Abstract
The Stereotype Content Model proposes that competence (or alternatively, agency) is a fundamental dimension of stereotypes. According to this model, beliefs about agency are partially due to the status relations between groups, such that high status groups are perceived to possess agency, whereas low status groups are perceived to lack agentic characteristics. Despite the considerable support for this model, the psychological processes that produce these stereotypes have not been fully explored. In the current studies, we examined whether the correspondence bias may be partially responsible for the stereotype that members of low status groups lack agentic characteristics, relative to those who belong to high status groups. Across both studies, a measure of the correspondence bias predicted such stereotypical beliefs, even after accounting for variables that are known to be associated with beliefs about high and low status groups. This effect was observed when beliefs about the status of groups were experimentally manipulated, and when we measured stereotypical beliefs about two sets of actual high and low status groups.
The Stereotype Content Model (Cuddy, Fiske, & Glick, 2008; Fiske, Cuddy, Glick, & Xu, 2002) proposes that beliefs about groups can typically be conceptualized along two dimensions, warmth and competence (or, alternatively, communion and agency; Abele, Cuddy, Judd, & Yzerbyt, 2008). Moreover, the Stereotype Content Model also offers a possible explanation for the source of such beliefs. The Social Structural Hypothesis (Fiske et al., 2002) holds that beliefs about groups along these two fundamental dimensions are a consequence of the structure of relations between groups. The degree of competition between groups is hypothesized to be associated with the perception that a group lacks warmth and communion, such that outgroups that are in conflict with one’s ingroup are perceived to lack warmth. Furthermore, the status relations between groups (i.e., high status vs. low status) are believed to be associated with perceptions of competence and agency, such that low status groups are perceived to lack these characteristics. The Stereotype Content Model and the Social Structural Hypothesis have received considerable empirical support and have been particularly successful in predicting affective reactions toward a variety of different groups (for a review, see Cuddy et al., 2008). The model has also received support in other domains as well, including interpersonal perception (Russell & Fiske, 2008) and a variety of cultural contexts (Cuddy et al., 2009).
Despite the body of literature supporting the Stereotype Content Model, and the hypotheses derived from it, the specific psychological mechanisms that create such beliefs have yet to be fully explored. Although it has been established that the structure of relations between groups (in terms of competition and status) does have a causal relationship with beliefs about communion and agency (Caprariello, Cuddy, & Fiske, 2009), the psychological processes that may be responsible for these changes have yet to be thoroughly documented. In the current studies, we explored whether the correspondence bias, the tendency to draw dispositional inferences from situationally constrained behavior (Gilbert & Malone, 1995), is one of the psychological mechanisms that produces stereotypical beliefs about groups. In particular, we focused on the role that the correspondence bias may play in producing the stereotypical perception that members of high status groups possess more agentic characteristics than members of low status groups.
Attributions About High and Low Status Groups
The attribution literature has long been of interest to researchers in the field of intergroup relations. For decades, researchers have observed that attributions do play a role in the development and maintenance of stereotypical perceptions of outgroups (e.g., Pettigrew, 1979). More specifically, the significance of attributions in generating stereotypical beliefs about high and low status groups has been examined in numerous studies (e.g., Cozzarelli, Wilkinson, & Tagler, 2001; Lemieux & Pratto, 2003). One variable that has been associated with stereotypical attributions about high and low status groups is the Belief in a Just World (Lerner & Miller, 1978). Individuals who score high on measures of Belief in a Just World embrace the ideology that the world is a relatively fair place, in which people generally receive the rewards and punishments they deserve and achieve a status that is commensurate with their actual ability. Thus, those who score high on the Belief in a Just World believe that there are relatively weak external constraints on the achievements of groups and individuals. Not surprisingly, individuals who score high on a measure of a Belief in a Just World tend to explain social inequalities, such as poverty, in terms of the individual characteristics of those who comprise these groups, rather than on external factors and broader societal trends (Campbell, Carr, & MacLachlan, 2001; Harper, Wagstaff, Newton, & Harrison, 1990).
