Font Size:
Three-way Correspondence Analysis for Ordinal-Nominal Variables
Last modified: 2017-05-20
Abstract
This paper presents some variants of three-way polynomial correspondence analysis to analyse associations in three-way contingency tables that are constructed from ordinal and nominal variables. Historically,  three-way correspondence analysis has been used for this purpose without regard to the ordinal structure of the variables. Recently, Lombardo et al.(2017) proposed  an alternate orthogonal basis of Emerson's polynomials for modelling interactions in three-way contingency tables. Here, we propose to consider \textit{hybrid} decompositions for modelling the cases  where not all variables are ordered.