetical person characterized by four similar and no dissimilar attitudes more favorably than one characterized by eight simi- lar and eight dissimilar attitudes. However, if the gross number of similar attitudes is the relevant factor, the evaluations should be re- versed. The results show that the proportion and not the gross number of similar attitudes held in common influences interpersonal attraction. Byrne and Griffitt ('7) obtained results that corroborate this finding and generalize it to high school and grammar school children as young as ten years old. Perhaps the most convincing study conducted in this area is that of Newcomb's (41). His subjects were 17 male students who had been living together for one year. All filled out lists indicating to what degree they liked the 16 other persons. It was found that the more similarly any pair of individuals rated the 15 remaining individuals, the more they indicated they liked each other. One may conclude that the perception of similar attitudes held toward other persons was related to interpersonal attraction between pairs of individuals. Newcomb writes "While I regard similarity of attitudes as a necessary rather than a sufficient condition, I believe that it accounts for more of the variance in interpersonal attraction than does any other single variable." Some evidence as to whether this is the case or not will be presented in the following para- graphs. Rokeach, Smith, and Evans (51) conducted the first study in which both attitude and racial similarity were simultaneously manipulated. Each hypothetical person consisted of a single attitude attributed to an individual of either the Negro or Caucasian race. Subjects were asked to rate the hypothetical persons as friends. The results indicate that both attitude similarity and racial similarity affected the ratings given the hypothetical persons and that of the two attitude similarity exerted a greater influence. Byrne and Wong (12) presented subjects with hypothetical persons that were either Negro or Caucasian and who either agreed or disagreed with the subjects on 26 issues. The dependent variables in this study were ratings of the hypothetical persons both as friends and as work partners in an experiment. The results reveal that when the hypothetical persons were evaluated as friends, attitude similarity governed responses while racial similarity had no significant effect. This tends to con- firm the findings of Rokeach, Smith and Evans. But when the hypothetical persons were evaluated as work partners, Caucasian hypothetical persons who agreed with the subjects were rated significantly higher than their Negro counterparts while the main effect of attitude similarity remained significant. It may be that when subjects feel that their responses may be of actual consequence, such as determining who they are to be paired with in future experiments, they are less likely to give the socially acceptable response and instead respond in congruence with their own feelings. It is suggested then that the greater the perceived consequences of a response the more valid that response is for indicating underlying attitudes. Byrne and McGrew ( 9) no doubt enhanced the credibility of the en- tire procedure by using a photograph to convey the race of the hypothet- ical person to the subjects. Attitude similarity was again imparted through a checklist purportedly filled out by the hypothetical persons. The latter held from one to eight attitudes in common with the subjects, who themselves were classified as either high or low on racial prejudice due to their scores on a previously administered paper and pencil measure (the Desegregation Scale of the General Survey of Public Opinion). Again the subjects' task was to rate the hypothetical persons first as friends and then as work partners. As in prior studies attitude similarity was found to be significantly related to ratings assigned in both situations while racial similarity was related to the subjects' ratings only when the hypothetical persons were viewed as work partners in an experiment. Possibly a more important finding of this study was that subjects inde- pendently designated as prejudiced behaved less in accord with the principle of Belief Congruence than subjects not so designated. Not only did they fail to discriminate on attitude similarity to the same degree as subjects designated low on racial prejudice, but they discriminated to a greater extent on racial similarity, rating Negroes significantly lower than Caucasians as work partners. This agreement between two methods of measuring the same underlying factor adds to the probable validity of both. With the sole exception of Newcomb's study, all the other investi- gations previously described confronted subjects with paper and pencil stimuli and requested paper and pencil responses. It is likely that these responses were of little consequence to the subjects as they failed to commit them to any future course of action. This criticism does not apply to the following study. Subjects in a study conducted by Rokeach and Mezzi (49) were either students, supposedly waiting to take part in a psychology experiment, or adults waiting to apply for jobs at state mental institutions. While waiting, each subject was engaged in a dis- cussion by four confederates of the experimenters. Two of the confed- erates were Caucasians while two were Negroes. One individual of each race agreed with the position taken by the naive subject while one in- dividual of each race disagreed. The naive students were then asked to select two of the other "subjects" to have coffee with them while the naive job applicants were asked to choose two of the other "job applicants" as future work partners. The results show that students selected pairs of "subjects" who agreed with them four times as fre- quently as pairs of "subjects" of the same race while job applicants chose other "job applicants" who they perceived to have similar atti- tudes 15 times as frequently as "job applicants" of the same race. This study, like those reviewed previously, appears to support the con- tention that when the hypothetical person or stranger is evaluated as a friend or work partner, attitude similarity affects the subjects' ratings to a greater degree than does racial similarity. Triandis (60) confronted his subjects with 16 hypothetical persons that varied on four attributes; race, social status, philosophy of life, and religion. Subjects were required to rate the hypothetical persons on a 15 item social distance scale which placed the hypothetical persons in roles such as friend, neighbor, and marriage partner. The results reveal that while all the variables significantly affect social distance scores, racial similarity accounts for approximately four times as much of the variance as does either attitude similarity or social status. Rokeach (47) attempted to explain the disparity between his own and Triandis' results by claiming that the belief component of Triandis' hypothetical persons was relatively weak and nebulous compared to the belief component of the hypothetical persons used in his own studies. Triandis felt the inconsistency in results was due to a different factor. He wrote "Had Rokeach considered something more than friendship, he would not have obtained his results." Thus Triandis was implying that the roles or situations the hypothetical persons were placed in deter- mined the differential effects of attitude and racial similarity. Triandis and Davis (61) had subjects evaluate hypothetical persons, who consisted of a race, sex, and single belief regarding civil rights, on four independent dimensions of behavioral intention, previously iso- lated by factor analysis. They found that whether the subjects admired the hypothetical persons (Formal Social Acceptance Dimension) was mainly contingent upon attitude similarity. But the hypothetical persons attractiveness as a friend (Friendship Acceptance Dimension), a neighbor (Social Distance Dimension), and as a marriage partner (Marital Accept- ance Dimension), was more strongly related to racial similarity. "...those situations that are the least intimate are Triandis concludes governed by belief similarity. The more intimate the behavior, the larger the weight given to the race component." He believes this to be the case because "the norms of behavior in our society are most clearly specified in the case of intimate behaviors...Thus the normative rejec- tion of persons who are racially different is found in its strongest form in the case of intimate behaviors." Apparently, in accord with Belief Congruence theory, Triandis believes individuals discriminate on race because established social sanctions often demand such behavior, and as these sanctions are differentially prominant in different situa- tions, the different situations come to evoke varying degrees of racial discrimination. Lohman and Reitzes (36) relate observations that indicate the powerful effect of the situation in determining behavior towards Negroes. In the former study subjects were both residents of a neighborhood in Chicago that was known to exclude Negroes and also members of an indus- trial union which had a policy of equal opportunity for members of both races. Lohman and Reitzes write "In both situations it was discovered that the individuals' generalized attitudes towards Negroes were inad- equate to explain actual behavior. For in the work situation...the union's position on racial questions was in control. In the neighbor- hood...the civic organizations' position of completely rejecting Negroes as potential neighbors was determinative. It is of particular interest to note that there was no statistical correlation between acceptance or rejection of Negroes on the job and acceptance or rejec- tion of Negroes in the neighborhood." Minard (39) reports similar find- ings. He found that Negro and Caucasian coal miners in the southern United States freely interacted when working in the mines, but would not associate with one another in other social contexts. These studies clearly reflect the effect of the situation on racially discriminatory behavior. Stein, Hardyk, and Smith (57) carried out a study which not only contributes more information as to which situations encourage racial dis- crimination, but also offers an explanation as to why previous studies had obtained contrary results. Each subject was presented with four by- pothetical persons that consisted of a single race and 20 attitudes 10 concerning how teenagers ought to behave. These attitudes were either similar to or dissimilar to those held by the subjects. The latter were required to indicate how friendly they felt towards the hypothetical persons and also to respond to them on an 11 item social distance scale. Further, each subject indicated how friendly he would feel towards a fifth hypothetical person described only as "a Negro teenager": The findings show that although the effects of both race and attitude simi- larity were significant, the latter accounted for more of the variance in the subjects' responses when they indicated how friendly they felt towards the hypothetical persons. In order to calculate the relative effects of racial and attitude similarity in various situations, each item on the social distance scale was analyzed separately. In every situation attitude similarity had a significant effect while the race of the hypothetical persons affected the subjects' responses in but three situations. These were "invite home to dinner," "live in the same apartment house," and "date my sister." The authors concluded "A race effect tends to occur on items that appear to involve publically visible relationships that are sensitive and controversial by prevailing cultur- al standards." Another very interesting result of this study was that friendliness ratings