THE COGNITIVE STRUCTURE OF STEREOTYPES

The Cognitive Structure of Stereotypes Running Head: THE COGNITIVE STRUCTURE OF STEREOTYPES

Investigating the Cognitive Structure of Stereotypes: Generic Beliefs about Groups Predict Social Judgments Better than Statistical Beliefs

Matthew D. Hammond* Victoria University of Wellington

Andrei Cimpian* New York University

* Corresponding Authors: [email protected] or [email protected]

This is a prepublication author copy. Please reference: Hammond, M. D., & Cimpian, A. (in press). Investigating the cognitive structure of stereotypes: Generic beliefs about groups predict social judgments better than statistical beliefs. Journal of Experimental Psychology: General. DOI: 10.1037/xge0000297

1

The Cognitive Structure of Stereotypes

2

Acknowledgements The authors acknowledge funding from the National Science Foundation for supporting this work (grant BCS-1530669 awarded to Cimpian) and thank Sapna Cheryan, Andrew Christy, Dov Cohen, Sarah-Jane Leslie, David Miller, Danny Osborne, Becca Schlegel, Susanna Stone, and the University of Illinois Cognitive Development Lab for their helpful feedback as we prepared this manuscript. This research was presented as a poster at the 2017 Convention of the Society for Personality and Social Psychology.


3

Abstract Stereotypes are typically defined as beliefs about groups, but this definition is underspecified. Beliefs about groups can be generic or statistical. Generic beliefs attribute features to entire groups (e.g., men are strong), whereas statistical beliefs encode the perceived prevalence of features (e.g., how common it is for men to be strong). The present research sought to determine whether generic or statistical beliefs are more central to the cognitive structure of stereotypes. Specifically, we tested whether generic or statistical beliefs are more influential in people’s social judgments, on the assumption that greater functional importance indicates greater centrality in stereotype structure. Relative to statistical beliefs, generic beliefs about social groups were significantly stronger predictors of expectations (Studies 1–3) and explanations (Study 4) for unfamiliar individuals’ traits. Additionally, consistent with prior evidence that generic beliefs are cognitively simpler than statistical beliefs, generic beliefs were particularly predictive of social judgments for participants with more intuitive (vs. analytic) cognitive styles and for participants higher (vs. lower) in authoritarianism, who tend to view outgroups in simplistic, all-or-none terms. The present studies suggest that generic beliefs about groups are more central than statistical beliefs to the cognitive structure of stereotypes.

Keywords: Stereotypes, Generic Beliefs, Statistical Beliefs, Social Judgment


4

Investigating the Cognitive Structure of Stereotypes: Generic Beliefs about Groups Predict Social Judgments Better than Statistical Beliefs What is the cognitive structure of stereotypes? Although social psychologists largely agree that stereotypes are individuals’ beliefs or generalizations about groups1 (e.g., Allport, 1954; Ashmore & Del Boca, 1981; Brewer, 1988; Devine, 1989; Fiske, Cuddy, Glick, & Xu, 2002; Judd & Park, 1993; Jussim, Crawford, & Rubinstein, 2015), this definition is underspecified. Cognitive science distinguishes between two types of beliefs about groups: generic and statistical (see Carlson & Pelletier, 1995; Cimpian, 2016; Gelman, 2004; Leslie, 2008; Prasada, 2000). Generic beliefs apply a feature to a group as a whole (e.g., men are strong), whereas statistical beliefs encode the perceived prevalence of a feature among the members of the group (e.g., how common it is for men to be strong). Are generic or statistical beliefs more central to the cognitive structure of stereotypes? We investigated this question by using a functional criterion: The type of belief that is central to stereotype structure should be the one that people use in their social judgments, such as expecting an individual to have certain traits based on their group membership. Investigating whether generic beliefs or statistical beliefs are more central to stereotype structure is important because these beliefs have distinct cognitive properties. Most prominently, as we detail below, generic beliefs are less contingent on evidence than statistical beliefs are (Bian & Cimpian, in press; Cimpian, Gelman, & Brandone, 2010; Leslie, 2008). Thus, a clearer understanding of the relative weight of these beliefs in the structure of people’s stereotypes will inform the current debate over whether these stereotypes are accurate representations of group characteristics (Jussim, 2012, in press;

Although some theorists define stereotypes as “exaggerated beliefs” about groups (Allport, 1954, p. 191), here we use a definition with no a priori assumptions about accuracy (see Jussim, 2012). We also focus on individual-level stereotypical beliefs rather than stereotypes that exist at the societal level (Judd & Park, 1993), and on explicit rather than implicit stereotypes (e.g., Devine, 1989; Greenwald & Banaji, 1995). 1


