The Influence of Teachers ’ Preferences on Children ’ s Social Status in Schools

This study reconceptualizes and operationalizes peer status in elementary school classrooms by considering both teachers’ and peers’ preferences. Within this framework, a sample of 1411 students (Grades 1-6) in Hong Kong were classified into four types based on combinations of their peers’ and teachers’ preference scores: (1) popular stars—those liked by both peers and teachers; (2) teachers’ pets—those liked by teachers but not by peers; (3) students’ heroes—those liked by peers but not by teachers; and (4) rejected—those disliked by both peers and teachers. Results of MANOVA tests revealed significant differences among the four status groups in their leadership, aggression, social withdrawal, and academic performance. Path analyses showed both similarities and differences among the four status groups in terms of students’ attributes and how they predicted teachers’ and peers’ preferences.


Introduction
Existing research has shown that children who receive predominately positive nominations and few negative nominations from their peers are classified as "popular" (Kochel, Ladd, & Rudolph, 2012;Cillessen & Rose, 2005;Marks, Babcock, Cillessen, & Crick, 2013).However, in addition to contact and interaction with peers, children also form relationships with their classroom teachers which may be positive or negative depending on teachers' expectations for children's behavior (Davis & Lease, 2007).Thus, it seems reasonable to suggest that children's social status in the classroom would be related to both their peers' and teachers' preferences.Although in many studies teachers have rated children's behavior and nominated their friends (e.g., Chen, Huang, Wang, & Chang, 2012;Newcomb, Bukowski, & Pattee, 1993), rarely have both peers' and teachers' perspectives been considered when determining children's relative status within their peer group.

Extending Sociometrically-Derived Peer Statuses
Children's popularity has been defined by a high number of personal liking nominations (Newcomb et al., 1993).Although widely used, popularity inferred from peer preference nominations may not provide a full and valid representation of children's social status (Cillessen & Rose, 2005).Studies of Chinese elementary school children have shown that teachers have a direct impact on children's status through their influence on children's social reputations and peers' evaluations (Chang, 2003;Chang et al., 2007;Lu & Chang, 2013).Studies carried out in North America found that children whose behavior was consistent with their teachers' beliefs and expectations, were rated more positively by their peers than were children whose behavior was not endorsed or was negatively viewed by teachers (Hughes, Cavell, & Wilson, 2001;La Fontana & Cillessen, 2002).Other studies have shown that children use teachers' appraisals of their classmates as a reference point for determining acceptable behavior, and then they evaluate their peers accordingly (Bierman, 2011;Casiglia, Coco, & Zappulla, 1998;La Fontana & Cillessen, 2002).
Teachers and peers may also have different schema for social status in mind (Davis & Lease, 2007).Peers may prefer children who are leaders in their group, while teachers may prefer children who are well-behaved and have outstanding academic performance.Accordingly, children who are admired by their peers may not be necessarily well liked by their teachers and vice versa.For example, students who receive special preference from teachers because of the good academic performance and cooperative or compliant behavior are not always liked by the peers (Babad, 2001).On the other hand, some aggressive children are not rejected and may even be popular among peers (Cillessen & Mayeux, 2004;Poulin & Boivin, 2000;Rulison, Gest, & Loken, 2013), whereas children who behave aggressively are usually disliked by teachers (Birch & Ladd, 1998;Chang, 2003;Wentzel, 1991Wentzel, , 1993)).
Therefore, to examine both teacher and peer influences on children's peer group relations, we modified the existing methodological framework to incorporate an additional dimension.Instead of social preference and social impact, children's social status was re-conceptualized using combinations of their teacher preference and peer preference scores.As shown in Figure 1, this yielded four types: (1) popular stars: children who were liked by both peers and teachers; (2) teacher's pets: children who were liked by teachers but not peers; (3) students' heroes: children who were liked by peers but not by teachers; and (4) rejected: children who were disliked by both peers and teachers.

