Table 3.
Under a larger sample size the comparison of logistic regression generalized estimating equation (GEE) models and linear regression GEE models and the proposed two-step method in detection and estimation of allele effect (n = 1000 clusters) and the difference in allele effect over strata (the interaction effect) (N =1200 clusters) and comparison of convergence rate between logistic and linear regression GEE models with a larger sample size. Data were generated using different assumptions on effect size (β1 = P1 − P0 from model (1) and its corresponding OR and β3 = (P11 − P10) − (P01 − P00)) from model (2) and the corresponding ratio OR1/OR0) and correlation among repeated events (ρ) was set to be 0.2. For details in parameter values, see descriptions in Tables 1 and 2. The simulations were repeated 1000 times.
| Binary logistic regression GEE | Linear regression GEE | Proposed method | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| True parameter |
Prop conv |
Prop sig among conv |
Prop sig among total |
Bias | β3 | Prop conv |
Prop sig among conv |
Prop sig among total |
Bias | Prop sig perm test |
Bias | |
| OR | 1.0 | 76% | 4.7% | 3.6% | 102% | 0.000 | 100% | 6.0% | 6.0% | 0.0% | 3.0% | 0.0% |
| 4.0 | 97% | 46.4% | 45.0% | 4.4% | 0.003 | 99% | 28.4% | 28.4% | 1.0% | 49.6% | 1.0% | |
| OR1/OR0 | 0.75 | 26% | 3.8% | 1.0% | 153% | 0.000 | 99% | 1.8% | 1.8% | 0.0% | 5.8% | 0.0% |
| 5.81 | 42% | 12.7% | 5.4% | −17.4% | 0.009 | 100% | 65.0% | 65.0% | −3.9% | 83.6% | −3.9% | |
| 10.4 | 43% | 33.2% | 14.2% | −6.0% | 0.019 | 100% | 96.2% | 96.2% | 2.2% | 99.0% | 2.2% | |