Journal of Personality and Social Psychology 1987, Vol. 52, No.

1,211-217

Copyright 1987 by the American Psychological Association, Inc. 0022.J5I4/87/S00.75

In Search of the False-Uniqueness Phenomenon: Fear and Estimates of Social Consensus

Jerry Suls and C. K. Wan
State University of New \brk at Albany

An examination of the social perception literature yields little evidence for the false-uniqueness phenomenon (Valins & Nisbett, 1972), the perception that one's attributes are more unique than is the case. In contrast, the tendency for individuals to project their own characteristics onto other people and assume that more people are like themselves is a robust phenomenon. One reason researchers may not have found false uniqueness is that they have not looked at the accuracy of consensus estimates. A close look at the results of Tabachnik, Crocker, and AUoy (1983) and Sanders and Mullen (1983), who did assess accuracy, suggests that people possessing undesirable attributes overestimate consensus, whereas people holding desirable attributes underestimate consensus. The latter pattern is a form of false uniqueness. In this study we looked at the accuracy of social consensus estimates in the context of psychological fears. A sample of subjects filled out an abbreviated version of a fear survey and made estimates of consensus. The results showed that both high- and low-fear respondents overestimated the incidence of high fear among their peers, but high-fear subjects were more inaccurate in their estimates. A false-uniqueness effect was found on the part of low-fear subjects, as they tended to underestimate the incidence of low fear among their peers. These findings are consistent with a motivational interpretation that emphasizes the individual's need to justify or normalize stigmatized behavior and to bolster perceived self-competence.

Social philosophers and social scientists have observed for some time that there is a persistent tension in social life between the desire to be similar to other people and the desire to be different or unique. For example, in the social science literature, Festinger's (1954; Suls & Miller, 1977) theory of social comparison emphasizes the need to be similar to others in order to make stable and accurate self-evaluations. In contrast, Snyder and Fromkin (1980) and Tesser (1980) have discussed situational and personal factors that motivate people to feel and act uniquely. In this respect, McGuire and his associates (McGuire & McGuire, 1982; McGuire, McGuire, Child, & Fujioka, 1978) have investigated the contents of the spontaneous selfconcept by categorizing responses to the query, "Tell us about yourself." McGuire found evidence for a distinctiveness principle: Attributes distinguishing the self from others in the general population or in the specific testing environment were especially likely to be mentioned by respondents. Most relevant in the present context, Valins and Nisbett (1972) proposed that many people who suffer from emotional disorders (e.g., depression, phobias) tend to see their problems as more unique than they are. For instance, Valins and Nisbett discussed the experience of new soldiers who, on joining long-established combat units, found themselves ostracized by the seasoned veterans. Not realizing that this behavior and their reactions to it were common to all those new to the battlefield, the "new guys" assumed there was something wrong with themselves, that their problems were unique. Valins and Nisbett recognized in their

1972 paper that only anecdotal evidence was available for false uniquenessthe perception that one's position or attributes are more uncommon than is actually the case.' Despite its intuitive appeal, little empirical evidence has come to light for a false-uniqueness effect since the Valins and Nisbett paper. In fact, there is little evidence that people possessing a minority attribute perceive their actual distinctiveness (i.e., perceive the true amount of support or consensus for their position). Instead, one of the most pervasive phenomena in the social perception literature is attributive projectionthe tendency for people to project onto other people characteristics that are identical to their own (Holmes, 1968; Murstein & Pryor, 1959).2 This phenomenon has also been referred to as the false-consensus effect by Ross, Greene, and House (1977; also see Gilovich, Jennings, & Jennings, 1983; Goethals, Allison, & Frost, 1979; Sherman, Presson, Chassin, Corty, & Olshavsky, 1983). The general finding in attributive-projection and false-consensus research is that respondents assume there are more people like themselves than do respondents with different attributes. In an early representative study, Secord, Backman, and Eachus (1964) had subjects rank 15 personality

Valins and Nisbett (1972) actually used the term false distinctive-

ness to refer to the mistaken assumption that one's attributes are uncommon. However, so as not to be confused with distinctiveness (information) in an attributional sense (Kelley, 1971), we prefer Jalse uniqueness.
2

Attributive projection assumes that the individual is consciously

aware of possessing the attribute. In Holmes's (1968) comprehensive Correspondence concerning this article should be addressed to Jerry Suls, Department of Psychology, State University of New 'Vbrk at Albany, Albany, New York 12222. review, he found little compelling evidence for the classic Freudian projection in which individuals project their own feelings into others but do not acknowledge those feelings in themselves.

