Race versus belief about race as determinants of attraction: Belief prejudice and two kinds of race prejudice

Race versus belief about race as determinants of attraction: Belief prejudice and two kinds of race prejudice

JOURNAL OF RESEARCH Race IN Versus 143-164 (1973) PERSONALITY 7, Belief About of Attraction: Kinds Belief of CLYDE HENDRICK Kent Race Ra...

1MB Sizes 1 Downloads 51 Views

JOURNAL

OF RESEARCH

Race

IN

Versus

143-164 (1973)

PERSONALITY

7,

Belief

About

of Attraction: Kinds

Belief of

CLYDE HENDRICK Kent

Race

Race Prejudice

and

Two

Prejudice1

AND DONNA State

as Determinants

K. RUMENIK

University

An experiment was conducted to assess the effects of actual racial membership versus similarity of beliefs about racial issues on interpersonal evaluations. Prejudiced and unprejudiced White male students observed a videotaped interaction of two Black and two White male actors discussing racial issues and then evaluated the actors on several scales. One White and one Black actor took a conservative stand on racial issues, and the other White and Black pair of actors took a liberal position. Results from the ratings showed strong belief prejudice effects. Prejudiced subjects were attracted to conservative actors while unprejudiced subjects were attracted to liberal actors. Additionally, there were two kinds of race effects. Prejudiced subjects preferred White to Black actors, but unprejudiced subjects showed a reverse bias, preferring Black to White actors. A tentative explanation for this unusual finding was proposed.

Numerous studies have explored the issue of whether belief similarity or racial membership is the more potent determinant of prejudice and discriminatory responses (Rokeach, Smith, & Evans, 1960). Most, of the studies have supported Rokeach’s contention that belief similarity is the more important variable. However, these studies are susceptible to a variety of methodological criticisms, including questioning the logic of comparison of the power of two independent variables (Byrne, 1969, p. 651, the unrealistic nature of the experimental situations (Sears & Abeles, ‘This research was supported by Grant MH 18652-01 to the first. author from the National Institute of Mental Health. The report is based on a master’s thesis submitted to Kent State University by the second author who is now at MacMurray College. Requests for reprints should be sent to Clyde Hendrick, Department of Psychology, Kent State University, Kent, Ohio, 44242. The authors’ thanks are due to the Department of Instructional Television at Kent State University for taping the interaction sequences. The authors are grateful to Donald Thigpen, Charles Englehart, Roger McElroy, and C. Scully Stikes for serving as actors in the experiment. 148 Copyright @ 1973 by Academic Press, Inc. All rights of reproduction in any form reserved.

RACE

VERSUS

BELIEF

149

1969, p. 273), mutual recriminations concerning the strength of the manipulations of the race and belief variables (Triandis, 1961; Rokeach, 1961)) and the possible reactive nature of the measurements in some of the experiments (Campbell, 1969, p. 374). Despite the methodological problems, the race-belief issue is important as Hendrick, Bixenstine, and Hawkins (1971) noted. If race prejudice is a special case of belief prejudice, then social policy considerations for the amelioration of race prejudice would be quite different than if the two types of prejudice are independent. In an attempt to solve some of the methodological problems and to provide a fair test of relative strength of the race and belief variables, Hendrick, Bixenstine, and Hawkins (1971) videotaped two White and two Black male actors discussing an issue (war in Vietnam). Both prowar and antiwar Xs viewed the tape and evaluated the actors on several scales. The results showed a great many effects due to belief similarity, but few effects due to race. These results strongly supported Rokeach’s theory. Further, it was difficult to argue that race was less powerfully manipulated than belief because the race of the actors was constantly associated visually on the videotape with the statement of their positions. Written comments obtained from Ss in the Hendrick et al. (1971) study indicated that they found the experiment interesting and involving. However, several Ss expressed the opinion that it would have been better to have presented the panel live before the audience. Such an experiment was subsequently conducted (Hendrick, Stikes, & Murray, 1972) in which an audience observed a panel of two Blacks and two Whites discussing a relevant local issue (ROTC on campus). Audience evaluations of the panel members showed strong effects of belief similarity on the issue, but few race effects. In an attempt to achieve still more realism and experimental impact the live discussion format was used in a series of three experiments (Hendrick, Stikes, Murray, & Puthoff, 1973). Instead of passive observation of the discussion, Ss were included as part of the discussion group. Each group consisted of five persons, one actual S and four confederates. Two confederates were Black and two were White. One White-Black pair took positions similar to the S’s position on the issue, and the other White-Black pair took opposing positions. The discussions were lively and highly involving. However, again the results showed powerful effects of belief similarity, but few effects due to the race of the actors. It seems clear that immediacy, subject involvement, and experimental realism will not, per se, elicit prejudiced racial responding. Further, a wide variety of measures have been used in this series of experiments, so that differential sensitivity of the rating measures does not appear to be

