The mechanisms underlying attributive projection' Gary Anton Chalus, Northwestern University

ABSTRACT

The experiment investigated the validity of the cognitive dissonance reduction and stimulus generalization explanations of attributive projection. Subjects in a high dissonance condition were informed that they possessed considerable hostility (a negative trait), while those in a low dissonance condition were told that they possessed a great deal of asseitiveness (a desirable trait). The dissonance mechanism received support in that (a) subjects' reports of their anxiety indicated that an ego threat was produced only in the high dissonance condition, as expected; and (b) hostility was projected onto others to a significantly greater degree than was assertiveness. The existence of only a weak relationship between degree of ego threat produced by the personality feedback and subsequent amount of projection, however, throws into question the validity of the hypothesized defensive function of projection. No support for the generalization mechanism was obtained since the subjects did not project a greater amount of either trait onto the similar targets (college students) than onto the relatively dissimilar targets (middleaged lawyers); possible reasons for this negative result were discussed, as well as suggestions for future research. Since Freud's introduction of the concept over 80 years ago (Freud, 1896/1962), the defense mechanism of projection has been widely regarded as one of the most fundamental of all human adaptive mechanisms. In recent decades, projection has received considerable empirical inquiry, but due to the fact that the 1. This research was supported in part by National Institutes of Health Grant 3 TOl HS 00049. This report is based on a dissertation submitted to the Graduate School of Northwestern University in partial fulfillment of the requirements for the PhD degree; the author would like to thank G. Peter Herman for his encouragement and valuable suggestions throughout all phases of this research. Requests for reprints should be sent to Gary Ghalus, Department of Gonsumer Information Services, Young & Rubicam Inc., 285 Madison Avenue, New York, New York 10017.

Mechanisms of projection

363

term was tradidonally defined in various nonsystematic ways by different writers, more confusion than clarity was generated. This situadon was aUeviated as a result of an influential review paper by Holmes (1968), which provided a theoredcal framework into which four logically possible varieties of projecdon were cast. One of the t5^es is called "attributive projecdon," which occurs when an individual attributes or projects onto others the same personality trait or momentary feehng which he consciously possesses. Attribudve projection appears to be the most frequently occurring variety of projecdon, and as the reviews by Hohnes (1968) and Chalus (1976b) indicate, it is the type for which the greatest amount of research evidence exists. However, the mechanisms which underlie attribudve projection have jof vfir heeja xJearly elaborated. Two mechanisms, more frequently than any others, have been proposed as responsible for the operation of attribudve projection: cognidve dissonance reduction and stimulus generalization. However, the relative contributing influence of these two mechanisms has never been studied within a single experimental design. The dissonance reduction hypothesis was first proposed and investigated by Bramel (1962, 1963), and is derived from Festinger's (1957) theory of cognidve dissonance. Dissonance theory as applied to attribudve projecdon states that if a person perceives that he possesses an undesirable trait a state of dissonance will be aroused in most individuals since the possession of a negative trait is inconsistent with a positive selfimage; the individual may then reduce this dissonance by projecting the undesirable trait primarily onto similar others. By doing so, the projector can convince himself that if similar and favorably evaluated others possess it as well, the trait might not be as undesirable as he had previously thought and/or he might not in fact possess the trait to a markedly greater degree than the targets of his projections. It should be clear that the dissonance reduction mechanism assumes a defensive or anxiety-reducing fimction of projection, and that it therefore applies only to negative traits. Much prior research has been conducted in an effort to validate the dissonance mechanism, but a thorough review and analysis of the literatvire by Chalus (1976b) has shown that various

364

Chalus

methodological and interpretive difficuldes have rendered all of these studies inconclusive, most notably those by Bramel (1962, 1963), Secord, Backman, and Eachus (1964), Edlow and Kiesler (1966), Epstein and Baron (1969), Gk)ldmann (1971), and Kreines and Bogart (1974). One of the most common problems is that the apparently supportive results of many of these studies are reinterpretable in terms of stimulus generalizadon effects. For example, Goldmann (1971) found that subjects who were led to believe that they possessed much hosdlity projected more of that trait than did those who presumably possessed little hostility. Although this result is consistent with what dissonance theory predicts, it could also have been due to the operadon of nondefensive stimulus generaUzation since a person who possesses a great deal of a trait has more of it to generalize onto others than one who does not. Thus, the degree of dissonance and degree of trait possession variables were confounded. The stimulus generalizadon explanation of attribudve projection states that this form of projection arises from our natural tendency to assume that others are essentially similar to ourselves by virtue of the fact that we all share a common humanity (e.g., Cameron & Magaret, 1951; CatteU, 1951; Horney, 1939). This explanation postulates that the greater the perceived similarity between the projector and the target on such variables as age, sex, social class, appearance, and occupation, the greater will be the generalization or projection of one's own traits and feelings onto that target. The "mechanistic" generaUzation mechanism thus applies to the projection of both positive and negadve characteristics, and it does not assume that projection serves an ego-defensive function. Finally, from the above it is apparent that both the generalization and dissonance mechanisms predict greater projection onto similar targets than onto dissimilar targets. The present study, therefore, attempted to assess the validity of these two mechanisms. The experiment was basically a 2 X 2 X 2 factorial design, the variables being degree of dissonance arousal, degree of similarity of the target figures, and sex. In the first part of the experiment, the subjects took a personality test and were led to believe via false feedback that they possessed either high levels of hostility (an undesirable, socially reprehensible, and thus dissonance-arousing trait) or high levels