211

212

JERRY SULS AND C. K. WAN

traits in terms of their desirability and the degree to which each trait was characteristic of themselves and a friend. Subjects were subsequently informed via bogus feedback that they possessed a trait they had previously rated as undesirable. After receiving this feedback, subjects attributed more of the trait to themselves and their friends on a second ranking. More recently, Ross et al. (1977) found that subjects estimated that a higher percentage of their peers possessed the personal attributes they used to describe themselves than those who did not. Ross et al.'s data indicate that attributive projection is a highly robust phenomenon because it was demonstrated over a wide range of personal attributes and behaviorsboth socially desirable and undesirable. Two items used by Ross et al. are of special interest to us here: "Am frequently depressed" and "Have difficulty controlling my temper." Although it may be expected that these items would be endorsed by a minority of (college student) subjects, which was the case, these subjects did not perceive their position as distinctive. Depressed and irritable respondents saw these attributes as more common among their peers than did their nondepressed and nonirritable counterparts. In a study of how depressed people perceive consensus for their attributes, Tabachnik, Crocker, and Alloy (1983) found that depressed subjects showed less false consensus (i.e., attributive projection) than did nondepressed subjects. However, both groups tended to see themselves as similar in attributes to the average college student (see Tabachnik et al., 1983, p. 692, Figure 1: The discrepancy score between depressed subjects' selfperceptions and their perceptions of the average college student bordered on zero). On the basis of Tabachnik et al. and Ross et al., it would appear that subjects possessing a negative attribute (such as depression or anger) do not perceive themselves to be relatively unique. However, whether one finds a uniqueness bias may depend on how it is denned and operationalized. In the studies just reviewed, estimates of consensus for an attribute by one group were contrasted with the estimates of consensus for that attribute by a group holding the alternative trait. In concrete terms, in Tabachnik et al., depressed subjects made estimates of the frequency of depressed individuals, and these estimates were compared with the estimates of the frequency of depressed individuals made by nondepressed subjects. However, it is possible that false uniqueness may emerge if one compares estimates of consensus with the frequency of the attribute or trait in the population. Only two studies have examined the accuracy of such estimates. Tabachnik et al. computed accuracy scores by taking the difference between subjects' estimates of the percentage of their peers who would endorse each attribute and the actual percentage of subjects in an independent sample who endorsed each attribute. Their results indicated that (a) depressed subjects were more inaccurate in an absolute sense (ignoring the direction of discrepancy) and (b) both depressed and nondepressed subjects overestimated the number of people characterized by depressive attributes. Also, although it was not statistically significant, there was a tendency for depressed subjects to overestimate the frequency of depressed attributes more often than did the nondepressed subjects. Results like these provide little support for the notion that persons possessing negative or stigmatized attributes perceive themselves as falsely unique. It

