9
Teaching practices and elementary classroom peer ecologies Scott D. Gest a, , Philip C. Rodkin b,1 a Pennsylvania State University, Department of Human Development & Family Studies, 110 Henderson South Building, University Park, PA 16802, USA b University of Illinois at Urbana-Champaign, Departments of Educational Psychology and Psychology, 232A Col. Wolfe School, Mail Code 422, 403 E. Healey St., Champaign, Illinois 61820, IL, USA abstract article info Available online 18 July 2011 Keywords: Peer networks Friendship Teaching practices Classroom Behavior management Teachers and students in 39 1st, 3rd and 5th grade classrooms participated in a study of teaching practices and classroom peer networks. Teachers reported on their attitudes towards aggression and withdrawal, provided rationales for their seating arrangements, and were observed on patterns of emotional and instructional support and classroom organization. Students were surveyed or interviewed for peer nominations of friend- ship, popularity, being cool, and aggressive and prosocial behavior, and from these nominations classroom- level indices were derived concerning the richness of positive and negative social ties among children, the egalitarian or hierarchical nature of social status hierarchies, and classroom norms for aggressive and prosocial behavior. Preliminary results suggest that teachers' attitudes towards social behavior are associated with the degree of liking and disliking their students express; teachers' grouping patterns are associated with numerous features of classroom social networks; and teachers' observed emotional support was associated with higher rates of friendship reciprocation. © 2011 Elsevier Inc. All rights reserved. Assertions that teachers play a crucial role in classroom peer relationships are not new (Cairns & Cairns, 1994; Gronlund, 1959; Lewin, 1943), but conceptual models that specify the processes involved are more recent (Farmer et al., 2006) and empirical evidence of these processes is uncommon. Part of the difculty in building an empirical literature linking teaching practices to classroom peer relations is the sheer diversity of teaching processes that could be studied and the equally numerous aspects of peer relationship patterns that could be impacted. Without a conceptual and measure- ment strategy to focus research efforts, there is a risk that claims of teacher inuence on classroom social processes could become an empty truism that lacks the specicity that could guide teacher professional development efforts. In the present paper, we aim to illustrate one strategy for focusing research efforts in this area by integrating recent advances in the measurement of teacherstudent interactions (Pianta, La Paro & Hamre, 2007), conceptual models of teacher inuence on social network dynamics (Farmer et al., 2006), and classroom-level measures of social networks (Ahn, Garandeau, & Rodkin, 2010; Garandeau, Ahn, & Rodkin, in press; Gest, Davidson, Rulison, Moody & Welsh, 2007). A central premise of our approach is that teachers inuence classroom social dynamics both indirectly, through general teaching practices, and more directly through active attempts to manage the social network; and that these effects can be seen not only in terms of the experiences of individual children, but also in the overall organization of peer network patterns. Lewin (1943) was certain that teachers had the opportunity and responsibility to optimize peer ecologies. During the time leading up to World War II, early studies of sociometry, or the science of the measurement of social relationships, focused creatively on how to quantify properties of peer social ecologies, relate these properties to educational success and failure, and build up a truly social, action- oriented theory of child social development (e.g., Bronfenbrenner, 1944; Jennings, 1943; Moreno, 1934). By mid-century, educational psychology texts emphasized that teachers could understand and manage peer social dynamics with the assistance of research-based technologies (e.g., Cronbach, 1954; Blair, Jones, & Simpson, 1954; see also Parsons, 1959). For instance, Gronlund (1959) in Sociometry in the Classroom wrote that the improvement of group effectiveness is of major concern to the classroom teacher(p. 234) and he proposed specic teaching strategies intended to reduce cliques and cleavagesand to improve the social integration of isolated and rejected children. Unfortunately, the theoretical and applied ambitions of this early work did not generate a rigorous empirical literature. Limitations in scale and computational ability precluded a detailed examination of quantitative variations in classroom-level features of peer ecologies. In the 1960s, the eld of social network analysis fell into a dark age(Freeman, 2004) during which theory and measurement of classroom social systems declined; and within the elds of child psychology and education, the group and the network suffered a period of neglect (Cairns, Xie, & Leung, 1998; Hanish & Rodkin, 2007). More recently, there is renewed emphasis on the variety of children's interpersonal relationships, with particular attention paid to group inuence and social power (Rubin, Bukowski, & Parker, 2006; Tseng & Seidman, 2007) and the implications of these dynamics for classroom management strategies (Farmer, 2000; Farmer et al., 2006). Journal of Applied Developmental Psychology 32 (2011) 288296 Corresponding author. Tel.: +1 814 865 3464; fax: +1 814 863 6207. E-mail addresses: [email protected] (S.D. Gest), [email protected] (P.C. Rodkin). 1 Tel.: +1 217 840 1550. 0193-3973/$ see front matter © 2011 Elsevier Inc. All rights reserved. doi:10.1016/j.appdev.2011.02.004 Contents lists available at ScienceDirect Journal of Applied Developmental Psychology

Teaching practices and elementary classroom peer ecologies

Embed Size (px)

Citation preview

Journal of Applied Developmental Psychology 32 (2011) 288–296

Contents lists available at ScienceDirect

Journal of Applied Developmental Psychology

Teaching practices and elementary classroom peer ecologies

Scott D. Gest a,⁎, Philip C. Rodkin b,1

a Pennsylvania State University, Department of Human Development & Family Studies, 110 Henderson South Building, University Park, PA 16802, USAb University of Illinois at Urbana-Champaign, Departments of Educational Psychology and Psychology, 232A Col. Wolfe School, Mail Code 422, 403 E. Healey St.,Champaign, Illinois 61820, IL, USA

⁎ Corresponding author. Tel.: +1 814 865 3464; fax:E-mail addresses: [email protected] (S.D. Gest), rodkin@

1 Tel.: +1 217 840 1550.

0193-3973/$ – see front matter © 2011 Elsevier Inc. Aldoi:10.1016/j.appdev.2011.02.004

a b s t r a c t

a r t i c l e i n f o

Available online 18 July 2011

Keywords:Peer networksFriendshipTeaching practicesClassroomBehavior management

Teachers and students in 39 1st, 3rd and 5th grade classrooms participated in a study of teaching practices andclassroom peer networks. Teachers reported on their attitudes towards aggression and withdrawal, providedrationales for their seating arrangements, and were observed on patterns of emotional and instructionalsupport and classroom organization. Students were surveyed or interviewed for peer nominations of friend-ship, popularity, being cool, and aggressive and prosocial behavior, and from these nominations classroom-level indices were derived concerning the richness of positive and negative social ties among children, theegalitarian or hierarchical nature of social status hierarchies, and classroom norms for aggressive and prosocialbehavior. Preliminary results suggest that teachers' attitudes towards social behavior are associated with thedegree of liking and disliking their students express; teachers' grouping patterns are associatedwith numerousfeatures of classroom social networks; and teachers' observed emotional support was associated with higherrates of friendship reciprocation.

+1 814 863 6207.illinois.edu (P.C. Rodkin).

l rights reserved.

© 2011 Elsevier Inc. All rights reserved.

Assertions that teachers play a crucial role in classroom peerrelationships are not new (Cairns & Cairns, 1994; Gronlund, 1959;Lewin, 1943), but conceptual models that specify the processesinvolved are more recent (Farmer et al., 2006) and empirical evidenceof these processes is uncommon. Part of the difficulty in building anempirical literature linking teaching practices to classroom peerrelations is the sheer diversity of teaching processes that could bestudied and the equally numerous aspects of peer relationshippatterns that could be impacted. Without a conceptual and measure-ment strategy to focus research efforts, there is a risk that claims ofteacher influence on classroom social processes could become anempty truism that lacks the specificity that could guide teacherprofessional development efforts. In the present paper, we aim toillustrate one strategy for focusing research efforts in this area byintegrating recent advances in the measurement of teacher–studentinteractions (Pianta, La Paro & Hamre, 2007), conceptual models ofteacher influence on social network dynamics (Farmer et al., 2006),and classroom-level measures of social networks (Ahn, Garandeau, &Rodkin, 2010; Garandeau, Ahn, & Rodkin, in press; Gest, Davidson,Rulison, Moody & Welsh, 2007). A central premise of our approach isthat teachers influence classroom social dynamics both indirectly,through general teaching practices, and more directly through activeattempts to manage the social network; and that these effects can beseen not only in terms of the experiences of individual children, butalso in the overall organization of peer network patterns.

