Monday, January 10, 2011

How A Nonexisting Gender Study Got Its Wings

I have followed write-ups of Catherine Hakim's nonexistent new study all over the UK mainstream media.

What's astonishing about the write-ups is that very few popularizers seem to have actually read Hakim's sermon/diatribe/essay to notice that she hasn't made any new study at all! Or any new survey!

What's equally astonishing is that people appear to have turned their logic sections off while reading some type of a press release. If, indeed, it were the case that more and more women marry for money (money!!!, even though Hakim looked at only education data), then what happens to all the men without money? We should see pronounced statistical hints of all this in the form of low-education men not getting married at all, for instance.

That something of this sort is eagerly publicized does tell us that anything which supports traditional sex roles as ingrained gets an easier pass to popularization than alternative kinds of studies. A nonexistent new study is exciting to talk about, actual existing new studies are not.

What is driving this? Mostly the bias of people "knowing" what they "know." There's a very strong prior belief bias among the audience of these pieces, created by the common popular myths of women as gold-diggers and/or frustrated housewives or nasty feminist harpies.

One example of the blinders that we all put on when it comes to gender comes from the Sunday Times article* which on the whole discusses Hakim's non-study quite well (including pointing out that it got rejected by an academic journal, probably because it's not a new study). Yet the survey data the article quotes is based on only surveying women:

Most mothers would prefer not to have the competing demands of family and job As the arguments rage, a YouGov survey for The Sunday Times suggests Hakim may have a point. It reveals that 64% of women of all political views, ages and locations, would have preferred to marry a man who earned more than them. But only 19% want a better educated man (against 62% who sought men with the same sort of education). More women (31%) think they are better educated than their spouses than vice versa (19%).

Asked whether, if money were not a worry, they would prefer to stay at home with their children, 55% said yes. And 53% agreed that society puts pressure on women with children to go to work.
Nothing wrong with using such data, EXCEPT as proof of a gender difference. When there's no data on how men might answer these questions, the women's answers cannot be used as proof of an existing difference between men and women.

But they are. Thus, we do not learn what percentage of men would stop working if money wasn't a worry**, for example, or what percentage of men might wish to marry a woman who earns more (though given that women still earn less, on average, than men, that question probably cannot be presented to men stripped of that context).

I doubt that the use of this study as somehow supporting Hakim's arguments is an example of an overt bias. It's much more likely that we are so utterly accustomed to juxtaposing family and work for women while completely ignoring family when it comes to men that the presentation of data on only one gender doesn't strike us as odd, even when it is used to make inferences about a gender difference.
* I had to pay for access to that one.
**That 45% of the women appear to have stated that they would continue working if money wasn't a worry seems a lot more surprising to me.