Aggression and Prosocial Behavior Predict Changes in Perceptions of Friendship Quality in Primary and Middle School Students

Home , Prosocial behavior

AGGRESSION AND PROSOCIAL BEHAVIOR PREDICT CHANGES IN

PERCEPTIONS OF FRIENDSHIP QUALITY IN PRIMARY AND MIDDLE SCHOOL

STUDENTS

Lauren Shawcross

Thesis Submitted to the Faculty of

The Charles E. Schmidt College of Science

In Partial Fulfillment of the Requirements for the Degree of

Master of Arts

Florida Atlantic University

Boca Raton, Florida

ACKNOWLEDGEMENTS

I would like express my gratitude for having Brett Laursen as my advisor. His guidance and support have been indispensable throughout the course of my graduate career in Psychology. I would also like to thank my committee members David Bjorklund and Dave Perry for their insight and thoughtful guidance during this project.

This data collection would not have been possible without help from my colleagues Gilly Bortman, Shirja Dirghangi, Ashley Richmond, Amy Hartl, Dan

Dickson, and Cody Hiatt. I am very thankful to have had the opportunity to work with such a dedicated team.

iv ABSTRACT

Author: Lauren Shawcross

Title: Aggression and Prosocial Behavior Predict Changes in Perceptions of Friendship Quality in Primary and Middle School Students

Institution: Florida Atlantic University

Thesis Advisor: Dr. Brett Laursen

Degree: Master of Arts

Year: 2015

This study examines whether aggression and prosocial behavior shape changes in perceptions of friendship quality within stable reciprocal best friend dyads. A longitudinal Actor-Partner Interdependence Model was used to investigate whether individual characteristics predict changes 6 to 12 weeks later in perceptions of relationship support and negativity. The sample included 76 same-sex dyads drawn from classrooms in grades 4 (M = 9.48 years) through 6 (M= 11.43 years) in two public schools in the United States.

The findings indicate that one friend’s initial relational aggression predicted increases in the other friend’s perceptions of relationship negativity, and one friend’s initial prosocial behavior predicted increases in the other friend’s perceptions of relationship social support. Thus, children’s prosocial and aggressive behaviors shaped their friends’ perceptions of the relationship but not one’s own perceptions.

v AGGRESSION AND PROSOCIAL BEHAVIOR PREDICT CHANGES IN

PERCEPTIONS OF FRIENDSHIP QUALITY IN PRIMARY AND MIDDLE SCHOOL

STUDENTS

List of Tables……………………………………………………………………………..ix

List of Figures…………………………………………………………………...………...x

Introduction…………………………………………………………………….……..…..1

Friendship Quality………………………………………….………………...……….1

Age and Gender Differences in Characteristics of Friendships………...………...3

Friend Influence………………………..………………………………………………3

Links Between Individual Behaviors and Perceptions of Friendship Quality……..….4

Measuring Friend Influence Using an Actor-Partner Interdependence Model………..6

Hypotheses………..……………………………………………………………..…….7

Method……………………………………………………………………………….…..10

Participants.……………………………………………………………………….…..10

Instruments……………………………………………………..…….…..……….…..10

Peer Nominations for Friendships…………………..……………………….…..10

Friendship Quality…………………………..…………...………………………11

Peer Reputation………………..………..……………………………………….11

Procedure………………………..……………………………………………………12

Plan of Analysis………………………..………………………………………...…...13

Does One Friend’s Initial Behavior Predict Changes in the Other Friend’s

vi Perception of the Quality of the Relationship?...... 14

Does One Friend’s Initial Behavior Predict Changes in His or Her Own

Perception of the Quality of the Relationship?...... 15

Does One Friend’s Initial Level of Prosocial Behavior or Aggressive Behavior

Predict Changes in the Other Friend’s Behavior?...... 15

Does One Friend’s Initial Perception of the Quality of the Relationship Predict

Changes in the Other Friend’s Perceptions of the Quality of the

Relationship?...... 16

Does Gender or School Moderate Patterns of Influence?...... 16

Results…………………………..………………………………………………………..17

Preliminary Analyses……………………..………………………………...…...……17

Longitudinal Indistinguishable APIMs for Social Support…………………………..18

Prosocial Behavior……………………..………………………………...…...….18

Relational Aggression………………..………………………………...…...... ….18

Physical Aggression………………..………………………………...…….....….19

Summary of Findings for Social Support…………………………....…….....….19

Longitudinal Indistinguishable APIMs for Negativity………………….……………19

Prosocial Behavior……………………..………………………………...…...….19

Relational Aggression………………..………………………………...…...... ….20

Physical Aggression………………..………………………………...…….....….20

Summary of Findings for Negativity…...…………………………....…….....….20

Discussion………………………..………………………………………………………22

Predicting Changes in Perceptions of Friendship Negativity……..………………….22

vii

Predicting Changes in Perceptions of Social Support……..………………………….24

Predicting Changes in Prosocial and Aggressive Behavior……..………...………….26

Implications.……………..……………………………………………………………27

Limitations……………..………………………………………………………..……28

Conclusion………………………………………………………...………………….30

Appendices…………………..…………………………………………..………………31

References…………………..…………………………….………..……………….……44

viii LIST OF TABLES

Table 1. Means, Standard Deviations, and Correlations Between Study

Variables……………………………………………………………..………32

Table 2. Within-Dyad Similarity: Intraclass Correlations on Study Variables….……33

LIST OF FIGURES

Figure 1. A longitudinal Actor-Partner Interdependence Model (APIM) for

indistinguishable dyads predicting changes in friendship quality

(i.e. social support or negativity) ……………….……..………...…...... 34

Figure 2. A longitudinal indistinguishable Actor-Partner Interdependence Model

(APIM) for prosocial behavior and social support………………..…...... 35

Figure 3. A longitudinal indistinguishable Actor-Partner Interdependence Model

(APIM) for relational aggression and social support………………..…...36

Figure 4. A longitudinal indistinguishable Actor-Partner Interdependence Model

(APIM) for physical aggression and social support……………………...37

Figure 5. A longitudinal indistinguishable Actor-Partner Interdependence Model

(APIM) for prosocial behavior and negativity…………………………...38

Figure 6. A longitudinal indistinguishable Actor-Partner Interdependence Model

(APIM) for relational aggression and negativity…………………...……39

Figure 7. A longitudinal indistinguishable Actor-Partner Interdependence Model

(APIM) for physical aggression and negativity………………………….40

x INTRODUCTION

Friendship quality has important implications for individual adjustment. It is linked to a number of adaptive outcomes such as sociability and leadership skills (Hartup

& Stevens, 1999). We know that friendship quality affects individual adjustment, but we know little about the factors that predict the quality of a friendship. Previous studies have indicated that best friends influence each other’s behavior (Wentzel, Barry, & Caldwell,

2004) and over-time associations between individual behaviors and friendship quality during adolescence (Cillssen, Jiang, West, & Laszkowski, 2005). The purpose of the present study is to determine whether aggressive behavior and prosocial behavior predict changes in friendship quality over the course of a semester.

Friendship Quality

Kerns defines friendship quality as “the essential character or nature of friendship” (1996). Over the years scholars have developed a variety of friendship quality scales to describe platonic peer relationships. In response to factor analyses indicating that friendship features are highly correlated within positive and negative categories

(Berndt, 1996), most recent friendship quality measures eschew specific categories or features of friendships in favor of broad scales that assess positive and negative dimensions (Berndt & Perry, 1986; Berndt & Keefe, 1992). Positive features of friendship tend to include closeness, companionship, help, and security (Bukowski,

1993). Negative features of friendship quality encompass conflict, criticism, and

1 competition (Berndt & McCanless, 2009). Positive and negative dimensions are only weakly correlated (Berndt, 2002).

Qualitative features of friendship play an important role in determining developmental outcomes (Hartup & Stevens, 1997). For example, social support is negatively correlated with depression and delinquency (Windle,1994) and is positively correlated with academic achievement (Cauce, 1986) and self-esteem (McGuire & Weisz,

1982). Friendship quality also influences social-emotional development and is associated with the observed behaviors of children with their friends (Brendgen, Markievicz, Doyle,

& Bukowski, 2001). For example, adolescent perception of social support from best friends is associated with emotional adjustment (Gauze, Bukowski, Aquan-Assee, &

Sippola, 1996), and with positive affect and relationship responsiveness (Brendgen et al.,

2001).

