Abstract
We propose a parametric robust procedure for comparing several means of dependent populations of counts. The validity of the proposed method requires no knowledge of the true underlying joint distributions so long as they have finite second moments. The efficacy of this novel technique is demonstrated by simulations and the analysis of a data set from the Taiwan National Health Insurance Research Database. Our new parametric robust method is also compared with the popular semi-parametric generalized estimating equations approach.
| Original language | English |
|---|---|
| Pages (from-to) | 2576-2585 |
| Number of pages | 10 |
| Journal | Statistics in Medicine |
| Volume | 27 |
| Issue number | 14 |
| DOIs | |
| State | Published - 30 Jun 2008 |
Keywords
- Correlated count data
- Multivariate negative binomial model
- Robust score test