The compensation effect (Kervyn et al., 2010) mainly occurs in relation to two fundamental dimensions of social judgment (i.e., warmth and competence, Fiske et al., 2002, 2007; Cuddy et al., 2007, 2008). Previous research suggests that the compensation effect exists among groups with different social status (Yzerbyt et al., 2005; Biernat et al., 2009). Specifically, when two groups are judged, high social status group members are more likely to be viewed as more competent but colder, whereas members of the low-status group are viewed as warmer but less competent. Most importantly, this tendency to appraise two groups with different social status in a compensatory way reflects a high degree of consensus across high- and low-status group members (Jost et al., 2005; Yzerbyt et al., 2005). For example, one study showed that both French (high-status group) and Belgian (low-status group) participants associated French individuals with high levels of competence and low levels of warmth and associated Belgian individuals with low levels of competence and high levels of warmth (Yzerbyt et al., 2005).

Following the traditional social identity approach, compensation has been interpreted as a social creativity strategy to re-establish positive social identity for low-status groups and as a magnanimity strategy to maintain positive social identity for high-status groups (Yzerbyt et al., 2008). When an ingroup’s low social status is unlikely to change, disadvantaged group members seek to bolster their standing on other comparative dimensions, although they have to acknowledge their ingroup inferiority on status-related dimensions (i.e., the social creativity strategy, van Knippenberg, 1978; Tajfel and Turner, 1986; Niens and Cairns, 2003). Members of an advantaged group whose superior status is fully ensured may attempt to appear non-discriminatory (Cambon et al., 2015) by showing no ingroup favoritism (Verkuyten and Reijerse, 2008) or even outgroup bias on such status-irrelevant dimensions as communion or warmth (Bettencourt et al., 2001; Leach et al., 2002; Vanbeselaere et al., 2006).

The above analysis indicates that compensation is a motivational strategy to maintain positive social identity when the status of advantaged and disadvantage groups is relatively stable. Therefore, we assumed that the emergence of compensation might rely on specific sociostructural conditions, and when the social structure changes, the compensation effect may disappear. Several empirical studies support our assumption. Terry (2003) finds that the social creativity and magnanimity strategies are usually adopted when the social structure is relatively unchangeable. Meanwhile, Cambon et al. (2015) find that perceived status legitimacy accounts to the indirect relationship between increased status difference and compensation effect. Specifically, the more asymmetrical social hierarchy is, the more legitimate people perceive their status and the more compensation they exhibit. However, this work was correlational in nature and failed to reveal causal relationship (Fiedler et al., 2011). Therefore, present research aimed to add to these literatures by exploring the effect of social structure on the compensation effect.

Given that group boundary permeability and the legitimacy of social hierarchy are key characteristics of social structure and closely related to stability of group status (Tajfel, 1978), the present research aimed to examine their effects on compensation. Permeability indicates the extent to which group members can leave one group and join another (Tajfel and Turner, 1986; Verkuyten and Reijerse, 2008). Legitimacy represents the extent to which privileged status is based on fair grounds (Tajfel, 1978). These two important characteristics determine people’s preferred responses to social hierarchy (Blair and Jost, 2003; Rubin et al., 2014; LeBlanc et al., 2015). Although there is no direct evidence of the effects of group boundary permeability and social hierarchy legitimacy on compensation, previous research regarding their influences on intergroup relations and identity management strategies informs our hypotheses.

Based on previous work, we predicted that high- and low-status group members would exhibit compensation when the group boundary was impermeable. On the one hand, the slim possibility of social mobility ensures the distinctive superiority of advantaged group members on status-relevant dimensions; therefore, they may show outgroup bias on status-irrelevant dimensions to manifest their magnanimity (Vanbeselaere et al., 2006). On the other hand, members of a disadvantaged group may re-establish positive social images for the ingroup and themselves by maintaining an advantage on comparative status-irrelevant dimensions while acknowledging ingroup inferiority on status-relative dimensions (Niens and Cairns, 2003) when they have no opportunities to change their group memberships. Therefore, it is likely that both high- and low-status group members will exhibit compensation when the group boundaries are impermeable.

