Methods for Handling Missing Item Values in Regression Models Using the National Survey on Drug Use and Health (NSDUH)

The purpose of this report is to guide analysts interested in fitting regression models using data from the National Survey on Drug Use and Health (NSDUH) by providing them with methods for handling missing item values in regression analyses (MIVRA). The report includes a theoretical review of existing MIVRA methods, a simulation study that evaluates several of the more promising methods using existing NSDUH datasets, and a final chapter where the results of both the theoretical review and the simulation study are synthesized into guidance for analysts via decision trees.