Social Behavior Profiles of the Four Status Groups
To validate this conceptual model, we first examined children's prosocial-leadership, aggression, and social withdrawal across the four status groups.These behaviors were of particular interest because they have been associated in the peer relations literature with children's peer acceptance or rejection (Bagwell, Newcomb, & Bukowski, 1998;Becker, McBurnett, Hinshaw, & Pfiffner, 2013) and teachers' preferences (Chang et al., 2004), although in different ways.For example, within their peer group, children who are rejected or accepted by their classmates may behave aggressively as well as socially withdrawn, while popular and well-accepted children are characterized almost exclusively by their prosocial-leadership behavior (Chang, 2003(Chang, , 2004;;Rulison et al., 2013).
Based on these findings, we developed several hypotheses.First, we predicted that children who were popular stars would have high ratings for prosocial leadership, which is preferred by both teachers and peers; low ratings for aggressive behavior, which is generally not tolerated by teachers; and low ratings for socially withdrawn behavior, which elicits mixed teacher reactions.
Second, we expected that aggressive behavior would be a defining characteristic that would contribute to students' heroes being liked by their peers but disliked by teachers.Among teachers, aggressive children are disliked because extra time and effort is needed to cope with their disruptive behavior and maintain class discipline (Bender & Smith, 1990).Thus, aggression may be the primary reason that teachers dislike student heroes although, as discussed below, they may have other desirable attributes.On the other hand, among peers, aggressive behavior may result in admiration and in turn, peers' acceptance.Children may view a peer's disruptive classroom behavior as an opposition or challenge to the teacher's authority.Thus, from the children's perspective, aggressive behavior does not present a negative image but instead represents bravery or heroism (Bukowski, Sippola, & Newcomb, 2000).
Third, we hypothesized that children who were classified as rejected (i.e., disliked by both peers and teachers) would have high ratings for social withdrawal or aggression and low ratings for prosocial leadership.Many rejected children acquire their status in part, because their bullying and aggressive behavior frequently disrupts and threatens peer interactions and relationships within the classroom.Generally, teachers dislike these children because of their troublesome conduct and peers dislike them either because they are aggressive or withdrawn.On the other hand, children who tend to be quiet, shy, or withdrawn seldom interact with peers (Rubin, Coplan, & Bowker, 2009;Rubin, Stewart, & Coplan, 1995).As a result, peers often ignore or actively reject them as companions and teachers have difficulty involving them in classroom activities.
Fourth, we predicted that social withdrawal would be the defining behavioral characteristic of teacher's pets (i.e., those liked by teachers but not peers).Teachers often sympathize with socially reticent children and tend to shelter them from their peers (Chang, 2003).In addition, due to their social withdrawal and isolation, teachers pets may experience considerably less peer contact than their classmates and thus are rejected or ignored by their peers, but are preferred by their teachers.

Academic Performance of the Four Status Groups
Apart from social interaction, children's school life consists primarily of academic activities.Thus, academic performance represents another dimension by which to profile the four status groups.Prior studies have suggested that teachers' behavior toward their students varies as a function of their academic performance.To some degree, academic performance is not only a sign of a student's intellectual competence, but it is also an indication of the teacher's effort and teaching abilities.From the teachers' point of view, students who actively participate, have a positive attitude towards their studies, and enjoy classroom activities, tend to make teachers feel good and motivated to develop a close relationship with them and vice versa (Birch & Ladd, 1997).Moreover, children who are capable of meeting the academic demands of school are more likely to form close relationships with their teachers because they are more cooperative with teachers and more motivated to carry out teachers' assignments than are low performing students (Birch & Ladd, 1997).Research supporting this view has found that highly motivated students were well liked by their teachers (Davis & Lease, 2007;Wentzel & Asher, 1995), and were often extended special privileges and considerable classroom freedom than were their counterparts.In the eyes of their classmates, these students are often referred to as teacher's pets.
A number of studies carried out in North America (Guay, Boivin, & Hodges, 1999;Wentzel & Asher, 1995) and China (Chen, Chang, & He, 2003;Chen, Chang, He, & Liu, 2005) reported an association between children's academic performance and their peer acceptance.In a longitudinal study, Welsh, Parke, Widaman, and O'Neil (2001) showed that children's academic performance in Grade 1 was positively correlated with peer liking, and negatively correlated with peer disliking in Grade 2. The same pattern was found from Grade 2 to Grade 3.These findings suggest that high academic achievement provides children with a positive social reputation and the associated prestige, both of which positively impact their popularity among peers.On the other hand, children with low academic performance are often rejected and negatively perceived by their peers (Zettergren, 2003).
Given that academic performance is highly valued in Chinese society, this may be a fairly robust predictor of Hong Kong children's social status.Research to support this contention is apparent in Chen, Dong, and Zhou's (1997) work which found that outstanding academic performance had a positive impact on children's peer nominations for positive behavior (e.g., prosocial-leadership), a determinate of teacher and peer preference.Therefore, we hypothesized that children who were classified as teacher's pets and popular stars would perform better academically than would students' heroes and rejected children.

