Table 2.

Simulated power, in %, of a sensitivity analysis of combined evidence in a case-control study, where there is no unmeasured confounder and $\mathrm{Pr}\left(Z_l=1\right)=1/3$. The response is simulated from $N(\beta ,1)$ if $Z_l=1$ and $N(0,1)$ if $Z_l=0$. There are 1000 narrow cases and 1000 marginal cases with 2000 controls. Based on 10.000 iterations. Fisher = Fisher’s combination method, tP = truncated product method with $\tilde{\alpha }=0.20$

		$\beta =0$		$\beta =0.2$		$\beta =0.4$		$\beta =0.6$
${\text{Γ}}_{nm}$	${\text{Γ}}_{bc}$	Fisher	tP	Fisher	tP	Fisher	tP	Fisher	tP
1	1	5	5	100	100	100	100	100	100
	1.5	0.6	1	25	26	100	100	100	100
	2	0.6	1	18	22	87	86	100	100
	2.5	0.6	1	18	22	75	80	100	100
1.25	1.25	0	0	48	51	100	100	100	100
	2	0	0	0	0.1	15	15	100	100
	2.75	0	0	0	0.1	3	5	69	66
	3.5	0	0	0	0.1	3	5	69	66
1.5	1.5	0	0	0.2	0.3	99.2	99.4	100	100
	2.5	0	0	0	0	0	0	54	52
	3.5	0	0	0	0	0	0	1	2
1.75	1.75	0	0	0	0	51	58	100	100
	2	0	0	0	0	2	3	100	100
	3.25	0	0	0	0	0	0	0	0
2	2	0	0	0	0	2	3	100	100
	2.5	0	0	0	0	0	0	36	43
	3	0	0	0	0	0	0	0.1	0.2
	3.5	0	0	0	0	0	0	0	0
2.25	2.25	0	0	0	0	0	0	88	91
	2.5	0	0	0	0	0	0	35	42
	3	0	0	0	0	0	0	0.1	0.2