Mechanisms of projection

365

of assertiveness (a positive and thus nonthreatening trait); this consdtuted the dissonance manipulation. (It will be noted that this procedure manipulated level of dissonance without confounding it with the amount of the relevant trait the subjects thought characteristic of themselves.) In the second part of the experiment, the subjects rated others for the degree to which they possessed either hostUity or assertiveness, basing their evaluadons on the other people's projective test responses. The description of the targets as either very similar to the subjects (other coUege students) or as reladvely dissimilar (middle-aged male lawyers) constituted the similarity manipuladon. From the degree to which the four experimental groups—generated by crossing the Dissonance and Similarity variables—exhibited projection, inferences about the validity of the dissonance and generalizadon explanations were made. The principal dissonance prediction was that there would be greater projection of hostility than of assertiveness onto both the student and lawyer targets, while the generalizadon explanation predicted greater projection of both traits onto the similar versus dissimilar targets. In addition to the above four groups, two control condidons were included. In one of these, subjects received low hostility feedback to determine whether the attributive projection of hostility was successfully elicited in the high hostility feedback condition. The other control group was included to test the dissonance theory assumption that the projection of negative traits serves an ego-defensive funcdon; accordingly, subjects in this group were not permitted to overdy project hostility, and the dissonance prediction was that the amount of anxiety reduction exhibited by this group (from immediately after the negadve feedback to immediately after the performance of a nonprojective task) would be significandy less than that shown by the high dissonance groups allowed to project hosdlity. To monitor the subjects' levels of anxiety throughout the study, repeated self-report state anxiety measures were administered. The specific hypotheses were that (a) the attributive projecdon of hostility would be elicited in the high hostility feedback condition, (b) evidence would be obtained in support of the vaUdity of both the dissonance reduction and generalizadon mechanisms, and (c) the projection of hostility would be found to serve a defensive function.

366

Chalus METHOD

Pretesting Pretesting data were obtained to determine the label to be applied to the dissimilar target category. Many prior studies on projection confounded the variables of similarity of the target figure and his overall favorability, thus potentially resulting in "halo" effects. To select an appropriate dissimilar target and therefore avoid the potential effect of this rating bias on the experimental subjects' projective responses, 24 Northwestern University undergraduates (17 males and 7 females) were asked to rate the typical or average middle-aged male member of nine occupational types (as well as what they saw as the typical male coUege student) for their degree of possession of six personality traits on an 11-point scale. On the basis of comparisons made between the ratings of the student versus those of the other nine people for the trait of "likeability" (an estimate of overaU favorability), it was concluded that the dissimilar targets would best be described as lawyers, who, unlike some of the other occupational types, were rated very similar to the average student in terms of likeability. (Additional data, obtained from 29 male and 44 female Northwestern students, revealed that, considering the traits exacdy as they were described to the subjects in the projection experiment, both the average male student and the average male lawyer were seen as significandy more assertive than hostile. Since the primary dissonance prediction is that there would be more projection of hostility than of assertiveness onto both targets, the present experiment thus appears to be a conservative test of the dissonance hypothesis.) Subjects The subjects were 60 male and 60 female Northwestern University students who participated to partially fulfUl their introductory psychology course experimental requirements. They were assigned randomly to the six experimental groups with the restriction that there be an equal number of males and females in each group. Block randomization was utilized in running the subjects, so that for each consecutive block of six subjects, each of the six groups was represented once. The subjects were run individually. Procedure

The experimental procedures for the six groups are outlined in Table 1. (The procedures followed in the four experimental groups generated by crossing the Dissonance and Similarity variables wiU

Mechanisms of projection Table I .

367

Summary of experimental design. Procedure

Group HiH-HiS HiH-LoS HiH-NPr LoH-HiS HiA-HiS HiA-LoS

Int. Int. Int. Int. Int. Int.

& & & & & &

AACL AACL AACL AACL AACL AACL

1 1 1 1 1 1

HiH HiH HiH LoH HiA HiA

& & & & & &

AACL AACL AACL AACL AACL AACL

2 2 2 2 2 2

HIS & LoS & NPr& His & His & LoS &

AACL AACL AACL AACL AACL AACL

3 3 3 3 3 3

S-RF S-RF S-RF S-RF S-RF S-RF

& & & & & &

AACL AACl AACL AACL AACL AACL

4 4 4 4 4 4

Note.—Int. := introduction, HiH = high hotility feedback, LoH = low hottility feedback, HiA = high assertiveneu feedback, HiS = high similarity targeto, LoS = low timilarity targeh, NPr = no proiection, and S-RF = seif-rating form postprojection questions. AACL refers to the Affect Adjective Check List.

