to causal mechanisms.” Rutter sees the search for causal mechanisms as primary and the assessment of genetic influence as secondary, while acknowledging the value of behavioral genetic studies for “a better understanding of the nature-nurture interplay in the development of socially relevant behaviors.”
Rutter’s chapter can be read as a call for behavioral genetics in a new key. It is possible to go beyond statistical path diagrams and measures of association. Broader opportunities are now opening up for including what Rutter calls “discriminating and sensitive measures of the environment” in behavioral genetic studies. Techniques for direct monitoring of environmental exposures and stimuli for individuals are in rapid development. They dovetail with approaches for making use of biological indicators of hormonal and other physiological responses, as described in the National Research Council’s 2000 volume Cells and Surveys, already mentioned. Such measures are particularly relevant for studies of fertility behavior, since hormonal signaling must play a role in so many pathways of influence.
The National Longitudinal Study of Adolescent Health (“Add Health”), whose first wave of home interviews were conducted in 1995, is providing researchers with data on familial correlations for a wide range of traits plausibly bearing on fertility. Jacobson and Rowe’s (1999) study of depressed moods in adolescents is an example. Where the outcomes of genetic predispositions are likely being modulated by social trends and circumstances, self-selection of individuals into particular niches has to be distinguished from causal influence. Information on school settings and experiences, peer-group composition, and parental treatment, such as Add Health collects, can give a basis for separating out genetic effects from the consequences of self-selection, social homogamy, and friends and family networks. For the topics central to this volume, later waves of the Add Health study should provide more extensive observations of union formation and childbearing as respondents grow toward adulthood.
Conclusions with regard to gene-environment interactions can be particularly convincing when genotypes are measured directly with molecular markers. Rowe at al. (2001) report one example involving numbers of marriages for women. Successes with such approaches are not yet frequent, but technology is improving. Supplementing twin and adoption studies, there are natural experiments taking place willy-nilly all around us which, with ingenuity, can be recognized and analyzed to advantage. The study of change over time stands to benefit from hard thinking about gene-environment correlations and feedbacks. Given such a range of opportunities, it is fair to expect that, for studies of fertility and family formation, the behavioral genetics of the future should bear little resemblance to the behavioral genetics of the past.