5

Jussim et al., 2015): Stereotype accuracy is possible if statistical beliefs are central to stereotype structure, but in doubt otherwise. Generic vs. Statistical Beliefs and Their Relation to Stereotypes The distinction between generic and statistical beliefs has been investigated extensively in cognitive psychology (e.g., Gelman, 2004; Gelman et al., 1998; Hampton, 2012; Khemlani, Leslie, & Glucksberg, 2012; Prasada, 2000), linguistics (e.g., Behrens, 2005; Carlson & Pelletier, 1995), and philosophy (e.g., Leslie, 2008; Sterken, 2015). Generic beliefs map features onto groups holistically, as if each group were an indivisible unit.2 These beliefs do not encode information about the prevalence of features within groups, like statistical beliefs do. Because the truth of generic beliefs is unmoored from statistical facts, generic beliefs are sometimes endorsed on the basis of minimal evidence or, conversely, rejected when there is substantial evidence. For example, people tend to endorse the generic statement that “sharks attack swimmers,” despite the rarity of shark attacks (Leslie, Khemlani, & Glucksberg, 2011; Prasada, Khemlani, Leslie, & Glucksberg, 2013). In contrast, people often disagree with the generic statement that “sharks are female” (Prasada et al., 2013), despite that ~50% of category members are female. This same discrepancy between generic beliefs and reality is observed for social groups (e.g., “African Americans are athletic” tends to be endorsed but not “African Americans are right-handed”; Gelman, Taylor, & Nguyen, 2004; Leslie, in press; Tasimi, Gelman, Cimpian, & Knobe, in press). High prevalence is neither necessary nor sufficient for endorsement of generic beliefs, even though—other things being equal—people are more likely to endorse generic beliefs about prevalent features than rare features (Prasada et al., 2013).

2

Note that believing all group members possess a feature is a statistical belief. Generic generalizations are cognitively distinct from universally-quantified generalizations (e.g., all, every) or other broad-scope quantified generalizations (e.g., most, typically; see Carlson & Pelletier, 1995; Gelman, 2004; Leslie, 2008).


6

Statistical beliefs encode only information about the perceived prevalence of traits. Accordingly, their endorsement is more closely tied to the relevant trait distributions. For example, whether statements such as “over half of African Americans are athletic” or “African Americans are typically athletic” are endorsed is entirely dependent on how common athleticism is perceived to be in this group. This information is both necessary and sufficient for endorsement of a statistical belief. Here, we test the relative weight of generic and statistical beliefs about groups in people’s stereotypes. Prior evidence suggests that generic beliefs are better retained in memory (Cimpian, 2016; Cimpian & Erickson, 2012; Leslie & Gelman, 2012) and easier to reason with (Hampton, 2012; Leslie et al., 2011; Leslie & Gelman, 2012) than statistical beliefs. Accordingly, we predict that people will rely on generic beliefs more than statistical beliefs in their social judgments, which—given our functional criterion—would signify that generic beliefs are more central to the structure of stereotypes. Moreover, if generic beliefs are a relatively low-effort way of approaching the social world, then we predict that generic beliefs should be relatively more privileged in stereotyped judgments for individuals who rely on low-effort, “all-or-none” reasoning. Evidence for the primacy of generic over statistical beliefs in stereotypes would bear on theories of stereotype content, including the current debate concerning whether stereotypes accurately represent reality (see Jussim, 2012). Given that people’s estimates of the prevalence of various group traits are close to objective benchmarks, some argue that stereotypes are accurate (e.g., Jussim, 2012, in press; Jussim et al., 2015). This argument, however, assumes that people’s stereotypes are statistical, encoding prevalence. However, if statistical beliefs are less central to the cognitive structure of stereotypes than generic beliefs (whose relationship to the evidence is more tenuous), then uncertainty remains about the accuracy of stereotypes.


7

The Present Studies Four studies tested our hypothesis that generic beliefs about stereotypical group/trait pairings (e.g., men/strong) are stronger predictors of participants’ social judgments than their statistical beliefs are. We conducted a broad test this hypothesis. First, we assessed two types of social judgments that have critical downstream consequences for social evaluations and interactions: participants’ expectations (Studies 1–3) and explanations (Study 4) for the traits of unfamiliar others (see Brewer, 1988; Dar-Nimrod & Heine, 2011; Devine, 1989; Fiske et al., 2002; Judd, James-Hawkins, Yzerbyt, & Kashima, 2005; Keller, 2005). Second, we compared social and non-social stimuli to test whether the primacy of generic beliefs in people’s judgments is a uniquely social phenomenon (Study 3). Finally, we explored individual-difference moderators (Studies 2–4). We expected that social judgments would rely more on generic beliefs, and less on statistical beliefs, for participants who (a) think intuitively rather than analytically (Frederick, 2005; Kahneman, 2011), or (b) are higher in authoritarianism and thus reason about outgroups in simplistic, all-or-none terms (e.g., Altemeyer, 1996; Sidanius & Pratto, 1999). STUDY 1 Method Data and analytic syntax are available on Open Science Framework: https://osf.io/s6zab/. Detailed information about measures, results, and excluded participants is provided in the Online Supplementary Materials (OSM). Participants Ninety-seven individuals (55% women), aged 20–83 (M = 36.54, SD = 13.18), were recruited via Amazon’s Mechanical Turk (MTurk; see Buhrmester, Kwang, & Gosling, 2011) for $0.90. Eligibility required living in the US and holding a >65% successful completion rate on MTurk. All studies included an attention-check: estimating the percentage of


