Ye, Ertefaie, Flory, Hennessy, and Small (YEFHS) proposed a new method, instrumented difference-in-differences, for dealing with unmeasured confounding. In this note, I connect and compare assumptions and identifications in instrumental variable (IV) and difference-in-differences (DID) methods with those in YEFHS, derive new identification results, and discuss different choices when extending such results to adjust for covariates.
Download full-text PDF |
Source |
---|---|
http://dx.doi.org/10.1111/biom.13781 | DOI Listing |
Enter search terms and have AI summaries delivered each week - change queries or unsubscribe any time!