assigned the hypothetical person described only as "A Negro teenager" correlated .15 with ratings given the Negro teenager with similar attitudes while correlating .62 with ratings assigned the Negro teenager with dissimilar attitudes. It appears then that subjects evaluate a hypothetical person described only as a Negro as though he held attitudes dissimilar to their own. If strong attitudinal compon- ents are not provided in the make-up of the hypothetical persons (this applies to the studies of Triandis, and Triandis and Davis), subjects 11 may assume attitude dissimilarity for Negro hypothetical persons, and evaluate them accordingly. This may explain why Triandis, and Triandis and Davis obtained relatively large race effects compared to studies conducted by other investigators. "I wouldn't mind if an educated Negro lived near me, but I wouldn't want trash, white or black (27)." No doubt an individual's socioeconomic class as well as his attitudes and race contribute to his attractiveness in various situations. Westie and Westie (68), working with a middle class Caucasian sample found that the higher the socioeconomic status of Negroes, the less social distance Caucasians felt towards them. Byrne, Clore, and Worchel (6) discovered that subjects indicated they felt sig- nificantly less friendly towards hypothetical persons of a lower socio- economic class than those of the same socioeconomic class as themselves. Triandis and Triandis (63) asked subjects to rate on a social distance scale hypothetical persons that varied on both race and socioeconomic class. Data analysis discloses that the race of the hypothetical person accounted for 77% of the variance in a composite social distance score while socioeconomic class accounted for 16%. Thus it seems that a person's race is more important than his socioeconomic status in deter- mining his overall rating on a social distance scale. Stein (56) examined the effects of attitude similarity, race, and socioeconomic status simultaneously by assigning some level of each of these attributes to hypothetical persons. The subjects' task was to indicate how friendly they might feel towards the hypothetical persons and also to indicate on an 11 item social distance scale in what situa- tions they would be willing to interact with them. Sixteen separate 12 groups of ninth grade students served as subjects. It was found that attitude similarity accounted for between 8% and 42% of the variance while race and socioeconomic status accounted for less than 5% of the variance in friendliness ratings. For total social distance scores, obtained from summing all the yes responses to the social distance scale, attitude similarity accounted for 6% to 33% of the variance, race accounted for 0% to 12% of the variance, and socioeconomic status was responsible for 0% to 9% of the variability. The results of analyzing each item on the social distance scale separately show that the race effect was greatest for "have a close relative marry," "invite home to ' and "neighbor on the street," but only in the first instance dinner,‘ was it stronger than the effect of attitude similarity. In light of previous findings these results are hardly surprising. Attitude simi- larity is more important than racial similarity which is more important than socioeconomic status in determining intentions to interact in most situations. In those situations where racial similarity appears to be the pre-eminent determinant of intentions to interact, societal sanctions facilitating racially discriminatory responses are assumed to be oper- ating. In studies in which a large race effect was obtained in specific situations, it was attributed to social sanctions salient in that situa- tion. Yet the presence of social sanctions facilitating racial discrim- ination in a situation was inferred from a large race effect. The reasoning is circular, and makes it quite impossible to disprove the principle of Belief Congruence. Whenever results are in accord with it, social sanctions are not thought to be operating. But when results show that much of the variance in responses is due to the race of the 13 hypothetical person, social sanctions are invoked. Clearly an independent measure of perceived social pressure to act in a racially discriminatory manner is required. In the present study such a measure was employed. In order to de- termine perceived social pressure for acting in a racially discriminatory manner when the hypothetical person was placed in the role of a friend, subjects were required to indicate how they thought their friends would react to them if they befriended a Negro. In order to determine per- ceived social pressure for acting in a racially discriminatory manner when the hypothetical person was placed in the role of a potential home purchaser, subjects were required to indicate how they thought their neighbors would react to having a Negro for a neighbor. Finally, to determine perceived social pressure for acting in a racially discrim- inatory manner when the hypothetical person was placed in the role of son-in-law, subjects were required to indicate how they thought their relatives would feel about a Negro marrying into the family. In each case subjects could select one of five possible responses indicating varying degrees of perceived social pressure for racial discrimination. The first hypothesis tested was that as perceived social pressure to act in a racially discriminatory manner increases across situations, actual discrimination on the factor of race increases. Although measures of perceived social pressure were acquired for three situations, the hypothetical persons were actually presented in four distinct social contexts. The fourth situation, for which no measure of perceived social pressure was acquired, was that of neighbor. Earlier studies have shown that when the hypothetical person is placed in the role of a neighbor subjects discriminate on race. However, there 14 is some reason to believe that racial discrimination will be less when the hypothetical persons are evaluated as neighbors than when they are evaluated as potential home purchasers. Reitzes (45) in a very recent survey, found that while 84% of the residents in a suburban community would either feel positively or at worst indifferent about having Negro neighbors, only 31% said they would sell their homes to Negroes. Fredrichs (22) picked up the same trend in a study conducted over a decade ago. Forty-three percent of his subjects expressed a willingness to accept Negroes as neighbors, but only 32% were willing to sell their homes to otherwise qualified Negro individuals. Based on these findings, the second hypothesis to be tested in this study was that subjects will discriminate on race to a significantly greater extent when the hypo- thetical person is evaluated as aypotential home purchaser than when being evaluated as a neighbor. It does not necessarily follow that the more one discriminates on race the less one discriminates on attitude similarity. The two may very well vary independently of one another. The most direct implication of the principle of Belief Congruence is that ratings of interpersonal attraction will be mainly influenced by attitude similarity except in situations in which there are strong societal sanctions to discriminate on race. Thus the third hypothesis to be tested was that there would be an inverse relationship betweenyperceived social_pressure to act in a racially discriminatory manner and the amount of variance accounted for by attitude similarityyin evaluations of the hypothetical persons. Perhaps one of the greatest shortcomings of almost all the studies focusing on the principal of Belief Congruence is that they were con- ducted exclusively with subjects that were either high school or college 15 students (invariably in introductory psychology courses) in an academic setting with an instrument that may have both revealed the purpose of the studies and allowed subjects to present themselves in any manner they desired. Students as a group are notably more liberal and far less prone to discriminate on race than are adults. It may be quite unjustifiable to generalize results garnered from such a distinct sample to the entire population in the form of a universal principle. This supposes, of course, that the responses given by the students accurately reflect their attitudes. This supposition itself may be invalid. Cook and Selltiz (16) write "We assume that with respect to many attitudes, the settings in which tests are usually administered tend to exert pressures in a constant direction. It seems reasonable that most respondents, presented with tests in an academic setting...will assume that the res- ponses which will place them in the most favorable light are those which represent them as well adjusted, unprejudiced, rational, open minded, and democratic." Thus the results of the studies reviewed may be due more to the "demand characteristics" (42) of the experimental situation than to the nature of the subjects' psyches. Cook and Selltiz continue "Susceptibility of overt response to distortion - that is, the possibil- ity of discrepancy between private and overt response - would seem to be a function of three characteristics of the instrument: the extent to which its purpose is apparent, the extent to which the implications of specific responses are clear, and the extent to which responses are subject to conscious control." The original Summated Differences Technique as designed by Westie (67) sharply limited the ability of sub- jects to represent themselves as tolerant by making it extremely diffi- cult to compare and equate responses to otherwise identical Negro and 16 Caucasian hypothetical persons.1 This difficulty was obtained by separat- ing a Negro hypothetical person from his Caucasian counterpart by at least 200 descriptions of other hypothetical persons. Subjects simply could not recall ratings assigned to comparable hypothetical persons of the other race. Hence subjects could not consciously control the amount of prejudice they displayed. Studies reviewed in this paper presented subjects with far fewer hypothetical persons to evaluate. This makes it considerably easier for the subjects to compare responses and thereby consciously project any image they feel will be well accepted. It is conceivable that the principle of Belief Congruence may merely be an artifact of the subjects, the experimental situation, and instruments used in obtaining the data upon which it is based. If the responses given to the paper and pencil instruement could be related to some sort of corresponding behavior, the validity of the Summated Differences Technique for identifying underlying attitudes, and consequently, the validity of the principle of Belief Congruence, would be enhanced. Rather than accepting the explanation of the findings offered in the previous paragraph, one would be compelled to conclude that subjects' responses are predominantly due to their own attitudes toward the hypothetical persons. The behavior observed in the present study was that of petition signing. Specifically subjects were asked to sign a petition which consisted of a pledge not to discriminate on race l. The Summated Differences Technique involves rating various hypothet- ical persons characterized in various ways on a one to nine scale. One indicates the most positive evaluation while nine indicates a strongly negative evaluation of the hypothetical person. A racial discrimination score can be determined for each subject by subtract- ing the sum of the ratings assigned all the Caucasian hypothetical persons from the ratings assigned the Negro hypothetical persons. The larger the resulting differences the more the subject discrim- inates against Negroes. 