be presented first, while those for the "low hostility feedback" and "no overt projection" control groups will be discussed later.) The introduction explained that the experiment was concemed with the process of "interpersonal perception," and was identical for all groups. To obtain a baseline measure of anxiety, all subjects then completed the first of four state anxiety measures, which in every case was Zuckerman's Affect Adjective Check List (Zuckerman & Lubin, 1965). After the first AACL, subjects were given information conceming the nature of the trait to be dealt with during the experiment. The subjects in the two high dissonance groups were told about hostility, while those in the two low dissonance groups were told about assertiveness. To indicate that hostility is an ego-alien trait, the former subjects were informed that the trait of hostility is highly correlated with a poor emotional adjustment, is indicative of disharmonious interpersonal relationships, and is predictive of occupational difficulties or dissatisfactions. On the other hand, subjects in the assertiveness groups were told that assertiveness is predictive of academic and occupational successes, that it is one of the most frequendy found traits in the effective group leader, and that it is indicative of a healthy psychological adjustment; thus, assertiveness was presented as a very desirable trait. All subjects were then told that the first part of the experiment involved measuring the degree to which they possessed the trait under consideration (either hostility or assertiveness) by means of the Rorschach Test. (The assessment procedure was adapted from that devised by Holmes and Houston [1971].) The subject's task was to look at each of the 10 Rorschach inkblots one at a time, and then select from a list of descriptions which one out of the five provided per inkblot best represents the "symbolic meaning" of the inkblot stimulus; all descriptions made available were in fact whole (W),

368

Chalus

good form (F-\-) responses (Beck et al., 1961). Immediately after the subject indicated each response choice to the experimenter, he was given feedback presumably indicating whether or not the response revealed hostile (or assertive) tendencies within himself. The feedback sequences were prearranged and independent of the subject's Rorschach responses. Those in the high dissonance groups received feedback indicating that their choices to seven inkblots (numbers 1, 3, 4, 6, 8, 9, and 10) were hostile responses, while subjects in the two assertiveness groups were informed that they made seven responses indicating assertiveness (to the same inkblots). After the subject received the feedback for his tenth inkblot response and recorded the total of his hostile (or assertive) responses on his "Response Scoring Sheet," his attention was directed to the lower half of the sheet, where there were located the supposed psychological meanings of various hostile (or assertive) response totals to the Rorschach Test. Those in the high dissonance groups were thereupon informed that they made "an exceptionally large number of hostile responses to the Rorschach Test," and that in fact only 5% of all college students will make as many as seven hostile responses. Analogously, those in the low dissonance groups found that only 5% of all students will make seven or more assertive responses. After reading the applicable part of the interpretation schedule, all subjects completed the second anxiety measure, supposedly since obtaining a measure of the subject's present feelings would enable the best interpretation of his Rorschach performance to be made. All of the experimental subjects were then told that since the second part of the experiment dealt with "the way in which people evaluate or perceive the personalities of other people," their task was to read a number of Thematic Apperception Test (TAT) stories and then evaluate the personalities of the people who made them up. Subjects were then given the three TAT stories to be read (Stein, 1955), along with the TAT cards which elicited the stories (cards 4, 11, and 14). (The TAT stories were identical for all five groups allowed to project; they were selected on the basis of data obtained during the pretesting session so that all the selected stories had been seen by the pretesting respondents as reflecting moderate—as opposed to extremely high or low—^levels of hostility and a similarly moderate degree of assertiveness.) After reading each TAT story, the subjects estimated the number of hostile (or assertive) responses each person would have made on the Rorschach test they just took, the possibilities ranging from 0 to 10, inclusive for each target. Half of the subjects in each dissonance condition were told that the people who constructed the stories were male Northwestern students (similar targets),

Mechanisms of projection

369

while to the other half the targets were described as middle-aged male lawyers (dissimilar targets). The ratings of the personalities of the target figures allowed the subject to project either hostility or assertiveness, and the sum of the hostile or assertive Rorschach responses attributed to the three people constituted the measure of attributive projection. Immediately following the projection opportunity, all subjects were asked to complete the third AACL. The subjects in both the experimental and the control groups were then asked to respond to two postprojection questions on a "Self-Rating Form." These items were included to test the dissonance proposition that the projection of negative traits onto similar others reduces dissonance by permitting the projector to convince himself either that he does not possess such a great amount of the trait and/ or that the trait is not as undesirable as he had believed prior to the projection. The first item had the subject rate on an 11-point scale the degree to which he believed that he possessed the trait of hostility (or assertiveness) as compared with other Northwestem students, while the other item asked the subject to rate the desirability of the relevant trait, again on an 11-point scale. (If this dissonance proposition is correct, the subjects in the high dissonance-no projection group should see themselves as possessing more hostility and should also consider it to be a more undesirable trait than those in the high dissonance groups allowed to project.) AU subjects then filled out the fourth and final AACL, which was obtained to allow for the possibility that the anxiety-reducing effect of attributive projection might take more time to manifest itself. The completion of this anxiety measure marked the end of the experiment, and subjects were then thoroughly debriefed. Subjects in the "low hostility feedback" control group were treated identically throughout the experiment to those in the high dissonance group with similar (student) targets, with the sole exception that these subjects received Rorschach Test feedback indicating that they made only one hostile response (to inkblot 5). The interpretation schedule part of their Response Scoring Sheet then informed these subjects that they made an "average number of hostile responses" to the Rorschach compared to other college students. The procedure followed in the "no projection" control group was the same as that employed in the other high dissonance groups, except that the subjects in this group were not allowed to overdy project hostility onto others. Rather, they were informed that the second part of the experiment dealt with "the ability of individuals to recall the contents of other people's projective test productions," and that after reading a number of TAT stories, they were to be questioned regarding their recall

