Journal of Personality and Social Psychology 197S, Vol. 32, No. S, 783-789

Dissimilarity and Attraction: When Difference Makes a Difference Joseph E. Crush Northern Illinois University Gerald L. Clore and Frank Costin University of Illinois at Urbana-Champaign Students in 93 university classes described themselves and their instructors on personality tests measuring traits relevant (ascendancy and personal relations) and irrelevant (sociability and cautiousness) to teaching skill. Dissimilarity (instructor being higher than student) on relevant trait dimensions was hypothesized to be instrumental for student satisfaction with an instructor's role performance. When students were classified as similar or dissimilar to their instructors on the personality dimensions, those most attracted to their instructors were those dissimilar on relevant (but not irrelevant) traits. The dissimilarity finding was not an artifact of differences in skill of instructor nor of differences in students' perceptions of their instructors' traits. A threedimensional system was devised to specify when similarity and dissimilarity should promote attraction.

Investigators using Byrne's (1971) attraction paradigm have generally found a strong relationship between attitude similarity and interpersonal attraction. The relationship has been obtained for different modes of attitude presentation (Byrne & Clore, 1966), different populations (Byrne, Griffitt, Hudgins, & Reeves, 1969), and different issues of varying levels of importance (Clore & Baldridge, 1968). Attraction has also been related to similarity in abilities (Zander & Havelin, 1960), emotional states (Zimbardo & Formica, 1963), economic status (Byrne, Clore, & Worchel, 1966), and personality traits (Byrne, Griffitt, & Stefaniak, 1967). A reinforcement-affect model of attraction (Clore & Byrne, 1974) was formulated as an interpretation of previous research and as a heuristic conception for new investigations. The focus of the model is interpersonal reinforcement generally, rather than similarity per se, but similarity has frequently served as the reinforcement in attraction studies. This research was supported by a research grant from the University of Illinois Research Board to Frank Costin, and preparation of the paper was supported by Research Grant MH 14510 from the National Institute of Mental Health, U. S. Public Health Service to Gerald L. Clore. Requests for reprints should be sent to Joseph E. Crush, Department of Psychology, Northern Illinois University, DeKalb. Illinois 60115.

Similarity research is linked to the concept of reinforcement by two correspondence rules. The "reinforcement rule" states that evaluation of a person is influenced by the reinforcements and punishments associated with that person. The "similarity rule" postulates that individuals generally find another person's views that are similar to their own to be reinforcing and find those dissimilar from their own to be punishing. Recent investigations report data that seem to conflict with the hypothesis that similarity is reinforcing and dissimilarity is punishing. For example, several studies have shown that similarity sometimes intensifies negative reactions (either avoidance or dislike) to persons portrayed as former mental patients (Novak & Lerner, 1968), obnoxious individuals (Taylor & Mettee, 1971), or drug addicts (Lerner & Agar, 1972). These studies suggest that under certain conditions dissimilar others may be more attractive than similar others. The present study was designed to answer two questions: (a) Under what conditions will dissimilarity lead to attraction? and (b) How are these conditions different from those in which similarity promotes attraction? The college classroom, where students inevitably judge their teachers, provided a naturalistic setting 'for the examination of

783

784

j. E. GKUSH, G. L. CLOKE, AND F. COSTIN

these questions. The hypothesis was that positive dissimilarity between teacher and student on personality traits that are relevant for effective teaching would produce higher ratings of teacher attraction than would similarity. By contrast, positive dissimilarity on traits irrelevant for the teaching role would have no such effect on teacher attraction.1 In the hypothesis, "positive dissimilarity" means that the teacher's personality score exceeds the students' scores. "Relevance" refers to the degree of relationship between a particular teacher trait and skill as a classroom instructor. The primary hypothesis was derived from the assumption that role and motivational factors can alter the usual link between similarity and attraction (Clore & Byrne, in press). Hunt (1965) has indicated that roles carry with them expectations for personality characteristics as well as behaviors. Since teachers have the role of directing the learning process, students may expect them to display levels of relevant traits that are higher than the students' own. This notion was tested by having students fill out two personality inventories, one for themselves and one for their ideal teacher. It was predicted that students would show a clear preference for their ideal teachers' personality scores to exceed their own. The investigation was concerned with one additional problem in attraction research. In a recent review, Fishbein and Ajzen (1972) pointed out that manipulations of similarity and dissimilarity are often confounded with evaluation. That is, subjects may show more attraction to similar persons because they are perceived as having more socially desirable characteristics than dissimilar persons. In view of this problem, the present experiment was designed to separate social desirability from similarity and positive dissimilarity. 1 Hypotheses concerning cither negative dissimilarity or personal attraction could not be tested unambiguously. A test of the effects of negative dissimilarity required students who scored a standard deviation above their teachers on the trail dimensions. Personal attraction was so highly related to teacher attraction (.64 £or Sample, 1, and .82 for Sample 2) that analyses of it were, not independent or informative.