In contrast, permeable group boundaries may elicit ingroup favoritism by advantaged group members and outgroup favoritism by disadvantaged group members on both status-relevant and status-irrelevant dimensions. One previous study finds that an open group boundary exposes threats to high-status group members’ positive distinctiveness; this motivates them to defend their group boundary and affirm their absolute superiority by differentiating themselves positively from disadvantaged group members (Scheepers and Ellemers, 2005). Moreover, permeable group boundaries also provide opportunities for low-status group members to change their group memberships; therefore, they may try to leave the disadvantaged ingroup and exhibit favoritism toward the advantaged outgroup (Niens and Cairns, 2003).

In addition to the effects of an impermeable group boundary, we expected that a legitimate social hierarchy may facilitate groups with different social status to engage in compensation. Specifically, when the social hierarchy is legitimate, members of advantaged groups need not resort to maintaining their undisputed privileged position (LeBlanc et al., 2015). Therefore, they may treat the disadvantaged outgroup in a relatively fair or even generous way (Leach et al., 2002; Nadler et al., 2009; Spears et al., 2010) to relieve strong normative pressures (Cambon et al., 2015). The legitimate social status quo justifies disadvantaged group members’ inferiority on status-relevant dimensions (Mullen et al., 1992; Bettencourt et al., 2001). Therefore, to re-establish the ingroup’s positive social image, disadvantaged group members are motivated to enhance the ingroup on the status-irrelevant dimensions that best define their identity (Tajfel and Turner, 1986). Based on this reasoning, we hypothesized that compensation would occur when the social hierarchy was legitimate.

The compensation effect will no longer exist when the social hierarchy is perceived as illegitimate. Because people’s perception of competence derives from the social status of groups (Fiske et al., 2002), the competence superiority of so-called high-status group members will no longer be acknowledged when the social status quo is illegitimate. Moreover, an illegitimate social hierarchy motivates disadvantaged group members to elevate the status of the ingroup to match that of high-status outgroups (Rubin et al., 2014) or even to exhibit ingroup favoritism in an instrumental way to reverse the intergroup status hierarchy and obtain a prestigious position (Bettencourt et al., 2001; Niens and Cairns, 2003; Scheepers et al., 2009). In addition, when the social hierarchy is perceived as illegitimate, advantaged group members may feel shameful and guilty about their exhibition of ingroup favoritism on status-relevant dimensions, so they are likely to exhibit more positive outgroup evaluation and egalitarianism (Shepherd et al., 2013a,b) to make them feel better. Some research (Kuang and Liu, 2012) has revealed that advantaged group members showed less discrimination to disadvantaged group members when they were told that the social hierarchy status quo was illegitimate (versus legitimate).

The present research was conducted with Chinese participants. In China, citizens are divided into either a rural or an urban category according to a spatial hierarchy. This categorization is based on the Hukou system, a register of the location of individuals’ permanent residences (Chan and Buckingham, 2008). People in the rural category live in the countryside and make their living by farming; those in the urban category live in cities and enjoy life-long benefits provided by the government. Therefore, in terms of social-economic status and social reputation, the urban category is seen as superior to the rural category (Zhang et al., 2014). In recent years, the Hukou system has been reformed to facilitate social integration and these changes have challenged the legitimacy of the existing social hierarchy (Kuang and Liu, 2012) and facilitated rural-to-urban mobility (Zhang et al., 2014). This current social situation provides a natural laboratory within which to explore the influences of legitimacy and permeability on compensation in high (i.e., urban category) and low (i.e., rural category) social status group members.