370

Chalus

Table 2. Mean anxiety scores over time. Group Period AACL AACL AACL AACL

1 2 3 4

HiH-HiS

HiH-LoS

HiH-NPr

LoH-HiS

HiA-HiS

HiA-LoS

6.05 7.45 6.95 6.60

5.10 7.45 5.30 5.05

6.65 9.35 6.75 7.05

7.15 6.10 6.80 6.50

6.70 6.45 6.75 5.55

6.40 5.90 5.70 5.20

Note.—n =: 20 per group mean.

of the contents of the stories. (The three stories read by these subjects [Stein, 1955] were elicited by TAT cards 1, 12M, and 2.) The reading of the TAT stories under the guise of a memory task in this control group prevented these subjects from projecting hostility overtly. To ideally inhibit, or at least not facilitate, the covert projection of hostility, three TAT stories were selected for this group which the pretesting respondents had judged as possessing relatively little hostile content. RESXJLTS

Effects of Feedback on Anxiety The means of the responses to the four state anxiety measures (AACL 1 through AACL 4) of the six independent groups are presented in Table 2. A between-within subjects analysis of variance on the first two anxiety measures from the two high dissonance-projection groups versus the two high assertiveness feedback groups revealed a significant Dissonance X Trials interaction, F(l,76) = 8.91, p < .005, as well as a significant main effect of Trials, F(l,76) = 3.96, p = .05, reflecting the overall increase in anxiety that was observed. A direct-difference i-test then applied to the AACL 1 and AACL 2 scores of the high dissonance-projection subjects indicated that a significant increase in level of anxiety occurred in these subjects, t (39) = 3.34, p < .002, while a corresponding test on the AACL scores of the subjects in the high assertiveness feedback groups revealed a nonsignificant decrease in anxiety, t (39) = .75. Thus, as expected, the high hostility feedback was successful in generating an ego threat in the high dissonance subjects, while the high assertiveness feedback did not produce such a threat. Finally, a betweenwithin analysis of variance performed on the AACL 1 and AACL 2 scores of the subjects in the low hostility feedback group in-

Mechanisms of projection Table 3.

371

Mean projection scores. Group

Sex Males Females

HiH-HiS

HiH-LoS

HiA-HiS

14.40

iXdif 14.10

11.30

HiA-LoS

11.80

LoH-HiS

7.80

Note.—n = 10 per cell mean.

dicated that this feedback resulted in a nonsignificant decrease in anxiety, F(l,18) = 1.76, p = .20, in accordance with expectations; both the main effect of Sex and the interaction of Sex X Trials did not approach significance. Projection Scores

The mean amounts of attributive projection exhibited in each of the eight group cells generated by the 3-variable factorial design (as well as the two cell means from the low hostility feedback condition) are shown in Table 3. To determine whether the experimental manipulation was successful in inducing the attributive projection of hostility, a 2 X 2 (Feedback X Sex) analysis of variance was performed on the projection scores of the subjects in the high hostility feedback-similar targets and the low hostility feedback-similar targets groups. The highly significant main effect of Feedback, F(l,36) = 28.25, p < .001, indicates that there was much greater projection in the high than in the low hostility feedback group as predicted, the overall means being 14.75 and 7.70, respectively; neither the main effect of Sex nor the interaction of Feedback X Sex approached significance. Since the basic projection effect was obtained, it is concluded that the results of the experiment are further interpretable, specifically in terms of the influence on the amount of projection of the Dissonance and Similarity variables. Accordingly, a T-score transformation was performed on the projection scores of the subjects in the four experimental groups, and a 2 X 2 X 2 (Dissonance X Similarity X Sex) analysis of variance was carried out on these transformed data. The analysis revealed the predicted significant main effect of the Dissonance variable, F(l,72) = 9.01, p < .005, indicating that subjects in the high dissonance condition projected more hostility (overall M = 14.52) than those in the low dissonance condition projected

372

Chalus

assertiveness (overall M = 11.30). Contrary to predictions, however, the main effect of Similarity was clearly not significant, F(l,72) = .08; the overall projection means in the high versus the low similarity targets groups were 13.08 and 12.75, respectively. Finally, the analysis revealed no main effect of Sex, F (1,72) = .00, and no interactions approached significance. Thus, the analysis of variance on the transformed projection scores provided support only for the vahdity of the dissonance explanation of the attributive projection of negative traits. Relationships Between Anxiety and Projection A prediction derived logically from the dissonance explanation is that subjects who were the most threatened by the receipt of high hostility feedback would be experiencing the greatest amount of dissonance and would thus feel the strongest need to reduce their dissonance by projecting hostility onto others. Taking the change in anxiety from AACL 1 to AACL 2 as a measure of threat, this proposition was tested by correlating the degree of ego threat experienced by each subject in the two high dissonance-projection groups combined with the amount of hostility subsequently projected by that subject. The nonsignificant product-moment correlation coefficient, r (38) = .17, however, indicates the existence of only a weak relationship between these two variables. The possibility was considered that a number of subjects in the high assertiveness groups would also experience some dissonance (and thus a need to project) in response to the personality feedback despite the fact that assertiveness was presented as a desirable trait. The corresponding analysis was therefore performed on the data from the subjects in the two high assertiveness feedback groups, but the obtained correlation between change in anxiety and amount of projection, r (38) = .18, again indicates the existence of a weak relationship. Because one of the variables in the above correlations consisted of changes on the AACL anxiety measure, and because difference scores are notoriously unreliable, one-way corrections for attenuation were performed on these correlations. In general, this procedure gives an estimate of the relationship which would have been obtained had the measures been perfectly reliable (Cuilford, 1954). As a result of this adjustment, the correlations