METHOD Subjects The teachers that served as the targets for judgment were graduate students at the University of Illinois at Urbana-Champaign. They taught courses in anthropology, economics, education, history, political science, sociology, and psychology. Senior faculty members were responsible for lectures that were presented live or on television; the teaching assistants were generally responsible for conducting discussion sections. The subjects were all of the students in these courses, and hence represented a broad spectrum of the campus population. Fifty-seven teachers (42 men and IS women) participated in the study on a voluntary basis. Wherever possible, two discussion sections for each teacher were included so that two distinct sets of personality traits could be investigated. If a teacher taught only one section, that section was used; if a teacher taught three or more sections, two sections were selected at random. Since most of the teaching assistants taught two sections, two separate samples were available. The total number of students providing complete data was 658 in Sample 1 and 720 in Sample 2. The number of classes was 46 in Sample 1 and 47 in Sample 2.

Instruments and Data Collection Personality traits. At the beginning of the semester students completed either the Gordon Personal Profile (Gordon, 1963b; Buros, 1970, pp. 1034-1037) or the Gordon Personal Inventory (Gordon, 1963a; Buros, 19VO, pp. 1032-1034). Both of these tests employ behaviorally stated items in a forced-choice format. All items within each test were initially selected by Gordon to be equivalent in social desirability. The students described their own personality traits and also used the same items to indicate the traits they would prefer in their teachers. The Gordon Personal Profile was used for Sample 1 and the Gordon Personal Inventory for Sample 2. The Gordon Personal Profile measures four dimensions of personality: ascendancy (selfconfidence), responsibility (performs tasks on time), emotional stability (free from anxiety), and sociability (grcgariousncss). The Gordon Personal Inventory measures a different set of four traits: cautiousness (acts deliberately), original thinking (enjoys thought-provoking discussion), personal relations (tolerant and understanding), and vigor (energetic). Trait scores on the Gordon Personal Profile can range from 0 to 36; scores on the Gordon Personal Inventory from 0 to 40. Teacher skill and teacher personality. At the end of the semester, students filled out a 25-item questionnaire about the frequency with which various kinds of classroom behavior had occurred during the semester. These responses were scored on a 5point scale from "almost always occurred" (5) to "almost never occurred" (1). Five items represented a skill factor (Isaacson, McKeachic, Milholland, Lin, Hofeller, Baerwaldt, & Zinn, 1964),

785

DISSIMILARITY AND ATTRACTION and were used as the measure of teacher skill in this study. After completing their ratings of classroom behavior, students described their teachers' personality traits as they had observed them during the course of the semester. The same test, cither the profile or the inventory, which was used at the beginning of the semester was used for this assessment. Attraction. As an index of teacher attraction the following three items were presented, and students indicated their degree of attraction on 7-point scales: (a) How much did you like your instructor as a teacher? (b) How much would you like to be in another course taught by your instructor? and (c) Mow much would you like to recommend this instructor to a friend who had to take the same course? The total score of these three items, ranging from "extremely dislike" (3) to "extremely like" (21), was taken as an index of teacher attraction.

Criteria for Similarity and Dissimilarity To obtain a reliable estimate of the teacher's classroom personality, independent of any one student's point of view, a composite score was computed on each trait by averaging the scores assigned to the teacher by his or her students. Similarity was defined as a difference of not more than two points separating the teacher's composite personality score and a particular student's score. Positive dissimilarity was defined as the teacher's composite score exceeding a particular student's score by at least five points. According to the normative data published by Gordon (1963a, 1963b), five points represent approximately one standard deviation for college populations. To avoid an arbitrary classification of students, additional restrictions were placed on the selection process. A student was not classified as similar unless his own score was within two points of both the score he personally assigned to the teacher and the teacher's composite score. Comparable criteria were used to assign subjects to the dissimilar group. Both the score assigned by the students to the teacher and the teacher's composite score had to be at least five points higher than the student's own score. These restrictions ensured compatibility between the hypothesized social comparison process (which must involve individual perception) and the empirical test of the process (which was based on composite perceptions of the teacher).

