@harlingg It’s probably not quite the type of example you’re looking for but my favorite IV example is the work Sonja Swanson has done on the per-protocol effect in the NORCCAP trial. Here’s a paper on bounding that effect: https://t.co/cuY82gR260
@edwardhkennedy @ProfMattFox Or if blinding isn’t good, people need to not be able to change their outcome risk because of knowing their assignment. Here’s an example that gets around some of the problems nicely: https://t.co/cuY82gR260
Bounding the per-protocol effect in randomized trials: an application to colorectal cancer screening https://t.co/ETeWjH2Qrf
Bounding the per-protocol effect in randomized trials: an application to colorectal cancer screening https://t.co/YMZf3IJttK