Lewin (1943) was certain that teachers had the opportunity andresponsibility to optimize peer ecologies. During the time leading upto World War II, early studies of sociometry, or the science of themeasurement of social relationships, focused creatively on how toquantify properties of peer social ecologies, relate these properties toeducational success and failure, and build up a truly social, action-oriented theory of child social development (e.g., Bronfenbrenner, 1944;Jennings, 1943; Moreno, 1934). Bymid-century, educational psychologytexts emphasized that teachers could understand andmanagepeer socialdynamics with the assistance of research-based technologies (e.g.,Cronbach, 1954; Blair, Jones, & Simpson, 1954; see also Parsons, 1959).For instance, Gronlund (1959) in Sociometry in the Classroomwrote that“the improvement of group effectiveness is of major concern to theclassroom teacher” (p. 234) and he proposed specific teaching strategiesintended to reduce “cliques and cleavages” and to improve the socialintegration of isolated and rejected children. Unfortunately, thetheoretical and applied ambitions of this early work did not generate arigorous empirical literature. Limitations in scale and computationalability precluded a detailed examination of quantitative variations inclassroom-level features of peer ecologies. In the 1960s, thefield of socialnetwork analysis fell into a “dark age” (Freeman, 2004) during whichtheory and measurement of classroom social systems declined; andwithin the fields of child psychology and education, the group and thenetwork sufferedaperiodofneglect (Cairns, Xie,& Leung, 1998;Hanish&Rodkin, 2007). More recently, there is renewed emphasis on the varietyof children's interpersonal relationships,with particular attention paid togroup influence and social power (Rubin, Bukowski, & Parker, 2006;Tseng & Seidman, 2007) and the implications of these dynamics forclassroom management strategies (Farmer, 2000; Farmer et al., 2006).

289S.D. Gest, P.C. Rodkin / Journal of Applied Developmental Psychology 32 (2011) 288–296

There is great potential in a new generation of social networkmethodologies and statistical software packages to advance measure-ment and theory of classroom social environments.

Our larger perspective on the classroom peer ecology as a centralfeature of the classroom context is illustrated in the figure below (seeFig. 1). We view features of the peer ecology at the classroom level asproximal determinants of youth outcomes (pathway A). Features ofclassroom peer ecologies, in turn, are determined partly by generalteaching practices (pathway B) and partly by network-relatedteaching practices (pathway C). We hypothesize that peer ecologiescharacterized by more egalitarian (vs. stratified) status hierarchies,strong group norms supporting academic achievement and prosocialbehavior, tightly knit positive social ties and little aggression-basedhomophily will be associated with more positive academic adjust-ment, less aggression and bullying, and stronger experiences of socialrelatedness (pathway A). We hypothesize that indicators of generalpatterns of teacher–student interactions will predict these positiveyouth outcomes in part due to their positive impact on the peerecology (pathway B). Finally, we hypothesize that teachers whodemonstrate more intentional management of social status and socialnetworks, who actively address aggression and bullying andwho havemore accurate knowledge of the peer ecology have a more positiveclassroom environment (pathway C), even after taking into accountthe effects of overall teaching quality. Although we focus on thedirectional linkages specified in Fig. 1, bidirectional (bottom-up)processes are surely possible.

We are in the process of working with schools, teachers, andstudents to examine the conceptual model we present in Fig. 1. In thepresent paper, we present preliminary results from data collectedduring the initial year of a multi-year study. We limit considerationhere to processes at the classroom level – teaching practices andfeatures of the peer ecology – and do not consider the individualyouth outcomes that presumably flow from these processes.

Classroom teaching practices

Teacher effects are as old as Pygmalion but they are woefullyunderstudied in the literature on children's peer relationships(Wentzel, 2009). In the present analysis we examined teacherattitudes towards aggression and withdrawal, “network-relatedteaching” such as teachers' use of grouping strategies for students,and classroom observations of teaching process.

We suspected that teachers' attitudes and beliefs about aggressionand withdrawal may impact peer social ecologies in the classroom.Chang (2003) found that when teachers were warm and caring,children were less rejecting of aggressive peers than when teachers

Fig. 1. Conceptual model of teaching practices, cla

had very negative beliefs about aggression. However, when teachers hadnegative beliefs about aggression, aggressive students perceived them-selves as socially efficacious. Chang's (2003) findings suggest thatteachers' best efforts at quashing aggression can unwittingly open theway for alternative authority structures conducive to the emergence ofhigh status aggressors (McFarland, 2001). Chang et al. (2007) examinedthe effects of teacher preference for particular students on behavior-preference relationships among1300primary school children fromHongKong. Teacher preferences mediated preference-behavior relationsamong younger primary school children, such that peers adopted theirteachers' dislike of aggressors (effects were weaker for preference-withdrawal and -prosocial relationships).

In our conceptual model in Fig. 1, we distinguish between generalteacher–student interaction patterns and “network-related” teaching.Network-related teaching refers to more specific features of teacher–student interaction that may have direct relevance to the developmentof the classroom peer ecology, and that may be more likely to reflectconscious choices or strategies adopted by the teacher to impact peerrelationships (Farmer, 2000; Farmer et al., 2006). In the present study,we focus on teachers' self-reported considerations in creating a seatingchart and in organizing small groups for purposes other than readinginstruction. For example, how important is it to use seating charts orsmall groups to promote or reinforce friendships (i.e., Farmer's “socialnetworkmanagement”)? If successful, sucheffortsmight be expected toincrease the richness of friendship ties in the classroom. Alternatively,how important is it to discourage potentially problematic patterns ofhomophily by separating students who may reinforce each other'sbehavior problems (Farmer's “directmanagement” of aggression); or bycreating groups that bring together children with diverse academicskills (Farmer's “social status management”)? Separating potentialproblembehavior youthmay enhance other children's opportunities forpositive peer interactions; and creating diverse-skill groups maycounteract the academic-homophily that may otherwise occur as theresult of instructional reading groups (Hallinan & Smith, 1989) andprovide alternative routes to social status.

General teacher–student interactionpatterns arepresumably relatedto a broad range of youth outcomes, partly through their presumedimpact on the peer ecology, but also throughmore direct processes. Thisapproach to studying general classroom-level teaching processes is bestexemplified by the Classroom Assessment Scoring System (CLASS;Pianta et al., 2007), which measures broad dimensions of EmotionalSupport, Instructional Support and Classroom Organization. Measuringgeneral patterns of teacher–student interaction rests on the sameconceptual foundation as individual-focused approaches to measuringthe teacher–student relationship (Baker, 2006; Pianta, 1992) in thatproximal teacher–student interactions are presumed to account for the

ssroom peer ecologies, and youth outcomes.

290 S.D. Gest, P.C. Rodkin / Journal of Applied Developmental Psychology 32 (2011) 288–296

impact of classroom environments on students, although fewer studieshave linked classroom-level measures of teacher–student interaction tostudents' social adjustment in the classroom. Existing research on theCLASShas focusedon its associationwith students' academic adjustment(e.g., Pianta, Belsky, Bandergrift, Houts & Morrison, 2008), but a recentstudy by Cappella, Neal and Atkins (2008) begins to link CLASSdimensions to classroom social dynamics: in elementary classroomsrated high in Productivity, there were stronger group norms supportingacademic effort and prosocial behavior. This suggests that even facets ofgeneral instructional interactions may have implications for children'ssocial relationships, and bidirectional relationships are also possible.General patterns of teacher–student interaction may not be undertakenwith any deliberate intent on the part of the teacher to have a specificimpact on the classroom peer ecology. But such patterns maynonetheless be relevant to classroom social dynamics: for example,emotional support provides a positive emotional context in whichsupportive peer interactions may develop; instructional support mayentail skilled use of peer learning structures that can foster newfriendships; and effective classroom organization can minimize oppor-tunities for disruptive peer behaviors.

