Events and Meetings of Italian Statistical Society, Advances in Latent Variables - Methods, Models and Applications

Font Size: 
Separating between- and within- group associations and effects for categorical variables
Marcel August Croon, Margot Bennink

Last modified: 2013-06-14


This paper discusses a particular application of latent variables models for the two-level analysis of categorical data. These models, which separate the between- and within-group associations between variables either measured at the lower or higher level in the multilevel hierarchy, allow the estimation of the effects of variables measured at the lower level on variables measured at the higher. In the last decades models of this kind were formulated for numeric variables but not for categorical variables. The present paper will show how similar models can be developed for the latter kind of variables, and this approach will be illustrated by an application to data collected in 2010 by the Banca d’Italia in the Survey of Italian Household Budgets on ownership of financial products.

Full Text: PDF