The Marshall Institute, a conservative think tank, recently released a report on “Funding Flows for Climate Change Research and Related Activities” which asserts that, “In today’s highly charged environment of climate change policy, efforts are often made to impugn the credibility of those engaged in the debate through assertions that their views are a product of financial relationships rather than sincerely held beliefs or objective research. All too frequently evidence of a financial tie is sufficient to condemn, without proof that the tie altered the views, opinions, or conclusions in any way.” After this complaint the Marshall Report joins this game by arguing that corporate funding ties are emphasized but “overlooked are concerns that public funding generates unwelcome pressures on scientists to conform to prevailing beliefs. Public funding is also said to breed alarmism and facilitate distortion in public discourse.” (Disclosure: The Marshall Report ranks my employer the University of Colorado as the top recipient of federal funds on climate change, and I have a few federal grants myself.) What should we make of claims of conflicts of interest? Are they restricted to financial conflicts? And can they be avoided?

Discussed below: Conflicts of interest do matter. They are not restricted only to financial considerations, and they cannot be avoided, only managed.

An article in yesterday’s USA Today illustrates why debate over conflicts of interest can be damaging to science and its role in policy. The article describes a just-published meta-study of health effects of a certain chemical in plastics, “Is it possible that a chemical’s effect is in the eye of the beholder? That’s the implication of a paper published this week in a prominent environmental health journal. It concerns a debate over the safety of low doses of a chemical used to make hard, clear plastics such as those found in baby bottles, food-storage containers and the lining of soda cans. When the plastic industry examines the health impact of a ubiquitous chemical called bisphenol A, everything’s fine. If the government or a university funds the study, there are big problems.”

The study, by Frederick S. vom Saal and Claude Hughes, (available here in PDF) made the interesting discovery that “Source of funding is highly correlated with positive or negative findings in published articles. For government-funded published studies, 94 / 104 (90%) report positive effects at doses of BPA below 50 mg/kg/day. No industry-funded studies (0 / 11 or 0%) report positive effects at these same doses.” This leads the authors to suggest two possible reasons for these findings, “1. Are government-funded scientists under real or perceived pressure to find or only publish data suggesting adverse outcomes? 2. Are industry-funded scientists under real or perceived pressure to find or only publish data suggesting negative outcomes?” vom Saal and Hughes clearly come to the conclusion that the government funded studies are those that are correct, while an industry spokesperson countered, “You can have 1,000 studies, but if they’re all weak, adding up weak evidence doesn’t necessarily give you strong evidence of anything, Jumping to who sponsored it is a way to dodge the facts.” And if we believe that sources of funding shape conclusions then we should not be surprised that the von Saal and Hughes study is government funded and the industry spokesperson works for, well, industry. This situation leads to the USA Today’s comment that research findings are “in the eye of the beholder,” a conclusion sure to infuriate many scientists who reject such relativistic post-modern claims to the subjectivity of knowledge.

It is important to recognize that the chase of chemical risk assessment (and climate change and many others) putatively scientific debates are real proxy wars over politics, that is, over a specific course of action, e.g., to regulate or not. It just so happens that the political battle is taking place through the language of science. A common assumption is that disclosure of sources of financial support is sufficient to address the influences of the conflicts of interest. But as the Marshall Institute report and USA Today article suggest, this may do little to shape a consensus on either science or action. The Marshall folks are suggesting that government studies are biased and vom Saal and Hughes are suggesting that industry is biases. But what if the reality is that everyone has biases and the world in not in fact black or white?

In an editorial in The Lancet (PDF) Richard Horton explains why disclosure of potential financial conflicts of interest alone is insufficient to deal with this issue, and begins to point to a more thoughtful treatment of this issue:

“… the case in favour of full disclosure rests, it seems to me, on three large fallacies. First, there is the fallacy of objectivity, the notion that scientific writing can be free from the common prejudices found in other literatures or journalism-or that if prejudice does exist it can be easily neutralised by a statement of disclosure. Yet interpretations of scientific data will always be refracted through the experiences and biases of the authors. Scientific writing can never escape from being a rhetorical exercise. Advocates of disclosure may argue that some of these potentially malign influences could be limited by focusing on the most serious conflicts. This brings us to the second fallacy-that it is financial conflicts of interest which “cause the most concern”. Financial conflicts may be the easiest to identify but they may not be the most influential. Academic, personal, and political rivalries and beliefs are less easily recognised, but each may affect an interpretation. Such biases render the declaration “conflict of interest: none” an impossibility. To put financial conflicts to the fore is to provide a smokescreen for more covert and possibly more influential commitments. The third fallacy is that disclosure can heal the wound inflicted by a financial conflict… ”

Horton suggests that there are a range of sources of potential conflict beyond the financial:

“Interests (commitments) facing an investigator

Professional (eg, personal, specialty, departmental, or institutional status) Financial (eg, personal reward; research funding) Patient-related (eg, as a personal physician; payment for study recruitment) Institutional (eg, ethics committee) Grant-related Regulatory (eg, FDA) Scientific publication Mass media Legal (eg, patent protection) Sociopolitical Public interest (eg, research support through taxes, charitable donations)”

Horton concludes that “The only way to minimise bias among interpretations is to allow maximum dialogue from all parties, irrespective of their interests.” He is correct in pointing toward procedural remedies. To this I would add that for scientists interested in limiting the morphing of scientific and politic debates, a focus on the choices available to decision makers (i.e., policy) and not just the science can help scientists to counter incentives to wage political battles through science. More in depth and thoughtful treatment of these issues can be found in the papers by Harrison, Oreskes and Herrick in the special issue of Environmental Science and Policy that I co-guest edited last year.