Relations Among Children's Social Behaviors and Status Group Classifications
The behavioral attributes which contribute to children's status in the classroom can be illustrated in a model and applied in the prediction of teachers' preferences and peer's acceptance.As discussed above, teachers favor academic high achievers, endorse prosocial leadership, dislike and reject aggression, and are somewhat sympathetic to social withdrawal.Based on this pattern of findings, we expected invariance across the four status groups in the association between teachers' preference and the three social behaviors and academic performance, respectively, and in the directions consistent with the above review.Among these three social behaviors, aggression may have the strongest negative impact on teachers' preference primarily because they hold aversive attitudes towards disruptive behavior (Chang, 2003).Children who are classified as rejected or as students' heroes may be more aggressive than those who are classified as teacher's pets and popular stars.Thus, we predicted that the negative impact of aggression on teachers' preference would be stronger for children who were classified as rejected or as students' heroes than for those classified as teacher's pets and popular stars.
Our conceptual model of expanded peer status may be able to explain some of these discrepant results.For example, aggression may evoke different peer reactions because the behavior has a different connotation depending on which children in the class are behaving aggressively, their relative status within the peer group, and their associated peer and teacher reputations.That is, students' heroes who aggress or rebel against teachers may be appreciated by peers and accorded the heroes' status.In this case, their aggressive behavior is perceived positively and contributes to their peer acceptance.On the other hand, aggressive behavior that involves direct physical harm to peers, such as hitting or fighting, generally leads to peer rejection.Accordingly, we predicted a negative association between children's aggressive behavior and peer acceptance for children who were classified as rejected, but predicted a positive association between aggressive behavior and peer acceptance for children classified as students' heroes.However, for all groups, we expected a positive association between prosocial leadership and peer acceptance, and a negative association between social withdrawal and peer acceptance.
As reviewed earlier, the relation between children's academic performance and their peer acceptance has been consistently positive in studies carried out with North American (Welsh et al., 2001) and Chinese children (Chen, He, & Li, 2004).However, given the strong emphasis on educational achievement in Chinese culture, the quality of children's peer interactions inevitably relies on their academic performance (Chen, Dong, & Zhou, 1997).As a result, we predicted that academic performance would be positively correlated with peer acceptance in the four status groups, albeit with some variation.That is, children who are high achievers may be the popular stars or the teacher's pets in their peer groups, whereas children who tend to perform poorly in their studies or give up on academics completely, may be rejected.Thus, we expected that the positive impact of academic performance on peer acceptance would be stronger for children classified as popular stars and teacher's pets than for the students' heroes and rejected children.