Social Dominance Orientation (Pratto, Sidanius, Stallworth, & Malle, 1994) is another variable that has been employed to understand stereotypical beliefs. This variable, which reflects an individual’s desire for group dominance and acceptance of inequality, has been associated with a number of stereotypical attributions about high and low status groups. For example, those who score high on measures of Social Dominance Orientation are more likely to hold prejudicial beliefs about low status minority groups, an effect that has been observed across a number of different studies and has been found in a number of different countries (Pratto, Sidanius, & Levin, 2006).
Taken together, both Belief in a Just World and Social Dominance Orientation are likely to function as system-justifying ideologies (Jost & Burgess, 2000) that may serve to perpetuate the stereotype that high status groups are indeed agentic, whereas individuals in low status groups are perceived to lack agency (Oldmeadow & Fiske, 2007). However, another phenomenon that may also be responsible for stereotypical attributions about groups is the correspondence bias, which has received less attention in stereotype research. The correspondence bias is the tendency to believe that the observed behavior of individuals corresponds to their internal traits and characteristics, rather than to external factors (Gilbert & Malone, 1995). Interestingly, this bias appears to be robust even under circumstances in which behavior is strongly constrained by situational factors, such as social roles. For example, in one well-known demonstration of the correspondence bias (Ross, Amabile, & Steinmetz, 1977), two individuals were randomly assigned to two different roles (a question “asker” and a “contestant” who answered the questions) and these roles strongly influenced the subsequent behavior of these two individuals. Nonetheless, this role-constrained behavior was largely attributed to the internal characteristics of the individuals, rather than to external features of the situation.
Several theorists have explicitly suggested that the correspondence bias may play a role in the development of stereotypes about groups (Conway, Pizzamiglio, & Mount, 1996; Fiske et al., 2002; Jones, 2006). In particular, the following hypothesis has been proposed: If the superior position of a high status group, relative to a lower status group, is attributed to the internal characteristics of the members of the group (e.g., the higher motivation of the high status group), rather than to external factors (e.g., lack of economic opportunity for the low status group), then the tendency to draw correspondent inferences should be associated with greater stereotyping of the high and low status groups. Thus, if the correspondence bias reflects a predisposition to explain outcomes in terms of the internal characteristics of individuals, rather than in terms of the external factors that constrain behavior, then the bias may indeed play a role in the formation of the stereotype that members of the high status group possess agentic characteristics, while the members of the low status group lack such characteristics. Although previous researchers have examined the general significance of internal attributions in generating stereotypical perceptions of high and low status groups (e.g., Zucker & Weiner, 1993), we know of no published study that has explicitly examined the relationship between measures of the correspondence bias, per se, and beliefs about the agency of high and low status groups. In the present studies, we examined whether such a relationship exists. Specifically we examined whether the correspondence bias may play a role in the stereotypical perception that members of high status groups are more likely to possess agentic characteristics, relative to members of low status groups.
Study 1
In the first study, we manipulated participants’ beliefs about the status of two groups, by having them listen to an audiotape describing two fictitious groups that ostensibly existed on a small Pacific island. In this description, which was based on one used by Conway, Pizzamiglio, and Mount (Study 3; 1996), one group was described as being higher and another as lower in status. We then assessed participants’ beliefs about the agency of these two groups. Prior to listening to the audiotaped descriptions of the two groups, we measured the degree to which participants drew correspondent inferences from role-constrained behavior (i.e., the correspondence bias), in what was described as an unrelated study. For this measure, which has been employed in a number of other studies that have used the “Role Effect” or “Quiz Role” paradigm (e.g., Block & Funder, 1986; Gawronski, 2003; Krull et al., 1999; Ross et al., 1977; Stalder, 2009; Yzerbyt, Rogier, & Fiske, 1998), participants watched an interaction in which one student, who was assigned by a coin flip to the role of “questioner,” composed a set of general knowledge questions and then quizzed a second student, who was randomly assigned to be the “contestant” and answer those questions. Participants then rated each student’s intellectual ability. Though the behavior of each individual is strongly constrained by the roles to which they have been randomly assigned, participants typically attribute their behavior to internal characteristics (i.e., they rate the questioner as intellectually superior to the contestant).
