On the Performance of Maximum Likelihood Versus Means and Variance Adjusted Weighted Least Squares Estimation in CFA

By André Beauducel, P. Herzberg. 2006

1,287 citations recorded by the source corpus.

Subfields: Computer Networks and Communications, Management Science and Operations Research, Statistics and Probability

View Paper: https://www.semanticscholar.org/paper/d15a1e85e80b0d47c128f545a00a3ff801005559

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Networks and Communications

In Management Science and Operations Research

In Statistics and Probability

Citations in the corpus

Directionality of the Associations of High School Expectancy-Value, Aspirations, and Attainment (2015, background): “…robust maximum likelihood (MLR) estimator, which is robust to the nesting of the students within schools and to the Likert nature of items including four or more answer categories (Mplus’s complex design option; Muthén & Muthén, 2012; e.g., Beauducel & Herzberg, 2006; Distefano & Motl, 2006).”
Methodological Measurement Fruitfulness of Exploratory Structural Equation Modeling (ESEM): New Approaches to Key Substantive Issues in Motivation and Engagement (2011, background): “…tests of fit that are robust in relation to nonnormality of observations and the use of categorical variables when there are at least four or more response categories (e.g., Beauducel & Herzberg, 2006; DiStefano, 2002; Dolan, 1994; Muthén & Kaplan, 1985; Rhemtulla, Brosseau-Liard, & Savalei, 2010).”
Examining Chi-Square Test Statistics Under Conditions of Large Model Size and Ordinal Data (2018, methodology): “ULSMV outperformed WLSMV, yielding better Type I error levels and less bias observed in standard errors, particularly if sample sizes were smaller, as ULS methods are not as sensitive as WLS-based methods to nonconvergence (Forero et al., 2009; Rhemtulla, Brosseau-Liard, & Savalei, 2012; Savalei & R”
The Effects of General and Specific Human Capital on Long–Term Growth and Failure of Newly Founded Businesses (2013, neutral): “This estimator is the most effective estimator in testing models that contain a continuous mediator (growth at t2) and a categorical dependent variable (failure) with robust standard errors (Beauducel & Herzberg, 2006).”
Relative Performance of Categorical Diagonally Weighted Least Squares and Robust Maximum Likelihood Estimation (2014, methodology): “Finally, Beauducel and Herzberg (2006) reported that standard errors obtained from adjusted cat-DWLS were uniformly lower than those from ML. However, empirical standard errors (i.e., standard deviations of parameter estimates across replications) were not reported in this study, so it is not clear ”
Current Methodological Considerations in Exploratory and Confirmatory Factor Analysis (2011, methodology): “…for ordinal data are robust continuous ML estimation, robust least squares (LS) estimation, and robust weighted least squares (WLS) estimation (Beauducel & Herzberg, 2006; Bentler & Yuan, 1999; Flora & Curran, 2004; Forero, Maydeu-Olivares, & Gallardo-Pujol, 2009; Holgado -Tello, Chacón…”
Confirmatory factor analysis with ordinal data: Comparing robust maximum likelihood and diagonally weighted least squares (2016, methodology): “…been found to produce much smaller standard errors of factor loadings than does mean-adjusted ML in LISREL, statistically equivalent to MLM in Mplus, (Yang-Wallentin et al., 2010) and WLSMV (Beauducel & Herzberg, 2006), indicating that uncorrected ML standard errors are generally underestimated.”
Testing Measurement Invariance and Comparing Latent Factor Means Within a Confirmatory Factor Analysis Framework (2011, background): “…previous methodological research has shed light on this issue by revealing that WLSMV may be more theoretically appropriate (Lubke & Muthén, 2004; Millsap & Yun-Tein, 2004) and may provide superior model estimation under certain conditions (Beauducel & Herzberg, 2006; Flora & Curran, 2004).”
Application of Exploratory Structural Equation Modeling to Evaluate the Academic Motivation Scale (2015, neutral): “The choice to rely on WLSMV estimation is linked to the fact that this estimator is more suited to the ordered-categorical nature of Likert scales than traditional maximum likelihood estimation (Beauducel & Herzberg, 2006; Finney, & DiStefano, 2006; Lubke & Muthén, 2004).”
Comparison of Different Estimation Methods for Categorical and Ordinal Data in Confirmatory Factor Analysis (2015, methodology): “Because categorical methodology can outperform continuous methodology with more than five categories (Beauducel & Herzberg, 2006).”
Conducting Measurement Invariance Tests with Ordinal Data: A Guide for Social Work Researchers (2015, methodology): “We specified WLSMV as the estimator (Beauducel & Herzberg, 2006; Flora & Curran, 2004), although Mplus will use WLSMV by default when categorical variables are specified.”