The Network of Relationships Inventory (Furman & Buhrmester, 1985) is one of the most common measures of relationship quality administered to children. The inventory addresses perceptions of psychological provisions (i.e. benefits or costs) of the relationship as opposed to social processes (i.e. interactions) that occur between friends.

The Social Provisions Version of the Network of Relationships Inventory assesses friendship perceptions divided into supportive features (e.g. enhancement of worth, reliable alliance, intimacy, instrumental help, companionship, nurturance and affection items) and negativity (e.g. antagonism and conflict items) that load onto two second- order factors (Adams & Laursen, 2007). From these two second-order factors, one can derive the perceived benefits (social support) and costs (negativity) of a relationship.

Many of positive features captured in the social support dimension are attributed to high-

quality friendships; low-quality friendships are high on negative features (Bagwell &

Schmidt, 2011).

Age and gender differences in characteristics of friendships. Some scholars posit that personal characteristics and behaviors become increasingly important to friendship quality with age because older children and adolescents typically have more options for friends than younger children and therefore can be more selective (Aboud &

Mendelson, 1996). Consistent with this assertion, perceptions of friendship social support increase from childhood to adolescence (Furman & Buhrmester, 1992). A four-year growth curve analysis of friendship perceptions indicated that perceptions of best friend positive qualities improve from middle to late adolescence (Way & Greene, 2006). In general, girls report higher intimacy in their same-sex best friendships before boys; however, the same story provided evidence that by late adolescence boys may “catch up” to girls in terms of social support.

Friend Influence

There are two reasons why children may be similar to their friends: they befriend children who are already similar to themselves (selection), and over the course of their friendship they influence each other’s behaviors through their interactions (socialization).

It is unclear to what degree best friends actively or passively reinforce one another’s aggression and prosocial behaviors. There is some evidence that selection may play a more important role in similarity between friends than socialization (Bot et al., 2005).

Scholars agree that peers influence each other on a variety of behaviors, and a number of studies show that the characteristics and behaviors of one’s friends predict important outcomes such as academic adjustment and health-risk behaviors (Berndt,

1999; Prinstein, Boergers, & Spirito, 2001). There is evidence that for certain behaviors, such as transition to alcohol use, one’s closest friend is more influential than the friendship group (Urberg, Değirmencioğlu, & Pilgrim, 1997). Less clear is the degree to which best friends influence one another’s prosocial and aggressive behaviors.

Links Between Individual Behaviors and Perceptions of Friendship Quality

Perceived friendship quality has been linked to the behaviors of each member of the dyad. Friends are expected to help and support one another. It follows that positive behaviors should predict to perceptions of social support. Friends are also expected to avoid inflicting harm on each other. It follows that aggressive behaviors should predict negative perceptions of the relationship.

Several studies reveal a concurrent positive correlation between adolescent prosocial behavior and best friendship quality (Markiewicz, Doyle, & Brendgen, 2001).

Two studies report that aggression is inversely related to concurrent social support; although, there is disagreement as to whether the association is stronger for boys or for girls (Fanti, Brookmeyer, Henrich, & Kuperminc, 2009; Cillessen et al., 2005).

Although there is strong evidence that prosocial and aggressive behaviors are concurrently correlated with friendship quality, over-time associations are not well- established. One study with same-sex middle school best friend dyads found that one child’s relational aggression was associated with increases in his or her own perceptions of positive friendship quality 11 months later but not with increases in the partner’s perception of positive friendship quality (Banny, et al., 2011). Thus, there were actor effects (an adolescent’s perceptions were influenced by his or her own behaviors), but

there were no partner effects (an adolescent’s perceptions were not influenced by his or her friend’s behaviors).

In another study, Grotpeter and Crick (1996) found that during childhood and early adolescence, relationally aggressive children report high levels of relational aggression within their best friendship. Paradoxically, relationally aggressive girls also report high levels of intimacy and exclusivity within their best friendships compared to non-relationally aggressive girls. Thus, relationally aggressive children may have friendships that are high on both social support and negativity. High levels of social support may be what allow relationally aggressive girls to maintain their best friendships in spite of the negativity resulting from their aggressive behavior.

With age, children increasingly comment on the prosocial behaviors of their peers

(Bernt, Hawkins, & Hoyle, 1996). Having a friend who is high on prosocial behavior predicts increases in one’s own prosocial behavior overtime (Wentzel et al., 2004).

Prosocial behaviors and intentions between friends increase between Grades 1 and 4

(Berndt, 1981). There is also evidence that adolescents in Grade 8 exhibit more prosocial behaviors with their friends compared to other classmates than students in Grades 4 and 6

(Berndt, 1985). These findings support the claim that across childhood and adolescence, equality in friendships is increasingly preferred over competition.

Perceptions of the positive and negative dimensions of children’s three closest friendships are quite stable across the school year, even when the students select different friends at different points during the year (Berndt & Keefe, 1992). This suggests that there is stability in how individuals view relationships, how they construct them, or both.

In his five-stage model of friendship development, Selman theorized that social support

from friends allows for increased intimacy and closeness during adolescence (1980). The positive social interactions and processes that contribute to the perceived benefit of social support should provide a positive feedback loop for increasing intimacy and support over time. Likewise, it is reasonable that negative behaviors, such as aggression, would create a negative feedback loop of conflict and disagreeable behaviors that spill over into intrapersonal and interpersonal skill development. Put simply, considerable variability between individuals but considerable stability within individuals is expected in perceptions of friendship quality.

Measuring Friend Influence Using an Actor-Partner Interdependence Model

This study incorporates the Actor-Partner Interdependence Model (APIM: Kenny,

Kashy, & Cook, 2006) for indistinguishable dyads to predict changes in friendship quality and individual behaviors. An indistinguishable APIM is used if dyad members are considered interchangeable, as is often the case in same sex friendships. The APIM used in this study has been modified for longitudinal data (Popp, Laursen, Burk, Kerr, &

Stattin, 2008). The longitudinal APIM can be used to measure changes in one friend’s behavior over time, predicted by initial levels of the other friend’s behavior. In the present study, one friend’s aggressive and prosocial behaviors are used to predict changes in the other friend’s perceptions of friendship quality.

APIM analyses are useful when examining friend dyads because they account for the non-independent nature of friend relationships. Friend reports of relationship quality and friend reports of individual behaviors are typically correlated, which violates assumptions of statistical independence. Longitudinal APIMs are able to partition the variance shared by friends from the variance unique to each friend. This allows the

investigator to examine the degree to which initial behaviors shape changes in one’s own perceptions of relationship quality and in friend’s perceptions of relationship quality.

Hypotheses

The present study addresses the following research questions: (1) Does one friend’s initial prosocial or aggressive behavior predict changes in the other friend’s perception of the quality of the relationship? (2) Does one friend’s initial prosocial or aggressive behavior predict changes in his or her own perception of the quality of the relationship? (3) Does one’s friend’s initial level of prosocial behavior or aggressive behavior predict changes in the other friend’s behavior? (4) Does one friend’s initial perception of the quality of the relationship predict changes in the other friend’s perceptions of the quality of the relationship? Each will be discussed in turn.

Does one friend’s initial prosocial or aggressive behavior predict changes in the other friend’s perception of the quality of the relationship? It is hypothesized that one friend’s prosocial behavior will predict increases in the other friend’s perception of social support and decreases in other friend’s perception negativity. It is also hypothesized that one friend’s aggressive behavior will predict increases in the other friend’s perceptions of negativity. Previous studies indicate that relational aggression is concurrently linked to high levels of intimacy and other positive features (Grotpeter & Crick, 1996); therefore, it is hypothesized that one friend’s relational aggression will also predict increases in the other friend’s perception of social support. Similar associations are not hypothesized for physical aggression because unlike relational aggression, physical aggression is not linked to high levels of intimate disclosure and closeness.

