March 21, 2023 - 5:10pm

A leading science journal’s endorsement of Joe Biden in 2020 resulted in a significant reduction in trust among Donald Trump-supporting readers, a new paper has found. At the same time, trust among Biden supporters rose only slightly.

Performed in late July and early August 2021, the experiment randomly assigned participants to receive information about Nature’s endorsement, while the control group were given irrelevant information about the journal’s new website design. Both Trump and Biden supporters were presented with the author’s summary of the messages conveyed in Nature’s 2020 editorial. 

Published on 14 October, the editorial outlined why the journal supported Biden for president. “Joe Biden,” it read, “is the nation’s best hope to begin to repair this damage to science and the truth — by virtue of his policies and his leadership record in office, as a former vice-president and as a senator.” Attacking Trump for his Covid response and for “promoting nationalism, isolationism and xenophobia”, the editorial claimed that only Biden could “restore trust in truth, in evidence, in science and in other institutions of democracy”. 

At the time, Nature‘s was one of many high-profile political endorsements by scientific journals, all of which lined up behind the man now occupying the White House. But the publication’s own findings this week reveal that its bid to galvanise support for Biden may have had the reverse effect. 

Participants were given a screenshot of the editorial’s title and first paragraph, and a link to the full text. They were then asked various follow-up questions.

Credit: Nature

As the table above shows, the strength of reaction to the endorsement was particularly acute among Trump supporters, with treated participants (i.e. those who viewed the endorsement) scaling much higher on negative attitudes towards Nature‘s impartiality and knowledge. For Biden supporters, there was a marginal uptick in positive attitudes.

Besides reducing trust in Nature among Trump supporters, the endorsement also had a cascading effect on their attitudes towards science and scientists more generally. Trump supporters were less likely to report high levels of trust towards US scientists than Biden supporters after the endorsement, and the gap is larger for treated participants. 

In addition, the shifts in trust in Nature resulted in lower demand for Covid-related information from the journal among Trump supporters. Results found that, at the height of the Delta variant surge in the US, the endorsement led to a -14.2 percentage point reduction in the frequency at which Trump supporters requested Nature articles, whereas the upswing for Biden supporters was negligible.

These results show that Nature’s endorsement had a hugely divisive effect on its readership. Treated Trump and Biden supporters became two to four times more polarised than the control participants on these stated measures of trust in Nature. Meanwhile, treated Trump supporters were 38% less likely than control Trump supporters to request stories from the publication’s website. What’s more, the endorsement had little effect on changing participants’ opinions about the two presidential candidates, rendering it ineffective.

In response to the research, Nature published an editorial defending its endorsement. Entitled ‘Should Nature endorse political candidates? Yes — when the occasion demands it’, it argued that “when individuals seeking office have a track record of causing harm, when they are transparently dismissive of facts and integrity, when they threaten scholarly autonomy, and when they are disdainful of cooperation and consensus, it becomes important to speak up”.

is UnHerd’s Newsroom editor.