also appears that they do not perceive their actual uniqueness either. In another study, Sanders and Mullen (1983), using several of Ross et al.'s items, compared subjects' consensus estimates with actual population figures. Sanders and Mullen (1983) found a consistent tendency for persons possessing a minority attribute (such as frequent depression and irritability) to overestimate the number of others who shared this attribute. Again, there was no suggestion that persons with negative attributes perceive themselves to be falsely unique (i.e., more unique than is the case) or veridically unique (i.e., as unique as they actually are). Rather, there was a consistent tendency for them to inflate or magnify support for their position. Ironically, there is evidence for false uniqueness on the part of the majority attribute holders. Although they did not emphasize it, Tabachnik et al. found that both depressed and nondepressed subjects underestimated the number of people who have nondepressive, "psychologically healthy" attributes ("1 am assertive," "I am cheerful"). Although it is not surprising, in light of the data just reviewed, for depressed subjects to underestimate the frequency of nondepressed people, the fact that nondepressed subjects underestimate the frequency of healthful attributes suggests these subjects see themselves as uniquein this case, falsely so. The point is made more strongly in Sanders and Mullen's (1983) data where subjects who actually shared a positive attribute (e.g., being able to control their temper) with a majority of their peers underestimated the prevalence of that attribute. Taking these past results together, an interesting picture emerges. Although it was initially expected that persons holding negative attributes might perceive themselves as falsely unique, it appears instead that persons holding positive attributes are more likely to perceive themselves as falsely unique. In contrast, persons with minority negative attributes tend to inflate support for their position. These results are compatible with a motivational perspective that emphasizes the individual's need to justify or normalize stigmatized behavior (see Sherman, Presson, & Chassin, 1984). For example, an individual who is highly fearful of heights may feel more comfortable believing that many others have the same problem. But for those who are not fearful, feelings of competence and esteem may be higher if they think there are very few strong, fearless people like themselves. In this article we report the results of a study that examined perceived consensus and uniqueness in the area of psychological fears. Such a study seemed to be needed for two reasons. First, there are only a few studies in the psychological literature that have examined the accuracy of consensus estimates. Our analysis and interpretation of the Tabachnik et al. and Sanders and Mullen studies are speculative and in need of additional tests. Second, it is not clear that the attributes examined in these past studies are the most appropriate for finding false uniqueness. Perhaps neither investigation found false uniqueness on the part of minority (undesirable) attribute holders because depression and irritability are relatively common in college populations. In the present study, by looking at both common and uncommon fears we will have a better opportunity to ascertain whether there are any conditions in which highly fearful persons perceive their actual uniqueness or overestimate their uniqueness.

FEAR AND CONSENSUS

213

We had two major aims in the present study. First we wanted to examine the prevalence of attributive projection among high- and low-fear subjects. On the basis of previous evidence, it was hypothesized that high-fear subjects would, relative to low-fear subjects, estimate that a larger proportion of their peers has high fear. Second, we wanted to examine the accuracy of high- and lowfear subjects' consensus estimates. On the basis of a motivational or ego-defensive perspective, it was hypothesized that low-fear subjects would tend to underestimate consensus for their attributes (i.e., false uniqueness) whereas high-fear subjects would overestimate consensus for their attributes. Note that the second hypothesis also suggested that both groups would overestimate the prevalence of high fear among their peers but that the high-fear group would be more in error (see alsoTabachniketal., 1983).

some, much, very much, terror). Respond to all items as honestly as you can. Each fear item was followed by the 7-point fear scale. This Likert format has been extensively used by behavioral therapists and researchers. On the next page of the questionnaire, subjects were asked to estimate what percentage of female undergraduates at the university had high fear, that is, much, verymuch, or terror for each of the eight items. After completing the questionnaire, subjects were asked to put it in the envelope that was provided before returning it to the experimenter.

Results Main Analyses The convention in behavioral treatment assessment with the Geer (1965) survey is to classify subjects checking one of the top three items (much, very much or terror) as highly fearful. The remainder of the subjects are considered low in fear or nonfearful. Typically, it is the subjects responding in the top three categories who are recruited for behavioral treatment (e.g., Lott & Murray, 1975). Following this convention, we classified subjects who checked one of the three top categories as high in fear and subjects who checked the other four as low in fear. The first aim of the research was to compare estimates of the frequency of high fear between low- and high-fear subjects. The mean estimates for each item are indicated in Table I , columns 3 and 7. Analyses of variance (ANOVAS) with high-low fear as the independent variable and population estimates as dependent variables were computed for each item. Table 1 shows that high-fear subjects made higher estimates of fear prevalence for all items, but the effect was statistically significant for Items 2, 6, and 8 (see column 10). Examination of Table I shows that analysis of the data for Item 5 ("fear of the opposite sex") was not appropriate given the very small number of subjects reporting high levels of fear of the opposite sex (n = 4). There were statistically significant attributive projection effects for three of the seven analyzable items. That is, subjects who reported that they were highly fearful tended to make higher estimates of the frequency of fear among their peers than did low-fear subjects. On first examination, it may seem surprising that attributive projection, which is said to be so pervasive (see Holmes, 1968), was not obtained on every item. However, as we will discuss later, this is not so surprising if the tendency on the part of lessfearful subjects to project their own views conflicts with the desire to feel unique in a positive way (i.e., see many others as more fearful). In the second stage of data analysis, we compared population estimates to the actual percentage of subjects in our sample reporting high fear (given the high-participation rate, we assumed our sample was representative of the student body). To accomplish this, absolute discrepancy scores were computed by sub-

