Fig. 1From: Inverse probability weighting to handle attrition in cohort studies: some guidance and a call for cautionScheme of the data-generation model for the simulation experiments. Seven covariates, differing in their association with the variables of interest (exposure and outcome variables) and the response variable, were generated. The strength of the associations (dashed arrows) between the variables of interest and the response variable (γx and γy) varied according to the scenarios described in Table 1Back to article page