8

Caucasians who are humans. Data totaled 2,917 ratings, exceeding power recommendations for multilevel models (Hox, 1998; West, Ryu, Kwok, & Cham, 2011). Materials and Procedure We identified 27 common stereotypes from prior research (see Table 1 and OSM for details) and assessed generic and statistical beliefs following standard practice (Cimpian, Brandone, & Gelman, 2010; Khemlani et al., 2012). Participants rated each stereotype on each measure below. Orders of measures and stereotypes were randomized, and participants completed a 1-min distracter task between measures. Generic Beliefs. Participants rated their agreement with the generic form of each stereotype (e.g., “Doctors are intelligent”; −3 = Strongly Disagree to 3 = Strongly Agree). Statistical Beliefs. Participants estimated a percentage for each stereotype (e.g., “What percentage of doctors are intelligent?”; 0–100 sliding scale).3 Social Judgments—Expectations. Participants rated the likelihood that an unfamiliar group member had each stereotypical trait (e.g., “Suppose that Person Y is a doctor. Is Person Y intelligent?”; −3 = Very Unlikely, 3 = Very Likely). Results Analytic Strategy In all studies, we tested the relative contribution of statistical beliefs and generic beliefs to social judgments by entering them simultaneously in multilevel models. Models accounted for the interdependence of each participant’s ratings across items and each item’s ratings across participants. We used Bayesian estimation with diffuse priors in Mplus 7.2 (Muthén & Muthén, 1998-2012), cross-classifying participants with items. (Results were also replicated using maximum-likelihood analyses [see OSM].) The 95% credibility intervals

3

A pilot study (see OSM) investigated a different type of statistical belief—the prevalence of a trait relative to other groups (i.e., distinctiveness). However, the distinctiveness of traits did not predict social judgments beyond generic beliefs and statistical beliefs about absolute prevalence.


9

reported below are distributions that contain the most likely parameter value with 95% probability. Bayesian p values are the smaller proportion of the posterior parameter’s distribution relative to zero. Accordingly, these p values have a maximum of .5 and signal significant evidence for the presence of an effect under a threshold of .025 (for reviews, see Gelman, Carlin, Stern, & Rubin, 2003; Kruschke, Aguinis, & Joo, 2012; Muthén, 2010). Across studies, variables were standardized to account for scale differences. Table 2 presents descriptive statistics and correlations. What Predicts Social Judgments? Both statistical beliefs (β = .315 [.277, .354], p < .001) and generic beliefs (β = .514 [.476, .552], p < .001) predicted unique variance in participants’ expectations about unfamiliar individuals’ traits.4 Supporting our hypothesis, participants’ generic beliefs were more strongly predictive of their expectation judgments than their statistical beliefs were (b = −.199 [−.269, −.129], p < .001). In supplementary analyses, the results reported above held when looking separately at stereotypes about race/ethnicity (12 items), gender (7 items), and professions (8 items; see Table 1), and were not moderated by participants’ tendency to provide socially desirable responses, assessed at the end of the study (see OSM). STUDY 2 In Study 2, we (a) assessed the generalizability of effects by using new stereotypes elicited from an independent sample of participants, (b) measured generic and statistical beliefs using identical response formats, and (c) explored cognitive style and authoritarianism as moderators. Because generic beliefs are lower-effort and more absolute than statistical beliefs, we expected them to be more closely related to social judgments for people with

4

Measuring expectations as likelihoods (i.e., statistical judgments) provides a conservative test of our hypothesis that generic beliefs are more predictive of social judgments than are statistical beliefs.