Method Subjects
One hundred forty-nine female undergraduate students at a large state university in the Northeast served as subjects.3 These respondents were all between 18 and 23 years of age and were solicited for participation in a variety of locations on campus, although predominantly in their dormitories. Participation was enlisted by three female experimenters who approached each prospective subject and described the study as a "survey on the incidence of various fears." The anonymity and brevity of the survey were emphasized so as to encourage maximum participation. If a subject agreed to participate, the survey instructions were explained and the survey and an envelope were left with the subject. In all cases, subjects filled out the questionnaire in privacy; the experimenter arranged to pick up the completed instrument within 15 min. On those occasions when the subject requested more time, the experimenter made an appointment to pick up the completed survey the next day. Because of these special arrangementssurvey brevity, anonymity, administration in a convenient setting, and same-sex experimenteronly 3% of those approached declined. Furthermore, of the total number of questionnaires collected (155), only 6 were so incomplete as to preclude including them in the data analysis.

Materials and Instructions

Subjects responded anonymously to an abbreviated version of the Fear Survey Schedule-II developed by Geer (FSS-II; 1965). The FSS-II consists of 51 items that list a broad sample of objects and situations that elicit fear, and it has been used extensively in studies of behavioral treatment of fears. The version of the scale used in this study consisted of eight items representative of the principal factors that have emerged from research with this instrument (Bernstein & Allen, 1969). The specific items are listed in Table 1. Our subjects read the following instructions before responding to the questionnaire: This inventory is designed primarily to explore the incidence of certain fears among college students. Your frank responses will enable us to better understand and establish new approaches to difficulties experienced by college students. Below you will find a list of situations or objects that may cause anxiety, uneasiness, or other unpleasant feelings. You are being asked to indicate the degree of fear, anxiety, or uneasiness that each of the situations or objects creates for you. For each item, put an X in the column that best describes your feeling of anxiety (none, very little, a little,

3 Only female subjects were used because we wanted to be assured of obtaining a reasonably high percentage of subjects falling into the highfear categories. On the basis of Geer's (1965) and other's (e.g., Bernstein & Allen, 1969) data, female samples report higher levels of fear than do male samples. We also expected they would be more eager to participate, thus helping to minimize attrition.

214

JERRY SULS AND C. K. WAN

Table 1 Estimates of Fear and Absolute and Directional Error for Respondents High and Low in Fear High-fear subjects Actual % Estimated reporting % of high Absolute Directional high fear n fear error error 20.8 31 50.0
29.2 31.8 118 ((30) = 6.6, 2.71 79)= 1.7, 11.7 ((61) = 3.9, p<.00i 14.7 ((59) = 4.5, p<.OOI
69

Low-fear subjects Estimated % of high Absolute fear error

49.8 29.0

High fear vs. low fear

Item 1. Being criticized

Directional error

Estimates

Absolute error

2. Snakes

53.7

80

56.4

27.5

42.8

23.1

30.8 ns ns ((117)= 13.4, p<.001 -10.9 fl\, 147) = 8.8, fl[l,147) = 4.1, ((68) = 19.4, p < .004 p < .04
9.8 ((85) = 3.8, p<.001 9.2 ((87) = 3.9,
JK.OOl

3. Speaking before a group 4. Not being a success 5. Being with a member of the opposite sex' 6. Spiders

41.9

62

53.6

22.4

86

51.7

21.3

ns

ns

40.5

60

55.2

24.7

88

49.7

19.0

ns

f\\, 146) = 5.1, p < .03

2.7 34.9

4 52

46.3 51.9

44.0 29.5

7.

