Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
J Abnorm Child Psychol. Author manuscript; available in PMC 2013 October 19.
Published in final edited form as:
PMCID: PMC3800108

The Best Friendships of Shy/Withdrawn Children: Prevalence, Stability, and Relationship Quality


The mutual best friendships of shy/withdrawn and control children were examined for prevalence, stability, best friend’s characteristics, and friendship quality. Using peer nominations of shy/socially withdrawn and aggressive behaviors, two groups of children were identified from a normative sample of fifth-grade children: shy/withdrawn (n = 169) and control (nonaggressive/nonwithdrawn; n = 163). Friendship nominations, teacher reports, and friendship quality data were gathered. Results revealed that shy/withdrawn children were as likely as control children to have mutual stable best friendships. Withdrawn children’s friends were more withdrawn and victimized than were the control children’s best friends; further, similarities in social withdrawal and peer victimization were revealed for withdrawn children and their friends. Withdrawn children and their friends reported lower friendship quality than did control children. Results highlight the importance of both quantitative and qualitative measures of friendship when considering relationships as risk and/or protective factors.

Keywords: social withdrawal, friendship, adjustment

Long ago, Jean Piaget and Harry Stack Sullivan posited that peer relationships provide children with a unique context for emotional and social development (Piaget, 1932; Sullivan, 1953). Piaget maintained that the symmetrical power relationship unique to peer relationships afforded children the opportunity to develop perspective-taking abilities, social competence, and advanced moral reasoning. Sullivan emphasized the intimacy of children’s same-gender chumships, arguing that such intimacy promotes identity development and contributes to later successes in romantic relationships. The implication of their arguments was that children who were not involved in peer relationships would miss out on developmental opportunities important for positive adjustment and growth (Rubin, Burgess, Kennedy, & Stewart, 2003).

Today, the peer relationships literature supports Piaget’s and Sullivan’s early contentions, clearly demonstrating the significance of peer relationships, particularly friendships, in children’s emotional and social development (see Rubin, Bukowski, & Parker, 1998 for a relevant review). For example, investigators have demonstrated that friendship promotes the development of perspective-taking and moral reasoning skills (e.g., McGuire & Weisz, 1982). Friendship also has been positively associated with measures of self-esteem and feelings of general self-worth (e.g., Bagwell, Newcomb, & Bukowski, 1998) and is considered an important source of social support, particularly under potentially stressful situations (Berndt & Keefe, 1995; Ladd, 1990). Further, researchers have shown that friendship can protect children from the negative externalizing and internalizing “costs” associated with peer victimization (e.g., Hodges, Boivin, Vitaro, & Bukowski, 1999).

It also has been shown that children who are without close peer relationships altogether or those who have difficulties with their peers, often experience difficulties of the social and emotional ilk. Specifically, researchers have demonstrated that children without friends suffer from loneliness (Brendgen, Vitaro, & Bukowski, 2000; Parker & Seal, 1996) and are often socially unskilled (Clark & Drewry, 1985). Chronic friendlessness during childhood has been associated with social timidity, sensitivity (Parker & Seal, 1996), and later internalizing problems (Ladd & Troop-Gordon, 2003). Like friendless children, rejected children report greater loneliness and more negative feelings about the self than do their nonrejected peers (e.g., Ladd & Troop-Gordon, 2003). Poorly accepted children also often have friendships that are qualitatively impoverished (Brendgen, Little, & Krappmann, 2000).

Although much attention has been paid to rejected children, few researchers have considered the close peer relationships of shy and socially withdrawn children. Yet, like peer-rejected children, shy and socially withdrawn children are often on the “outside” or the periphery of the social scene. It is true that the reasons why shy and socially withdrawn children are not involved with their peers differ from those of many rejected children. Rather than being actively isolated by the peer group, shy and socially withdrawn children often actively isolate themselves because they are socially anxious in the company of others (e.g. Gazelle & Rudolph, 2004). Thus, shy and socially withdrawn children may often be physically removed and isolated from their peers and may miss out on the many benefits of close peer relationships.

There is a considerable amount of research demonstrating that shyness and social withdrawal are associated with significant psychosocial maladjustment and difficulties with peers (see Rubin et al., 2003 for a review). Specifically, socially withdrawn children and young adolescents who are socially anxious and shy (that is, they passively withdraw from peers; Hymel, Rubin, Rowden, & LeMare, 1990) have negative self-perceptions of their social competence and their relationships, experience loneliness, and suffer from depressive symptoms (e.g., Boivin & Hymel, 1997; Gazelle & Ladd, 2003; Nilzon & Palmerus, 1998; Rubin, Chen, McDougall, Bowker, & McKinnon, 1995). Also, relative to nonwithdrawn children, young socially anxious, shy, withdrawn children are socially unskilled (e.g., Gazelle & Rudolph, 2004; Stewart & Rubin, 1995). It is important to note that shyness and social withdrawal become increasingly salient and negative to peers with age. In early childhood, shy and socially withdrawn children do not appear to be rejected by their classmates (Hart et al., 2000; Ladd & Burgess, 1999); however, by mid-to-late childhood and early adolescence, many shy, socially withdrawn children do become rejected by their peers (Boivin, Hymel, & Bukowski, 1995; Rubin, Chen, & Hymel, 1993). Moreover, shy/withdrawn children are also often the targets of peer victimization (Hanish & Guerra, 2004; Hodges, Malone, & Perry, 1997; Olweus, 1993).

While it is clear from the aforementioned findings that shyness and social withdrawal are associated with adjustment difficulties at both the individual (e.g., internalizing problems) and group (e.g., peer rejection) levels of social complexity, less is known about the relation between shyness and social withdrawal and adjustment at the dyadic relationship level (Hinde, 1987). Shy and socially withdrawn children may remove themselves from groups of peers due to discomfort and social anxiety; it is not known whether this discomfort is also overwhelming and intolerable when in the company of just a few or even one close peer. Also, if shy and socially withdrawn children do acquire and maintain best friendships, then who do they befriend, and how can their friendships be characterized? Are their friendships long-lasting? Are they of high or low relationship quality? Few researchers have considered social withdrawal at the dyadic level of social complexity (e.g., Schneider, 1999); as a result, very little is known about the friendships of socially withdrawn children. Given the many putative advantages of participation in friendship and the risks associated with not being involved in such dyadic relationships, an examination of shy and socially withdrawn children’s participation in such relationships could further our understanding of the specific risks associated with being shy and socially withdrawn during late childhood.

The association between social withdrawal and rejection by the peer group at-large is well documented in the developmental psychology literature (Newcomb, Bukowski, & Pattee, 1993); however, it is not clear whether some of the reported difficulties associated with shyness and social withdrawal can be explained by shy/withdrawn children’s involvement, or lack thereof, in close dyadic relationships. Does their social withdrawal preclude them from garnering the many advantages of friendship? Such information could increase our knowledge regarding the social “costs” of being shy and socially withdrawn and could in turn, improve the success of interventions. To date, fewer than a handful of investigators have examined the close friendships of shy and socially withdrawn children (e.g., Fordham & Stevenson-Hinde, 1999; Schneider, 1999). These studies either focus on the period of early childhood or they comprise rather small samples of withdrawn children. Thus, the overarching goal of the present study was to examine the characteristics of the friendships of shy and socially withdrawn children during their final year of elementary school (approximately 10 years of age).

