Abstract
A simulation study is conducted in a community intervention setting. Several methods of stratified analysis of clustered binary data are compared in terms of empirical significance and empirical power levels. They are the Mantel-Haenszel test statistic (χMH2), the adjusted Mantel-Haenszel test statistic of Donald-Donner (χDD2), Rao-Scott (χRSN2 and χRSP2), and Zhang-Boos (χZBN2 and χZBP2), Wald (χW2), robust Wald (χRW2), score (χS2), robust score (χRS2), and the test statistic based on generalized linear mixed model (GLMM) (χGLMM2. When ρ≠0, χMH2 has inflated type I error, and it should not be used when observations are correlated. The results also warn of the use of χRSN2 and χRW2 due to their poor performance in terms of empirical significance level. χZBP2 and χGLMM2 have better empirical significance levels as compared to other statistics; however, χZBP2 tends to have lower empirical powers than other statistics when the number of clusters (N) is less than 24. χRSP2 provides the highest empirical powers when ρ≥0.1 and N ≤ 12. When ρ≤0.01, we recommend the use of χRS2 and χGLMM2 since they have better overall performance in terms of empirical significance levels and empirical power levels.
Original language | English (US) |
---|---|
Pages (from-to) | 2205-2216 |
Number of pages | 12 |
Journal | Statistics in Medicine |
Volume | 22 |
Issue number | 13 |
DOIs | |
State | Published - Jul 15 2003 |
Keywords
- Community intervention
- Correlated binary data
- Simulation
- Stratified analysis
ASJC Scopus subject areas
- Epidemiology
- Statistics and Probability