Classroom peer ecology

Our notion of the peer ecology (Rodkin & Gest, 2011; Rodkin &Hodges, 2003) borrows from Lewin's (1935) idea of a lifespace andBronfenbrenner's (1979) description of multileveled ecologies. Weconceptualize the classroompeer ecologyas amicrosystem, “theultimatemechanism throughwhichdevelopmentoccurs” (Bronfenbrenner, 1996,p. xv), which involves children interacting with, influencing, andsocializing one another. Teachers are not part of the peer ecology butcan impact and be impacted by it as they attempt to guide studentstowards successful learning and adjustment. Classroom peer ecologiesprovide social regularities (cf., Seidman, 1988) that organize and givestability to children's interpersonal behavior.

In the present paper we focus on three aspects of the peer ecology:(a) the overall richness of positive and negative social ties among youthin the classroom, (b) the degree to which these patterns suggest arelatively egalitarian vs. hierarchical social structure, and (c) behavioralnorms with respect to aggressive and prosocial behavior. First, it haslong been thought that classrooms with rich positive ties betweenchildren and many positive social relations lead to better academic andbehavioral outcomes (Bronfenbrenner, 1979; Roseth, Johnson, &Johnson, 2008). For example, Gronlund (1959, p. 234) wrote that “thesecurity arising frommutual acceptance in an integrated group frees thegroup members from emotional conflicts and enables them to directtheir energies towards the group task.” Second, more egalitarian ordemocratic peer ecologies are thought to be superior to those in whichsocial prestige is concentrated among just a few (e.g., Lewin, Lippitt, &White, 1939). For example, in Sherif's (1956) classic study of teams ofboys who competed and collaborated with one another at the RobbersCave summer camp, teams with more egalitarian friendship structures,where most children received a similar proportion of friendshipnominations, outperformed teams in which one or two boys receivedmost of the friendship nominations. Negative interpersonal dynamicssuch as aggression and bullying are particularly sensitive to statushierarchy dynamics: for example, Schäfer, Korn, Brodbeck, Wolke, andSchulz (2005) found that elementary classrooms that presented anearlier hierarchical structure promoted victimization by establishingfixed social positions, therefore making victimization more stablerelative to classrooms that were relatively fluid and egalitarian in thedistribution of social status.

Finally, classrooms differ in the social support they provide toaggressive and prosocial behavior (Boivin, Dodge, & Coie, 1995; Chang,2004; Wright, Giammarino & Parad, 1986). This is significant becausechildren in classrooms where aggression is normative tend to becomemore aggressive themselves, even in future years (Henry et al., 2000;

Kellam, Ling, Merisca, Brown, & Ialongo, 1998; Thomas et al., 2006).Weexamined classroom descriptive norms, which is essentially the averagelevel of behavior reported, or more specifically the proportion ofclassmates nominated by children as fitting an aggressive or prosocialpeer nomination. In addition,we considerednorm salience, or thedegreeto which aggressive and prosocial behavior is associated with socialstatus (i.e., being popular or cool) in a particular classroom. Thepopularity of aggressive behavior, and not just its frequency, wasunderscored in a recent report by Dijkstra, Lindenberg, and Veenstra(2008), who found that in Dutch secondary school classrooms wherebullies were popular, the association between peer rejection andbullying by any child was attenuated. Classroom norms may be setprimarily by just a few popular children, who are likely interconnected(Witvliet et al., 2010), and these influential popular associations whendirected towards antisocial and prosocial behavior can affect the peersocial ecology as a whole.

In summary, our aim in this study is to examine the extent towhich teachers' attitudes towards aggression and withdrawal andnetwork- and general patterns of teaching processes are associatedwith three key features of their classroom peer ecologies. Based on thelimited empirical literature, we had a limited number of specificexpectations. With regard to the overall richness of positive andnegative social ties among children, we expected that general levels ofteacher emotional support would provide a positive emotionalcontext for the development of rich friendship ties, and that teachers'explicit efforts to promote friendships and to reduce behavioraldisruptions by separating students at risk of behavior problems wouldfoster a setting more conducive to friendship development. Withregard to classroom status hierarchies, we expected that teachers whosought to counteract the predominance of academic skills as a socialorganizer – structurally built in to most classrooms in the form ofinstructional reading groups – would have less prominent statushierarchies. Finally, we expected that peer norms supportingprosocial behavior and opposing aggression would be most evidentin classrooms where teachers provided emotional support (modelingprosocial behavior) and took specific action to manage aggression(separating youth at risk for behavior problems).

Methods

Participants

Participants were teachers and students from 39 classrooms in 1st(N = 15), 3rd (N = 14) and 5th (N = 10) grade across two researchsites. One site consisted of schools in a small to mid-sized city (pop.70,000) in Illinois in the Midwestern U.S., with significant racial/ethnic diversity (43% African American, 8% Asian, and 3% Hispanic).Approximately 44% of all students in the district are designated aslow-income. The other site is located in a rural area in Pennsylvania inthe northeastern U.S. The population served by the schools is over99% European American, with 35% classified as economicallydisadvantaged.

Procedures

All 1st, 3rd and 5th grade teachers at the participating schools wereinvited to participate in the study. Forty-one of the forty-two eligibleteachers agreed to participate by signing an informed consent form thathad been approved by the respective University Institutional ReviewBoards at each site. All children in the classrooms of consenting teacherswere invited to participate in the study. An informed parental consentform approved by each site's University IRB was sent home with eachstudent. A classroom pizza party was offered to any classroom in which90% of students returned a consent form that had been signed by aparent, regardless of the percentage of parents who agreed to ordeclined their child's participation. Over 90% of all parental consent

291S.D. Gest, P.C. Rodkin / Journal of Applied Developmental Psychology 32 (2011) 288–296

formswere returned in all but two classrooms. Overall, 76.0% (596/784)of all children in the participating classrooms received parental consentto participate in the study. The two classrooms with parental consentrates below 60% were excluded from analyses because classroom-levelmeasures of the peer ecology have diminished validity when they arebased on a small proportion of individuals in the setting. Forcomparison, in the largest national study of peer networks, Add Health,a looser criterion of 50% participation is used.

Two waves of data were collected in the mid-to-late portion of theschool year (January and April at the first site, March and May at thesecond site). At each wave, teachers completed a packet ofquestionnaires, and students who received parental consent wereinvited to complete surveys in individual interviews (1st Grade) or ingroup-administered paper-and-pencil surveys (3rd and 5th Grades).On the date of the survey, student assent to participation was soughtbefore surveys were administered, orally for 1st grade students and inwritten form for 3rd and 5th grade students.

Measures of teaching practices

Teacher attitudes toward aggression and withdrawalTeacher attitudes about aggressive and shy or withdrawn behavior

were measured with items adapted from the scale developed byChang (2003). Six items describing disapproval of aggression formedan internally consistent scale (α = .61): “Aggression must bepunished”, “I feel angry when students take advantage of others”, “Ifeel angry when students bully others”, “Bullies must be disciplined”,“I tolerate aggression” (reversed) and “It's normal for kids to beaggressive sometimes” (reversed). Three items describing support forshy-withdrawal formed an internally consistent scale (α = .78):“Extra patience is needed with those who are reticent andwithdrawn”, “I take extra care of those who withdraw from peers”,and “I am protective of those who are shy and timid”.