Our first research question— what is the prevalence of friendship among shy and socially withdrawn children? Most children and adolescents have mutual best friendships; for example, Parker and Asher (1993), reported that approximately 70% of children have a best friend — at least when friendship is defined as a reciprocated nomination among three possible friendship nominations. Given these findings, one might expect that many shy and socially withdrawn children may not lack best friendships. Indeed, although few studies pertaining to prevalence exist, it has been found that children characterized by their social wariness and withdrawal have as many mutual friendships as their nonwithdrawn classmates. For example, some researchers have indicated that the prevalence of best friendships among young socially withdrawn children is nonsignificantly different from that of nonwithdrawn children (Ladd & Burgess, 1999). But, as noted above, social withdrawal in early childhood is neither a particularly negative nor salient characteristic. Thus, the Ladd and Burgess finding may not be too surprising.

In his research with slightly older children, Schneider (1999) found that 60% of withdrawn 8- and 9-year-olds had a reciprocated friendship. Notably, Schneider allowed children to nominate an unlimited number of “good friends,” and while precedence was given to those dyads in which both members nominated one another as the first good friend, he also included other mutual friendships. Given the reputed advantages of friendship noted above, it would appear important to extend previous studies of the prevalence of mutually nominated friendships among withdrawn children. Thus, the present study was unique in its focus on the best friendships of withdrawn children, especially given that best friendships or close friendships have been shown to influence children’s adjustment more than other good (but not “best”) friends or the peer group at-large (Berndt, 1999). We hypothesized that withdrawn children would be less likely than nonwithdrawn children to have a mutual best friendship during late childhood; we reasoned that the strong relation between social withdrawal and peer rejection at this age may lead peers to judge the shy/withdrawn child as being less attractive as a potential best friend, thereby decreasing the likelihood that the shy/withdrawn child would experience a mutual best friendship. However, based on the extant literature regarding the benefits of friendship involvement, we also expected that those shy/withdrawn children who had a best friendship would be less victimized, more socially accepted, and viewed as more socially competent than those shy/withdrawn children without such relationships.

Who do shy and socially withdrawn children befriend?

In existing studies of peer relationships, researchers have considered the identity or characteristics of children’s best friends. They have done so because it may be that the company children keep influences their psychosocial adjustment. The “homophily” hypothesis suggests that children are attracted to, and become friends with, others who are similar to them. Haselager, Hartup, van Lieshout, and Riksen-Walraven (1998) found behavioral similarities to be greater between friends than non-friends in late childhood, with the relation particularly strong for aggressive and antisocial behaviors. These researchers also found that children’s friendships were with others who resembled them in terms of prosocial behavior and shyness/dependency. Haselager et al. suggested that the developmental significance of friendship might vary as a function of the characteristics of the child, the friend, and the similarity between the two. Specifically, they posited that friendship might function protectively when children are similar to their friends in terms of prosocial tendencies. Alternatively, friendship can be a risk factor when there are behavioral concordances in aggression and displays of antisocial behaviors, a view that has been supported in empirical studies by Dishion and colleagues (e.g., Dishion, Andrews, & Crosby, 1995).

Importantly, these findings also provide support for the “reputationalsalience hypothesis,” or the notion that similarities between friends vary according to the salience of the attribute in determining the social reputations of the children involved (Haselager et al., 1998). In other words, it would be expected that aggressive children would be more likely to have friends who were highly aggressive, and withdrawn children would have friends who were highly withdrawn because such similarities would be consistent with their reputations. Moreover, in light of recent research suggesting that withdrawn children are frequently victimized by their peers (e.g., Hanish & Guerra, 2004; Hodges et al., 1997), it may be the case that the best friends of withdrawn children likewise experience being bullied. Whereas some researchers have investigated this hypothesis for aggressive children (e.g. Poulin & Boivin, 2000), it has not yet been examined for shy and socially withdrawn children and their best friends.

Just as the prevalence of best friendships and the characteristics of the best friends are important to consider, so too are the relationship qualities of these friendships (Rubin et al., 1998). Therefore, we asked in the present study: Are the best friendships of shy and socially withdrawn children stable? Are there qualitative differences in the friendships of socially withdrawn and nonwithdrawn children? Although several researchers have examined the effects of a stable friendship on psychosocial functioning and adjustment (Berndt, Hawkins, & Jiao, 1999; Parker & Seal, 1996), few have studied the relations between individual characteristics of children and friendship maintenance (Ladd & Burgess, 1999). Consequently, little is known about the stability of the friendships of socially wary and withdrawn children. Schneider (1999) found no differences in the stability of friendship dyads comprising zero, one, or two withdrawn children across the school year. Unfortunately, his sample of dyads that included at least one withdrawn child was rather small, thereby limiting the generalizability of his findings. Schneider noted, however, that withdrawn children were less communicative with their best friends than were nonwithdrawn children. Given that verbal communication is necessary for intimate disclosure and important for the maintenance of close personal relationships, it appears reasonable to expect that socially withdrawn children may struggle to maintain their friendships across the school year.

In terms of relationship quality, we were interested specifically in whether children’s subjective perceptions of friendship quality varied as a function of their shy, withdrawn status. Some researchers who have studied friendship quality in relation to shyness and social withdrawal have found that socially withdrawn children have friendships that are relatively high in relationship quality (Fordham & Stevenson-Hinde, 1999; Schneider, 1999). However, as noted above, results from the Schneider study also indicated lowered levels of verbal communication in the friendships of shy/withdrawn children. Further examination of the qualities of shy/withdrawn children’s friendships is required to better understand the ways in which shyness might influence relationship qualities. We reasoned that the overall quiet and reserved nature of withdrawn children and their reported verbal reticence, during the developmental period of late childhood when intimacy and self-disclosure in close relationships become central, would negatively influence the amount of intimate exchange, help, and companionship within the friendship, and the overall quality of the relationship. The exchange of ideas is not only important to mutual understanding of thoughts, feelings, and emotions but also for the planning of everyday friendship activities involving fun and recreation.

Lastly, it may be the case that important gender differences exist in the best friendships of shy and socially withdrawn children. Socially withdrawn boys and girls may experience different successes and difficulties in their best friendships. Researchers have shown that it may be more socially acceptable to be a shy and quiet girl than it is to be a shy and quiet boy (Hinde, Stevenson-Hinde, & Tamplin, 1985). This finding may be especially true once children move into early adolescence and conformity and adherence to gender roles becomes particularly important (Ruble & Martin, 1998). There is some evidence to support the suggestion that social withdrawal is associated with greater psychosocial maladjustment, for boys than girls, particularly during adolescence (e.g., Caspi, Elder, & Bem, 1988). Clearly, however, additional studies are needed to examine gender differences in the correlates of social withdrawal and shyness.


