Breadth of Peer Relationships 51
Running head: EXTERNALIZING PRESCHOOLERS’ SOCIAL BREADTH
The Breadth of Peer Relationships among Preschoolers: An Application of the Q-Connectivity Method to Externalizing Behavior
Laura D. Hanish, Carol Lynn Martin, Richard A. Fabes, and Hélène Barcelo
Arizona State University
Keywords: Aggression, Q-connectivity method, social networks, preschool
In press: Child Development
Final paper submitted Oct. 19, 2007
Abstract
Preschoolers’ (M age = 48.7 months) social breadth (the number of peers children interact with and the frequency of these interactions), and the relations between breadth and externalizing behaviors, were examined using the Q-connectivity method. After considering age, sex, and classroom effects on social breadth, children’s externalizing behaviors were studied in relation to: levels of social breadth across variations in frequency of exposure to peers, and to patterns of decline in social breadth. Externalizing preschoolers occasionally interacted with most of their classmates. However, their individual peer networks moderately to rapidly became selective and they engaged in repeated interactions with a relatively small number of peers. Although there were no sex effects, significant age and classroom effects were obtained.
The Breadth of Peer Relationships among Preschoolers: An Application of the Q-Connectivity Method to Externalizing Behavior
One of the most interesting, but least studied, aspects of children’s peer relationships is the substantial variation in the breadth of children’s peer affiliations. Some children interact broadly with many peers whereas others are more selective. To complicate matters, relationship patterns change over time: friendships form, persist for short or long periods of time, and often dissolve, and peers move into or out of social networks. Consequently, there is both stability and fluidity, as well as marked individual differences, in children’s peer affiliations (Cairns, Leung, Buchanan, & Cairns, 1995; Estell, Cairns, Farmer, & Cairns, 2002; Martin, Fabes, Hanish, & Hollenstein, 2005). It is this variability in children’s social breadth (i.e., the number of peers that each child interacts with and the frequency with which each child interacts with every peer) that we highlight in the present study. To do so, we introduce the Q-connectivity method (Barcelo & Laubenbacher, 2005), to calculate preschoolers’ social breadth. The Q-connectivity method is a variation on social network methods that involves computing the frequency with which each child interacts with every other peer and using this information to build each child’s social network at increasing levels of exposure to peers.
We expected to find individual differences in children’s social breadth and we hypothesized that variations in social breadth would be related to children’s externalizing behavior. We focused specifically on children’s externalizing characteristics because, as research on the heterogeneity of aggressive children has highlighted (e.g., Rodkin, Farmer, Pearl, & Van Acker, 2000), the fact that some aggressive children maintain successful peer relationships whereas others have more difficulty speaks to the need to better understand the peer contexts of these children. This issue is directly relevant to the need to better understand peer contagion processes. The conditions that are necessary for peer contagion to occur remain to be documented; for example, it is not well understood whether externalizing behaviors can be spread with just a few contacts with an aggressive child or whether extended interactions are required. Examining the social breadth of externalizing children is important to this area of study because it represents their potential range of influence.
A Developmental-Contextual Framework for Conceptualizing the Relations between Social Breadth and Externalizing Behaviors in Early Childhood
Social breadth refers to the range of peer affiliates—that is, the number of peers that each child interacts with—and the frequency with which each child interacts with every peer. As such, social breadth represents a construct that is distinct from friendship, social status, or membership in social networks because it reflects the entire social landscape of children’s peer affiliations and is not limited to a particular type or quality of social interaction, although it may relate to these factors. The roots of this construct can be seen in the early work of Challman (1932) and Waldrop and Halverson (1975).
We speculate that social breadth is an organizing feature of peer relationships across the developmental life course. We also speculate that the meaning of social breadth will vary according to developmental level. For example, early in life, social breadth may reflect increasing maturity in social development. Evidence for this supposition comes from research that has shown that, across early childhood, children evidence increasingly greater preference for and competence in engaging their peers (Rubin, Bukowski, & Parker, 1998). Moreover, preschoolers’ social networks become larger, denser, and more organized with age (Strayer & Santos, 1996; Vespo, Kerns, & O'Connor, 1996). Additionally, during this time, parents and teachers expect increasing sociability, and children respond to these expectations and pressures by seeking out peer interactions. However, there is still considerable variability and change in these social tendencies during early childhood, making it an ideal time to study the correlates associated with the evolution of social breadth.