Teacher-reported grouping strategiesTeachers completed surveys in which they were asked to rate the

importance of each of several considerations when they (a) createdthe classroom seating chart and (b) assigned students to small groupsfor purposes other than reading instruction. For each groupingcontext, the teacher was asked “Please rate how important each ofthe following considerations was when you (created the seatingchart/assigned students to a small group)”. Each consideration wasrated on a 5-point Likert scale with response options ranging from Notat all important (= 1) to Somewhat important (= 3) and Veryimportant (= 5). The importance of each consideration was reliablycorrelated across the two grouping contexts (cross-context correla-tions are reported below), so we averaged teachers' ratings for eachitem across the two contexts to arrive at one score each forconsideration. Two of the items addressed academic considerations:“To place students together who have diverse skill levels” (r = .39)and “to place students together who have similar skill levels” (r =.35). Two other items described friendship considerations: “to placechildren together with others who are not yet their friends (e.g., topromote new friendships and social connections)” (r = .51), and “Toplace children together who are already friends (e.g., to respectstudent preferences and/or reinforce those friendships)” (r = .66).One item described a behavior management consideration: “Toseparate students who might pose behavior problems if they were(seated next to each other/in the same group)” (r = .44).

Observed teacher behaviorThe Classroom Assessment Scoring System (CLASS; Pianta, La Paro,

& Hamre, 2006) is an observational measure that assesses 10dimensions of teacher–student interaction. These 10 dimensions aresummarized in three higher-order composite scores. EmotionalSupport includes four dimensions: Positive Climate (emotional

connection with students), Negative Climate (anger, hostility oraggression), Teacher Sensitivity (responsivity to student needs) andRegard for Student Perspectives. Instructional Support includes threedimensions: Concept Development (instructional discussions focusedon higher-order thinking), Quality of Feedback (extending studentlearning), and LanguageModeling (facilitate and encourage). Classroomorganization includes dimensions that assess “processes related to theorganization and management of students' behavior, time andattention” (Pianta, Hamre, & Stuhlman, 2003, p. 3): Behavior Manage-ment (monitoring, preventing and redirecting behavior), Productivity(clear and efficient routines), and Instructional Learning Formats(facilitation of varied and effective learning activities). Each CLASSitem is rated on a 7-point Likert scale. Observers complete four 30-minute rating cycles during a visit to the classroom. Each cycle consistsof an observation period of 20 consecutive minutes followed by 10 minto complete the 10 ratings. Ratings for each item are then averagedacross the four cycles. At each site, a team of research assistants wastrained and certifiedby certifiedCLASS trainers. Fordata collection, pairsof trained observers visited each classroomon the samemorning duringthe first 2.5 h of the school day. The two observers completed all fourcycles of observations independently, and the two observers' scoreswere averaged for analysis. The specific observers who were pairedwere rotated across classrooms to prevent the emergence of systematicdifferences in rater-pairs. Pairs of observers were employed because adetailed reliability study of the CLASS revealed that variance associatedwith raters is more pronounced than variance due to the number ofdifferent days observed or the number of observation cycles per day(Raudenbush,Martinez, Bloom, Zhu & Lin, 2007). Intraclass correlations(ICCs), computed separately at each research site, indicated acceptableinter-rater agreement for the three higher-order scores (ICCs range .71to .85).

Measures of classroom peer ecology

The student survey included a series of questions about classmatesthat required respondents to identify classmates (“students in yourclass”) who met particular descriptors. Items of interest for thecurrent analyses focused on: friendship, liking, disliking, perceivedpopularity, aggressive behavior and prosocial behavior. Specific itemsare described below in the context of the indices derived from them.For each item, respondents identified as few or as many classmates asthey wished. Respondents identified peers by stating their names (1stGrade) or by circling their names on a printed survey (3rd and 5thGrades).Given the modest sample size, we sought to maximizemeasurement reliability and reduce the number of statistical tests byaggregating data across items and waves whenever possible. Forindices aggregated across multiple peer report items, we report theinternal consistency of items within each wave, and the stability ofeach index across waves, r12.

Positive and negative social tiesPositive and negative social ties were defined by reports of liking

(“These are the kids Iwould likemost to playwith”), disliking (“These arethe kids I would like least to play with”), and friendships (“Who are yourfriends in this classroom?”). Three indices summarized classroom-levelpatterns of liking and disliking. Liking density (r12 = .81) was the sum ofall “like most” ties existing in the classroom divided by the total possiblenumber of such ties [i.e., n (n− 1), where n is the number of studentsparticipating in the survey]. Disliking density (r12 = .74) was computedsimilarly. To capture the overall valence of classroom liking anddisliking, the liking/disliking ratio (r12 = .71) was computed by dividingliking density by disliking density. Two additional indices summarizedpatterns of friendship nominations. Friendship density (r12 = .76) wascomputed in the samemanner as the other densitymeasures. Friendshipreciprocity (r12= .66)was the proportion of all friendship ties thatwerereciprocated.

292 S.D. Gest, P.C. Rodkin / Journal of Applied Developmental Psychology 32 (2011) 288–296

Status hierarchiesClassroom status hierarchies were identified based on the

distribution of scores for six peer-nomination items measuring liking,disliking and perceived popularity. In addition to the liking anddisliking nominations described above, four items described perceivedpopularity: “these are the most popular kids in my class”, “these arethe kids in my class who are not popular”, “these are the coolest kidsin my class”, and “everyone listens to what these kids have to say”. Foreach of these six items, we quantified the degree to whichnominations received were concentrated among one or a fewindividuals or more equally spread across classmates. Specifically,each student's in-degree centrality was computed (i.e., total numberof nominations received), and the distribution of in-degree centralityscores in each classroom was summarized in a centralization index(Wasserman & Faust, 1994). A centralization score is intuitivelysimilar to a variance score in that it summarizes the amount of spreadin a distribution, but is scaled so that a value of 0 indicates amaximally egalitarian setting (i.e., all students receive the samenumber of nominations) and a value of 1 indicates a maximallyhierarchical setting (i.e., one student receives all of the nominationsmade); this scaling facilitates comparisons across classrooms.

For each item, we computed centralization both for the classroomas a whole and separately for girls and for boys (due to researchsuggesting possible sex differences in status dynamics; Adler & Adler,1998). At the whole-classroom level, the centralization scores for thesix relevant items were moderately to strongly intercorrelated andformed highly internally consistent factors (α1 = .84, α2 = .81) thatwere strongly correlated across waves (rw12 = .85). The within-gender centralization indices were not as consistently intercorrelatedwithin wave and so no within-wave composite scores were created.However, centralization scores for two items were reliably correlatedacross waves for both girls and boys, and so these items are analyzedindividually: “not popular” r12g = .35, r12b = .47; “coolest” r12g = .52,r12b = .47, all p b .05.

Behavioral normsFor each classroom, we examined both descriptive norms, summa-

rizing how common particular behaviors were, and norm salience,characterizing the extent to which a particular behavior wasassociated with peer status. Both types of norms were calculatedwith respect to peer-nominated aggression and prosocial behavior.The four aggression items were “these kids starts fights; they pushother kids around, or hit them or kick them”, “these kids get intotrouble; they don't follow the rules, don't pay attention and talk backto the teacher”, “says mean things about other people” and “these kidsmake fun of people; they like to make fun of kids and embarrass themin front of other people”; and “these kids say mean things to otherkids, and they spread nasty rumors about other kids”. The twoprosocial behavior items were “these kids cooperate; they pitch in,share and give everyone a turn” and “these kids are always willing todo something nice for somebody else”. Descriptive norms werecalculated in terms of the density of each peer-nominated behavior.Density scores for items in each behavioral domain were averagedwithin-wave (aggression α1 = .93, α2 = .90; prosocial behavior r1 =.92, r2 = .87), and then averaged across waves (aggression r12 = .80;prosocial r12 = .82).

