Abstract:
|
Multiple imputation is a popular method for addressing missing data, but its implementation is difficult when data have a multilevel structure and one or more variables are systematically missing. This might occur in meta-analysis of individual participant data, where some variables are never observed in some studies. In these cases, valid imputation must account for both relationships between variables and correlation within studies. Proposed methods for multilevel imputation include specifying a full joint model and multiple imputation with chained equations (MICE). While MICE is attractive for its ease of implementation, there is little existing work describing conditions under which this is a valid alternative to specifying the full joint model. We use simulation and data from studies of traumatic brain injury to compare MICE and a fully joint model for imputing systematically missing values in an individual patient meta-analysis. We show that MICE imputations tend to produce biased regression coefficients when variables are highly correlated, particularly when between-group variance is high. The full joint model provides estimates that are less efficient but unbiased.
|