I particularly like @Sander Greenland’s 2005 ‘Multiple-Bias Modeling’ discussion paper, mentioned here.
Inverse probability weighting for treatment selection and loss to follow up in observational studies
4 Likes