10

intuitive (vs. analytic) cognitive styles (Frederick, 2005) and for people higher in authoritarianism (Altemeyer, 1996; Sidanius & Pratto, 1999). Method Participants 185 individuals (68% women; ages 18–78, M = 37.13, SD = 13.11) provided 5,499 ratings. The sample size was increased to provide sufficient power for the moderation tests. Procedure and Materials Study 2 matched Study 1 except that (a) participants rated stereotypes generated by an independent MTurk sample (n = 33; see Table 1 and OSM for full details); (b) generic beliefs were assessed with the same sliding-scale format used for statistical beliefs (Strongly Disagree to Strongly Agree); and (3) we assessed two moderators. Moderator—Cognitive Style. An adapted version of the Cognitive Reflection Test (Frederick, 2005; see OSM) was administered. High scores indicated deliberative, analytic thinking. Moderator—Authoritarianism. We averaged responses (α = .88) to short-form scales of right-wing authoritarianism (e.g., “It is always better to trust the judgment of the proper authorities in government…”; 6 items; Altemeyer, 1996) and social dominance orientation (e.g., “Inferior groups should stay in their place”; 6 items; Sidanius & Pratto, 1999). These measures were combined to assess a simplistic, authoritarian orientation toward social groups. (Similar patterns of moderation emerged using separate scales.) Results Analytic Strategy In Studies 2 and 3, we capitalized on a strength of Bayesian analyses by setting informative priors based on estimates from Study 1 (variances multiplied by four; see Muthén, 2010). This approach accumulates evidence rather than repeatedly assuming a null


11

hypothesis. Nonetheless, across studies, the expected results also emerged with diffuse priors (see OSM). What Predicts Social Judgments? Participants’ expectations about individuals’ traits were predicted by both their statistical beliefs (β = .329 [.304, .354], p < .001) and their generic beliefs (β = .458 [.433, .484], p < .001). Again supporting our hypothesis, generic beliefs were significantly more predictive of social judgments (b = −.129 [−.175, −.084], p < .001). Moderators Cognitive style moderated the relationship of statistical beliefs (p = .008), but not generic beliefs (p = .358), with social judgments (see OSM for more detail and additional analyses). As expected, analytic thinkers’ (+1 SD) social judgments (b = .366 [.324, .408], p < .001) were more closely related to their statistical beliefs than were intuitive thinkers’ judgments (–1 SD; b = .296 [.257, .335], p < .001). Authoritarianism moderated the relationship of both generic and statistical beliefs with social judgments (ps = .001). As expected, high-authoritarian (+1 SD) participants’ judgments (b = .521 [.478, .565], p < .001) were more strongly related to their generic beliefs than were low-authoritarian (–1 SD) participants’ judgments (b = .428 [.389, .467], p < .001). Moreover, high-authoritarian participants (b = .283 [.242, .324], p < .001) relied on statistical beliefs less than low-authoritarian participants (b = .379 [.339, .419], p < .001). STUDY 3 In Study 3, we (a) assessed statistical beliefs with a simpler question, and (b) compared social and non-social stimuli. Research on generic beliefs has seldom compared these domains systematically (although see Tasimi et al., in press), so we did not have strong a priori predictions about whether the primacy of generic beliefs would extend to judgments about non-social stimuli.


12

Method Participants 214 individuals (54% women; ages 18–71; M = 35.13, SD = 11.10) provided 5,800 ratings. Procedure and Materials Procedures matched Study 2, except that (a) we simplified the wording for statistical beliefs (“Out of all of the [category members], how many [have feature]?” None to All sliding scale), and (b) we randomly assigned participants to rate social (N = 98) or non-social (N = 96) stimuli. The non-social stimuli (e.g., “barns are red,” “ducks lay eggs”) were adapted from Khemlani et al. (2012; see OSM). Results What Predicts Social and Non-Social Judgments? Social judgments. Both statistical beliefs (β = .330 [.299, .362], p < .001) and generic beliefs (β = .451 [.420, .483], p < .001) predicted participants’ expectations about the traits of individual people. As in Studies 1 and 2, generic beliefs were a significantly stronger predictor of social judgments (b = −.121 [−.176, −.065], p < .001). Non-social judgments. Both statistical beliefs (β = .605 [.574, .636], p < .001) and generic beliefs (β = .322 [.292, .353], p < .001) predicted expectations about the features of individual artifacts and animals. In contrast to social expectations, statistical beliefs were a significantly stronger predictor of participants’ non-social expectations (b = .283 [.227, .338], p < .001). Thus, it seems generic beliefs are particularly influential when making predictions about social (vs. non-social) targets—at least given the items tested here. Moderators Social Judgments. Cognitive style moderated the relationship of generic beliefs (p < .001), but not statistical beliefs (p = .163), with participants’ judgments of social others.


