ABSTRACT
This chapter examines the analysis of binomially distributed data where the binary responses are independent, with special attention given to situations where the binary responses are obtained from clusters. It explores the logistic transformation and outlines the relationship between the regression coefficient and the odds ratio parameter. The chapter describes approaches to modeling correlated binary data and investigates population average and cluster specific models. Correlated binary data from longitudinal studies where repeated measures of a binary outcome are gathered from independent samples of individuals need special modeling strategies, simply because the nature of the within-cluster correlation is different. Measures of goodness of fit are statistical tools used to explore the extent to which the fitted responses obtained from the postulated model compare with the observed data. Clearly, the fit is good if there is a good agreement between the fitted and the observed data.
