Abstract

A recently proposed method for factor analysis of discrete data is extended to better handle overdispersion. Three empirical examples from veterinary sciences, musicology and agriculture are investigated, involving true count data as well as ordinal data. Comparisons are made with results from related statistical techniques, e.g., principal component analysis.

Keywords

Poisson distribution; Negative binomial distribution; Model selection

Published in

Communications in Statistics - Simulation and Computation
2021,
Publisher: TAYLOR & FRANCIS INC

Authors' information