Does one friend’s initial prosocial or aggressive behavior predict changes in his or her own perception of the quality of the relationship? It is hypothesized that as a child’s prosocial behavior increases there should be commensurate increases in his or her own perceptions of social support and decreases in his or her own perceptions of negativity within a friendship. One study found that adolescents’ with higher prosocial behavior had higher overall perceptions of the relationship quality with their best friend (Markiewicz,

Doyle, & Brendgen, 2001). It is also hypothesized that one friend’s relational aggression would predict increases in his or her own perceptions of social support. This would replicate Banny and colleagues (2011) findings that adolescents with higher levels of relational aggression reported increases in their own perceptions of positive friendship quality. In contrast, physical aggression is not hypothesized to have the same compensatory benefits and should not predict increases in perceptions of social support because physical aggression is not associated with the same type of subtle, manipulative behavior as relational aggression. Finally, it is hypothesized that one friend’s aggression

(relational or physical) should predict increases in his or her own perceptions of negativity. Aggression is typically associated with disagreements and negative relationship features (Cillessen et al., 2005; Dishion, Andrews, & Crosby, 1995), which suggests that aggressive individuals should have more negative conflict with their friends.

Does one’s friend’s initial level of prosocial behavior or aggressive behavior predict changes in the other friend’s behavior? It is hypothesized that one friend’s prosocial behavior will predict increases in his or her best friend’s prosocial behavior.

Stable best friends’ prosocial behavior should increase in similarity over time, consistant with results from a study following of high school students who reported that friend

prosocial behavior at time 1 was related to the subject’s prosocial goal pursuit and prosocial behavior at time 2 (Barry & Wentzel, 2006). Finally, aggressive behaviors at time 1 were hypothesized to lead to increases in friend aggressive behaviors.

Does one friend’s initial perception of the quality of the relationship predict changes in the other friend’s perceptions of the quality of the relationship? It is hypothesized that initial high levels of social support on the part of one friend will predict increases in the other friend’s perception of social support, and that one friend’s initial perceptions of negativity will predict increases the other friend’s perception of negativity.

Berndt has argued that children with high levels of negativity in their friendships tend to have an interactional style that contributes to negativity and conflict (Berndt, 2000).

Perceptions of quality should be contagious in close relationships because there is some evidence that best friendship quality predicts happiness (Demir, Ozdemir & Weitekamp,

2007), and that happy, satisfied people influence others’ satisfaction (Murray, Holmes, &

Griffin, 1996).

9 METHOD

Participants

This study was part of a larger project on friendship that included 450 students from two public schools representative of the state of Florida in terms of ethnicity and family income. This study was limited to participants who were involved in reciprocally nominated best friendships that were stable across two time points. The final sample of participants included 152 students in Grades 4 through 6. Of this total, there were 34 fourth grade boys (22%), 30 fourth grade girls (20%), 18 fifth grade boys (12%), 26 fifth grade girls (17%), 12 sixth grade boys (8%), and 32 sixth grade girls (21%). At the initial assessment, fourth graders ranged in age from 9 to 11 years old (M= 9.48, SD = 0.54), fifth graders ranged in age from 10 to 11 years old (M=10.55, SD=0.50), and sixth graders ranged in age from 11 to 13 years old (M=11.43, SD=0.58). The sample consisted of 32 male (42%) and 44 female (58%) best friend dyads.

Instruments

Participants completed the “My Classmates” questionnaire at two time points.

Participants answered questions about themselves, their best friend relationships, and their same-sex classmates. The present study focused on characteristics of individuals assessed through peer nominations and perceptions of friendship quality assessed through self-reports.

Peer nominations for friendships. Participants were provided with a roster of same-sex participants in the same class (Grades 4 and 5) or grade (Grade 6). Participants

10 were instructed to circle the names of all of their friends and then rank order them. No ties were permitted (see Appendix A). Unlimited nominations were possible.

Reciprocated best friendships were defined as dyads in which two participants nominated each other as first or second best friend. Stable best friends were defined as dyads in which both friends nominated one another as a first or second best friend at both time points. The friendship rank constraint was necessary because children only completed friendship inventories describing their 1st and 2nd ranked friends.

Friendship Quality. Participants completed a short version of the Network of

Relationships Inventory (Furman & Buhrmester, 1985) describing perceptions of the quality of the relationship with each of the first two nominated best friends (see Appendix

B). Items were rated on a scale ranging from 1 (little or none) to 5 (the most). Social support was measured by 6 items describing positive provisions of the relationship.

Negativity was measured by 5 items describing negative provisions of the relationship.

For each scale, items were summed and averaged. Internal reliability was good for social support (α = 0.82 to α =0.84 ) and adequate for negativity (α = 0.74 to α = 0.90).

Peer Reputation. Participants completed a modified version of the Extended

Class Play (Burgess, Wojslawowicz, Rubin, Rose-Krasnor, & Booth, 2003). Each participant was asked to nominate an unlimited number of same-sex classmates or grade mates who best fit each descriptor (see Appendix C). For each item, students were provided with a roster of same-sex participants by class (Grades 4 and 5) or by grade

(Grade 6) and asked to nominate those who best fit the description.

Two nomination items were used to measure prosocial behavior (Masten,

Morison, & Pellegrini, 1985): “Someone who makes sure everyone is treated equally”

and “Someone who helps others with their problems.” Internal reliability was good (α =

0.89 to α = 0.89). Two items were used to calculate relational aggression (Bukowski, et al., 2009; Velasquz, Santo, Saldariagga, Lopez, & Bukowski, 2010): “Someone who talks bad about others behind their backs to hurt them” and “Someone who tries to keep others out of the group.” Internal reliability was good (α = 0.88 to α = 0.86). Two nomination items assessed physical aggression (Bukowski, Schwartzman, Santo, Bagwell, & Adams,

2009): “Someone who hurts others physically” and “Someone who hits, pushes, or shoves.” Internal reliability was adequate (α = 0.76 to α = 0.85). For each item, the number of nominations a student received was summed and divided by the number of potential nominators to create a proportion score. For each variable, the proportion scores for the two items were averaged.

Procedure

Letters were sent home from school to parents describing the study and requesting consent for participation. Parental consent and student assent were required for participation.

Valid sociometric data requires participation rates of at least 60% within class and sex (Grades 4 and 5) or within grade and sex (Grade 6) (Cillessen, 2009). Participating girls in 9 out of 10 classrooms in Grade 4, 8 out of 10 classrooms in Grade 5, and 11 out of 11 classes in Grade 6 met this criterion. Participating boys in 8 out of 10 classrooms in

Grade 4, 6 out of 10 classrooms in Grade 5, and 4 out of 11 classes in Grade 6 met this criterion.

Students were assessed at two time points, 6 (n = 32) or 12 (n = 120) weeks apart depending on the school. Trained research assistants administered questionnaires to the

participating students during forty-five minute blocks during school hours. Students were assessed in small groups with testing barriers for privacy. There were no missing data.

Plan of Analysis

A longitudinal, indistinguishable Actor-Partner Interdependence Model (APIM) was used to determine whether individual characteristics and behaviors predict changes in friendship quality. Analyses were conducted using AMOS v22 (Arbuckle, 2013). In indistinguishable APIM analyses, the means, variances, covariances, and path weights are fixed to be equal between friends, because partners are considered interchangeable.

Figures 1 depicts fully saturated measurement models for social support and negativity. Paths a and b represent over time, within-individual stability. Path a depicts the stability of behavioral characteristics (i.e. prosocial behavior, relational aggression, or physical aggression) from time 1 to time 2. Path b shows the stability of friendship quality (i.e. negativity or social support) from time 1 to time 2. Actor effects are associations within subject between constructs over time (paths c and d). Path c represents the within-individual influence that one friend’s time 1 behavior (i.e. prosocial behavior, relational aggression, or physical aggression) has on his or her own time 2 perceptions of friendship quality (i.e. social support or negativity). Path d represents the within-individual influence that one friend’s time 1 perception of friendship quality (e.g. social support or negativity) has on his or her own time 2 behavior (i.e. prosocial behavior, relational aggression, or physical aggression).

Partner paths are associations between friends between constructs over time and associations between friends within constructs overtime (paths e, f, g, and h). Path e represents the between-friend influence that one friend’s time 1 behavior (i.e. prosocial

behavior, relational aggression, or physical aggression) has on the other friend’s time 2 perception of friendship quality (i.e. social support or negativity). Path f represents the influence of one friend’s time 1 reports of friendship quality (e.g. social support or negativity) on the other friend’s time 2 behavior (i.e. prosocial behavior, relational aggression, or physical aggression). Path g represents the between-friend influence of one friend’s time 1 perceptions of friendship quality (i.e. social support or negativity) on the other friend’s time 2 perception of friendship quality (i.e. social support or negativity). Path h depicts the between-friend influence of one friend’s time 1 behavior

(i.e. prosocial behavior, relational aggression, or physical aggression) on the other friend’s time 2 behavior (i.e. prosocial behavior, relational aggression, or physical aggression).