Being selfconscious

20.4

30

54.7

35.4

8. Making mistakes

21.5

32

58.0

40.0

44.0 145 17.0 97 ((51) = 4.1, p<.00\ 34.4 117 ((29) = 7.3, p<.001 36.6 117 ((31) = 8.1, p<.00\

45.3 43.3

42.7 20.0

50.2

31.8

46.8

27.5

42.7 f\\, 147) = 3.8, f ( l , 1 4 7 ) = 13.6, 8.4 p< .05 p< .003 ((96) = 3.7, p<.00\ ns ns 29.8 ((116)= 14.2, /x.OOl F(l, 147) = 6. 5, F(\, 147)= 11.7, 25.3 ((116)= 12.9, p<.OI /x.OOl p<.001

Note. The ( values in columns 5 and 9 indicate the results of comparison with 0 (two-tailed test). The /"values in columns 10 and 11 represent results of contrast of high- and low-fear groups. " There were too few high-fear subjects on this item to perform an analysis of variance.

tracting the actual percentage from each subject's estimate. Mean absolute discrepancy scores were computed for the highand low-fear groups for each item. The mean discrepancy scores are shown in Table 1, columns 4 and 8, for high- and low-fear subjects, respectively. To assess the degree of inaccuracy, analyses of variance contrasting the absolute discrepancy scores of the high- and low-fear groups were conducted. The results are listed in Table 1, column 11, and demonstrate that high-fear subjects were significantly more in error for four items (Items 2, 4, 6, 8). Although not statistically significant, high-fear subjects tended to be more in error on the other items as well. It is clear that both groups were in error about fear prevalence; however, the high-fear group was generally more in error. In the third stage of data analysis, we computed directional discrepancy scores for each subject, taking into account the sign and discrepancy between estimates and the actual population percentage. The directional error scores for high- and low-fear subjects are indicated in Table 1, columns 5 and 9, respectively. In all but one case (high-fear subjects for Item 2), / tests contrasting these scores with zero (no error) were significant. Note that Item 2 (snakes) was the only item for which the majority of subjects fell into the high-fear category. This was also the only item for which there was a negative directional error score; that is, low-fear subjects underestimated the prevalence of high fear. In all other cases, both high- and low-fear subjects significantly

overestimated the prevalence of fear among their peers. Thus, it can be seen that subjects did not exclusively project their own feelings but, rather, overestimated the incidence of a socially undesirable, inappropriate behavior. Analyses of variance were not computed to contrast the directional error scores of the high- and low-fear groups. Because directional error was calculated by subtracting the actual population frequency from estimates made by high- and low-fear subjects, this analysis was essentially the same as that reported in column 10. (Subtracting the population figure means subtracting a constant from both high- and low-fear estimates, so contrasting the high- and low-fear estimates yields the same relative results as reported in column 10.) Column 10, as well as entries in columns 5 and 9, indicate that the directional error scores of the high-fear subjects were significantly greater for three of the analyzable items (Items 2, 6, and 8). In these three cases, the high-fear group overestimated the prevalence of fear significantly more than did the low-fear group. However, there was a trend across all items for the high-fear subjects to overestimate more than the low-fear subjects. These results indicated that highly fearful subjects did not perceive false uniqueness or the actual uniqueness of their behavior. Instead, they tended to perceive their behavior as consensual, estimating that at least 50% of their peers felt as they did (with the exception of Item 5:46%, n = 4).

FEAR AND CONSENSUS The directional error scores of the low-fear subjects indicated a false-uniqueness bias on their part. For example, for the fear of spiders, low-fear subjects overestimated the incidence of high fear by 8.4%, which means they underestimated their level of (low) fear by 8.4%. Hence, the low-fear respondents perceived their own (desirable) position as less common than was actually the case. An examination of the directional error scores of the low-fear subjects in column 9 shows that these subjects consistently overestimated the prevalence of high fear, which indicates that they underestimated the percentage of their peers sharing low fear with them. The only reversal of this tendency was on Item 2 where the majority of subjects reported high levels of fear. Thus, as predicted, we found evidence of false uniqueness on the part of subjects low in fear and inflated consensus on the part of subjects high in fear. Table 2 Estimates of High-Fear by Respondents in High-Fear Categories

215

Item

Category much fear very much fear + terror much fear very much fear terror much fear very much fear terror much fear very much fear + terror much fear very much fear terror much fear very much fear + terror much fear very much fear + terror

Estimated % having high fear

50.2 49.5 56.1 52.0 62.3 50.5 53.3 70.0 50.2 59.7 47.1 48.5 64.1 51.8 61.7 53.2 75.4

