Do peers influence children's skill development in preschool?

Abstract

Currently, a disjuncture exists between the economic literature on young children's outcomes from early education or child care and the literature on schooling outcomes of older children and adolescents. Peer effects have been found to be both theoretically important and empirically significant in school settings, yet the effects of peers have not been incorporated into research estimating the child development production function. In this study, we estimate the value-added effects of peer abilities on the educational outcomes of a probability sample of four year olds who attended Head Start, publicly subsidized pre-kindergarten, or private preschool in Georgia. We use a longitudinal data set that includes measures of preschool quality, child and family characteristics, peer abilities, and assessments of the skills of pre-kindergarteners both before and after attending preschool. The ability level of the peers in a child's classroom has direct and positive effects on the child's cognitive skills, pre-reading skills, and expressive language skills after controlling for preschool resources, family characteristics, and the child's skills at the beginning of preschool. Neither time spent on discipline, nor contextual effects of classroom composition, nor teachers’ motivation appear to be the mechanisms that explain the influences of peers on children's skill development.

Introduction

An obvious disjuncture exists in the economic research literature on the development of young children and the educational outcomes of older, school-age children and adolescents. The outcomes of schooling are held to be a function of the ability of the student, the family and home influences on the child, the school inputs, and peers in the classroom (Hanushek, 1979; Hanushek, Kain, Markman, & Rivkin, 2003; Wolfe & Summers, 1977; Zimmer & Toma, 2000). Moreover, some theoretical work on the education market argues that the effects of schooling are principally derived from peer effects and efficient sorting of children (Epple & Romano, 1998). In contrast, the early education and child care literature has largely omitted the effects of peers on the developmental outcomes of the young children in these environments. For example, in a recent study of the effect sizes attributable to the quality of child care environments, the Duncan Early Childcare Research Network and Duncan (2003) posit four factors that influence children's development status: quantity and quality of care, resources available in the child's home, child characteristics, and family attributes.

The omission of peer influences from estimates of the child development production function appears to be particularly consequential in that it inhibits a complete understanding of how the developmental skills of the children are produced and may lead to omitted variable bias which may exaggerate the estimates of other coefficients in the model. It is entirely plausible that the effects of peers may be particularly large in early education and care environments where more interaction between peers occurs throughout the day and instructional methods deemphasize whole group activities and teacher-directed instruction. For example, extant research suggests that child development is enhanced by teachers who believe in and use child-initiated instruction within their classrooms (Marcon, 1999). Child-initiated instruction allows the children to choose their own activities and the other children with whom they will participate for large parts of the day. Frequent interactions between children can stimulate development, not only for language and communications skills but also for social and problem-solving skills. In addition, preschool quality is judged to be higher when children can access classroom instructional resources for large periods of the day, under the theory that children will visit centers within the classroom and develop skills through their interactions with the other children and center resources (Harms, Clifford, & Cryer, 1998). However, we currently have no direct tests of the extent to which the quality of the environment or the influence of peers within that environment yield greater skill development.

In early educational environments where teacher-directed activities are often kept to a minimum, the skill development of children may directly relate to the skills of their peers. Higher skilled peers who have a larger vocabulary, ability to express themselves, greater familiarity with print materials, and well-developed social skills could stimulate skill development among the other children within their preschool environment, primarily within their classrooms. But peer effects can arise from sources other than direct transmission from one's peers (Betts & Shkolnik, 1999). For example, peers with more highly developed behavioral skills may cause less disruption of the class and allow teachers to focus their attention on stimulating the children's development. In addition, peers with higher levels of skills could allow teachers to increase the pace and level of content that they introduce during their interactions with the children. Teachers who introduce content just at the point when the children are capable of assimilating it or encourage the children to practice skills they have just acquired may enhance the cognitive and language skills of the children within their classrooms. Further, teachers may become more motivated when they perceive that the children in their class have higher abilities and may change their instruction in ways that produce benefits. These hypotheses that set forth plausible explanations of ways that peer abilities may affect children's development motivate the inclusion of peer effects in evaluations of the child development production function. Moreover, including peer effects may serve as a check on the potential for omitted variable bias in the estimates of the effects of preschool inputs, such as the quality of preschool or center-based child care. When the abilities of peers are omitted from the models, it is possible that the effects of quality are biased upwards and the effects magnified due to misspecification.

Moreover, peer influences bear implications for early education policies. If peer effects are shown to influence children's development at an early age, policies that raise the average ability level of the child's peers may produce incremental improvements to children's development. Research on peer effects for school-age children has shown that disadvantaged children benefit most from peers with higher ability levels (Summers & Wolfe, 1977; Zimmer & Toma, 2000). Currently, Head Start, the largest early education program with appropriations of approximately $6.7 billion nationally (Butler & Gish, 2003), targets children from economically disadvantaged families whose development is significantly behind that of other children (US Department of Health and Human Services, 2002; Henry et al., 2004). For example, Henry et al. report that four year olds entering Head Start were over 10 points (0.68 standard deviation units) behind other children entering private preschool in terms of cognitive skills and 16 points (1.1 standard deviation units) behind in terms of pre-reading skills (2004). If the ability level of peers have positive effects on the development of children with initial skill deficits, it may be advantageous to develop policies with greater mixing of children with different abilities than can be achieved in early education programs targeted exclusively toward disadvantaged children. However, if highly skilled peers do not affect the skill development of children, the mixing of children with different abilities may assume less importance in policy decisions.

In this study, for what is (to our knowledge) the first time, the effects of peer abilities on the development of children in preschool are estimated. We use data gathered on a probability sample of children attending preschool in Georgia, including children in private preschool, publicly subsidized pre-kindergarten, and Head Start. We develop estimates of the abilities of the children in each child's preschool classroom, which include measures of cognitive skills, pre-reading skills, and expressive language skills. We test the effects of peer abilities on children's skills at the beginning of preschool and after the end of preschool using value-added models that control for family characteristics and preschool inputs. In addition, we test some alternative specifications to assess the robustness of the estimates and the mechanisms through which peer effects register their effects. In the next section, we present the theory underlying the production of skills in early education settings—the production function. In Section 3, we describe the sample, measures, and analysis techniques. In Section 4, the findings are presented. Finally, we draw some conclusions.