On Thursday, a British TV report about South Africa included a pretty shocking statistic. The report said that one woman or child is raped there every 26 seconds.
South Africa's problem with crime, including rape, is well known. But that number sounded awfully high to the staff at Africa Check, a non-profit group devoted to accuracy in news from or about Africa.
They asked Sky News what the deal was but they didn't get a reply.
Which, of course, made some of us even more curious.
Turns out, Sky News isn't the first media company to use this figure. The New York Times published the same statistic in a 2012 op-ed, citing Doctors Without Borders, also known as MSF. But used the wrong link.
A lot of Google searching turned up a 2009 MSF report about rape around the world.
MSF used the same source that the BBC did three years earlier in 2006. And the BBC used the same source as another article, from 2003. And before that, in 2000, the same fact turned up in a CBS report — with no citation at all.
We also found a lot of articles that called South Africa the "rape capital of the world" — and called out Khayelitsha, a township in Cape Town that's pictured here, as the place with the highest rates of rape in that alleged "capital." (Africa Check has a good debunking of that claim — and some wise words against calling any place a "rape capital.")
And then, because of the CBS story, the same fact turned up again — in an article by a law professor published in the Georgetown Journal of Gender and the Law.
Finally, we came across what appeared to be the original source: an actress you may have heard of named Charlize Theron.
In 1999, Theron — who is South African — filmed an anti-rape message called "Real Men Don't Rape." In the 45-second spot, Theron says, "Every 26 seconds a woman is raped in South Africa."
That figure was an extrapolation by a rape crisis center Theron made the ad for. There's no mention of that figure on their website today, but an archived page explains their thinking — and notes how complicated it can be to estimate rape prevalence. Extrapolations like this one often sound logical — but they're rarely considered statistically accurate by, well, statisticians.