RESULTS Reliability and Validity of Teacher Attraction To determine the reliability of the attraction measure, the ratings of 200 students were selected at random from the two samples. The mean correlation between each item and the sum of the other two items was .84. Crush and Costin (in press) have conceptualized the measure of teacher attraction

as a "recommendation" of the teacher to other students. As a check on the validity of that conception, correlations were computed between teacher skill and attraction. The correlation was .76 for Sample 1 (n = 6S8), and .73 for Sample 2 (n - 720). Since one would expect a recommendation of the teacher to reflect an evaluation of the teacher's skill, these correlations provided evidence for the validity of the attraction measure. Relevance of Personality Traits Table 1 provides information about the personality traits which are most and least relevant for effective teaching. Traits with the largest significant beta weights were selected as relevant traits; traits with the smallest and nonsignificant beta weights were selected as irrelevant traits. For Sample 1, ascendancy was selected as a relevant 'trait and sociability as an irrelevant trait. For Sample 2, personal relations was selected as a relevant trait and cautiousness as an irrelevant one. These traits were selected on the basis of multiple regression analyses which treated the means of intact classes as individual subjects. However, very similar results were obtained when the scores of individual subjects were analyzed. In addition, Costin and Crush (1973) showed that these traits, ascendancy, sociability, personal relations, TABLE 1 CORRELATIONS AND BETA WRIGHTS FOR PREDICTING TEACHER SKILL PROM TEACHER PERSONALITY TRAITK Teacher skill Teacher personality trail r

B

Sample 1 Ascendancy Responsibility Emotional stability Sociability

.82* .39* .61* .53*

.81 .24 -.02 -.01

4.63*" 1.76 -.25 -.13

Sample 2 Cautiousness Original thinking Personal relations Vigor

.16 .58* .51* .56*

-.13 .22 .46 .38

-1.00 1.64 3.54** 3.00*"

Note. Whole classes were treated as individual subjects: i = 46 for Sample 1, n = 4-7 for Sample 2. * / » < .01. **/> < .001.

J. E. GRUSH, G. L. CLOEE, AND F. COSTIN

786

and cautiousness, had the following respective correlations with teacher skill: .67, .22, .60, and .08. Thus, there is ample evidence that Sample 1 and Sample 2 each provided a relevant and irrelevant personality 'trait for a test of the major hypothesis.

TABLK3 TEACHER ATTRACTION MEANS Teacher iicrHonalily trait Level of student similarity Low

Medium

nigh

Sample \- teacher ascendancy

Primary Hypothesis The hypothesis stated that a condition of positive dissimilarity between teacher and student on relevant traits would lead to more teacher attraction than would a condition of similarity and that a condition of positive dissimilarity on irrelevant traits would have no effect on attraction. These predictions were tested in a series of 3 X 2 analyses of variance. To control for social desirability, teachers were trichotomized into low, medium, and high levels of trait possession on the basis of their composite personality scores. This categorization created a factor referred to as teacher personality. The second factor was TABLE 2 ANALYSES OK VARIANCE TOR THREE LEVELS oi' TEACHF.R PERSONALITY AND Two LF.VELS or SIMILARITY ON TEACHER ATTRACTION Source

Sample 1 teacher ascendancy Teacher personality (A) Similarity (B) A X B

126.45

133.33 11.19

7.24 7.64 .64

.001 .007 .53

Sample 1 — teacher sociability Teacher personality (A) Similarity (B) AXB

225.45 4.48 5.16

8.25 .16 .19

.001 .69 .83

Sample 2—teacher personal relations Teacher personality (A) Similarity (B) AXB

331.75 163.79 5.45

18.48 9.13 .30

.001 .003 .74

Sample 2—teacher cautiousness

Teacher personality (A) Similarity (J!) A X U

83.56 .15 .18

Note. All ciror terms have df = 102.