150

HENDRICK

AND

RUMENIK

a critical factor (e.g., Triandis & Davis, 1965). The general lack of race effects in most studies poses something of an anomaly, because such lack of effects contrasts strongly with the racial and ethnic hostility which exists in the society at large. One possible solution to this problem was provided recently by Dienstbier (1972) who criticized previous research on the race-belief issue. He argued that a meaningful manipulation of the belief variable should consist of manipulated beliefs about race. “The hypothesis underlying the research of this paper is simply that it is not possible to consider realistic and racially relevant belief manipulations to be independent of the race manipulations; these two types of variables are, when realistically presented, interactive” (Dienstbier, 1972, p. 152). The direct implication is that race prejudice effects are elicited most strongly, and perhaps only when the focus of the belief dimension is race per se. Prejudice may be based on and elicited, not so much by physical presence or actions of outgroup members, but by a stereotyped concept of the outgroup held by the prejudiced individual. For prejudice to occur the stereotype must be activated. Since a large component of such stereotypes is linguistic and symbolic, it follows that the outgroup stereotype will be activated most powerfully by linguistic devices; e.g., discussion of racial issues. In this view prejudice is more a matter of the activating power of words rather than the physical presence or salience of specific outgroup members. Prejudiced responding should, therefore, be elicited when a race concept is activated within the context of a group discussion about race. It is in this sense that Dienstbier meant that race and belief manipulations are inherently interactive. The present experiment was conducted in order to test the hypothesis that racially prejudiced attitudes will be activated by a racially relevant discussion and will result in racially prejudiced evaluations. The experiment was similar in format to the Hendrick, Bixenstine, and Hawkins (1971) study. Subjects who were either high or low in initial prejudice toward Blacks viewed two White and two Black videotaped actors discussing racial issues. One White and one Black actor took a liberal stand on racial issues, and the other pair of actors took a conservative stand. The design allowed assessment of belief similarity effects as well as effects due to race of actors. METHOD Under the guise of serving in an experiment on accuracy of personality impressions, male college students viewed one of two videotaped versions of four young men discussing racial issues. After viewing the discussion 8s rated each of the four stimulus persons on several measures.

RACE

Stimulus

VERSUS

BELIEF

151

Tapes

The same four actors in the previous experiment (Hendrick, Bixenstine, & Hawkins, 1971) also served as actors in the present experiment. Use of the same actors provided a valuable control for comparison of the results of the present experiment against previous results. The men were videotaped sitting in a semicircle conducting a discussion of racial issues. One White and one Black took a strong liberal stand on race, while the second White and Black pair took a strong conservative stand. After the first session liberal and conservative roles were reversed among the actors, and a second session was taped. Each session lasted about 20 min. The Black actors will be designated B, and B2, and the White actors WI and W,. The seating arrangement for both tapes was B,, W,, WI, B1. Actors B1 and W, always took the same stand on the race issue (liberal in Tape 1, conservative in Tape 2), and Bz and W, took the opposing stand. For each tape beginning with B, and proceeding from left to right, each actor made a 1-min statement of his beliefs. After these initial statements the actors engaged in “free” discussions with each other. No attempt was made to write a specific script, although the actors spent several hours in rehearsal prior to the taping session. The net result was a powerful and rather dramatic confrontation of the actors with each other on the race issue in what appeared to be spontaneous and genuine presentation of beliefs. The topics discussed varied widely and included black unity, black values, poverty. institutionalized racism, housing, violence, and education. The liberal position supported the beliefs of the Black movement as they are typically expressed today, while the conservative position maintained a traditional stereotypical viewpoint. For example, on the issue of poverty the liberal White and Black position condemned welfare as presently handled, maintaining that it keeps Blacks poor while enriching bureaucrats who administer the welfare system. The conservative position, on the other hand, condemned blacks for not wanting to work, for failing to hold jobs, and in general for failing to live up to the values required for regular employment. In summary, the White conservative was presented as a traditionalist but he did not express blatant racist opinions per se. The Black conservative subscribed to middleclass occupational and moral values and was in complete agreement with the White conservative on the issues. Both the White and Black liberal positions were presented as strongly liberal, but not to the point of extreme radicalism. An informal content analysis of the tapes indicated that the positions expressed in Tape 2 were slightly stronger than the comparable presentation in Tape 1. Subjects

and Procedure

The Ss were 40 males selected from a large pool of introductory psycho’ogp students who were administered a 53-item questionnaire early in the quarter. This questionnaire included 16 items (each rated on a g-point scale) relevant to racial attitudes. As examples, six of these items were: 1. Institutional racism and exploitation are realities of American society. 2. The “Black movement” is infringing too much upon other people’s freedom. 3. The majority of Blacks are lazy and don’t want to work as indicated by number of them on welfare. 4. Cries of police brutality in the Black community seem unjustified,

the

152 5. The 6. The done.