Finally, we also wished to see whether the correspondence bias could uniquely predict stereotypes about the agency of high and low status groups, even after we accounted for system-justifying ideologies that are known to be associated with beliefs about low status groups—Belief in a Just World and Social Dominance Orientation. We predicted that our measure of the correspondence bias, the Role Effect, would account for unique variance in perceptions of agency about low status groups, which could not be explained by either a Belief in Just World (BJW) or Social Dominance Orientation (SDO).
Method
Participants
Participants were 36 White undergraduates (32 women and 4 men) enrolled in an introductory psychology class in the Eastern United States. Participants received course credit for their participation.
Design
Within-subjects experimental design, in which the status of two fictitious groups was manipulated.
Procedure
Participants were greeted by two experimenters and informed that they would be participating in two different studies. One experimenter then introduced the first study, which was described as a study of person perception.
Participants then watched a videotape of two female students who were described as participants from an earlier stage of the study. The videotaped interaction between these two individuals replicated the procedure used by Block and Funder (1986), called the Role Effect, in which one student was assigned—by a coin flip—to the role of questioner and a second student was assigned to the role of contestant (Ross et al., 1977). Following their assignment to roles, the video showed the questioner formulate a list of general knowledge questions and then subsequently quiz the contestant on these questions. The contestant then attempted to answer each question. Participants then rated both individuals in terms of their “General knowledge” and their “Memory for single, isolated facts.” They also evaluated the questioner in terms of her “Ability to make up general knowledge questions” and the contestant in terms of her “Ability to answer general knowledge questions made up by others.” These ratings were made on a 1–100 scale, where higher ratings indicated greater intellectual ability.
Next, the first experimenter left the room and the second experimenter explained that participants would now be taking part in a study of Pacific Island cultures. This cover story appeared to be effective; no participants guessed that the two parts of the study were actually related to one another. Participants then listened to the audiotaped description of the two groups, based on Conway et al. (Study 3; 1996), which described a culture that ostensibly exists on a small atoll in the Pacific Ocean. This culture was comprised of two groups, the Bwisi and Mwangai, who support themselves through subsistence agriculture and fishing. The Bwisi was described as a higher status group, and the Mwangai was described as being lower in status. For example, the higher status of the Bwisi, relative to the Mwangai, was reflected by the location and style of their respective huts. The recording stated that “the cluster of Mwangai huts is clearly visible closer to the shoreline, while the Bwisi huts are a little further inland on higher ground, under the protection of palm trees, and closer to the cultivated gardens. In addition, the roofs of the Bwisi huts are thatched with a special kind of grass which gives them a characteristic deep red color; in turn, this color contrasts with the pale yellow of the roofs on the Mwangai’s huts.” After listening to the descriptions of the groups, participants then completed the dependent measures, which were also based on Conway et al. (1996). These measures included a set of characteristics that assessed the degree to which participants attributed five agentic characteristics (i.e., aggressive, assertive, dominant, forceful, has leadership qualities) to individuals within each group. These ratings were made on a 7-point scale, with “never true/almost never true” and “always true/almost always true” as anchors. We then averaged together participants ratings of each group on these five dimensions, such that higher scores reflected greater perceptions of agency.
As manipulation checks, we asked participants to indicate which group was higher in status and to evaluate each group on a 101-point feelings thermometer, with “extremely favorable” and “extremely unfavorable” as anchors. These measures allowed us to determine if participants were aware of the differential status of the two groups and to ensure that our status manipulation was not confounded by differences in the valence of each group, as assessed by the feelings thermometer. Finally, participants completed an eight-item version of the Belief in a Just World scale (Lipkus, Dalbert, & Siegler, 1996) and a 14-item version of the Social Dominance Orientation scale (Pratto et al., 1994). Responses on both were indicated on a 7-point scale, and for each scale participants’ responses were averaged together to create two scale scores, where higher scores indicated a stronger Belief in a Just World (α = .87) and a greater Social Dominance Orientation (α = .84).