17 in the selling of their homes.2 Subjects were told that the petition, allegedly sponsored by the Urban League, was to be given to the mayor of Lansing in order to show popular support in the Caucasian community for the principle of open-housing. They were unaware that records were being kept of their behavior and that the petition was related to a psychology experiment in which they were soon to take part. The fourth hypothesis tested was that individuals who signed the petition would discriminate significantly less on the factor of racei when evaluating hypothetical persons as potential home purchasers, than those individ- uals who refused to sign the petition. It would be interesting to know if those subjects who discriminate relatively little against Negroes when the hypothetical persons are placed in the role of home purchasers are more likely to say that they would sell their homes to any financially qualified buyer, Negro or Caucasian, than subjects who discriminate to a greater extent. It is possible that even relatively small racial discrimination scores indi- cate that the subjects would be hesitant to sell their homes to Negroes. A separate questionnaire, asking subjects if they would sell their homes to any qualified buyers, regardless of race, was administered at the same session in which they evaluated the hypothetical persons. ‘Thg fifth hypothesis tested was that there was a significant relationship between subjects' racial discrimination scores obtained when the hypo- theticalypersons were placed in the role of home purchaser and their 2. Subjects in the present study were 43 middle-class Caucasian home- owners living in an all-white neighborhood in Lansing. I‘L‘ [I‘ll .I ‘II‘ [II [ll 7‘“ .1 ‘ II {II 18 ppofessed willipgness to sell their homes to any financiallypqualified buyeriiregardless of race. More specificallyijit was hypothesized that those who said they would sell their homes to any qualified buyer would have significantly smaller racial discrimination scores than those sub- jects who said they would not sell their homes to any qgalified buyer. Pettigrew (43) writes ”Along the continuum of prejudice theories, two extreme positions have been popular. One strongly emphasizes the personality of the bigot and neglects his cultural milieu; the other views intolerance as a mere reflection of cultural norms and neglects individual differences." If the degree to which one discriminates is due mainly to one's personality, it would be expected that the largest proportion of the variance in racial discrimination scores would be re- lated to individual differences. On the other hand, if the situation in which an individual encounters a Negro is mainly responsible for the amount of racial discrimination, then the largest proportion of variance in racial discrimination scores ought to be accounted for by situational differences. The sixth hypothesis to be tested is that the situation accounts for more of the variance in racial discrimination scores than does the individuals'ypersonalities. If the sixth hypothesis is confirmed,it reveals only that the mag- nitude of racial discrimination is related more strongly to the social context in which a member of a minority group is encountered than to the particular personality of the individual. Still it may be valid to call some bigots while referring to others as tolerant if the former consist- ently discriminate on race more than other individuals in several situa- tions and the latter discriminate to a lesser extent than other individu- als given any of several situations. If individuals are consistent in 19 the degree to which they discriminate, relative to others, across situa- tions, large correlations between the magnitude of racial discrimination scores in one situation and those in another should be obtained. If, however, individuals vary the degree to which they discriminate on race, relative to others, as a function of the situation in which the hypo- thetical person is placed, low correlations would be expected. If this were the case the appelation ”bigot" would hardly be justified as one might be more tolerant than his peers when it comes to accepting Negroes as neighbors and yet respond more prejudicially than others when friend- ship is considered. The seventh hypothesis to be tested was that indi- viduals who discriminate more than average in one situation also discrim- inate more than average in other situations. The purpose of the final four hypotheses to be tested in this study was to determine exactly which factors the adult homeowners discriminate on within each specific situation. The results of the tests generated from these hypotheses are not intended to support or refute any partic- ular theory but only to provide information. The eighth hypothesis tested was that when the hypothetical persons were placed in the role of friend only the factor of attitude similarity would prove significant. The ninth hypothesis tested was that when the hypothetical persons were encountered in the role of neighbors, both attitude similarity and the race x attitude similarity interaction would attain significance. It was supposed that Negroes who held similar attitudes would be evaluated sig- nificantly lower than their Caucasian counterparts. The tenth hypothesis tested was that when the hypothetical persons were placed in the role of potential home purchasers all three main effects, ract, attitude, simi- larity, and socioeconomic status would be significant. The eleventh I . [\‘III‘II l‘l Il.lll1ll I'll 20 hypothesis tested was that when the subjects were required to evaluate the hypothetical persons as sons-in-law, not only would the three main effects be significant, but also the triple interaction would reach significance. This would be due, it was felt, to the inordinately favorable rating given the high status, similar attitude Caucasian rela- tive to other hypothetical persons. SECOND STUDY In the first study variability in perceived social pressure to dis- criminate against Negroes was examined across situations. The question asked was, do subjects, as a group, discriminate more in those situa- tions in which perceived social pressure to discriminate on race is greatest? In the study to be discussed next, the effects of perceived social pressure to discriminate against Negroes was examined with refer- ence to individual subjects within situations. In contrast to the first study the question becomes, within a given situation, do subjects who perceive much social pressure to discriminate against Negroes actually discriminate to a greater degree than subjects who perceive less social pressure? If so, it might be supposed that viewing one's friends, neighbors, or relatives as prejudiced might cause one to refrain from acting in a racially non-discriminatory manner. If fear of engendering the ill will of one's neighbors for example, prevents one from acting in a non-discriminatory manner, manipulating one's perceptions of his neighbors' feelings about Negroes ought to result in a change in one's behavior. An attempt to determine whether this is or is not the case constitutes the greater part of the second investigation. Katz (31) gave high school students a scale designed to measure anti-Negro prejudice. Each subject not only filled out the scale for himself but also filled it out as he thought his average classmate 21 would. The ten most tolerant students who had a mean prejudice score of 9.01 characterized their average classmate by a score of 8.51 while the ten least tolerant students, who had a mean prejudice score of 4.62, saw the average prejudice score of their peers as being 5.05. In fact, the mean prejudice score of the class was 7.33. Wallen (66) asked subjects to respond either "yes" or "no" to three statements of opinion and to estimate the percentage of students on campus who would respond in the affirmative to each. He analyzed his results by calculating bi-serial correlations between the subjects' own yes or no responses and the pro- portion of the population they perceived as replying positively. The resulting correlations were .56, .39, and .45 indicating that there is a moderately strong relationship between the subjects' own opinions and their perception of what proportion of the population agrees with the three statements of opinion. The results of a study conducted by Travers (58) tends to confirm these findings. In describing the results "...there is a persistent of studies such as these Campbell (13) writes correlation between a person's own attitude and his estimate of group opinion. While a few persons may chronically underestimate the popu- larity of their own opinions, the prevailing tendency is to overestimate the size of the group agreeing with oneself." The results of these studies may be explained in one of two ways. First, one might reasonably suppose that subjects project their own atti- tudes on to their peers and therefore perceive their peers as being more similar to themselves than they actually are. An equally tenable argu- ment is that the subjects' perceptionsof group opinion, rather than being caused by the subjects' attitudes, are somehow responsible for them. The next three studies to be reviewed all present evidence 22 supporting this position. Newcomb (40) relates a study in which measures on the liberal- conservative dimension were acquired for women who were students at Bennington College during the 1930's. Approximately twenty-five years later the political orientations of these same women were measured again. Of the twenty-two women who were classified as extreme liberals in the 1930's, eight were so classified in 1961. Of these eight, 75% had po- litically liberal husbands. Seven of the women who originally were evaluated as liberals were later evaluated as moderates. Of these seven, 58% had politically liberal husbands. Finally, for the three women who at first had been deemed liberal and who in 1961 were rated as conserva- tives, 100% had politically conservative husbands. The results of this study lend credence to the contention that the attitudes one holds are in part contingent upon the attitudes popular in one's immediate environ- ment. This proposition receives further support from the findings of Liu (35). His subjects were 196 Catholic individuals who had recently moved to Florida from various northern communities. They each were required to fill out a Southern Identification Scale and a Race Attitudes Scale. The results reveal that scores on both these measures were significantly related to length of residency in the south. The longer an individual had lived in the south, the more he identified with the south and the more he discriminated against Negroes. The question now arises, is merely living among others who hold certain attitudes sufficient to influence an individual's attitudes or must one identify with a particular group of people before their views exert an influence upon one's own? The following study provides an answer to this question. 