The Accuracy of Multilevel Structural Equation Modeling With Pseudobalanced Groups and Small Samples

By Joop J. Hox, Cora J. M. Maas. 2001

409 citations recorded by the source corpus.

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

View Paper: https://www.semanticscholar.org/paper/217907b3510318437593b2e31fcc0208db33a993

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

Citations in the corpus

Transformational leadership and leader moral orientation: Contrasting an ethic of justice and an ethic of care (2010, methodology): “…Maximum Likelihood (Muthén, 1989, 1994) estimation algorithm provides a good maximum likelihood approximation, simulation studies show it to be effective only for group sample sizes of at least 100 (Hox & Mass, 2001); smaller group sample sizes have been associated with biased standard errors.”
Sample Size Limits for Estimating Upper Level Mediation Models Using Multilevel SEM (2013, methodology): “…number of clusters, G, to use when estimating MMs. High proportions of nonconvergent cases, and biased parameter and standard error estimates have been associated with small numbers of clusters in previous simulation studies (Hox & Maas, 2001; Julian, 2001; Stapleton, 2002; Ryu & West, 2009).”
The Incorporation of Sample Weights Into Multilevel Structural Equation Models (2002, background): “It also has been suggested that this fitting function can be used in the general unbalanced case (Hox & Maas, 2001; Muthén, 1994).”
Multilevel Mediation With Small Samples: A Cautionary Note on the Multilevel Structural Equation Modeling Framework (2017, background): “This type of question is best assessed through ML-SEM, although sample size requirements increase as a function of model complexity.”
A general multilevel SEM framework for assessing multilevel mediation. (2010, neutral): “Regarding the number of units to collect, Hox and Maas (2001) suggested that at least 100 clusters are required for good performance of Muthe´n’s maximum likelihood (MUML) estimator used with Muthe´n’s two-matrix method, echoing Muthe´n’s (1989) recommendation that at least 50 to 100 clusters…”
Actual environments do affect motivation and psychological adjustment: A test of self-determination theory in a natural setting (2008, background): “Therefore, not all parameter estimates were significant at p \ .05 (they were at pone-tailed \ .10) (see Hox and Maas 2001 for an estimation of the bias in parameter estimates with a small group sample size in a multilevel structural equation analysis).”
Dimensions of Socio-economic Status and their Relationship to Mathematics and Science Achievement at Individual and Collective Levels (2003, background): “Differences in sample size, intraclass correlation, and correlations between the SES indicators in different countries may influence the model fit statistics (Hox & Maas, 2001).”
The effect of estimation method and sample size in multilevel structural equation modeling (2010, background): “The other results of Hox and Maas (2001) with respect to the pseudobalanced method replicate well in this simulation, and correspond to Yuan and Hayashi’s (2005) conclusions.”
Transformational Leadership and Leaders’ Mode of Care Reasoning (2011, neutral): “Alternate approximations using the Muthén (1989, 1994) Maximum Likelihood estimation algorithm are also problematic when groups are smaller than 100; these are associated with biased standard errors (Hox and Mass 2001).”
Sample sizes for multilevel modeling (2002, positive): “A recent simulation study on multilevel structural equation modeling ( Hox & Maas, 2001 )”
Alternative Methods for Assessing Mediation in Multilevel Data: The Advantages of Multilevel SEM (2011, background): “Assuming that researchers adhere to common sample size recommendations for MLM ( Hox & Maas, 2001 ; Maas & Hox, 2005 ; Snijders & Bosker, 1999 ) when using MSEM (discussed later), these problems are minimized.”