Research data in medicine, public health, sociology, education, law or psychology are often shown as a series of 2×2 tables. Sometimes, as in a multicenter clinical trial, this is their natural form. In other cases, the data are stratified at the analysis stage to adjust for the confounding effect of one or more factors. With the spotlight on logit or odds ratio models for a series of 2 × 2 tables, the specific aims of this chapter are:

• To present three statistical models for several 2 × 2 tables arising from a product binomial design. These models are: the common odds ratio model, the heterogenous odds ratios model and the trend in odds ratios model.