Friday, March 07, 2014

SAT And The Dropped Essay Requirement. Gender Implications.

The Washington Post tells us that the essay requirement, which was only fairly recently added, will now be made optional:

The SAT college admission test will no longer require a timed essay, will dwell less on fancy vocabulary and will return to the familiar 1600-point scoring scale in a major overhaul intended to open doors to higher education for students who are now shut out.


Out, too, will be a much-reviled rule that deducts a quarter-point for each wrong answer to multiple-choice questions, deterring random guesses.

Whatever the actual reasons for these changes -- and they may be good ones, who knows -- there's an outcome which is not good for female students taking the test.  Have a look at the tables showing the 2013 scores on the written test, on the critical reading test and on the mathematics test.  You might check both the 99th percentiles, the top scorers, and then scroll down to the means and standard deviations on the tests.*

By making the essay an optional one, the College Board is keeping the tests where the male mean is somewhat higher than the female mean compulsory, while making the test where the female mean is somewhat higher than the male mean optional.  A similar impact could be created by not deducting anything for guessing if men are more likely to use that strategy than women**.

Removing the essay requirement also means that those who use the upper tail theory about tests to explain why there are fewer women than men in positions of power don't have to try to explain away the writing test results, because they will no longer be there to be cited in that debate.

*Note, also, that more women than men take the test, and this matters for all the means in those tables I link to.  From an older article (before the essay requirement was added) about this effect:

Gender Differences. The number of women taking the SAT has exceeded the number of men for more than two decades. In the class of 1996, there were 75,529 more women than men taking the SAT. These "additional" women were less likely to have taken rigorous academic course work than other students who took the SAT. In 1996, average SAT verbal scores were 503 for women and 507 for men. This four-point difference is what one might reasonably expect given that there are so many more women taking the test. In fact, research has shown that the differences in the self-selected population taking the test explains the difference in average verbal scores between men and women. If equal numbers of men and women took the SAT, the average verbal score for women would actually be somewhat higher than the average verbal score for men.

**My apologies for getting that one wrong.  Removing the penalty for guessing may benefit women if the results of that study generalize.