The purpose of this project is to conduct research on statistical methodology problems which arise in relationship to the Institute's activities in the field of epidemiology. Epidemiologic data, because of its observational nature, is subject to biases and other analytic difficulties not seen in experimental data. In addition, such data can be quite complex, with many complicating factors to account for. The objective of this project is both to broaden understanding of the uses and limitations of currently employed study designs and corresponding analyses, and to develop and evaluate new techniques for statistical analyses of epidemiological studies. Current research on this project involves (1) the development of methods for the assessment of fertility through the use of time-to-pregnancy data (i.e., number of menstrual cycles required to achieve pregnancy); (2) an assessment of some of the techniques currently in use for the analysis of spontaneous abortions to determine whether these techniques adequately handle the various sorts of biases which are involved in this type of data; (3) the development of techniques for assessing the relationship between two processes (in this case, heart rate and respiration) and for using this relationship (or disturbances thereof) to study conditions such as sudden infant death syndrome.
