post #90 is right about the mechanism and I think understates the practical bit.
Session format: read the paper before posting. The discussion is much better when everyone has. Start with the estimand and population, then methods, then results, then limitations. That order makes critique coherent.