1
Department of Statistics, Tarbiat Modares University
2
Iranian Statistical Research Institute
3
Department of Statistics, Shahid Beheshti University
Abstract
In the analysis of longitudinal data with count responses, Poisson or negative binomial random effects models with normal distribution assumptions are usually used. In some applications, the distribution of random effects may not be normal. However, misspecification of the distribution of random effects may reduce the efficiency of existing estimators. In this paper, the generalized log-gamma distribution, which includes the normal distribution as a special case, is used as the random effects distribution assumption. Since the frequentist analysis of the model faces complex calculations, Bayesian analysis of this model is introduced and used to analyze several real data sets. The performance of the proposed model has also been evaluated in a simulation study.
Baghfalaki, T., Kamreei, M., Shabak, A. & Ganjali, M. (2021). Bayesian approach to analyzing count data in longitudinal studies using the generalized Poisson-log-gamma model. Andishe_ye Amari, 25(2), 83-95.
MLA
Baghfalaki, T., Kamreei, M., Shabak, A., & Ganjali, M. "Bayesian approach to analyzing count data in longitudinal studies using the generalized Poisson-log-gamma model", Andishe_ye Amari, 25, 2, 2021, 83-95.
HARVARD
Baghfalaki T., Kamreei M., Shabak A., Ganjali M. (2021). 'Bayesian approach to analyzing count data in longitudinal studies using the generalized Poisson-log-gamma model', Andishe_ye Amari, 25(2), pp. 83-95.
CHICAGO
T. Baghfalaki, M. Kamreei, A. Shabak & M. Ganjali, "Bayesian approach to analyzing count data in longitudinal studies using the generalized Poisson-log-gamma model," Andishe_ye Amari, 25 2 (2021): 83-95,
VANCOUVER
Baghfalaki T., Kamreei M., Shabak A., Ganjali M. Bayesian approach to analyzing count data in longitudinal studies using the generalized Poisson-log-gamma model. Andishe_ye Amari. 2021;25(2):83-95 (In Persian).