Norm salience scores were calculated as follows. First, at eachwave, individual student composite scores for each behavior werecalculated as the average number of nominations received for therelevant items (aggressionα1= .92,α2= .91; prosocial r1= .81, r2=.79). Next, within each classroom, these individual-level behaviorscores were correlated with the individual-level scores for each of thesix items measuring different facets of classroom status hierarchies(described in the Status hierarchies section). These within-classroomcorrelations were highly consistent across the different status items,so they were averaged to create a single norm salience score for each

behavior at each wave (aggressionα1 = .90,α2 = .91; prosocialα1 =.81, α2 = .92). Finally, because these norm salience scores weremoderately to strongly correlated across waves (aggression r12 = .83;prosocial r12 = .58), scores were averaged across waves to arrive at asingle norm salience score for each behavior.

Results

Preliminary analyses

Preliminary analyses focused on identifying appropriate controlvariables tomaximize the interpretability of any associations betweenteaching practices and measures of the peer ecology. Class size isimportant because network measures of density and reciprocity arenecessarily related to network size. Specifically, when class sizeincreases by one, children do not systematically name more friends,yet the denominator for density measures increases by n − 1, drivingdown density measures. Reciprocity rates are lower in largernetworks for similar reasons. In the current study, class size wasmoderately negatively related to measures based on density andreciprocity (r = −.41 to r = −.61). Class size was also negativelycorrelated with indices of network centralization (hierarchy) becauseit is more difficult for a few students to monopolize nominations inlarge classes (r = −.28 to r = −.45). Class size was not reliablycorrelated with norm salience (r = −.17).

Grade level is relevant for network measures because olderchildren typically identify more classmates in response to each peernomination item, producing positive associations with density-basedmeasures. As expected, grade level was positively correlated withdensity and reciprocity measures (r= .39 to r= .65). Grade level wasalso positively correlated with network centralization (r = .38 to r =.49), but was not reliably correlatedwith norm salience (r= .11 to r=−.31).

Research site was used as a proxy for the several potentiallyimportant demographic differences in the two communities sampledfor this study: in comparison to the Illinois community, thePennsylvania community was more rural, characterized by greaterracial/ethnic homogeneity, lower rates of poverty, and had smallerclass sizes. Site was not systematically associated with differences inmeasures of density, reciprocity or centralization, |r| b .30; but normsalience scores differed across communities. For example, prosocialbehavior was positively associated with peer status in classrooms inboth communities (MPA = .71, MIL = .38), but this approval wassignificantly stronger in the Illinois community, F(1,38) = 16.1, p b

.001. To account for these potentially confounding effects, wecomputed partial correlations between each measure of teachingpractices and each of the measures of the peer ecology, controlling forclass size, grade level and research site.

Descriptive statistics

Teachers generally rated themselves as disapproving of aggressivebehavior and providing support for shy–withdrawn students (e.g.,Mean ratings above 4 on the 5-point scale; see Table 1), but the strengthof these feelings ranged from moderate to very strong. In terms ofgrouping strategies, the teachers in this sample described separatingstudents with behavior problems as a strong consideration. Promotingacademic diversity and new friendships were also moderate to strongconsiderations. Teachers generally did not place a strong emphasis oncreating groups that were academically homogenous or that wereorganized around existing friendships. Consistent with other researchusing the CLASS, teachers were observed to provide moderately highlevels of Emotional Support and Classroom Organization (Mean scoresabove scale mid-point), but only moderate levels of InstructionalSupport (around scale midpoint).

Table 1Descriptive statistics for measures of teaching practices and classroom peer ecologies.

Percentile

Mean SD 10th 90th

Teaching practicesTeacher-rated attitudes

Disapproval of aggression 4.27 .67 3.33 5.00Support for shy–withdrawal 4.16 .68 3.00 5.00

Teacher-rated grouping strategiesPromote new friendships 3.11 .87 2.00 4.50Reinforce existing friendships 2.32 .85 1.00 3.00Promote academic homophily 2.32 .95 1.00 3.50Promote academic diversity 3.66 .99 2.00 5.00Separate behavior problems 4.58 .76 3.40 5.00

Observed teacher behaviorEmotional support 5.26 .52 4.53 6.03Classroom organization 5.09 .54 4.45 5.87Instructional support 3.78 .70 2.96 4.83

Classroom peer ecologiesPositive and negative social ties

Liking density .24 .08 .15 .39Disliking density .22 .07 .14 .31Liking/disliking ratio 1.19 .43 .72 1.74Friendship density .36 .15 .19 .58Friendship reciprocity .38 .14 .17 .55

Status hierarchyWhole classroom .28 .08 .19 .41Girls — Not popular .35 .17 .18 .64Girls — Coolest .33 .13 .18 .52Boys — Not popular .35 .15 .16 .59Boys — Coolest .33 .14 .19 .58

Behavioral normsDescriptive: Aggression density .17 .06 .12 .27Descriptive: Prosocial density .30 .12 .17 .47Norm salience: Aggression −.17 .34 −.57 .30Norm salience: Prosocial .56 .31 .14 .90

Table 2Teaching practices and the richness of positive and negative classroom ties.

Likingdensity

Dislikingdensity

Liking/disliking ratio

Friendshipdensity

Friendshipreciprocity

Teacher-rated attitudesDisapproval of aggression −.02 −.25 .14 −.02 −.16Support for shy–withdrawal

.15 −.38⁎ .44⁎ .01 −.07

Teacher-rated grouping strategiesPromote new friendships −.16 −.22 .06 −.25 −.30+

Reinforce existingfriendships

−.01 −.07 .03 .00 −.18

Promote academichomophily

−.22 .10 −.30+ −.07 −.05

Promote academicdiversity

.03 −.22 .24 .03 .10

Separate behaviorproblems

.24 −.33+ .40⁎ .41⁎ −.02

Observed teacher behaviorEmotional support .18 .08 .14 .21 .39⁎

Classroom organization −.02 −.11 .16 −.05 .11Instructional support −.07 −.09 .11 −.05 .08

Note. Entries are partial correlations, controlling for grade levels (1st, 3rd, and 5th),number of students in classroom and research site. N = 39.+p b .10. ⁎p b .05.

Table 3Teaching practices and classroom social status hierarchies.

Girls Boys

Teaching practices Wholeclassroom

Notpopular

Cool Notpopular

Cool

Teacher-rated attitudes

293S.D. Gest, P.C. Rodkin / Journal of Applied Developmental Psychology 32 (2011) 288–296

In terms of peer network features, the average ratio of liking todisliking was greater than 1.0, t(37) = 2.74, p b .001, indicating thatclassrooms were generally characterized by a preponderance ofpositive rather than negative social ties, but this tendency variedwidely: several classrooms had more disliking than liking (ratiosbelow 1.0), and others had nearly twice as much liking as disliking.Status hierarchies were generallymodest and similar inmagnitude forgirls and boys. The density of peer-nominated prosocial behavior andacademic competence was significantly higher than the density ofpeer-nominated aggression, F(1,37) = 54.34 and F(1,37) = 65.77,both p b .001. Norm salience scores for aggression were generallynegative, t(38) = −15.14, p b .001, indicating that higher levels ofaggression were generally associated with lower peer status; but notethat in some classrooms there was a weak positive associationbetween aggression and peer status. Prosocial behavior and academiccompetence were positively associated with peer status in everyclassroom, but the strength of these norms varied from very weak tovery strong.

Disapproval of aggression −.24 −.11 −.01 .05 −.15Support for shy–withdrawal .13 −.12 .05 .14 .28

Teacher-rated grouping strategiesPromote new friendships −.42⁎ −.42⁎ −.48⁎⁎ −.10 −.26Reinforce existing friendships −.33+ −.26 −.26 −.20 −.29Promote academic homophily −.03 .16 .06 .16 −.15Promote academic diversity −.07 −.07 −.36⁎ −.01 −.21Separate behavior problems −.16 −.03 −.23 .01 −.28

Observed teacher behaviorEmotional support .11 .30+ .18 .06 .01Classroom organization −.17 .09 −.06 −.20 −.15Instructional support −.28 .04 −.21 −.39⁎ −.24

Note. Entries are partial correlations, controlling for grade level (1st, 3rd, and 5th),number of students in classroom and research site. N = 39.+p b .10. ⁎p b .05. ⁎⁎p b .01.

