Use of conditional and marginal odds-ratios for analysing familial aggregation of binary data

We propose a method for measuring familial aggregation of binary outcomes t hat incorporates unconditional, i.e., marginal, odds-ratios, for measuring association between outcomes within generations while using conditional odd s-ratios to measure association between generations. This method may be use d together with marginal odds-ratio models to explore the strength of assoc iation between individuals within a generation while accounting for the out comes of predecessors. Interpretation of resulting odds-ratios is straightf orward and a number of features of the method are discussed. An example, us ing smoking status data from Busselton Health Study families, is used to il lustrate our approach. (C) 2000 Wiley-Liss, Inc.