1 Introduction

This chapter introduces a technique for specifying and estimating paths between dependent variables, so called non-recursive models. Uses of this technique include: modeling social interactions, for example, sibling competition and cooperation; testing for direction of causation in bivariate data, e.g., whether life events cause depression or vice versa; and developmental models for longitudinal or repeated measurements. Models for sibling interaction have been popular in genetics for some time (Eaves, 1976b), and the reader should see Carey (1986b) for a thorough treatment of the problem in the context of variable family size. Here we provide an introductory outline and application for the restricted case of pairs of twins, and we assume no effects of other siblings in the family. We further confine our treatment to sibling interactions within variables. Although multivariate sibling interactions (such as aggression in one twin causing depression in the cotwin) may in the long run prove to be more important than those within variables, they are beyond the scope of this introductory text. Section 8.2 provides a summary of the basic univariate genetic model without interaction. The extension to include sibling interaction is described in Section 8.3. Details on the consequences of sibling interaction on the variation and covariation are discussed in Section 8.4