STUDY 1
We have proposed that some scientific conceptions of personality reflect an essentialist understanding of human attributes, and that this understanding may also be reflected in laypeople’s implicit personality theories. Study 1 was conducted to determine whether essentialist beliefs that had been found to covary in research on social categories (Haslam et al., 2000, 2002) also cohered in beliefs about personality. Simply put, we asked whether beliefs about the discreteness, biological basis, immutability, informativeness, homogeneity, and inherence of personality characteristics form a coherent set, such that characteristics judged high on each property also tend to be essentialized on the others.We also asked how the structure of this expected covariation should be described.
Previous research and theory offer two alternative models of the structure of essentialist beliefs. Gelman’s (2003) work implies a unifactorial structure, as she formed a composite essentialism score by combining all of her items. Alternatively, previous work by Haslam et al. (2000, 2002) favors a model containing distinct and orthogonal dimensions of naturalness and entitativity consistent with Rothbart andTaylor’s (1992) description of essentialism as having distinct aspects of inalterability and inductive potential. However, this work exclusively addresses social categories, and the two-dimensional structure may not apply in the personality domain. If a different structure applies in that domain, then essentialist thinking might have different determinants and implications from those documented in the study of social categories. For example, essentialist beliefs about personality might not have the same links to prejudice and devaluation. A unifactorial structure would also cast doubt on the applicability of concepts of natural kind and entitativity in the personality domain. Given the plausibility of both one- and two-dimensional models, we made no specific structural predictions beyond an expectation of overall covariation among the essentialist beliefs.
Study 1 also tests three hypotheses about possible correlates of personality essentialism. Little is known about why some social distinctions are essentialized more than others and some possibilities—for example, visible morphology for race and gender (Rothbart & Taylor, 1992), descent and endogamy for ethnicity (Gil-White, 2001), abrupt transformation (Keil et al., 1999) for disease—do not apply to personality characteristics. Our hypotheses were, therefore, speculative. First, we hypothesized that affective personality characteristics should tend to be more essentialized than others because, according to folk psychology, emotions are intimately linked to the person’s biology (D’Andrade, 1987). As a result of this embodiment, emotion-related personality characteristics should be understood in a more naturalized manner than others. This prediction about lay conceptions of personality accords with scientific conceptions of temperament, which is normally understood to be composed of emotional traits that are substantially heritable and biologically based (Clark & Watson, 1999). Our second and third hypotheses derive from the recent work of Leyens et al. (2000, 2001), who argued that people selectively attribute a distinctively human essence or nature to themselves and their ingroup. By implication, personality characteristics should be essentialized if they are understood to be aspects of human nature. Because human nature is a normative concept, representing valued and in principle, widely shared human attributes, personality characteristics that are understood as aspects of it should be relatively desirable and prevalent. Thus, positive (Hypothesis 2) and prevalent
(Hypothesis 3) characteristics should be essentialized more than others because they are more likely to be seen as elements of human nature.
Neither of these hypotheses is self-evident. Ethnic and sexual minorities—groups both devalued and of low prevalence—have been the focus of much past research on essentialist beliefs. It has often been argued that essentialist beliefs play an important role in prejudice (e.g., Allport, 1954; Rothbart & Taylor, 1992), implying that they are associated with the attribution of undesirable characteristics; and one study finds that more essentialized social groups tend to have lower social status (Haslam et al., 2000). Nevertheless, the logic of the work of Leyens et al. (2000, 2001) is that aspects of normative human nature—of which personality characteristics are more likely candidates than particular ethnic or sexual identities—should tend to be essentialized.
