Robustness issues in multilevel regression analysis

By Cora J. M. Maas, Joop J. Hox. 2004

879 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/7b2a0f91b553217eebd642b1a523a05d087521e7

How researchers see this paper

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

In Statistics and Probability

Citations in the corpus

Multilevel Modelling of Country Effects: A Cautionary Tale (2016, background): “Recent studies include Bell et al. (2014), Maas and Hox (2004, 2005), and Stegmueller (2013).”
The Impact of Small Cluster Size on Multilevel Models: A Monte Carlo Examination of Two-Level Models with Binary and Continuous Predictors (2010, background): “Studies consistently showed little to no bias in the estimates of the fixed effects (Clarke & Wheaton, 2007; Maas & Hox, 2004, 2005; Mok, 1995; Newsom & Nishishiba, 2002).”
Regression Analysis of Country Effects Using Multilevel Data: A Cautionary Tale (2013, background): “The evidence to date for linear models indicates that OLS, GLS and FML estimates of the parameters associated with fixed predictors (β and γ) are unbiased even if the number of groups is as small as 10 (Hox 2010; Maas and Hox 2004).”
A solution to dependency: using multilevel analysis to accommodate nested data (2014, methodology): “…at least 53% of the 314 examined articles clearly concerned nested data, of which 44% specifically reported the number of observations per cluster with a minimum of five observations per cluster (that is, for robust multilevel analysis a minimum of five observations per cluster is required 11,12 ).”
The Accuracy of Significance Tests for Slope Variance Components in Multilevel Random Coefficient Models (2009, methodology): “The residual correlation between the intercept and slope variances was set to 0 (Maas & Hox, 2004) to aid in model convergence.2 Following Hox (2002) and Raudenbush and Liu (2000), we manipulated the variance of the Level 1 slope by varying the degree to which it was related to the Level 2…”
Cluster Size in Multilevel Models: The Impact of Sparse Data Structures on Point and Interval Estimates in Two-Level Models (2008, background): “This rule of thumb is commonly cited (see, for example, Hox, 1998; Maas & Hox, 2002; Maas & Hox, 2004) and was further developed by Hox (1998) who recommended a minimum of 20 observations (level-1) for 50 groups (level-2) when examining interactions across levels.”
The impact of venture capital financing method on SME performance and internationalization (2011, methodology): “The Huber-White method is appropriate for small samples (>30) such as ours ( Maas and Hox 2004 ).”
An Application of the What Works Clearinghouse Standards for Evaluating Single-Subject Research (2013, methodology): “A sandwich-type correction was included in these models to assist with making estimates more robust to violations of normality (Maas & Hox, 2004; Yuan & Bentler, 2002).”
Robustness of linear mixed‐effects models to violations of distributional assumptions (2020, background): “Unfortunately, very few studies have assessed random effect variances and predictions (e.g. Maas & Hox, 2004; McCulloch & Neuhaus, 2011; Verbeke & Lesaffre, 1996); for those who did, the estimation of the group variance is usually accurate, but random effect coefficients (best linear unbiased…”
Are Individual-Level and Country-Level Value Structures Different? Testing Hofstede’s Legacy With the Schwartz Value Survey (2010, background): “Analogously, sampling at the country level could have affected estimates of isomorphism (e.g., Maas & Hox, 2004).”