Thread by @mikestabile

The problem with bad studies is they get replicated as headlines with no context. Yesterday, the British Journal of Criminology (@CrimeandJustice) published a study that purports to detail the frequency of sexual violence in video descriptions. 1/ academic.oup.com/bjc/advance-ar…

According to the study, 12% of the videos on the homepage of the major tube sites show some form of "sexual violence" — a headline that's since been replicated globally. 2/

For example, this one in @thetimes 3/ thetimes.co.uk/article/porn-s…

Or this one from @BBCNews and which has since been syndicated by @YahooNews and tons others. bbc.com/news/technolog… 4/

Except the study *doesn't* exactly present a convincing case. Maybe that's why the Times doesn't link directly to it. 5/

The issue is how the researchers defined "sexual violence." In this case, the study used all sorts of common porn phrases that *don't* imply any violence. The most popular "sexual violence" term they identified was ... "teen." 6/

The also counted a video as sexually violent every time they saw "step" ("Stepbrother""stepmother" etc) or some sort of family taboo. And words like "expose" or "money" 7/

And counted it as sexually violent title used a physically aggressive terms like "spanked," "ploughed," or "pounded." 8/

Not surprisingly, these were all pretty common terms in titles. But you wouldn't know that from the coverage, which focused on words like rape (1 in 131,000 videos) and "assault" (present in 4 titles, or .00003% of the data set) that were incredibly rare 9/

Beyond that, there's no attempt to differentiate the description from the actual content shown (Was it actually forced or was it consensual BDSM — or someone using sensational text?) Was it actually her stepmother?? 10/

This is all tremendously irresponsible, both by the news outlets and the researchers! There's an infamous study from 2011 that purports to demonstrate the same thing — in 88% of videos! — that ends up in every damned article and suffers the same flaws. 11/

At a time when legislators in multiple countries, including the US, are proposing the potential censorship of large amounts of adult content, and the deplatforming of vast numbers of sex workers, this is beyond irresponsible. These results end up in ACTUAL BILLS! 12/

As much as we need media literacy for the public, we need sex literacy for researchers and scientific study literacy for journalists. It takes literally five seconds of looking at the study to identify possible flaws 13/13

Here’s another one. The disturbing part of the study is how bad it is. @jkwritesstuff @VICE

Another terrible take — this time from @DailyMailUK @jwillchad. Besides regurgitating the bad data, they also quote study authors making even broader claims that are entirely unrelated to and unsupported by their data.

To be clear, I don't blame journalists, especially not those on a general beat. No one trains them to look at data, or to question study conclusions. They're just reporting what is released — it's news! — but it can allow those with bad data to game the system.

Ah, the next stage of the Bad Study fact-laundering cycle:

Headlines that replace the study's *actual* conclusions w/ random ideas floated by previous journalists.

And words the study authors claim *suggest* violence now magically become *actual* "coercive" content.

Paging the Utah legislature.