Gender of Siblings, Cognitive Achievement, and Academic Performance: Familial and Nonfamilial Influences on Children

Abstract

In this article, I use data on twins to obtain estimates of the degree of shared familial influence on cognitive achievement and academic performance. The results indicate that there is considerable shared influence on cognitive achievement but much less for academic performance. The model estimated also indicates a considerable degree of sibling symmetry in both outcomes. The only significant difference rests between oppositegender twins. For these twins, nonshared sources of influence are more important than for samegender twins. Finally, the results indicate that the regression of academic performance on cognitive achievement is not biased by unmeasured familial characteristics. Key Words: academic achievement, cognitive achievement, gender, siblings, twins. Researchers have long been interested in the influence of the family on the achievement of children. Unfortunately, with conventional research designs, it is difficult to separate familial from nonfamilial influences on outcomes of interest. In this article, I use data on twins to estimate the relative contribution of shared (familial) and nonshared (nonfamilial) influences on cognitive achievement and academic performance. I estimate a series of models that tests the degree of sibling symmetry in these outcomes and decomposes variance into shared and nonshared components. I pay particular attention to the role played by the gender of individual twins and the gender composition of twin pairs in affecting cognitive achievement and academic performance. The article is organized as follows. First, I outline the methodological approach taken, including a discussion of the statistical model employed. Second, I describe cognitive achievement and academic performance as outcomes deserving research attention, especially because they might vary according to the gender of twins and gender composition of twin pairs. Third, I describe the sample and measures used. Fourth, I present results based on the modelling exercise. I conclude with a discussion of the significance of the results. THE FAMILY AS A NATURAL EXPERIMENT The study of the influence of the family on the achievement of children generally has proceeded by selecting a sample of children and measuring various characteristics of their families and the outcomes of interest (here, cognitive achievement and academic performance). In this standard design, variation between children on covariates tapping family characteristics are correlated with the outcome of interest (e.g., the effect of family income on education obtained, the effect of mother's labor market participation on children's reading ability). This design poorly captures the influence of families, however, because it is unlikely that all covariates associated with familial influence are adequately measured. At best, this type of analysis identifies only a portion of the common familial influence and leaves a residual (an error term) that is composed of both familial and nonfamilial influences. The standard design also has assumed, if even implicitly, that family background has the same effect (i.e., a symmetry effect) on all siblings in a family. This assumption results from using data that contain only one child per family (or from ignoring the fact that some children in a data base may come from the same family). Without comparing children within a family, researchers are forced to assume that a child from a given family is indistinguishable from other children in that family. Thus, potentially important variations in family processes affecting child well-being are ignored. Such variation in family processes, if not taken into account, may influence estimates of the amount of variation that lies within versus between families. What is needed is an approach that will identify the total effect of all shared (genetic and environmental) influences on child well-being, if not the specific components of shared influence. …