Popularity and Acceptance as Distinct Dimensions of Social Standing for Chinese Children in Hong Kong
This research was supported by an Earmarked Research Grant (CUHK4620/05H) from the Research Grants Council, Hong Kong Special Administrative Region, China and by the University of Southern Carolina's fund for Advancing Scholarship in the Humanities and Social Sciences.
Abstract
This study attempted to validate distinctions between popularity and social acceptance in the cultural context of Hong Kong. We recruited 280 Chinese children (132 girls, 148 boys, mean age = 9.5) from Hong Kong primary schools. These children completed a peer nomination inventory assessing popularity, social acceptance, social rejection, aggression, peer victimization, and social behavior. Consistent with research conducted in western samples, we found that social acceptance was correlated primarily with positive behavioral characteristics (i.e., assertiveness-leadership and low levels of submissiveness-withdrawal). In contrast, popularity was associated with a more mixed pattern of features including high levels of aggression. The overall pattern of findings closely replicates past research conducted in North American and European settings.
Introduction
Research on children's social development has often incorporated a focus on children who are well liked or highly accepted by their peers (Coie & Dodge, 1983; Coie, Dodge, & Coppotelli, 1982). In this work, social acceptance is typically assessed with peer nomination items that require children to identify classmates who they like or would prefer as playmates. The resulting indices are associated primarily with positive behavioral tendencies. Not surprisingly, children who are well accepted by their peers are generally characterized by prosocial attributes. These children are described as being helpful, friendly, and sociable (Wentzel & Erdley, 1993).
From a somewhat different perspective, sociological theories on peer group organization have focused on the construct of popularity (Adler & Adler, 1998; Eder, 1985; Merten, 1997). Popularity is generally conceptualized as a shared recognition among peers that a particular child has achieved prestige, visibility, or high social standing (Adler, Kless, & Adler, 1992). This aspect of status is associated with a more mixed pattern of behavioral features than is social acceptance (Lease, Kennedy, & Axelrod, 2002). Popular children seem to be characterized by both prosocial attributes and aggression (LaFontana & Cillessen, 1998). For some popular children, aggressive or domineering behaviors may represent important tools for reaching and maintaining high standing in the peer group (Butcher, 1986; Luthar & McMahon, 1996).
On a theoretical level, popularity and social acceptance might be viewed as closely related constructs. Consistent with this suggestion, there are moderately strong associations between these two dimensions of social standing across a wide period of development (Cillessen & Mayeux, 2004). Still, by the middle years of childhood, being well accepted and being popular are not equivalent social outcomes (Gorman, Kim, & Schimmelbusch, 2002; LaFontana & Cillessen, 1998, 1999). In fact, some children who are identified as popular by their peers are not especially well liked (Parkhurst & Hopmeyer, 1998). There may also be differences in the implications of popularity and social acceptance for longer-term adjustment. Although social acceptance is likely to have either positive or neutral implications for development, popularity has been linked to negative outcomes in a number of studies. For example, popularity is predictive of unexplained absences from school (Schwartz, Gorman, Nakamoto, & McKay, 2006), sexual experimentation (Prinstein, Meade, & Cohen, 2003), and alcohol use (Mayeux, Sandstrom, & Cillessen, 2008). Under some circumstances, popularity in high-status peer networks can require conformity with maladaptive peer group norms (e.g., negative attitudes toward school; Farmer, Estell, Bishop, O'Neal, & Cairns, 2003; Schwartz et al., 2006).
The empirical evidence supporting a multidimensional perspective on social standing is compelling. Nonetheless, one potential limitation of the research in this domain is that it has been restricted primarily to western contexts. Indeed, we are unaware of any study conducted outside North America and Europe that includes assessments of both social acceptance and popularity. This limitation is noteworthy because an exclusive focus on western children's peer groups could limit the generalizability of existing theoretical perspectives (Weisz, McCarty, Eastman, Chaiyasit, & Suwanlert, 1997). It is not yet clear whether the distinction between social acceptance and popularity replicates across cultural contexts.
In the current investigation, we examined the correlates of popularity and social acceptance for Chinese children attending two Hong Kong primary schools. The social environment in this setting is quite complex. While under British jurisdiction, Hong Kong had extensive contact with other Asian cultures and with western cultures (Cheung-Blunden & Juang, 2008). The social, economic, legal, and political systems of these varied cultures have affected life in Hong Kong (Huque, Tao, Wilding, & Wilding, 1997). Even after return to Chinese control, Hong Kong society has been partially shaped by these external influences (Cheung-Blunden & Juang, 2008).
Despite the population's exposure to outside cultures, traditional Chinese values continue to serve as a predominant socializing factor for many of Hong Kong's children. For example, researchers have described similarities in the basic structure of parenting across Mainland China and Hong Kong (Yau & Smetana, 2003). In both contexts, there is a strong expectation of obedience to adult caregivers and compliance with Confucian principles (Berndt, Cheung, Lau, Hau, & Lew, 1993). As a broad generalization, Hong Kong's collectivistic value system emphasizes the maintenance of group well-being over individual interests (Yau & Smetana, 2003). To this end, socialization focuses on self-restraint, interdependent sense of self, and co-operative behavior (Bond, 1996; Ho, 1986).
