Estimating Adjusted Associations between Random Effects from Multilevel Models: The Reffadjust Package. (March 2014)

• Record Type: Journal Article
• Authors: Palmer, Tom M.; Macdonald-Wallis, Corrie M.; Lawlor, Debbie A.; Tilling, Kate
• Is Part Of: Stata journal. Volume 14:Number 1(2014)
• Page Start: 119
• Page End: 140
• Publication Date: 2014-03
• Subjects: st0327 -- reffadjust -- reffadjustsim -- reffadjust4nlcom -- meqrlogit -- meqr-poisson -- mixed -- multilevel models -- runmlwin -- xtmelogit -- xtmepoisson -- xtmixed -- ALSPAC; Statistics -- Periodicals; Statistics -- Computer programs -- Periodicals; 001.422
• DOI: 10.1177/1536867X1401400109
• Languages: English
• ISSNs: 1536-867X

Abstract

We describe a method to estimate associations between random effects from multilevel models. We provide two new postestimation commands, reffadjustsim andreffadjust4nlcom, which are distributed as thereffadjust package. These commands produce the estimates and their associated confidence intervals. The commands are used after official Stata multilevel model estimation commandsmixed, meqrlogit, andmeqrpoisson (formerly namedxtmixed, xtmelogit, andxtmepoisson, respectively, before Stata 13) and with models fit in the MLwiN statistical software package via therunmlwin command. We demonstrate our commands with several simulated datasets and for a bivariate outcome model investigating the relationship between weight and mean arterial pressure in pregnant women using data from the Avon Longitudinal Study of Parents and Children. Our method and commands help to improve the interpretability of estimated random-effects variance components from multilevel models.