23 Siegel and Siegel (55) determined that students living in high status, off campus row houses had significantly higher F-Scale scores than students living in dormitories. In effect the two groups had different norms. After their freshman year, all twenty-eight of the girls who served as subjects in this study wanted to move into a row house, but only nine were able to do so. For these nine girls, their membership group and their reference group were identical during their second year of college. At the end of their sophomore year, eleven of the girls who had lived in the dormitory still desired to move to a row house for their junior year. For these girls, their membership group and their reference group were disparate during their sophomore year. The remaining eight girls indicated that they wished to remain in a dorm- itory for their junior year. For these girls membership and reference groups were originally disparate at the beginning of their sophomore year, but by the end they had become congruent through a change in ref- erence groups. All subjects filled out the F-Scale at the beginning and at the end of their sophomore year. The results indicate that those girls who actually lived in the row houses during their second year had a significantly smaller reduction in their F-Scale scores than subjects in either of the other two groups. Subjects who wished to move to a row house at the end of their second year, after spending that year in a dormitory, had a significantly smaller reduction in their F-Scale scores than did subjects who indicated they wished to continue living in a dormitory. The authors' remark "The imposition of a membership group does have some effect on an individual's attitudes, even when the im- posed group is not accepted by the individual as his reference group." This point is extremely important to the present study, in which an 24 attempt was made to manipulate the subjects' perceptions of their neighbors' attitudes. It follows from the Siegels' findings that manip- ulating the attitudes of one's neighbors should affect one's own atti- tudes, whether or not one views the neighbors as a reference group. An individual ought to be affected by his neighbors' views merely because he is a member of the community. Only in the experimental situation can a causative rather than merely a correlative relationship be established between the attitude of a group, as perceived by the individual, and the individual's own atti- tudes. Asch (l) asked subjects to match a stimulus line with three other comparison lines that were presented. By themselves, subjects selected the correct match among comparison lines 99% of the time, so the correct response was obvious. In the experimental situation the naive subject gave his response after seven other subjects, who were actually confederates of the experimenter, responded. On twelve of eighteen trials the confederates gave obviously incorrect responses. The effect of perceived social pressure was represented by the number of incorrect matchings made by the naive subject. In three experimental groups the average number of errors per subject ranged from 4.1 to 5.1 while over one quarter of the subjects gave the conformity response at least eight times or more. Clearly the judgments of others influenced the judgments of the individuals. In a second study the naive subject was required to write down his selections rather than announcing them publicly before the group. This introduced a measure of privacy into the experimental situation and the mean number of errors committed by subjects was reduced to 1.5. Thus it appears that privacy of response reduces the effect of perceived social pressure. This finding was 25 confirmed by Deutsch and Gerard (18). This is important because subjects in the present study gave their responses, both to the questionnaires and to the petition, in the privacy of their homes. Note, however, that even when privacy prevailed, significantly more errors were made than when subjects were unaware of the responses of the other individuals. This indicates that merely being aware of the judgments of others is sufficient to influence one's own judgment. The one weakness of the experiments just related, relative to the present study, is that they dealt with judgments as to the length of a line rather than with attitudes. If their findings do not generalize to the realm of attitudes, their findings may be irrelevant to the present investigation. Wheeler and Jordan (69) had students either agree or disagree with fifty statements of opinion. The same students were con- fronted with the same task one week later to determine chance answer changes. After another week had passed subjects were again required to indicate agreement or disagreement with the items, but before doing so they were informed how the majority of the students had responded during previous sessions. The results show that after group opinion was made known a significantly greater proportion of answers were changed towards it than changed in the same direction by chance during the earlier administration. Raven (44) in a similar study, had subjects give their opinions as to how responsible a hypothetical teenager was for murdering an elderly lady. Ratings could range from one, indicating the subject felt the youth was entirely responsible for his actions, to seven, indicating that the subjects felt the youth's environment was responsible for his actions. The majority of the subjects chose ratings between four and 26 six. The subjects were then presented with fictitious group norms that made it appear as though the majority of the group felt the teenager was personally responsible for his crime. All subjects were requested to make another judgment as to how responsible the youth was for his crime. This second rating was not done publically. Thirty-three percent of the subjects whose initial opinions deviated by at least two positions from the fictitious group norm moved toward that norm when given the oppor- tunity to change their opinion, while only 15% of the subjects whose initial opinion was relatively congruent with the group norm changed their opinion to make it more congruent. Both this study, and the one directly preceeding it support the proposition that the subjects' per- ceptions of group opinion are to some degree responsible for the opinions of the subjects. Campbell and Stanley (14) write "Correlation does not necessarily indicate causation, but a causal law...does imply correlation." Thus finding a correlation between a measure of perceived social pressure and a measure of racial discrimination is a necessary but not a suffi- cient condition for positing that the former causes the latter. Per- ceived social pressure to behave in a racially discriminatory manner was inferred from the subjects' perceptions of how their friends, neighbors, and relatives viewed Negroes. It was assumed that if these groups were seen as having a negative attitude towards Negroes, sub- jects would be reluctant to interact with Negroes for fear of engaging their ill will. The first hyppthesis to be tested in this study was that when the hypothetical persons were evaluated as friends, there would be a significant relationship between racial discrimination scores and perceived social pressure to behave in a racially discriminatory 27 manner. Similarly, the second hypothesis tested was that when the hypothetical persons were rated as potential home purchasers, there would be a significant relationship between racial discrimination scores and perceived social pressure to behave in a racially discriminatory manner. Finally, the third hypothesis tested was that when the hypo- thetical persons were placed in the role of son-in-law, there would be a significant relationship between racial discrimination scores and per- ceived social pressure to discriminate against Negroes. A second measure from which the subjects' attitudes toward Negroes might be inferred is their petition signing behavior. Subjects either endorsed or refused to endorse a petition purportedly sponsored by the Urban League, which read in part, "We the undersigned residents and homeowners of Lansing, Michigan, pledge not to discriminate on race or ethnic background in the selling or renting of our homes." This indi- cator of the subjects' attitude towards selling their homes to Negroes has several advantages over the customary paper and pencil instruments used to measure attitudes. Cook and Selltiz (16) comment "To the extent that the purportedly unstaged situations are accepted as genuine, the respondent will not see them as designed to get information about his attitudes; thus one possible source of pressure to give responses that are likely to be considered desirable is eliminated." In the present study subjects had no idea that records were being kept of who signed and who failed to sign the petition or that the petition was associated with a psychology experiment in which they later were to participate. Secondly, Defleur and Westie (17) write "In American society, the affix- ing of one's signature to a document is a particularly significant act. The signing of checks, contracts, agreements and the like is clearly 28 understood to indicate a binding obligation on the part of the signer to abide by the provisions of the document." It is quite conceivable that the subjects may have felt that the petition, which was to have been presented to the mayor of Lansing, might have some influence on affecting public policy towards integration. Cook and Selltiz (14) remark "When responses are expected to have real-life consequences, the anticipation of such consequences may counterbalance the wish to make a good impres- sion." Therefore, because of the subjects' unawareness of being evalu- ated and because of the greater conceivable consequences of the behavior, petition signing is thought to be more free of variance due to the subjects' desire to make a good impression than most paper and pencil measures. Hence it is felt to be a more valid measure of the subjects' attitude toward selling their homes to Negroes than the paper and pencil Summated Differences Technique. The fourth hypothesis tested was that those who sigped the petition would perceive their neighbors as signifi- cantly less prejudiced than those who refused to sign the_petition. If this should turn out to be the case, it would be supposed that those who signed the petition perceived less social pressure to discriminate against Negroes in the role of home purchasers than those who refused to sign the petition. If perceived social pressure causes some subjects to refrain from signing the petition, changing their perception of their neighbors' feelings towards Negroes ought to result in a change in their petition signing behavior. In the present study 21 individuals failed to sign the first petition and later participated in the remainder of the study. A month after the study had been completed they received a letter from the experimenter informing them of the results of the study in which 29 they participated. The aspect of the findings that was emphasized most was that when the hypothetical persons were rated as neighbors, the subjects did not discriminate on race. Therefore, the letter explained, it appeared as though the residents of the neighborhood did not object to having Negroes for neighbors. Now eighteen other individuals who failed to sign the first petition and who did not participate in the study also received the letter. They served as a comparison group. A third group of 21 individuals who failed to sign the first petition and who did not participate in the study did not receive the letter. They served as a control group. Within a week after the hopefully persuasive communication was sent, all subjects were again given the opportunity to sign a petition similar to the first. The fifth hypothesis tested was that those individuals who failed to sign the first petition, who did not take part in the study, but who did receive the letter would sigp the secondypetition more freqpently than those subjects who failed to sign the first petition,ywho did not take part in the study, and who did not receive the letter. In order to determine if simply taking part in a psychological study, in which subjects were exposed to hypothetical equal status Negroes, affected the subjects frequency of petition sign- ing, a comparison was made between those individuals who failed to sign the first petition, who participated in the study, and who received the letter and those individuals who failed to sign the first petition, who did not participate in the study, and who received the letter. No speci- fic hypothesis was put forth. Finally, in order to determine if the combined effects of receiving the