Developmental Differences among Status Groups
We also examined developmental differences in children's social status classifications.There is some evidence to suggest that teachers may be more likely to provide physical care and nurturance to younger children which promotes a close teacher-student relationship (Lynch & Cicchetti, 1997).On the other hand, with the onset of puberty and the increased importance of peer relations, older children may be eager to assert their autonomy (Berk, 1997;Bukowski et al., 2000;La Fontana & Cillessen, 2010).Thus, we predicted that children classified as teacher's pets would be higher among the younger (first and second graders) than among the older children (third to sixth graders) and we expected that children classified as students' heroes would be higher among the older rather than the younger children.
Apart from examining age differences in children's status group classification, we were also interested in exploring whether children's social behavior, differentiated by age, would be associated with their status group classification.For younger children, aggressive behavior may be associated with peer rejection and teacher disliking.Yet, this may not hold true for older children because overt aggressive behavior decreases with age and other social behaviors become more prominent.Therefore, we predicted that aggressive behavior would have a stronger negative impact on children's peer preference in the lower than in the higher grades, whereas we expected the pattern for social withdrawal would be the reverse.Social withdrawal among younger children may not be considered a problematic behavior or not to an extent that leads to peer rejection (Chen et al., 1995;Coplan, Prakash, O'Neil, & Armer, 2004), while older children who are withdrawn, are likely to be rejected by their peers (Chang et al., 2004).Therefore, social withdrawal was expected to have a stronger negative impact on peer preference among children in the higher grades than in lower grades.

Gender Differences
Gender was not a focus of the present study.However, potential gender differences were examined.Current research on teacher-child relationships suggests that teachers maintain closer and more positive relationships with girls than with boys in kindergarten (Birch & Ladd, 1998), throughout elementary school (Hamre & Pianta, 2001), and middle school (Davis & Lease, 2007).To some extent, this may be attributed to the notion that girls are more likely to conform to adult values, including those of their teachers (Maccoby, 2000) than are boys.On the other hand, the teacher-child relationship may be more predictive of peer preference for boys than girls.Some studies have found that teachers' ratings of their students were more strongly related to boys' popularity than to girls' popularity (La Greca, 1981), yet other researchers have reported few gender differences (White & Jones, 2000;White & Kistner, 1992).Given these discrepant findings we took a more exploratory approach.We predicted that more girls than boys would be classified as teachers' pets, especially for younger children, whereas more boys than girls would be classified as students' heroes or rejected.In addition, we expected that gender differences within the four status groups would be consistent with prior results based on sociometric methods.For example, independent of status group membership, girls were expected to have higher academic performance than boys, boys were expected to be more aggressive than girls, and the relation between aggression and peer acceptance was predicted to be more positive for boys than for girls.

The Present Study
In the present study we operationalized Hong Kong children's peer status in their primary school classrooms according to four classifications (i.e., popular stars, teachers' pets, students' heroes, and rejected; see Figure 1).To validate these classifications, we examined status-group differences and similarities in children's social behaviors and academic performance, and explored the relations among children's social behaviors and academic performance and their teachers' and peers' preferences.We also considered grade-related developmental differences and possible gender interactions.

Participants
Students and teachers were recruited from three randomly selected primary schools in Hong Kong.The sample consisted of 1411 children (59% boys; aged 6 to 14 years (M=9.69;SD=2.00; 56% between 10 and 14) from 54, grade 1 to grade 6 classes and their 54 classroom teachers.All three schools were government-funded and located in public housing estates; representing a lower middle class background in Hong Kong.The schools were considered to be typical of the Hong Kong setting in terms of children's academic performance and social class.Parents, teachers and students were told the purpose of the study was to understand children's social behavior.Parents' and teachers' consents were obtained for all participants.

Teacher and Peer Preference
Teachers were asked to rate their preference for each student in their classroom using a 5-point scale (1=least preferred; 5=most preferred).Teachers' ratings were standardized within class and a total teacher preference score was calculated for each child.
In each class, students were given a roster and asked to circle the names of all their friends.This was an unlimited within-class nomination of friends.This approach has been shown to yield results similar to those from limited nominations (Bukowski, Pizzamiglio, Newcomeb, & Hoza, 1996).Peers' ratings were standardized within each class and a total peer preference score was calculated for each child.

