Saturday, October 10, 2015

Litany of Tainted Research

Most of what Luigi Zingales has to say about the Litan affair is right on target: reputation effects are not sufficient to prevent researchers from shading their conclusions to satisfy funders.  Litan himself was guilty of failure to disclose his sponsors, the Capital Group, for testimony he presented to Congress, irrespective of the merits or otherwise of his work: who is paying for research is relevant to forming a judgment on it, and not only whether this or that variable was properly identified.  And Senator Warren was right to call him out.

Still, there are problems of omission and commission with Zingales’ piece.  Commission #1: I strongly disagree that public funding of research raises the same issues as private.  Says Z: “Eliminating private funding will leave research completely in public funding’s hands.  It will not eliminate the bias; it will simply tilt it in the direction of the government.”  Really, Zingales has been hanging out in Chicago too long.  As a sometimes recipient of public funding (disclosure), I think it’s ludicrous to say that the interest of the “government” bears on how us grantees do our work.  Seriously, what government interest does NSF represent?  In any case, there is a legal basis for exposing bias and malfeasance in public research support that does not exist in the private sector.

Commission #2: “....reputational incentives work relatively well only for academic papers that circulate widely in the relevant academic community and are independently scrutinized in peer review.”  I wish.  The reality is that peer scrutiny is not a process that separates the worthy from the unworthy in an objective, reliable manner.  Just look at any journal: quite a bit of shady stuff survives peer review and gets published, even if their results are not replicable.  In addition, as any author knows, there are a myriad of detailed matters in research that can be resolved in a variety of ways.  You can push the boundaries of a category a little bit this way or that, make or not make a restrictive assumption, and so on.  Your peers are not going to write you off just because your choices lean a little in a particular direction as long as you are technically sound.  The moral of the story: disclosure of funding sources is essential for all published research, no matter how and to whom it is disseminated.

Omission: Zingales provides two suggestions for reform, advance posting of expert testimony at least two weeks before a Congressional hearing and public disclosure of the identities of expert witnesses.  But the most important reform is missing: we need a single, consolidated source of information on all publishing economists disclosing all their sources of funding.  This could be a website where any journalist, congressional staffer or interested member of the public could find out if a given author had taken money that might influence his or her work.  Beginning with the AEA, professional associations and journals could make accurate listing in this database a condition for participation or submission.  Who’s paying you should not be treated as a trade secret, not if you claim to be contributing to the store of human knowledge.


Ragout said...

You're mistaken to claim that "Litan himself was guilty of failure to disclose his sponsors, the Capital Group". Actually, the cover page of Litan's paper states: "Funding for this paper was provided by the Capital Group."

Peter Dorman said...

Ragout: You're right, I'm wrong. The problem was not that Litan didn't disclose funding, but that he didn't indicate that the study itself was commissioned by the Capital Group. In this case, then, a funding disclosure system would not have made a difference, although I think it's still a good idea.

Thanks for the correction.