To examine the influence of social structure change on the compensation effect, three studies were conducted. As there is no previous research on the compensation effect with Chinese participants, it was first necessary to replicate the compensation effect in two categories of Chinese citizens (Study 1). We hypothesized that both high- (urban) and low-status (rural) group members would perceive the ingroup and outgroup in a compensatory way; that is, the urban group would be viewed as more competent than the rural group, whereas the rural group would be viewed as warmer than the urban group (Hypothesis 1). Then we examined how group boundary permeability (Study 2) and legitimacy of social status (Study 3) affected the compensation effect. According to the analysis above, both high- and low-status members should engage in compensation when the group boundary is impermeable (Hypothesis 2) and the social hierarchy is legitimate (Hypothesis 3).

A post hoc power analysis conducted using GPower (Faul et al., 2007) indicated that the achieved power of the compensation effect approached to 100% in this study, suggesting the sample size was large enough.

We conducted a 2 (Judge category: urban, rural) × 2 (Target category: urban, rural) × 2 (Dimension: warmth, competence) mixed-model analysis of variance (ANOVA) with the first factor varying between participants to analyze the compensation effect. A significant main effect of Target category showed that the rural group (M = 4.16, SD = 0.35) were evaluated more positively on general ratings than the urban group (M = 3.98, SD = 0.24), F(1,55) = 7.68, p = 0.008, η² = 0.12. We also found a significant effect of Dimension, F(1,55) = 28.23, p < 0.001, η² = 0.34, indicating that warmth ratings (M = 3.93, SD = 0.27) were generally lower than competence ratings (M = 4.20, SD = 0.28).

Most importantly, the expected Target category by Dimension interaction was significant, F(1,54) = 206.32, p < 0.001, η² = 0.79. More competence was attributed to individuals in the urban group (M = 4.69, SD = 0.46) than to those in the rural group (M = 3.72, SD = 0.48), t(56) = 9.88, p < 0.001, Cohen’s d = 2.08. In contrast, more warmth was attributed to individuals in the rural group (M = 4.60, SD = 0.53) than to those in the urban group (M = 3.27, SD = 0.45), t(56) = 12.14, p < 0.001, Cohen’s d = 2.71 (see Figure 1). Meanwhile, urban target was rated higher on competence than on warmth [t(56) = 13.95, p < 0.001, Cohen’s d = 3.12], while rural target was rated higher on warmth than on competence [t(56) = 9.22, p < 0.001, Cohen’s d = 1.74]. Thus, the Target category by Dimension interaction demonstrated the emergence of compensatory pattern of social perception.

The present study replicated previous findings and tested whether the response patterns observed by Yzerbyt et al. (2008) and Kervyn et al. (2016) emerged in Chinese participants. Study 1 revealed the existence of compensation among Chinese participants; specifically, both high-status (urban) and low-status (rural) participants perceived advantaged group members as more competent but colder than disadvantaged group members. These results were consistent with previous findings (Jost et al., 2005; Biernat et al., 2009) and suggested that compensatory stereotypes were acknowledged across high- and low-status groups.

A post hoc power analysis indicated that the achieved power of the compensation effect approached to 100% with the sample size of 129 in this study, suggesting the sample size was large enough.

A post hoc power analysis indicated that the achieved power of the compensation effect approached to 100% with the sample size of 120 in this study, suggesting the sample size was large enough.

The present research adds to existing research in several ways. First, most previous studies have only validated the existence of compensation (e.g., Kervyn et al., 2008; Biernat et al., 2009) and our findings further suggest that compensation does not always persist. The manipulation of social structure factors in the present research revealed that group boundary permeability and social hierarchy legitimacy are situational factors that affect compensation. Second, the present research combined a social identity perspective with stereotype content and provided evidence for the effect of social structure changes on strategies to manage identity. The expression of compensation reflected social creativity and magnanimity strategies to rebuild or maintain positive social identity for low-and high-status groups, respectively (Yzerbyt et al., 2008). The present research indicates that these strategies are adopted when the possibility of intergroup mobility is minimal or when the social hierarchy is considered legitimate. In contrast, the complementary pattern to evaluate each other on warmth and competence no longer existed if the group boundary was open or the legitimacy of social hierarchy was denied. These findings support the suggestion that social creativity and magnanimity strategies are used more readily under relatively unchangeable social conditions (Terry, 2003).