Social Behaviors
Assessments of children's social behaviors were obtained using a peer nomination measure for Chinese children developed by Schwartz, Chang, and Farver (2001).Using this measure, students were asked to nominate three children in their class who best fit the descriptions of three types of social behavior: prosocial-leadership (i.e., kids who are leaders, are helpful, are listened to when speaking up, they organize plays, and lead others (alpha=.87));aggression (i.e., kids who start fights, hit or push, bully, say mean things to, and disobey rules (alpha=.97));and social withdrawal (i.e., kids who are often alone, are shy and quiet, avoid social interaction, do not speak much, and do not join others in play (alpha=.92)).Summary scores for these three items were generated for each child based on the frequencies of being nominated for the relevant items and standardized within class.

Academic Performance
Academic performance was based on students' average scores in the Chinese language, English, and Mathematics at the end of the school term (maximum mark=100).Students' scores in each subject were first standardized within the grade of their corresponding school, then summed, and divided by three.

Procedure
Two research assistants administered the questionnaires in each classroom in a standardized manner without the presence of the classroom teachers or other adults.Students were informed that their participation was voluntary and their responses would be kept confidential.For students in Grade 1 and 2, the research assistants read each item aloud to ensure they understood the procedure.Children were given McDonald's Restaurant coupons for their participation.

Peer Status Classifications
We determined children's peer status classification using a strategy developed by Gormon et al. (2002).First, we standardized both the peer nomination scores and the teacher preference ratings within classrooms.In order not to lose cases, we used the mean rather than half a standard deviation above the mean (Gormon et al., 2002) as the cutting score.Using this method, 407 children (29% of the sample) with positive scores for both their teachers' preference ratings and peers' nominations were classified as popular stars (52% were girls); 382 (27% of the sample) with positive teacher preference scores and negative peer nomination scores were classified as teacher's pets (49% were girls); 219 (16% of the sample) with positive peer nomination scores and negative teacher preference scores were classified as students' heroes (29% were girls); and 403 (28% of the sample) who had negative scores on both teachers' and peers' preference were classified as rejected (31% were girls).These classifications are presented in Figure 2 showing the scatterplot of standardized teacher preference by peer preference scores.
Figure 2. Scatter plot of standardized teacher preference by peer preference scores

Developmental Differences Among Status Groups
To examine gender and developmental differences in children's status classifications, percentage distribution were calculated.The results showed that more children were classified as students' heroes in grades 3 to 6 (17%) than in grades 1 and 2 (11%).For both grades, more boys (70%) were classified as students' heroes than were girls.The percentage of popular stars was also moderately higher in grades 3 to 6 (30%) than in grades 1 and 2 (26%).However, within the two grade categories, there were no gender differences.The percentage of teachers' pets was moderately higher in grades 1 and 2 (31%) than in grades 3 to 6 (25%) and there was a gender-by-grade interaction.For grades 3 to 6, 54% of the teachers' pets were girls, whereas only 39% of the girls in grades 1 and 2 were classified as such.Finally, more children were classified as rejected in grades 1 and 2 (33%) than in grades 3 to 6 (27%), and for both grade levels more boys (70%) were classified as rejected.
There were no significant results for grade level and the three social behaviors.However, there was a significant grade level by status group interaction for social withdrawal, F (3, 1395) =5.29, p<.01.Students' heroes (.12) and popular stars (-.26) in the lower grades had higher social withdrawal scores than did teachers' pets and rejected children in the higher grades (Means=-.70,-.72 for the status groups respectively); (Means=.72,1.12 for the two status groups respectively) (Means=.14 and .68 for the two status groups respectively).(.35) and teachers' pets (.17) had higher academic performance than did the students' heroes (-.07) and rejected children (-.48).There were no significant interactions for grade level and gender by status group, or gender by status group.