3.70 .01 .01

.03 .9-4 .99

15.50 16.56

Similarity Dissimilarity

13.72 17.00

Sample 1 -teacher sociability Similarity Dissimilarity

12.94 11.6V

14.56 14.67

17.33 17.28

Sample 2- -teacher personal relations Similarity Dissimilarity Sample 2 Similarity Dissimilarity

11.33 14.56

11.78 13.44

16.78 19.28

teacher cautiousness

13.89 14.11

16.89 16.94

14.72 14.66

similarity, and it had two levels: similarity and positive dissimilarity. The selection of a student's rating to represent either similarity or positive dissimilarity followed pre-established criteria. The possible number of subjects were further restricted by two stipulations: (a) From a given teacher's classroom, an equal number of students were selected for the similarity and positive dissimilarity conditions and (b) the number of observations for each cell of the 3 X 2 design were to be equal. These restrictions produced an n of 18 for each cell. The results of 3 X 2 analyses of variance, with attraction as the dependent measure, are reported in Table 2. The factor of teacher personality showed a significant main effect on attraction for all four of the personality variables investigated. Generally, the higher a teacher's composite personality score on any trait dimension, the higher the teacher was rated on attraction. By contrast, significant main effects for the factor of similarity were observed only for ascendancy and personal relations, the traits that were relevant for effective teaching. The teacher attraction means in Table 3 show

DISSIMILARITY AND ATTRACTION TABLE 4 Cowi'ARrsoN OF STUDENTS' OWN AND PREFERRED TEACHER PERSONALITY TRAITS Personality trail

Students' own

Teacher preferred

Difference

7"

Sample 1 Ascendancy Responsibility Emotional stability Sociability

20.69 23.83 23.24 20.39

25.60 28.00 28.01 21.84

4.91 4.17 4.77 1.45

15.29* 13.77* 14.98* 4.40*

Sample 2 Cautiousness Original thinking Personal relations Vigor

23.86 24.34 24.03 23.95

25.41 28.25 31.69 24.95

1.55 3.91 7.66 1 .00

8.13* 17.20* 25.75* 4.02*

Note. Whole classe.s were treated as individual subjects: n = -16 for Sample 1, n ~ 47 for Sample 2. *p = .002.

that students in the positive dissimilarity conditions rated the teachers higher on teacher attraction than did students in the similarity conditions. The irrelevant traits of sociability and cautiousness displayed no significant main effects for the factor of similarity. These results represented strong support for all aspects of the principal hypothesis. Secondary Hypothesis The secondary hypothesis predicted that students would prefer their teachers' scores on the personality dimensions to exceed their own. This hypothesis was tested by comparing the students' own personality scores with those resulting from students' descriptions of their preferred teacher. The results of this analysis are presented in Table 4. As can be seen there, students consistently and significantly preferred that their teachers be higher on all the trait dimensions than were the students themselves. These results provided evidence consistent with the rationale for the major hypothesis. Alternative Explanation A more parsimonious explanation than one based on student expectations was also considered. Since composite teacher scores were involved in testing the major hypothesis, it is possible that the dissimilarity effect was due to perceptual differences. That is, if dissimilar students attributed more of the relevant traits to their teachers than did similar students, this alone could account for the dissimilarity effects. To exclude this possible interpreta-

tion, a second series of 3 X 2 analyses of variance was conducted. The factors of teacher personality and similarity were the same as in previous analyses; the dependent measures were the actual teacher personality scores assigned by the respective students. The results showed no main effects for similarity, no interactions, and nearly identical means for the similar and dissimilar students. An explanation of the data based on perceptual differences between similar and dissimilar students was unwarranted. DISCUSSION The results showed that students dissimilar to their teachers on role-relevant traits (ascendancy and personal relations) liked their teachers more than did students who were similar. In addition, students similar and dissimilar to their teachers on irrelevant traits (sociability and cautiousness) showed no differences in teacher attraction. Predictions were based on the hypothesis that students would engage in social comparison processes when making their attraction judgments. Presumably, this comparison proceeded somewhat as follows: "If teachers are to direct me in the learning process, they ought to possess more of those traits important for effective teaching than I do; if they only possess the same amount as 1 do, how can I learn anything from them?" In this regard, students showed a distinct preference at the beginning of the semester for teachers to display higher levels of the traits than themselves. This difference was most pro-