The overarching goal of the present study was to compare the best friendships of withdrawn children with those of controlchildren in terms of prevalence, best friends’ characteristics, stability, and relationship quality. We examined the best friendships of children during late childhood, a developmental period during which few studies focusing on socially withdrawn children have been conducted, and explored possible gender differences. We hypothesized that the best friendships of withdrawn children would differ quantitatively and qualitatively from the best friendships of control children. Compared with the control children, we expected withdrawn children to (a) be less likely to have a mutual stable best friendship; (b) have greater similarities with their best friends in terms of social withdrawal and victimization; and (c) have friendships that were lower in relationship quality. Given the putative benefits of friendship involvement, we hypothesized that shy/withdrawn children with mutual best friendships would be rated by their peers as less victimized, and more socially accepted and socially skilled than would shy/withdrawn children without mutual best friendships. We also hypothesized that shy/withdrawn boys would demonstrate greater difficulties at the dyadic level than would shy/withdrawn girls.



Participants were drawn from a large normative sample of fifth graders from eight public elementary schools. The sample consisted of 827 children (406 boys and 421 girls). The mean age of the sample was 10.33 years (SD = 0.52), and all participants had parental consent (consent rate = 84%). There were two phases of data collection: (1) assessments in the schools and (2) an assessment in the laboratory. Participants were assessed in the schools on two different occasions: the beginning of the year (October; Time 1) and approximately 7 months later near the end of the school year (May; Time 2). Between the two time points, 24 children (9 boys and 15 girls) moved to different schools that were not participating in the study; therefore, these children were not involved in Time 2 data collection. Attrition analyses revealed nonsignificant differences between these 24 children and the longitudinal participants (N = 804) on peer- and teacher-reports of social and emotional behaviors (see below for description of ECP and TCRS measures).

School data collection had a dual purpose: (a) to obtain friendship nominations and determine mutuality or reciprocities; and (b) to obtain peer nominations of social-behavioral characteristics and then identify withdrawn and control groups (see description of classification criteria below). Additional questionnaire data were collected during the laboratory visits, which typically occurred between the Time 1 and the Time 2 school assessments.


Research assistants administered two questionnaires in group format in classrooms or larger schoolrooms. The children were informed that their answers were confidential and were instructed not to discuss their answers with classmates. Each session lasted approximately 1 hr. The first questionnaire involved friendship nominations and the second questionnaire was an extended version of the Revised Class Play (see below). The teachers of each participating classroom completed the Teacher–Child Rating Scale (Hightower et al., 1986) on all children for whom parental permission was obtained (Time 1 only). After target groups of children were identified or classified as shy/withdrawn, and control (nonaggressive/nonwithdrawn), they came to the university laboratory with their mutual best friend; at the university the target children and their mutual best friends completed a series of questionnaires, including a measure pertaining to friendship quality. Only those targeted children who had a mutual best friendship participated in the laboratory assessment.

School Measures

Friendship Nominations (Bukowski, Hoza, and Boivin, 1994)

Participants were asked to write the names of their “very best friend” and their “second best friend” at their school. Children could only name same-gender friends in their grade, and only mutual (reciprocated) best friendships were subsequently considered. Children were considered “best friends” if they were each other’s very best or second best friend choice. The identification of a best friendship is similar to procedures used in other studies focused on best friendships (e.g., Parker & Asher, 1993). Although children could nominate any same-gender child in their grade as their best friend, only participating children completed the friendship nominations; therefore, it was impossible to determine whether a friendship was reciprocated when a nonparticipating child was identified as a best friend. Thus, 62 children (38 boys and 24 girls) were excluded from friendship prevalence and stability analyses because both of their best friend nominations were of nonparticipating children or their best friends moved. It is important to note that the majority of children nominated two participating children (Time 1: 76% and Time 2: 80%). Friendship prevalence analyses for the total sample revealed that 60% of the children (496 children: 217 boys and 279 girls) at Time 1 and 59% of children (491 children: 220 boys and 271 girls) at Time 2 had a mutual best friendship.

Child Behaviors

Following completion of the friendship nomination questionnaire, participants completed an extended version of the Revised Class Play (RCP; Masten, Morison, & Pellegrini, 1985). The children were instructed to pretend to be the directors of an imaginary class play and to nominate their classmates for various positive and negative roles. The children were provided with a list of their classmates who were participating in the study and were instructed to choose one boy and one girl for each role, but the same person could be selected for more than one role. Only same-gender nominations for participating children were considered to eliminate possible gender-stereotyping. All item scores were standardized within gender and within classroom in order to adjust for the number of nominations received and also the number of nominators. Items were added to the original RCP to more fully capture different types of aggression (e.g., someone who spreads rumors), and to better distinguish between peer victimization and active isolation or social withdrawal (e.g., someone who prefers to be alone). An exploratory principal components factor analysis with varimax rotation yielded five orthogonal factors: aggression, shyness/withdrawal, victimization/exclusion, prosocial behaviors, and popularity/sociability. The standardized item scores were summed to yield five different total factor scores for each participant. It should be noted that this extended measure, the Extended Class Play (ECP), was found both valid and reliable using the present sample of fifth-grade children across two time points (Burgess, Rubin, Wojslawowicz, Rose-Krasnor, & Booth, 2003). The five-factor model was supported by confirmatory factor analyses. The alphas for the factors were: aggression (7-items): .91; shyness/withdrawal (4-items): .82; victimization/exclusion (8-items): .87; popularity/sociability (5-items): .87; and prosocial behaviors (6-items): .82.

Teacher–Child Rating Scale (T-CRS; Hightower et al., 1986)

The T-CRS questionnaire was used as a measure of socioemotional and behavioral functioning in the classroom. Reliability and validity of the scale have been demonstrated previously (Hightower et al., 1986). The T-CRS yields three problem subscales, learning problems, acting-out, and shy-anxious, and three competency sub-scales, task orientation, assertive social skills, and frustration tolerance. For the problem behaviors, teachers rated each child on a 5-point scale ranging from 1 = not a problem to 5 = very serious problem. For the competency behaviors, the teachers were asked to rate how well each item described the child, ranging from 1 = not at all to 5 = very well, and higher scores were indicative of greater competence. Alphas for the subscales in the present study ranged from .87 to .95. The Learning Problems and Task Orientation subscales were not of interest in the present study.

Identification of Risk and Control Groups

For the Shy/Withdrawn group, we identified those children whose ECP Shyness/Withdrawal scores were in the top 33% (or approximately, 1 SD above the mean) and whose Aggression scores were in the bottom 50% for their gender and grade. A Control group was identified from among those children whose Aggression scores and Shyness/Withdrawal scores fell in the bottom 50% for their gender group and grade. These procedures, including cutoffs, were very similar to those used in previous studies of shyness and social withdrawal (e.g., Ladd & Burgess, 1999; Rubin et al., 1993). All classifications were made at Time 1 according to peer nomination data. The only Time 2 data considered in the present study were friendship nominations.

The Shy/Withdrawn group consisted of 169 children (75 boys and 94 girls) whose mean standardized Shyness/Withdrawal score was 1.04 and whose mean standardized Aggression score was − 0.54. The Control group comprised 163 children (93 boys and 70 girls), whose mean standardized Aggression score was − 0.52 and mean standardized Shyness/Withdrawal score was −0.51.