Does one friend’s initial behavior predict changes in the other friend’s perception of the quality of the relationship? Support for the hypothesis that one friend’s prosocial behavior predicts increases in the other friend’s perception of social support will be found if path e is positive and statistically significant (Figure 1). Support for the hypothesis that one friend’s prosocial behavior predicts decreases in the other friend’s perception negativity of will be found if path e is negative and statistically significant (Figure 1).

Support for the hypothesis that one friend’s relational aggression predicts increases in the other friend’s perception of social support will be found if path e is positive and statistically significant (Figure 1). Support for the hypothesis that one friend’s relational aggression predicts increases in the other friend’s perception of negativity will be found if path e is positive and statistically significant (Figure 1).

Support for the hypothesis that one friend’s physical aggression predicts decreases in the other friend’s perception of social support will be found if partner path e is negative and statistically significant (Figure 1). Support for the hypothesis that one friend’s physical aggression predicts increases in the other friend’s perception of negativity will be found if partner path e is positive and statistically significant (Figure

1).

Does one friend’s initial behavior predict changes in his or her own perception of the quality of the relationship? Support for the hypothesis that one friend’s physical aggression predicts decreases in his or her own perceptions of social support will be found if actor path c is negative and statistically significant (Figures 1).

Support for the hypothesis that one friend’s relational aggression physical predicts increases in his or her own perceptions of social support will be found if actor path c is positive and statistically significant (Figures 1). Support for the hypothesis that one friend’s aggression (relational or physical) predicts increases in his or her own perceptions of negativity will be found if actor path c is positive and statistically significant (Figures 1).

Support for the hypothesis that one friend’s prosocial behavior predicts increases his or her own perception of social support will be found if actor path c is positive and statistically significant (Figure 1). Support for the hypothesis that one friend’s prosocial behavior predicts decreases his or her own perception of negativity will be found if actor path c is negative and statistically significant (Figure 1).

Does one’s friend’s initial level of prosocial behavior or aggressive behavior predict changes in the other friend’s behavior? Support for the hypothesis that one

friend’s prosocial behavior predicts increases in his or her friend’s prosocial behavior will be found if partner path e is positive and statistically significant (Figure 1).

Support for the hypothesis that one friend’s aggression (e.g. relational or physical) predicts increases in his or her friend’s aggression (e.g. relational or physical) would be found if partner path e is positive and statistically significant (Figure 1).

Does one friend’s initial perception of the quality of the relationship predict changes in the other friend’s perceptions of the quality of the relationship?

Support for the hypothesis that one friend’s initial perception of social support predicts increases in his or her best friend’s perception of social support will be found if partner path g is positive and statistically significant (Figure 1).

Support for the hypothesis that one friend’s initial perceptions of negativity predict increases in his or best friend’s perception of negativity will be found if partner path g is positive and statistically significant (Figure 1).

Does gender or school moderate patterns of influence? Six multiple group analyses will determine whether strength of association differences exist as a function of school or gender (Figures 2, 3, 4, 5, 6, and 7). Influence paths will be constrained to be equal across models, and model fit will be examined to determine if there are statistically significant differences between boys and girls or between school groups. A significant χ2 difference between groups will indicate poor model fit, and that the patterns of association differed. A non-significant χ2 difference between groups will indicate that the influence paths did not significantly differ.

16 RESULTS

Preliminary Analyses

Table 1 presents the bivariate correlations between study variables. Each time 1 variable was positively correlated with itself at time 2. Prosocial behavior at time 1 was negatively associated with both types of aggression, concurrently and over time. Both types of aggression were positively associated with one another, concurrently and over time. Relational aggression was positively correlated with negativity at time 1 but not at time 2 or over time. Social support and negativity were negatively correlated at time 2 but not at time 1 or over time.

Table 2 describes the intraclass correlations (interpreted as r2) between friends within dyads. Statistically significant correlations emerged for social support at both times and for negativity at time 2.

Three separate 3 (grade) X 2 (sex) repeated measures ANOVAS were conducted with prosocial behavior, relational aggression, and physical aggression as dependent variables. Time was factor in the ANOVA, and the effect of time was not statistically significant. For the purpose of these analyses, one member was randomly selected from each dyad to avoid statistical bias due to non-independence.

For prosocial behavior, there was a significant main effect of grade, F(2, 70)=

9.68, p<.001, ηp2=.217. Grades 4 (M=.49, SD=.26) and 5 (M =.53 SD=.27) reported significantly higher prosocial behavior than Grade 6 (M=.24, SD= .09). There were neither main effects nor interactions involving gender on the pattern of results.

17 For relational aggression, there were neither main effects nor interactions involving gender or grade.

For physical aggression, there was a significant main effect of gender, F(1,

70)=5.85, p<.05, ηp2=.08. Males (M=.05, SD=.09) reported significantly higher physical aggression than females (M =.01 SD=.03). There were no main effects involving grade, and there were no statistically significant interactions involving grade or gender.

Table 1 displays the means and standard deviations for the study variables at times 1 and 2.

Longitudinal Indistinguishable APIMs for Social Support

Prosocial Behavior. Figure 2 presents the results for a longitudinal APIM analysis describing associations between prosocial behavior and social support. Two out of four partner paths were statistically significant. One friend’s time 1 prosocial behavior was positively associated with the other friend’s time 2 perceived social support (β= .16, p=.01). Higher initial levels of one friend’s prosocial behavior predicted greater increases in the other friend’s perceived social support. In addition, one friend’s prosocial behavior at time 1 was positively associated with the other friend’s prosocial behavior at time 2

(β= .16, p<.01). Higher initial levels of one friend’s prosocial behavior predicted increases in the other friend’s prosocial behavior. There were statistically significant stability paths for prosocial behavior (β= .7, p<.01) and social support (β= .63, p<.01).

There were no statistically significant actor paths.

Relational Aggression. Figure 3 shows the results for the APIM analysis describing associations between relational aggression and social support. None of the partner paths reached conventional levels of statistical significance. There were

borderline partner effects indicating that one friend’s time 1 relational aggression was negatively associated with the other friend’s time 2 perceived social support (β= -.11, p=.095). Higher initial levels of one friend’s relational aggression predicted greater decreases in the other friend’s perceived social support. There were statistically significant stability paths for relational aggression (β= .67, p<.01 and social support (β=

.64, p<.01). There were no statistically significant actor paths.

Physical Aggression. Figure 4 shows the results for the APIM analysis describing associations between physical aggression and social support. One out of the four partner paths was statistically significant. One friend’s time 1 physical aggression was negatively associated with the other friend’s time 2 physical aggression (β= -.12, p<.05). Higher initial levels of one friend’s physical aggression predicted decreases in the other friend’s physical aggression. There were statistically significant stability paths for physical aggression (β= .77, p<.01) and social support (β= .64, p<.01). There were no statistically significant actor paths.

Summary of findings for Social Support. There were partner paths from one friend’s prosocial behavior to the other friend’s prosocial behavior, and from one friend’s physical aggression to the other friend’s physical aggression. One additional partner path emerged: one friend’s initial prosocial behavior predicted changes in the other friend’s perceptions of social support. There were no actor paths in any model.

Longitudinal Indistinguishable APIMs for Negativity

Prosocial Behavior. Figure 5 shows the results for the APIM analysis describing associations between prosocial behavior and negativity. One out of the four partner paths was statistically significant. One friend’s prosocial behavior at time 1 was positively

associated with the other friend’s prosocial behavior at time 2 (β= .16, p<.01). Higher initial levels of one friend’s prosocial behavior predicted increases in the other friend’s prosocial behavior. There were statistically significant stability paths for prosocial behavior (β= .72, p<.01 and negativity (β= .53, p<.01). There were no statistically significant actor paths.

Relational Aggression. Figure 6 shows the results for the APIM analysis describing associations between relational aggression and negativity. One out of the four partner paths was statistically significant. One friend’s time 1 relational aggression was positively associated with the other friend’s time 2 perceived negativity (β= .14, p=.03).

