Bad journalism, good science

The Washington Post reports that hurricanes with female names kill more people than hurricanes with male names, because people take them less seriously.  This sounds like amazing hard proof  of sexism in America… except that the data dates back to 1950, when all hurricanes were female.  They didn’t start using male names until 1979.

I tracked down the actual article, and it isn’t any better.  They brag about using multifactor analysis taking into account the gender associated with the name and minimum pressure, which I assume is a measure of severity of the storm.  What they do not account for is the year the storm took place.  So the empirical data that female names are associated with higher casualties could be explained entirely by improved building standards and warning systems.

They’re not that dumb, of course.  They did a set of lab experiments that would usually really impress me, asking people how they’d react to various hurricanes of different characteristics and varying the gender of an otherwise identical hurricane.  People really do seem to respond differently to the hypotheticals when the hurricane gender changes. I was all set to dismiss this anyway, because leaving out the detail about the change in naming patterns was so ridiculous I couldn’t trust them.  But then I found an addendum, noticing exactly my concern.  The sample showed the same pattern, with less statistical significance on account of the smaller size.  So my brilliant plan to demonstrate my intelligence by tearing others down has been foiled, and I am forced to once again confront the unpleasant reality of sexism in America.