Professional Distancing in Academia
When scholars are targeted for speech acts, what happens to their productivity and influence? A recent paper by Kris Gulati and Lorenzo Palladini addresses this question, using publications as a measure of productivity and citations as a measure of influence. The information on targeting comes from the Scholars Under Fire database assembled by the Foundation for Individual Rights and Expression (FIRE), which I have discussed in an earlier post.1
The authors argue that the targeting of scholars for speech acts reduces both productivity and influence, lowering the number of published papers by one-fifth and citations to work published prior to the controversy by four percent. These are claims about causal effects—cancellation is seen as a treatment that leads to lower output and impact.
Since the assignment of scholars to control and treatment groups is not random, the attribution of causality requires some ingenuity and care. To identify the impact on productivity, one needs to construct a control group of scholars with initially similar publication profiles, to see if targeting leads to a divergence in the output trajectories of one group relative to the other. To identify the impact on influence, the two treatment and control groups must be composed of publications rather than scholars. The manner in which these groups are constructed (described below) is creative and compelling.
First consider productivity. Here the authors show that the output of treated scholars drops after targeting relative to that of a group of matched scholars with initially similar publication counts. Ideally, one would want a control group that is composed of scholars who risked but evaded targeting, but this is not really possible with the available data. That is, the control group was populated by scholars who either did not hold the controversial opinions expressed by those in the treatment group, or engaged in strategic self-censorship to avoid sanction. This might lead some to question the claim of causality, though personally I find it intuitive and plausible. Being targeted for speech acts is distracting and stressful; it would be surprising if productivity did not decline in response.
Now consider influence. Here the unit of observation is not a scholar but a publication—the treatment group is composed of papers by targeted scholars that were published before the targeting incident took place. This is important, because papers written after the incident may have suffered from a decline in quality (for reasons similar to those responsible for the decline in productivity). The control group is then constructed by randomly selecting articles that appeared in the same journal and issue as the papers in the treatment group. This controls for paper quality in an extremely convincing manner.
The authors show that articles in the treatment and control groups have similar initial citation profiles, but a gap opens up once a scholar is targeted. Furthermore, they find that this reduction in citations comes primarily from professional distancing by those who are close to the targeted scholars, where closeness is determined by proximity in a co-authorship network or by shared institutional affiliation. They interpret this finding as follows: “close scholars likely face greater reputational risks from continued association and may face pressure to distance themselves from the… scholar who is embroiled in a scandal. By reducing citations to controversial colleagues, these scholars can signal professional distance while minimizing potential career costs.”
If I were reviewing this paper for a journal, I would ask the authors to consider the following two points.
First, while the effects on scholar influence are negative on average, the size and sign of these effects may be sensitive to the speech topic and the political direction from which the targeting arrives. There has been a sharp change over the past couple of years along both these dimensions. A few years ago speech related to race and gender was most likely to be sanctioned, and scholars were targeted primarily from the left. More recently, it is speech related to Israel/Palestine that is most frequently targeted, with attacks coming predominantly from the right.2 It is worth exploring whether the causal effects of targeting on influence are stable across topics and the source of attack.
The second point is related to the broader implications of the study. The authors point out, correctly, that “freedom of speech plays a crucial role in the pursuit of knowledge.” But the sanctioning of scholars is not the same as the suppression of ideas. Attempts at censorship sometimes backfire, resulting in heightened attention and more robust debate. Ideas that are initially suppressed may continue to smoulder and burst into the open when the time is ripe. Some of the most severely punished scholars eventually joined the ranks of the most influential. Socrates and Galileo come to mind. This is cold comfort for the sanctioned scholars, of course, since they suffer severe reputational and material harms. But in the long run, people are easier to suppress than knowledge itself.