Method Participants. Seventy-three undergraduate psychology students (58 women, 15 men), mean age 21.5 years (SD = 1.6), participated in the study as part of a laboratory session. Materials. All participants completed a questionnaire in which they rated 80 personality descriptors. The descriptors were systematically selected to yield a broad and evaluatively diverse sample based on an inclusive understanding of personality that extends beyond standard trait models (see Table 1). Forty terms were sampled from adjectival markers of the Big Five trait dimensions developed by John and Srivastava (1999), taking 4 descriptors from each pole of every dimension. Twenty terms were sampled from Schwartz’s (1992) value taxonomy, taking 2 from each value segment. Ten terms were derived from the Diagnostic and Statistical Manual of Mental Disorders (4th ed.; American Psychiatric Association, 1994) personality disorders, taking 1 emblematic term for each disorder. Finally, 10 negative terms were sampled from Benet-Martínez and Waller’s (2002) fivedimensional model of highly evaluative terms. Participants rated the personality descriptors on subsets of nine items. Six of these items assessed essentialist beliefs and were based on the Essentialist Beliefs Scale developed by Haslam et al. (2000, 2002). That scale’s original “uniformity” item was modified to refer to the cross-situational consistency of behavior rather than the similarity of category members because both refer to forms of homogeneity. All items were rated on 7-point Likert-type scales (1 = strongly disagree and 7 = strongly agree) with the exception of biological basis, and the items were written as follows: Discreteness: “People either have this characteristic or they do not: those who have it are a distinct type of person” Biological basis: “To what extent is this characteristic based on the person’s biological or genetic make-up? Write one of the following percentages in the space provided” (percentages from 0 to 100 in increments of 10) Immutability: “It is easy to change this characteristic: it is not a fixed attribute of the person” (reverse scored) Informativeness: “This characteristic has broad ramifications: it influences people’s behavior in a wide variety of situations and in many aspects of their lives” Consistency: “People who have this characteristic will tend to display it in a consistent manner, showing it in different situations and with different people” Inherence: “This characteristic is a deeply-rooted part of the personality: it lies deep within the person and underlies the person’s behavior” Three additional items were written to assess variables hypothesized to be correlates of essentialist beliefs about personality. These items assessed the social desirability, population prevalence, and affectivity of personality characteristics: Desirability: “How desirable or positive is this personality characteristic?” (1 = extremely undesirable and 7 = extremely desirable)
Prevalence: “What percentage of the general population could reasonably be described as having this characteristic? (please write a number in the space provided)” Affectivity: “Some personality characteristics are primarily emotional (i.e., about emotions, moods and feelings) and some are primarily cognitive (i.e., about beliefs and ways of thinking). Rate the characteristics on the extent to which they are primarily emotional or cognitive” (1 = primarily emotional and 7 = primarily cognitive; reverse scored) Each participant completed a three-page questionnaire containing a cover sheet and two rating pages, each listing the 80 personality descriptors and with one item printed at the top. Design and procedure. Each participant’s questionnaire contained a randomly assigned pair of the nine items, requiring a total of 160 ratings. The order of items in each pair was also randomized. The 80 personality descriptors were presented in a standard random order or its exact reverse, and these two orders were also randomly assigned within each questionnaire. Consequently, each item was equally likely to be rated by each participant, in either first or second position, with either of two descriptor orderings. Participants completed the questionnaire in a classroom setting, in three groups of 18 to 29. Most finished within 20 minutes, after which they were extensively debriefed.
Results
The focus of the study was on shared essentialist beliefs about personality and on differences among the 80 personality descriptors rather than among participants, so ratings of the descriptors were aggregated across the subsample of participants (mean n = 16.2) who rated them on each item. Consequently, mean ratings of each descriptor on each of the nine items served as the basis for the data analyses. These aggregated ratings are meaningful to the extent that participants agree in their ratings. If agreement is adequate, a testable requirement, then the aggregated ratings yield higher levels of reliability than the individual ratings on which they are based. Aggregation is particularly important where ratings are apt to be somewhat difficult to make, as in the sometimes complicated conceptual judgments (i.e., treating each rater as akin to a test item), are presented in Table 2. Reliabilities were generally moderate to very strong (mean α = 0.77), with ratings of immutability alone being weak (α = 0.40). Table 2 also presents mean aggregated ratings of the 80 descriptors. Mean ratings on the Likert-scaled items fall close to the scale midpoint of 4. “Biological makeup” was attributed a substantial role in the personality characteristics (mean 36.2%), which were generally judged to apply to a large minority of the population (mean 42.9%). One-way ANOVAs with post hoc Scheffé comparisons (Bonferroni corrected) indicated that the descriptor types differed on four of the six