The early childhood period is a particularly important developmental period to consider sex differences in social breadth. Research has suggested that, beginning around kindergarten-age, boys maintain larger groups than girls. However, among preschoolers, there are fewer clear sex differences in group size, and both boys and girls tend to interact with peers in small groups (Ruble, Martin, & Berenbaum, 2006). These findings, however, pertain to the size of friendship networks, and may or may not translate to social breadth. Thus, it is necessary to describe the extent to which there are sex differences in social breadth in early childhood.
It is important to note that social breadth is a construct that does not reside entirely within the child. Rather, children’s social breadth reflects their own characteristics as well as those of their peers. Thus, the degree to which children exhibit greater or lesser social breadth may be as much dependent upon the mix of available peers and the match between their own characteristics and behavioral styles and those of their peers as it is an indicator of their underlying propensity to engage with others. This means that children’s tendencies to engage with peers may vary from classroom to classroom, depending upon the classroom peer culture; consequently, we would expect to find classroom-related variations in social breadth as well as individual differences in social breadth that are related in meaningful ways to children’s characteristics.
Looking beyond contextual and demographic characteristics, we presume that children’s externalizing behaviors also contribute to social breadth. Externalizing behaviors are particularly salient in young children’s social lives, and it is not surprising that the peer relationships of aggressive children tend to be sparser and less stable over time. For instance, van den Oord, Rispens, Goudena, and Vermande (2000), who collected sociometric nomination data with over 1000 preschool children, found that children rated by teachers as having more problem behaviors (e.g., aggression, attention problems, and fear) were less likely to be members of cliques and had peer relationships that were less transitive (i.e., children were disliked by their friends’ friends). Similarly, Snyder, Horsch, and Childs (1997) found that aggressive preschoolers, compared to nonaggressive preschoolers, developed fewer relationships with peers that were stable over time. Additionally, studies have demonstrated that aggressive children have lower social preference scores and are more likely to be rejected by peers than their nonaggressive counterparts (Olson, 1992; Snyder et al., 1997). Thus, we expected social breadth to be negatively related to externalizing behaviors, including physical aggression, attention problems, and anger, among young children.
The Use of the Q-Connectivity Method to Measure Social Breadth
To measure social breadth, we used a recently developed variation on social network methods called the Q-connectivity method (Barcelo & Laubenbacher, 2005). This is a mathematical method of modeling complex systems and their dynamics, and it is derived from the Q-connectivity theory for simplicial complexes and graphs (in which the graphs are diagrams with vertices and edges). The method has previously been applied in non-social areas, such as search-and-rescue operations (Barcelo & Laubenbacher, 2005). From a social developmental perspective, the Q-connectivity method makes it possible to measure aspects of peer relationships that have been difficult to study using existing methods because it allows for an assessment of the changing structure of individual children’s interaction patterns with all available peers. This is made possible by Q, which represents exposure; thus, as Q increases, exposure to peers may also increase.
The Q-connectivity method is marked by two unique features that make this a significant innovation for the study of peer relationships. First, the Q-connectivity method maps networks individually for each child in the sample, rather than for a group of children—e.g., classmates—as a whole as is often done with traditional social network procedures (e.g., Cairns et al., 1988; Richards & Rice, 1981). This is a useful distinction to make because, when networks are mapped for a group as a whole, information about networks of those children who are isolated or peripheral is sacrificed in exchange for a parsimonious representation of the social structure of the group as a whole. Thus, by using the Q-connectivity method, each child’s unique patterns of interaction with all available peers are assessed.
