|
Abstract:
|
Zero-inflation models are commonly used in count data applications with a disproportionate number of zeros. We develop a Bayesian approach to allow for inflation at multiple values, in a repeated measures setting. Models accounting for repeated measures zero-inflated count data have previously been developed, as well as for cross sectional data with multiple inflation points. We extend these models using a Binomial distribution and random effects with application to a study measuring days of substance use in the last 3 months, where we observed a large number of non-users and daily users, with the rest of the population falling somewhere in between.
|