Wide-sweeping cultural dichotomies, such as the distinction between collectivism and individualism, have been criticized for oversimplifying the tremendous diversity within and across settings (Fiske, 2002; Miller, 2002). Still, Mainland China and Hong Kong are often conceptualized as prototypes of societies with collectivistic orientations (Triandis, 1995). There is some empirical support for this perspective. Oyserman, Coon, and Kemmelmeier (2002) conducted a meta-analysis examining value systems in different cultural contexts. These researchers found that samples recruited from Hong Kong and China had relatively large effect sizes on indices assessing endorsement of collectivistic ideals as well as small effect sizes on indices assessing individualism.
At first consideration, collectivistic values and an orientation toward Confucianism might seem to be somewhat inconsistent with the western construct of popularity. Popularity in western children's peer groups involves aggression, dominating control over peers, and social manipulation (Lease, Musgrove, & Axelrod, 2002; Prinstein & Cillessen, 2003; Rose, Swenson, & Waller, 2004). To the extent that harmony with others is a central social objective, a child who is overly assertive or manipulative could be behaving in a manner that is in opposition to larger societal values. In settings where interdependence is an organizing principle, such behaviors may be unlikely to lead to prestige or high standing with peers.
An alternative hypothesis, and the one that we will endorse, might emphasize the idea that popularity does not require a child to engage in behaviors that are evaluated positively by most classmates or that are consistent with larger societal norms. Instead, popular children are likely to be those whose behavior incorporates the ‘Machiavellian’ tendencies needed to control others (Hawley, 2003). Although popularity does require sophisticated social skills, popular youths may maintain their status through behavioral strategies that peers view as aversive (Parkhurst & Hopmeyer, 1998). Under these conditions, popularity could reflect a form of social dominance that is distinct from acceptance in both North American and Hong Kong children's peer groups.
To test these competing perspectives, we examined associations between children's behavioral reputations and their social standing in the peer group during middle childhood. Our objective was to replicate findings from existing research in western contexts to the previously unexplored setting of Hong Kong. More specifically, we sought to examine evidence that popularity and social acceptance are associated with distinct patterns of correlates for Chinese children in Hong Kong schools. Similar to past investigations in North American and European children's peer groups, we expected that social acceptance would be correlated primarily with indicators of positive relationships with peers and adaptive social behavior. On the other hand, we predicted that popularity would be linked to a mixed pattern of positive and negative behavioral features. That is, we expected that popular youths would be high on indicators of social competence but also characterized by relational and overt aggression.
Our focus on children's social behavior was multifaceted, and we attempted to tap a number of dimensions of functioning with peers. Because previous investigators have portrayed popular youth as socially skilled and able to influence peers (Cillessen & Mayeux, 2004), we included assessments of leadership and assertiveness. Likewise, indicators of withdrawal and submissiveness tend to be negatively associated with popularity in western samples (Lease, Kennedy, & Axelrod, 2002), perhaps because popularity requires sociability and an assertive disposition. Therefore, we included items designed to index submissiveness and social withdrawal. Finally, we examined aggressive behavior given consistent findings in the literature regarding links between aggression and popularity (Schwartz et al., 2006).
As a related research goal, we explored evidence that there are distinct subgroups of popular youth. Past researchers have hypothesized that there are multiple behavioral pathways to popularity (Cillessen & Mayeux, 2004), and cluster analytic studies conducted in western children's peer groups have consistently yielded evidence for distinct behavioral subtypes of popular children (Lease, Musgrove, & Axelrod, 2002; Rodkin, Farmer, Pearl, & Van Acker, 2000). Some popular children rely on aggressive or socially manipulative strategies and hence are actively disliked by their peers (de Bruyn & Cillessen, 2006). Other popular children are characterized by assertive (but not aggressive) behavioral tendencies and are well liked as a result. We sought to determine if a similar two-cluster structure exists in the context of Hong Kong.
A final set of issues examined in this article relates to the implication of gender for social adjustment in Hong Kong children's peer groups. As an exploratory goal, we considered gender differences in relational aggression. When girls behave aggressively in North American peer groups, they tend to rely on strategies such as exclusion, gossiping, and spreading rumors (Crick & Grotpeter, 1995). Consequently, girls score relatively high on indices assessing relationally aggressive behavior (Cillessen & Mayeux, 2004). The pattern for Chinese children may be somewhat different, perhaps reflecting strong social sanctions against girls' aggressive behavior in this cultural context. Schwartz, Chang, and Farver (2001) examined aggression and victimization for children attending an elementary school in Tianjin, China. These investigators reported higher scores for boys than girls on indices assessing both overt and relational aggression (obtained via self-reports, teacher ratings, and peer nominations). We attempted to replicate this pattern of effects in the Hong Kong setting.