essentialist belief items. In every case, the highly evaluative terms were essentialized less than the trait, value, and personality disorder descriptors. Predictably, the highly evaluative and personality disorder descriptors were judged to be more deviant—infrequent and undesirable—than the value and trait descriptors. Intercorrelations among the nine items across the 80 descriptors are presented in Table 3, which indicates that the essentialist belief items are intercorrelated as hypothesized. Ten of the 15 correlations are statistically significant with a mean value of 0.34, which rises to 0.49 when correlations are disattenuated for unreliability. A principal components analysis indicated that the six essentialist belief items composed a single factor (48.9% explained variance) based on the scree test (eigenvalues 2.94, 1.34, 0.75, 0.39, 0.37, 0.22). All items loaded substantially on the factor (> 0.42) with the exception of immutability (0.21), a finding that might be explained by the item’s unreliable assessment. The unifactorial structure was strongly supported by an iterative factor analysis (eigenvalues 2.68, 0.95, 0.25, 0.10, 0.03, 0.00). This structure also held when analyses were restricted to the 40 Big Five traits. Factor scores of the 80 descriptors from the first principal component were therefore employed as an essentialism index, which summarizes the extent to which the descriptors were essentialized on the six items. Scores of the descriptors on this index, averaged across Studies 1 and 2, are presented inTable 1. The essentialism index was used to test the study hypotheses. Consistent with prediction, more desirable personality characteristics were essentialized more, r = 0.54, p < .001, as were characteristics rated as more prevalent, r = 0.50, p < .001. Contrary to prediction, however, more emotion-based characteristics were not more essentialized, r = 0.10, ns, although Table 3 indicates that they were judged to be more biologically based and inherent. Inspection of Table 3 also shows that desirability, prevalence, and affectivity are intercorrelated, rendering univariate tests potentially unreliable. Consequently, a simultaneous multiple regression analysis was conducted with the three variables serving as independent variables. As a set, these items powerfully predicted the essentialism index, F(3,76) = 14.1, p < .0001, R = 0.60, and each item had an independent effect (desirability: β = 0.39, p = .003; prevalence: β = 0.26, p = .044; affectivity: β = 0.21, p = .031). Thus a multivariate analysis supports all three hypotheses.
Discussion
The findings of Study 1 offer substantial support for its hypotheses. The essentialist belief items covaried strongly and in a unifactorial manner, indicating that people’s thinking about personality reflects a coherent essentialist ontology that is systematically applied more to some characteristics than others. Thus, a set of beliefs that had been found to covary in studies of social categories also covaried for characteristics that are usually seen as personal rather than social, as dimensions rather than categories, and as adjectives rather than nouns. Simply stated, personality characteristics understood to be biologically based, discrete, immutable, informative, consistent, or deeply rooted also tend to be understood in all of the other ways as well. In addition, the extent to which personality characteristics were essentialized was predictable from their desirability, prevalence, and emotional basis. The last effect was relatively weak and did not emerge on a univariate test, but it implies that affective characteristics are, as D’Andrade (1987) proposed, more directly linked in folk psychology to our underlying, embodied nature than more cognitive characteristics. The desirability and prevalence effects were more robust. Although essentialist beliefs are often invoked to account for beliefs about deviant and devalued minorities, in the personality domain, they appear to be held primarily about valued and widely shared characteristics. The work of Leyens et al. (2000, 2001) on “human nature” offers a way to understand these effects. Almost by definition, our fundamental nature as human beings should be revealed in what we tend to have in common and in what we value. By this account, desirable and prevalent characteristics are especially essentialized because they constitute aspects of human nature. Leyens et al. (2000) listed intelligence and language as candidates for distinctively human essences, and it is interesting that the 2nd and 3rdmost essentialized descriptors in Study 1 were intelligent and talkative. Several other highly essentialized descriptors—independent (1st), creative (4th), imaginative (6th), ambitious (7th)—also arguably represent distinctively human attributes of autonomy, originality, transcendence of the senses, andfuture-orientedstriving.1
STUDY 2
Study 1 demonstrates that people distinguish among personality characteristics in a coherent essentialist fashion and supports three hypotheses about factors that may contribute to the greater essentializing of particular characteristics. However, it remains to be seen whether essentializing personality has a bearing on social cognition and behavior. If a coherent essentialist understanding is held for some characteristics more than others, but this understanding has no further implications, then Study 1’s findings have limited relevance. One plausible general hypothesis about the implications of essentialist beliefs is that more essentialized personality characteristics should be particularly important bases for social inference and judgment. This hypothesis can be justified on several grounds. First, highly essentialized characteristics should be understood as informative and, hence, valuable for making wide-ranging inferences about