HENDRICK

“American Dream” applies current emphasis on Black

AND

RUMENIK

only to White people. history and Black studies

programs

seems

ovcrl)

The items were designed to tap current specific issues as well as more general affective reactions. An attempt was made to include a sufficient number of issues so that differential response to the entire set could justifiably be designated as “prejudiced” and “unprejudiced” racial responding. Some but not all of the items were relevant to the topics discussed by the actors in the videotaped presentations. Likewise, more issues were included in the taped discussions than were assc>ssed by thr attitude items. Some overlap of items with discussion topics was desirable (indeed. necessary) to provide a basis for Ss’ who viewed thr tapes to discriminate differcncc~a in belief similarity relative to the actors on the topic, of race. Ten of the attitude items were phrased in an anti-Black (pt’cjudiccd) direction. and six were in a pro direction. The scoring on the latter items was reversed and no overall sum score was derived whieh could range from l&144. the higher the score the more prejudice exhibited. On the basis of these sum scores. male students wercl recruited to serve in the experiment. Ten high- and 10 low-prejudiced Ss vicmrtl Tape 1, and another 10 high- and 10 low-prejudiced Ss viewed Tape 2. Three of the subjects were Rlack and were replaced with White 8s. The mpan prcajudicr s:‘ores were: Tape 1, high prejudice 105.3, low prejudice 43.9; Tape 2, high prejudice 104.7. low prejudice 42.6. An analysis of variance of these scores showed that thr main rffrc( of 8s prejudice was signifirant (F(1,36) = 407.63, p < .OOl), but nrithcr the main effect of tapes nor the interaction was significant. Thus, assignment of Ss to t,aprs on the basis of their initial prejudice scores was successful. The procedure followed that of Hendrick. Rixenstinc. and Hawkins (1971, p, 252) quite closely. Briefly, the cover story indicated to Ss that the expcrimrnt was roncerned with accuracy of personality impressions. The nature of the issue to br discussed was indicated, as well as the fact that both liberal and conservative positions would be presented (in order “to generate some depth of feeling in the interaction”). Both Blacks and Whites were presented in order ‘(to ensure grnerality of the trait of perceptual accuracy.” After observing one of the tapes Ss completed an “interpersonal prrception questionnaire” which contained 21 rating measures for each of the four actors. The measures assessed liking, dominance in the discussion, value of thr contribution. clearness of presentation, similarity of the actor to the S, eight trait adjertivts. and an eight-item social distance scale. Each rating was made on a O-point scale with values ranging from -4 to +4. The endpoints (and for the adjcctivrs. the midpoints also) of each scale were appropriately labeled. While Ss were making their ratings. a large photo of each actor was on display on top of the television monitor. Each actor was numbered both on the photo and on the rating forms. For further distinctiveness the color of the paper of the rating form alternated between red and white for different2 actors. Several of the 24 possible orders of presentation of t.hp four actors were used, and Ss were randomly assigned to orders. After the booklets were completed Ss were asked to complete a one-page form which ascertained prior acquaintance with the actors and assessed suspicion of the procedures. None of the Ss indicated any suspicion. After this form was finished a complete explanation of the experiment was given. The complete design of the experiment was a 2 X 2 X 2 X 2 factorial, including Tapes (1 and 2), 8s’ initial prejudice (high and low). actors’ race (White and Black).

RACE

VERSUS

BELIEF

153

and actors position on the race issue (liberal and conservative). Tapes and Ss’ prejudice were between-S variables, and actors’ race and actors’ belief positionswere within-S variables. RESULTS There were 21 different rating measures including five general evaluative questions, eight trait ratings, and eight social distance items. Each rating was transformed to a I- to g-point basis for ease in computation. A fourway analysis of variance was performed on the data for each measure. The summary F ratios are shown in Table 1. Each analysis yielded 15 orthogonal effects which are shown in the rows of the table. Each of the 21 rating measuresis shown in the column headings. For each effect the number of measures whic.h achieved significance for that effect is shown in the last column of Table 1. Since the table contains 315 separate F tests, several significant effects would be expected by chance alone. Consequently an effect was not interpreted unless more than three of the rating measures achieved significance on that effect By this criterion seven of the 15 effects in the analyses of variance required interpretation. The results will be presented for those seven effects. Prior to that presentation, however, it is desirable to demonstrate the adequacy of the manipulations. Adequacy of Manipulations The main effect of tapes (A) was significant for only one measure, indicating at a gross level at least, that the two versions of the tape resulted in about the same general level of mean ratings. Subjects’ prejudice (B) was significant only on the social distance items, but these differences seemedtheoretically relevant and will be discussedbelow. The tapes X 8s’ prejudice (A X B) interaction was significant on only one item, indicating adequacy of the assignment of Ss to the two between-S variables. The main effect of actors’ belief position (C) was significant on only two measures. This result indicates that, the strength of the liberal and conservative arguments were evenly matched with respect to the liberal and conservative belief positions of the Ss. Otherwise, many of the measures would have shown a significant main effect of actors’ position. This argument is generally true for both tapes since only three of the A X C interactions were significant. Tapes (A) were involved in two triple interactions (A X B X C, A X C X D), but as indicated below these interactions were not problematic. They were due primarily to the somewhat stronger arguments in Tape 2 noted in the Method section. The rating question “to what extent do you agree with the position of this person on the issue” served as a direct check on the adequacy of the

