New paper in top journal finds evidence of pervasive bias in scientific research

A new paper reviewing current empirical literature on academic freedom in science has been published in the Proceedings of the National Academy of Sciences, one of the world’s most prestigious scientific journals. 

The paper is authored by Dr Cory Clark (corresponding author), a behavioural scientist at the University of Pennsylvania specialising in cognitive biases, along with 38 coauthors.  

The paper defines scientific censorship as “actions aimed at obstructing particular scientific ideas from reaching an audience for reasons other than low scientific quality”. It reviews empirical evidence of the causes and extent of contemporary scientific censorship and makes a number of recommendations. It argues for two claims in particular: first, that much scientific censorship today is carried out by scientists themselves; and second, that it is actuated largely by prosocial concerns, that is, concerns to benefit or avoid harming others.  

CAF takes no view on the work presented in the paper. It merely notes that scholarly research into academic freedom is of high importance and likely to be of interest to all those concerned with the issue. Certain particularly significant findings of the paper are summarised below. 

Censorship in academic journals 

The paper notes that there is a growing body of evidence that acceptance procedures for scientific journals are subject to bias. Many of the criteria on which submissions to journals are assessed, such as “novelty, interest [and] “fit””, are ambiguous and partially subjective. This ambiguity and subjectivity creates a space in which the biases of editors and peer reviewers may operate, intentionally or unintentionally.  

The authors present evidence that such biases are operating. Studies conducted by Abramowitz et al. and Mahoney have found that peer reviewers evaluate research more favourably when findings support their prior beliefs, theoretical orientations, and political views. The authors argue that the problem may often be exacerbated by reviewers being unaware of their biases.  

At the editorial level, the authors note that the world’s top scientific journal, Nature, has issued guidance stating that research may be unsuitable for publication because of its potential to cause harm to individuals not involved in the research itself. The harmfulness of research is of course extremely difficult to assess, and to date minimal research has been done on this topic. The authors contend that this guidance grants the editors of Nature “vast leeway to censor high-quality research that offends their own moral sensibilities”. 

The extent of censorship, and of support for censorship 

The authors find that the incidence of academic staff at American universities being sanctioned for expressing controversial ideas has increased over the last ten years. The authors also note the work of Honeycutt et al. in which 25% of staff at US four-year colleges and universities reported being “very” or “extremely” likely to self-censor in academic publications. 91% reported being at least somewhat likely to self-censor in publications, meetings, presentations, or on social media. The authors refer to the work of Inbar and Lammers which has found that “many academics report willingness to discriminate against conservatives in hiring, promotions, grants, and publications”.  

“Harm motivations” for censorship 

The authors argue that the desire to prevent harm is a significant cause of support for censorship. They note that individuals are particularly supportive of censorship when they believe that publication may influence the views of others (Rojas et al.), and furthermore, that individuals in the UK and the US are more likely to support censorship of research which presents historically disadvantaged groups in a negative light (Clark). Individuals who overestimate potential harms from publication are also more censorious, and overestimation of the harms from publication is common (Clark). However, overall little work has been done on the potential harms of publishing research. 

Suggested measures 

The authors suggest a number of measures to aid the further study of censorship in scientific research, with a view ultimately to reducing it. They note that much of the current research into scientific censorship is carried on by academics working in non-profits and is not submitted to peer-reviewed journals. They urge that more research should be undertaken, and undertaken within academia itself. They also suggest that: 

  1. Feedback from peer reviewers and editors should be made public, with names redacted as appropriate. This, they argue, would enable levels of censorship to be measured, and give journals an incentive to compete to minimise bias. They argue that similar measures should be taken with respect to submissions to learned societies and conferences. 
  1. Scholars should audit scientific journals and institutions for the extent to which they prioritise considerations other than the quality of the work submitted. This could provide firm evidence of censorship in these bodies.  

The claims and recommendations of this paper are open to debate. What is undoubtedly true is that scholarly research into academic freedom is of the highest importance.