behavior. Second, they should be viewed as homogeneous (consistent) in their manifestations and, hence, likely to yield relatively accurate behavioral predictions across situations. Third, the perceived immutability of essentialized characteristics should make them reliable sources of social inferences over time. Fourth, their inherence and discreteness should make highly essentialized characteristics seem to capture fundamental and defining aspects of people. For all of these reasons, more essentialized personality characteristics might be understood as particularly fundamental—high in generality, significance, and reliability. If essentialized personality characteristics are especially important for social cognition and behavior, this importance could be manifest in several ways. First, essentialized characteristics might be viewed as particularly central to personal identity, judged to be core aspects of the person that define who he or she is. Allport (1937) raised this issue of trait centrality to refer to the fact that particular traits are especially prominent within persons. In this spirit, Trafimow and Finlay (2001) assessed trait importance by having people rate how important certain traits were for how they thought about themselves. Essentialized personality characteristics might be central in a more general way, seen as especially salient and identity-determining for whoever possesses them. Asecond, more interpersonal manifestation of a characteristic’s importance is its perceived utility for judging other people. More “relationally important” characteristics, in this sense, are those that people believe to be most useful or valuable to know in evaluating potential inter- actants. Relational importance as defined here resembles “trait centrality” as employed by Asch (1946) in his pioneering research on impression formation. More central traits, in his sense, are those that exert a relatively potent effect on impressions, and we might expect that these are the traits that social perceivers would want to know about in judging an unfamiliar person. If essentialized personality characteristics are high in relational importance, then people would be especially keen to find evidence pertaining to them when making such judgments.
A third form of importance involves communication. The pragmatics of communication suggest that people should normally endeavor to communicate the most relevant and informative material to their interlocutors (Grice, 1975). By implication, if they are describing a person to someone else, people should select or emphasize those personal attributes that are most vital to know. If essentialized personality characteristics are particularly important in this sense, then they should be more communicable (Schaller, Conway, & Tanchuk, 2002). Study 2 was designed to test the broad hypothesis that more essentialized personality characteristics are judged to be more important. Importance was assessed as centrality, relational importance, and communicability, as defined above. All three subsidiary hypotheses were also expected to hold when other factors that might influence social importance were statistically controlled. For instance, a personality characteristic’s importance for some purposes might in part be a function of its desirability or of its prevalence, as examined in Study 1. Communicators might tend to mention a person’s more positive attributes out of politeness, whereas undesirable characteristics might be considered particularly important in forming impressions (Hamilton & Zanna, 1972). Characteristics that are central to identity might be those that are particularly distinctive (i.e., low prevalence; Blanton & Christie, 2003). Thus, we predicted that essentialized characteristics would be judged to have high
importance independent of their social desirability and prevalence. Finally, Study 2 aimed to replicate the structure of essentialist beliefs about personality, employing the same items and characteristics as in Study 1.
Method
Participants. Eighty-one undergraduate psychology students (64 women, 17 men), mean age 22.3 years (SD = 4.2), participated in the study as part of a laboratory session. Materials. Participants rated the 80 personality descriptors from Study 1 on questionnaires that closely resembled those completed in that study. The same six essentialist belief items were employed, but participants also rated three new items assessing the importance of the personality characteristics. Importance was operationalized in three distinct ways: a characteristic’s centrality to the identity of the person who possesses it, its relational importance (i.e., the extent to which it would be desirable for judging a prospective interactant), and its communicability when describing someone who had it to another person. These items were written as follows: Centrality: “This characteristic is a central aspect of a person’s personality: if you have it, it defines who you are” (1 = strongly disagree and 7 = strongly agree) Relational importance: “Imagine you are about to meet someone new who you will have to interact with for a substantial period of time. How important would it be for you to know that the person has this characteristic?” (1 = not at all important and 5 = extremely important) Communicability: “If you knew that a person had this characteristic, how likely would you be to mention it when describing the person to someone else?” (1 = very unlikely and 7 = very likely) Design and procedure. The questionnaires were constructed according to the same design as in Study 1. Again participants rated a randomly assigned and ordered pair of the nine items, with two possible reversed orderings of the 80 personality descriptors. Participants completed the questionnaires in a classroom setting, in four groups of 15 to 27 participants, followed by debriefing.
