Two related, long-standing issues that have come to the surface in the disputes over the Colorado report are how much sexual harassment is going on and how much it should be widely publicized. For instance, everyone seems to agree now that there were fifteen reported cases of harassment in Colorado, of which two were substantiated, and that lack of substantiation does not entail lack of truth. But there the agreement ends. At least a few people seem to feel that we should therefore assume that something like two cases took place, and many others appear to assume that all fifteen accusations, and more, were probably accurate.
Moreover, both sides of the issue seem to rely on different background assumptions in support of their views. Those who would err on the high side justify their claims by reference to the fact that ‘everyone knows’ that harassment levels in philosophy are generally very high, while those on the other side have a corresponding skepticism about this. And as far as I’ve seen, those who believe in high harassment numbers tend to be the keenest on promoting public and general discussion of the harassment problems.
This is an important issue because serious problems come with a significantly mistaken assessment, whether it’s too low or too high. Many problems that could stem from a significant underestimation are well-known:
- the harassment problem may not seem as serious as it is, leading to systemic toleration;
large-scale collective action against harassment might not be motivated when needed;
some female philosophers, frustrated by inaction in the face of harassment, may choose to leave the discipline (which is bad in itself, but in turn worsens the discipline for everyone who remains behind); and
women might join the discipline or a particular department without being adequately warned about a real harassment threat, which neglects their safety.
However, all actual and potential philosophers, including and in some ways especially women, can also be harmed by a significantly excessive estimation of the problem:
- women who have not faced any harassment but are given an significantly inflated sense of the extent of the problem might pre-emptively switch out of philosophy to avoid it;
women who have a greatly exaggerated sense of harassment (after reading the Slate and Gawker pieces, say) might reasonably choose not to take philosophy courses in the first place;
valuable time, energy and resources that could otherwise be devoted to the promotion of philosophy might be diverted to dealing with problems whose actual scale would demand a different sort of solution, and the proposed solutions may needlessly hamper the philosophical culture of a department;
female philosophers who erroneously see their environment as hostile will likely feel considerable stress as a result;
an excessive reinforcement of taboos against the sometimes nebulous actions that can constitute harassment can make social and professional interactions awkward and stressful for all parties;
an excessively high estimation of the extent of harassment might well lead to the accusation and censure of innocent parties; and
a general sense that reports of harassment levels are inflated can lead to a ‘crying wolf’ effect, against the interests of the subjects of harassment.
Since bad results follow from inaccurate judgments on both the high and the low side, there is no side of caution to err on. We need to aim for accuracy. But here the real problems begin.
If we accept only reports of harassment that have been substantiated, then we will miss all the cases that are not reported and the cases that did not come with adequate evidence. Since harassment (I presume) typically occurs in private, one would imagine that many genuine cases of harassment cannot be established on the available evidence. So we cannot rightly restrict our estimation to substantiated cases.
On the other hand, we surely cannot just assume that all or most accusations are correct. Nor does it seem right, as I’ve often heard suggested, to base our assumptions on a reading of the ‘What It’s Like To Be A Woman In Philosophy’ blog. Whatever the merits of that blog in facilitating discussion of some of the worst cases of harassment — and I have no wish to dispute those merits — it would be ludicrous to treat its contents as an accurate survey of the extent of harassment in the field, nor does it seem intended to play that role.
In case the reasons for this are not obvious, please imagine a parallel case. I enjoy riding my bicycle to work and have never had any problems doing so. But someone recommends that, to be safe, I should look at the findings of a blog called ‘What It’s Like To Be A Cyclist’. Whatever its founders’ intentions, this blog serves mostly as a repository of terrible cycling stories. Tales abound of being cut off in traffic, skidding in puddles, having one’s bike stolen, and so on. A number of stories detail visits to the hospital after serious bike mishaps and even the death of fellow bicyclists in accidents. The blog makes no attempt at a comprehensive survey of cyclists, but just accepts whatever stories are submitted; and this of course greatly biases the results toward negative cycling experiences. Moreover, the reports are sent in anonymously and posted without verification or the ability for readers who might have witnessed some bike accidents to add skeptical details, so there is no safeguard against exaggeration or a single, possibly somewhat unhinged, commenter submitting hundreds of stories. As I say, a blog like this may have several legitimate purposes; but as a source of objective information about the advisability of my riding my bicycle to work, it is simply not the correct tool for the job. I might easily come away with the impression that all bicyclists run into serious problems sooner or later when in fact only 10% do and, in my city, only 2%.
Finally, splitting the difference between the highest and lowest estimates commits the fallacy of the mean and should be avoided for all the familiar reasons (a vast skewing of one side’s view of the data will also shift the mean-point significantly, and so on).
So what should we believe, and what should we do?
(I realize that this problem may already have been addressed by a survey that would pass basic social sciences protocols. I just haven’t seen it and I’m a little skeptical about its existence due to the widespread appeal to the ‘What It’s Like…’ blog. If you know about such a survey, please post the details for us.)