Children's Social Behavior, Academic Performance, and Peer and Teacher Preferences
Invariance tests in path analyses were conducted to examine whether the influence of children's social behaviors and academic performance on peer and teacher preference would differ among the four status groups and to explore grade-related differences in these associations.Chi-square difference tests were used to compare a constrained model where a specific path was constrained to be equal across the four status groups, against an unconstrained model where all the paths were freely estimated.Table 3 shows the path coefficients for the four status groups from the unconstrained model.The invariance tests yielded significant chi-square differences with respect to the paths leading from prosocial leadership ( 2 =21.81, df=3, p<.001), social withdrawal ( 2 =16.32, df=3, p<.001), and academic performance ( 2 =8.36, df=3, p<.05) to peer preference.For prosocial-leadership, the rejected group differed significantly from the other three groups.As shown in Table 3, the  for the rejected group was .69,whereas those of other status groups were around .30.For social withdrawal, 's for the teachers' pets (=-.17) and the rejected children (-.31) were lower than the 's for the popular stars (-.07) and students' heroes (-.01).The path from academic performance to peer preference was statistically significant only for teachers' pets (.11) but not the rest three status groups.
Among the paths leading to teacher preference, only the path from aggression differed among the four status groups,  2 =33.68, df=3, p<.001.This was significant for the popular stars (-.17), students' heroes (-.29) and the rejected children (-.45), but not for teacher's pets (-.07).When comparing the popular stars and teachers' pets with the rejected children and students' heroes, there was a significant difference.The negative impact of aggression on teacher preference for the popular stars and teachers' pets was weaker than those for the rejected children and students' heroes.
With regard to developmental differences Table 4 shows the path coefficients of the four status groups separated by the lower (Grades 1 and 2) and higher grades (Grades 3 to 6).A grade invariance test within each status group was conducted to examine whether the paths fit both lower and higher grades equally well.Some grade-by-status interactions were found.For example, among the higher grades, there were significant differences in the paths leading from leadership ( 2 =13.58, df=3, p<.01), aggression ( 2 =11.08, df=3, p<.05) and social withdrawal ( 2 =14.18, df=3, p<.01) to peer preference.As shown in Table 3, the rejected group differed from the other three groups among the higher grades.However, for the lower grades, there were no differences among the four status groups.
Similarly, significant chi-square differences were found in the paths leading from aggression ( 2 =10.95, df=3, p<.05) and academic performance ( 2 =8.50, df=3, p<.05) to teacher preference among the higher but not the lower grades.Again, the rejected group had the highest negative coefficient (=-.41).Grade differences were also found for the rejected group; i.e., the paths leading from aggression ( 2 =10.95, df=3, p<.05) and from social withdrawal ( 2 =8.5, df=3, p<.05) to peer acceptance were significant.The negative impact of aggression (-.25) and social withdrawal (-.36) on peer acceptance among higher grades was stronger than those for lower grades (-.18 and -.14 respectively for aggression and social withdrawal).