Results
As in Study 1, ratings were aggregated across the subsamples of participants who rated each item (mean n = 18.0). Reliabilities and mean ratings across the 80 personality descriptors are presented in Table 4. All reliabilities were satisfactory (mean α = 0.79), indicating that the aggregate ratings offer reliable estimates of shared beliefs about the personality descriptors. The reliability of immutability ratings was notably better than in Study 1. As in Study 1, one-way ANOVAs with post hoc Scheffé comparisons (Bonferroni corrected) indicated that the highly evaluative personality descriptors scored lower than the other descriptor types on four of the six essentialist belief items. The same pattern held for the centrality and communicability items. Correlations among the mean ratings across the 80 personality descriptors are presented in Table 5, which again reveals substantial intercorrelations among the six essentialist belief items (mean r = 0.50). Immutability correlated much more strongly with the other essentialist belief items than in Study 1, probably because of its improved reliability. A principal components analysis indicated that a single factor accounted for the 6 items (59.1% explained variance), according to both the Kaiser criterion and the scree test (eigenvalues 3.54, 0.95, 0.69, 0.46, 0.21, 0.16). Every item loaded powerfully (> 0.59) on this factor. To pursue the structure of essentialist beliefs about personality further, a confirmatory factor analysis compared the adequacy of a onefactor model with models containing two correlated or orthogonal factors corresponding to the natural kind and entitativity factors obtained previously (Haslam et al., 2000). The one-factor model (χ2/df = 5.52, root mean square error of approximation [RMSEA] = .239) fit substantially and significantly better than the model containing two orthogonal factors (χ2/df = 10.93, RMSEA = .355) and equally well as the model containing two correlated factors (χ2/df = 5.44, RMSEA = .237). As fit values for the two better fitting models did not differ significantly, parsimony supports the one-factor model, especially as the two factors were so highly correlated (r = 0.83) as to question their distinctness. Thus, scores on the first principal component were again employed as an essentialism index. This index correlated 0.91 with the Study 1 index, demonstrating that the component is highly replicable and yields a robust assessment of essentialist beliefs. Similarly, all of the essentialist belief items correlated very highly with the corresponding items in Study 1 (rs ranged from 0.74 to 0.85), with the exception of immutability (r = 0.39). Study 2’s additional hypotheses predict that more essentialized personality characteristics should be more important, as operationalized by the centrality, relational importance, and communicability items. These items correlated well across the sample of personality descriptors (mean r = 0.47), consistent with them all representing forms of importance. Univariate tests of the hypothesized associations were uniformly supportive. The essentialism index correlated 0.82 (p < .0001) with centrality, 0.27 (p = .014) with relational importance, and 0.69 (p < .0001) with communicability. Thus, more essentialized personality characteristics were consistently judged to be relatively important. The univariate associations between the importance items and the essentialism index might be complicated by other variables not measured in Study 2. Indeed, the desirability and prevalence ratings derived from Study 1 were strongly correlated with centrality (rs = .51 and .51, ps < .001) and communicability (rs = .53 and .42, ps < .001). To determine whether desirability and prevalence might account for the correlations between the essentialism index and the importance items, several simultaneous multiple regression analyses were conducted. In these analyses, the importance items were dependent variables and the essentialism index (from Study 2), and desirability and prevalence (from Study 1) were the independent variables. In each case, the association between essentialist beliefs and importance remained significant after desirability and prevalence were statistically controlled. Standardized beta weights for the essentialism index in the centrality, relational importance, and communicability analyses were 0.74 (p < .0001), 0.31 (p = .018), and 0.56 (p < .0001), respectively. More important
personality characteristics, therefore, appear to be more essentialized, independent of their evaluative charge and judged prevalence. Desirability independently predicted relational importance (β = –0.32, p = .044) and communicability (β = 0.24, p = .044) in these analyses, implying that people are more concerned to know about undesirable characteristics of prospective interactants and more inclined to mention desirable characteristics when describing someone to a third party.