In a related vein, we considered gender as a potential moderator in the association between aggression and popularity. There is some evidence that in western children's peer groups, girls are more likely than boys to rely on relationally aggressive behavior to reach social dominance (Cillessen & Mayeux, 2004; Rose et al., 2004). However, these findings have not been fully consistent across studies (LaFontana & Cillessen, 2002) and warrant further exploration in the Hong Kong context.
The Current Study
We conducted our analyses as part of a larger investigation focusing on the correlates of social rejection and peer group victimization in Hong Kong primary schools (Tom, Schwartz, & Chang, 2005), although the described dataset has not been reported in any existing publications. We chose to focus on middle childhood because distinctions between popularity and social acceptance first become apparent in western children's peer groups during this developmental stage. Longitudinal research conducted in North America has demonstrated that popularity and acceptance become progressively more distinct through adolescence (Cillessen & Mayeux, 2004) but begin to have unique patterns of correlates by the later years of elementary school (Rodkin et al., 2000). Individual differences in aggression, a critically important correlate of popularity, also stabilize during this period of childhood (Eron, 1987; Olweus, 1979). Finally, middle childhood represents an important transitional period, with peers beginning to emerge as an increasingly significant socializing influence (Brown, 1990).
Method
Participants
Participants were 280 Chinese children (132 girls, 148 boys, mean age = 9.5) recruited from six third- and fourth-grade classrooms in two Hong Kong elementary schools. Class sizes in these schools were large by North American standards but were typical of Hong Kong. The schools served families living in publicly supported housing, implying that the children were from lower- to lower middle-class socioeconomic backgrounds.
We obtained written parental consent for each participating youth. Parents were reminded that their child's involvement was purely voluntary and that the study was not part of regular schoolwork. Of the eligible students, 95 percent returned positive parental permission, agreed to participate in the project, and were present at school during the data collection.
Procedure
Data were collected using a peer nomination inventory. The inventory was derived from the existing bully–victim literature including past work in the Chinese cultural context (Schwartz et al., 2001). The items listed in the current report are English back translations of Chinese items used during administration.
The peer nomination inventory contained a series of descriptors of children's social behavior, treatment by peers, and standing with peers. The inventory was group administered in all classrooms by trained research assistants. Students were given a class roster and asked to circle the names of three students in their class for each descriptor. Nominations were not restricted within gender (i.e., boys could nominate girls and girls could nominate boys). In each classroom, one research assistant read standard instructions aloud while another research assistant walked around to help students.
Measures
Popularity. Our efforts to generate a peer nomination item that taps popularity were complicated by translation issues. Popularity, as it is conceptualized in the English language, does not directly correspond to a specific Chinese term. Instead, we relied on a Chinese word (i.e., 
) that translates to ‘prestige, popular trust’ (Foreign Language Teaching and Research Press, 1997). We consulted with multiple native speakers who had graduate training in psychology. These consultants agreed that the identified term delineates children who occupy a high social position and have a leadership role in the peer group ecology. Similar to the western notion of popularity, this construct was also viewed as having both positive and negative connotations. For later analysis, we generated a popularity summary score from the total number of nominations received for the resulting item, standardized within class.
Social Acceptance and Rejection. Children nominated the three peers they liked most in their classroom and the three peers they liked least. Social acceptance and social rejection summary scores were then generated from the total number of nominations received for each item, standardized within class.
Aggression. We used three peer nomination items to assess overt aggression (‘someone who starts fights with other kids’, ‘someone who pushes or hits other kids’, ‘someone who bullies other kids’; α= .93) and two items to assess relational aggression (‘someone who gossips or says mean things about other kids', ‘someone who tries to leave other kids out of play to hurt their feelings’; r= .82, p < .001). For later analysis, we calculated the total number of nominations received for each item, standardized within class. We then generated relational and overt aggression variables from the mean of the respective items for each subtype.
We should acknowledge that our measurement strategy was designed to optimize content validity rather than discriminant validity. That is, to enhance identification of children with social problems, we selected assessment items that tap a broad range of relevant behaviors rather than focusing on narrow distinctions between subtypes. One implication of this strategy is that there are likely to be statistical associations between the relational and overt scales because the items are purposely worded in a manner that could tap overlapping behaviors. Indeed, confirmatory factor analyses conducted with similar items have generally yielded evidence for a unidimensional scale (Schwartz, Farver, Chang, & Lee-Shim, 2002; Schwartz et al., 2001). Nonetheless, we opted to calculate separate relational and overt scales because one of our exploratory objectives was to examine gender differences in aggression subtypes. In addition, aggression subtypes have specific relevance for the construct of popularity in western peer groups. Previous researchers have hypothesized that popular youth are particularly likely to use relational forms of aggression such as exclusion and spreading rumors to maintain their high positions in the peer group hierarchy (Cillessen & Mayeux, 2004).