13

Analytic thinkers’ expectations (b = .401 [.359, .444], p < .001) were less reliant on generic beliefs relative to intuitive thinkers’ expectations (b = .516 [.473, .559], p < .001). Similarly, authoritarianism was a significant moderator for generic (p < .001), but not statistical (p = .043), beliefs: The relationship between generic beliefs and expectations of social others was stronger for high-authoritarians (b = .569 [.501, .627], p < .001) than for low-authoritarians (b = .383 [.326, .462], p < .001). Non-social Judgments. Cognitive style moderated the relationship of both generic and statistical beliefs with participants’ non-social judgments (ps < .013). Consistent with prior patterns, analytic thinkers’ expectations about non-social stimuli (b = .251 [.205, .297], p < .001) showed a weaker relationship with generic beliefs than did intuitive thinkers’ expectations (b = .324 [.280, .366], p < .001). Moreover, analytic thinkers’ expectations about non-social stimuli (b = .685 [.642, .729], p < .001) were more strongly related to their statistical beliefs than were intuitive thinkers’ expectations (b = .547 [.504, .590], p < .001). Authoritarianism was not a significant moderator (ps > .06), which was unsurprising given the non-social context. STUDY 4 In Study 4, we extended our predictions to another consequential social judgment: participants’ biological explanations for the stereotypical traits of social others. These explanations often foster ingroup favoritism and outgroup antipathy (Dar-Nimrod & Heine, 2011). Method Participants 177 individuals (67% women; ages 18–75; M = 37.32, SD = 12.80) provided 5,258 ratings. Procedure and Materials


14

Study 4 was identical to Study 2, except that the social judgments were participants’ biological explanations for stereotypical traits (e.g., “Person Y is Asian and is smart. To what extent is Person Y’s being smart due to their biology?”; −3 = Not at All to 3 = A Great Deal). Results What Predicts Social Judgments? Both statistical beliefs (β = .086 [.044, .127], p < .001) and generic beliefs (β = .174 [.133, .215], p < .001) predicted greater endorsement of biological explanations. Again, generic beliefs were a significantly stronger predictor than were statistical beliefs (b = −.088 [−.164, −.013], p = .011). Moderators Cognitive style moderated the relationship of generic beliefs (p = .003), but not statistical beliefs (p = .232), with social judgments. As expected, intuitive thinkers’ endorsement of biological explanations for stereotypical traits (b = .235 [.180, .290], p < .001) was more strongly predicted by their generic beliefs than was analytic thinkers’ endorsement (b = .127 [.071, .183], p < .001). Authoritarianism moderated the relationship of generic beliefs (p < .001), but not statistical beliefs (p = .269), with social judgments. As expected, high-authoritarians’ biological explanations (b = .255 [.201, .309], p < .001) were more closely related to their generic beliefs than were low-authoritarians’ explanations (b = .108 [.053, .161], p < .001). General Discussion We examined the cognitive structure of stereotypes by testing whether their generic or statistical aspect was more predictive of social judgments. Across four studies, generic beliefs were consistently the stronger predictor of people’s expectations and explanations for the traits of social others. Thus, although in principle stereotypes consist of both generic and statistical beliefs about groups, our evidence suggests that generic beliefs are more central to


15

stereotype structure—the more powerful and active component of stereotypes. Generic beliefs may be primary in stereotyping in part because they are cognitively easy (e.g., Cimpian, 2016; Leslie, 2008), consistent with the idea that stereotypes serve to simplify the social world (e.g., Bodenhausen, 1990). Although the patterns of moderation differed across studies, people with more intuitive (vs. analytic) cognitive styles generally relied more on generic beliefs and less on statistical beliefs in their judgments. (Exploratory analyses described in the OSM suggested that generic beliefs were also more influential in older [vs. younger] participants, arguably because information-processing resources decrease with age [e.g., Salthouse, 2015].) The simplistic, one-dimensional quality of generic beliefs is also likely why they were more prominent in the judgments of people with authoritarian tendencies, who tend to be hostile toward outgroups (e.g., Altemeyer, 1996). The heightened prominence of generic beliefs for participants with less-than-rational and authoritarian dispositions has implications for stereotype accuracy. If the core components of stereotypes were the product of cognitive processes that veridically tracked reality (e.g., Jussim, 2012), then generic beliefs should be overweighted by participants with more (vs. less) rational thinking dispositions. The fact that they were instead relatively overweighted in the judgments of participants with low-effort, rigid thinking styles is further evidence against a close match between stereotypes and reality. This work also contributes to theories on stereotype content, which have so far identified the dimensions of warmth and competence as fundamental (Fiske et al., 2002; Judd et al., 2005). Incorporating the generic/statistical distinction into this theorizing could present several advantages. For example, the generic content of stereotypes may explain why evidence fails to falsify them (e.g., Chambers, Graham, & Turner, 2008; Cimpian, Brandone, & Gelman, 2010) and why stereotypes are often accompanied by essentialist overtones (e.g., Cimpian & Markman, 2011; Cimpian & Salomon, 2014; Rhodes, Leslie, Saunders, Dunham,