Second, the Q-connectivity method capitalizes on the dynamic nature of peer interactions by repeatedly assessing the structure of children’s peer interactions at differing levels of Q, thus providing an estimate of children’s exposure to every peer. In the present study, we relied on repeated observations of children’s peer interaction partners obtained over the course of a semester. This is an important feature of the Q-connectivity method because it provides a means for measuring the fluctuations that occur in children’s peer interaction partners across a period of time. In comparison, traditional social network methods most often rely on data obtained using self-reports or sociometric nominations collected at one discrete time; thus, the resulting social networks represent aggregated views of relationships that provide relatively static pictures of children’s affiliation patterns (for recent exceptions to this see Moody, McFarland, & Bender-deMoll, 2005 and Snijders, 2005).
By applying the Q-connectivity method to the study of social breadth, we are able to calculate multiple measures of social breadth for each child that change as a function of exposure to peers (see the methods section for details about how this is operationalized). On average, children should have a few interactions (low exposure) with many peers, but more frequent interactions (high exposure) with fewer peers. By measuring social breadth at lower and higher levels of exposure, we are able to: (a) examine variations in the structure and meaning of social breadth and (b) assess the drop-off of social breadth as we move across a continuum of how often children were seen playing together.
The Present Study
We described preschoolers’ social breadth along a dimension representing the relative exposure to peers and examined this in relation to three aspects of externalizing behavior—namely, physical aggression, attention problems, and anger. We tested two sets of hypotheses regarding the relations between externalizing behaviors and social breadth. Both sets of hypotheses take into consideration the drop-off of social breadth as the criterion for peer exposure is increased. However, this drop-off was conceptualized differently for both hypotheses. This allowed us to explore the relations between externalizing behaviors and: (a) variations in social breadth scores along a dimension of increased peer exposure (Hypothesis 1) and (b) variations in the patterns of decline (trajectories) of social breadth scores as peer exposure increased (Hypothesis 2).
Hypothesis 1. We examined the relations between externalizing behavior and social breadth at each of several consecutive levels of peer exposure. In general, we expected to find a negative relation between externalizing behaviors (physical aggression, attention problems, and anger) and social breadth, which would indicate that externalizing children interact with a smaller set of peers than do non-externalizing children. This hypothesis is supported by previous studies that have shown negative relations between externalizing behavior and other indicators of the quality of young children’s peer relationships (Snyder et al., 1997; van den Oord et al., 2000). However, we also expected that the relations would be weak and nonsignificant at low exposure because these reflect relatively short-lived peer relationships (i.e., children are only required to interact a small number of times to be counted) but significantly negative at medium to high levels of exposure because these reflect more enduring peer relationships (i.e., children are required to interact more frequently to be counted). Such a finding would support the idea that externalizing children have relationships with peers that are less stable than those of non-externalizing children (Snyder et al., 1997).
Hypothesis 2. We also considered the possibility that there are systematic variations in the patterns of drop-off of social breadth—that children can be distinguished according to their trajectories of social breadth as peer exposure increases. For instance, some children may have a relatively high intercept, followed by a slowly declining slope, whereas others may have a relatively high intercept followed by a steeply declining slope. Still others may have a relatively low intercept with trajectories that similarly vary in steepness. We expected that these variations in patterns could be meaningfully explained by children’s externalizing behaviors. An implicit assumption underlying this idea is that a person-oriented approach can be useful in identifying and categorizing variations in patterns among children (Magnusson, 1998). This type of approach has been successfully used in prior research on children’s peer relationships to describe and explain the heterogeneity in children’s social relationships (e.g., Rodkin et al., 2000).
We hypothesized that children could be clustered according to systematic variations in their trajectories of social breadth and that these clusters would be meaningfully differentiated by both intercepts and slopes. In particular, we expected to find one group evidencing a moderate to high intercept (which would indicate moderate to high social breadth at low peer exposure ) with a steeply declining slope as exposure increases (which would indicate low stability of peer relationships). We further expected that this pattern of social breadth would characterize the children who exhibit relatively high rates of externalizing behaviors. This hypothesis is consistent with prior research that has demonstrated that the quality and duration of externalizing children’s peer relationships tend to be lower than those of more competent children (Snyder et al., 1997).