Results and Discussion
We conducted a manipulation check to determine whether participants perceived the two groups to vary in status. Consistent with the purpose of the manipulation, 32 of our 36 participants correctly identified that the Bwisi was the higher status group. The four participants who incorrectly identified the Mwangai as the higher status group were excluded from subsequent analyses, which resulted in a sample of 32 participants. To determine whether the status manipulation also produced evaluative differences in perceptions of the two groups, which we hoped it would not, we examined participants’ rating of each group on the feelings thermometer. The high status group (M = 69.6) was not evaluated significantly differently than the low status group (M = 71.4), t (31) = 0.67, p = .51. Thus the status manipulation appeared to be effective; most participants were aware of the differential status of each group, and the manipulation did not produce evaluative differences.
To measure the effect of our status manipulation on perceptions of agency, we compared the perceived agency of the high and low status groups. Consistent with our prediction, participants rated the high status group as significantly higher in agency (M = 4.48) than the low status group (M = 3.05), t (31) = 5.89, p < .001. Thus we replicated the basic effect found by Conway et al. (1996), in which higher status groups were perceived to possess more agentic characteristics than lower status groups.
Relationship between perceptions of agency and the correspondence bias
To construct a measure of the intellectual ability of the questioner and the contestant, we averaged together participants’ ratings of each individual along three dimensions (i.e., general knowledge, memory, ability to ask/answer questions). Replicating previous research employing the Role Effect paradigm, participants attributed greater intellectual ability to the questioner (M = 74.1), relative to the contestant (M = 39.3), t (31) = 9.61, p < .001. Next, we subtracted the ratings of the contestant from the ratings of questioner; the resulting score (M = 34.8) reflected the magnitude of the Role Effect (higher scores indicated a greater Role Effect). This measure had good reliability, α = .84.
Our primary hypothesis was that individuals who showed a greater correspondence bias would be less likely to attribute agentic characteristics to low status groups, relative to high status groups. In order to examine the relationship between the Role Effect and these differences in agency stereotyping, we computed a single score which reflected the tendency to stereotype the high status group as possessing greater agency than the low status group. This was accomplished by subtracting the ratings of agency of the low status group from the high status group. The resulting measure (M = 1.43) was reliable (α = .79) and significantly greater than zero, t (31) = 10.53, p < .001, indicating a significant difference in agency stereotyping of the groups (see Table 1). Most importantly, the correlation between the Role Effect and the difference in agency stereotyping was significant, r (30) = .39, p = .03. This result indicates that participants who tended to exhibit a greater Role Effect were more likely to stereotype the low status group as lacking agentic characteristics, relative to the high status group. Thus, our primary hypothesis was supported. 1
Intercorrelations among dependent measures from Study 1 (N = 32).
Note. Asterisks indicate statistically significant Pearson correlation coefficients, p< .05 (two-tailed).
Finally, we wished to examine whether the Role Effect would offer predictive power over two variables that are often used to predict perceptions of high and low status groups, Belief in a Just World scale (BJW) and Social Dominance Orientation (SDO). In order to examine this question, we conducted a regression analysis in which we used the Role Effect, BJW, and SDO to predict the difference in agency stereotyping of the high and low status groups. Since we predicted that high scores on each attributional measure would be associated with greater differences in agency stereotyping, one-tailed tests were used to evaluate the significance of each attributional variable. Also, the predictor variables were entered simultaneously, so that we could determine whether each variable could account for unique variance in perceptions of agency. We observed that the Role Effect was a significant predictor (β = .40, p = .02), whereas BJW (β = .03, p = .43) and SDO (β = .04, p = .42) were not. Thus the Role Effect did account for unique variance in agency stereotyping, such that those who displayed a greater Role Effect were more likely to attribute agentic characteristics to the members of the high status group, relative to the low status group.