Peer Victimization. We used three peer nomination items (‘someone who gets pushed around by other kids', ‘someone who gets picked on by other kids’, ‘someone who gets bullied by other kids’) to assess overt victimization and two peer nomination items (‘someone who has mean things said about them by other kids', ‘someone who gets excluded from play’; r= .54, p < .001) to assess relational victimization. For later analysis, we calculated the total number of nominations received for each item, standardized within class. We then generated a summary victimization variable from the mean of the five items (α= .89).
Submissive-withdrawn Behavior. To assess this dimension of children's social behavior, we used six peer nomination items (‘someone who would rather be alone than be with other kids', ‘someone who is shy and timid’, ‘someone who stays alone and rarely plays with others’, ‘someone who is submissive to others’, ‘someone who wants to play with others but is afraid to join in’, ‘someone who does not like talking’). For later analysis, we calculated the total number of nominations received for each item, standardized within class. We then generated a summary submissive-withdrawal variable from the mean of the five items (α= .90).
We opted to include a relatively large number of items that tap withdrawn-submissive social reputations as a reflection of ongoing controversies regarding the role of inhibited or passive behavior in the Chinese cultural context (see Schwartz et al., 2001; Tom et al., 2005). Investigators have suggested that inhibited dispositions may be linked to positive social outcomes for Chinese children (Chen, Rubin, & Li, 1995; Chen, Rubin, & Sun, 1992), although such effects probably do not hold when anxious-avoidant subtypes of withdrawal are emphasized (Hart et al., 2000). In any case, we sought to emphasize assessment of these behaviors in our peer nomination inventory given the prominent role of this construct in past publications.
Assertiveness-leadership. We included two peer nomination items to assess assertiveness-leadership (‘someone who is a good leader', ‘someone who is assertive and can stand up for himself/herself without using aggression’). For later analysis, we calculated the total number of nominations received for each item, standardized within class. We then generated a summary assertiveness-leadership variable from the mean of the two items (r= .63, p < .001). These specific items were taken directly from previous data collections in Mainland China (Abou-ezzeddine et al., 2007; Xu, Farver, Schwartz, & Chang, 2003) and correlate well with assessments obtained via other data sources (i.e., teacher ratings; Schwartz et al., 2001).
Results
Bivariate Relations
Bivariate correlations between the variables are summarized in Table 1. We evaluated these effects using a relatively conservative Type I error rate of .005 in order to maintain overall error rates for the study. As shown, popularity and social acceptance were positively correlated with a medium effect size. Popularity was associated with a mixed pattern of positive and negative features, replicating past research conducted in western contexts (Cillessen & Mayeux, 2004; Schwartz et al., 2006). On the one hand, popularity was correlated with relational aggression (with a small effect size). On the other hand, popularity was associated with indicators of assertiveness-leadership and low levels of submissiveness-withdrawal. The effect for assertiveness-leadership was large whereas the effect for submissiveness-withdrawal was small.
| Variables | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 |
|---|---|---|---|---|---|---|---|---|---|---|
| 1. Popularity | — | .41 * | .12 | −.12 | .11 | .13 | .24 * | −.17* | .71 * | −.08 |
| 2. Social acceptance | — | −.26* | −.23* | −.16 | −.12 | −.05 | −.24* | .45 * | .02 | |
| 3. Social rejection | — | .41 * | .69 * | .63 * | .63 * | .17 * | −.07 | −.24* | ||
| 4. Overt victimization | — | .53 * | .28 * | .17 * | .65 * | −.24* | −.23* | |||
| 5. Relational victimization | — | .61 * | .67 * | .23 * | −.04 | −.20* | ||||
| 6. Overt aggression | — | .74 * | .03 | −.04 | −.47* | |||||
| 7. Relational aggression | — | −.06 | .05 | −.32* | ||||||
| 8. Submissiveness-withdrawal | — | −.24* | .01 | |||||||
| 9. Assertiveness-leadership | — | .02 | ||||||||
| 10. Gendera | — |
- a Gender is coded 0 = male and 1 = female.
- * p < .005. Figures in bold were corrected (changing from negative to positive) on 8 December after online publication.
Social acceptance was linked exclusively to positive behavioral features including assertiveness and low levels of submissiveness-withdrawal. Social acceptance was also negatively correlated with social rejection and overt victimization. The effects were in the small to medium range.
To examine gender effects, we ‘dummy coded’ gender as a dichotomous variable (0 =‘male’, 1 =‘female’). Thus, positive correlations between a variable and gender indicate higher scores for females than males. Likewise, negative correlations indicate higher scores for males than females. Boys had higher scores than girls for both subtypes of aggression, both subtypes of victimization, and social rejection. In other words, our analyses replicated past findings in Mainland China (Schwartz et al., 2001), with boys receiving higher scores than girls on peer nomination indices for relational aggression and victimization.