Associations between teaching practices and classroom peer ecologies

Positive and negative social tiesTeachers who reported being more sympathetic and attentive to

the needs of shy–withdrawn students had classrooms with lowerdisliking density and a higher ratio of liking to disliking (see Table 2).In classrooms where teachers reported that separating students whoposed behavior problems was a major consideration in creatingseating charts and small groups, there was a higher ratio of liking todisliking and a higher density of friendships. Finally, teachers whowere observed to provide a higher level of emotional support tostudents had classrooms with higher rates of friendship reciprocity.

Status hierarchiesTeachers who reported that fostering new friendships was a major

consideration in creating seating charts and small groups hadclassrooms with a less pronounced status hierarchy (see Table 3).Among girls, this strategy was associated with a more equaldistribution of nominations as “not popular” and “coolest”. Teacherswho emphasized creating academically diverse groups also had lesspronounced “cool” hierarchies among girls. Teachers who wereobserved to provide higher levels of instructional support had lesspronounced “not popular” hierarchies among boys.

Behavioral normsThe strength of teachers' disapproval of aggression was negatively

associated with the density of peer-nominated aggression (i.e.,descriptive norms for aggression) suggesting that peers perceivedless aggression in classrooms where teachers expressed the strongestdisapproval of aggression (see Table 4). Fostering new friendshipsthrough seating and small-group arrangements was positively

Table 4Teaching practices and classroom behavioral norms.

Descriptive norms Norm salience

Agg Pro Agg Pro

Teacher-rated attitudesDisapproval of aggression −.43⁎ .12 −.10 .20Support for shy–withdrawal −.21 .00 −.01 −.12

Teacher-rated grouping strategiesPromote new friendships −.18 −.21 .43⁎ −.47⁎⁎

Reinforce existing friendships .04 .08 .34+ −.24Promote academic homophily −.02 −.13 .17 .11Promote academic diversity −.07 .17 .28+ −.35⁎

Separate behavior problems .12 .22 .03 −.21

Observed teacher behaviorEmotional support .12 .23 −.02 −.18Classroom organization −.15 .03 .04 −.20Instructional support .07 .07 .20 −.41⁎

Note. Entries are partial correlations, controlling for grade levels (1st, 3rd, and 5th),number of students in classroom and research site. N = 39.+p b .10. ⁎p b .05. ⁎⁎p b .01.

294 S.D. Gest, P.C. Rodkin / Journal of Applied Developmental Psychology 32 (2011) 288–296

associated with aggression norm salience, indicating less peerdisapproval of aggression; and negatively associated with prosocialnorm salience, indicating weaker peer approval for prosocialbehavior. Similarly, making efforts to promote academic diversityand observer-rated instructional support were negatively associatedwith prosocial norm salience, indicating weaker peer approval forprosocial behavior.

Discussion

This study provides a simple but direct test of several long-standingpropositions about teachers' influence on the classroom social environ-ment. Much of the historical literature on these issues has been limitedto broad theoretical assertions (Lewin, 1943) or a combination ofanecdotal and small-scale empirical reports (Gronlund, 1959). Morerecent efforts to understand the relational context of the classroomhavetypically focused on the individual level of analysis. These studiesdemonstrate that within-classroom variations in teachers' perceptionsof their relationships with students contribute to important academicand social outcomes across elementary school (e.g., Baker, 2006), butthey do not illuminate how between-teacher differences in objectivelyobserved teaching strategies or relational processes relate to classroom-level peer processes. In contrast, a smaller but compelling literaturefocuses on how interventions designed to systematically alter broadpatterns of teacher–student interaction can alter students' experience ofthe classroom or school-level social environment (Battistich, Schaps,Watson & Solomon, 1996). These efforts involve setting-level efforts toalter specific teaching practices, but the key outcome variables aretypically individual student perceptions of the social environment thatare aggregated to serve as setting-level indicators (Tseng & Seidman,2007). In this context, one advantage of the current study is the focus onmeasures of teaching processes and classroom friendship patterns thatare inherently at the setting level, and thus focus attention on between-classroomdifferences. The setting-level research design allows us to usesimple correlations to test several longstanding hypotheses regardingthe association between teaching practices and classroom-level peersocial dynamics.

Consistent with Chang's research (2003; Chang et al., 2007),teachers' self-rated attitudes regarding aggression and shy–withdrawnbehavior were associated with classroom level social processes.Teachers who expressed strong empathy and support for shy–withdrawn students had classrooms with a lower density of dislikingand a higher ratio of liking to disliking, while teachers who expressedthe strongest disapproval of aggression had classroomswhere studentsperceived fewer classmates to be aggressive.

Teachers' self-reported considerations in creating classroom seatingcharts and in comprising small groups for social and instructionalpurposes (other than reading groups) were related to several aspects ofclassroom friendships. Separating students who might pose behaviorproblems was an important consideration for most teachers, yetvariations in the emphasis that teachers placed on this strategy werenonetheless associatedwith classroom-level patterns of liking, dislikingand friendship. Teachers who placed the strongest emphasis onseparating students with behavior problems had students whoexpressed a stronger predominance of liking over disliking, andreported denser friendship networks. It may be that, to the extent thatteachers are successful in preventing behavioral disruptions byseparating these youth, they reducepeer conflicts that underlie dislikingrelationships and provide a context in which positive social ties canemerge. This represents an active form of social network managementthat may be a direct response to patterns of aggression in the classroom(Farmer, 2000).

Teachers varied widely in the extent to which they attempted tofoster new friendships through seating arrangements and smallgroup assignments. This strategy was associated with a complexprofile of peer network features. On the one hand, attempting topromote new friendships was associated with a less pronouncedclassroom status hierarchy, and in particular, a less hierchicallystratified status system among girls. This could reflect the fact thatteachers' active attempts to provide opportunities for newaffiliationsdisrupts the complex status dynamics related to inclusion andexclusion among girls (Adler & Adler, 1998), resulting in less rigidhierarchies. In this regard, it is also noteworthy that this strategywasassociated with a trend (p b .10) toward lower friendship reciprocity,which would be consistent with a less “settled” friendship network.On the other hand, promoting new friendships was also associatedwith a less desirable profile of norm salience: less disapproval ofaggression and weaker approval of prosocial behavior. Given ourcross-sectional research design, one explanation for this unexpectedfinding is that teachers who become concerned about the peer normsin their class may begin to attempt to manipulate friendshipdynamics through grouping practices. Of course, the effectivenessof direct attempts by teachers to foster friendship ties would seem todepend on teachers having an accurate understanding of existingfriendships patterns in the classroom. Yet existing research suggeststhat many teachers have a poor understanding of classroomfriendship patterns (Gest, 2006; Pearl, Leung, Van Acker, Farmer, &Rodkin, 2007). If that is true, then “setting the stage” for suchrelationships to emerge naturally (by establishing a positiveemotional tone and reducing the likelihood of classroom disrup-tions)may bemore effective than attempting to engineer friendshipsmore directly. Peer relationships are a powerful source of socialinfluence on children, and are just as likely (if not more likely) to bemanaged ineffectively by adults than to be productively engaged(Dishion & Tipsord, 2011; Farrington & Ttofi, 2009; Larson, Hansen, &Moneta, 2006; McFarland, 2001).

Teachers' attempts to create academically diverse groups wereassociated with a similar effect as the strategy of promoting newfriendships: a more egalitarian status hierarchy among girls, and lessdesirable patterns of norm salience. Farmer et al. (2006) emphasize theimportance of providing students with multiple prosocial routes tosocial status to prevent students from disengaging or resorting todisruptive behavior to gainpeer status. Given that classroom friendshipscan emerge from instructional reading groups (Hallinan & Smith, 1989)and that academic achievement is a correlate of social status inelementary-age youth (Rubin et al., 2006), teachers who seek toorganize classrooms along alternative linesmayweaken the emergenceof status hierarchies related to achievement (cf., Rosenholtz & Simpson,1984). The fact that this was only true for girls suggests thatachievement may be a stronger organizer of friendship patternsamong girls than among boys.

