• Treatment/control randomization is IV

    Just a quick follow-up to my prior post on IV. I used a (flawed) treatment/control randomization as an example. My guess is that many people are not aware that the randomization (the coin flip) is an instrumental variable. If this is not clear to you, I highly suggest reading Steve Pizer’s paper on this exact point. His (and my) hope is that it makes IV much more accessible to health services researchers.

    • Yeah, it’s actually the only instrumental variable that I feel comfortable with. In medical research I am very worried about other approaches and worry that they could result in serious amounts of bais due to the complexity of medical problems. Physician preference and geogrpahy are both examples that require fairly tough assumptions to really make work.

      • @Joseph – I’m with you, but only insofar as RTCs provide the best evidence. I won’t reject other types of studies if they’re all we have. What’s left? Guessing?

        Unfortunately we do not have the luxury or the money to do RTCs for everything of interest or import. Then what? I’d rather take a body of work with seemingly good instruments than a bunch of quite bad (and useless) studies that show only correlation. I did a whole series on this. Some observational studies show that Medicaid is associated with worse health outcomes than no insurance at all. Are we to believe that Medicaid makes people even sicker? No! And IV studies show us something closer to the truth, or so I believe (Medicaid is good for health). We’ll never randomize to Medicaid or not. This is the best we’ll get. There are many other examples.