Re: 'Methods of covariate selection: directed acyclic graphs and the change-in-estimate procedure'.

摘要

I read with great interest the recent paper by Weng et al. (1), in which the authors compared a directed acyclic graph (DAG) full model and 3 DAG and change-in-estimate combined procedures. I wish to comment on the reported sampling distribution-based standard error. I hope this comment can help to clarify why the results on sampling distribution-based standard errors reported by- Weng et al. (1) are not consistent with the theoretical results established in the paper by Robinson and Jewell (2).