Bad Methodology and Bad Reporting Mar Trump Survey

A rule of thumb for researchers: If you create a super-smart algorithm to determine, say, the best movie of 2015, and you come back with Mortdecai, that might be a sign that there’s something wrong with your research methods—not that the American people inexplicably failed to appreciate the genius of Johnny Depp mugging for the camera with a pencil thin moustache and “hilarious” British accent. Likewise, if you design a study to find the most pro-Donald Trump colleges in America, and your “research” yields that those campuses include such redoubts of right-wingery (just kidding!) as Scripps College, Wesleyan, and Hampshire, there might be something screwy with your methods.

Yet when uCribs, a website that helps college kids find apartments, released a survey that found just those results, some in the media ran with it. The website of the Pulitzer Prize-winning Oregonian
newspaper, for example, credulously ran an article announcing which “Colleges love . . .Trump the most.”

The uCribs methodology was laughably bad:

We used data from Facebook’s advertiser targeting platform to analyze both the number of students on Facebook who are currently enrolled at each college, as well as those who have indicated interest in Trump on the social media network. In our final tally, we used some discretion in removing schools with limited available data and also removed some small and specialized universities to create a list of mostly recognizable colleges on both ends of the spectrum.

In other words, that guy from your Media Studies class who is always posting online about how Trump is “just like Hitler” was counted as a Trump supporter by uCribs. (Oh, and uCribs also summarily threw out a bunch of colleges.)

Stand by for uCribs’s next blockbuster report, which will find that Trump’s biggest celebrity supporters include Susan Sarandon, Mark Ruffalo, and Cher.

Related Content