Full Contingency Tables, Logits, and Split Contingency Tables

摘要: Three methods of fitting log-linear models to multivariate contingency-table data with one dichotomous variable are discussed. Logit analysis is commonly used when a full contingency table s dimensions regarded as rates dimension -1. The split-table method treats the same two separate tables each We show that can be generalized approach: which fitted by it include both mutually exclusive subsets other methods. Even logit permits model choice fitted, iterative proportional set sufficient configurations has advantage requiring neither matrix inversion nor substitution an arbitrary value in empty elementary cells.