Fowler urges skepticism, rigor, and openness in the pursuit of evidence-informed policymaking.
October 02, 2025
Ted Gregory
Professor Anthony Fowler

In 2005, four prominent climate scientists received extensive, favorable media coverage after publishing a paper in Science claiming that climate change was causing a significant increase in severe hurricanes.

The problem was that the scientists’ study contained serious flaws. After the paper was published, critics found that if the range of years analyzed was expanded, or the cutoff used to classify severe hurricanes was changed, the increase in severe hurricanes disappeared.

“Several of the authors have actually come out publicly and said they no longer believe this result,” Anthony Fowler, the Sydney A. Stein Jr. Professor at Harris, told about 200 people in the school’s Keller Center in September. “They essentially said that we really don’t know for sure if climate change is causing an increase in hurricanes or not, but you shouldn’t cite our paper as evidence that it is.”

Fowler, who is editor-in-chief of the Quarterly Journal of Political Science and host of Not Another Politics Podcast, cited the example and several others in a thought-provoking, nearly hourlong Aims of Public Policy Address.

The annual address has become a signature event of Harris’ Welcome Week for incoming students at Harris. Fowler’s talk, “The Politicization of Science and the Challenges for Evidence-informed Policymaking,”  was synthesized from his 2023 article in Skeptic and a 2021 book, Thinking Clearly With Data, that he co-authored with Ethan Bueno de Mesquita, AB’96, dean and Sydney Stein Professor at Harris.

Bueno de Mesquita, who introduced Fowler, called the Aims of Public Policy Address the “intellectual table setting” for new students at Harris. Presented every year since 2013, the address offers “some of the big issues we’re going to face conceptually” and underscores “how we treat the Harris master’s program as a really serious intellectual enterprise,” Bueno de Mesquita added.

Nine politicization problems

Before citing the hurricane study example, Fowler discussed what he sees as a fundamental issue for evidence-based policymaking: sometimes scientific consensus does not occur through rigorous analyses. Instead, consensus can be the result of flawed research, insufficient or selective vetting, or a combination of problematic factors.

He cited nine such “problems”: (1) herding, which is where scientists take cues from each other, leading them to “herd” on the wrong answer; (2) scientists assuming a position is the consensus view without providing evidence that it is actually the consensus view; (3) relying on a survey of scientists which cannot show how certain scientists are in their beliefs; (4) scientific studies may be biased, and errors in different studies do not cancel each other out; (5) the tendency of journal editors and peer-reviewers to accept results that align with their predispositions rather than those which conflict with conventional wisdom; (6) a scientist’s tendency to uphold consensus to gain the acceptance of peers, who may have some control over their career; (7) a conflation of values and science, as science alone cannot dictate which policies to implement, which also requires the values of others, including citizens, elected officials, and bureaucrats; (8) politicization of a topic); or, (9) actual fraud, where scientists manipulate the data.

Peppering his talk with examples, Fowler closed by providing guidance on how to avoid those problems. Skepticism and rigorous analysis are necessary, he said, “but you should not be a nihilist.”

COVID origin controversy

One of the more newsworthy examples Fowler noted when discussing how career incentives can detract from impartial, comprehensive analysis was the controversial effort to find the origins of COVID-19.

To support his contention, Fowler presented slides of emails from an author of a prominent article, published in March 2020 in the journal Nature Medicine, in which the researchers dismiss speculation that the virus originated in a lab and escaped—something called “the lab-leak theory.”

In a February 8, 2020 email, lead author Kristian G. Andersen wrote that although the research team was “focused on trying to disprove any type of lab theory… the scientific evidence isn’t conclusive enough to say that we have high confidence in any of the three main theories considered.”

Andersen went on to oppose publishing the research document and in a February 20 email, stated that the research team had been unable to refute the lab-leak theory, “and the possibility must be considered as a serious scientific theory….”

By March 17, the authors wrote, “we do not believe that any type of laboratory-based scenario is plausible.”

That conclusion aligned with the virology community’s desire that the pandemic outbreak had natural, zoonotic origins and hadn’t been the result of a lab leak. With the benefit of hindsight, Fowler said, we know that virologists probably were over-confident in that conclusion, “because now, a lot of scientists think that the lab leak was at least highly plausible, if not the most likely scenario.”

Fowler said the saga suggests that when the virologists embarked on their research, they might have been more concerned about protecting the reputation of virology and producing a result that would please their peers—and enhance their career standing—than reaching a scientifically-sound conclusion.

“The main path to success is by pleasing your fellow peers,” Fowler said. “It’s an interesting challenge, because on the one hand, your peers presumably are the most qualified to evaluate work that you’re doing. But on the other hand, they might be the ones with the most vested interest in the results going one way or another. That’s a real source of tension.”

Greatest threat to policy research

Fowler said he hoped that research fraud is “a relatively rare problem” but was worth mentioning. He cited three examples.

In one, a doctoral student at MIT received considerable media attention for his 2024 paper claiming that his analysis of data showed that using AI significantly accelerated discoveries in material science. The attention generated scrutiny from MIT professors who in May 2025 stated they “have no confidence in the validity of the data and in the veracity of the research.” 

In another example, a widely publicized 2014 study by a UCLA researcher and Columbia University professor contended that gay canvassers had a large, enduring effect on generating support for gay marriage. In May 2015, Science, which had published the study, retracted it due to concerns about the validity of the research.

Fowler’s final example of fraud involved behavioral scientists from Harvard and Duke who, he said, committed fraud “across multiple studies.” “Ironically,” he added, “many of those studies were about the topic of dishonesty.”

But the greatest threat to the credibility of policy-relevant research, Fowler said toward the end of his address, emanates from political pressures and ideological biases in the policy community.

“All of the previous concerns, all of the previous challenges for the scientific enterprise, are exacerbated when a scientific question is particularly relevant for public policy,” he said.

The imposing challenge for public policy researchers “is that we all have our own personal values and preferences,” Fowler added. “We all have different ideological preferences. We all have our ideas about what the right policy should be, and that is going to influence how we analyze data. That’s naturally going to be an extra challenge that we have to think about when deciding whether particular evidence is persuasive.”

Politicization influences policies in other ways, Fowler added. The Trump administration currently is targeting universities, threatening to cut federal funding, which might affect research results, he said.

“I think we should do everything we can to push back against” the tendency to skew results to gain favor with any presidential administration, Fowler said.

First-year Harris students Sadhika Thapa and Amen Bekele sat next to each other during Fowler’s talk. Thapa called it “eye-opening” and said it will prompt her to keep a more open mind when reading research and claims of a certain result.

Bekele, who worked in a lab before enrolling at Harris, said Fowler’s comments resonated with her.

“I used to read a lot of papers and work with people who really wanted results to come out of nothing,” she said. That environment drew her from a research career track and into an interest in municipal finance.

Thapa also said she will give more consideration to opposing views when researching and writing her papers. Bekele said she’ll focus her reading of studies on the results section instead of the abstract and conclusion.

Two important questions

Avoiding pitfalls of politicization, Fowler said, is a matter of pushing back against a scientific consensus and maintaining a commitment to rigorous policy analysis. It comes down to asking two questions when students scrutinize a study, he said.

First, does the finding align with the predispositions or political leaning of the research community from which it came? Second, if the researchers had done the same analysis and come up with a different—even an exact opposite—conclusion, would the study be equally convincing and a paper been written about it?

“To the extent that you’re trying to convince somebody of the truth,” Fowler said, “you should convince them on the basis of the evidence, on the merits of the argument, not on the basis of authority or scientific consensus.”

Previous Aims of Public Policy Addresses