Study 2
The first study did indeed demonstrate that stereotypes of agency were associated with the correspondence bias, in which the status of groups was experimentally manipulated. However, the study employed only fictitious groups. Therefore in our second study, we examined whether the correspondence bias was associated with beliefs about two sets of actual groups in the United States—the wealthy and the poor, as well as women and men. We chose to use these groups because there is a significant amount of literature that indicates that attributions play an important role in generating stereotypes about the individuals within these groups (e.g., Cozzarelli et al., 2001; Heilman & Haynes, 2005). As in Study 1, we hypothesized that some individuals are more likely to demonstrate the correspondence bias than others, and that these individuals should be more likely to engage in agency stereotyping of high and low status groups. We therefore predicted that the degree to which individuals make internal attributions for role-constrained behavior (i.e., the Role Effect) should be associated with agency stereotypes about class (i.e., the wealthy and the poor) and gender (i.e., women and men). In other words, we expected that those individuals who are more likely to demonstrate the correspondence bias should be more likely to infer that members of the high status groups (i.e., the wealthy and men) are more agentic than members of the low status groups (i.e., the poor and women). As in Study 1, we also examined whether the Role Effect would predict agency stereotyping, even after we accounted for other measures that are known to be linked to beliefs about low status groups. So we again administered the Belief in a Just World scale (BJW) and a measure of Social Dominance Orientation (SDO).
Finally, when examining stereotypes of “real” groups, the agency dimension of stereotypes is not likely to be truly independent of communion (i.e., warmth), which is the other major dimension of stereotypes according to the stereotype content model. That is, these two fundamental dimensions—agency and communion—cannot be orthogonally manipulated when studying actual groups. Therefore, in Study 2 we conducted a set of multiple regression analyses that statistically controlled for the impact of communal stereotypes, so that we could more clearly examine the variables that predicted agency stereotypes. These analyses allowed us to determine which variables could predict stereotypes of agency, after already accounting for perceptions of communion. Our hypothesis for these analyses was that the Role Effect would remain a significant predictor of the difference in agency stereotypes of high and low status groups, even after controlling for the perceived communion of the groups.
Method
Participants
Participants were 72 undergraduates (60 women and 12 men) enrolled in an introductory psychology class in the Eastern United States. Participants received course credit for their participation. Seven participants were excluded for failing to complete the dependent measures, resulting in a final sample of 65 participants.
Procedure
As in Study 1, participants were greeted by two experimenters, were informed that they would be participating in two different studies, and were told that the first study dealt with person perception. Participants again watched a videotaped interaction in which one student, who was assigned by a coin flip to the role of questioner, quizzed a second student on a set of general knowledge questions. Participants then rated each student on the same measures used in Study 1.
Following the ratings of the individuals on the videotape, the first experimenter left the room, and the second experimenter explained that participants would now be completing a social attitudes questionnaire for a different study. Participants then completed a measure of class stereotyping, similar to the measure employed in Study 1; they rated wealthy people and poor people on a set of five agentic traits (i.e., aggressive, assertive, dominant, forceful, has leadership qualities). These ratings were made on a 7-point scale, with “never true/almost never true” and “always true/almost always true” as anchors. We then averaged together participants’ ratings of each group on these five dimensions, such that higher scores reflected greater perceptions of agency. To assess stereotypes of communion for class, participants also rated the wealthy and the poor on a set of five communal characteristics (i.e., compassionate, gentle, sensitive, sympathetic, tender), using the same scale and anchors that were used to assess agency stereotyping. These five items were then averaged together, so that higher scores reflected greater communion stereotyping. To measure gender stereotypes, participants also rated women and men on the same scales of agency and communion stereotyping. Participants then completed the same measures of Belief in a Just World and Social Dominance Orientation used in Study 1.
Results and Discussion
Descriptive statistics
Stereotypes of agency were consistent with expectations; the wealthy were perceived to be higher in agency (M = 5.09) than the poor (M = 3.70), t (64) = 9.73, p < .001, and men were perceived to be higher in agency (M = 5.07) than women (M = 3.91), t (64) = 12.67, p < .001. To construct a measure of agency stereotyping, which could be correlated with the Role Effect, we subtracted the ratings of agency of the low status group from the high status group (see Table 2). For both class (M = 1.39) and gender (M = 1.16), the resulting scores were significantly greater than zero (t’s > 9, p’s < .001), indicating significant differences in agency stereotyping of the high and low status groups. These measures had acceptable reliability for both class (α = .79) and gender (α = .66).