Laboratory Visit

Based on the school measures of friendship, all targeted children who had a mutual friendship were invited to the university laboratory in the spring of their fifth-grade year. If a targeted child had two mutual school-based best friendships, the child was invited to visit with his or her “very best friend” choice. During the visit, the children and their mutual best friendship completed a number of questionnaires, including the Friendship Quality Questionnaire-Revised (FQQ; Parker & Asher, 1993). A research assistant administered instructions and questionnaires individually to each child. For analyses involving friendship quality, when two targeted withdrawn children shared a mutual best friendship and visited the laboratory together, the child with the higher Shyness/Withdrawal score was selected as a “target,” and the other child was placed in the corresponding “best friend” group for the present study. In other words, if two withdrawn children shared a mutual best friendship, the child with the higher withdrawal score was selected to be the target and the other child was placed in the “best friends of withdrawn children” group. If two control children shared a mutual best friendship, the child with the lower withdrawal score was selected to be the target and the other child was placed in the “best friends of control children” group. If a shy/withdrawn and control child shared a friendship (13 instances), the targeted child was randomly chosen, and the other child was placed in the best friend group. Given these criteria, 47 shy/withdrawn (22 boys and 25 girls) and 48 control children (22 boys and 26 girls) visited the laboratory with their best friends. Approximately 60% of these children visited the laboratory with their “very best friend” choice considered in the “Characteristics of the Best Friends of Shy/Withdrawn Children” analyses below.

Demographic data were collected from the laboratory sample only. Approximately 60% of the children who visited the laboratory were European American, 15% African-American, 15% Asian-American, and 10% Latin American. Sixty-eight percent of their mothers (68% of the fathers) had a university degree, 21% some college education (13% of the fathers), and 9% had high school and vocational education (12% of the fathers). With respect to the shy/withdrawn and control groups of children who visited the laboratory, analyses revealed nonsignificant demographic differences between the two groups in terms of race and SES, as measured by mother’s and father’s highest level of education. Within-group analyses revealed nonsignificant differences on the ECP and TCRS variables between those targeted children who visited the laboratory and those who did not.

Laboratory Measures

Friendship Quality Questionnaire-Revised (FQQ; Parker & Asher, 1993)

The FQQ was used to assess the child’s self-perceived quality of friendship with his/her best friend. The 40-item FQQ yields six subscales in the areas of companionship/recreation, validation/caring, help/guidance, intimate disclosure, conflict/betrayal, and conflict resolution (alpha = .73–.90). The conflict/betrayal factor was reverse-scored; higher scores indicated greater perceived friendship quality on all of the subscales. A total friendship quality score was computed for each child by adding the mean scores of the subscales (with the exception of the conflict and betrayal scale).


Preliminary Analyses

Prior to examining the best friends of the children identified as shy/withdrawn and control, a series of 2 (Group: Shy/Withdrawn, Control) × 2 (Gender) ANOVAs were computed to examine the characteristics of the target groups themselves. These analyses were conducted to ascertain that the target groups indeed differed statistically from each other on the Extended Class Play (ECP) factors of shyness/withdrawal and aggression. Moreover, between-group differences on the victimization/exclusion variable were examined (see Table I for means and standard deviations).

Table I
Mean ECP and TCRS Scores as a Function of Group Status (Shy/Withdrawn and Control)

As expected, shy/withdrawn children were rated by their peers as significantly more withdrawn than were control children, F(1, 332) = 444.23, p < .001. The groups did not differ significantly in their levels of aggression, F(1, 332) = 2.22, ns. Shy/withdrawn children were reported by their peers to be significantly more victimized and excluded than were the control children, F(1, 332) = 88.07, p < .001. There were nonsignificant Group × Gender interaction effects.5

The two target groups also were compared on teacher-reports of shy/anxious and acting-out behaviors (TCRS; Hightower et al., 1986). A 2 (Group: Shy/Withdrawn, Control) × 2 (Gender) ANOVAs were computed (see Table I for means and standard deviations).

Analyses revealed a significant Group × Gender interaction effect for the acting-out variable, F(1, 332) = 4.6, p < .03. Follow-up t tests, conducted separately by gender, revealed significant group differences for boys only. Teachers rated Control boys (M = 8.37, SD = 3.27) as demonstrating significantly more acting-out behaviors than Shy/Withdrawn boys (M = 6.96, SD = 1.97).

Analyses also revealed a significant Group main effect for shy/anxious behaviors, F(1, 332) = 22.69, p < .001. As expected, teachers reported shy/withdrawn children as demonstrating significantly more shy and anxious behaviors than control children. A significant gender effect was also revealed, F(1, 328) = 5.22, p < .02; girls were rated by their teachers as demonstrating greater shy and anxious behaviors (Girls: M = 10.82, SD = 5.19; Boys: M = 9.41, SD = 3.73) than were the boys.

School Assessment: Prevalence of Friendship

The first question concerned whether shy/withdrawn children were as likely as control children to have best friendships. Children were considered “best friends” if they were each other’s very best or second best friend choice. At Time 1, 65% of shy/withdrawn children (37 boys and 69 girls) and 70% of control children (57 boys and 55 girls) had a mutual very best friend. Chi-square analysis revealed no significant differences in the likelihood that shy/withdrawn and control children would have a mutual best friendship, (2, N = 324) = 1.06, ns. Also, there were no significant differences when data were examined separately by gender. (Eight children were excluded from analyses because the two friends they nominated were not participating in the project).

At Time 2, 63% of the shy/withdrawn children (41 boys and 65 girls) and 72% of the control children (63 boys and 55 girls) had a mutual best friend. Chi-square analysis revealed that the groups did not differ, (2, N = 318) = 2.78, ns; no significant differences were revealed when analyses were conducted separately by gender (At Time 2, 14 children were excluded from analyses because the two friends they nominated were not participating in the project).

School Assessment: Comparison of Shy/Withdrawn Children with and Without Best Friendships

To compare the social acceptance and social competence of shy/withdrawn children with mutual best friendships with those without such relationships, we conducted a series of t tests, using the popularity/sociability, prosocial behaviors, and victimization/exclusion ECP factors and the social skills and frustration tolerance TCRS subscales as dependent variables. The ability to regulate frustrating and negative emotions is often considered an index of social competence (Rubin et al., 1998). (It should be noted that a series of preliminary 2 (Group: Shy/Withdrawn with best friend, Shy/Withdrawn without best friend) × 2 (Gender) ANOVAs were conducted; results revealed no significant Gender and Group × Gender interaction effects; and t tests were then performed to maximize power.) This set of analyses involved 106 shy/withdrawn with mutual best friends and 58 shy/withdrawn without mutual best friends.

Results showed that as expected, shy/withdrawn children with mutual best friends were more popular and sociable, t(162) = 2.32, p < .02, than were the shy/withdrawn children without mutual best friends. There were nonsignificant group differences on the prosocial behaviors and victimization/exclusion variables as well as on the TCRS variables.

School Assessment: Characteristics of the Best Friends of Shy/Withdrawn Children

To consider the characteristics of the best friends of shy/withdrawn and control children as a group, a series of 2 (Group: Best Friend of Shy/Withdrawn (N = 97, 33 boys), Best Friend of Control (N = 98, 48 boys)) × 2 (Gender) ANOVAs were performed on the ECP peer nominations of aggression, shyness/withdrawal, victimization/exclusion, prosocial behaviors, and popularity/sociability (see Table II for means and standard deviations). We chose these variables based on the above mentioned shyness and social withdrawal literatures (Rubin et al., 2003); while similarities between shy/withdrawn children and their best friends in terms of shyness and victimization were expected, we also explored possible similarities in sociable and socially competent behaviors and aggression. For these analyses, the “very best friend” data were used. However, because the questionnaire allowed the children to nominate a “very best” and “second best” friend, there were some cases in which the child had two mutual best friends and we were forced to choose only one for analyses (71 children had two mutual best friends). In these instances, the friend nominated by the target child as the “very best” friend (top choice) was used for the analyses. Nine shy/withdrawn and 14 control children were excluded from these analyses because their mutual best friend shared a mutual “very best” friendship with another child (e.g., both of these children nominated each other as “very best friend”). This ensured that no child was involved in more than one mutual best friendship for analyses.

