Multidimensional Contingency Tables
- 1.2k Downloads
Cross-classifications involving more than two variables are a natural generalization of the tables discussed in Chapters 6 and 7. From the generalized linear model (Poisson regression) point of view, this merely corresponds to incorporating more (nominal or ordinal) predictors into the model, and presents no particular difficulties. The structure in a multidimensional contingency table model, however, and the implied forms of association and independence that loglinear models imply, make it important to study in some detail the analysis of higher-dimensional categorical data. We start with what is in many ways the simplest generalization, the 2 × 2 × K table, since many of the general issues involved arise in this simpler context. We will then generalize results to I × J × K and eventually four- and higher-dimensional tables.
KeywordsRespiratory Syncytial Virus Death Penalty Conditional Independence Recall Rate Loglinear Model
Unable to display preview. Download preview PDF.