ABSTRACT
One of the greatest challenges in the nonexperimental sciences is the drawing of causal inferences from observational data (Greenland, Pearl, & Robins, 1999; Henán, Hernández-Diaz, Werler, & Mitchell, 2002; Robins, 1997, 2001). is is especially true of developmental research where it is very dicult to identify the unique contribution of any single factor against the backdrop of the multitude of factors with which it is correlated. Developmental research on parental inuences provides a useful case in point. e standard approach to investigating parent eects has been to study intact nuclear families, but this design does not allow for an unequivocal determination of whether the observed eects are mediated by parental contributions to the environments of the children they rear or by the genes they transmit. Scarr and McCartney (1983) were among the rst to comprehensively characterize the nature and implications of the confounding of genetic and environmental eects for developmental research. ey showed in particular that confounding in studies purporting to investigate parent eects, which they labeled passive genotype-environment correlation, renders ndings from most socialization research uninterpretable. An example helps to illustrate the nature of the inferential limitations. ere is a substantial body of research showing that the biological children of antisocial parents have relatively high rates of conduct disorder (Lahey et al., 1988). Nonetheless, the consistent observation of this association does little to help us understand the extent to which the heightened risk of ospring conduct disorder arises because antisocial parents are ineective in socializing their children or because antisocial behavior is heritable so that antisocial parents transmit genes that increase their ospring’s vulnerability for antisocial behavior (Hicks, Krueger, Iacono, McGue, & Patrick, 2004).