Tapes (A) Subjects’ prejudice (B) AXB Actors’ belief posit.ion (C) AxC BXC AXBXC Actors’ race (D) AxD BxD AXBXD Belief X race (C X I)) AXCXD BXCXD AXBXCXD

1 2:5 .5 1.0 1.0 61.9* 3.2 .7 3.6 6.7* .9 14.7* 2.3 3.2 .5

Perceived similarity of beliefs

TABLE

1

.l .5 2.2 .2 _*3 38.8* 8.7* 3.2 .O 22.6* .2 32.4* 13:2* 2.8 .1

Liking 1.3 .O .4 2.0 .9 9.6* 2.5 81.4* 1.2 .2 1.9 24.0* 10.9* .6 1.8

Dominance in discussion 1.7 .O 1 :o .O 25.7* 9.7* 9.6* .6 .l 3.3 .O 2.6 1.3 .9

Value of contribution .9 .6 1.4 .3 .O 4.5* 3.4 7.9* .O 2.9 .O 2.6 7.0* 1.3 1.6

Clearness of position .l .9 1 :5 .2 4.0* 5.1* .O 10.0* .8 1.8 10.1* .2 3.5 5.2*

Opinionated

SUMMARY F I~ATIOS FOR EACH BOATING ~~~RURE*

1.2 .O 2.6 1.4 .7 14.1* 9.4* 8.6* 1.1 3.3 .3 10.4* .<5 .2 3.6

Intelligent --

9 1.3 .8 2.0 .6 20.1s 1.7 1.6 .6 3.3 .8 44.6* .l .2 .6

Openminded

.2 1 :8 9.6* 4.1* 17.4* 4.1* .O 1.7 5.6* 6 32.7* .2 5 :7

Friendly

1.5 1.2 6.9* .1 8.9* .7 .6 4.3* .4 1.6 .l 6.0* 1 :1 .l

Relaxed

a Degrees of freedom * p < .05.

-

.l 13.8* 10.2* .6 3.9 .S .O .O 4.3* 12.0* .3

W) AXC BxC AXBXC Actors’ race (D) AXD BXD AxBXD Belief x race (C X D> AxCXD BxCXD AXBXCXD

7.2* 2.6 .5 4.0* 2.5 1.1 .1 46.8* 1.6 .O .4

7.3* .o .4 8.2*

Courteous

= 1,36 for each effect.

22.1* 4.9* 6.8* .6 .9 .2 4.3% .8 .3 2.4

2

3.0 .o .o 3.5

2.5 3 1 .2 .2

Subjects’ prejudice (B) -4 x B Act,ors’ belief position

Tapes (A)

Informed

Sincere

.O 48.3* 10.1* .O . .3 29.4* .l 29.7* .9 .2 .7

1.0 7.5* 3 3 2.4

Would work with

.O 34.8* 12.2* .6 .I 29.0* .1 16.4* .l .2 1.9

1.7 6.3* 1.5 .2

Would party with

TABLE

.5 63.1* 8.0* .O .O 45.0* 1.3 22.2* 3.2 .4 .4

.8 19.2* 1.3 .O

Have home to dinner

.9 50.4* 9.8* 4.0* .O 38.7* .O 18.6* .O .3 .O

2.0 15.9* .8 1.0

Would like next door

1 (Continued)

.4 60.1* 15.3* .O .a 42.6* .9 14.6* .6 .O .l

3.4 5.8* 1.5 .5

Have in my fraternity

.9 62.5* 17.8* .l .l 31.6* 3.3 18.1* 2.2 .2 .6

2.4 3.6 1.6 .l

room with ._____

C!OITIl

Share

.3 55.5* 9.0* 5.9* .O 38.4* 1.0 24.6* 7.6* .1 .O

1.1 2.6 .7 .l

Have as a close friend

3.5 32.1* 5.7* 11.6* .l 19.6* .2 5.2* 8.1* 3.3 .O

.5 10.4* .2 .3

Have date my sister

0 18 6 1 1

19 15 10 1 11

3

1 6 1 2

Number of measures significant

E E: g

% 2

s

$ mo

156

HENDRICK

AND

TABLE MEAN

RATINGS

RUMENIK

2

OF PERCEIVED

SIMILARITY”

