Competence and warmth in context: The compensatory nature of stereotypic views of national groups

In two experiments we show that the context in which groups are perceived influences how they are judged in a compensatory manner on the fundamental dimensions of social judgment, that is, warmth and competence. We manipulate the type of country (high in competence and low in warmth vs. high in warmth and low in competence) to which a target country is compared. Our data show that the target country is perceived as warmer and less competent when the comparison country is stereotypically high (vs. low) in competence and low (vs. high) in warmth. We also found compensation correlationally across targets and across dimensions in that the higher the comparison country is rated on one of the two dimensions, the higher the target country is rated on the other. Compensation effects are shown to affect judgments of both the ingroup (Experiment 1) and an outgroup (Experiment 2). Our results shed new light on context effects in group judgments as well as on the compensatory relation of the two fundamental dimensions of social judgment. Copyright # 2008 John Wiley & Sons, Ltd. Research has identified two fundamental dimensions of social perception. Although different names have been used, there is wide agreement on the common core of those dimensions (Abele & Wojciszke, 2007). Here, we use the labels warmth and competence (Fiske, Cuddy, Glick, & Xu, 2002). Recent experimental work by Judd, James-Hawkins, Yzerbyt, & Kashima (2005) provides clear evidence that social perception is characterized by a compensatory relation between competence and warmth: when a social target is seen as higher than another on one dimension, this target will likely be perceived as lower than the other on the second dimension. Our question is whether the use of a different comparison group in the context alters the evaluation of a target group in a way that demonstrates compensation. Compensation in Social Judgments In a study that examined the dimensions of competence and warmth in a full ingroup–outgroup design, Yzerbyt, Provost, and Corneille (2005) asked French and Belgian respondents to indicate how they perceived their own and the other group in terms of competence and warmth. The data strongly supported the compensation effect in that both groups of respondents described the French as more competent than the Belgians but also the Belgians as warmer than the French. Using a more controlled setting, Cuddy, Fiske, and Glick (2004) had their participants examine several individual profiles in the context of a personnel evaluation procedure. Female professionals with children were viewed as warmer than European Journal of Social Psychology Eur. J. Soc. Psychol. 38, 1175–1183 (2008) Published online in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/ejsp.526 *Correspondence to: Nicolas Kervyn, Department of Psychology, Université Catholique de Louvain, 10, Pl. Cardinal Mercier, Louvain-la-Neuve, Belgium 1348. E-mail: [email protected] Copyright # 2008 John Wiley & Sons, Ltd. Received 28 September 2007 Accepted 29 February 2008 competent but also as warmer and as less competent than female professionals without children (Cuddy, Norton, & Fiske, 2005). Given the dearth of experimental work on these two dimensions in the context of intergroup relations, Judd et al. (2005) conducted a series of experiments investigating the relation between competence and warmth and the factors that may influence this relationship. Participants received lists of behaviors allegedly describing two different groups. High competence behaviors were attributed to the members of one group and low competence behaviors to the members of the other group. Each participant judged both groups on various scales that were related to warmth and competence. Unsurprisingly, the high competent group was judged to be more competent than the low competent one. Of more interest, the high competence group was also judged as less warm than the low competence group. Similar effects were observed on competence when warmth was manipulated (Experiment 2). These results emerged despite the fact that when the behaviors were pre-tested (Judd et al., 2005) to verify that, for instance, competence relevant behaviors conveyed competence information but not warmth, a small positive correlation materialized between their pretest ratings on competence and warmth. Thus, even though more competent behaviors were judged as slightly warmer than less competent ones (replicating the halo effect, Rosenberg, Nelson, & Vivekananthan, 1968), when two groups were described, one with high and the other with low competent behaviors, they were judged as differing in warmth in the opposite direction. Importantly, Judd et al. (2005, Experiment 4) showed that a comparative context is a necessary condition for compensation. When participants were presented with either the high or the low competence group, halo rather than compensation was observed. Specifically, the high competence group was rated as warmer than the low competence group. Finally, Judd et al. (2005, Experiment 5) found compensation even when participants were led to believe that they belonged to one of the groups. Assessing Compensation Compensation effects (Cuddy et al., 2004; Judd et al., 2005; Yzerbyt et al., 2005) have mainly been examined at the mean level. Evidence for compensation rests on the finding that participants see one group to be higher than the other on one dimension while rating this same group lower than the other group on the second dimension. However, compensation may not only materialize under the form of a negative relationship between warmth and competence at the mean level but also in terms of a negative relationship between the judgments at the respondent level. Specifically, the participants that differentiate more the two groups on one dimension should also differentiate them more on the other dimension but in the opposite direction. Such a correlation was examined by Judd et al. (2005) and, although the results were at times marginal, the predicted negative correlation emerged in all four studies in which two groups were compared. Rather than looking at the correlation that involves within-dimension differences across groups, an alternative way to examine the compensation effect at the correlational level is to compute the correlation across dimensions and across groups. Compensation would be found if the higher one group is rated on one of the two dimensions, the higher the other group is rated on the other dimension. The compensation pattern between warmth and competence would therefore translate into a positive correlation across groups and across dimensions. Yzerbyt et al. (2005) computed and found such a positive correlation. In a recent study, Demoulin, Geeraert, and Yzerbyt (2007) examined how the perception of exchange students of their homeand host-country evolved through time. These authors recommended computing a series of regression models in which one group’s standing on one dimension (e.g. warmth) is predicted by the same group’s standing on the other dimension (e.g. competence) and by the other group’s position on each of the two dimensions. Results again confirmed the presence of compensation in that the evaluation of one group on one dimension was positively predicted by the evaluation of the other group on the other dimension. In the present work, we examine compensation both by looking at mean levels of judgments as well as correlations across dimensions and across groups. The Impact of Contexts on Stereotype Content Research suggests that, within a given population, there is wide agreement on how social groups in general and nations in particular are perceived (Schneider, 2004). This consensus holds for the evaluations on the two fundamental dimensions Copyright # 2008 John Wiley & Sons, Ltd. Eur. J. Soc. Psychol. 38, 1175–1183 (2008) DOI: 10.1002/ejsp 1176 Nicolas Kervyn et al.