Popularity and Social Acceptance as Independent Dimensions
Next, we conducted a series of hierarchical regression analyses to examine independent relations between each of the dimensions of social standing and the remaining peer nomination variables and to examine the potential moderating role of gender. For each of these models, we entered the main effects of popularity, social acceptance, and gender on the first step. On the second step, we entered the two-way interactions for popularity by gender, social acceptance by gender, and popularity by social acceptance (i.e., all possible two-way interactions). Variables were entered simultaneously at each step, and steps were entered sequentially. To maintain experiment error rates, we did not consider the significance of individual regression parameters within steps unless the ΔR2 associated with the full step reached significance.
The results of these analyses are summarized in Table 2. As shown, each of the models was significant overall. The variables entered on the first step of the model (i.e., the main-effect terms) significantly incremented R2 in each case. Popularity had positive associations with social rejection, assertiveness-leadership, relational and overt aggression, and relational victimization. Social acceptance was positively associated with assertiveness-leadership and negatively correlated with social rejection, submissiveness-withdrawal, and relational and overt victimization. Thus, the overall pattern of main-effects again supports the hypothesis that popularity is linked to a mixed pattern of correlates whereas social acceptance is linked primarily to positive features.
| Outcome variable | Step | Effects in model | β | sr 2 | Step Δr2 |
|---|---|---|---|---|---|
| Social rejection | 1 | Popularity | .25*** | .05 | |
| Social acceptance | −.36*** | .11 | |||
| Gender | −.21*** | .04 | .17*** | ||
| 2 | Popularity × gender | −.05 | .00 | ||
| Social acceptance × gender | .07 | .00 | |||
| Popularity × social acceptance | −.09 | .01 | .01 | ||
| Full Model: F(6, 273) = 10.25, p < .001, R2= .18 | |||||
| Overt victimization | 1 | Popularity | −.06 | .00 | |
| Social acceptance | −.20** | .03 | |||
| Gender | −.23*** | .05 | .11*** | ||
| 2 | Popularity × gender | .00 | .00 | ||
| Social acceptance × gender | .08 | .01 | |||
| Popularity × social acceptance | −.00 | .00 | .01 | ||
| Full Model: F(6, 273) = 5.79, p < .001, R2= .11 | |||||
| Relational victimization | 1 | Popularity | .20** | .03 | |
| Social acceptance | −.24*** | .05 | |||
| Gender | −.18** | .03 | .10*** | ||
| 2 | Popularity × gender | −.00 | .00 | ||
| Social acceptance × gender | .01 | .00 | |||
| Popularity × social acceptance | −.10 | .01 | .01 | ||
| Full Model: F(6, 273) = 5.33, p < .001, R2= .10 | |||||
| Overt aggression | 1 | Popularity | .17** | .02 | |
| Social acceptance | −.18** | .03 | |||
| Gender | −.45*** | .20 | .25*** | ||
| 2 | Popularity × gender | −.08 | .01 | ||
| Social acceptance × gender | .07 | .00 | |||
| Popularity × social acceptance | −.11 | .01 | .02 | ||
| Full Model: F(6, 273) = 16.96, p < .001, R2= .27 | |||||
| Relational aggression | 1 | Popularity | .28 *** | .06 | |
| Social acceptance | −.15* | .02 | |||
| Gender | −.29*** | .09 | .17*** | ||
| 2 | Popularity × gender | −.07 | .00 | ||
| Social acceptance × gender | −.05 | .00 | |||
| Popularity × social acceptance | −.14* | .01 | .03* | ||
| Full Model: F(6, 273) = 10.85, p < .001, R2= .19 | |||||
| Submissiveness-withdrawal | 1 | Popularity | −.08 | .01 | |
| Social acceptance | −.21** | .04 | |||
| Gender | .01 | .00 | .06*** | ||
| 2 | Popularity × gender | −.08 | .01 | ||
| Social acceptance × gender | .13 | .01 | |||
| Popularity × social acceptance | .08 | .00 | .02 | ||
| Full Model: F(6, 273) = 4.02, p < .001, R2= .08 | |||||
| Assertiveness-leadership | 1 | Popularity | .64 *** | .34 | |
| Social acceptance | .19 *** | .03 | |||
| Gender | .07 | .00 | .54*** | ||
| 2 | Popularity × gender | .13 ** | .01 | ||
| Social acceptance × gender | −.08 | .00 | |||
| Popularity × social acceptance | .03 | .00 | .02* | ||
| Full Model: F(6, 273) = 57.74, p < .001, R2= .56 | |||||
- Note: sr2 is the squared semi-partial correlation coefficient, the percentage of variance accounted for uniquely by the parameter at time of entry. Step Δr2 is the R-square change for the full step, the increment in variance accounted for when the combined variables are entered at each step.
- * p < .05,
- ** p < .01,
- *** p < .001. Figures in bold were corrected (changing from negative to positive) on 8 December after online publication.
