Bayesian Factor Analysis By Gibbs Sampling and Iterated Conditional Modes

A Bayesian factor analysis model was proposed by Press and Shigemasu in which factor scores, factor loadings, and disturbance variances and covariances were estimated in closed form using a large sample approximation for one of the terms in the posterior distribution. This paper shows that by using Gibbs sampling or Iterated Conditional Modes approaches to estimation instead of the large sample approximation, we can obtain improved point estimators in small samples. All three methods of estimation used in an example and compared. Bayesian Factor Analysis By Gibbs Sampling and Iterated Conditional Modes Author(s) names removed. Key Words: Press-Shigemasu, scores, loadings. Abstract A Bayesian factor analysis model was proposed by Press and Shigemasu in which factor scores, factor loadings, and disturbance variances and covariances were estimated in closed form using a large sample approximation for one of the terms in the posterior distribution. This paper shows that by using Gibbs sa..