788

J. E. GKUSH, G. L. CLORE, AND F. COSTIN

nounced for the role-relevant traits (ascendancy, 4.91; personal relations, 7.66) and less prominent for the irrelevant traits (sociability, 1.4S; cautiousness, l.S.S). Therefore, positive dissimilarity seemed to mediate attraction for relevant traits because dissimilar students had their role expectations fulfilled, while similar students did not. For irrelevant traits, expectations that teachers be higher were weaker, and hence dissimilarity on these traits did not lead to attraction. The interpretation that the differential effects of dissimilarity were due to relevance (cf. Levinger & Breedlove, 1966) is bolstered by the fact that the same pattern of results emerged for two different samples of students judging two different sets of personality traits. The present study 'found a true dissimilarity effect rather than a simple tendency for teachers higher on socially desirable traits to be liked (Fishbein & Ajzen, 1972). Once teachers were divided into low, medium, and high groups on the basis of their composite personality scores, an equal number of students from a teacher's class was assigned to the similarity and positive dissimilarity conditions. Thus, regardless of teacher personality score (low, medium, or high), the same teachers who were especially liked by dissimilar students were liked less well by similar students. The fact that dissimilarity functioned differently for relevant and irrelevant traits also cannot be easily attributed to social desirability. All traits were initially selected because they were equivalent in social desirability (Gordon, 1963a, 1963b). Also, the main effects for teacher personality on all traits suggest that the traits were equally socially desirable in the present context as well. Implications The results of this study seem to conflict with research in which similarity is found to be rewarding and dissimilarity to be punishing. To resolve this apparent conflict it is necessary to examine the specific conditions under which similarity and dissimilarity effects are obtained. Similarity studies employing the Byrne (1971) paradigm can be described as a social comparison process where: (a) The judge

and person to be judged are not related through some role (i.e., they are strangers or anonymous), and the status relationship between the two is of a peer-peer type; (b) the presumed purpose of social comparison is for the judge to determine the appropriateness of his own views or personal characteristics; and (c) the mechanism operative in this deliberation is consensual validation. Under these conditions, a judge will like an anonymous peer who shares his views because similarity provides a testimonial to the appropriateness of the judge's own position. The present research differed from the usual similarity studies in all three conditions. The teacher and student shared a role relationship in which the teacher was assigned higher status; the purpose of social comparison was to evaluate the teacher's role performance; and the mechanism was an expectation that teachers display higher levels of role-relevant traits than students. Under these conditions, positive dissimilarity became the criterion for teacher attraction. Other dissimilarity studies have also changed conditions. Some have altered the purpose of social comparison (e.g., Novak & Lerner, 1968; Lerner & Agar, 1972), others have changed the mechanism or rule (e.g., Taylor & Mettee, 1971), while still others have altered both purpose and rule (e.g., Hendrick & Brown, 1971). Two conclusions emerge from this analysis. First, there is no necessary contradiction between similarity and dissimilarity studies. Since dissimilarity effects are found under different conditions than similarity effects, the two sets of findings are not directly comparable. Second, the scope of the "similarity rule" can be clarified. The rule that similarity will be reinforcing and dissimilarity punishing applies only to studies that conform to the boundary conditions summarized earlier for Byrne's (1971) paradigm. The "similarity rule" will be applicable to other investigations to the extent that they are conducted under those kinds of conditions. Conclusion It would be desirable to extend the reinforcement-affect model of attraction (Clore & Byrne, 1974) to include similarity and dis-