Table II
Mean ECP and TCRS Scores as a Function of Best Friend Group Status (Best Friend of Shy/Withdrawn and Control)

Results of these “best friend” analyses indicated a significant main effect for the group variable (Best friends of Shy/Withdrawn, Best friends of Control) for shyness/withdrawal, F(1, 195) = 4.55, p < .03 and victimization/exclusion, F(1, 195) = 5.89, p < .02. When compared with the best friends of control children, the best friends of shy/withdrawn children were significantly more shy and withdrawn as well as more victimized and excluded by their peers. The best friend groups did not differ in terms of prosocial behaviors, popularity/sociability and aggression scores. There were no significant Group × Gender interactions.

Additional analyses examined whether teachers rated the best friends of shy/withdrawn and control children differently in terms of shy/anxious and socially competent behaviors (social skills, frustration tolerance). Scores were analyzed using a series of 2 (Group: Best friend of Withdrawn, Best friend of Control) × 2 (Gender) ANOVAs. There were no significant best friend group main effects on all teacher-report variables (see Table II for means and standard deviations).

Similarities Between Targeted Children and Their Best Friends

To test the homophily hypothesis within dyads involving shy/withdrawn and control children, behavioral similarity was assessed by Pearson product-moment correlation coefficients computed for each ECP peer nomination variable.6 Similar analyses have been conducted in the past to examine similarities between friends (e.g., Haselager et al., 1998). The correlations were computed separately within group, and also separately by gender, and are shown in Tables III and andIVIV.

Table III
Correlations Between ECP Scores of Target Children and Scores of Their Best Friends, Separately by Gender
Table IV
Correlations Between TCRS Scores of Target Children and Scores of Their Best Friends, Separately by Gender

For the shy/withdrawn children and their best friends, significant dyadic correlations were found for victimization/exclusion, prosocial behaviors, and popularity/sociability. The control children were most similar to their best friends in terms of victimization/exclusion. All other correlations for each group were nonsignificant. It is important to note that these correlations are descriptive of variables on which each of the friendship groups were similar; significant correlations do not suggest, however, that the groups had higher scores on these variables.

Similarities also varied between targeted children and their best friends when correlations were conducted separately by gender. For the shy/withdrawn group, both withdrawn boys’ and girls’ prosocial behaviors scores were significantly associated with those of their best friends, but only the withdrawn boys had significant associations with their best friend’s scores on popularity/sociability and victimization/exclusion. When data were examined separately by gender, there were no significant dyadic correlations for the control boys or girls.

A series of analyses involving Fisher’s r to z transformations revealed that the differences between these correlations (similarities between shy/withdrawn children and their best friends and control children and their best friends) were significant for the prosocial behaviors variable only (z = 2.36, p < .02). When these correlations on the prosocial behaviors variable for shy/withdrawn boys and their best friends were compared to those between control boys and their best friends, the differences approached significance, z = 1.87, p < .06. There were no other significant group or group by gender differences in the magnitude of the behavioral similarities.

We also examined behavioral similarities between targeted children and their best friends in terms of teacher-reports of social behaviors (acting-out, shy/anxious behaviors) and social competence (social skills, frustration tolerance; see Table IV). Similarities between targeted children and their best friends varied by group. For the withdrawn children and their best friends, significant dyadic correlations were found for all TCRS variables. The control children were significantly similar to their best friends in terms of frustration tolerance and social assertive skills. Fisher’s r to z transformations revealed that the differences between these correlations neared significance for the acting-out variable only (z = 1.81, p < .07); analyses revealed no other significant differences.

Shy/withdrawn boys had significant associations with their best friends on shyness/anxiety and social assertive skills. Shy/withdrawn girls were similar to their best friends in terms of acting-out behaviors and frustration tolerance. Significant dyadic correlations were found only for control boys in terms of social assertiveness. Significant dyadic correlations were revealed for control girls for frustration tolerance. Fisher’s r to z transformations failed to reveal any significant group by gender differences.

School Assessment: Stability of Friendship

One-hundred and twelve control children had a mutual best friendship at Time 1—Fall (five of these children were excluded from analyses because their best friend moved or they themselves had moved). Seventy percent (35 boys and 43 girls) of the control children had a stable best friendship (Fall-to-Spring). One-hundred and six shy/withdrawn children had a mutual best friend at Time 1—Fall; however, five of these children moved or their best friend had moved. Sixty-nine percent (24 boys and 46 girls) of the shy/withdrawn children shared the same friendship across the school year. Comparisons of the extent to which shy/withdrawn and control children’s friendships were stable did not reach significance; within-gender chi-square comparisons were likewise nonsignificant. It is important to note that if a targeted child possessed two mutual best friendships at Time 1, he or she was considered to have a stable best friendship if one of these best friendships was maintained (Wojslawowicz, Rubin, Burgess, Booth-LaForce, & Rose-Krasnor, in press).

Laboratory Assessment: Friendship Quality

We examined whether children’s perceptions of friendship quality were related to group status. A series of 2 (Group: Shy/Withdrawn, Control) × 2 (Gender) ANOVAs were performed to consider differences between the target groups and also between the groups of best friends (best friend of withdrawn, best friend of control). The children completed the FQQ with specific reference to the quality of the relationship they had with the friend who visited the laboratory with them (see Tables V and andVIVI for means and standard deviations).

Table V
Mean FQQ Scores for Target Children
Table VI
Mean FQQ Scores for Best Friends of Target Children

A significant group main effect was obtained for the FQQ constructs of help and guidance, F(1, 95) = 7.15, p < .01; intimate exchange, F(1, 95) = 5.19, p < .03; conflict resolution, F(1, 95) = 8.22, p < .001; and the total FQQ summary score, F(1, 95) = 8.03, p < .01. Withdrawn children rated their best friendships as significantly lower than control children on each of these dimensions of friendship quality.

There were gender main effects for validation and caring, F(1, 95) = 5.98, p < .02; help and guidance, F(1, 95) = 4.25, p < .04; intimate disclosure, F(1, 95) = 25.96, p < .001; conflict and betrayal, F(1, 95) = 5.12, p < .03; and total FQQ, F(1, 95) = 10.83, p < .001. Girls reported higher levels of friendship quality than did boys on the following variables: validation/caring: M = 3.98, SD = 0.77 and M = 4.31, SD = 0.55; help/guidance: M = 3.57, SD = 0.83 and M = 3.90, SD = 0.78; intimate disclosure: M = 3.17, SD = 1.02 and M = 4.05, SD = 0.70; conflict and betrayal: M = 4.15, SD = 0.53 and M = 4.39, SD = 0.50; and total FQQ, M = 3.68, SD = 0.72 and M = 4.09, SD = 0.54 for boys and girls, respectively. There were no significant Group × Gender interactions.

