Plausibility of multivariate normality assumption when multiply imputing non-Gaussian continuous outcomes: a simulation assessment

157 citations recorded by the source corpus.

How researchers see this paper

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

Should a Normal Imputation Model be Modified to Impute Skewed Variables (2013, background): “Simulations confirm that FN imputation yields consistent estimates for and under a variety of non-normal distributions (Rubin and Schenker 1986; He and Raghunathan 2006; Demirtas and Hedeker 2008), but yields biased estimates of many percentiles (Demirtas and Hedeker 2008; He and Raghunathan 2006; D”
New Confidence Intervals and Bias Comparisons Show That Maximum Likelihood Can Beat Multiple Imputation in Small Samples (2013, background): “…of missing-data estimators, which have most commonly examined sample sizes down to n=25 (Lipsitz, Parzen, & Zhao, 2002; Wagstaff & Harel, 2011), although two studies went as low as n=10 (Barnard & Rubin, 1999; von Hippel, 2013b) and two didn’t go below n=40 (Demirtas et al., 2008; Reiter, 2007).”
Comparison of missing value imputation methods in time series: the case of Turkish meteorological data (2013, background): “Note however that according to Demirtas et al. (2008) and Schafer (1997), the violation of the normality assumption may not cause invalid estimations.”