Hazards models for human population biology.

Population biologists often wish to examine the effects of continuous biological variables such as measures of growth, body size,nutritional status, and exposure to environmental risk factors on discrete vital events like birth, onset of disease, and death. Traditional statistical analyses are unable to accommodate some complexities which develop in studying such effects, including the censoring of observations and explanatory variables which change over time. Most traditional methods also provide only empirical rather than etiologic models of pertinent processes. Statistical techniques on hazards analysis deemed especially appropriate for studying biodemographic events and processes are reviewed. A general likelihood framework is also presented which allows the efficient estimation and testing of etiologic hazards models. Past attempts to model biological processes underlying age patterns of fertility and mortality are reviewed, while discussion is had on how the hazards framework may be adapted to study the quantitative genetics of vital events.