Mechanisms of projection

373

between change in anxiety and amount of projection in the high versus low dissonance conditions rose only to .21 and .23, respectively, neither of which are significant (in botii, p < .20). To obtain an estimate of the general relationship between degree of ego threat and amount of projection regardless of the desirability of the specific trait under consideration, the above corrected correlations were combined according to the "weighted z-transformation" procedure suggested by Snedecor and Cochran (1967). Averaging the correlations in this manner yielded an overall association which was significant due to increased degrees of freedom, r (76) = .22, p = .05. However, this correlation is still very low at an absolute level (accounting as it does for only 5% of the variance), and it must therefore throw into question the validity of the assumption that the attributive projection of negative traits serves a manifestly anxiety-reducing function. Dissonance theory would also predict that subjects in the high dissonance groups who projected the most would have subsequently shown the greatest reductions in anxiety from AACL 2 (immediately after feedback) to AACL 3 (immediately after projection opportunity), since the purpose of projection in this case was presumably to minimize dissonance (anxiety). If the larger positive numbers reflect greater anxiety reductions from AACL 2 to AACL 3, then a positive correlation should be expected between amount of projection of hostility and degree of anxiety reduction. However, the nonsignificant correlation, r (38) = .10, reveals no evidence for the anticipated relationship, and application of the one-way correction for attenuation increased the correlation to only .15, which, clearly, is still not significant. An unexpected finding regarding the anxiety data was that subjects in die high dissonance-no projection control group produced significantly higher AACL 2 scores than the combined subjects from the high dissonance-projection groups, F(l,56) = 4.84, p < .05. There is no immediately apparent explanation for the greater anxiety of the "no projection" subjects as all of the high dissonance groups had presumably been treated identically through the AACL 2 measure. A test of the defensive properties of projection in terms of the hypothesized differential decrease in anxiety from AACL 2 to AACL 3 between the projection and

374

Chalus

Table 4. Mean postprojection question scores. Group Question

HiH-HiS

HiH-LoS

HiH-NPr

LoH-HiS

HiA-HiS

HiA-LoS

Post 1 Post 2

6.55 4.15

7.55 3.50

6.40 3.30

6.80 2.60

4.60 7.50

4.00 8.45

Note.—The lower the number, the greater the self-perceived possession of the trait (Post 1), and the greater the undesirability of the trait (Post 2). n =: 20 per group mean.

no projection groups would therefore be uninterpretable, since such a test would presuppose statistical equivalence among all the high dissonance groups on the AACL 2 measure of anxiety. Postprojection Question Responses The means of the responses of the six independent groups to the two postprojection questions are presented in Table 4. The scores of the subjects in the high dissonance-no projection group in response to the first postprojection question were compared to those of the subjects in the high dissonance groups allowed to project by means of a 3 X 2 (Croups X Sex) analysis of variance. The nonsignificant main effect of Croups, F(2,54) = 1.28, indicates that, contrary to the dissonance prediction, subjects in the "no projection" control group did not rate themselves as possessing more hostility than did those in the two projection groups. Similarly, an analogous Groups X Sex analysis of variance on the data from the second postprojection question revealed that the "no projection" control subjects did not rate the trait of hostility as being more undesirable than did the subjects from the high dissonance-projection groups, F(2,54) = .64, a result which is again contrary to the dissonance prediction. DISCUSSION

Since the present study was concerned with two distinguishable issues (the validity of each of the two mechanisms alleged to underlie attributive projection), these topics will be discussed separately. The Dissonance Mechanism One of the purposes of this investigation was to provide data on the validity of the dissonance reduction explanation of the