Behavioral Profiles
In the present study, we proposed and tested a re-conceptualized model for examining children's peer group status.We classified children into four groups using a combination of their teachers' and peers' preference scores to obtain what we refer to as popular stars, teacher's pets, students' heroes, and rejected children.Consistent with our hypotheses, the results showed that each status group was characterized by particular social behaviors.Children who were classified as popular stars were rated high for prosocial-leadership, teachers' pets were rated high for social withdrawal, students' heroes were rated high for aggressive behavior and prosocial-leadership, and rejected children were rated high for aggressive and social withdrawn behavior.
For the popular stars and students' heroes, prosocial-leadership was an attribute of their behavioral profiles.Undoubtedly, the ability to lead others is revered and appreciated but as the results showed, there was a clear difference in leadership style which distinguished popular stars from students' heroes.Popular stars' leadership status among their peers may have been supported by their role as class representatives or monitors.In Hong Kong primary schools, teachers choose students who have good conduct and outstanding academic performance to serve as class representatives.Thus, as the teacher sanctioned positive role models, the popular stars were likely to be positively viewed by most if not all of their peers and teachers.
Students' heroes on the other hand, may have derived their leadership status by challenging their classroom teachers' authority.In this case, the peer group endorsed rather than rejected the aggressive behavior characteristic of the students' heroes.It is possible that some degree of social intelligence allows these children to establish and maintain high peer status (Salmivalli et al., 2000).By using both aggressive and prosocial means of control, students' heroes can in effect, become the center of their social group, and well-liked by peers, despite their aggressive behavior (Hawley, 2003) and regardless of teachers' preferences.Although popular stars and students' heroes might have similar high levels of social competence, Rodkin and Roisman (2010) found that teachers tended to evaluate students' heroes (tough children) as having poorer social skills than popular stars ("model" children).This evaluation might result from teachers' frustration with students' heroes.Peers who emphasize peer interaction quality might view both popular stars and students' heroes as socially competent.
Among the children classified as teacher's pets and rejected social withdrawal was a defining feature of their behavioral profiles.In contrast to studies reporting an association between social withdrawal and peer disliking; other research suggests that it may also be associated with sympathy (Craig, Henderson, & Murphy, 2000) on the part of teachers, particularly in the Chinese context.In Chinese schools, students are expected to keep silent during class and to follow the teacher's instructions.Therefore, students who seldom talk in class and sit quietly by themselves (i.e., behaviors often associated with social withdrawal) may be labeled as good students by their teachers.Teachers may give these students additional attention such that they become teacher's pets.However, not all students who are socially withdrawn will attract a teacher's attention.Some may be viewed as being too dependent on the teacher or timid in other social situations, and they are disliked by their teachers and rejected by peers for this behavior.Thus, it would appear that the teachers' view and interpretation of a particular child's withdrawn behavior can play an important role in determining the child's status within the peer group.
As predicted, children classified as popular stars and teacher's pets had better academic performance than children who were classified as students' heroes or rejected.This finding is consistent with the literature showing that high academic achievers were usually popular among peers and teachers and were often designated as teachers' pets (Babad, 2000;Chiu, Lee, & Liang, 2013;Moore et al., 2012;Tal & Babad, 1989, 1990).Because students who perform well academically give teachers much satisfaction, it was not surprising to find that teachers showed a preference towards them.On the other hand, the students' heroes and the rejected children not only had high negative behavior ratings but also poor academic performance which was reflected in their negative teacher preference scores.Children's academic performance appears to be a critical indicator of teacher preference.
There was also an interesting pattern of gender difference which was somewhat consistent with our predictions.
In line with our findings that boys had higher ratings than girls for aggressive behavior and social withdrawal, 70% of the children classified as students' heroes or rejected were boys, and in the lower grades more boys than girls were classified as teachers' pets.These results may also reflect the teachers' perspectives in terms valued classroom demeanor and the associated expectations for boys and girls.Typically, in Chinese contexts, boys are perceived as being more active and avid participators in class lessons and activities than are girls.Thus, boys who are not particularly disruptive or aggressive, may receive more preferential treatment from their teachers than girls are generally viewed as being typically shy and quiet.However, among higher grades, boys in general may become less attractive to teachers due to their disruptive behavior and poorer academic performance as compared with girls (Guo & Chang, 2003) which might account for more girls being classified as teachers' pets in the higher than in the lower grades.

Developmental Differences
The results of our tests for developmental differences in children's status groups partially supported our hypotheses.More specifically, in contrast to the lower grades, we found in grades 3 to 6 more children were classified as popular stars and students' heroes and less were classified as rejected.These findings suggest that children's peer preference increased with age and could be associated with the widening social circles characteristic of older children.With maturity, children may learn that tolerance and acceptance are necessary to maintain friendships and thus they were less likely to reject many of their peers.Furthermore, during the late primary school years, children may become more independent and autonomous; a key factor that may have influenced their appreciation of and preference for children like students' heroes who challenge authority figures, such as their classroom teachers.In fact, aggressive and assertive behaviors were sometimes associated with peer acceptance (Xu, Farver, Schwartz, & Chang, 2003).This was particularly true for male students of upper grades (Guo & Chang, 2003;Zeller, Vannatta, Schafer, & Noll, 2003).Zeller et al. (2003) further suggested that older children in general view aggressive behaviors less negatively than younger children do.This also might account for the higher proportion of students' heroes in the higher grade levels.
The results also showed a significant grade by status interaction for children's social withdrawal.In the higher grade levels, children classified as popular stars and students' heroes were less socially withdrawn than were children in the lower grade levels.By contrast, children classified as teacher's pets and rejected were more socially withdrawn in the higher than in the lower grade levels.This finding implies that socially withdrawn children in the higher grades were rejected by their peers.As children develop, their behavioral profiles may remain stable or change.In fact, many young children may be socially withdrawn, but when they reach higher grade levels, this behavior diminishes and they are less shy or submissive.Therefore, for children in the lower grades, social withdrawal may not necessarily lead to peer rejection, but it is undesirable among children in the higher grades.