Higher initial levels of one friend’s relationally aggressive behavior predicted greater increases in the other friend’s time 2 perceived negativity. There were statistically significant stability paths for relational aggression (β= .7, p<.01) and negativity (β= .54, p<.01). There were no statistically significant actor paths.

Physical Aggression. Figure 7 shows the results for the APIM analysis describing associations between physical aggression and negativity. One out of the four partner paths was statistically significant. One friend’s time 1 physical aggression was negatively associated with the other friend’s time 2 physical aggression (β= -.11, p<.05). Higher initial levels of one friend’s physically aggressive behavior predicted decreases in the other friend’s time 2 physically aggressive behavior. There were statistically significant stability paths for physical aggression (β= .78, p<.01 and negativity (β= .54, p<.01).

There were no statistically significant actor paths.

Summary of Findings for Negativity. As was found for social support, there were partner paths from one friend’s prosocial behavior to the other friend’s prosoical

behavior, and from one friend’s physical aggression to the other friend’s physical aggression. One additional partner path emerged: one friend’s initial relationally aggressive behavior predicted changes in the other friend’s perceptions of negativity.

There were no actor paths in any model.

21 DISCUSSION

This study was designed to examine the interplay between friend behavior and perceptions of friendship quality. Four goals were addressed. The first three concerned changes in perceptions of the quality of friendships. For example, does one friend’s behavior alter the other friend’s view of the friendship? I found that the relationally aggressive and prosocial behavior of one friend predicted changes in the other friend’s perceptions of the relationship, but one friend’s behavior did not predict changes in his or her own perceptions of the relationship. Prosocial behavior led to an increase in perceptions of social support; whereas, relational aggression led to an increase in perceptions of negativity. The fourth goal concerned how friends influence each other’s prosocial and aggressive behavior. I found that one friend’s prosocial behavior predicted increases in the other friend’s prosocial behavior. However unexpectedly, one friend’s physical aggression predicted decreases in the other friend’s physical aggression.

Predicting Changes in Perceptions of Friendship Negativity

Results from the present study indicated that individual levels of relational aggression predict increases in the friend’s perceptions of negativity but not one’s own perceptions of negativity. This is in line with previous studies that indicate that relational aggression adversely impacts friendship quality (Cillissen et al, 2005). Children with friends who are high on relational aggression report increasing negativity and heightened friendship difficulties.

There are several ways in which relational aggression can damage friendship quality. Relationally aggressive children may use positive aspects of friendship, such as intimate disclosure, to control their friends (Crick & Grotpeter, 1996). Relational aggression is defined as subtle, manipulative behavior that aims to harm others. It is logical that the person in the relationship who feels like he or she is being manipulated would experience increases in negativity. Consistent with this view are studies indicating, that dissimilarity on relational aggression predicts friendship dissolution (Ellis &

Zarbaany, 2007). For these middle school children, the negative effects of conflict may be exacerbated due to aggressive strategies of their friends such as revenge or exclusion.

Although some conflict may be a constructive means of communication and offer opportunities to develop social skills (Laursen & Pursell, 2009), not all children perceive and handle conflicts in ways that improve relationship outcomes. One study showed that physically and relationally aggressive boys have more intense negative affect during conflict with their friends (Bagwell & Coie, 2004). Another study found that verbally aggressive strategies in response to conflict are linked to low friendship quality (Rose &

Asher, 1999). Thus, relationally aggressive children may respond to conflict in ways that increase their best friends’ perceptions of negativity but not their own.

In the present study, physical aggression did not predict changes in negativity.

This was not what I had expected. It was assumed that children whose best friends were high on physical aggression would experience increases in negativity consistent with results from a previous study (Brendgen et al., 2002). There is some evidence, however, that physically aggressive children do not perceive increases in negativity, even as observers’ ratings of friendship quality were significantly worse for aggressive boys

compared to non-aggressive boys (Bagwell & Coie, 2004). A number of other studies have found that children high on aggression have more frequent and acute conflicts with their friends (Cillessen et al., 2005; Coie et al., 1999), but perhaps children do not see this as detrimental to their relationships. Or perhaps the relationships with the worst outcomes dissolved.

Initial friendship negativity did not predicted changes in negativity 6 to 12 weeks later. This was unexpected because in a study examining relationship quality in marital relationships, wives of men who reported high levels negativity early in their marriage reported decreases in satisfaction and increases in their own negativity (Huston &

Vangelisti, 1991). This suggests that negativity is particularly contagious within relationships. Although with adults, there is evidence that initial relationship quality may predict later changes in relationship quality, the present study did not find that children’s initial negative perceptions predicted changes in quality 6 to 12 weeks later. This may be because negativity is not as contagious within children’s friendships or because the timeframe of the study was too short to capture significant changes. Null findings also may have emerged because relationships that significantly changed in terms of negativity dissolved, leaving only stable ones for analyses.

Predicting Changes in Perceptions of Friendship Social Support

I expected positive social behaviors would be linked to increased perceptions of positive friendship features. In the present study, one friend’s prosocial behavior predicted increases in the other friend’s perceived social support. Children who are high on prosocial behavior are willing to help others, and this adaptive behavior is linked with positive social adjustment (Crick, 1996). Prosocial children often are well-liked by peers

(Ladd & Profilet, 1996). Friends of prosocial children report increases in social support because prosocial behaviors such as helping, sharing, cooperating, and comforting others are types of positive interactions that make an individual feel cared for and connected

(Rose-Krasnor & Denham, 2009). Children high on prosocial behaviors interact with their friends in socially supportive ways that increase the perceptions of positive features of friendship. Prosocial behaviors are rewarding and apparently make friends feel good about their relationship.

One friend’s aggressive behavior did not predict changes in the other friend’s perceptions of social support. A previous study with older adolescents (Grades 6 and 8), found that individuals high on relational aggression reported increases in social support one year later, but their friends did not (Banny et al., 2011). It is possible that as perceived negativity increases the relationally aggressive friend engages in more socially supportive behavior in order to maintain the friendship. Relationally aggressive children may use intimate disclosure or other types of socially supportive interactions to increase closeness after a conflict with their best friend. It is possible that children high on relational aggression report increases in social support over the course of many months as

Banny and colleagues found (2011), but over shorter time periods, aggressive behavior does not significantly predict changes in quality perceptions.

Initial perceptions of social support did not predict changes in perceptions of social support 6 to 12 weeks later. This was unexpected because previous studies concerning adult romantic relationships and friendships have indicated that positivity begets positivity. A study of college students found that having a high-quality relationship with one’s best friend was the best relationship feature that predicted

happiness (Demir, Ozdemir & Weitekamp, 2006). In romantic relationships too, positive perceptions of one’s partner predict greater relationship satisfaction (Murray, Holmes, &

Griffin, 1996). Although previous research with adults suggests that positive views of a relationship may lead to higher relationship quality, the present study did not find that initial social support predicted changes in friendship quality. Null findings may have emerged because associations between social support and friendship quality operate differently for children and adults. It is also possible that the timeframe for the present study was too short to identify significant changes.

Predicting Changes in Prosocial and Aggressive Behavior

This is not the first study to find that one child’s prosocial behavior influences his or her friend’s prosocial behavior. Higher levels of prosocial behavior have been linked to increases in a friend’s prosocial behavior (Barry & Wentzel, 2006). Earlier studies, however, have focused on adolescents. The present study is the first to demonstrate this link among primary school children. Extending the findings downward is important because the results imply that the identity of friends during primary school can have an impact on developing positive behaviors related to social competence across middle school and high school for longer than previously suspected. Future studies should determine whether behavior changes in primary school children are moderated by social goals as has been demonstrated with high school students.

In the present study, children whose best friends were high on prosocial behavior experienced increases in their own prosocial behavior. There is some evidence that children in high quality friendships have more influence over each other than children in low quality friendships (Berndt, 2002; Barry & Wentzel, 2006). Given that children with

prosocial friends reported increases in social support, it is perhaps not surprising that their own prosocial behavior was also influenced by their friend’s behavior.

The present study did not find that the friends of children high on aggressive behavior became more aggressive. This may be due to the fact that children in low quality friendships are not as easily influenced to become like their friends (Berndt,

2002). One study that followed students from primary school to middle school found that children’s unilateral friends have a greater influence on aggressive behavior than their reciprocated friends do (Adams, Bukowski, & Bagwell, 2005).