Mechanisms of projection

375

attributive projection of negative traits. Support for this mechanism came primarily from two interrelated findings. First, the fact that the high hostility feedback significantly increased the subjects' levels of anxiety while the high assertiveness feedback resTilted in a nonsignificant anxiety decrease supports the interpretation that a dissonance-arousing ego threat was effected in the high hostility condition but not in the high assertiveness condition, as was expected. And second, tiie principal hypothesis derived from the dissonance mechanism was borne out: subjects in the high dissonance condition projected significantly more hostility than subjects in the low dissonance condition projected assertiveness. consistent with the dissonance predictions. For example, there was not a significantly greater amount of hostility projected in the high dissonance-similar targets group than in the high dissonance-dissimilar targets group; this differential projection effect was predicted since the dissonance explanation states that dissonance is especiaUy reduced by projecting undesirable traits onto similar and well-liked others (i.e., peer group members) rather than onto relatively dissimilar and potentially disliked others. However, the obtaining of this effect was at no time considered crucial, and it is also possible that the student and lawyer target figures were not perceived as dissimilar enough to allow for significant differential projection, a factor which has greater importance for the interpretation of the projection results in terms of the vahdity of the stimulus generalization explanation, to be discussed later. The responses to the two postprojection questions were also contrary to the dissonance predictions since (a) the high dissonance-no projection subjects did not rate themselves as possessing more of the trait of hostihty than did those in the high dissonance-projection groups, and (b) the former subjects failed to view hostility as a more undesirable trait than did the latter subjects. These negative results, however, would be expected if the "no projection" control group subjects projected hostility covertly and if such covert projection effectively reduced their experienced dissonance. In support of this possibility should be noted Stevens and Reitz's (1970) finding that 75% of their "no projection" subjects subsequently reported that they projected covertly anyway; another relevant

376

Chalus

item is the finding in the present study that the high dissonanceno projection subjects were unexpectedly significantly more anxious immediately after the false feedback than were the high dissonance-projection subjects, thereby probably indicating the presence of a stronger state of dissonance in the former subjects and a consequently greater motivation to covertly project hostility. It is concluded that the above considerations render uncertain the interpretation of the differential projection and postprojection question data in terms of the validity of the dissonance mechanism. The findings regarding the presumed defensive function of attributive projection were essentially negative. Considering first the association between degree of ego threat (change in anxiety) produced by the false personality feedback and subsequent amount of projection, it was only when the obtained correlations were corrected for attenuation and then combined across dissonance conditions that the resulting correlation just barely attained significance. The dissonance theory explanation probably should have predicted a stronger relationship in that the most highly threatened subjects should have felt the strongest need to use the defense of projection to reduce their dissonance. A related finding was the absence of a significant relationship between amount of projection of hostility and subsequent degree of anxiety reduction, even after the obtained correlation had been corrected for attenuation. This relationship would be expected since subjects who projected hostility the most should have been the most successful in reducing their levels of anxiety. In attempting to account for this negative result, the possibility should be considered that the anxiety generated by the negative feedback was ephemeral, and that it had dissipated by the time the projection phase of the experiment was completed, thereby obscuring the expected relationship. Nevertheless, it is apparent that this is a post hoc explanation which requires empirical validation before it can be confidently accepted. The results of this study conceming the defensive function of attributive projection can be compared with those of three other relevant interpretable studies (Bennett & Holmes, 1975; Holmes & Houston, 1971; Zemore & Creenough, 1973). Although projection was successfully elicited in all three studies, the evidence pertaining to the presumed defensive function of projection dif-

Mechanisms of projection

fered across studies. Holmes and Houston (1971) found no significant relationship in female subjects between degree of ego threat and amount of projection of psychological maladjustment, and no relation between amount of projection and subsequent degree of anxiety reduction, the results being consistent for both AACL and pulse rate measures of anxiety. Bennett and Holmes (1975) reported that female subjects who were informed that they failed on a test of intellectual abilities and who were then permitted to project failure onto others reported less postprojection anxiety on the AACL than failure subjects who were not allowed to project, results apparently in support of an anxiety-reducing function of projection. However, interpretation of this aspect of the study becomes ambiguous when it is considered that (a) although the postprojection anxiety difference between groups was reported as significant at the .10 level of confidence, it was not indicated whether this difference was also significant at the conventionally-acceptable .05 level; (b) the above anxiety difference was 2.02 AACL points, but fully 46% of this difference was already present immediately after the feedback but before the projection opportvuiity; and (c) the pulse rate data indicated that the failure-projection subjects were slightly and nonsignificantly more anxious after their projection opportunity than were the failure-no projection subjects at a comparable period in the experiment. Finally, Zemore and Creenough (1973) found that male subjects who were told that they scored high on a test of femininity did not exhibit differential decreases in AACL-measured anxiety depending on whether or not they were permitted to project femininity onto another male. However, when later given the opportunity to compensate for the threat (by performing physical exercise to demonstrate their strength and thus their masculinity), the ego threat-no projection subjects exhibited significantly more compensatory behavior than the ego threat-projection subjects, indicating that the threat had persisted only in the former subjects. The conclusion from these studies, as well as from the results of the present study, is that certain threats to self-esteem and the subjects' attempts to cope with them may produce effects too subtle to be readily detected by rather global self-report and physiological measures of anxiety. Other data are consistent with this interpretation; for example, both Holmes (1971) and

