I don’t want to stand in the way of a potentially useful analysis, whose merits could be assessed after publication. To some extent the applicant has addressed my earlier caveat about the use of a surrogate measure of treatment expectation, by stating that the analysis will depend on there being a direct record of such expectation embedded in the IPD. I don’t know how likely that is, and to find out will require full processing and deidentification of the 9 datasets requested. I am not sure if the applicant has the support of a full statistical team in this exploration – “This project is for fulfillment of an undergraduate senior thesis.”
Despite my reservations, I would be happy to accept if others feel that this sufficiently meets YODA criteria.