Guest Post: How Statistics on Anti-Semitic Tweets might Mislead – Justin Tiehen

“Out of 2.6 million anti-semitic tweets sent from Aug 2015-July 2016, 70% came from just 1,600 accounts”. So Ezra Klein tweeted recently, with a link to Mathew Yglesias’s Vox.com discussion of a new report from the Anti-Defamation League containing this finding.

ezra-klein-tweet

According to some, the inference to draw is that while the surge in anti-Semitism is troubling and not to be minimized, the fact that such a high percentage of the offending tweets was written by such a comparatively small group of people (just 1,600) is encouraging. It suggests that those writing anti-Semitic tweets are “not exactly a mass social movement” according to Yglesias. Julia Galef, cofounder of the Center for Applied Rationality, described the result as “good news”.

But I didn’t find the statistic very informative. For one thing, I have almost no intuitions at all as to whether the 70%-1,600 statistic is unusual. How few accounts are responsible for 70% of all Seattle Seahawks tweets, or 70% of all Ghostbusters tweets, or 70% of all Libertarian Tweets? Is the 1,600 figure for such topics unusually low, about average, or fairly high? Without such comparisons, I have no idea what to infer.

Or, cutting more deeply, imagine the following. Start with a scenario in which there is a distribution of anti-Semitic tweets you regard as troubling—I’ll leave it up to you what the distribution is exactly. Now add to the scenario a single person, a remarkably proficient anti-Semite. This anti-Semite matches the total quantity of anti-Semitic tweets written by all other people, and then writes one more. Has adding this proficient anti-Semite made the scenario better, made it less troubling than it started off? After all, it’s now the case that more than 50% of all anti-Semitic tweets are written by just a single person, and one person is “not exactly a mass social movement,” in Yglesias’s words. But this is absurd: surely we haven’t made the scenario better in this way.

More realistically, a scenario in which lots and lots of people are writing occasional anti-Semitic tweets while a few people are writing tons of them (70% worth) doesn’t seem like much of an improvement on a scenario with the same overall number of people writing anti-Semitic tweets but with a more equal distribution. If this is right, the 70%-1,600 figure seems like the wrong way to try to get a handle on the extent of the problem.

All that said, I do find somewhat plausible the hypothesis that the recent upsurge in Twitter anti-Semitism is mostly due to a limited group of people. Even prior to looking at any serious evidence, I think there’s a decent chance this is true. My claim is only that when we do look at the serious evidence, I would like to have something more informative than the statistic in question.


Justin Tiehen is Associate Professor of Philosophy at the University of Puget Sound who works primarily in the philosophy of mind and metaphysics. He also maintains an impressive list of Trumpsplanations on Twitter.

Advertisements

One thought on “Guest Post: How Statistics on Anti-Semitic Tweets might Mislead – Justin Tiehen

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s