letter and participating in the study had a significant effect on the frequency of signing the second petition, those who failed to sign the first petition, who participated in the 30 study, and who received the letter were compared with those who failed to sign the first petition, did not participate in the study, and did not receive the letter. Again no specific hypothesis was formulated. Along with determining the effect of the letter on those individ- uals who failed to sign the first petition, an attempt was made to discover if receiving the letter had any effect on those persons who signed the first petition. It is reasonable to suppose that all of those who endorsed the first petition did not do so with equal convic- tion. No doubt some of these individuals, due to changes in their attitudes, would fail to endorse the second petition.3 Three groups of subjects were established to ascertain if receiving the letter bolstered one's attitude (as inferred from petition signing) towards selling one's home to any qualified individual, regardless of race. The first group of individuals had signed the first petition, participated in the study, and received the letter while the second group of individuals was simi- lar except that they had not participated in the study. The third group of individuals had signed the petition but had neither taken part in the study nor received the letter. The sixth hypothesis tested was that those individuals who had signed the first petition and received the letter would more frequently sign the second petition than those 3. A second factor that would lead to changes in frequency of endorse- ment from the first to the second petition is regression towards the mean. Campbell and Stanley (14) write "Regression effects are in- evitable accompaniments of imperfect test - retest correlations for groups selected for their extremity. Regression is more generally a function of the degree of correlation between scores on the same test administered at two different points in time. The lower the correlation the greater the regression towards the mean." Table 18 shows that the reliability of petition signing behavior was rt= .39. 31 individuals who had signed the firsthpetition but had not received the letter. To determine whether taking part in the study affected frequency of signing the second petition a comparison was made between those per- sons who signed the first petition, participated in the study, and who received the letter and those who signed the first petition, received the letter, but did not participate in the study. In order to determine the combined effects of taking part in the study and receiving the letter on those persons who signed the first petition, those who both participated in the study and received the letter were compared to those who did not participate in the study and who did not receive the letter. THIRD STUDY The third study is actually an elaboration of the fourth hypothesis in the first study. 'In that hypothesis it was posited that when the hypothetical persons were evaluated as home purchasers, a relationship would exist between racial discrimination scores and petition signing behavior. The discrimination scores might be interpreted as verbal ex- pressions of behavioral intentions in that low scores may be thought to indicate a willingness on the part of the subject to sell his home to a Negro while high scores might be thought to indicate a reluctance to engage in the same behavior. In the third study an attempt was made to discover what variables affect the strength of the relationships be- tween verbal expression of behavioral intention and overt behavior. Rokeach (48) writes "an attitude is a relatively enduring organiza- tion of beliefs around an object or situation predisposing one to res- pond in some preferential manner." Like other definitions of the same concept, this definition implicitly contains the hypothesis that behav- ior toward a specific object in a specific situation is related to an underlying constellation of attitudes. Given that the same attitudes 32 become salient whenever a subject in a psychology experiment is con- fronted with a specific attitude object in a specific situation, usually through the vehicle of an attitude survey, as compared to when he actu- ally encounters the attitude object in that very same situation in the course of his everyday affairs, we might reasonably expect that the behavior exhibited in both instances to be congruent. Unfortunately this expectation has rarely been realized. Far more frequently the results of the very few studies (38) done in this area support the proposition that there is no relationship between verbal behavior toward an attitude object placed in a specific situation and overt behavior manifest toward the identical object - situation stimulus. La Pierre (33) found that although over 200 proprietors of auto camps and restaurants in fact did provide service to a Chinese couple, approx- imately 90% of them replied to a questionnaire that they would not accept Chinese as guests. Kutner, Wilkins, and Yarrow (32) found that in fact managers of 11 restaurants served a racially mixed group of three women, but when subsequently for reservations for a racially mixed group, six of these managers declined to make them while all eleven accepted reservations over the phone for an all white group made the same day. Saenger and Gilbert (53) discovered that there was no differ- ence between subjects who made purchases from white sales clerks as com- pared to black sales clerks in their responses to the question, "what would you think if all New York department stores hired Negro sales persons?" Unlike the three studies just mentioned, in which overt behavior was significantly less racially discriminatory than corresponding verbal behavior, Linn (34) reports a study in which overt behavior reflected 33 more racial discrimination than did the subjects' verbal behavior. While verbally indicating, on the average, that they would allow a picture of themselves and a Negro of the opposite sex to be placed in the school paper, the subjects were only willing to sign picture re- lease forms that would allow the picture to be displayed in journals read exclusively by psychologists and sociologists. McGrew (38) relates a study in which ten landlords both advertised apartments for rent and said they would rent to Negroes when questioned by phone, although six of them in fact refused to do so when approached by a Negro couple. Caucasian couples, coming one hour before and one hour after the arrival of the Negro couple, were shown the apartments for rent without hesita- tion. A common virtue of the studies heretofore presented is that the situation in which the attitude object was depicted when measuring verbal behavior was very similar if not identical to the situation in which the attitude object was actually encountered. Other studies at- tempting to relate verbal behavior to overt behavior have been less precise and consequently might not be expected to find a relationship between the two. Berg (2) administered the Ethnocentrism Scale, the Fascism Scale, and Bogardus' Social Distance Scale to obtain measures of opinion with regard to Negroes. He found that those who gave verbal responses from which high anti-Negro prejudice might be inferred were no less influenced in the autokenetic situation by the judgments of a Negro confederate as compared to the judgments of a Caucasian confederate than those who gave verbal responses from which low anti-Negro prejudice might be inferred. One investigation which shows a relationship between verbal 34 behavior and subsequent overt behavior is that of Defleur and Westie (17). In order to get a measure of verbal behavior all subjects were asked to complete the Summated Differences Scale. By adding the differ- ences in ratings given to Negroes and Caucasians of the same occupation- al status placed in the same situations, it was possible to determine to what extent subjects verbally discriminate on the factor of race. Overt behavior was measured by the degree to which a subject would allow a picture of himself with a Negro of the opposite sex to be circulated. This was indicated by the actual signing or refusal to sign of various picture release forms arranged on a Guttman type scale. A significant relationship existed between verbal discrimination against Negroes and refusing to sign picture release forms. This was the only study that could be found that supported the contention that corresponding verbal and objectively measured overt behavior vary together. One of two alternative conclusions may be drawn from the research reviewed. First one might conclude that attitudes are not related to behavior since two measures of behavior, purportedly contingent upon identical underlying attitudes, did not themselves correspond. On the other hand one might conclude that the reason no relationship was found between verbal and overt behavior was not that attitudes and behavior are independent, but that different attitudes had become salient when measuring the two behaviors. This would be due to the fact that differ- ent stimuli were presented to the subjects on the two occasions when measurements of verbal and overt behavior were garnered. For example, Berg elicited verbal expressions of behavioral intentions towards atti- tude objects (Negroes) placed in several situations (those depicted on the Social Distance Scale) and attempted to predict behavior toward the 35 same attitude object in the situation of a psychological experiment. LaPiere asked his subjects how they would react to Chinese when, in fact, what they actually were confronted with on many occasions was a Chinese couple plus a Caucasian, LaPiere himself. The studies of Linn and McGrew manage to present the attitude ob- jects in the identical situations as those in which the subjects encountered them in their everyday affairs. However, the demand charac- teristics (42) of the situations in which the verbal expressions of be- havioral intention were obtained may have created evaluation apprehen- sion (52) which was lacking in the situation in which overt behavior was measured, resulting in the incongruency. Linn points out that the liberal norms of the university subculture, which were absent or less salient when overt behavior was measured, no doubt influenced the verbal expressions of behavioral intentions of his subjects. McGrew, while acquiring the subjects' verbal expression of behavioral intention, in- formed them that it was unlawful to discriminate on race in renting apartments, thus applying social pressure to facilitate a non-discrim- inatory response. The same pressure was absent when overt behavior was measured. Non-identical stimulation in the situations in which verbal and overt behavior were elicited, evoking different attitude constellations resulting in an incongruency between verbal and overt behavior cannot account for the failure of verbal expressions of behavioral intentions and overt behavior to correlate in all of the studies which showed no relationship. Further, in the lone study where the relationship held (Defleur and Westie) the situation in which the attitude object was placed in the questionnaire, did not accurately reflect the situation 36 in which the subjects actually encountered it. Thus failure to present attitude objects in identical situations when attempting to equate to sorts of behavioral responses to them does not preclude establishing a relationship. Yet it would seem that the more similar the stimuli in the situation in which one measures verbal expression of behavioral intention to the stimuli in the situation in which one measures overt behavior, the greater the chance that the behaviors will correspond. Theoretically, this is because the same or very similar attitude con- stellations