The variables entered on step 2 (i.e., the interaction terms) significantly incremented model R2 in the prediction of assertiveness and relational aggression. For assertiveness, there was a significant popularity by gender interaction. To decompose this effect, we specified regression models predicting assertiveness from popularity and social acceptance, separately by gender. However, this analysis did not produce a theoretically meaningful pattern of findings with significant associations between popularity and assertiveness for both boys, β= .75, p < .005, and girls, β= .52, p < .005. That is, the effects were significant for both boys and girls, although they were larger for boys.
For relational aggression, we did not find any significant gender interactions, although there was a significant popularity by social acceptance two-way interaction. To decompose this effect, we used procedures recommended by Aiken and West (1991). We algebraically fixed the level of social acceptance at high (one SD above the mean), medium (the mean), and low (one SD below the mean) levels. We then examined the relation between popularity (with gender controlled) and relational aggression at each level of the moderator. The association between popularity and relational aggression declined in magnitude as the level of acceptance moved from low, β= .45, p < .005, to medium, β= .35, p < .005, to high, β= .24, p < .005. However, the effect reached significance at all levels of acceptance.
Subgroups of Popular Children
Finally, we examined evidence that there are two broad subgroups of popular children (one subgroup that is well liked and characterized by positive behavioral features and a second subgroup that is rejected and characterized by aggressive behavior). For these analyses, we included only the children in the sample who had a popularity score of .50 SD or higher (N = 54). We selected this cutoff in an attempt to balance competing concerns regarding cell-size and theoretical significance. We wanted to ensure a sufficient number of children for analysis while also retaining only those participants who were popular relative to their peers.
We focused our analysis on variable configurations for assertiveness-leadership, submissiveness-withdrawal, aggression, victimization, and social acceptance/rejection. We used a K-means clustering procedure, with a two-group solution selected a priori (consistent with the confirmatory nature of our analysis). The results of the cluster solution are summarized in Table 3. As shown, the pattern of differences between the two clusters provided support for our theory. The first cluster (N = 14) was characterized primarily by elevations on social acceptance and assertiveness-leadership. The second cluster (N = 40) had elevations on social rejection, relational victimization, and the aggression variables. A series of post hoc t tests indicated that differences between the clusters were significant for each variable (all ps < .005). The differences on the disliking score may be particularly noteworthy as the effect sizes appear to exceed those reported in studies based on western samples by a sizeable margin (Lease, Musgrove, & Axelrod, 2002).
| Variable | Cluster 1 (N = 14) | Cluster 2 (N = 40) | ||
|---|---|---|---|---|
| M | SD | M | SD | |
| 1. Social acceptance | .98 | 1.04 | −.23 | 1.06 |
| 2. Social rejection | −.31 | .55 | 1.83 | .91 |
| 3. Overt victimization | −.46 | .30 | .42 | .98 |
| 4. Relational victimization | −.27 | .46 | 1.37 | .75 |
| 5. Overt aggression | −.34 | .45 | 1.35 | 1.06 |
| 6. Relational aggression | −.11 | .71 | 1.97 | .92 |
| 7. Submissiveness-withdrawal | −.36 | .35 | .00 | .94 |
| 8. Assertiveness-leadership | 1.52 | 1.01 | −.02 | .78 |
- Note: Scores for the clusters are significantly different for each variable at .05 or better. Figures in bold were corrected (changing from negative to positive) on 8 December after online publication.
Discussion
Research conducted in western contexts has provided convincing evidence that popularity and social acceptance are partially independent dimensions of children's social experience (Parkhurst & Hopmeyer, 1998). In North American and European schools, popularity and acceptance are only moderately associated and have distinct patterns of correlates (Schwartz et al., 2006). Our findings offer an important extension to this work by demonstrating that there are similar effects for Chinese children in Hong Kong. In so far as we are aware, this study is the first to validate distinctions between social acceptance and popularity in an Asian setting.
The results of this project indicate that there is considerable correspondence in the correlates of the two forms of social standing across western and Hong Kong peer groups. Replicating investigations conducted in North America and Europe, we found that social acceptance was linked exclusively to positive behavioral features and other indicators of adjustment with peers. More specifically, well-accepted children had social reputations that included high levels of assertiveness-leadership and low levels of submissiveness-withdrawal. Social acceptance was also negatively correlated with social rejection and peer victimization. In contrast, popularity was related to a more mixed pattern of features that incorporated both prosocial and aggressive behaviors (particularly relational aggression).
Our analyses also highlight the particular relevance of relational subtypes of aggression for popularity in the Hong Kong setting. There was a positive bivariate correlation between popularity and relational aggression, although a corresponding correlation between popularity and overt aggression did not reach significance. As in western children's peer groups, popular youth in Hong Kong may employ subtle manipulative behaviors to maintain their high standing with peers (Hawley, 2003), such as exclusion and other related strategies (Prinstein & Cillessen, 2003). Interestingly, we found that these associations were attenuated at high levels of social acceptance. In the Hong Kong context, well-liked children may rely on a wider range of behaviors to achieve prestige with their peers.