16

& Cimpian, in press), which in turn foster prejudice and discrimination (Dar-Nimrod & Heine, 2011). More broadly, generic and statistical ways of reasoning about categories differ in terms of how resource-intensive they are, how early they develop, and how robust the resulting memory traces are (for a summary, see Cimpian, 2016)— which have as-yet unexplored consequences for stereotype content and processes. The generic/statistical distinction is relevant to applied issues as well, such as reducing intergroup prejudice. If generic beliefs are more central to stereotypes, then interventions to reduce prejudice that target the generic component of stereotypes (e.g., considering multiple group memberships; Prati, Crisp, Meleady, & Rubini, 2016) should be more effective than interventions that target the statistical component (e.g., presenting stereotype-disconfirming evidence; Weber & Crocker, 1983). Future work could test this prediction by comparing the effectiveness of prejudice-reduction interventions aimed at the generic vs. the statistical elements of stereotypes. Finally, Study 3 suggested that the privileged role of generic beliefs in people’s judgments may be a uniquely social phenomenon. One potential explanation for this domain difference is the difference in attributes typically represented about social and non-social categories. Social stereotypes are often about psychological traits such as being intelligent, hot-tempered, or shy (e.g., Fiske et al., 2002), which are inherently more ambiguous than many of the attributes represented about non-social categories (e.g., whether a barn is red; see Supplementary Table 5). As a result, encoding and using statistical information may be more feasible for non-social categories, and—conversely—generic beliefs may be uniquely privileged for social stereotypes because the complexities encountered in this domain force greater reliance on cognitive shortcuts. In sum, this work introduces the key distinction between generic and statistical beliefs and illustrates the primacy of generic beliefs in the cognitive structure of stereotypes. This


17

distinction also adds to theory on stereotype content, suggesting new ways of thinking about stereotype accuracy.


18

References Allport, G. W. (1954). The nature of prejudice. Cambridge, MA: Perseus Books Altemeyer, B. (1996). The authoritarian spectre. Cambridge, MA: Harvard University Press. Ashmore, R. D., & Del Boca, F. K. (1981). Conceptual approaches to stereotypes and stereotyping. In D. L. Hamilton (Ed.), Cognitive processes in stereotyping and intergroup behavior (pp. 1-35). Hillsdale, NJ: Erlbaum. Behrens, L. (2005). Genericity from a cross-linguistic perspective. Linguistics, 43, 275-344. Bodenhausen, G. V. (1990). Stereotypes as judgmental heuristics: Evidence of circadian variations in discrimination. Psychological Science, 1, 319-322. Bian, L., & Cimpian, A. (in press). Are stereotypes accurate? A perspective from the cognitive science of concepts. Behavioral and Brain Sciences. Brewer, M. B. (1988). A dual process model of impression formation. In T. K. Srull & R. S. Wyer (Eds.), Advances in social cognition (Vol. 1, pp. 1–36). Hillsdale, NJ: Erlbaum. Buhrmester, M., Kwang, T., & Gosling, S. D. (2011). Amazon’s Mechanical Turk: A new source of cheap, yet high-quality, data? Perspectives on Psychological Science, 6, 3-5. Carlson, G. N., & Pelletier, F. J. (1995). The generic book. Chicago: University of Chicago Press. Chambers, C. G., Graham, S. A., & Turner, J. N. (2008). When hearsay trumps evidence: How generic language guides preschoolers’ inferences about unfamiliar things. Language and Cognitive Processes, 23, 749-766. Cimpian, A. (2016). The privileged status of category representations in early development. Child Development Perspectives, 10, 99-104. Cimpian, A., Brandone, A. C., & Gelman, S. A. (2010). Generic statements require little evidence for acceptance but have powerful implications. Cognitive Science, 34(8), 1452-1482.


19

Cimpian, A., & Erickson, L. C. (2012). The effect of generic statements on children's causal attributions: Questions of mechanism. Developmental Psychology, 48, 159-170. Cimpian, A., Gelman, S. A., & Brandone, A. C. (2010). Theory-based considerations influence the interpretation of generic sentences. Language and Cognitive Processes, 25, 261-276. Cimpian, A., & Markman, E. M. (2011). The generic/nongeneric distinction influences how children interpret new information about social others. Child Development, 82, 471492. Cimpian, A., & Salomon, E. (2014). The inherence heuristic: An intuitive means of making sense of the world, and a potential precursor to psychological essentialism. Behavioral and Brain Sciences, 37(5), 461-480. Dar-Nimrod, I., & Heine, S. J. (2011). Genetic essentialism: On the deceptive determinism of DNA. Psychological Bulletin, 137, 800-818. Devine, P. G. (1989). Stereotypes and prejudice: Their automatic and controlled components. Journal of Personality and Social Psychology, 56, 5-18. Fiske, S. T., Cuddy, A. C., Glick, P., & Xu, J. (2002). A model of (often mixed) stereotype content: Competence and warmth respectively follow from perceived status and competition. Journal of Personality and Social Psychology, 82, 878-902. Frederick, S. (2005). Cognitive reflection and decision making. Journal of Economic Perspectives, 19, 25-42. Gelman, A., Carlin, J. B., Stern, H. S., & Rubin, D. B. (2003). Bayesian data analysis (2nd ed.). New York: Chapman & Hall/CRC. Gelman, S. A. (2004). Learning words for kinds: generic noun phrases in acquisition. In D. G. Hall and S. R. Waxman (Eds.), Weaving a Lexicon (pp. 445-483). Cambridge, MA: MIT Press.