Discussion
The findings of Study 2 clearly support the broad hypothesis that essentialized personality characteristics are judged to be particularly important and central attributes of persons. Their relatively high importance generalized across intrapersonal (centrality) and interpersonal (relational importance and communicability) domains and was not reducible to established determinants of importance, such as desirability and distinctiveness (i.e., low prevalence). Essentialist beliefs, therefore, may represent a previously unexamined determinant of the importance of personality characteristics, which may be relevant to researchers on identity, impression formation, and interpersonal communication. In short, part of what makes some attributes particularly influential in
these domains may be the ontological assumptions that people make about them. Just as people judge trait adjectives to imply greater behavioral stability and consistency than action verbs (Semin & Fiedler, 1988), they also judge some adjectives to refer to more fundamental and essence-like attributes than others.
Although it is plausible that the extent to which personality attributes are essentialized plays a causal role in determining their importance and, hence, their role in social cognition and identity, the causal arrow might point in the opposite direction. More important personality attributes, or attributes that are most often used in social judgment and communication, might come to be more essentialized. Kashima (2004), for example, demonstrated how communication can increase the perceived entitativity and essentialism of social groups. Consequently, the findings of this study, such as the association between essentialism and communicability, cannot be confidently interpreted as revealing a causal role for essentialist beliefs in this domain.
GENERAL DISCUSSION
The present studies strongly suggest that a coherent set of essentialist beliefs are held about personality characteristics, just as previous research finds them to be held about a variety of social categories. The studies, therefore, help to extend the range of domains in which psychological essentialism applies. In doing so, they join a small quantity of work that primarily examines children’s understandings of traits (e.g., Giles, 2003; Yuill & Pearson, 1998), focuses on single elements of essentialist beliefs such as inherence (e.g., Semin&Krahé, 1987), or addresses individual differences in essence-related beliefs about personality as a whole rather than differences among personality characteristics (e.g., Levy et al., 1998). Moreover, the studies demonstrate that essentialist beliefs about personality are not only held, but are also held preferentially for particular kinds of characteristics and have several social-cognitive implications. Just as academic trait theory presents a somewhat essentialist view of traits, so do implicit personality theories apply essentialist assumptions, as Gelman (1992) proposed. Although the two studies provide good support for the coherence of essentialist beliefs about personality, the unifactorial structure of these beliefs differs from the two-dimensional structure obtained in previous work on social categories (Haslam et al., 2000, 2002). Different belief structures may hold in the personality and social category domains, such that the distinct entitativity and natural kind dimensions obtained in the latter are somehow fused in the former. Personality characteristics that are “naturalized” (i.e., understood to be biologically based, fixed, and discrete) also appear to be understood as underlying entities (i.e., inherent, informative, and consistent), whereas these two understandings are largely independent for social categories. Reasons for this apparent divergence are unclear, but it supports the view that essentialist beliefs may be differently organized in different domains. One interpretation of the tendency for people to essentialize some personality characteristics more than others is that they treat some characteristics as more “real” than others. If this is the case, and a realist view of personality characteristics (Yuill, 1992) is applied selectively, then it is little surprise that characteristics judged to be more real are considered more important. Social perceivers and actors are unlikely to base their identities, judgments, or communications on characteristics that they believe to be superficial, flimsy, and subjective. However, we would argue that essentializing a personality characteristic goes beyond adopting a realist view of it. Characteristics could be seen as real underlying causes of behaviors without also being conceptualized as biological, immutable, or discrete. The fact that these beliefs cohere with those that more directly represent the realist view (i.e., inherence, consistency) indicates that essentializing a personality characteristic is not simply treating it as real. Instead, it amounts to treating the characteristic as a real essence. It does seem to be the case that some kinds of personality characteristics are essentialized substantially less than others. Notably, the highly evaluative personality characteristics (Benet-Martínez&Waller, 2002) received very low ratings on the essentialist belief items in both studies. Although undesirable and relatively rare characteristics were generally essentialized less than others, the highly evaluative terms were still essentialized less than the personality disorder descriptors, despite similar perceived deviance. One interpretation of this finding is that the highly evaluative personality terms were judged to be more in the (jaundiced) eye of the beholder than in the person beheld. Thus, these terms, most of which are somewhat abusive, may be seen less as inhering, enduring, consistent, and biologically based features of persons and more as context-specific assessments made by external observers. Although highly evaluative terms may represent important but neglected forms of personality description (Benet-Martínez & Waller, 2002), people appear to perceive them to be quite different in nature from other personality attributes.
Evaluative Implications of Essentialist Beliefs Research and theory have tended to emphasize the dark side of essentialist thinking about difference.