DISSIMILARITY AND ATTRACTION

similarity results under one theoretical umbrella. The factor that unites the apparently discrepant similarity and dissimilarity findings is the underlying role of reinforcement. While similarity has frequently been found to be reinforcing (the "similarity rule"), the core of the model is not similarity but affective reinforcement whatever its source (the "reinforcement rule"). Hence, the model could be extended by replacing the "similarity rule" with a detailed specification of the conditions which lead similarity and dissimilarity to be mediational for attraction. Such a specification could take into account variations along the following dimensions: (a) the role and status of judge and target, which may vary from nonfunctional roles of peer status to functional roles of disparate status; (b) the purpose of social comparison, which might vary from a judgment of the goodness of one's own attributes to satisfaction with another's role performance; and (c) the mechanism used to validate the judgment, which might vary from consensual validation to an appeal to some social norm or other standard. At one end of these dimensions are situations in which similarity should be reinforcing; at the other end are situations in which dissimilarity could be reinforcing. REFERENCES Kuros, 0. K. Personality tests and reviews. Highland Park, N.J.: Gryphon Press, 1970. Byrne, D. The attraction paradigm. New YorkAcademic Press, 1971. Byrne, D., & Clore, G. L., Jr. Predicting interpersonal attraction towards strangers presented in three different stimulus modes. Psychonomic Science, 1966, 4, 239-240. Byrne, D., Clore, G. L., Jr., & Worchcl, P. Effect of economic similarity-dissimilarity on interpersonal attraction. Journal of Personality and Social Psychology, 1966, 4, 220-224. Byrne, D., Griflitt, W., Hudgins, W., & Reeves, K. Attitude similarity-dissimilarity and attraction: Generality beyond the college sophomore. Journal of Social Psychology, 1969, 79, 155-161. Byrne, D., Griffitt, W., & Stefaniak, D. Attraction and similarity of personality characteristics. Journal of Personality and Social Psychology, 1967, 5, 82-90. Clore, G. L., & Baldridgc, B. Interpersonal attraction: The role of agreement and topic interest.

789

Journal of Personality and Social Psychology, 1968, 9, 340-346. Clore, G. L., & Byrne, D. A reinforcement-affect model of attraction. In T. L. Huston (Ed.), Perspectives on interpersonal attraction. New York: Academic Press, 1974. Clore, G. L., & Byrne, D. The process of personality interaction. In R. B. Cattell and R. M. Dreger (Eds.), Handbook of modern personality theory. New York: Naiburg, in press. Costin, F., & Crush, J. E. Personality correlates of teacher-student behavior in the college classroom. Journal of Educational Psychology, 1973, 65, 3544. Fishbcin, M., & Ajzcn, I. Attitudes and opinions. Annual Review of Psychology, 1972, 23, 487-544. Gordon, L. V. Manual, Gordon Personal Inventory (Rev. ed.). New York: Harcourt, Brace & World, 1963. (a) Gordon, L. V. Manual, Gordon Personal Profile (Rev. ed.). New York: Harcourt, Brace & World, 1963. (b) Crush, J. E., & Coslin, F. The student as consumer of the teaching process. American Educational Research Journal, in press. Hendrick, C., & Brown, S. R. Introversion, extraversion, and interpersonal attraction. Journal of Personality and Social Psychology, 1971, 20, 31-36. Hunt, R. G. Role and role conflict. In H. J. Hartley and G. E. Holloway (Eds.), Focus on change and the school administrator. Buffalo: State University of XTcw York, School of Education, 1965. Isaacson, R. L., McKeachie, W. J., Milholland, J. E., Lin, Y. G., Hotelier, M., Baerwaldt, J. W., & Zinn, K. L. Dimensions of student evaluation of teaching. Journal of Educational Psychology, 1964, 55, 344-351. Lerncr, M. J., & Agar, E. The consequences of perceived similarity: Attraction and rejection, approach and avoidance. Journal of Experimental Research in Personality, 1972, 6, 69-75. Levinger, C., & Breedlove, J. Interpersonal attraction and agreement: A study of marriage partners. Journal of Personality and Social Psychology, 1966, 3, 367-372. Novak, D. W., & Lerner, M. J. Rejection as a consequence of perceived similarity. Journal of Personality and Social Psychology, 1968, 9, 147-152. Taylor, S. E., & Mettee, IX R. When similarity breeds contempt. Journal of Personality and Social Psychology, 1971, 20, 75-81. Zander, A., & Havelin, A. Social comparison and interpersonal attraction. Human Relations, 1960, 13, 21-32. Zimbardo, P., & Formica, R. Emotional comparison and self-esteem as determinants of affiliation. Journal of Personality, 1963, 31, 141-162. (Received August 12, 1974)

Dissimilarity and attraction: when difference makes a difference.

Students in 93 university classes described themselves and their instructors on personality tests measuring traits relevant (ascendancy and personal r...
590KB Sizes 0 Downloads 0 Views