Comparison of confirmatory factor analysis estimation methods on mixedformat data

Abstract

Weighted least squares (WLS), weighted least squares mean-andvariance-adjusted (WLSMV), unweighted least squares mean-and-varianceadjusted (ULSMV), maximum likelihood (ML), robust maximum likelihood(MLR) and Bayesian estimation methods were compared in mixed item responsetype data via Monte Carlo simulation. The percentage of polytomous items,distribution of polytomous items, categories of polytomous items, average factorloading, sample size and test length conditions were manipulated. ULSMV andWLSMV were found to be the more accurate methods under all simulationconditions. All methods except WLS had acceptable relative bias and relativestandard error bias. No method gives accurate results with small sample sizes andlow factor loading, however, the ULSMV method can be recommended toresearchers because it gives more appropriate results in all conditions.