Allport (1954) drew attention to the link between beliefs in a group essence and prejudice, and Rothbart and Taylor (1992) argued that viewing social categories as natural kinds accentuates group differences. Yzerbyt et al. (1997) argued that essentialist beliefs rationalize unjust social arrangements, Levy et al. (1998) demonstrated that “entity theorists” are especially apt to endorse stereotypes, and Haslam et al. (2000) found that devalued social groups tend to be more essentialized than others. Our findings contribute to a growing awareness that essentializing human differences is not invariably negative in its implications. Leyens et al. (2000, 2001) demonstrated that a distinctively human essence is preferentially attributed to favored ingroups, and Castano (2004) showed that people prefer to belong to essentialized ingroups. Haslam et al. (2002) found that some essentialist beliefs were associated with more pro-gay attitudes, and Verkuyten (2003) showed how essentialist discourse can have progressive aspects for members of ethnic minorities. The present study, therefore, adds to a body of work that shows essentialist beliefs to have complicated implications for understanding difference. One possible way to resolve these complexities is to propose that essentialist beliefs have different dynamics and determinants in different domains. It is clear that many stigmatized social categories tend to be essentialized and likely that essentialist beliefs contribute to their devaluation by rationalizing their low status and accentuating their distinctiveness. It now seems equally clear that personality characteristics that are valued and normative tend to be essentialized, and that essentialized characteristics are perceived to be socially important. These conclusions are consistent with Dunning’s (1995) finding that people self-enhance most on traits that they believe to be stable and important. This rather stark difference in the implications of essentialist beliefs may hinge on the kind of nature that is being ascribed to social categories as distinct from personal attributes. Personality characteristics appear to be essentialized to the extent that they partake of human nature. All people can be seen as intelligent or imaginative, at least to some degree. Most social categories, in
contrast, cannot easily be understood as aspects of an encompassing human nature. They are, instead, ways in which some humans are perceived to be categorically different from others. Being female or White cannot be a core aspect of what it is to be human in a world where many people are neither, so whatever nature members of each category share cannot be a broadly humannature. Where social categories are concerned, then, essentialist beliefs may tend to impute a nature that deepens and legitimates a social division and, therefore, have largely negative and exclusive implications. Where personality characteristics are concerned, however, essentialist beliefs may tend to impute a human nature that is positive and inclusive in its implications.
Essentialized personality characteristics are valued foundations for social identity and judgment. Essentialized social categories, in contrast, are often devalued forms of social distinction. Some limitations of the present studies must also be acknowledged. Although the hypothesized associations between essentializing and social desirability, prevalence, centrality, social importance, and communicability were all large, they are likely to overestimate the associations implicitly perceived by individual participants. Aggregated ratings capture the shared variance in participants’ beliefs about personality characteristics—beliefs that do indeed appear to be reasonably shared and coherent—but do not allow the inference that all participants held the same beliefs with equal strength or a comparable structure. Individual differences in this domain surely exist, and the work of Dweck and colleagues (e.g., Levy et al., 1998) demonstrates their importance. Our studies complement this work by focusing on shared beliefs and on differences among personality characteristics, and they also suggest that individual differences in essentialist beliefs as a whole might well be investigated, and not just beliefs about immutability.
Another limitation concerns our exclusive use of selfratings. Although this is probably unavoidable for assessing essentialist beliefs and simple judgments about personality characteristics (i.e., their desirability, prevalence, and affectivity), the ratings of centrality, relational importance, and communicability in Study 2 may not adequately capture actual tendencies to employ the personality characteristics in identity construction, impression formation, and communication. Future research should examine whether essentialized personality characteristics are preferentially employed when these activities are assessed in more direct and behavioral ways. Ideally, also, future research would develop experimental methods for manipulating essentialist beliefs rather than rely on correlational methods. Despite these limitations, our studies make several potentially important contributions. First, they establish that implicit personality theories make essentialist assumptions about at least some personality characteristics, and that these assumptions have a similar structure to those that are held about some social categories. Second, they indicate that essentialist beliefs are most strongly held about normative personality characteristics, so that positive characteristics are anchored deeply within human nature. By implication, negative characteristics are seen as relatively superficial, inconsistent, and unreal. Finally, our findings imply that essentialist beliefs represent a new, but potentially powerful, determinant of the importance that personality characteristics have in social cognition and communication.