The testing companies have a motive to prove a lack of bias and they use large averages. Individual colleges can do a quick regression on their own data. I’ve done this, while working for an Educational Policy Research group. The college for which I did this showed the SAT had very little power to predict who would do well in that college. It was a state university with a fairly diverse student body, albeit probably not typical of the country, because the state’s demographics aren’t. The study was never formalized or published, but I know the report was circulated internally and that a few months later, the university announced it was going test-optional. Over the years, I’ve seen news articles suggesting that other colleges have done the same thing, the same way — quietly.
I was in favor of the SATs before I did the research. But a good hard look at the data made me change my mind. While I did not look at racial bias (my remit was predicting success in the college, based on admissions data and the SAT was so useless I promptly discarded it as a predictor), my bet it that if colleges find an underprediction based on one demographic group (and there’s gender, race and hispanic/not) then they have a potential court case if it gets out. They could get sued for discriminating against other groups, because they accept the underpredicted group with lower scores. If they don’t correct for underprediction, the underpredicted group can sue, if they can prove the college knows of the underprediction. This is why I think the SAT is disappearing. On average, it might be a decent predictor, but unbiased prediction for every demographic combination in every college? I’m skeptical.
We need to find ways of identifying kids who will be well-served in college (and make it affordable for them), but using a test as bad as the SAT is not a good way to do this. The SAT subject tests are far better.
Using the Accuplacer tests would be, for the college I looked at, a pretty decent metric. Kids who could pass the Accuplacer Algebra test were much better prepared for college. The thing that makes me mad is that kids go to mediocre schools, get Bs in math and fail the Accuplacer Algebra test, meaning they have to pay college tuition for remedial math that earns them no credit towards graduation. If they wanted to do any subject that requires basic math (most of the better paying majors), they start a year behind and won’t graduate in 4 years.
If the Accuplacer tests replaced the SAT math test, then high schools would be far more motivated to ensure their students could pass it. Because the real disgrace is the education these kids get.