378

Chalus

Chalus (1976a) observed compensatory behavior in response to a mascuhnity threat despite the fact that self-report and pulse rate measures failed to reveal the presence of higher levels of anxiety in the threatened than in the nonthreatened subjects. Taken jointly, then, the accumulated evidence appears to permit the conclusion that the projection of negative traits can serve an ego-defensive function, but that the defensive effect operates in a subtle manner. Consequently, the dissonance theory explanation of attributive projection, as well as all other theories which may postulate a defensive function of this variety of projection, is seen as receiving modest support from these studies. Clearly, though, additional evidence of a more unambiguous nature must be acquired before this conclusion can be strengthened. In view of all the above data on the vahdity of the dissonance mechanism, it is concluded that this explanation of the attributive projection of negative traits has been supported. This conclusion is drawn primarily because the major dissonance prediction of the experiment was bome out, that is, that the high hostility subjects would project more than the high assertiveness subjects. The failure to find strong confirmatory evidence with regard to the anxiety-projection relationships is disappointing, yet this does not negate the occurrence of the major predicted dissonance finding. The Generalization Mechanism The other proposed explanation of attributive projection investigated in this study was stimulus generalization, which predicts greater projection onto similar versus dissimilar targets regardless of the desirability of the trait being projected. No support for the generahzation mechanism was obtained since the subjects did not project a greater amount of either hostility or assertiveness onto the similar targets (students) than onto the dissimilar targets (lawyers). The study which has acquired the strongest evidence in support of the validity of this mechanism, therefore, appears to remain that of Feshbach, Singer, and Feshbach (1963). In brief, this study found that male and female college students, after being made angry by insulting experimenter remarks, attributed more anger than they had previously onto pictures of similar

Mechanisms of projection

379

others (coUege-age males) but not onto pictures of relatively dissimilar others (12-year-old boys), whereas control subjects who were not subjected to the anger-arousing comments tended to decrease their attributions of anger onto both target categories. A consideration of the major differences between the design of the present study versus that of the study by Feshbach et al. may help to explain the difference in experimental results. First, the present study employed hostility and assertiveness as traits as opposed to Feshbach et al.'s use of anger as a state; the well-known fact that emotional states are transitory while enduring personality traits are not may make estimations of the feehngs of others more subjective or ambiguous than evaluations of their stable personality characteristics. Thus, the degree of projection may have been less readily influenced by the similarity variable in the present study. Another difference is that whereas Feshbach et al. had their subjects judge people depicted in photographs, the subjects in the present investigation used probably more concrete information (TAT stories) in making their ratings; again, it appears that the present study may have been less sensitive to the influence of the generalization mechanism since the use of specific personality test information may permit a more definite judgment about another's personality to be made than the use of probably less reveahng photographs. The final major difference between the two experiments is that the dissimilar targets were middle-aged male lawyers versus 12-yearold boys in the present and in Feshbach et al.'s study, respectively; if it is accepted that lawyers are closer to college students on a stimulus generalization similarity gradient than are children (perhaps because only students and lawyers share a common adulthood), then a smaller difference in amount of projection onto the similar versus dissimilar targets should have been expected in the present study. It is therefore possible that the lawyer targets were not perceived as dissimilar enough from the student targets to allow a differential projection effect to operate. Additional experimentation is clearly necessary, however, to ascertain whether the above speculations are valid. As part of this additional research, it is hoped that by varying the manipulated emotional states and the targets onto whom projection is made possible the generality of the original findings of Feshbach et al. (1963) will be extended. It is concluded from the preceding

380

Chalus

considerations that the results of the present study do not necessarily disconfirm the validity of the generalization mechanism. Prior research evidence exists which lends credence to this conclusion. This evidence relies on the fact that the attributive projection of acceptable, desirable traits is not amenable to explanation in terms of dissonance theory (since the possession of desirable traits does not arouse dissonance), while it is perfectly consistent with a stimulus generahzation process. For example, the attributive projection of happiness (Coldings, 1954), sexual arousal (Criffitt, 1973), and friendliness, likeability, and generosity (Wyhe, 1957) have all been demonstrated. Similarly, the projection of relatively nonpsychological characteristics such as age (Mintz, 1956) and height (Hinckley & Rethlingshafer, 1951; Ward, 1967) appears to be explainable in terms of the generahzation mechanism, but not in terms of the dissonance mechanism. In reference to the results of the present study, it should be noted that if stimulus generahzation contributes to the projection of both positive and neutral traits, it is logical to assume that it at least partially underlies the projection of negative traits as well. A final idea for future research would be to test the generalization mechanism's logical prediction that the tendency to engage in attributive projection should decrease as the age of the subject increases. Both Cattell (1951) and Feshbach and Singer (1957) have made this prediction, asserting that it takes time to acquire the appropriate discriminations between oneself and other people, discriminations which prevent the unwarranted inferences inherent in attributive projection from occurring. Such a result would be conceptually consistent with Wright's (1942) finding of an extremely high degree of attributive projection of generosity by young children and also with Bieri's (1955) finding that college students with complex cognitive structures evidenced less projective behavior than students with relatively simple structures. REFERENCES

Beck, S. J., Beck, A. G., Levitt, E. E., & Molish, H. B. Rorschach's test, I— Basic processes, 3rd Edition. New York: Gnme & Stratton, 1961. Bennett, D. H., & Holmes, D. S. Influence of denial (situation redefinition) and projection on anxiety associated with threat to self-esteem. Joumd of Persorudtiy and Sodd Psychology, 1975, 32, 915-921.

