Industry funding of scientific research was in the news last week. The most prominent story was that Coca-Cola is funding obesity research. But I had an interesting Twitter exchange about a different story: Clif Bar and Organic Valley are funding research on organic agriculture at UW-Madison.
To paraphrase the Tweet from Tamar Haspel at the top of the Twitter exchange, it’s hard to see why funding from Clif Bar and Organic Valley for organic research is substantively different funding from Monsanto or Syngenta for agricultural biotechnology research. If one is bad, then so is the other; or conversely, if one is perfectly acceptable, then so is the other. Or, more generally, how do we tell whether the values of funders are having an illegitimate influence on research?
Brandon Horvath offered an interesting response to that question, in a tweet which seems to have disappeared. As I recall, his criterion was something like “being able to publish unfavorable findings.” That is, if a research group can publish findings that are unfavorable to their funders’ interests, then the influence of the funders’ values is legitimate.
There’s a lot that I could say about this proposal. For this post, I’m going to assume that Horvath’s proposal is a good one, and that this kind of independence in publishing does give reason to think that the influence of funders’ values is legitimate.
What I want to consider are some epistemic challenges for this proposal. That is, I want to consider how people outside the research group can get evidence that Horvath’s condition is satisfied (or not) in any particular case.
First, Horvath’s condition involves a counterfactual: it’s about what would happen if the research group decided to try to publish unfavorable findings. If a research group does, or has, published unfavorable findings, then it’s relatively easy to show that the condition is satisfied: “we found that this plant-incorporated pesticide didn’t work very well, even though Syngenta gave us millions of dollars to develop it.”
But what if the research group hasn’t published any unfavorable findings? This might be because values are having an illegitimate influence: maybe the researchers have produced some unfavorable findings, but the funders aren’t letting them publish; or maybe the researchers are illegitimately conducting their research in such a way that the research produces only favorable findings. On the other hand, maybe the influence of funders’ values is perfectly legitimate, and it just happens that the researchers have only produced favorable findings.
In other words, if the research group hasn’t published only favorable findings, then Horvath’s condition might be satisfied. But it also might be violated. In this case, we would have to look at other kinds of evidence to determine whether the condition is satisfied. For example, we might need to see some kind of legal document attesting that the researchers have funding independence.
The second challenge has to do with the public accessibility of this evidence. By “public” here I mean the general public, and especially people who don’t have access to paywalled scientific journals. My point here is that even if Horvath’s condition is satisfied and publications of unfavorable findings provide good evidence that the condition is satisfied, because members of the general public don’t have access to this evidence, they’re still not in a good position to conclude that the condition really is satisfied.
There’s a lot in that sentence, so let’s consider an example. We suppose that the research group has published unfavorable findings in several closed-access peer-reviewed scientific articles. Academics and other researchers generally have access to the journals that published these articles through institutional subscriptions. For example, university libraries will pay a subscription fee to the publisher, giving everyone affiliated with the university access to the journal. However, because the journals are closed-access, people without this kind of individual access can only access the publications if they pay the publisher. Access typically costs around 30 USD per individual article. That might seem relatively affordable for one or two articles, but it’s prohibitively expensive if you’re trying to comb through all of a research group’s publications looking for unfavorable findings.
In effect, members of the general public don’t have access to closed-access scientific publications. So if these publications are the only evidence that Horvath’s condition is satisfied, the general public is not in a good position to judge whether industry funding is legitimate or not.
I see at least two ways for researchers to respond to this epistemic challenge. First, they can publish in open-access journals, or otherwise make their unfavorable findings freely available to the general public. (Many publishers allow authors to make pre-copyediting versions of their papers freely available in public depositories and personal websites.) Second, they can work with journalists and other science communicators. Though the exactly rules vary from publisher to publisher, authors are generally allowed to provide individuals with copies of the published versions of their papers. With these copies, a science communicator could prepare a publicly-accessible review and summary of the research, documenting the fact that the research group has published unfavorable findings. (Researchers could do this themselves, of course. But the summary might be better received if it were prepared by an independent author.)
Quick note: Tumblr doesn’t have any native commenting facilities, and I haven’t gone through the trouble of setting up Disqus with my current theme. If you’d like to make a public comment, click here to send me an email. I’ll post comments or links to responses here or in future posts.