Actors’ belief position and race Conservative Initial prejudice of subjects

Liberal

Tape

White

Black

White

Black

1 2

5.7 7.0 2.6 1.7

7.8 6.8 3 .9 2.2

4.9 4.2 6.4 7.8

2.6 2. 1 6.1 7.6

Prejudiced Unprejudiced

1

2

0 Each mean is based on 10 observations. The ratings were for the question “To what extent do you agree with the position of this person on the issue?”

belief manipulation. The means for this measure are shown in Table 2 and the summary of the analysis of variance is shown in the first column of Table 1. Perusal of Table 2 indicates that the manipulation of belief similarity was successful. Prejudiced Ss perceived greater similarity to the conservative actors (both White and Black) than to the liberal actors on both Tape 1 and Tape 2. Just the reverse was true for unprejudiced Ss. They perceived greater similarity to the liberal than to the conservative actors. Subjects’ prejudice X actors’ position (B X C) tests the significance of these differences. Table 1 indicates that the effect was significant (F(1,36) = 61.9, p < .Ol). Two other interaction effects (B X D, C X D) on the question assessingperceived similarity were significant. These effects are of some interest and will be discussed in later sections. Effects of Belief Similarity

(B X C)

Information about the effects of similarity of racial beliefs is given most directly by the interaction of Ss’ prejudice with actors’ belief positions (B X C) Inspection of the B X C row in Table 1 indicates that this effect was significant in 19 of the 21 analyses. The relevant means are shown in Table 3. The pattern of the means indicates that for every measure except one there was a rating bias in favor of belief similarity. Prejudiced Ss rated the conservative actors more favorably than the unprejudiced Ss. However, the unprejudiced Ss rated the liberal actors more favorably than the prejudiced Ss. The one exception is shown in the last row of the table. Unprejudiced Ss seemedmore willing than the prejudiced Ss to have anyone date their sister, either liberal or conservative. It is interesting to note that the differences between prejudiced and unprejudiced Ss were somewhat stronger for liberal than for conservative actors. As shown in Table 1, tapes interacted significantly with belief similarity

RACE

VERSUS

TABLE MEAN

RATINGS

OF THE ACTORS

Measure Rating questions 1 Perceived similarity 2 Liking 3 Dominance in discussion 4 Value of contribution 5 Clearness of position Trait ratings Opinionated Intelligent Open-minded Friendly Relaxed Sincere Informed Courteous Social distance items Work with Party with Invite to dinner Live next door Member of fraternity Share dorm room Close friend Date sister

3

BY PREJUDICED

Conservative Prejudiced subjects

157

BELIEF

AND UNPREJUDICED

actors

Unprejudiced subjects

Liberal

SUBJECTS

actors

Prejudiced subjects

Unprejudiced subjects

6.8 6.7 6.2

2.6 4.5 5.3

3.5 4.2 5.7

7.0 6.7 6.6

7.2 7.5

5.8 6.6

5.8 7.1

7.3 7.3

6.7 7.3 5.9 6.9 6.2 7.1 6.4 6.7

6.4 6.2 4.3 6.0 5.7 6.6 4.6 6.2

6.3 5.9 4.3 5.2 5.8 6.2 5.4 5.4

7.2 6.9 5.2 6.3 5.9 7.8 7.1 5.9

7.7 7.3 6.5 6.9 7.0 6.4 6.5 4.8

6.8 6.7 6.1 6.9 5.8 4.9 4.6 5.1

5.3 5.4 4.1 4.7 4.4 3.8 3.9 3.3

8.3 8.3 8.3 8.5 7.9 7.3 7.3 6.8

(A x B x C) on 15 of the measures. A study of these interactions revealed that they were due to an exaggerated 8s’ prejudice X actors? position (B x C) interaction in Tape 2 relative to Tape 1. The form of the B x C interaction tended to be the same for both tapes, however. The exaggerated B x C effect for Tape 2 follows directly from the earlier statement that the actors’ position statements were more extreme in Tape 2. Thus, the A X B X C interaction did not pose any particular interpretative problems. Main Effects of Subjects’ Prejudice (B) There were significant main effects of Ss’ prejudice on six measures, and all of them were social distance items. The direction of the differences was very consistent. Prejudiced Ss were more unwilling than unprejudiced

158

HENDRICK

AND

RUMENIK

Ss to work with the actors, party with, or dine with them, live next door to them, belong to the same fraternity, or have the actors date their sister. These results indicate a relative social rejection of others generally (regardless of race) by prejudiced individuals. These data confirm an earlier subsidiary finding by Hendrick, Bixenstine, and Hawkins (1971) that prejudiced Ss dislike others, regardless of race, more than unprejudiced Ss. It is an interesting anomaly that prejudiced attitudes covary with a desire to avoid social contact. The problem with such avoidance is t,hat it reduces the probability of receiving new information which might change the initially prejudiced attitude.