20

Gelman, S. A., Coley, J. D., Rosengren, K., Hartman, E., & Pappas, A. (1998). Beyond labeling: The role of maternal input in the acquisition of richly-structured categories. Monographs of the Society for Research in Child Development, 63, 1-157. Gelman, S. A., Taylor, M. G., & Nguyen, S. P. (2004). Mother-child conversations about gender: Understanding the acquisition of essentialist beliefs. Monographs of the Society for Research in Child Development, 69, 1-127. Greenwald, A. G., & Banaji, M. R. (1995). Implicit social cognition: Attitudes, self-esteem, and stereotypes. Psychological Review, 102, 4-27. Hampton, J. A. (2012). Thinking intuitively: The rich (and at times illogical) world of concepts. Current Directions in Psychological Science, 21(6), 398-402. Hox, J. J. (1998). Multilevel modeling: When and why. In I. Balderjahn, R. Mathar & M. Schader (Eds.), Classification, data analysis, and data highways (pp. 147-154). New York: Springer Verlag. Judd, C., James-Hawkins, L., Yzerbyt, V., & Kashima, Y. (2005). Fundamental dimensions of social judgment: Understanding the relations between judgments of competence and warmth. Journal of Personality and Social Psychology, 89, 899-913. Judd, C. M., & Park, B. (1993). Definition and assessment of accuracy in social stereotypes. Psychological Review, 100, 109-128. Jussim, L. (2012). Social perception and social reality: Why accuracy dominates bias and self-fulfilling prophecy. New York: Oxford University Press. Jussim, L. (in press). Précis of “Social perception and social reality: Why accuracy dominates bias and self-fulfilling prophecy.” Behavioral and Brain Sciences. Jussim, L., Crawford, J.T., & Rubinstein, R. S. (2015). Stereotype (in)accuracy in perceptions of groups and individuals. Current Directions in Psychological Science, 24, 490-497. Kahneman, D. (2011). Thinking, fast and slow. New York: Farrar, Straus and Giroux.


21

Keller, J. (2005). In genes we trust: the biological component of psychological essentialism and its relationship to mechanisms of motivated social cognition. Journal of Personality and Social Psychology, 88, 686-702. Khemlani, S., Leslie, S. J., & Glucksberg, S. (2012). Inferences about members of kinds: The generics hypothesis. Language and Cognitive Processes, 27, 887-900. Kruschke, J. K., Aguinis, H., & Joo, H. (2012). The time has come: Bayesian methods for data analysis in the organizational sciences. Organizational Research Methods, 15, 722-752. Leslie, S. J. (2008). Generics: Cognition and acquisition. Philosophical Review, 117, 1-47. Leslie, S. J. (in press). The original sin of cognition: Fear, prejudice and generalization. Journal of Philosophy. Leslie, S. J., & Gelman, S. A. (2012). Quantified statements are recalled as generics: Evidence from preschool children and adults. Cognitive Psychology, 64, 186-214. Leslie, S. J., Khemlani, S., & Glucksberg, S. (2011). Do all ducks lay eggs? The generic overgeneralization effect. Journal of Memory and Language, 65, 15-31. Muthén, B. (2010) Bayesian Analysis in Mplus: A Brief Introduction. Mplus Technical Report. Available from http://www.statmodel.com Muthén, L. & Muthén, B. (1998-2012). Mplus User’s Guide. Los Angeles, CA: Muthén & Muthén. Prasada, S. (2000). Acquiring generic knowledge. Trends in Cognitive Sciences, 4, 66-72. Prasada, S., Khemlani, S., Leslie, S. J., & Glucksberg, S. (2013). Conceptual distinctions amongst generics. Cognition, 126, 405-422. Prati, F., Crisp, R. J., Meleady, R., & Rubini, M. (2016). Humanizing outgroups through multiple categorization: The roles of individuation and threat. Personality and Social Psychology Bulletin, 42, 526-539.


