The following HTML text is provided to enhance online
readability. Many aspects of typography translate only awkwardly to HTML.
Please use the page image
as the authoritative form to ensure accuracy.
pects associated with collecting genetic information on specific alleles in conjunction with one or more existing national surveys. Until now, there has been a dearth of thinking on how genetic variables are best introduced into sophisticated mortality models. The availability of even a few genetic markers in large health and demographic data sets would be likely, in my view, to generate rapid improvements in the repertory of models. The central technical problem in recent survey-based research on the impact of health services, interventions, and behavioral changes has been the problem of self-selection bias. Approaches that come under the broad heading of "instrumental variable techniques" are enjoying a vogue in econometrics but remain substantively questionable, because convincing instrumental variables, actually satisfying the statistical assumptions of the models, are hardly to be found. Genetic indicators offer the best hope for obtaining usable instrumental variables and so for beginning to sort out the tangle of selection bias and causation.
The most urgent need for genetic indicators in large national surveys is for confirmatory studies of causal effects of genes on medical conditions. The epidemiological samples on which searches for genes with causal influences are conducted are almost inevitably subject to grave potential selection biases and a large variety of potential confounding factors. Control groups are difficult to arrange. Population-based surveys rich in socioeconomic and demographic background information are going to be essential to confirm the effects on longevity, disability, dementia, and other health-status outcomes as candidate genes are identified. This research will not be easy, because many hard-to-measure factors like dietary patterns are likely to interact with and overshadow the genetic variables. The essential contribution is not to help in the process of screening for candidate genes but rather to detect spurious candidates and sort out the gene-environment-behavior interactions that must underlie many of the observable effects. The inclusion of biomarkers in nationally representative surveys with control variables for potential confounding factors is an essential step in confirming the true role in longevity and health of isolable genetic influences.
The title Biodemography is meant to evoke the idea of a marriage of the disciplines of biology and demography. A mixer of metaphors might reasonably ask whether this marriage is to be grounded in mutual affection and commitment or is to be more in the nature of a practically advantageous but cautious arrangement for cohabitation. For neither partner is this a first union. The theories and findings from each discipline discussed in this book are bound to be somewhat in the nature of stepchildren to the other partner. But stepchildren are increasingly central to families in the world around us. The ideas about longevity put forward from the perspectives of biology and demography in the following pages deserve committed nurturance and promise lasting rewards.