are involved in each situation. In the present study each subject indicated on a questionnaire whether or not they would sell their home to any financially qualified buyer, regardless of race and whether or not they would sign a petition supporting open housing. As the second question more accurately re- flected the situation in which the behavior (i.e. petition signing) was elicited than the first, it was theorized that the attitudes underlying responses to the second question would more closely approximate the atti- tudes salient in the behavioral situation than attitudes underlying res- ponses to the first question. Thus, one hypothesis tested was that a stronger relationship would exist between verbally indicating one would sign a petition supporting open housing and actually signing such a petition, than between verbally indicating one would sell his home to any financially qualified buyer, regardless of race, and the same be- havior. A second hypothesis tested was that verbally indicating one would sell his home to any financially qualified buyer would be signifi- cantly related to signing a petition supporting open housing. A third ,hypothesis tested was that verbally indicating one would sign a petition supporting open housing would be significantly related to signing such a petition. METHOD The 1960 census (65) reveals that the population of Lansing, Michigan is approximately 110,000 persons, of whom 100,000 are Caucasians and 7,000 are Negroes. Before beginning the study it was necessary to find a middle class exclusively white neighborhood. As the study was conducted almost a decade after the census was taken, that survey was considered virtually useless for determining the present racial composi- tion of the various neighborhoods. As an alternative procedure, indi- viduals active in civil rights were queried as to what areas of the city might be all white. Through information furnished by Dr. Robert Green of Michigan State University the area known as Old Oaks was selected. Within this area, five contiguous streets were arbitrarily chosen from which to draw subjects. All of the living units on these streets, as well as in the remainder of the neighborhood, are single family dwellings. Those who were later to serve as subjects were initially contacted in their homes by coeds at Michigan State University who presented them- selves as volunteer workers for the Urban League. Each individual was asked to sign a petition (Appendix A). Each coed attempted to make it clear to the homeowners that they were not obligated to sign the peti- tion, but that if it accurately reflected their own views, their signa- ture would be greatly appreciated (Appendix B). By signing the petition the individuals pledged not to discriminate on race or ethnic background 37 38 in the selling or renting of their homes. All homes on the designated five streets were called on but once. Individuals who were not home obviously could neither sign nor refuse to sign the petition and there- fore were eliminated from the possible subject pool. As a result of this procedure seventy-three individuals signed the petition while seventy-three refused. From these persons were selected twenty couple who signed the petition and twenty couples that failed to do so. Homes at which both the husband and wife had either signed or refused to sign were selected because in the later stages of the study it would save time to be able to administer questionnaires to two persons at once. Forty households, then, were sent letters (Appendix C) in which it was explained that a study concerning the generation gap was being conducted by the psychol- ogy department at Michigan State. Those who received the letter were asked to participate in the study so as to provide an adult sample to be compared with the students at the University. Two weeks after the petition had been circulated the experimenter began calling on those households that had received the letter. A total of forty-three individuals, representing twenty-three sep- arate households, consented to take part in the study. TWenty-two of them had previously signed the petition while twenty-one had refused to sign. Of those who had signed, twelve were males and ten were females. Of those who refused to sign, ten were males and eleven were females. Sixteen of those who endorsed the petition were Protestant while six were Catholics. Of those who did not sign the petition, four were Cath- olic and seventeen Protestant. The average age of both groups was forty- five. The mean number of years of formal education for those who signed was fifteen whereas for those who refused to sign the corresponding figure was one-half a year less. 39 During the first meeting with the subjects the Adult Attitude and Interest Survey (Appendix D) was administered. Its purpose was to ascer- tain the subject's attitudes toward five issues that were of current interest and thought to be important to the subjects. After completing the questionnaire, all subjects agreed to have the experimenter return in about a month with several more questionnaires. Having determined the attitudes of the subjects, the next task was to construct hypothetical persons for them to evaluate. Eight hypo- thetical persons, who each represented some combination of race (either Caucasian or Negro), socioeconomic status (high or low), and attitude similarity (similar or dissimilar) were tailor-made for each subject. Rather than consisting of checkmarks on a questionnaire, as was the case in most previous studies, each hypothetical person was described to the subject in a six sentence paragraph. Within the first two sentences of the paragraph the hypothetical person's race, family, occupation and educational background were described. In the remaining four sentences his attitudes towards four of the very same issues the subjects had earlier responded to were presented. As the set of eight hypothetical persons was to be presented to the subjects four successive times, four separate booklets were constructed (Appendix E). These were identical except for their first page, which served to place the hypothetical persons in the roles in which the subjects were to evaluate them. First 4. The eight hypothetical persons with which each subject was confronted were; Negro-Similar Attitudes - High Status, Negro-Similar Attitudes- Low Status, Negro-Dissimilar Attitudes-High Status, Negro-Dissimilar Attitudes-Low Status, Caucasian-Similar Attitudes-High Status, Caucasian-Similar Attitudes-Low Status, Caucasian-Dissimilar Atti- tudes-High Status, and Caucasian-Dissimilar Attitudes-Low Status. 40 the hypothetical persons were to be evaluated as friends, then as neigh- bors, thirdly as potential home purchasers, and finally as potential sons-in-law. Below each paragraph describing a single hypothetical per- son were the numbers one to nine. By circling one the subjects could indicate extreme liking toward the hypothetical person. By circling nine the subjects could indicate an intense disliking toward the hypo- thetical person in whatever particular situation he was placed. By subtracting the sum of the ratings given the four Caucasian hypothetical persons from the sum of the ratings given the four Negro hypothetical persons it would be possible to generate four racial discrimination scores per subject, corresponding to the four situations in which the hypothetical persons were placed. Two other questionnaires were also constructed at this time. The first (Appendix F) was designed to tap the subjects' perceptions of their friends, neighbors, and relatives feelings towards Negroes from which perceived social pressure for behaving in a racially discriminatory manner could be inferred. On the first page of this questionnaire sub- jects could indicate how they thought their friends would react to them if they befriended a Negro. Alternative responses ranged from "it would probably raise their impression of me a notch or two as most of them. would sort of admire me" to "all my friends would feel very badly about my associating with a Negro and would very likely devalue me and subse- quently quit seeing me." On the second page subjects could indicate how they thought their neighbors felt about having Negroes as neighbors. Alternative responses ranged from "they would prefer to live in an inte- grated community and thus would be very happy to have a Negro on the block" to "the people would be very unhappy about Negroes moving in and 41 probably would react violently." On the third page subjects could indi- cate how they thought their relatives would feel about a Negro marrying into the family. Again the alternatives ranged from those from which no social pressure facilitating discrimination against Negroes could be inferred to those from which could be inferred much social pressure for racial discrimination. The second questionnaire (Appendix G) was de- signed to obtain verbal expressions of behavioral intention that varied in their degree of correspondence to the behavior of signing the peti- tion supporting open housing. The two relevant items on this question- naire were "I would sign a petition supporting open housing" and "if actually selling my home, I would sell it to any financially qualified individual, Negro or Caucasian." The latter item also served a second function in that it provided an independent measure of the subjects' willingness to sell their homes to Negroes. Eight weeks after the subjects had seen the first petition - six weeks after they had filled out the first questionnaire, the experi- menter returned with the questionnaires just described. Before being asked to complete them, the subjects were reminded that the purpose of the study was to measure the generation gap by comparing their attitudes to those of the students at Michigan State. The subjects were then asked to evaluate the hypothetical persons, first as friends, then as neighbors, as home purchasers, and finally as sons-in-law. Subjects were instructed to read the description of each hypothetical person as many times as they liked, but once having made their evaluation, not to refer back to it. As soon as the subjects finished one booklet, it was taken from them and they were given another to complete. This was done to prevent them from looking back to determine what rating they had 42 assigned a particular hypothetical person previously. Upon completing their evaluations of the hypothetical persons, the subjects were given the two remaining forms to fill out. The first being that which tapped perceived social pressure to discriminate against Negroes while the second measured verbal expressions of behavioral intention. As husbands and wives were usually filling out the questionnaires simultaneously, they sat apart from one another so that the responses of one had no influence on the responses of the other. After finishing all three questionnaires the subjects were thanked for participating in the study and told that they would be sent a preliminary report of the findings. Most of them seemed genuinely interested in having this information. At this point the subjects perceived the study to be over. One month later the forty-three subjects received a letter (Appendix H) which explained that the true purpose of the study was to determine what traits middle class homeowners deem important in select- ing friends, neighbors, persons to sell their homes to, and potential sons-in-law. Among the other results reported,_particular emphasis was given to the finding that homeowners in Old Oaks did not discriminate against Negroes when they were evaluated as neighbors. It was this in- formation that was designed to manipulate the subjects' perceptions of their neighbors' feelings towards having Negroes as neighbors. In addi- tion to those who had participated in the study, twenty-one persons who had refused