Implications

The present study indicates that one child’s relational aggression and prosocial behavior impact his or her best friends’ perceptions of the relationship during a time period as brief as 6 weeks. This is important because it shows that individual behaviors can significantly improve or worsen a relationship over a short timeframe. This study provides a first step in helping teachers, clinicians, and parents to understand the important role that a friend’s behavior can have over perceptions of relationship quality, which, in turn, can impact individual adjustment.

This information could be transferred to potential interventions that help children understand how their behaviors are perceived by peers. It may help clinicians select behavior-targeted interventions to help children who have low-quality friendships.

Teachers could also use this information to help children who are experiencing peer difficulties interact with peers in ways that will help them develop social competences and minimize contact with those who will exacerbate problem behaviors. Findings from

the present study may also remind parents to encourage prosocial behaviors in their children and to discourage children from affiliating with relationally aggressive peers.

Future studies should track friendships over longer intervals to investigate the longer term outlook of children who have worsening or improving friendships. Children with improving friendships are likely to cultivate positive behaviors and social competences that benefit their socioemotional and academic adjustment. There is evidence that improvements in friendship quality predict decreases in problem behaviors such as aggression, depression, and anxiety (Pittman & Richmond, 2010). Children with worsening friendships may exhibit negative behaviors that strain their relationships and interfere with the development of social competence. Increases in negativity have been linked to decreases in psychosocial adjustment (Bagwell et al., 2005) and self-esteem

(Keefe & Berndt, 1996).

Future research is needed to show if the effects of individual behaviors are exacerbated or attenuated over extended time frames, to determine their impact on friendship stability. The present study did not consider the possibility that some friendships deteriorated to the point where the dissolved completely because of increasing negativity. Although some studies have found no link between friendship quality and friendship stability during adolescence (Bowker, 2004), others have found that social support predicts friendship stability (Bukowski, Hoza, & Newcomb, 1994), and negativity is linked to instability (Ladd, Kochenderf, & Coleman, 1996). Additional studies are needed to clarify how changes in friendship quality perceptions influence relationship outcomes.

Limitations

Several limitations merit mention. First, power was low for testing moderators.

With only 76 dyads, there was limited power to adequately investigate age or gender differences. Second, participation rates varied by age and sex. Participation rates were lowest for boys and Grade 6 subjects. Rates were especially low for Grade 5 and 6 males.

As a consequence, the results may be somewhat more representative of younger subjects and girls. Third, the study was limited to very best friends. Only the first and second best friend nominations could be used because quality measures were only administered to each participant’s two best friends. It may be that aggression is lower and prosocial behavior is higher in these friendships than in lower ranked friendships. It is likely that there are certain traits that influence changes in relationship quality outside the realm of one’s very best friendships. For example, relational aggression may prove more problematic for lower ranked friendships because there may not be the same degree of intimacy and disclosure between members of the dyad.

A challenge for this type of research is that friendship dissolution is a common occurrence. Approximately half of all friendships dissolve over the course of a school year (Bowker, 2004; Degirmencioglu, Urberg, Tolson, & Richard, 1998). As a consequence it is likely that friendships that deteriorated in quality were not included in this study because they were not stable across the two time points. Findings for the consequences of negative behaviors and negativity may have been understated as a result.

Peer-reported measures of behaviors may not be the best measure of aggression or prosocial behavior within a close relationship. Children may be more prosocial and less aggressive with best friends than with other peers. For example, the physical aggression of a child who victimizes others in the class (but not necessarily their friends) may not be

as strong of a predictor for negativity within the friendship as physical aggression directed toward the friend. The degree to which peer reports of behavior are correlated with friend reports of behavior in a relationship are unknown.

Conclusion

Children and adolescents with prosocial best friends enjoy friendships that improve, in quality over time. Those with relationally aggressive friends must endure friendships that decline in quality over time. This is important because it shows that the behaviors of children’s best friends impact the quality of the relationship. We know that friendship quality is linked to psychosocial and behavioral adjustment (Vitaro, Boivin, &

Bukowski, 2009). It follows that the behaviors of a child’s best friend in primary school may play an important role not only in his or her own perceptions of peer relationships, but also in a variety of outcomes related to socioemotional competence.

APPENDICES

Table 1.

Means, Standard Deviations, and Correlations Between Study Variables

Variable 1 2 3 4 5 6 7 8 9 10 M SD Time 1

1. Social Support - 3.97 0.67

2. Negativity -.02-- - 1.31 0.46

3. Prosocial Behavior -.11- -.14- - 0.40 0.22

4. Relational Aggression -.12-- - .22** -.44** - 0.09 0.12

32 5. Physical Aggre ssion -.00-- -.03-- -.34** - .40** - 0.02 0.05

Time 2 6. Social Support -.66** -.02- -.07- -.02-- -.06- - 3.99 0.72

7. Negativity -.09- - .55** -.09- -.13-- -.07- -.23** - 1.33 0.43

8. Prosocial Behavior -.13- -.09- -.77** -.33** -.29** -.12 -.03 - 0.42 0.25

9. Relational Aggression .13- .08- -.39** -.67** - .35** .11- -.04- -.37** - 0.10 0.12

10. Physical Aggression -.05- .09- -.24** -.47** - .76** -.10-- -.11- -.26** -.52** - 0.03 0.08

Note. N= 152 (76 best friend dyads). Social Support and Negativity were rated on a scale from 1 (little or none) to 5 (the most). *p<.05, **p<.01, two-tailed

Table 2.

Within-Dyad Similarity: Intraclass Correlations on Study Variables

Variable Time 1 Time 2 1. Social Support .18* .27**

2. Negativity .09 .29**

3. Prosocial Behavior .62** .69**

33 4. Relational Aggression .55** .59**

5. Physical Aggression .26* .11

Note. N= 152 (76 best friend dyads). Social Support and Negativity were rated on a scale from 1 (little or none) to 5 (the most).

*p<.05, **p<.01, two-tailed

Figure 1. A longitudinal Actor-Partner Interdependence Model (APIM) for indistinguishable dyads predicting changes in friendship quality (i.e. social support or negativity)

Time 1 Time 2

Friend 1 Friend 1 a e1 Behavior Behavior c d s f w m v e h j x

Friend 1 b Friend 1 y Friendship Friendship e2 z g Quality Quality r 34 n u i

Friend 2 g Friend 2 y Friendship e3 z b Friendship Quality Quality r

w n u x e f h i

c d Friend 2 a Friend 2 e4 Behavior Behavior s m v j Note. The model includes stability paths (a and b), within-friend influence paths (c and d), between-friend influence paths (e, f, g, and

h), within-friend correlations (w and x), and between-friend correlations (y and z).

Figure 2. A longitudinal indistinguishable Actor-Partner Interdependence Model (APIM) for prosocial behavior and social support

Time 1 Time 2 (r² =.61 ) .70** Friend 1 Friend 1 -.07 [.60, .78] Prosocial Prosocial e1 [-.22, .06] -.06 Behavioral Behavioral .16* [-.17, .05] .16** -.11 [.05, .26] -.11 [.05, .26] [-.28, .04] (r² =.46) [-.24, .11]

.63** .04 Friend 1 Friend 1 .00 Social [.51, .73] e2 [-.15, .21] .09 Social [-.18, .20] Support -.01 [-.05, .21] Support 35 ] .42** .18 [-.11, .08] .20 .52** [-.02, .43] [.32, .70] [.20, .62] [-.03, .36] (r² =.46) -.01 Friend 2 [-.11, .08] .09 .04 Friend 2 .00 Social .63** [-.05, .21] e3 [-.15, .21] Social [-.18, .20] Support [.51, .73] Support -.11 -.11 .16** -.06 [-.24, .11] [-.28, .04] [.05, .26] [-.17, .05] (r² = .61) .16* Friend 2 [.05, .26] -.07 Friend 2 Prosocial [-.22, .06] e4 Prosocial Behavior .70** Behavior

[.60, .78] Note. N=152 (76 friend dyads). For each path, 95% confidence intervals are presented in brackets, *p<.05, **p<.01, two-tailed.

