Statistical analysis of nonlinear factor analysis models

This paper presents a Bayesian analysis of a general nonlinear factor analy sis model. Both the non-informative and conjugate prior distributions are c onsidered. A hybrid algorithm that combines the Metropolis-Hastings algorit hm and the Gibbs sampler is implemented to produce direct estimates of the latent factor scores and the structural parameters. Standard errors estimat es as well as a goodness-of-fit statistic for assessing the posited model a re presented. To illustrate the methodology, results from a simulation stud y and a real example are reported.