1. Being criticized
2. Snakes

21 10 28 29 23 36 19 7 28 32 19 20 13 21 9 25 7

3. Speaking before a group 4. Not being a success 6. Spiders

Supplementary Analyses
Before discussing the implications of these results, we think it best to address some possible criticisms of the results. It might be argued that by combining subjects who responded much, very much, and terror, we may have "washed out" a uniqueness effect on the part of the most fearful subjects. To explore this possibility, we examined the distribution of responses across our high-fear respondents. We found there were a sufficient number of subjects to make three categoriesmuch, very much, and terrorfor Items 2, 3, and 6. For Items 1,4, 7, and 8 we established only two categoriesmuch and very much + terror (there were too few terror subjects to make a separate group). Item 5 could not be used because only 4 subjects responded in the top three categories. Table 2 lists the population estimates of the high-fear categories. This table indicates that the highest estimates were made by the groups highest in fear. The only exception is Hem I where the mean for the much fear category was larger than the higher combined category by less than one percentage point. There appears to be no evidence that combining subjects at the high end of the continuum with subjects somewhat lower in fear obscured a uniqueness effect. Instead, we see that as subjects' fear level increased, so did their perceptions of fear in the population. Still another possibility is that only persons seeking psychological treatment may perceive themselves as falsely unique. The present study collected other supplementary data that bear on this issue. The questionnaire asked subjects if they would desire psychological help if it was available and inexpensive. Subjects were categorized into high-fear-treatment, high-fearno-treatment, low-fear-treatment, and low-fear-no-treatment categories. Analyses of variance were then computed on the population estimates for each item. Although there were some significant differences between high- and low-fear subjects (paralleling earlier analyses), there were no systematic differences between the estimates of subjects who desired treatment and those who did not (it is worth noting that 39.6% of the high-fear subjects indicated a desire for treatment averaged across items). Thus, there was no evidence that a desire for treatment engenders perceptions of uniqueness. These supplementary data are also relevant to a recent investigation by Snyder and Ingram (1983) who gave high-anxiety

7. Being selfconscious 8. Making mistakes

and normal subjects feedback that they possessed test anxiety and information about whether the problem was common or not (consensus information). Snyder and Ingram found that test-anxious subjects were more likely to seek help when they learned that their problems were common. In contrast, the present study did not find that high-fear subjects who were interested in treatment made higher consensus estimates than those subjects not interested in treatment. But these results are not necessarily contradictory. Recall that our subjects made up their own estimates, whereas Snyder and Ingram (1983) provided information that was presumably systematically collected by psychologists. Taking the two data sets together, it appears that in order to induce people to seek professional psychological assistance, it may be necessary to provide consensus information from experts.

Discussion
The present study's results showed that subjects both high and low in fear overestimated the incidence of high fear among their peers; however, high-fear subjects were more inaccurate in their estimates of fear prevalence; that is, there was evidence of attributive projection (or false consensus; Goethals et al., 1979; Ross et al., 1977) in that high-fear subjects made higher population estimates than did low-fear subjects. There was no evidence that high-fear subjects perceived their position as either falsely unique or that they perceived the actual uniqueness of their (minority) position. Instead, they tended to inflate consensus for their position. But there was evidence of a false-uniqueness position on the part of the low-fear subjects. By overestimating the frequency of high fear, the low-fear subjects were essentially underestimating the incidence of low fear among their peers. These results are consistent with a motivational explanation.