At first consideration, the observed association between relational aggression and popularity might seem inconsistent with a view of Hong Kong as a collectivistic society. Why would domineering or manipulative behaviors be rewarded with social prestige in a context where children are exposed to values that emphasize group harmony and interdependence? One potential explanation for these findings might be found in Triandis's (1995) suggestion that collectivistic cultures are organized along vertical and horizontal dimensions. In vertical collectivism, individuals see themselves as unique actors contributing to the functioning of the group. Marked disparities in social power and behaviors designed to maintain social hierarchies (e.g., relational aggression) can be acceptable in so far as all members of a group contribute to the functioning of the collective. In horizontal collectivism, individuals within the group are viewed as part of a larger entity and social stratification is not desirable.
From this perspective, popularity in Hong Kong peer groups might reflect a context characterized by vertical collectivism. High-status youth may be those who serve a central organizing role in the peer group hierarchy. In fact, peer group hierarchies are encouraged in Hong Kong classrooms, with specific children formally appointed to leadership positions by instructors. We suspect that these practices underlie the strong association we observed between leadership-assertiveness scores and popularity. Children who rise to the top of the peer group hierarchy through whatever means may be seen as playing an important role in the classroom social ecology. Moreover, a vertical organization to Hong Kong society could represent an important area of overlap with some western cultural contexts.
We should also acknowledge that distinctions between collectivism and individualism have been the subject of recent controversy. For example, Fiske (2002) highlighted measurement problems and argued that cross-cultural typologies fail to take into account heterogeneity within setting. He also identified other validity concerns, noting that simple dichotomies cannot capture the intricacies of the underlying processes. In addition, theorists have wondered if a focus on cultural subtypes can fully capture the range of variability across societies (Miller, 2002). With regard to the current study, these discussions highlight the complexities involved in understanding the influence of context on the determinants of popularity. A characterization of Hong Kong society as ‘collectivistic’ might be overly simplistic.
Other theoretical viewpoints emphasize the universality of particular social phenomena. Most notably, ethological formulations have portrayed social dominance as a species-wide process that is likely to be pertinent across cultures. For example, Hawley and colleagues (e.g., Hawley, Little, & Card, 2008) have described aggression as a social strategy that is selected for by evolutionary pressures and can be successfully used to achieve status in many different social situations. An ethological model might predict that aggression and popularity would be linked even in a collectivistic culture, but it could also acknowledge that contextual factors might shape the specific topology of the relevant behaviors.
Regardless of the underlying mechanisms, it is noteworthy that we found a moderately strong association between relational aggression and popularity during middle childhood in our Hong Kong sample. Rose et al. (2004) concluded that such links do not emerge until the early years of adolescence in western children's peer groups. On the other hand, Cillessen and Mayeux (2004) found that relationally aggressive behavior was associated with popularity by the final years of elementary school. Unfortunately, the cross-sectional nature of the current study does not allow for conclusions regarding potential developmental shifts in Hong Kong peer groups.
It is also the case that some popular children in our sample were not characterized by social reputations that included high levels of either subtype of aggression. Cluster analyses revealed evidence for two distinct subgroups of popular youth. One of two subgroups comprised children who had elevations on reputational variables assessing relational and overt aggression, relational victimization, and peer rejection. In contrast, the other subgroup had relatively low aggression scores but had elevations on indices assessing assertiveness-leadership and liking by peers. To some extent, these findings replicate past work in western settings (Lease, Kennedy, & Axelrod, 2002; Rodkin et al., 2000). Research in North America and Europe suggests that there may be multiple pathways to popularity, with some high-status children relying primarily on non-aggressive strategies (de Bruyn & Cillessen, 2006).
With regard to our cluster analytic studies, one interesting set of results related to negative indicators of social relationships. In our Hong Kong sample, popular children who were aggressive were actually highly disliked and had social reputations of being the frequent targets of victimization. The relevant effect sizes were, in fact, more substantial than we would have predicted based on past cluster analyses conducted in western settings (Lease, Musgrove, & Axelrod, 2002). We suspect that this pattern reflects the strong negative valence that aggression has in Hong Kong schools. Relational aggression may be an efficient strategy for achieving social dominance over peers, but such behaviors are unlikely to lead to acceptance or liking.
In western children's peer groups, peer victimization and aggression tend to be reciprocal processes, with some children emerging as both frequent recipients and initiators of aggression (Schwartz, 2000; Schwartz, Proctor, & Chien, 2001). In so far as popular youth in any setting tend to rely on aggressive strategies, they will likely find themselves involved in social interactions that might potentiate retaliatory acts by peers. Popularity and victimization could be linked in the Hong Kong context as a result of such cyclical processes.
