Biostatistics (2019) 20, 2, pp. 332–346
doi:10.1093/biostatistics/kxy005
In the original article, there were errors in simulation calculations that have since been corrected. Table 2 has now been amended.
Table 2.
The proportion of simulations that returned significant results for each intervention effect of interest (
)
Relative risk ( ) |
Test-positive fraction | Odds ratio | GEE | Random effects |
|---|---|---|---|---|
| 1 | 0.0497 | 0.0749 | 0.0779 | 0.0743 |
| 0.6 | 0.4916 | 0.5795 | 0.5936 | 0.6143 |
| 0.5 | 0.7498 | 0.8238 | 0.8266 | 0.8445 |
| 0.4 | 0.9298 | 0.9620 | 0.9603 | 0.9670 |
| 0.3 | 0.9951 | 0.9985 | 0.9983 | 0.9988 |
The GEE assumed an exchangeable correlation matrix. Each approach was applied to the results of the 10 000 random intervention allocations with 1000 cases and 4000 controls (
).
The corrected table appears below.