to sign the first petition and who had not taken part in the study along with eighteen individuals who had signed the first peti- tion but had not taken part in the study also received the letter. Added to the total sampling frame for the next stage of the study were twenty-one persons who did not receive the letter, had refused to sign TABLE 1 THE VARIOUS EXPERIMENTAL GROUPS Received lst Petition Completed Received Received Groups Signed Did Not Sign Questionnaires 2nd Letter 2nd Petition 1 X X X X 2 X X X X 3 X X X 4 X X X 5 X X 6 X X 43 44 the first petition, and had not participated in the study along with twenty-three persons who did not receive the letter, had signed the first petition, and had not participated in the study. Table 1 depicts the treatments administered to each of the six groups employed in the study. One week after the letter was sent, coeds from Michigan State University representing themselves as members of the Organization for Equal Opportunity began calling on the various groups of persons des- cribed in the previous paragraph. Each individual was asked to sign a petition (Appendix I) supporting open housing if it reflected their own views. By returning several times to the homes of those persons who were not in when first called upon, 94% of those designated to be con- tacted were actually given the chance to either sign or refuse to sign the petition. Roughly thirteen weeks after seeing the first petition the subjects had the opportunity to sign the second, almost identical petition. At this point, the mechanics of the study were concluded. RESULTS To test the first hypothesis that discrimination against Negroes in different situations is related to perceived social pressure to act in a racially discriminatory manner, it was first necessary to ascertain if perceived social pressure to act in a racially discriminatory manner was differentially salient across situations. Figure 1 depicts the subjects' perceptions of their friends reactions to them supposing that they be- friended a Negro. Thirty-four (79%) of the subjects chose alternative B, indicating they thought their friends would think no less of them if they befriended a Negro. It might be inferred that these persons perceive no social pressure to discriminate against Negroes as friends. Six (14%) chose alternative C, indicating that they thought their friends would prefer that they kept all white company, but at the same time, would not impose any sanctions upon them for befriending a Negro. It might be inferred that these persons perceive some social pressure to discrimin- ate against Negroes as friends. Three (7%) subjects selected alternative D indicating that they believe their friends would strongly object to them befriending a Negro. These persons, no doubt, perceive considerable social pressure to discriminate against Negroes when choosing friends. Figure 2 depicts the subjects' perceptions of their neighbors' attitudes towards having Negroes as neighbors. Three (7%) of the subjects chose alternative B, revealing that they did not see their neighbors as ob- jecting to Negroes as neighbors. It might be inferred that these persons perceive no social pressure to discriminate against Negroes when selling their homes. Thirty-two (74%) of the subjects chose alternative C 45 4O 32 24 16 A B C D E Little perceived Much perceived social pressure social pressure Figure l. Perceived Social Pressure in the Situation of Having a Negro for a Friend 46 40 32 24 16 A B C D E Little perceived Much perceived social pressure social pressure Figure 2. Perceived Social Pressure in the Situation of Having a Negro Purchase One's Home 47 48 indicating that although they thought their neighbors would treat a Negro family cordially, they would still prefer Caucasian neighbors. These persons, no doubt, perceive some social pressure not to sell their homes to Negroes. Eight (19%) of the subjects selected alternative D indicating that they think their neighbors would object strongly to having Negroes as neighbors. It may be assumed that these subjects perceive considerable social pressure not to sell their homes to Negroes. Figure 3 depicts the subjects' perceptions of their relatives' reactions if a member of the family should marry a Negro. Two (4%) of the subjects chose alternative B indicating they perceived their relatives as fully accepting a Negro into the family. It might be inferred that these per- sons do not perceive social pressure against having their progeny marry- ing a Negro. Six (14%) of the subjects selected alternative C indicating that they felt their relatives would prefer the family to remain all white. These persons perceive some social pressure against having their progeny marry a Negro. Twenty-eight (65%) of the subjects chose alter- native D while seven (17%) chose alternative E. These individuals see their relatives as strongly objecting to having a Negro in the family and therefore perceive much social pressure against having their children marry a Negro. From an examination of the three figures 35 appears that the subjects, as a group, perceive more social pressure to discriminate against Negroes as members of their families than as potential home pur- chasers while perceiving more social pressure to discriminate against Negroes as potential home purchasers than as friends. In order to test these relationships in a more formal manner, two t—tests for matched groups were conducted. 40 32 24 16 A B C D E Little perceived Much perceived social pressure social pressure Figure 3. Perceived Social Pressure in the Situation of Having a Negro for a Son-in-Law 49 50 TABLE 2 Perceived Social Pressure to Discriminate Aginst Negroes in Various Situations - Inferred from Attitudes Attributed to Friends, Neighbors, and Relatives 3': 32 t df P Friend 2.27 .38 7.00 42 p<.001 Home Purchaser 3.11 .28 7.29 42 P<3001 Son-in-law 3.93 .49 t-tests for matched groups Before doing this however, it was necessary to assign numerical values to the alternatives measuring perceived social pressure to discriminate against Negroes. Alternative A was given the value 1, B = 2, C = 3, D = 4, and E was made equal to 5. The larger the number the greater the perceived social pressure to discriminate against Negroes. The mean per- ceived social pressure score when Negroes were considered for family membership was 3.93 which was significantly greater than the perceived social pressure score (3.11) generated when Negroes were considered as potential home purchasers (t = 7.29, df = 42, p<.001). The latter score was significantly greater than the perceived social pressure score (2.27) when Negroes were placed in the role of friends (t = 7.00, df = 42, p<.001). Given these results, it was expected that the factor of race would become more and more important in determining the evaluations of the hy- pothetical persons as the role in which they were placed changed from that of friend to home purchaser, to son-in-law. Table 3 discloses that this is in fact the case. Approximately .3% of the variance in ratings Table 3 AMOUNT OF VARIANCE ACCOUNTED FOR BY THE TRAITS HYPOTHETICAL PERSONS IN FOUR SITUATIONS OF THE Race (A) Attitudes (B) Status (C) Subjects A x B A x C B x C (D) A x B x C Residual Interactions Friend .3% 64.7% 7.0% 28.0% Neighbor 56.5% .7% 13.0% 29.0% Home Purchaser 1.8% 35.3% 3.5% 19.0% 40.0% Son-in-Law 18.3% 19.1% 4.0% 15.0% 4.4% 2.0% 1.4% .8% 35.0% 51 52 assigned to the hypothetical persons in the role of friend was due to their race while 1.8% of the variance ratings assigned to the hypotheti- cal persons placed in the role of potential home purchasers was due to the factor of race. About 18% of the variance ratings assigned to the hpothetical persons when evaluated as sons-in-law was accounted for by their race.5 Thus as the role the hypothetical persons are placed in changes from friend to home purchaser to son-in-law the race of the hy- pothetical persons comes to have a greater bearing on the ratings given by the subjects. In order to determine if these differences in the strength of the race factor are significant, four racial discrimination scores were cal- culated for each subject corresponding to the four situations in which the hypothetical persons were evaluated. Because a constant of 9 was added to all of the racial discrimination scores (so that there would be no negative numbers), the value 9 represented a lack of discrimination on the factor of race. Numbers less than 9 indicated discrimination against Caucasians while numbers greater than 9 indicated discrimination against Negroes. 5. The values presented in Table 3 were calculated by dividing the sums of squares for each effect by the total sums of squares. This pro- cedure yields a rough estimate of the proportion of the tOtal vari— ance accounted for by each particular effect. 53 TABLE 4 Discrimination on the Race Factor as a Function of the Situation in Which the Hypothetical Persons are Evaluated 2 i s t df P Friend 7.79 8.27 4.19 42 p<.001 Home Purchaser 12.11 34.07 5.56 42 p<.001 Son-in-Law 18.55 43.84 t-tests for matched groups, one-tailed The first comparison depicted in Table 4 shows that subjects discriminate significantly more against Negroes when rating the hypothetical persons as home purchasers than when rating them as friends (t = 4.19, df = 42, p<.001) while the second comparison reveals that subjects discriminate significantly more against Negroes when rating the hypothetical persons as sons-in-law than as home purchasers (t = 5.56, df = 42, p<.001). It may be concluded that middle class Caucasian homeowners discriminate sig- nificantly more against Negroes when they are thought of as sons-in-law than when thought of as home purchasers while discriminating signi- ficantly more against Negroes as home purchasers than as friends. The first hypothesis is strongly supported. The second hypothesis holds that subjects will discriminate on race to a significantly greater extent when the hypothetical persons are evaluated as potential home purchasers than when evaluated as neighbors. Table 3 shows that when the hypothetical persons were evaluated as neighbors, none of the variance in ratings was accounted for by race while 1.8% of the variance in ratings was due to the race of the hypo- thetical persons when they were evaluated as home purchasers. 54 TABLE 5 Negroes as Neighbors and Negroes as Home Purchasers — The Differential Significance of Race as a Function of the Situation i 32 t df P Neighbor 9.06 7.22 3.61 42 p<.01 Home-Purchaser 12.11 34.07 t-test for matched groups, one tailed The mean racial discrimination scores presented in Table 5 reveals that the subjects discriminate significantly more against Negroes when the hypothetical persons are evaluated as home purchasers than when they are evaluated as neighbors (t = 3.61, df = 42, p<.01). The second hypothesis is strongly supported. The third hypothesis tested was that there would be an inverse re- lationship between perceived social pressure to act in a racially dis— criminatory manner and the amount of variance accounted for by attitude similarity in evaluations of the hypothetical persons. Earlier in this section it was established that perceived social pressure to discrimin- ate against Negroes increases significantly as the role in which the Negro is encountered changes from that of friend to home purchaser to son-in-law. Table 3 discloses that about 65% of the variance in ratings was accounted for by attitude similarity when the hypothetical persons were evaluated as friends while 35% of the variance was accounted for by this factor when the hypothetical persons were evaluated as home purchasers. Only 19% of the variance in ratings was due to the attitudes attributed to the hypothetical persons when they were evaluated as sons-in-law. To determine if attitude similarity had a significantly 55 greater effect in some situations as opposed to others, three belief discrimination scores were calculated for each subject corresponding to three of the situations in which the hypothetical persons were evaluated.6 TABLE 6 Discrimination on the Attitude Factor as a Function of the Situation in Which the Hypothetical Persons are Evaluated X 32 t df Significance Friend 27.67 69.72 3.28 42 p<,01 Home Purchaser 23.13 77.40 3.13 42 p<.01 Son-in-Law 18.60 47.10 t-tests for matched groups, one tailed Table 6 reveals that the subjects discriminate significantly more against hypothetical persons who espouse dissimilar attitudes when these hypo- thetical persons are evaluated as friends than.when they are evaluated as home purchasers (t = 3.28, df = 42, p<301). Further, the subjects discriminate significantly more against hypothetical persons who hold dissimilar attitudes when they are evaluated as home purchasers than when they are evaluated as sons-in-law (t = 3.13, df = 42, p<.01). 