Gender Bender

The AAUW spin sisters strike again

Agenda-driven politicized junk social science at its worst.

"Thus, while discrimination cannot be measured directly, it is reasonable to assume that this pay gap is the product of discrimination."

Oh, baloney.  And to start that sentence with "Thus" is simply hilarious -- as is the preceding confident assertion that "all the [other] factors known to affect earnings" have been accounted for.  The problem, of course, are all the things you didn't know that you didn't know, or suspected but had no way to measure. Assertions about having accounted for "all the other factors" is pure bilge.  (See explanations of global temperature change that do not include, for example, the activity of the Sun (!) as a causal factor. The mind reels.)

We're deep in follies of regression analysis here:  Attributing the residual, unexplained variance to your preferred cause, which you admit you can't measure.  It's hanging your pet name on an unknown. OK, instead of calling it "sex discrimination," I'm going to call it "Timmy."

But what you certainly do have is plenty of measurement error (random, systematic, and everything in between) in many or most of your measures, some or many of which will be self-reports of some kind. Fortunately, people have perfect memories and never fudge the truth -- and I am Marie of Romania.

Problem is, measurement error lowers correlations, and that creates unexplained variance.  Until somebody demonstrates otherwise, I'll assume "Timmy" is just another name for measurement error.

Byron