Conflicts of interest in climate change science
A new pre-print. It's time for professional norms to step it up
In 1981, the young Tennessee Representative, Al Gore, led the charge in Congress to understand a series of fraud cases in biomedical science. Testimony by prestigious researchers illustrated a scientific culture that resented policymaker inquiry about research misconduct believing that the methods of science would inevitably bring to light and correct any bad behavior.
During the hearing, Congressional members were annoyed that at the same time researchers testified to the problem of misconduct being limited and manageable through informal community norms, they also gave evidence that even among the witnesses present there was more fraud going on than Congress knew.
Gore observed that,
In one short hour, the subcommittee went from the Olympian heights of a nonproblem to the depths of a potential perjury.
Gore called it “scientific schizophrenia.” He continued,
It is my belief, however, that all of us, including the scientific community, will be better off for the examination of its dark corners.
As the decade wore on and several high profile misconduct cases came to light the issue of researcher conflicts of interest became a central focus.
The federal government implemented conflicts of interest disclosure rules for those receiving grant funding from Health and Human Services in 1995. In 2009, the National Academy of Medicine (then, the Institute of Medicine) released a report reviewing the issue of conflicts of research in medicine and opened with the statement that, “Hardly a week goes by without a news story about conflicts of interest in medicine.” In 2010, Congress passed the Physician Payments Sunshine Act requiring certain companies to keep track of their financial arrangements with physicians and report them to a central repository.
If conflicts of interest is such a pervasive issue in medicine we should expect, in the very least, for it to be a concern issue of concern in other areas of research with substantial industry or political interest.
And yet, mum’s the word from the scientific community for improved disclosure of researcher conflicts of interest in climate change research despite decades of serious controversy around some of its leading practices, lapses in scientific integrity, entire areas of technological practice overtly oriented around advocacy and litigation, and knock-on issues in finance, insurance, energy and seemingly, well-being in young people.
That the need for better disclosure practices is not a central topic among climate scientists and journals points to the same dynamics that became apparent at the Congressional hearing led by Al Gore in 1981.
Good news!
My colleagues and I have a new preprint available for review and comment taking a first step towards understanding these dynamics in climate change research:
(I’m having trouble with the SocArXiv site. You can access the paper here until the site starts working again).
Here are some key take-aways:
We analyzed 82 peer-reviewed articles on the relationship between climate change and the geophysical properties of hurricanes published between 1994 and 2023 to determine whether conflicts of interest disclosures, funding support, or author affiliation are associated with study outcomes or recommendations.
We chose studies using a search that focused our attention on the literature engaging the detection and attribution area of study.
NGO funding was a significant predictor for an article to find a positive association between1 climate change and geophysical characteristics of hurricanes as a research outcome.
Our most important finding is that
None of the 331 authors disclosed COIs. Not a single one.
This is surprising because
COI disclosures in other areas of research, such as bioscience, range from about 17% to 33%
There is quite a bit to indicate that researchers have conflicts of interest. In just our study sample we found that
An author holds a patent relevant to the research reported and is an advisor to a risk analytics firm and a financial company
An author is an advisor to a climate risk analytics firm
An author is a non-resident scholar for an insurance industry association
One or more authors developed research methods in collaboration with an advocacy organization to pursue climate litigation
Given the confluence between government, academia, industry, and NGOs in climate research, many researchers may have non-financial COIs they are not disclosing.
To promote objectivity, transparency, and trust in climate science, journals that publish climate change research should clearly state that authors must disclose financial and non-financial COIs and provide clear processes for doing so.
The International Committee of Medical Journal Editors provides a good template for author disclosure of financial and non-financial conflicts of interest. Even better is a central repository to reduce time burden on authors and make disclosures accessible across journals.
All of this matters because a robust body of literature demonstrates correlation between research outcomes and researcher conflicts of interest.
This happens for a lot of reasons unrelated to fraud (though that is one reason). The choice of research question, analytical methods deployed, and interpretation of results impact research outcomes and those choices may be influenced by conflicts of interest.
When assessing the impacts of conflicts of interest on research, ethicists David Resnik (a coauthor on the preprint) and Kevin Elliott point to five factors to consider, which I amend to consider politics:
Do sponsors, institutions, or researchers have a significant financial or political stake in the outcome of a study?
Do the financial interests of the sponsor, institution, or researcher coincide with the goal of conducting research that is objective and reliable?
Do the sponsors, institutions, or researchers have a history of biasing research in order to promote their financial or political goals?
How easy is it to manipulate the research in order to achieve financial or political goals?
Are there oversight mechanisms in place that minimize bias and make it more difficult to manipulate research successfully?
Based on these factors much of climate change research is highly susceptible to bias from conflicts of interests.
For example, there is a lot of money and at stake for researchers with conflicts of interest with from risk analytics firms. There are substantial policy considerations at stake for researchers with non-financial conflicts of interest with political actors, regulatory agencies, and NGOs. Extreme event attribution methods are particularly susceptible to bias from research conflicts of interest because of the range of choice in each stage of the method(s) and the relevance of the outcome for media, litigation, and ‘loss and damage’ financing.
It’s important to know that a conflicts of interest disclosure is not a statement of the morality of the researcher or the scientific merit of the publication. However, because bias and assumptions in research are impossible to remove disclosure provides important information for judging the credibility of research and to assess the state of knowledge on a specific topic.
As our research shows though, in order to truly provide better understanding about the impact of conflicts of interest on climate change research it is necessary for authors to, in fact, disclose their conflicts. This means that professional organizations and journals have to step up and help researchers understand what financial and non-financial conflicts of interests are and how to disclosure them.
Historically, concern about conflicts of interest in research originates with individuals who know a research area well and its main players. My hope is that our latest work is just the first step in increased examination of conflicts of interests in climate science and improved professional norms of disclosure.
There is a lot of work to do. 2
It’s important to know that our work is not a technical assessment of the detection and attribution science regarding hurricanes. Our coding took at face value researcher claims in the abstract and conclusion section. Our method considered if researchers claimed to document a change in hurricane characteristics that they then explained or in some way associated with “climate change.” (see footnote 2 for further thoughts)
I also learned from this research process that the format for climate change reporting would benefit from a certain level of standardization so that methods, results, and outcomes are articulated clearly to support meta-analysis and systematic reviews of the literature for assessment reporting.
As a health scientist and researcher COI, fake studies, and fraudulent ones are on my daily radar. I always suspected that climate science is no exception to the humongous degree of biased papers that we see in science in general. I have seen the influence if NGOs in my area, too. From villifying tuna fishing to evangelizing "planetary health diets". I encourage you to keep up your sleuthing work. It is essential and highly necessary.
This is heroic work. I find this issue, along with many others, tedious and boring and so far from the basic science that I can barely bother to read the details.
But, alas, the details are totally necessary.
I prefer to always state the existence of the elephant being still in the room, Carbon dioxide in the atmosphere goes up and down as the ocean temperature goes up and down, first the ocean and then the atmospheric carbon dioxide. Temperature sets CO2 level, CO2 does not set ocean temperature level.