Intercorrelations among dependent measures from Study 2 (N = 65).
Note. Asterisks indicate statistically significant Pearson correlation coefficients, p< .05 (two-tailed).
Unsurprisingly, though not explicitly predicted, the poor were perceived to be higher in communion (M = 4.34) than the wealthy (M = 3.64), t (64) = 6.98, p < .001, and women were perceived to be higher in communion (M = 5.00) than men (M = 3.81), t (64) = 11.88, p < .001. To construct a measure of communion stereotypes, we subtracted the ratings of communion of the low status group from the high status group (see Table 2). For both class (M = -0.70) and gender (M = -1.20), the resulting scores were significantly less than zero (t’s > 5, p’s < .001) indicating significant differences in communion stereotyping of the high and low status groups. The measures were reliable for both class (α = .77) and gender (α = .70). With respect to the Role Effect, participants again demonstrated the Role Effect by attributing greater intellectual ability to the questioner (M = 76.6), on average, relative to the contestant (M = 43.9), t (64) = 14.9, p < .001. We constructed a measure of the Role Effect that was identical to that used in Study 1 (M = 32.7), which was again reliable (α = .73).
The role effect and stereotypes
We correlated the Role Effect with participants’ stereotypes of class and gender (see Table 2). As predicted, the Role Effect was positively correlated with agency stereotypes about class, r (63) = .27, p = .03, indicating that those participants who showed a greater Role Effect were more likely to perceive the wealthy as more agentic than the poor. The results for gender stereotypes were similar; the Role Effect was also significantly correlated with agency stereotypes about gender, r (63) = .25, p = .05, such that participants who showed a large Role Effect also tended to stereotype men as being more agentic than women. Conversely, the Role Effect was not correlated with class stereotypes of communion, r (63) = -.04, p = .75, or with gender stereotypes of communion, r (63) = -.03, p = .81. 2
Next, we wished to examine whether the Role Effect would be a unique predictor of class and gender stereotypes. That is, could the Role Effect explain variance in stereotypes, which could not be explained by the other attributional variables? To address this question we conducted two regressions, similar to those in Study 1, in which the three attributional variables (i.e., the Role Effect, BJW, and SDO) were used to predict differences in agency stereotyping of the high and low status groups, for both class and gender. Since we predicted that high scores on each attributional measure would be associated with greater class and gender stereotyping, one-tailed tests were again used to evaluate the significance of each attributional variable. Also, to statistically control for the impact of communion stereotyping, we included the measure of communion in the two regressions. Thus, the measure of communion stereotyping was used as a covariate in each analysis. In both regressions all four predictors (the attributional variables and the covariate) were simultaneously entered into the regression model.
In the first regression, we predicted differences in agency stereotyping for class (i.e., the wealthy and the poor) from the Role Effect, the BJW, and SDO, with communion stereotypes serving as a covariate. The covariate was significant (β = -.41, p < .001), indicating that perceptions of agency and communion were not orthogonal. In terms of the attributional variables, BJW was a significant predictor (β = .42, p < .001), while SDO was a marginally significant predictor (β = .15, p = .10). Most important, the Role Effect was a significant predictor of differences in agency stereotyping (β = .20, p = .04). This result indicated that the Role Effect explained unique variance in participants’ tendency to stereotype the wealthy as more agentic than the poor, even after statistically controlling for communion stereotyping.
In the second regression, we examined whether the same group of variables could also predict differences in agency stereotypes of women and men. Once again, communion stereotyping was included as a covariate and was significant (β = -.55, p < .001). BJW was a marginally significant predictor of the difference in agency stereotyping (β = .16, p = .07) while SDO was not (β = -.03, p = .38). As expected, the Role Effect was a significant predictor of agency stereotypes of men and women (β = .21, p = .02). This result indicated that the Role Effect explained unique variance in participants’ tendency to stereotype men as more agentic than women, even after controlling for perceptions of communion. Thus overall, the results of Study 2, in which actual stereotypes of “real” groups were assessed, closely mirrored the results of Study 1, in which the status of two groups was experimentally manipulated.
