RGLS and RLS in Covariance Structure Analysis

This paper assesses the performance of regularized generalized least squares (RGLS) and reweighted (RLS) methodologies in a confirmatory factor analysis model. Normal theory maximum likelihood (ML) (GLS) statistics are based on large sample statistical theory. However, ML GLS goodness-of-fit tests often make incorrect decisions true model, when size is small. The novel methods RGLS RLS aim to correct over-rejection by under-rejection GLS. Both outperform samples small, yet no studies have compared their relative performance. A Monte Carlo simulation study was carried out examine these two methods. We find that equivalent N ≥70; whereas <70, outperforms RGLS. clearly with ≤400. Nonetheless, adopting mean variance adjusted test for non-normal data, slightly RLS. Power analyses found generally showed small loss power performed better than