Wednesday, September 2, 2009

Tidbits and Wrap Up on The Bell Curve

1. The Bush Administration made a major push regarding tranparency of institutional performance both in the health care arena, "scorecards" for hospitals, and in the educational arena, scorecards for schools. Almost everyone is for the idea at the broad strokes level, but the devil is in the details. In particular, in the Hospital case, some of the measures don't control for the prior health of the patient. When the very best hospitals attract the sickest patients, what happens to the performance measures.? I don't know those data very well. There are analogous data for College graduation rates. There is substantial variation across the Big Ten. We're tied for third (with Penn State). Northwestern is first. Michigan is second. Our (six year) graduation rates are in the low 80s. Some other schools in the Big Ten have graduation rates in the 60s. But we are more selective than those schools and family income may vary as a separate explanatory variable. If you don't control for those things using the graduation rate as a university performance measure may confound inputs with outputs.

2. I can't recall the discussion exactly, but the conclusion was reached in class that the Hospital in Cincinatti gained the trust of the families of patients by releasing performance information, though that information was pretty damning. I don't believe, however, that we dealt properly with the complexity of the case, because during a previous period information was concealed. In that case, the revealing of the information could potentially harm trust rather than restore it. Their sincerity in pursuing reforms mattered a lot to get the outcome they did achieve.

3. I have not yet made the case for the connection between learning of the individual professional and change at the organizational level. Most professionals, I believe, would give at least lip service to the notion that part of their credo is to keep learning on the job. Schon calls it "learning in action." If that learning professional happens to be in a strong leadership position in her organization, then the individual learning might translate into organizational change for that reason. If however, the professional is more isolated, the knowledge may not diffuse readily. You may be interested to learn that there is an organization to promote scholarship of teaching and learning. There are some people quite active in that organization on Campus. But to date the systemic diffusion of their work has been modest, at best.

4. There is always sampling error as a factor to explain performance variation. You can be confident that this sort of "noise" is not that big a deal if the sample size is sufficient and if the samples are properly constructed (actual observations are truly a random draw). The other factors that explain variation in hospital performance are at the low end: out and out negligence, falling behind currently established best practice, and likely being poor about sharing information on these fronts (because if the information was shared well how could the practices persist)? On the high end Gawande explains the variation by constant innovation in practice. These practice innovations are not discussed much in general public discussion about improvement in health care, where the focus is on new technologies (they aren't new anymore but think about MRI or CT scan), new drugs, and the like. Those inventions don't come from individual practitioners but rather from large corporate research labs (and the pure science at University labs that precedes the corporate research). So Gawande is making an argument that we didn't really fully develop, to distinguish from practitioner innovation from this other sort of innovation. His argument is to take more seriously the former. It is rather important and we tend to ignore it.

No comments:

Post a Comment