Figure 3. A longitudinal indistinguishable Actor-Partner Interdependence Model (APIM) for relational aggression and social support

Time 1 Time 2 (r² = .45) .67** Friend 1 Friend 1 -.01 [.46, .82] Relational Relational e1 [-.20, .22] Aggression .06 Aggression -.03 -.11 [-.11, .17] .12 .12 [-.18, .15] [-.21, -.01] [-.05, .30] (r² = .45) [-.05, .33]

Friend 1 .64 Friend 1 -.06 -.03 Social e2 [.52, .75] .08 Social [-.22, .13] Support 36 [-.23, .18] -.05 [-.05, .20] Support

.37** .18 [-.17, .07] .17 .37 [-.06, .40] [.15, .53] [.12, .57] [-.03, .36] -.05 (r² = .45)

Friend 2 [-.17, .07] .08 -.03 Friend 2 -.06 Social .64 [-.05, .20] e3 [-.23, .18] Social [-.22, .13] Support [.52, .75] Support .12 .12 -.03 .06 [-.05, .33] [-.05, .30] [-.18, .15] -.11 [-.11, .17] (r² = .45)

[-.21, -.01] Friend 2 -.01 Friend 2 Relational [-.20, .22] e4 Relational Aggression .67** Aggression

[.46, .82]

Note. N=152 (76 friend dyads). For each path, 95% confidence intervals are presented in brackets, *p<.05, **p<.01, two-tailed

Figure 4. A longitudinal indistinguishable Actor-Partner Interdependence Model (APIM) for physical aggression and social support

Time 1 Time 2 (r² = .60) .77** Friend 1 Friend 1 -.05 [.55, .88] Physical Physical e1 [-.20, .07] Aggression -.04 Aggression -.12* -.04 [-.14, .06] -.04 .00 [-.30, .00] [-.14, .05] [-.18, .17] (r² = .44) [-.24, .21]

Friend 1 .64** Friend 1 -.14 -.09 Social e2 [.52, .75] .06 Social [-.26, .00] Support [-.24, .08] -.08 [-.05, .20] Support 37 .20 .18 [-.17, .07] .17 .35** [-.02, .39] [.10, .55] [ -.06, .64] [-.02, .39] -.08 (r² = .44) [-.17, .07] Friend 2 .06 -.09 Friend 2 -.14 Social .64** [-.05, .20] e3 [-.24, .08] Social [-.26, .00] Support [.52, .75] Support -.04 .00 -.12* -.04 [-.24, .21] [-.18, .17] [-.30, .00] -.04 [-.14, .06] (r² = .60)

[-.14, .05] Friend 2 -.05 Friend 2 Physical [-.20, .07] e4 Physical Aggression .77** Aggression

[.55, .88]

Note. N=152 (76 friend dyads). For each path, 95% confidence intervals are presented in brackets, *p<.05, **p<.01, two-tailed

Figure 5. A longitudinal indistinguishable Actor-Partner Interdependence Model (APIM) for prosocial behavior and negativity

Time 1 Time 2 (r² = .61) .72** Friend 1 Friend 1 .04 [.63, .80] Prosocial Prosocial e1 [-.10, .21] Behavior .03 Behavior -.08 [-.07, .12] .08 .-14 .16** [-.23, .05] [.06, .26] (r² = .32) [-.05, .22] [-.30, .04]

-.17* .53** Friend 1 Friend 1 -.06 [-.30, -.04] [.36, .66] e2 [-.22, .11] Negativity .12 Negativity .06 [-.03, .24] 38 .42** .09 [-.18, .04] .20 .52** [-.07, .40] [.33, .68] [.18, .64] [-.12, .34] .06 (r² = .32) [-.18, .04] .12 -.17* Friend 2 53** [-.03, .24] Friend 2 e3 -.06 Negativity [-.22, .11] [-.30, -.04] [.36, .66] Negativity .08 -.14 .16** .03 [-.05, .22] [-.30, .04] [.06, .26] -.08 [--.07, .12] (r² = .61)

[-.23, .05] Friend 2 .04 Friend 2 Prosocial [-.10, .21] e4 Prosocial Behavior .72** Behavior

[.63, .80]

Note. N=152 (76 friend dyads). For each path, 95% confidence intervals are presented in brackets, *p<.05, **p<.01, two-tailed.

Figure 6. A longitudinal indistinguishable Actor-Partner Interdependence Model (APIM) for relational aggression and negativity

Time 1 Time 2 (r² = .45) .70** Friend 1 Friend 1 -.06 [.50, .82] Relational Relational e1 [-.21, .08] Aggression -.06 Aggression .14* [-.14, .04] .00 .22** -.03 [-.06, .33] [-.17, .16] (r² = .33) [-.20, .21] [.02, .40]

.54** Friend 1 Friend 1 .13 .12 [.37, .67] e2 [-.12, .33] Negativity .09 Negativity [-.05, .22] -.07 [-.06, .22] 39 .37** .09 [-.18, .04] .21 .35** -.03, .41] [.14, .51] [.12, .57] [-.12, .27] -.07 (r² = .33)

[-.18, .04] .09 .12 Friend 2 54** [-.06, .22] Friend 2 e3 .13 Negativity [-.12, .33] [-.05, .22] [.37, .67] Negativity .00 .22** -.03 -.06 [-.2, .21] [.02, .40] [-.17, .16] .14* [-.14, .04] (r² = .45)

[-.06, .33] Friend 2 -.06 Friend 2 Relational [-.21, .08] e4 Relational Aggression .70** Aggression

[.50, .82]

Note. N=152 (76 friend dyads). For each path, 95% confidence intervals are presented in brackets, *p<.05, **p<.01, two-tailed.

Figure 7. A longitudinal indistinguishable Actor-Partner Interdependence Model (APIM) for physical aggression and negativity

Time 1 Time 2

(r² = .59) .78** Friend 1 [.56, .89] Friend 1 Physical .03 Physical e1 [-.07, .18] .06 Aggression Aggression -.11* .09 [-.06, .18] .06 .04 [-.28, -.02] [-.06, .26] [-.06, .18] (r² = .33) [-.16, .29]

Friend 1 .54** .21 -.02 Friend 1 e2 Negativity [.37, .67] .12 [-.01, .42]

40 Negativity [-.12, .11] -.03 [-.02, .24] .38 .20 .09 [-.12, .04] .18 [-.06, .41] [.16, .56] [ -.06, .64] [-.12, .34] -.03 (r² = .33)

[-.12, .04] .12 -.02 Friend 2 .21 .54** [-.02, .24] Friend 2 e3 [-.12, .11] Negativity [-.01, .42] [.37, .67] Negativity .06 .04 -.11* .06 [-.16, .29] [-.06, .18] [-.28, -.02] .09 [-.06, .18] (r² = .59)

[-.06, .26] Friend 2 .03 Friend 2 Physical [-.07, .18] e4 Physical Aggression .78** Aggression [.56, .89]

Note. N=152 (76 friend dyads). For each path, 95% confidence intervals are presented in brackets, *p<.05, **p<.01, two-tailed

Appendix A

Peer Nominations for Friendships

Each participant will be provided a roster of the same-sex participants within their class. They will then be asked to nominate and rank their friends. For example, boys will be given the following directions:

1. Circle the names of your friends 2. Rank your friends: Put “1” next to the name of your 1st best friend Put “2” next to your 2nd best friend. Keep going for as many friends as you have. Make sure you give numbers for as many friends as you circle. No ties!

Participants will also be provided with a blank space to write the name(s) of any other same-sex students in their grade that they are friends with at their school.

In addition, participants will be provided with two check boxes. The directions for the first box (for boys) are: “Check this box if there is not a boy in your grade at school who is your friend.” The directions for the second box are: “Check this box if you have 1 or more friends outside your grade at school.”

Appendix B

Friendship Quality

Participants will rate friendship quality with each of their top two nominated friends.

1. How much does this person like or approve of the things you do? (positive) 2. How much do you and this person argue with each other? (negative) 3. How sure are you that this relationship will last? (positive) 4. How much do you and this person get on eacther’s nerves? (negative) 5. How often do you play around and have fun with this person? (positive) 6. How often do you and this person get mad at or fight with each other? (negative) 7. How much does this person help you figure things out or fix things? (positive) 8. How much do you and this person get annoyed with each other? (negative) 9. How often do you tell this person things you don’t want others to know? (positive) 10. How much do you and this person hassle or nag one another? (negative) 11. How much does this person really care about you? (positive)

Response format: (1) Little or none (2) Somewhat (3) Very much (4) Extremely much (5) The most

Appendix C

Behavioral Measures (Peer Assessment)

Participants will be provided with a roster of same-sex participants in their class for each item. The directions for this section are: “Each statement describes people. Check the box for all of the students each statement describes. Pick the students that best fit each description. If no one fits the description, check the box that says NONE, and do not check any other box. Do not pick yourself for any of the descriptions. You can check as many names as you think fit.”