216

JERRY SULS AND C. K. WAN

By overestimating the prevalence of fear among their peers, individuals high in fear can feel more comfortable"others feel as I do." Attribution processes (Kelley, 1971) also probably come into play. High-consensus information ("Many others act or feel the same way") should yield an inference that the situation rather than the individual is the primary causal agent of the behavior. As Snyder and Ingram (1983) noted, "the psychological problem is externalized, which should be reassuring" (p. 1118). In contrast, if a behavior has a low consensus that may lead to the perception, the individual is chiefly responsible thus, internalizing the problem. The results for the low-fear subjects can also be given a motivational interpretation. By perceiving dissimilarity between the self and a target population (i.e., overestimating the incidence of high fear), low-fear subjects can boost their own feelings of competence and psychological health. Also, perceiving high fear as pervasive, but being unfearful oneself, can, via the augmentation principle (Kelley, 1971), lead to the attribution to one's special competence and strength ("By virtue of my special capabilities, 1 can resist situational pressures to be afraid"). In general, the results were consistent with the arguments made in the introduction that high-fear subjects would be more in error and overestimate the incidence of fear among their peers. In contrast, low-fear subjects by overestimating high-fear incidence underestimate consensual support for their position. One result of the investigation, which may be surprising, is that evidence for attributive projection was not as strong as a reading of past literature may suggest. However, our failure to find strong differences is explicable: Low-fear subjects have more to gain if they inflate their population estimates of dissimilar (lessfortunate) others, thus rendering their estimates closer to those of the high-fear subjects. This perspective is interesting because, unlike psychoanalytic approaches to projection that emphasize the motivational forces on the part of those with phobias and neuroses, the present study suggests that both fearful and unfearful respondents are motivated to make self-serving estimates. The data also bear on some more general issues. Although there is a large body of literature on projection, with rare exceptions investigators have compared perceptions to actual or sample figures in order to establish accuracy. The question of accuracy takes on special relevance in the discussion of false uniqueness. On the basis of the present results, it appears that false-uniqueness effects will emerge when comparisons of estimates are made with actual sample figures. These comparisons are also necessary to determine whether individuals holding minority attributes perceive their actual uniqueness. As we saw, there is little evidence that persons accurately construe their position, at least with respect to psychological fears. It may appear surprising that we found no evidence for Valins and Nisbett's (1972) intuitively reasonable notion that disturbed subjects should perceive themselves as more unique than they really are. However, perhaps the discrepancy is more apparent than real. Valins and Nisbett emphasized the role of uniqueness in the development of fear or other maladjustments. Perhaps perceptions of one's uniqueness occur early on in the development of emotional and behavioral disorders. With time, however, individuals may, for ego defensive reasons, come to inflate support for their problems. These comments suggest the

need for a longitudinal study of the etiology of fear and its relation to perceived social consensus. Another possibility is that fearful subjects refer themselves for treatment when they recognize their uniqueness. This suggests the need to study the social consensus estimates of those who not only have the desire for psychological or psychiatric treatment but actually enter treatment. We are currently conducting such a study. In the beginning of this article, we mentioned the conflicting tensions between the desire to be the same as others and the desire to be distinctive. The present study found evidence for both tendencies, which appear to be best explained in motivational or ego-defensive terms (see Goethals, 1986; Tesser, 1980; Tesser & Campbell, 1983). Highly fearful subjects protect their esteem by perceiving that there are more people like themselves. Individuals low in fear bolster their competence by perceiving themselves as more unique than they are. The results do not lend themselves as easily to a cognitive interpretation. On the one hand, high-fear subjects may overestimate their position because it is highly available (Kahneman & Tversky, 1973; Nisbett & Ross, 1980). But why should lowfear subjects overestimate the prevalence of fear (thus underestimating their own position) when their own opinion should be most available? One possibility is that exposure to a single or to a few fearful persons because of their novelty and vividness may cause these experiences to stand out in memory. At this time, however, empirical support for the vividness heuristic is limited (see Taylor & Thompson, 1982). In any case, a number of commentators (see Tetlock & Levi, 1982) have suggested that it may not be possible to resolve the controversy between cognitive and motivational approaches. Perhaps a more fruitful avenue of study would be to examine the specific dimensions and situations in which people prefer to perceive themselves as the same or different from others (Snyder & Fromkin, 1980). References
Bernstein, D. A., & Allen, G. J. (1969). Fear survey schedule (II): Normative data and factor analysis based upon a large college sample. Behavior Research and Therapy, 7, 403-408. Festinger, L. (1954). A theory of social comparison processes. Human Relations, 7, 117-140. Geei; J. H. (1965). The development of a scale to measure fear. Behavior Research and Therapy, 3. 45-53. Gilovich, T., Jennings, D. L., & Jennings. S. (1983). Causal analysis and estimates of consensus: Undermining the false consensus effect. Journal of Personality and Social Psychology, 45. 550-559. Goethals, G. R. (1986). Constructing and ignoring social reality: Positive self-appraisal consistent estimates of consensus. In J. Olson, C. P. Herman, & M. Zanna (Eds.), Relative deprivation and social comparison: The Ontario Symposium (pp. 137-157). Hillsdale, NJ: Erlbaum. Goethals, G., Allison, S. J., & Frost, M. (1979). Perceptions of the magnitude and diversity of social support. Journal of Experimental Social Psychology. 15, 570-581. Holmes, D. S. (1968). Dimension of projection. Psychological Bulletin, 69, 248-268. Kahneman, D., & Tversky, A. (1973). On the psychology of prediction. Psychological Review. 80, 237-251. Kelley, H. H. (1971). The process of causal attributions. American Psychologist, 28, 107-128.

