Joint (k, l)-hurdle random effects models for mixed longitudinal power series and normal outcomes. Issue 12 (18th June 2019)

• Record Type: Journal Article
• Authors: Razie, F.; Bahrami Samani, E.
• Is Part Of: Communications in statistics. Volume 48:Issue 12(2019)
• Page Start: 3025
• Page End: 3043
• Publication Date: 2019-06-18
• Subjects: Mixed correlated outcomes -- longitudinal study -- (k -- l)-hurdle power series distribution -- random effects -- non-ignorable missing outcomes; Mathematical statistics -- Periodicals; Mathematics; Statistics; 519.2
• DOI: 10.1080/03610926.2018.1473601
• Languages: English
• ISSNs: 0361-0926

Abstract

Abstract: A random effects model for analyzing mixed longitudinal normal and count outcomes with and without the possibility of non ignorable missing outcomes is presented. The count response is inflated in two points ( k and l ) and the ( k, l )-Hurdle power series is used as its distribution. The new distribution contains, as special submodels, several important distributions which are discussed, such as ( k, l )-Hurdle Poisson and ( k, l )-Hurdle negative binomial and ( k, l )-Hurdle binomial distributions among others. Random effects are used to take into account the correlation between longitudinal outcomes and inflation parameters. A full likelihood-based approach is used to yield maximum likelihood estimates of the model parameters. A simulation study is performed in which for count outcome ( k, l )-Hurdle Poisson, ( k, l )-Hurdle negative binomial and ( k, l )-Hurdle binomial distributions are considered. To illustrate the application of such modelling the longitudinal data of body mass index and the number of joint damage are analyzed.