Analyses focusing on the best friends’ reports of friendship quality revealed significant group main effects for FQQ constructs of companionship and recreation, F(1, 95) = 3.95, p < .05; help and guidance, F(1, 95) = 7.72, p < .01; and the total FQQ summary score, F(1, 95) = 6.28, p < .01. The best friends of shy/withdrawn children reported their best friendships to be less fun and help/guidance and to be lower in overall friendship quality than did the best friends of control children.


Previous research has demonstrated that shy and socially withdrawn children are often on the “periphery” of the social scene and less engaged with peers than their nonwithdrawn classmates (Rubin et al., 2003). In late childhood, this withdrawal is viewed negatively by peers and is a strong predictor of peer rejection (Newcomb et al., 1993). The present study extended prior research by its investigation of shy and socially withdrawn children’s involvement in close dyadic relationships, specifically best friendships. The present research substantially furthers our knowledge of shy/withdrawn children’s experiences in dyadic relationships by revealing qualitative (behavioral similarities, friendship quality) but not quantitative (prevalence, stability) differences between the best friendships of shy/withdrawn and nonwithdrawn children. Although shy/withdrawn children seemed to be as likely as nonwithdrawn children to have mutual, stable best friendships, their best friendship experiences were less positive.

To begin with our first research question, contrary to our expectations, shy and socially withdrawn children were as likely as control children to have mutual best friendships. These children’s social withdrawal from the peer group at-large did not appear to influence their involvement in close dyadic friendships in the age group under study, suggesting that the company of just one peer may be less anxiety-provoking for these children than the company of many peers. These findings are consistent with results reported earlier by Ladd and Burgess (1999) and Schneider (1999). In contrast to the Schneider’s study, however, all the friendships in the present study were “best” friendships, demonstrating that shy and socially withdrawn children during late childhood are able not only to establish good friendships but also are able to form the closest, and most influential type of friendships, best friendships (Berndt, 1999). Although these results are consistent with prior research focused on the good friendships of withdrawn children, an important direction for future research would be to examine the prevalence of best friendships in shy and socially withdrawn adolescents. As shyness and social withdrawal become even more salient and negative to peers, shy and socially withdrawn adolescents may experience greater difficulty forming best friendships. Such studies focused on shyness during the adolescent developmental period may also reveal significant gender differences, which were not found in the present study. It seems likely that decreased acceptance of shyness, particularly for boys as they move into adolescence, could negatively influence the friendship formation process (Hinde et al., 1985).

The data revealed that shy/withdrawn children with best friends were viewed as more sociable than those without best friends. Specifically, shy/withdrawn children with a mutual best friendship were rated by their peers as more popular/sociable than were shy/withdrawn children without best friendships. These findings strongly suggest that as is the case with nonwithdrawn children, social competence plays an important role in the best friendship formation process for shy/withdrawn children (Rubin et al., 1998). Prior studies have demonstrated that socially competent behaviors, like the sociable behaviors measured in the present study, promote the formation of friendships, whereas socially incompetent behaviors, such as aggression, function in a disruptive manner (Rubin et al., 1998). Importantly, although these behaviors may not improve shy/withdrawn children’s standing with the larger peer group, frequent displays of sociable behaviors may have caused peers to view these better adjusted shy/withdrawn children as more attractive potential friends.

Although friendship involvement does appear to help shy/withdrawn children to be viewed as more involved in the peer group, evidence was revealed suggesting that the benefit of best friendship involvement may be limited for shy/withdrawn children. To begin with, results from behavioral similarity analyses demonstrated significant behavioral similarities between withdrawn children and their best friends. Specifically, significant similarities were revealed for peer-reports of victimization and exclusion, prosocial and popular/sociable behaviors and teacher-reports of acting-out and shy/anxious behaviors, frustration tolerance, and assertive social skills. Furthermore, findings indicated that both withdrawn children and their best friends were more victimized and excluded than were targeted control children and their respective best friends. Results also indicated no significant differences between shy/withdrawn children who had best friends and those who did not in their levels of peer victimization. In addition, the best friends of shy/withdrawn children were more shy and withdrawn than were the best friends of control children, results that provide support for the homophily (“birds of a feather”) hypothesis for withdrawn children. Taken together, these results strongly suggest that the behavioral similarities between shy/withdrawn children and their mutual best friends may decrease the benefits of friendship involvement and also the protective power of friendship, particularly in regard to peer victimization (Hodges et al., 1999). Haselager et al. (1998) suggested that friendships may not be protective for children when the similarities between best friends indicate shared maladaptation. Accordingly, it seems likely that a “misery loves company” scenario exists for many shy, withdrawn children. The coping of two shy children who both feel poorly about themselves and their social worlds may lead to increases in internalizing problems, much the same way that co-rumination within the friendships of adolescent girls predicts later internalizing difficulties (Rose, 2002). Furthermore, it may be that children who are shy and withdrawn are less able to defend themselves and also their friends from peer victimization. The social timidity and wariness of these children may make them even “easier” targets; two “easy” targets that spend time together may be more inviting to bullies and their henchmen than only one. For example, it may be the case that successful victimization of two withdrawn children garners more peer attention than the victimization of a single child. Future research might focus longitudinally on the friendships of shy and socially withdrawn children because a long-lasting friendship with another shy/withdrawn child may exacerbate existing difficulties. On the other hand, a stable, high-quality friendship may prove helpful and comforting for shy/withdrawn children.

It is important to note that whereas peers reported behavioral differences between the best friends of shy/withdrawn and the control children, teachers did not. The teachers, however, judged the targeted shy/withdrawn children to be shyer and anxious and less socially assertive than were the targeted control children. In addition to being more socially withdrawn, an inspection of the means and standard deviations on all adjustment variables for the targeted shy/withdrawn children and their mutual best friends shows the group of targeted withdrawn children as experiencing even greater adjustment difficulties than their respective best friends. Therefore, the difficulties of the best friends may be less obvious to teachers than those of the shy and socially withdrawn children. These findings are consistent with the argument that many teachers are simply unaware of the more subtle difficulties that many children experience. In fact, until recently researchers overlooked shy and socially withdrawn children in part because relative to their aggressive counterparts, parents and teachers alike viewed shy and socially withdrawn children as quiet, well-behaved, and studious (Rubin et al., 2003). However, as noted previously, it is now well-known to researchers that shy and socially withdrawn children experience many difficulties, particularly those of an internalizing nature (e.g., Gazelle & Rudolph, 2004). The findings in the present study suggest that the best friends of withdrawn children may also be at risk for the internalizing problems associated with shyness and social withdrawal during late childhood, and yet, their internalized difficulties may go unnoticed to many adults. Indeed, in the present study, targeted children were selected to be more socially withdrawn than their best friends. An important direction for future studies may be to examine the behavioral similarity between shy/withdrawn children and their best friends longitudinally. If the best friends of socially withdrawn children become more withdrawn over time, it may be important to include these best friends in intervention efforts directed at withdrawn children.

