Sunday, May 23, 2010

New NIH disclosure rules proposed

New disclosure rules regarding potential financial conflicts of interest may be on the way for NIH-funded researchers.

I support this sort of disclosure but worry that the public discussion of issues of bias in research funded by government grants and contracts tends to focus almost exclusively on issues related to researchers who receive funding from private firms.

Also important, but rarely discussed, are the incentives that face researchers to pander to the perceived objectives of the government agency funding the research. Two good examples from the Bush II administration are the Reading First program and abstinence-only sex education, which administration was heavily invested in and where it pretty clearly had an answer it wanted to hear from the evaluations.

These sorts of issues, which come up regardless of the party in power, are just as real as the private sector related concerns that appear to motivate the NIH proposals. When it became clear that the Reading First evaluation was not going to produce a shiny positive and statistically significant impact estimate, the contractor got, justifiably, a bit nervous. As a way of covering themselves, I think they had nearly every economist who had ever written a paper using a regression discontinuity design (including yours truly) review and sign off on the design and implementation of the Reading First study.

Also potentially important are biases related to individual ideology, as when individuals with strong normative views related to the environment sort into doing climate research.

I am not sure quite how one goes about developing institutions to alert the research consuming public to these issues, but they are surely worthy of note.

Hat tip: Nancy Herlocher, UM Economics' amazing computer poobah.

1 comment:

sutirthabagchi said...

I have thought about writing a blog on this myself and am thrilled to see that you picked up on it. Yes, when you have a lab which is called (and I am making this up) "Laboratory for Study of Human-Induced Climate Change" what is the likelihood that a scientist from that lab will come out and say that climate change is not happening. Indeed, what is the likelihood that he will starve himself and his family and put himself out of a job?

In a different context, commenting on media bias, I have suggested that (and here I quote from my blog)
"how about having a similar disclosure norm for the political journalists who cover the political news segments. At the end of a 1 hour segment run by CNN on the upcoming presidential elections, we might have a disclosure on the lines of : "Campbell Brown: Democratic; Wolf Blitzer: Democratic; Anderson Cooper: Democratic; Rolland Martin: Democratic". And then, the audience, people like yourself who expect the media to be unbiased, could make up their minds about what they listened to was balanced media coverage or not. If this sounds like too drastic a plan that curtails individual privacy, we could at least move to a system where at the end of each segment, we have a statement on the lines of: "Of the 56 (I am making a number up, for illustrative purposes) journalists who were a part of the crew, 39 of them were registered Democrats, 6 were registered Republicans and the remaining were unaffiliated with either party." I am certain that the audience would be enlightened by such information."