Main Effects of Actors’ Race (D) Table 1 indicates that the main effect of actors’ race was significant on 10 of the measures. The direction of the means for these effects was not entirely consistent. Blacks were perceived as dominating the discussion, but they were also perceived as making a more valuable contribution, and as stating their position more clearly. Blacks were rated as more intelligent, more relaxed, and more informed than Whites, but t,hey were also rated as less courteous. On the social distance items Blacks were preferred over Whites as a close friend, but they were less preferred than Whites for a next door neighbor or to date one’s sister. There were seven interaction effects involving race. Four of these effects (A X D, A X B X D, B X C X D, A X B X C X D) were not interpreted becausethe effects were significant on so few of the rating measures. Of the three remaining effects which were significant for several rating measures (B X D, C X D, A X C X D), the interaction of Ss’ prejudice X actors’ race (B X D) is most relevant to the issueof racial discrimination per se. This effect is considered below.

Race Bias Effects (B X D) Table 1 indicates that the B X D interaction was significant on 11 of the rating measures. The B X D means for each of these 11 measures are given in Table 4. The relevant measureswere perceived similarity, liking, friendly, and the eight social distance items. Table 4 reveals some interesting and unusual effects. These effects are most clearly observed on the social distance items. Consider first the S’s’ ratings of the White actors. On all eight social distance items prejudiced and unprejudiced Ss did not differ in their ratings of White actors. However, when the Black actors are considered, in every instance the Black actors were rated significantly less favorably by prejudiced than by unprejudiced Ss. That this difference is a race effect may he seen by comparing column differences. On all eight social distance items the prejudiced Ss rated the Black actors significantly

159

RACE VERSUS BELIEF TABLE

4

MEAN RATINGS OF WHITE AND BLACK ACTORS Perceived

similarity

Actors’ race

Prej . SS

White Black

5.5a 4.8* Work

Liking Unprej ss

. Actors’ race

Prej.

4.6* 5.ob

White Black

5.70 5.2*

with

Friendly

SS

Party

Unprej ss 5. o* 6.2a

. Actors’ race

Prej. ss

Unprej. SS

White Black

6.3” 5.90

5.90 6.40

with

Invite

to dinner

Actors’ race

Prej. ss

Unprej ss

. Actors’ race

Prej. ss

Unprej. ss

Actors’ race

Prej. SS

Unprej. ss

White Black

7.oa 6.w

7.00 8.1~

White Black

6.80 5.8*

6.80 8.2C

White Black

6.1a 4.56

6.5a 7.9

Live

next

door

Member

of fraternity

Share

dorm

room

Actors’ race

Prej. ss

Unprej. ss

Actors’ race

Prej. ss

Unprej. ss

Actors’ race

Prej. ss

Unprej. ss

White Black

6.8a 4.8b

7.2” 8 .2c

Whit,e Black

6.4” 4.9*

6.0a 7.@

White Black

5.9. 4.3”

5.4a 6.8c

Close friend

Date

sister

Actors’ race

Prej. ss

Unprej. ss

Actors’ race

Prej. ss

TJnprej. ss

White Black

5.6” 4.8*

5.00 6.g

White Black

5.4a 2.7*

5.80 6.1n

Not,e. Means in a column or in a row other at the .05 level by the Newman-Keuls

with

different test.

superscripts

differed

from

each

more unfavorably than the White actors. However, column comparisons for unprejudiced Ss show that on seven of the eight measuresBlack actors were rated significantly more favorably than White actors. The same effect is shown for liking, except that prejudiced Ss also rated White actors more favorably than unprejudiced Ss. The same pattern held for friendly, but the differences were not large enough for the Newman-Keuls comparisons to reach significance. These data, then, show two kinds of discriminatory race effects. Prejudiced Ss were biased against Blacks relative to Whites. However, unprejudiced Ss showed a reverse bias. They were biased against Whites relative to Blacks. Further, these differences cannot be attributed to differences in perceived similarity. As indicated in Table 4, prejudiced Ss

160

HENDRICK

AND

RUMENIK

perceived White actors as more similar than did the unprejudiced Ss. Also, White actors were perceived as more similar than Black actors by prejudiced 8s. However, the pattern of the interaction for perceived similarity was quite different than the pattern for the social distance items. The similarity means for White actors were significantly different, but this was not true for White actors on the social distance items, The means for Black actors were not significantly different for perceived similarity, but means for Black actors were different on the social distance items. The column means for unprejudiced 5’s were not significantly different on perceived similarity, but they were different on the social distance items. Thus, it is reasonable to conclude that the two types of race effects were genuine race bias effects, and were not due merely to differences in perceived similarity. Actors’

Belief Position

X Actors’ Race (C X D)