The overlap between peer victimization and popularity also resonates with some themes in the ethnographic literature. Ethnographers have described the aggressive jockeying for power that occurs with the context of high-status cliques (Adler & Adler, 1998). Popular youths in Hong Kong and other settings may emerge as victims of relational aggression due to this back-and-forth process.
Our analyses focusing on gender effects also produced some interesting findings. Similar to past research conducted with Chinese children (Schwartz et al., 2001), we found higher levels of relational aggression among boys than among girls. We hesitate to reach strong conclusions given some basic limitations in our measures (i.e., relational aggression was assessed with only two peer nomination items), but this pattern of results could provide important clues regarding differences in gender roles across settings. For example, some social contexts may have particularly strong sanctions for aggressive behavior (relational or overt) by girls.
Regardless of gender differences in mean levels of aggression, we did not find evidence that gender moderates associations between aggression and popularity. In western samples, relational aggression tends to be more strongly linked to popularity for girls than boys, although these effects do not tend to emerge until the early years of adolescence. Previous researchers have suggested that such findings might reflect the central nature of relational aggression in girls' interactions (Cillessen & Mayeux, 2004). In contrast, we did not find any evidence that gender moderates associations between aggression and popularity in the Hong Kong setting. It is possible that our results reflect the tendency for relationally aggressive behaviors to be relatively common among boys in this setting.
Caveats and Future Directions
The findings of this project offer an important contribution to the existing research on popularity in children's peer groups, but a number of potential shortcomings should be acknowledged. One critical issue reflects the complexities of the Hong Kong context. Families in Hong Kong often have strong ties to traditional Chinese cultural values (Yau & Smetana, 2003). Still, children in Hong Kong are likely to have greater exposure to western value systems than children living in other parts of China (Cheung-Blunden & Juang, 2008). There are also characteristics of the school system that reflect earlier structures associated with British control. For these reasons, we think replication of our findings to other sections of China might be a worthwhile goal.
A larger point is that the findings of the current project do not shed light on potential differences between settings. Analyses conducted within setting are a necessary starting point for a wider program of investigation examining the impact of contextual features. A within-setting design might also allow for some conclusions regarding replication of past patterns of findings. Nevertheless, comparative statements regarding differences in social processes should be made with extreme care.
One reason that we have avoided strong inferences regarding cultural processes is that there is certain to be considerable heterogeneity within context. Researchers can sometimes identify broad value systems that tend to be prevalent in particular social settings. It is a far more difficult task to assess the specific influences to which each child has been exposed. In our own sample, we can make some very basic statements about the nature of Hong Kong society, but there is likely to be marked variability in the extent to which each child is influenced by traditional Chinese values.
A related set of issues reflects the subtle difficulties with language and translation. As a first step in generalizing research to new cultural contexts, researchers must assume conceptual equivalence in the underlying constructs. For the current study, interpretation of our findings will be problematic if the concept of popularity is not meaningful in the setting of Hong Kong. Unfortunately, the challenges that we experienced identifying Chinese words that tap popularity and high standing emphasize the importance of these concerns. Much remains to be learned about how Hong Kong children view and understand social prestige (as it is operationalized in studies conducted with western samples) in school peer groups. In the meantime, a conservative interpretation of the present study might be that the findings simply provide evidence for a multidimensional perspective on high standing in Hong Kong peer groups. Our findings certainly do not establish conceptual equivalence in schemas regarding popularity across contexts.
Beyond these considerable complexities, there are other more basic design limitations in the project. Because the goals of the initial data collection did not incorporate a focus on subtypes of victimization and aggression, our assessments did not optimize discriminant validity between subtypes of behavior. For this reason, we included a relatively small number of items for overt and relational aggression and the correlations between the scales were high. As a related issue, all of our variables were derived from a peer nomination inventory. A multi-informant approach, perhaps including teacher rating data, would have been more optimal.
Likewise, further research with a wider range of age groups might prove to be informative. We focused our analyses on a narrow period of middle childhood based on previous investigations conducted with North American peer groups. The findings from these studies suggest that popularity and social acceptance are moderately inter-correlated during the later years of elementary school. As adolescence unfolds, disparities between the two aspects of social standing become somewhat more pronounced and gender differences emerge as a significant factor (Cillessen & Mayeux, 2004; Rose et al., 2004). Because of the limitations inherent in cross-sectional designs, we are not yet able to determine whether similar processes are relevant in the setting of Hong Kong.
In summary, the current study adds to the existing literature on popularity and social acceptance by validating distinctions between these two forms of high social standing in the Hong Kong context. Consistent with research conducted in western samples, we found that social acceptance was correlated primarily with positive behavioral characteristics and indicators of adjustment with peers. Popularity was associated with a more mixed pattern of behaviors (including aggression, assertiveness-leadership, and low levels of submissiveness-withdrawal). Moreover, cluster analyses revealed evidence for distinct subgroups of popular youth. Further research conducted with longitudinal designs and multi-informant assessments seems warranted.
