In this paper I focus on the fixed model for exploratory factor analysis,in which common factors are treated as unknown fixed parameters.Let X be an n-individuals × p-variables data matrix,whose columns are centered as In X =0p with In the n × 1 vector of ones and 0p the p × 1 zero vector.The fixed factor model for X can be written as X =F A + U,using F for an n-individuals × m-factors matrix of common factor scores,A for a p × m loading matrix,and U for an n × p unique factor matrix filled with the realizations of 1 × p random error vector u.Here,1n F =0m,n > p > m,rank(F A ) =m,E(u) =0p,and E(uu) =ψ with ψ the diagonal matrix whose diagonal elements take the positive values called uniqueness.The purpose of fixed factor analysis is to estimate parameter matrices F,A,and ψ.Their joint estimation is,however,known to be difficult:the likelihood function of F,A,and ψ following from the normality assumption of u does not have an upper bound (Anderson & Rubin,1956).