Like prior research onrejected children (Parker & Asher, 1993), findings in the present study demonstrated that the best friendships of shy and socially withdrawn children were qualitatively poorer than the best friendships of children who were not removed from the peer group at-large. As expected, shy/withdrawn children rated their best friendships as significantly lower in help and guidance, intimate disclosure, conflict resolution, and overall friendship quality than did control children. The present study extended prior research (Fordham & Stevenson-Hinde, 1999; Schneider, 1999) by examining the perspectives of both the targeted child and their mutual best friend on the quality of the friendship. Findings indicated that the best friends of withdrawn children viewed their friendships as relatively low in fun and helpfulness, as well as in overall friendship quality. Perhaps the lower friendship quality reported by withdrawn children was, in large part, a function of their shy and timid behavioral style. In the present study, shyness and social withdrawal was operationalized by four items—someone who is very shy; someone who doesn’t talk much or talk quietly; someone who hardly starts conversations; and someone who gets nervous about participating in class discussions. If these children’s shyness and verbal reticence persist when in the context of a close friendship, then it may explain the lower relationship quality. It may be very difficult to have fun with or feel close to a friend who offers few ideas and suggestions, little help and guidance, and only a small amount of overall emotional and social support. These findings contrast with those of prior studies, which suggest that the friendships of shy/withdrawn children are relatively high in relationship quality (Fordham & Stevenson-Hinde, 1999; Schneider, 1999). This inconsistency between our reported findings and those of Schneider, for example, may be accounted for by the direct impact of communicative reticence and social anxiety on different types of friendships. Further, unlike the other two published studies, the children in the present study reported on the quality of their mutual best friendships. These findings suggest that shy/withdrawn children and their best friends may benefit from interventions focused on the dyad, such as those involving “pair therapy” (Selman, 1997; Selman & Schultz, 1990).

Intimacy is especially important for children’s best friendships (Berndt, 1999). It is possible, therefore, that verbal reticence and quiescence is more detrimental to the quality of best friendships than other friendships. Despite the lesser quality of the friendships of socially withdrawn children, nearly 70% of their best friendships were maintained across the school year. Perhaps, then, by middle and late childhood, socially withdrawn children have learned that relationships with age-mates are precious and that just having a friend is of significance. Perhaps, having a friend, regardless of its quality, may buffer socially withdrawn children from feelings of loneliness. Thus, once a best friendship is established, shy/withdrawn children may do their very best to maintain it, regardless of its “costs.”

A number of limitations should be noted. First, friendship identifications were limited to same-sex, same-age, same-school participating children. As such, many important friendships may have been missed, including “neighborhood-” and “leisure-activity-”-based friendships and friendships that are with opposite-sex children but platonic in nature. These types of friendships may prove important sources of social and emotional support for children who experience difficulties with the larger peer group. Moreover, due to the large number of analyses in the present study (without any correction for chance), caution should be used when interpreting findings. Future research is needed to determine whether these findings emerge with larger samples of socially withdrawn and nonwithdrawn children.

Lastly, in future studies, researchers would do well to consider factors within the family when examining the relation between withdrawal and friendship. For example, the parents of withdrawn children often have been described as overprotective or oversolicitous (Rubin & Burgess, 2002). As such, some withdrawn children may seek friends who are similar to their parents in behavioral style (e.g., intrusive and controlling), whereas others may choose friends who are more similar to them (e.g., withdrawn and submissive). A more complete understanding of children’s peer relationships may require consideration of the family environment and the interplay between the two.


The authors would like to thank the children, parents, and teachers who participated in the study as well as Alli Buskirk, Charissa Cheah, Stacey Chuffo, Kathleen Dwyer, Erin Galloway, Jon Goldner, Sue Hartman, Amy Kennedy, Angel Kim, Sarrit Kovacs, Alison Levitch, Abby Moorman, Andre Peri, Margro Purple, Joshua Rubin, Erin Shockey, and Bridget Trame who assisted in data collection and input. The research reported in this manuscript was supported by National Institute of Mental Health grant # MH58116 to Kenneth H. Rubin.


5Additional analyses with Time 2 ECP data were performed to determine whether the groups were based on fairly stable behavioral characteristics. Evidence for behavioral stability was revealed; results indicated that withdrawn children were significantly more withdrawn (p < .001) and victimized (p < .001) and significantly less popular and sociable (p < .001) than control children at Time 2.

6Behavioral similarity analyses were also conducted for each peer nomination variable for the entire sample (Time 1 = 495 children and their respective best friends; Time 2 = 491 children and their respective best friends). At Time 1, the similarity was greatest between children and their best friends on the prosocial behaviors variable (r = .29, p < .001), followed by a correlation of .23 (p < .001) for the popular/sociable variable, and a correlation of .21 between children and their best friends’ victimization scores (p < .001). Significant correlations also emerged for the aggression (r = .17, p < .001) and shyness/withdrawal (r = .17, p<.001) variables. At Time 2, the strongest correlation was for the victimization variable (r = .24, p < .001) and also for the popular/sociable (r = .22, p < .001) and prosocial behaviors variable (r = .18, p < .001). Also, analyses yielded significant correlations for the aggression (r = .11, p < .02) and shyness/withdrawal (r = .10, p < .03) variables.