Table 1 indicates that this interaction effect was significant in 18 of the analyses of the rating measures.The relevant triple interaction (A X C X D) was significant in six of the analyses. The nature of the C X D interaction will be illustrated with the measure of perceived similarity. Liberal White actors were rated as more similar (5.8) than liberal Black actors (4.6)) but conservative White actors were rated as lesssimilar (4.3) than conservative Black actors (5.2). In general, the means for most of the other measuresfollowed the same pattern. Although the differences on the other measures may be attributed to differences in perceived similarity, the problem remains of explaining the differences in perceived similarity. Careful study of the tapes did not reveal an objective basis for this difference. Liberal White and Black actors mutually agreed with each other and used the same types of arguments. The same was true for the conservative White and Black actors. We believe that the explanation lies in a perceptual contrast effect created by 8s’ expectations about what Blacks and Whites should say, relative to what the actors actually said. For example, the Black conservative actor may have sounded like the voice of sweet reason to White Ss, counseling the virtues of traditional middle-class society. However, the same arguments from the mouth of a White conservative may have subtly connoted racism to Ss in a way that simply could not be true for the Black conservative actor. Subjects would, therefore, tend to perceive relative dissimilarity to the White conservative since they would not wish to be identified as racist. The liberal White, on the other hand, expressed liberal pro-Black views. Such expressions probably tend to be normative on most college campuses. The Black liberal, although expressing about

RACE

VERSUS

BELIEF

161

the same arguments, may have been perceived as “militant,” particularly when contrasted with the Black conservative on the panel. The net result was a kind of perceptual hypocrisy in which differing attributions of similarity (and liking, etc.) were made on the basis of skin color in conjunction with what Ss expected the belief positions ought to entail. White liberalism is the norm on campus, thus it is “good.” Conservative Whites are by definition prejudiced Whites. Conservative Blacks seem to accept White mores, and are therefore ‘Lgood,” while liberal Blacks are really militant, and by implication rejecting of Whites. This interpretation requires some caution. The 11 significant main effects for race (D) might mean that Black actors objectively differed in performance (e.g., on dominance), as opposed to merely being perceived as different by White subjects. This possibility is an alternative to the perceptual contrast argument suggested above. However, certain patterns of the data argue against the “objective difference” interpretation of the C X D interaction. First, it has been noted that liberal and conservative arguments were stronger in Tape 2 resulting in exaggerated effects for belief similarity (B X C) in the relevant triple interaction (A X B X C). If this exaggeration were in fact due to objective behavioral differences between Blacks and Whites (e.g., Blacks more dominant, less courteous, etc.), it should be reflected by an interaction of actor’s race with tapes (A X D) . This result did not occur, however, since the A X D interaction was significant for only one item. Thus, the differences in ratings resulting from differences in extremity of argument between the two tapes were apparently not due to ‘Lobjective” behavioral differences between White and Black actors across the two tapes. On the other hand, if the belief X race (C X D) effects were due to a perceptual contrast effect as noted above, this contrast should be enhanced by stronger differences in arguments between White and Black actors. The relevant interaction term (A X C X D) shown in Table 1 shows significance for six measures. This analysis indicates that the effect we have interpreted as contrast (C X D) was enhanced when stronger arguments were made in Tape 2. The net pattern of results suggeststhat something like perceptual contrast, rather than objective behavioral differences, was responsible for the C X D interactions. It should be noted that, whatever the final interpretation of the significant C X D interactions, the unusual race bias effects shown by the B X D interactions are unaffected. There is no way in which the B X D effects could be confounded by the C XD effects. This conclusion is particularly viable since the A X B X D and A X B X C X D interactions were nonsignificant.