Mechanisms of projection

381

Bieri J. Cognitive complexity-simplicity and predictive behavior. Jourrud of Abnormd and Socid Psychology, 1955, 51, 263-268. Bramel, D. A dissonance theory approach to defensive projection. Joumd of Abnormd and Sodd Psychology, 1962, 64, 121-129. Bramel, D. Selection of a target for defensive projection. Joumd of Abnormd and Sodd Psychology, 1963, 66, 318-324. Cameron, N., & Magaret, A. Behavior pathology. Boston: Houghton Mifflin, 1951. Cattell, R. B. Principles of design in "projective" or misperception tests of personality. In H. H. Anderson & G. L. Anderson (Eds.), An iraroduction to projective techniques. Englewood Cliffs, N.J.: Prentice-Hall, 1951. Chalus, G. A. Defensive compensation as a response to ego threat: A replication. Psychologicd Reports, 1976, 38, 699-702. (a) Chalus, G. A. The relationship between paranoia and defensive projection, and the mechanisms alleged to underlie attributive projection (Doctoral dissertation, Northwestem University, 1976). Dissertation Abstracts IrOematiorud, 1977, 37, 3600B. (University Microfilms No. 77-1227.) (b) Edlow, D. W., & Kiesler, C. A. Ease of denial and defensive projection. Jourruil of Experimentd Socid Psychology, 1966, 2, 56-69. Epstein, R., & Baron, R. M. Cognitive dissonance and projected hostility toward outgroups. Joumd of Sodd Psychology, 1969, 79, 171-182. Feshbach, S., & Singer, R. D. The effects of fear arousal and suppression of fear upon social perception. Jourrud of Abnomud and Social Psychology, 1957, 55, 283-288. Feshbach, S., Singer, R. D., & Feshbach, N. Effects of anger arousal and similarity upon the attribution of hostility to pictorial stimuli. Jourrud of Consulting Psychology, 1963, 27, 248-252. Festinger, L. A theory of cognitive dissorumce. Stanford, CaUf.: Stanford University Press, 1957. Freud, S. Furliier remarks on the neuro-psychoses of defence. In J. Strachey (Ed. and Trans.), The standard edition of the complete psychologicd works of Sigmund Freud, (Vol. 3). London: Hogarth Press, 1962. (Originally published, 1896.) Goldings, H. J. On the avowal and projection of happiness. Jourrud of Persondity, 1954, 23, 30-47. Goldmann, T. S. Projection, paranoia, and cognitive dissonance (Doctoral dissertation, Pennsylvania State University, 1971). Dissertation Abstracts Intemationd, 1972, 32, 6678B. (University Microfilms No. 72-13, 855.) Gdffitt, W. Response to erotica and the projection of response to erotica in the opposite sex. Joumd of Experimentd, Research in Persorudity, 1973, 6, 330-338. Guilford, J. P. Psychometric methods (2nd Edition). New York: McGrawHill, 1954. Hinckley, E. D., & Rethlingshafer, D. Value judgments of heights of men by college students. Jourrud of Psychology, 1951, 31, 257-262. Holmes, D. S. Dimensions of projection. Psychologicd Bulletin, 1968, 69, 248-268. Holmes, D. S. Compensation for ego threat: Two experiments. Joumd of Persondity and Sodd Psychology, 1971, 18, 234-237. Holmes, D. S., & Houston, B. K. The defensive function of projection. Journal of Persorudity arid Sodd Psychology, 1971, 20, 208-213. Homey, K. New ways in psychoarudysis. New York: Norton, 1939. Kreines, D. C , & Bogart, K. Defensive projection and the reduction of dissonance. Jourrud of Sodd Psychdogy, 1974, 92, 103-108. Mintz, E. An example of assimilative projection. Jourrud of Almormal tmd Sodd Psychology, 1956, 52, 279-280.

382

Chalus

Secord, P. F., Backman, C. W., & Eachus, H. T. Effects of imbalance in the self-concept on the perception of persons. Joumd of Abnormal and Social Psychology, 1964, 68, 442-446. Snedecor, G. W., & Cochran, W. G. Statisticd methods, 6th Edition. Ames, Iowa: Iowa State University Press, 1967. Stein, M. I. The Thematic Apperception Test. Gambridge, Mass.: AddisonWesley, 1955. Stevens, H. A., & Reitz, W. E. An experimental investigation of projection as a defense mechanism. Joumd of CUnicd Psychology, 1970, 26, 152-154. Ward, C. D. Own h e i ^ t , sex, and liking in the judgment of the heights of others. Joumd of Personality, 1967, 35, 381-401. Wright, B. A. Altruism in children and the perceived conduct of others. Joumd of Abnormd and Social Psychology, 1942, 37, 218-233. Wylie, R. C. Some relationships between defensiveness and self-concept discrepancies. Joumd of Persondity, 1957, 25, 600-616. Zemore, R. W., & Greenough, T. J. Reduction of ego threat following attributive projection. Proceedings of the 81st Annud Convention of the American Psychologicd Assodation, 1973, 8, 343-344. (Summary.) Zuckerman, M., & Lubin, B. Manud for the Multiple Affect Adjective Check List. San Diego, Galif.: Educational and Industrial Testing Service, 1965. Manuscript received February 23, 1977.

The mechanisms underlying attributive projection.

The mechanisms underlying attributive projection' Gary Anton Chalus, Northwestern University ABSTRACT The experiment investigated the validity of th...
1MB Sizes 0 Downloads 0 Views