Multiple Imputation of Zero-Inflated and Overdispersed Multilevel Count Data | Dr. Kristian Kleinke

Multiple Imputation of Zero-Inflated and Overdispersed Multilevel Count Data

Abstract

Throughout the last couple of years multiple imputation (MI) has become a popular and widely accepted method to address the missing data problem. However, currently existing multiple imputation software has limitations regarding incomplete count data, especially with regard to certain kinds of multilevel count data: We present a multiple imputation solution for ordinary and overdispersed zero-inflated clustered count data based on a two-level hurdle model using a Bayesian regression approach within a chained equations multiple imputation framework (Raghunathan, Lepkowski, van Hoewyk, & Solenberger, 2001; van Buuren, Brand, Groothuis-Oudshoorn, & Rubin, 2006; van Buuren & Groothuis-Oudshoorn, 2011).

Publication
Technical Report