Leifur Halldór Ásgeirsson5d ago
Rabbit Cohen
You know that thing about how men are radically more likely to divorce their wives when they fall ill than women are to divorce their sick husbands? Well, fortunately, it turns out that isn't true. The study has been retracted due to a huge error. https://retractionwatch.com/2015/07/21/to-our-horror-widely-reported-study-suggesting-divorce-is-more-likely-when-wives-fall-ill-gets-axed/
“To our horror”: Widely reported study suggesting divorce is more likely when wives fall ill gets axed

A widely reported finding that the risk of divorce increases when wives fall ill — but not when men do — is invalid, thanks to a short string of mistaken coding that negates the origina…

Retraction Watch
Jun 20 at 11:18pmWeb
Show threadRabbit Cohen5d ago
I know this was retracted a while back, but people are still quoting the retracted study like it's fact.
Show threadRachel Rawlings5d ago
@BathysphereHat People have a confirmation bias based on a sample size of Newt Gingrich.
Show threadDr Alice Violett5d ago
@BathysphereHat Boris Johnson is an outlier and shouldn't have been counted...
Show threadEl Duvelle5d ago

@BathysphereHat
To be more specific, it is still true but only for one type of disease (heart diseases):

"What we find in the corrected analysis is we still see evidence that when wives become sick marriages are at an elevated risk of divorce, whereas we don’t see any relationship between divorce and husbands’ illness. We see this in a very specific case, which is in the onset of heart problems. So basically its a more nuanced finding.”

Show threadRabbit Cohen5d ago
@elduvelle Yeah, that part is very odd.
Show threadJohn Mark Ockerbloom4d ago

@BathysphereHat @elduvelle I haven't checked the paper, but when it turns out an original hypothesis only seems to apply to a specific un-predicted sub-category, that's often a coincidence. When potentially *many* subcategories are possible in an experiment, one would expect a small number of those sub-categories to have unusually high occurrences of the hypothesized effect, simply by chance.

XKCD's "green jelly beans" cartoon is a memorable illustration of the fallacy: https://xkcd.com/882/

Significant

xkcd
Show threadRabbit Cohen4d ago
@JMarkOckerbloom @elduvelle That makes sense, thanks!
Show threadEl Duvelle4d ago

@BathysphereHat @JMarkOckerbloom That's a known effect when doing multiple comparisons though - people generally do corrections to compensate for that and surely the authors must have done that... I might check when I get a chance

Edit: I looked (at the retracted paper, it hasn't been corrected yet) and I don't really understand the type of stats they use so can't say. There is no mention of multiple comparison correction, but sometimes that's already taken into account in the statistical test.

Show threadDaniel Carosone5d ago
@BathysphereHat … in 2015!
Show threadAlexanderVI5d ago
@BathysphereHat I suspect the retraction will take at least an order of magnitude longer to make it around the world
Show threadIrenes (many)5d ago
@BathysphereHat oh, wow. good to know - we do remember when it was originally published.
Show threadRabbit Cohen3d ago
@ireneista Yeah, it was truly a disturbing conclusion, and it's a big relief knowing it's not true.
Show threadantimagic heart-sickness4d ago
@BathysphereHat just blurted out "WHAT!?" out loud at this post