The present findings have some practical implications. In recent years, Chinese citizens have experienced the reform of the Hukou system. These reforms have challenged the fairness of the existing social hierarchy (Kuang and Liu, 2012) and facilitated rural-to-urban mobility (Zhang et al., 2014), and several tentative official policies have been tested in some places (Chan and Buckingham, 2008). The present research found that the salience of a permeable group boundary and illegitimate social hierarchy inhibited the occurrence of the relatively harmonious intergroup relations, (i.e., compensation) and increased disadvantaged group members’ favoritism toward the ingroup. However, these findings do not convince us of the infeasibility of the Hukou system reform or the desirability of maintaining an unfair status quo. The social changes resulting from the Hukou system reform have offered some advantages for low-status group members and decreased advantaged group members’ discrimination toward low-status group members (see Zhang et al., 2014). This dilemma requires the creation of more suitable ways to introduce social structure changes that do not arouse new prejudice and that allow shifts to the status quo satisfy everybody.

Findings of Study 2 indicated that the effect of a permeable group boundary on intergroup perception was more complex than we had expected. We found that disadvantaged group members exhibited ingroup favoritism on both dimensions when the group boundary was perceived as permeable. Although this contradicts some previous research (e.g., Lalonde and Silverman, 1993), there are also research indicating that permeable group boundary does not necessarily result in individual mobility among disadvantaged group members (e.g., Ellemers et al., 1992). Jackson (Jackson et al., 1996, Experiment 2) supposed that not all disadvantage group members but only those less identifiers choose to leave their ingroup when upward mobility is possible. So, when rural Chinese realize they can migrate to cities by their efforts, they may increase their confidence in ingroup capability and identification with ingroup, which makes them remain loyal to ingroup and take actions favorable to ingroup (Blair and Jost, 2003).

Meanwhile, our Study 2 results also showed that a permeable group boundary decreased rather than increased the advantaged group’s ingroup favoritism, which is contrary to some previous research (e.g., Scheepers and Ellemers, 2005). Agreeing with previous researchers (Zhang et al., 2014), we think this inconsistency may result from the different nature of group boundaries. The group boundary between rural and urban Chinese is mainly caused by geographical classification other than biological classification (e.g., skin color or race). The rural-to-urban mobility may increase the homogeneity between rural and urban Chinese and promote urban Chinese to perceive a more common ingroup identity with rural residence (Zhang et al., 2014), which then reduces prejudice and discrimination (Gaertner et al., 1993; Florack et al., 2003).

Some issues need to be addressed in future studies. First, our participants were urban and rural undergraduates, so further research on other groups naturally differing in status is needed to examine whether our findings can be extended to other kinds of Chinese social groups. Second, as the accuracy of stereotype has attracted more and more attention, more research should be conducted to test whether what we found is the compensation effect or just accurate social perception about rural and urban Chinese. As proposed by Jussim et al. (2015), we can compare people’s descriptive beliefs about the groups to the objective criteria that establish group characteristics. We can also measure and compare rural and urban people’s motivation under different conditions or test whether the motivation mediates the relationship between social structure and the compensation effect. Third, the curvilinear relation between groups’ perceived competence and warmth (i.e., warmth peaks at average competence; Imhoff and Koch, in press; Koch et al., 2016) suggests that the compensation effect only applies to groups with high and average competence, but not to groups with high and low or average and low competence. Future research should examine this boundary condition of the compensation effect. Finally, future studies should use more concise and unequivocal essay manipulations of status permeability and illegitimacy to examine whether it is actually these two or other aspects of social structure that led to the breakdown of the compensation effect in the studies reported here.