295S.D. Gest, P.C. Rodkin / Journal of Applied Developmental Psychology 32 (2011) 288–296

Teachers who were rated as providing high levels of emotionalsupport to students had classrooms with higher rates of friendshipreciprocation. To receive high ratings for emotional support, teachershad to demonstrate a positive emotional connection to students, lowlevels of anger or hostility, high levels of responsiveness to studentneeds and positive regard for student perspectives. At the simplestlevel, teachers who display these behaviors are modeling many of thevery relational skills that characterize children's mutual friendships(Hartup, 1996). Beyond simple modeling, teachers who establish awarm and supportive classroom environment are providing a contextwithin which children can safely explore and practice positive peerrelational skills.

Instructional Support, evidenced by discussions oriented aroundhigher-order thinking, high-quality feedback to students and facili-tating complex language use, was associated with less pronouncedstatus hierarchies among boys and weaker support for prosocialbehavior. We are reluctant to interpret the association with boys'status hierarchies given that this effect emerged for only one of thetwo status dimensions. The association between instructional supportand weaker support for prosocial behavior was not anticipated. Onepossibility is that teachers who provide a particularly stimulatingclassroom environment may provide a context in which a wider rangeof student talents or competencies provide routes to peer status,thereby attenuating the association between prosocial behavior andstatus.

Classroom Organization was not associated with any of thefeatures of peer networks but it remains plausible as a correlate offriendship patterns in that a highly disorganized classroom mightmake it difficult to establish new friendship ties or to establish smoothreciprocal interactions; and Capella et al. (2008) found links betweena component of classroom organization (Productivity) and prosocialbehavior. It is possible that wewould have observed similar patterns ifwe had tested the narrow-band dimensions underlying classroomorganization: given the small sample of classrooms in the presentstudy, we opted not to do so in order to minimize the number ofstatistical tests.

As noted above, a unique strength of this study is the use ofclassroom-level measures of teaching processes and the organizationof classroom peer networks. There are also several importantlimitations. Though the study involved the participation of hundredsof youth from a very diverse range of classrooms and schools, the keyanalyses were based on only 39 classrooms, which provided limitedstatistical power, and virtually no power to detect weak butconceptually interesting effects. Moreover, while we were able tocontrol for the main effects of grade level on the friendship networkindices, we could not test whether associations between teachingpractices and classroom friendship patterns varied across develop-mental level or across classrooms with varying racial/ethnic compo-sition. Developmental differences are not self-evident: for example,young children as individuals may be more open to direct teacherinfluence, but their peer networks may be only loosely coupled andtherefore more difficult to manipulate; whereas older individualchildren may be less open to teacher influence, but their more tightlycoupled peer networkmay bemore responsive to activemanagement.The present report is from the first phase of a larger research programthat will involve collecting similar data in a larger number ofclassrooms that should allow for tests of these possibilities. With alarger sample of classrooms and schools, we will havemore, andmorereliable, findings to report on dynamic associations between teachingpractices, classroom peer ecologies, and also individual youthoutcomes.

Finally, the premise of this paper is that the direction of influenceflows from teaching practices to peer experiences (and not thereverse), but as noted in our interpretation of findings regarding normsalience, it is possible that teachers adopt different teaching strategiesin response to the peer network dynamics of their classroom (e.g.,

adopting a more positive emotional tone in response to a highlyreciprocal classroom friendship network). In framing our introductionaround the role of teachers in shaping peer environments, we arefollowing a long tradition in educational, social and developmentalpsychology and special education (Farmer, 2000; Gronlund, 1959;Lewin, 1943). Longitudinal data on both teaching practices andclassroom friendship patterns could permit useful empirical explorationof these possibilities.

Overall, these findings provide promising initial empiricalevidence consistent with the longstanding view that teachersexercise an “invisible hand” in the classroom that may play a keyrole in shaping patterns of classroom friendships. Key issues forfuture research include clarifying the relative strength and interplaybetween teachers' indirect influence over networks (general teach-er–student interactions) and their more direct attempts to managenetworks, including the degree to which teachers themselves areaware of their influence. By combining direct observations ofteaching practices with classroom-level indices of peer networkpatterns derived from social network analysis, it should be possibleto build a rigorous empirical literature on these issues to complementthe existing rich theoretical traditions.

References

Adler, P. A., & Adler, P. (1998). Peer power: Preadolescent culture and identity. NewBrunswick, NJ: Rutgers University Press.

Ahn, H. -J., Garandeau, C. F., & Rodkin, P. C. (2010). Effects of classroom embeddednessand density on the social status of aggressive and victimized children. Journal ofEarly Adolescence, 30, 76−101.

Baker, J. A. (2006). Contributions of teacher–child relationships to positive schooladjustment during elementary school. Journal of School Psychology, 44, 221−229.

Battistich, V., Schaps, E., Watson, M., & Solomon, D. (1996). Prevention effects of the ChildDevelopment Project: Early findings from an ongoing multisite demonstration trial.Journal of Adolescent Research, 11, 12−35.

Blair, G. M., Jones, R. S., & Simpson, R. H. (1954). Educational psychology. New York, NY:Macmillan.

Boivin, M., Dodge, K. A., & Coie, J. D. (1995). Individual-group behavioral similarity andpeer status in experimental play groups of boys: The social misfit revisited. Journalof Social and Personality Psychology, 69, 269−279.

Bronfenbrenner, U. (1944). A constant frame of reference for sociometric research: IIExperiment and inference. Sociometry, 7, 40−75.

Bronfenbrenner, U. (1979). The ecology of human development: Experiments by natureand design. Cambridge, MA: Harvard University Press.

Bronfenbrenner, U. (1996). Foreword. In R. B. Cairns, G. H. ElderJr., & E. J. Costello (Eds.),Developmental Science (pp. ix−xvii). New York, NY: Cambridge University Press.

Cairns, R. B., & Cairns, B. D. (1994). Lifelines and risks: Pathways of youth in our time.Cambridge, UK: Cambridge University Press.

Cairns, R. B., Xie, H., & Leung, M. -C. (1998). The popularity of friendship and the neglectof social networks. New Directions for Child Development, 80, 25−53.

Capella, E., Neal, J. W., & Atkins, M. (2008). Links to learning: Teacher practices andstudent social networks in urban classrooms. Paper Presented at the 2008 AnnualMeeting of the American Educational Research Association, New York, NY.

Chang, L. (2003). Variable effects of children's aggression, social withdrawal, and prosocialleadership as functions of teacher beliefs and behaviors. Child Development, 74,535−548.

Chang, L. (2004). The role of classroom norms in contextualizing the relations ofchildren's social behaviors to peer acceptance. Developmental Psychology, 40,691−702.

Chang, L., Liu, H., Fung, K. Y., Wang, Y., Wen, Z., Li, H., & Farver, J. M. (2007). Themediating and moderating effects of teacher preference on the relations betweenstudents' social behaviors and peer acceptance. Merrill-Palmer Quarterly, 53,603−630.

Cronbach, L. J. (1954). Educational psychology. New York, NY: Harcourt, Brace.Dijkstra, J. K., Lindenberg, S. M., & Veenstra, R. (2008). Beyond the classroom norm: The

influence of bullying of popular adolescents and its relation to peer acceptance andrejection. Journal of Abnormal Child Psychology, 36, 1289−1299.

Dishion, T. J., & Tipsord, J. M. (2011). Peer contagion in child and adolescent social andemotional development. Annual Review of Psychology, 62, 189−214.

Farmer, T. W. (2000). Social dynamics of aggressive and disruptive behavior in school:Implications for behavior consultation. Journal of Educational and PsychologicalConsultation, 11, 299−322.

