Abstract
The fit of marginal models to longitudinal data should include modelling all extra variation among responses and covariates. This paper proposes a Partitioned Method of Valid Moments marginal regression model for binary outcomes with Bayes method while using lagged coefficients. Time-dependent covariates are factored in through composite likelihoods. A simulation study highlights the properties of the model coefficients. Modeling cognitive impairment diagnosis in NACC Alzheimer clinical data are demonstrated. Sensitivity analyses are conducted to evaluate the impact of the prior distribution on the posterior inferences.
Original language | English (US) |
---|---|
Pages (from-to) | 2701-2718 |
Number of pages | 18 |
Journal | Computational Statistics |
Volume | 36 |
Issue number | 4 |
DOIs | |
State | Published - Dec 2021 |
Keywords
- Bayes interval estimates
- Binary outcomes
- Logistic regression
- Longitudinal data
- Valid moments
ASJC Scopus subject areas
- Statistics and Probability
- Statistics, Probability and Uncertainty
- Computational Mathematics