Predictions of Teacher and Peer Preferences among Status Groups
Results of the path analyses showed that aggression had a negative impact on teacher and peer preference, regardless of the children's peer status classification.However, at the same time, aggression had a differential impact on teachers' preferences across the four status groups.In line with our hypotheses, the negative impact of aggression on teacher preference was stronger for students' heroes and rejected children when compared to popular stars and teacher's pets.Thus, the degree and nature of aggressive behavior seems to vary among children within the peer status groups.For example, children classified as rejected were rated as being more physically aggressive (e.g., hitting or pushing others) as well as relationally aggressive (e.g., says mean things to others) than were the popular stars and teacher's pets.From the teachers' point of view, classroom aggression interferes with their work, and these behaviors not only challenge their authority (Kedar-Voivodas, 1983) but also threaten the personal safety of others.Teachers have the responsibility to protect their students from physical injury inflicted by other students and there may be considerable pressure on the teachers when they must deal with it.This appeared to be reflected in the stronger impact of aggression on teachers' preferences for the rejected children than for their counterparts.
The current study represents an initial step in understanding how the combined influence of peers' and teachers' preferences can shape children's social status in their primary school classrooms.Clearly, peers and teachers make significant contributions to children's status within their peer groups and additional work is needed to further clarify the mechanisms of this relation.
However, the current findings should be interpreted in light of several limitations.First, children's social behavior was measured using peers' nominations and the use of multiple informants could broaden and enhance the description of children's behavioral profiles.Peer nominations could be a source of bias because some students may not have been familiar with the social behavior of all of their classmates and they instead nominated only those who were well-known within the class.For example, they might nominate someone as aggressive if he/she had been frequently punished or singled out by their teachers, rather than actually identifying, which student did in fact behave aggressively.To strengthen associations between teacher preferences and children's aggressive behavior, future research should use more objective measures of students' social behaviors and classroom observations.If children's classroom or playground behaviors could be video taped, detailed behavioral profiles of children in each group could be obtained.Also, teachers' ratings of children's social behaviors could also be included.
Future studies might also collect similar data with adolescents to examine developmental trends in the salience of peer and teacher preferences on children's status within their peer groups.Moreover, longitudinal studies could examine change over time in children's behavior and status.Existing work with peer status (Cillessen & Mayeux, 2004;Pettit, Clawson, Dodge, & Bates, 1996) has shown that children's peer status may improve or worsen across grades.Factors which may contribute to these changes are of research interest, particularly in Chinese populations where little longitudinal work has been done.
In conclusion, this study is among the first to incorporate teachers' preferences into a system to determine children's peer group status.By investigating the behavioral profiles of students in the four status groups, this project began to isolate some important aspects of these behavioral profiles.Clearly, teachers' impact on children's peer status deserves additional research attention.

Figure 1 .
Figure 1.Conceptual model for determining the four categories of sociometric status

Table 1 .
Means and standard deviations for the variables Behavior and Academic Performance Profiles of the Four Peer Status Groups

Table 2 .
Mean level of children's social behaviors and academic performance by status group Univariate analyses examining children's academic performance revealed a significant effect for status group, F (3, 1395) =64.60, p<.001.Post hoc tests showed that popular stars

Table 3 .
Standardized path coefficients from children's social behaviors and academic performance to peers' and teacher's preference by status group

Table 4 .
Standardized path coefficients from children's social behaviors and academic performance to peers' and teacher's preference by grade and status group