162

HENDRICK

AND

RUMENIK

DISCUSSION

The results of the present experiment demonstrated once more the powerful effect of belief similarity-dissimilarity in eliciting differences in evaluations. Similar actors were preferred to dissimilar actors on 19 of the 21 rating measures. These belief prejudice effects provide further confirmation of Rokeach’s theory. Since a number of studies have demonstrated such effects, the present data require little comment in this regard. The most interesting results in the present experiment were the race bias effects revealed in the interactions of Ss’ prejudice X actors’ race (B x D). These effects confirmed the prediction based on Dienstbier’s (19721 theory that race prejudice effects would be obtained when racial attitudes were activated by a salient racial discussion. The results were as predicted for initially prejudiced Ss. However, the reverse prejudice effects in favor of Blacks by initially unprejudiced Ss were unanticipated. It is of some interest whether this anomalous effect was a “genuine” reverse prejudice effect. One possible criticism is that the two kinds of race bias effects might have reflected rating dispositions of the two types of Xs. High-prejudiced Ss may have been disposed toward a positive rating bias, and low-prejudiced Ss toward a negative bias independently of the presumed activating effects of the taped racial discussion. However, data presented by Hendrick, Bixenstine, and Hawkins i 1971, p. 256) rule out this interpretation. In that study high- and low-prejudiced Ss rated the same four actors used in the present study when the audio portion of the tape was turned off. Results showed that high-prejudiced Ss liked all actors less than lowprejudiced Ss, but a prejudice X race of actors interaction was not obtained. One may conclude that the taped discussion of race in the present study was necessary to activate the two different kinds of prejudiced reactions. Thus, it seems reasonable to conclude that the effects were genuine and did not reflect some type of rating bias. An attempt was made to explicate more fully the nature of the two prejudice effects by examining the responsesof the Ss to the initial questionnaire which was used to categorize Ss into high- and low-prejudiced groups. In addition to the 16 racially relevant it,ems there were 37 other items on the questionnaire assessingattitudes on 17 different issues.Prejudiced S’sdiffered significantly from unprejudiced Ss on 24 of the 37 items. The nature of these differences may be summarized as follows. Prejudiced Ss were more “warlike” than unprejudiced Ss. Prejudiced Ss were more favorable toward the Vietnam war, ROTC, the draft, defense spending, and serving in the army. With respect to the role of women, prejudiced Ss tended to feel less strongly that married women have a right to work, and they felt that woman’s proper place is in the home. Prejudiced Ss

RACE

VERSUS

BELIEF

163

were more willing to draft women (if they insist, on equality), and they felt that real fulfillment for a woman is achieved only through having a home and a family. Prejudiced Ss were more unfavorable toward the use of drugs, permissive abortion laws, hippies, freedom of speech, burning draft cards, guaranteed annual income, and the dispensation of birth control information on campus. Prejudiced Ss were more inclined to feel that the government should stay out of the area of population control. They were less inclined to rate higher education as irrelevant, to condemn technological progress, and to disown pride in being associated with the United States. These broad differences in attitudes seemed to reflect two very different underlying value structures. Prejudiced Ss were conservative, resisted change, and subscribed to traditional values. The unprejudiced Ss embraced change in such areas as new roles for women, individual freedom in the use of drugs, etc. Their ratings indicated a distinct antisystem, antigovernment bias, and by implication their ratings stressed the value of the individual and equality between individuals. If the prejudiced Ss oversubscribed to a traditional conservative ideology, it is fair to state that the unprejudiced Ss also oversubscribed to a newer liberal ideology. It may well be that the explanation of the reverse prejudice effect lies in the oversubscription. Unprejudiced Ss may “lean over backward” in an attempt to be absolutely equalitarian in their relationships with others, and particularly with minority group members. In so doing they may actually discriminate in favor of minority group members. Whatever the ultimate explanation for the reverse prejudice effect, on a strict operational basis it would appear that a substantial percentage of White, male college students now agree that “Black is beautiful.” Whether this percentage will increase during the next several years, and the implications of such a bias for relationships between the races are fascinating questions for future research. REFERENCES BYRNE, D. Attitudes and attraction. In L. Berkowitz (Ed.), Advances in experimental social psychology. Vol. 4. New York: Academic Press, 1969. CAMPBELL, D. T. Prospective: Artifact and control. In R. Rosenthal & R. L. Rosnow (Eds.), Artifact in behavioral research. New York: Academic Press, 1969. DIENSTBIER, R. A. A modified belief theory of prejudice emphasizing the mutual causality of racial prejudice and anticipated belief differences. Psychological Review, 1972, 79, 146160. HENDRICK, C., BIXENSTINE, V. E., & HAWKINS, G. Race versus belief similarity as determinants of attraction: A search for a fair test. Journal of Personality and Social Psychology, 1971, 17, 250-258. HENDRICK, C., STIKES, C. S., & MURRAY, E. J. Race versus belief similarity as de-

HENDRICK

164

terminants

RUMENIK

in a live interaction setting. Journal of Experimental 1972, 6, 162168. HENDRICK, C., STIKES, C. S., MURRAY, E. J., & PUTHOFF, C. Race versus belief as determinants of attraction in a group interaction context. Memory and Cognition, 1973, 1, 4146. ROKEACH, M. Belief versus race as determinants of social distance: Comments on Triandis’ paper. Journal of Abnormal and Social Psychology, 1961, 62, 187-188. ROKEACH, M., SMITH, P., & EVANS, R. Two kinds of prejudice or one? In M. Rokeach (Ed.), The open and closed mind. New York: Basic Books, 1960. SEARS, D. O., & ARELES, R. P. Attitudes and opinions. Annual Review of Psychology, 1969, 20, 253-288. TRIANDIS, H. C. A note on Rokeach’s theory of prejudice. Journal of Abnormal and Social Psychology, 1961,62, 184-186. TRIANDIS, H. C., & DAVIS, E. E. Race and belief as determinants of behavioral intentions. Journal of Personality and Social Psychology, 1965, 2, 715-725. Research

of attraction

AND

in Personality,