General Discussion
According to the Stereotype Content Model, competence (or agency) is a fundamental dimension of stereotypes. Moreover, the Social Structural Hypothesis holds that beliefs about agency are partially due to the status relations between groups, such that high status groups are more likely to be perceived to possess agentic characteristics, whereas low status groups are perceived to lack agency. Despite the considerable support for this model and its corresponding hypotheses, the psychological processes that produce these stereotypes about high and low status groups have not been fully explored. In the current studies, we sought to examine whether the correspondence bias may be partially responsible for agency stereotypes about high and low status groups. Across both studies, a measure of the correspondence bias—the Role Effect—predicted such stereotypical beliefs, even when we accounted for other variables that are known to be associated with beliefs about high and low status groups. Moreover, the Role Effect predicted stereotypes of agency even after controlling for perceptions of communion.
We also examined the relationship of system-justifying ideologies to the correspondence bias and stereotypes of agency. Although neither the Belief in a Just World nor Social Dominance Orientation were significantly correlated with our measure of the correspondence bias, we did find some support for the significance of these ideologies. Belief in a Just World was correlated with both agency and communion stereotypes about class, whereas Social Dominance Orientation was correlated with communal stereotypes about class. Neither of these variables appeared to be strongly correlated with gender stereotypes, though Belief in a Just World was a marginally significant predictor of agency stereotypes about gender. Thus it appears that system-justifying ideologies, in addition to the correspondence bias, do seem to play some role in explaining the effects predicted by the Social Structural Hypothesis, which is consistent with the results of previous research (Oldmeadow & Fiske, 2007).
We believe that our results have significant implications for the development of stereotypes, and suggest that the correspondence bias likely plays a role in the formation of beliefs about high and low status groups. However, certain aspects of our design were correlational in nature. For example, while we experimentally manipulated status in Study 1, one of the key findings from this study (that there was a relationship between the Role Effect and agency stereotypes) was correlational. Also since Study 2 focused on stereotypes of actual groups rather than fictitious group, it employed, by necessity, a correlational design. Despite the correlational nature of some aspects of the designs of these two studies, our results nonetheless indicate that the correspondence bias is associated with the tendency to stereotype high status groups as more agentic than low status groups. Those who are more likely to display the correspondence bias are also more likely to stereotype the members of high status groups as more agentic, relative to those who belong to low status groups.
Interestingly, across both studies we did not observe that the Role Effect was significantly correlated with either Belief in a Just World or Social Dominance Orientation. We find this result to be of particular importance because it suggests that some stereotypes about high and low status groups may not have their origin solely in system-justifying ideology. Rather, the results of the current study suggest that the correspondence bias, per se, may play an important role in the development of stereotypical beliefs. That is, the perception of high status groups as possessing agency, and low status groups as lacking agency, may—in part—be a consequence of everyday attributional processes, rather than only a reflection of an individual’s ideological motive to justify inequality. In order to explore this possibility more fully, future researchers could examine the relationship among system-justifying ideologies, the correspondence bias, and a broader range of social stereotypes. If the role of the correspondence bias in the formation of stereotypes is robust, then the effects we have observed may have broad implications. Stereotypes may, at least in part, be a byproduct of a fairly commonplace, perhaps ubiquitous, attributional bias.
Footnotes
Acknowledgements
We would like to thank the members of the Connecticut College Social Psychology Research Group for their guidance and feedback throughout the course of this project. We are especially grateful for the assistance of Natalie Porter, Sarah Clapp, Rachel Hess, Adriane Muzzy, and Emily Ricketson; they were indispensable in assembling the stimuli for the present studies. We would also like to thank Sarah Pincus for her assistance in preparing the manuscript. Finally, we would like to thank Joan Chrisler for her helpful comments on an earlier version of this paper.