  • Bagwell CL, Newcomb AF, Bukowski WM. Preadolescent friendship and peer rejection as predictors of adult adjustment. Child Development. 1998;69(1):140–153. [PubMed]
  • Berndt T. Friendship quality and social development. Current Directions in Psychological Science. 1999;11(1):7–10.
  • Berndt TJ, Hawkins JA, Jiao Z. Influences of friends and friendships on adjustment to junior high school. Merrill-Palmer Quarterly. 1999;45:13–41.
  • Berndt TJ, Keefe K. Friends’ influence on adolescents’ adjustment to school. Child Development. 1995;66:1313–1329. [PubMed]
  • Boivin M, Hymel S. Peer experiences and social self-perceptions: A sequential model. Developmental Psychology. 1997;33(1):135–145. [PubMed]
  • Boivin M, Hymel S, Bukowski W. The roles of social withdrawal, peer rejection, and victimization by peers in predicting loneliness and depressed mood in childhood. Development and Psychopathology. 1995;7:765–785.
  • Brendgen M, Little TD, Krappmann L. Rejected children and their friends: A shared evaluation of friendship quality? Merrill-Palmer Quarterly. 2000;46(1):45–70.
  • Brendgen M, Vitaro F, Bukowski WM. Deviant friends and early adolescents’ emotional and behavioral adjustment. Journal of Research on Adolescence. 2000;10(2):173–189.
  • Bukowski WM, Hoza B, Boivin M. Measuring friendship quality during pre- and early adolescence: The development and psychometric properties of the Friendship Qualities Scale. Journal of Social and Personal Relationships. 1994;11(3):472–484.
  • Burgess KB, Rubin KH, Wojslawowicz J, Rose-Krasnor L, Booth C. The “Extended Class Play”: A Longitudinal Study of its factor structure, reliability, and validity; Tampa, FL. Paper presented at the Biennial Meeting of the Society for Research in Child Development.2003.
  • Caspi A, Elder GH, Bem DJ. Moving away from the world: Life-course patterns of shy children. Developmental Psychology. 1988;24(6):824–831.
  • Clark ML, Drewry DL. Similarity and reciprocity in the friendships of elementary school children. Child Study Journal. 1985;15(4):251–264.
  • Dishion TJ, Andrews DW, Crosby L. Antisocial boys and their friends in early adolescence: Relationship characteristics, quality, and interactional process. Child Development. 1995;66:139–151. [PubMed]
  • Fordham K, Stevenson-Hinde J. Shyness, friendship quality, and adjustment during middle childhood. Journal of Child Psychology and Psychiatry and Allied Disciplines. 1999;40(5):757–768. [PubMed]
  • Gazelle H, Ladd GW. Anxious solitude and peer exclusion: A diathesis–stress model of internalizing trajectories in childhood. Child Development. 2003;74(1):257–278. [PubMed]
  • Gazelle H, Rudolph KD. Moving toward and moving away from the world: Social approach and avoidance trajectories in anxious youth. Child Development. 2004;75(3):829–849. [PubMed]
  • Hanish LD, Guerra NG. Aggressive victims, passive victims, and bullies: Developmental continuity or developmental change. Merrill-Palmer Quarterly. 2004;50:17–38.
  • Hart CH, Yang C, Nelson LJ, Robinson C, Olsen J, Porter C, Jin S, Olsen S, Wu P. Peer acceptance in early childhood and subtypes of socially withdrawn behaviour in China, Russia, and the United States. International Journal of Behavioral Development. 2000;24(1):73–81.
  • Haselager GJT, Hartup WH, van Lieshout CFM, Riksen-Walraven JMA. Similarities between friends and nonfriends in middle childhood. Child Development. 1998;69(4):1198–1208.
  • Hightower AD, Work WC, Cowen E, Lotyczewski BS, Spinell AP, Guare JC, Rohrbeck CA. The teacher–child rating scale: A brief objective measure of elementary children’s school problem behaviors and competencies. School Psychology Review. 1986;15(3):393–409.
  • Hinde RA. Individuals, relationships, and culture: Links between ethology and the social sciences. Cambridge: Cambridge University Press; 1987.
  • Hinde RA, Stevenson-Hinde J, Tamplin A. Characteristics of 3- to 4-year-olds assessed at home and their interactions in preschool. Developmental Psychology. 1985;21(1):130–140.
  • Hodges EVE, Boivin M, Vitaro F, Bukowski WM. The power of friendship protection against an escalating cycle of peer victimization. Developmental Psychology. 1999;35(1):94–101. [PubMed]
  • Hodges EVE, Malone MJ, Perry DG. Individual risk and social risk as interacting determinants of victimization in the peer group. Developmental Psychology. 1997;33:1032–1039. [PubMed]
  • Hymel S, Rubin K, Rowden L, LeMare L. Children’s peer relationships: Longitudinal prediction of internalizing and externalizing problems from middle to late childhood. Child Development. 1990;61(6):2004–2021.
  • Ladd GW. Having friends, keeping friends, making friends, and being liked by peers in the classroom: Predictors of children’s early school adjustment? Child Development. 1990;61:1081–1100. [PubMed]
  • Ladd GW, Burgess KB. Charting the relationship trajectories of aggressive, withdrawn, and aggressive/withdrawn children during early grade school. Child Development. 1999;70:910–929. [PubMed]
  • Ladd GW, Troop-Gordon W. The role of chronic peer difficulties in the development of children’s psychological adjustment problems. Child Development. 2003;74(5):1344–1367. [PubMed]
  • Masten AS, Morison P, Pellegrini DS. A revised class play method of peer assessment. Developmental Psychology. 1985;21(3):523–533.
  • McGuire KD, Weisz JR. Social cognition and behavioral correlates of preadolescent chumship. Child Development. 1982;53:1478–1484.
  • Newcomb AF, Bukowski WM, Pattee L. Children’s peer relations: A meta-analytic review of popular, rejected, neglected, controversial, and average sociometric status. Psychological Bulletin. 1993;113:99–128. [PubMed]
  • Nilzon K, Palmerus K. Anxiety and withdrawal of depressed 9–11 year olds three years later: A longitudinal study. School Psychology International. 1998;19(4):341–349.
  • Olweus D. Victimization by peers: Antecedents and long-term outcomes. In: Rubin KH, Asendorpf JB, editors. Social withdrawal, inhibition and shyness in childhood. New Jersey: Lawrence Erlbaum Associates; 1993. pp. 315–341.
  • Parker JG, Asher SR. Friendship and friendship quality in middle childhood: Links with peer group acceptance and feelings of loneliness and social dissatisfaction. Developmental Psychology. 1993;29:611–621.
  • Parker JG, Seal J. Forming, losing, renewing, and replacing friendships: Applying temporal parameters to the assessment of children’s friendship experiences. Child Development. 1996;67:2248–2268.
  • Piaget J. The moral judgment of the child. Oxford: Harcourt; 1932.
  • Poulin F, Boivin M. The role of proactive and reactive aggression in the formation and development of boys’ friendships. Developmental Psychology. 2000;36:233–240. [PubMed]
  • Rose AJ. Co-rumination in the friendships of girls and boys. Child Development. 2002;73(6):1830–1843. [PubMed]
  • Rubin KH, Bukowski W, Parker JG. Peer interactions, relationships, and groups. In: Eisenberg N, Damon W, editors. Handbook of child psychology: Social, emotional, and personality development. 5. Vol. 3. New York: Wiley; 1998. pp. 619–700.
  • Rubin KH, Burgess K. Parents of aggressive and withdrawn children. In: Bornstein M, editor. Handbook of parenting. 2. Vol. 1. Hillsdale, NJ: Lawrence Erlbaum Associates; 2002. pp. 383–418.
  • Rubin K, Burgess K, Kennedy A, Stewart S. Social withdrawal in childhood. In: Mash E, Barkley R, editors. Child psychopathology. 2. New York: Guilford Press; 2003. pp. 372–406.
  • Rubin KH, Chen X, Hymel S. The socio-emotional characteristics of extremely aggressive and extremely withdrawn children. Merrill-Palmer Quarterly. 1993;39:518–534.
  • Rubin KH, Chen X, McDougall P, Bowker A, McKinnon J. The Waterloo longitudinal project: Predicting internalizing and externalizing problems in adolescence. Development and Psychopathology. 1995;7:751–764.
  • Ruble DN, Martin CL. Gender development. In: Eisenberg N, editor. Handbook of child psychology: Social, emotional, and personality development. Vol. 3. New York: Wiley; 1998. pp. 933–1016.
  • Schneider B. A multimethod exploration of the friendships of children considered socially withdrawn by their school peers. Journal of Abnormal Child Psychology. 1999;27(2):115–123. [PubMed]
  • Selman R. Toward a practical theory. In: Selman R, Watts, Schultz L, editors. Fostering friendship: Pair therapy for treatment and prevention. Hawthorne, NY: Gruyter; 1997. pp. 53–74.
  • Selman R, Schultz LH. Making a friend in youth: Developmental theory and pair therapy. Chicago: University of Chicago Press; 1990.
  • Stewart SL, Rubin KH. The social problem-solving skills of anxious-withdrawn children. Development and Psychopathology. 1995;7(2):323–336.
  • Sullivan HS. The interpersonal theory of psychiatry. New York: Norton; 1953.
  • Wojslawowicz JC, Rubin KH, Burgess KB, Booth-LaForce C, Rose-Krasnor L. Behavioral characteristics associated with stable and fluid best friendship patterns during middle childhood. Merrill-Palmer Quarterly. in press.