6. These were calculated in the same way as racial discrimination scores. The sum of the ratings assigned the four hypothetical persons who held similar attitudes to those of the subject was subtracted from the sum of the ratings assigned to the four hypothetical persons who held dissimilar attitudes. A constant of 9 was then added to each score so that the belief discrimination scores would be comparable to the racial discrimination scores. Belief discrimination scores were not calculated for the subjects ratings of the hypothetical persons placed in the role of neighbor as this score was not necessary for the testing of any particular hypothesis. 56 As a result of these findings, the third hypothesis is strongly supported. The fourth hypothesis tested was that individuals who signed the petition would discriminate significantly less on the factor of race, when evaluating the hypothetical persons as potential home purchasers, than those individuals who refused to sign the petition. TABLE 7 Racial Discrimination Scores of 21 Subjects who Refused to Sign a Petition for Open Housing Compared to 22 Subjects who did Sign Such a Petition When the Hypothetical Persons were Evaluated as a Home Purchaser 'X 82 t df Significance Signed Petition 10.09 14.91 2.44 41 p<.01 Refused to Sign 14.23 43.36 One tailed t-test Table 7 shows that the twenty-one subjects who refused to sign the peti- tion generated a significantly higher average racial discrimination score when rating the hypothetical persons in the role of home purchasers than did the twenty-two subjects who did sign the petition (t = 2.44, df = 41, p<,01). Consequently the fourth hypothesis is strongly supported. It is of interest to know whether those who signed the petition differed significantly from those who refused to sign in evaluating the hypothetical persons when they were placed in the roles of friend, neigh- bor, and son-in-law. It might be expected that those who refused to sign the petition might discriminate against Negroes to a greater degree in all of the situations in which the hypothetical persons were en- countered. 57 TABLE 8 Racial Discrimination Scores of 21 Subjects who Refused to Sign a Petition for Open Housing Compared to 22 Subjects who did Sign Such a Petition When the Hypothetical Persons were Evaluated as Friends X 32 t df Sighificance Signed Petition 7.86 5.95 -.16 41 N S Refused to Sign 7.71 11.49 One tailed t-test Table 8 discloses that there is no difference between the racial discrim- ination scores of those who signed the petition and those who refused to sign when the hypothetical persons were placed in the role of friends. TABLE 9 Racial Discrimination Scores of 21 Subjects who Refused to Sign a Petition for Open Housing Compared to 22 Subjects who did Sign Such a Petition When the Hypothetical Persons were Evaluated as Neighbors 'X S2 t df Significance Signed Petition 8.72 5.10 .86 41 N S Refused to Sign 9.42 9.06 One tailed t-test Similarly, Table 9 shows that there is no difference between the racial discrimination scores of those who signed the petition and those who refused to sign when the hypothetical persons were placed in the role of neighbors. 58 TABLE 10 Racial Discrimination Scores of 21 Subjects who Refused to Sign a Petition for Open Housing Compared to 22 Subjects who did Sign such a Petition when the Hypothetical Persons were Evaluated as Sons-in-Law 2 ‘X S t df Significance Signed Petition 17.09 50.26 1.72 41 p<.05 Refused to Sign 20.47 31.02 One tailed t-test On the other hand, Table 10 reveals that there is a significant difference between the racial discrimination scores of those who signed the petition and those who refused to sign when the hypothetical persons were placed in the role of sons-in-law (t = 1.72, df = 41, p<.05). The fifth hypothesis tested was that those who said they would sell their homes to any financially qualified buyer would have significantly smaller racial discrimination scores than those subjects who said they would not sell their homes to any financially qualified buyer, when the hypothetical persons were rated as home purchasers. TABLE 11 Relationship Between Stating "I would sell my home to any financially qualified buyer, regardless of race" and Racial Discrimination Scores Given the Hypothetical Persons as Potential Home Purchasers '2 S t df Significance I would sell 10.93 28.66 1.98 41 p<,05 I would not sell 14.85 35.36 One tailed t-test 59 As the figures presented in Table 11 reveal, those indivduals who said they would not sell their homes to any qualified buyer generated signi- ficantly larger racial discrimination scores when evaluating the hypo- thetical persons as home purchasers than those individuals who said they would sell their homes to any qualified buyer, Negro or Caucasian (t = 1.98, df = 41, p<.05). Thus the data confirm the fifth hypothesis. The sixth hypothesis to be tested is that the situation or role in which the hypothetical persons are placed accounts for more of the vari- ance in racial discrimination scores than does the individuals' person- alities. A 4 x 43 analysis of variance was conducfixlwith each column representing one of the four situations in which the hypothetical persons were placed while each row represented one of the 43 subjects. A single racial discrimination score was available for each of the resulting 172 cells. To determine what proportion of the total variance is due to the individuals' personalities, the mean square of error was subtracted from the mean square of rows and the difference was divided by the number of columns multiplied by the number of observations per cell. The result- ing value was termed the estimated row variance, 0:3, and was divided by the estimated total variance, o-t, which itself consisted of the esti- mated variances of rows, columns, and error summed. To determine what proportion of the total variance is due to the situations, the mean square of error was subtracted from the mean square of columns and the difference was divided by the number of rows multiplied by the number of observations per cell. The resulting value was termed the esti- 2 mated column variance, oz , and was divided by the estimated total 60 7 variance, described previously (28). TABLE 12 Proportion of the Total Variance in Racial Discrimination Scores Accounted for by the Situation in which the Hypothetical Person was Placed and Accounted for by Individual Differences SS df M S o— % of Variance Situations 2934 3 978 22.25 47% Subjects 1487 42 35.40 3.57 8% Error and 2661 126 21.11 21.11 Interaction 7. Sheffe (54) shows that the Expected Mean ngare of columns in a mixe model consists of error variance, o-e , nteraction variance OCAB , and variance due to column effects, o—A . The Expected Mean Squ re of rows on the other hand consists of only two factors - o—e , error variance, and variance due to the main effect of rows o—R . Because there is but one observation per cell the Expected Mean Square of error in th's particular design contains not only variance due to error, o-e , but als? variance due to the inter- action of the two main effects, O—AB . To the degree that there are interaction effects, subtracting the obtained Mean Square of error from the Mean Square of rows, as was done, will yield an overly conservative estimate of the amount of variance accounted for by rows, as interaction effects as well as error variance are being subtracted from that main effect. Winer (70) presents a way of separating the sumszof squares due to error into its two component parts, o-e and o-AB . This was done and the resulting Mean Square of error gas reduced from 21.11 to 18. The la ter value was then used when o-R was estimated. The value of o-R becomes 5.8 and the row effect comes to account for 11% of the total variance. An alternative method that can be used to de- termine the proportion of variance accounted for is to divide the Sums of Squares for each effect by the total sums of squares. Using this method, the situations account for 41% of the total variance while differences between subjects account for 20%. Re— gardless of the method used, it appears as though the situations account for more of the variance in racial discrimination scores than do individual differences between the subjects. 61 In Table 12 it is seen that while differences between subjects, which may be thought of as personality differences, account for 8% of the total variance in racial discrimination scores, differences in the situation account for 47% of the total variance. Although no significance tests were conducted between these two values, the available data clearly supports the sixth hypothesis. The seventh hypothesis tested was that individuals who disciminate against Negroes more than average in one situation also discriminate against Negroes more than average in other situations. If this hypo- thesis is true and subjects are consistent in the amount they discrim- inate against Negroes relative to the other subjects across situations, large correlations should be obtained between racial discrimination scores in the various situations. TABLE 13 Subjects' Consistency in Discriminating Against Negroes in Four Situations as Expressed by Correlations Between Racial Discrimination Scores Friend Neighbor Home Purchaser Son-in-Law Friend - .20 -.07 .06 Neighbor - - .36* .27 Home Purchaser - - - .38* Son-in-Law - - - _ * p mo oossom 133 m000.0 >H>.0 ®0H.0 mza.o Hmm.o :H0.0 m000.0 Nam.o eoesoausemam a... MH.0 ON.N NN.N mm.o mm.>mm 00.0 oaemseeem e mam mN.N N: om.m N: ma.N N: mN.o a mm.a N: Na.m a mo.a N: me.m N Ne.a N: Nm.N a Na.m N: Hm.om N mm.m N: om.m:ma a ma.N N: Ho.o a osmswm new: .mm mommUHmz mm.momm mm.am mm.moa mm.am mN.o mm.ms Na.m mm.os mN.m em.ms so.a ms.oma am.om :m.omm om.msma me.om ao.o Hmpoa nom< u N Amputeesmv a 00m< n N 0mm 00m u N om 00< u N o< 0m<_n N m< no u N Amspmpmv 0 am u N Aheespaee mo monsom 13h Hdgoa comet N Ampoefinst a nom< u N om< mom a N om ao< u N o< mm< u N m< no u N Aespepmv N am I N Ameespaepev m a<_u N Aoommv < mam mN.NNmN 0:.N N: mm.ooa mooo.o mm.m a:.ma Na mm.mmm 0:.N N: mN.ooa Nem.o Na.o H:.o a H:.o mm.N N: ss.mm moo.o oo.m s:.w a ss.m me.a N: es.ms oom.o oN.a NN.N a NN.N m:.H N: NH.N@ emo.o mm.m Na.m H Na.m mm.m N: ow.msm mooo.o mm.ma em.:oa a sm.soa Ne.w N: so.mom mooo.o NN.NHH om.Nmoa a om.Nmoa ma.: N: mm.ssa Noo.o mm.NH mH.Nm N mH.Nm mocdowwanwflm oapmfipspmm osmdum new: mm. monsoon No 85w mmm no moanom 135 m000.0 H00.0 m000.0 m000.0 m000.0 m000.0 m000.0 m000.0 N0.m :N.NH mm.0a Ns.ma is: Nw.mm Hm.HHH mm.mm mm.a sw.m em.a so.:m mm.N se.Nm mm.m ma.mm mm.m ma.NNH Ne.N mm.oaa mo.a so.mmm se.m m:.m0m woeeeausamsm usemspepm e osnmmw new: mzm N: Na 3ma .ho.mmm ma.omN ma.m0m HSNOB nom<.u N Ampoennsmv a nom< u N om< mom a N om ao< u N o< nme u N m< no u N Amaeepmv 0 am I N Ameeepspp mo mohdow 136 AUG 6 1959‘ MICHIG GNA NSTATE UNIVE ESR ITY LIB BERARI ES III II III” III II IIII IIII