22

Rhodes, M., Leslie, S. J., Saunders, K., Dunham, Y., & Cimpian, A. (in press). How does social essentialism affect the development of inter-group relations? Developmental Science. Salthouse, T. A. (2015). Individual differences in working memory and aging. In R. H. Logie & R. G. Morris (Eds.), Working memory and ageing (pp. 1-20). New York: Psychology Press. Sidanius, J., & Pratto, F. (1999). Social dominance: An intergroup theory of social hierarchies and oppression. New York: Cambridge University Press. Sterken, R. K. (2015). Generics, content and cognitive bias. Analytic Philosophy, 56(1), 7593. Tasimi, A., Gelman, S. A., Cimpian, A., & Knobe, J. (in press). Differences in the evaluation of generic statements about human and non-human categories. Cognitive Science. Weber, R., & Crocker, J. (1983). Cognitive processes in the revision of stereotypic beliefs. Journal of Personality and Social Psychology, 45, 961-977. West, S. G., Ryu, E., Kwok, O. M., & Cham, H. (2011). Multilevel modeling: Current and future applications in personality research. Journal of Personality, 79, 2-50.

The Cognitive Structure of Stereotypes Table 1 The Items Used in Studies 1–4 (in Generic Form) Study 1

Studies 2, 3, and 4

1. African Americans are athletic

1. Asian people are bad drivers

2. African Americans are musical

2. Asian people are good at math

3. African Americans are undisciplined

3. Asian people are smart

4. African Americans are violent

4. Black people are athletic

5. Asians are ethnocentric

5. Black people are lazy

6. Asians are hardworking

6. Black people are unintelligent

7. Asians are intelligent

7. Black people are violent

8. Asians are introverted

8. Blondes are dumb

9. Caucasians are racist

9. British people have bad teeth

10. Doctors are hardworking

10. Fat people are lazy

11. Doctors are intelligent

11. French people are snobs

12. Jewish people are cheap

12. Irish people are alcoholics

13. Jewish people are ethnocentric

13. Jewish people are cheap

14. Jewish people are intelligent

14. Jewish people are rich

15. Lawyers are competitive

15. Men are selfish

16. Lawyers are dishonest

16. Men are strong

17. Lawyers are greedy

17. Mexicans are hard-working

18. Men are assertive

18. Mexicans are lazy

19. Men are intellectual

19. Muslims are terrorists

20. Men are logical

20. Native Americans are alcoholics

21. Men are self-reliant

21. Politicians are liars

22. Politicians are dishonest

22. Poor people are lazy

23. Teachers are hardworking

23. Redheads have a temper

24. Teachers are intelligent

24. Rock stars do drugs

25. Women are compassionate

25. Southerners are ignorant

26. Women are honest

26. Teenagers are irresponsible

27. Women are sensitive

27. White people are racist 28. White people are rich 29. Women are delicate 30. Women are unintelligent

Note. Stereotypes in Study 1 were selected from prior research on stereotype endorsement. Stereotypes in Studies 2, 3, and 4 were generated by an independent sample of participants. See OSM for item means and standard deviations.

23


24

Table 2 Descriptive Statistics and Correlations across Measures in Studies 1–4 Mean (SD)

1.

2.

3.

4.

60.79 (22.96)a

−

2. Generic Beliefs

0.73 (1.35)b

.73*

−

3. Social Judgment (Expectation)

0.71 (1.25)b

.70*

.76*

4. Social Desirability

6.95 (3.10)c

.03

.01

1. Statistical Beliefs

37.89 (26.52)a

−

2. Generic Beliefs

39.91 (29.25)a

.72*

−


−0.21 (1.48)b

.66*

.70*

−

4. Authoritarianism

−1.36 (1.36)b

.13*

.20*

.10*

5. Cognitive Style

1.65 (1.13)d

−.03*


39.61 (26.37)a

−

2. Generic Beliefs

42.60 (29.34)a

.72*

−


−0.24 (1.47)b

.67*

.71*

−

4. Authoritarianism

−1.33 (1.07)b

.28*

.17*

.15*

−

5. Cognitive Style

1.74 (1.07)d

−.17*

−.08*

−.06*

−.18*


34.60 (25.25)a

−

2. Generic Beliefs

35.41 (29.63)a

.73*

−

3. Social Judgment (Biological Expl.)

−2.12 (1.46)b

.26*

.29*

−

4. Authoritarianism

−1.52 (1.10)b

.17*

.17*

.15*

5. Cognitive Style

1.72 (1.13)d

−.05*

Study 1 1. Statistical Beliefs

− −.01

−

Study 2

−.01

−.02

− −.19*

Study 3 (Social Items)

Study 4

−.01

−.01

− −.21*

Note. Superscripts represent scale of measurement. a = 100-point sliding scales, b = −3 to 3 Likert-type scales, c = 0 to 13 scale, and d = 0 to 3 scale. See Supplementary Tables 4 and 5 in the OSM for descriptive statistics for the non-social items in Study 3. *p < .05.