Contingency table estimation of genetic parameters and disease risks.

Abstract

Many genetic variables, categorical in nature, allow for analysis by use of contingency table techniques and thereby lead to maximum likelihood estimators. This paper describes these techniques for the classical direct effect model under Mendelian segregation to illustrate how one obtains maximum likelihood estimators for penetrance probabilities and recurrence risks when the data appear in the form of a semi-symmetric intraclass contingency table. The method presented for the 2(3) table is based exclusively on the generalized cross-product ratio and the Pearson chi-squared statistic, and thereby avoids complex statistical methodology and unwieldy computations.