Table 2.

Estimated beta coefficients, exponentiated beta coefficients and standard errors derived from 5000 samples of 1000 individuals sampled from a population with the same characteristics as the sample population shown in Figure 1 and described in the text

Approach	Parametric assumption	Average beta	Exp (average beta)	Average standard error
Crude regression	Log-binomial	0.290	1.336	0.1889
Adjusted regression	Log-binomial	0.686	1.986	0.1882
Multiple imputation	Logistic monotone	0.684	1.982	0.1967
Parametric g-formula	Log-binomial	0.686	1.986	0.1882