Table 1 Proportion of protocol deviations under different choices of α0 in (1) by arm under “strong confounding”
From: Interval-cohort designs and bias in the estimation of per-protocol effects: a simulation study
Scenario | Arm | α 0 | Cumulative proportion deviated | |
|---|---|---|---|---|
0 | Z=1 | 4.0 | 41% | |
Z=0 | -6.5 | 41% | ||
1 | Z=1 | 5.0 | 21% | |
Z=0 | -7.5 | 20% | ||
2 | Z=1 | 6.0 | 9% | |
Z=0 | -8.5 | 8% |