Simulation scenarios: n(x) is the number of subjects per treatment and OR(x) the odds ratio of the risk of harm under treatment 1 over treatment 0 for subjects with covariates x. Pr[Y(0,x) =1] = 0.05 Independently of x. Scenario 2 differs only with respect to n((0,0)) = 1000 instead of 100 from scenario 1.

Table 3: Simulation scenarios: $n (x)$ is the number of subjects per treatment and OR(x) the odds ratio of the risk of harm under treatment 1 over treatment 0 for subjects with covariates $\Pr [Y (0, x) = 1] = 0.05$ . Independently of x. Scenario 2 differs only with respect to $n ((0, 0)) = 1000$ instead of 100 from scenario 1.

Scenario	$x_{1}$	$x_{2}$	$n (x)$	$OR (x)$
1	1	1	100	2.0
	1	0	100	1.5
	0	1	100	1.2
	0	0	100	1.0
2	0	0	1000	1.0