The principal aims of this project are to study a number of statistical problems associated with the analysis of epidemiologic data. Since odds ratios form the nucleus of many epidemiologic analyses, this research addresses issues in modeling building with odds ratios and more generally with relative risk or log-linear modeling. The research contrasts selected statistical estimation procedures. Computer and theoretical techniques are used to develop and compare the methodology. In addition, the methodology is applied to selected cancer case-control data sets, and these data sets are utilized in the comparisons and evaluations of the methods developed. In addition to methods associated with odds ratios, this grant proposes topics in analysis of survivorship data. Epidemiologic data often involve an outcome variable which represents the time to a specified event such as development of cancer. In this study, certain problems associated with the analysis of survivorship data are addressed and will be carefully studied through analytical and applied techniques. Where possible, data from existing case- control and the SEER registry at the University of Iowa will be utilized in the applications and comparisons. This research is a continuation of the previous two years of research in which a number of statistical problems have been studied. These problems are associated with methods for handling sparse multidimensional data. Such data are typical in epidemiological research and improved methods of analysis are needed. This research will lead to improved statistical methodology for the analysis of complex epidemiologic data. The issues studied have immediate implications for the analysis of such data.
Showing the most recent 10 out of 14 publications
