Table 1.

Parameter values assumed in simulation studies. The models used for the generation of binary exposure X and binary outcome Y are respectively logit(P(X=1))=η₀+η₁C₁+η₂C₂ and logit(P(Y=1))=γ₀+γ₁C₁+γ₂C₂+βX, where confounders C₁∼Bin(1,0.1) and C₂∼Bin(1,0.6)

	Exposure model				Outcome model
Setting	η0	η1	η2		γ0	γ1	γ2	β
I	$log\frac{0.5}{0.5}$	2	2		$log\frac{0.3}{0.7}$	2	2	log(1), log(3),log(1/3)
II	$log\frac{0.5}{0.5}$	2	2		$log\frac{0.3}{0.7}$	2	−2	log(1)
III	$log\frac{0.5}{0.5}$	2	2		$log\frac{0.3}{0.7}$	−2	−2	log(1)