Tuesday, September 21, 2010

The Reverse Gender Gap in Wages Among Young Workers

This article completely slipped past me, and it should not have done so, because it's such an excellent example of a subtly biased presentation of research findings.

Its major thesis is that a certain group of women is out-earning their male contemporaries: Women between the ages 22 to 30 who are single, childless and who live in large metropolitan areas earn more than men between the ages 22 to 30 who are also single, childless and who live in large metropolitan areas.

Note what is held constant (standardized) in that comparison and what is not:
1. Age is held constant. Older workers are excluded.
2. Marital status is held constant. Married workers are excluded.
3. Parental status is held constant. Workers with children are excluded.
4. The level of urbanization is held constant: Workers not living in metropolitan areas are excluded.

All of these choices work in one direction, by the way. The gender gap has always been the smallest among young workers, among single workers, among childless workers and among workers who live in large metropolitan areas. These are exactly the standardizations I would choose if I wanted to prove that women earn more than men, in at least some groups!

Then note what is NOT held constant: Education. In fact, we have no idea if we are comparing men with only a high school diploma to women with a college degree, say. The proper analysis would not just add a paragraph telling us that the reason for these difference is most likely the higher levels of education young women in metropolitan areas have (as this article does).

The proper analysis would compare workers who are young, single, childless, metropolitan AND who have a particular level of education: less than high school, high school diploma, an undergraduate degree, a post-graduate degree and so on.

By not doing that the article fails to show that the women it studies actually earn more than men when all the relevant variables are held constant.

Then, of course, the article does state this:

The shift in earnings power started showing up in a few big cities a few years ago and has become widespread. It isn't true for all women in their 20s working full time — overall, they earn 90% of what all men in their 20s make — just for those who don't marry or have kids.

Bolds are mine.

Note the way this article has been constructed. The finding that women in their twenties working full time earn 90% of what men in their twenties working full time do is hidden in the middle of the piece!

That's pretty biased, in my view. Then note that one could have done a completely different article about the gender gap in earnings between young workers by only looking at those who are married and who do have children (the exact opposite of what this article chose to do). If the overall gender gap is 90% and if single and childless women indeed earn more than their male counterparts, there must be pretty sizable gender gap in favor of men among those who are married and/or have children.
ETA: It is very important to look at the gender gap in wages after taking into account all the other variables which affect earnings. If we don't do that we are not talking about a difference which is directly about gender. Both sides in the political debates make that mistake far too often as I point out in my gender gap series (available at the website given at the top of this blog).