1. Someone who talks bad about others behind their backs to hurt them 2. Someone who hits, pushes, or shoves other people 3. Someone who makes sure everyone is treated equally 4. Someone who hurts others physically 5. Someone who tries to keep others out of the group 6. Someone who helps others with their problems

REFERENCES

Adams, R. E., & Laursen, B. (2007). The correlates of conflict: disagreement is not

necessarily detrimental. Journal of Family Psychology, 21(3), 445.

Adams, R., Bukowski, W., & Bagwell, C. (2005). Stability of aggression during early

adolescence as moderated by reciprocated friendship status and friend's

aggression. International Journal of Behavioral Development, 29(2), 139-145.

Bukowski, W. M., Schwartzman, A., Santo, J., Bagwell, C., & Adams, R. (2009).

Reactivity and distortions in the self: Narcissism, types of aggression, and the

functioning of the hypothalamic–pituitary–adrenal axis during early

adolescence. Development and psychopathology, 21(04), 1249-1262.

Arbuckle, J. L. (2013). IBM® SPSS® Amos™ (Version 22) [Computer Program].

Armonk, NY: IBM Corp.

Bagwell, C. L., & Coie, J. D. (2004). The best friendships of aggressive boys:

Relationship quality, conflict management, and rule-breaking behavior. Journal of

Experimental Child Psychology, 88(1), 5-24.

Bagwell, C. L., Bender, S. E., Andreassi, C. L., Kinoshita, T. L., Montarello, S. A., &

Muller, J. G. (2005). Friendship quality and perceived relationship changes

predict psychosocial adjustment in early adulthood. Journal of Social and

Personal Relationships, 22(2), 235-254.

Bagwell, C. & Schmidt, M.E. (2011). Friendships in Childhood and Adolescence. New

York: Guildford Press.

Banny, A. M., Heilbron, N., Ames, A., & Prinstein, M. J. (2011). Relational benefits of

relational aggression: Adaptive and maladaptive associations with adolescent

friendship quality.Developmental Psychology, 47(4), 1153-1166.

doi:10.1037/a0022546

Barry, C. M., & Wentzel, K. R. (2006). Friend influence on prosocial behavior: The role

of motivational factors and friendship characteristics. Developmental Psychology,

42(1), 153.

Berndt, T. J. (1981). Age changes and changes over time in prosocial intentions and

behavior between friends. Developmental Psychology, 17(4), 408.

Berndt, T. J. (1985). Prosocial behavior between friends in middle childhood and early

adolescence. The Journal of Early Adolescence, 5(3), 307-317.

Berndt, T. J. (1999). Friends' influence on students' adjustment to school. Educational

Psychologist, 34(1), 15-28.

Berndt, T. J. (2002). Friendship quality and social development. Current Directions in

Psychological Science, 11, 7-10.

Berndt, T. J., Hawkins, J. A., & Hoyle, S. G. (1986). Changes in friendship during a

school year: Effects on children's and adolescents' impressions of friendship and

sharing with friends. Child Development, 1284-1297.

Berndt, T. J., & Perry, T. B. (1986).Children’s perceptions of friendships as supportive

relationships. Developmental Psychology, 22, 640-648.

Berndt, T. J., & Keefe, K. E. U. N. H. O. (1992). Friends’ influence on adolescents’

perceptions of themselves at school. Student perceptions in the classroom, 51-73.

Bowker, A. (2004). Predicting friendship stability during early adolescence. The Journal

of Early Adolescence, 24(2), 85-112.

Bukowski, W. M., Hoza, B., & Newcomb, A. F. (1994). Using rating scale and

nomination techniques to measure friendship and popularity. Journal of Social

and Personal Relationships, 11(3), 485-488.

Burgess, K. B., Wojslawowicz, J. C., Rubin, K. H., Rose-Krasnor, L., & Booth, C. L.

(2003, April). The ‘‘Extended Class Play’’: A longitudinal study of its factor

structure, reliability, and validity. Presented at the biennial meeting of the Society

for Research in Child Development, Tampa, FL.

Cillessen, A. H. N. (2009). Sociometric Methods. In K. H. Rubin, W. M. Bukowski, & B.

Laursen (Eds.), Handbook of Peer Interactions, Relationships, and Groups (82-

99). New York: Guilford.

Cillessen, A. H. N., Jiang, X. L., West, T. V., & Laszkowski, D. K. (2005). Predictors of

dyadic friendship quality in adolescence. International Journal of Behavioral

Development, 29, 165–172.

Değirmencioğlu, S. M., Urberg, K. A., Tolson, J. M., & Richard, P. (1998). Adolescent

friendship networks: Continuity and change over the school year.Merrill-Palmer

Quarterly (1982-), 313-337.

Demir, M., Özdemir, M., & Weitekamp, L. A. (2007). Looking to happy tomorrows with

friends: Best and close friendships as they predict happiness.Journal of Happiness

Studies, 8(2), 243-271.

Dishion, T. J., Andrews, D. W., & Crosby, L. (1995). Antisocial boys and their friends in

early adolescence: Relationship characteristics, quality, and interactional process.

Child Development, 66(1), 139-151.

Ellis, W. & Zarbatany, L. (2007). Explaining Friendship Formation and Friendship

Stability: The Role of Children’s and Friends’ Aggression and Victimization.

Merrill-Palmer Quarterly, 53(1), 79–104.

Furman, W., & Buhrmester, D. (1985). Children's perceptions of the personal

relationships in their social networks. Developmental Psychology, 21(6), 1016-

1024.

Grotpeter, J. K., & Crick, N. R. (1996). Relational aggression, overt aggression, and

friendship. Child development, 67(5), 2328-2338.

Hartup, W. W., & Stevens, N. (1997). Friendships and adaptation in the life course.

Psychological Bulletin, 121, 355-370.

Keefe, K., & Berndt, T. J. (1996). Relations of friendship quality to self-esteem in early

adolescence. The Journal of Early Adolescence, 16(1), 110-129.

Kenny, D. A., Kashy, D. A., & Cook, W. L. (2006). Dyadic data analysis. New York:

Guilford Press.

Ladd, G. W., Kochenderfer, B. J., & Coleman, C. C. (1996). Friendship quality as a

predictor of young children's early school adjustment. Child development, 67(3),

1103-1118.

Ladd, G. W., & Profilet, S. M. (1996). The Child Behavior Scale: A teacher-report

measure of young children's aggressive, withdrawn, and prosocial

behaviors. Developmental psychology, 32(6), 1008.

Murray, S. L., Holmes, J. G., & Griffin, D. W. (1996). The self-fulfilling nature of

positive illusions in romantic relationships: love is not blind, but

prescient.Journal of personality and social psychology, 71(6), 1155.

Prinstein, M. J., Boergers, J., & Spirito, A. (2001). Adolescents' and their friends' health-

risk behavior: Factors that alter or add to peer influence. Journal of Pediatric

Psychology, 26(5), 287-298.

Salvas, M.-C., Vitaro, F., Brendgen, M., Lacourse, É., Boivin, M. and Tremblay, R. E.

(2011), Interplay between Friends' Aggression and Friendship Quality in the

Development of Child Aggression during the Early School Years. Social

Development, 20: 645–663. doi: 10.1111/j.1467-9507.2010.00592.x

Selman, R. L. (1980). The growth of interpersonal understanding (p. 24). New York:

Academic Press.

Urberg, K. A., Değirmencioğlu, S. M., & Pilgrim, C. (1997). Close friend and group

influence on adolescent cigarette smoking and alcohol use. Developmental

Psychology, 33(5), 834.

Vitaro, F., Boivin, M., & Bukowski, W. M. (2009). The role of friendship in child and

adolescent psychosocial development. Handbook of peer interactions,

relationships, and groups, 568-585.

Wentzel, K. R., Barry, C. M., & Caldwell, K. A. (2004). Friendships in Middle School:

Influences on Motivation and School Adjustment. Journal of educational

psychology, 96(2), 195.