Farmer, T. W., Goforth, J. B., Hives, J., Aaron, A., Jackson, F., & Sgammato, A. (2006).Competence enhancement behavior management. Preventing School Failure, 50(3),39−44.

Farrington, D. P., & Ttofi, M. M. (2009). School-based programs to reduce bullying andvictimization. Campbell Systematic Reviews, 6.

Freeman, L. C. (2004). The development of social network analysis: A study in the sociologyof science. Vancouver, BC: Booksurge.

296 S.D. Gest, P.C. Rodkin / Journal of Applied Developmental Psychology 32 (2011) 288–296

Garandeau, C. F., Ahn, H.- J., & Rodkin, P. C. (in press). The social status of aggressionacross classroom contexts: The role of peer status hierarchy, classroom academiclevel, and grade. Developmental Psychology.

Gest, S. D. (2006). Teacher reports of children's friendships and social groups:Agreement with peer reports and implications for studying peer similarity. SocialDevelopment, 15, 248−259.

Gest, S. D., Davidson, A. J., Rulison, K. L., Moody, J., & Welsh, J. A. (2007). Features ofgroups and status hierarchies in girls' and boys' early adolescent peer networks. InP. C. Rodkin & L. D. Hanish (Eds.), Social Network Analysis and Children's PeerRelationships (pp. 43−60). San Francisco, CA: Jossey Bass.

Gronlund, N. E. (1959). Sociometry in the classroom. New York, NY: Harper & Brothers.Hallinan, M. T., & Smith, S. S. (1989). Classroom characteristics and student friendship

cliques. Social Forces, 67(4), 898−919.Hanish, L. D., & Rodkin, P. C. (2007). Bridging children's social development and social

network analysis. In P. C. Rodkin & L. D. Hanish (Eds.), Social Network Analysis andChildren's Peer Relationships (pp. 1−8). San Francisco, CA: Jossey-Bass.

Hartup, W. W. (1996). The company they keep: Friendships and their developmentalsignificance. Child Development, 67, 1−13.

Henry, D., Guerra, N., Huesmann, R., Tolan, P., Van Acker, R., & Eron, L. (2000).Normative influences on aggression in urban elementary school classrooms.American Journal of Community Psychology, 28(1), 59−81.

Jennings, H. (1943). Leadership and isolation: A study of personality in inter-personalrelations. New York, NY: Longmans, Green.

Kellam, S., Ling, X., Merisca, R., Brown, C. B., & Ialongo, N. (1998). The effects of the level ofaggression in the first grade classroom on the course and malleability of aggressivebehavior into middle school. Development and Psychopathology, 10, 165−185.

Larson, R. W., Hansen, D. M., & Moneta, G. (2006). Differing profiles of developmentalexperiences across types of organized youth activities. Developmental Psychology,42, 849−863.

Lewin, K. (1935). A dynamic theory of personality. New York, NY: McGraw-Hill.Lewin, K. (1943). Psychology and the process of group living. The Journal of Social

Psychology, 17, 113−131.Lewin,K., Lippitt, R.,&White, R. K. (1939). Patterns of aggressivebehavior inexperimentally

created “social climates”. The Journal of Social Psychology, 10, 271−299.McFarland, D. A. (2001). Student resistance: How the formal and informal organization

of classrooms facilitate everyday forms of student defiance. The American Journal ofSociology, 107, 612−678.

Moreno, J. L. (1934). Who shall survive? A new approach to the problem of humaninterrelations. Washington, DC: Nervous and Mental Disease Publishing.

Parsons, T. (1959). The school class as a social system: Some of its functions in Americansociety. Harvard Educational Review, 29, 297−318.

Pearl, R., Leung, M. -C., Van Acker, R., Farmer, T. W., & Rodkin, P. C. (2007). Fourth- andfifth-grade teachers' awareness of their classrooms' social networks. ElementarySchool Journal, 108, 25−39.

Pianta, R. C. (1992). The student–teacher relationship scale. Charlottesville, VA:University of Virginia.

Pianta, R. C., Belsky, J., Vandergrift, R. H., Houts, R., & Morrison, F. J. (2008). Classroomeffects on children's achievement trajectories in elementary school. AmericanEducation Research Journal, 45(2), 365−397.

Pianta, R. C., Hamre, B. K., & Stuhlman, M. (2003). Relationships between teachers andchildren. In W. Reynolds & G. Miller (Eds.), Comprehensive Handbook of Psychology(vol. 7) Educational Psychology (pp. 199–234). New York: John Wiley & Sons.

Pianta, R. C., La Paro, K. M., & Hamre, B. K. (2006). The Classroom Assessment ScoringSystem (CLASS). Brookes Publishing.

Pianta, R. C., La Paro, K. M., & Hamre, B. K. (2007). The Classroom Assessment ScoringSystem (CLASS). Baltimore, MD: Brookes Publishing.

Raudenbush, S. W., Martinez, A., & Spybrook, J. (2007). Strategies for improvingprecision in group-randomized experiments. Educational Evaluation and PolicyAnalysis, 29(1), 5−29.

Rodkin, P. C., & Gest, S. D. (2011). Teaching practices, classroom peer ecologies, andbullying behaviors among schoolchildren. In D. L. Espelage S. Swearer (Eds.),Bullying in North American Schools (2nd ed., pp. 75-90) New York, NY: Routledge.

Rodkin, P. C., & Hodges, E. V. E. (2003). Bullies and victims in the peer ecology: Fourquestions for psychologists and school professionals. School Psychology Review, 32,384−400.

Rosenholtz, S. J., & Simpson, C. (1984). Classroom organization and studentstratification. Elementary School Journal, 85, 21−37.

Roseth, C. J., Johnson, D. W., & Johnson, R. T. (2008). Promoting early adolescents'achievement and peer relationships: The effects of cooperative, competitive, andindividualistic goal structures. Psychological Bulletin, 134, 223−246.

Rubin, K. H., Bukowski, W., & Parker, J. G. (2006). Peers interactions, relationships, andinteractions. In W. Damon & R. Lerner (Series Eds.) & N. Eisenberg (Vol. Ed.),Handbook of child psychology: Social, emotional, and personality development (6thed., vol. 3, pp. 571–645). New York, NY: Wiley.

Schäfer, M., Korn, S., Brodbeck, F., Wolke, D., & Schulz, H. (2005). Bullying roles inchanging contexts: The stability of victim and bully roles from primary tosecondary school. International Journal of Behavioral Development, 29(4), 323−335.

Seidman, E. (1988). Back to the future, community psychology: Unfolding a theory ofsocial intervention. American Journal of Community Psychology, 16(1), 3−24.

Sherif, M. (1956). Experiments in group conflict. Scientific American, 195, 54−58.Thomas, D. E., Bierman, K. L. & the Conduct Problems Prevention Research Group.

(2006). The impact of classroom aggression on the development of aggressivebehavior problems in children. Development and Psychopathology, 18, 471−487.

Tseng, V., & Seidman, E. (2007). A systems framework for understanding social settings.American Journal of Community Psychology, 39, 217−228.

Wasserman, S., & Faust, K. (1994). Social network analysis: Methods and applications.Structural analysis in the social sciences. New York, NY: Cambridge UniversityPress.

Wentzel, K. (2009). Peers and academic functioning at school. In K. H. Rubin, W. M.Bukowski, & B. Laursen (Eds.), Handbook of Peer Interactions, Relationships andGroups (pp. 531−547). New York, NY: Guilford.

Witvliet, M., Olthof, T., Hoeksma, J. B., Goossens, F. A., Smits, M. S. I., & Koot, H. M.(2010). Peer group affiliation of children: The role of perceived popularity,likeability, and behavioral similarity in bullying. Social Development, 19,285−303.

Wright, J. C., Giammarino, M., & Parad, H. W. (1986). Social status in small groups:Individual-group similarity and the social “misfit”. Journal of Personality and SocialPsychology, 50, 523−536.