FEAR AND CONSENSUS Lott, D. R., & Murray, E. J. (1975). The effect of expectancy manipulation on outcome in systematic desensitization. Psychotherapy: Theory. Response and Practice, 12, 28-32. McGuire, W. J., & McGuire, C. (1982). Significant others in self-space. In J. Suls (Ed.), Psychological perspectives on the self (\o\. l.pp. 7196). Hillsdale, NJ: Erlbaum. McGuire, W. J., McGuire, C., Child, P., & Fujioka, T. (1978). Salience of ethnicity in the spontaneous self-concept as a function of one's ethnic distinctiveness in the social environment. Journal of Personality and Social Psychology. 36, 511-520. Murstein, B. I., & Ftyor, R. S. (1959). The concept of projection: A review. Psychological Bulletin, 56, 353-374. Nisbett, R., & Ross, L. (1980). Human inference: Strategies and shortcomings of social judgment. Englewood Cliffs, NJ: Ftentice-Hall. Ross, L., Greene, D., & House, P. (1977). The false consensus effect: An egocentric bias in social perception and attribution processes. Journal of Experimental Social Psychology, 13, 279-301. Sanders, G. S., & Mullen, B. (1983). Accuracy in perceptions of consensus: Differential tendencies of people with majority and minority positions. European Journal of Social Psychology, li. 57-70. Secord, P. F., Backman, C. W., & Eachus, H. T. (1964). Effects of imbalance in the self-concept on the perception of persons. Journal of Abnormal and Social Psychology, 68,442-446. Sherman, S. J., Presson, C. C, & Chassin, L. (1984). Mechanisms underlying the false consensus effect: The special role of threats to the self. Personality and Social Psychology Bulletin, 10.127-138. Sherman, S. J., Presson, C. C., Chassin, L., Corty, E., & Olshavsky, R. (1983). The false consensus effect in estimates of smoking prevalence. Personality and Social Psychology Bulletin, 9, 279-301.

217

Snyder, C. R., & Fromkin, H. (1980). Uniqueness: The human pursuit of difference. New York: Plenum Press. Snyder, C. R., & Ingram, R. E. (1983). "Company motivates the miserable": The impact of consensus information on help seeking for psychological problems. Journal of Personality and Social Psychology, 45,1118-1126. Suls, J., & Miller, R. L. (Eds.). (1977). Social comparison processes: Theoretical and empirical perspectives. Washington. DC: Hemisphere. Tabachnik, N., Crocker, J., & Alloy, L. (1983). Depression, social comparison, and the false consensus effect. Journal of Personality and Social Psychology, 45, 688-699. Taylor, S. E., & Thompson, S. C. (1982). Stalking the elusive "vividness" effect. Psychological Review. 89, 15 5-181. Tesser, A. (1980). Self-esteem maintenance in family dynamics. Journal of Personality and Social Psychology, 39, 77-91. Tesser, A., & Campbell, J. (1983). Self-definition and self-evaluation maintenance. In J. Suls & A. G. Greenwald (Eds.), Psychological perspectives on the self(Vol 2, pp. 1-32). Hillsdale, NJ: Erlbaum. Tetlock, P. E., & Levi, A. (1982). Attribution bias: On the inconclusiveness of the cognition-motivation debate. Journal of Experimental Social Psychology. 18, 68-88. Valins, S., & Nisbett, R. E. (1972). Attribution processes in the development and treatment of emotional disorders. In E. E. Jones, D. E. Kanouse, H. H. Kelley, R. E. Nisbett, S. Valins, & B. B. Weiner (Eds.), Attribution: Perceiving the causes of behavior (pp, 137-150) Morristown, NJ: General Learning Press. Received September 20, 1984 Revision received March 5, 1985