What's a testing opponent's worst nightmare? The use of the SAT by employers after a potential employee has graduated from college. The Volokh Conspiracy notes that the Wall Street Journal has reported on this new trend; some employers are allegedly requiring the submission of SAT scores along with a college transcript for entry-level positions.
I don't have a subscription to the online version so I can't reach the original article, but here's what The Volokh Conspiracy - and one reader - had to say:
SATs For Life? My colleague Lloyd Cohen passes along these thoughts:
------------------
A most interesting article in yesterday's Wall Street Journal. Apparently, there is an increasing trend among employers to ask applicants for their SATs long after graduation from college. I find this heartening. The SAT is fundamentally an IQ test. While not the only measure of likely productivity on the job, intelligence is probably the most powerful and robust predictor. In the past college of attendance, major, and grades, while always subject to unreliability, were more powerful indices of both intelligence and other productive inputs than they are now. As these other predictors have become more debased it is good to see that the market is responding.
--------------
Precisely because of its value as an IQ test there are those who wish to transform the SAT into something else, and wish to narrow its employment. One wonders how--if they succeeed--the market will respond.
Of course, the SAT does have some limitations as an IQ test, not least that taking a preparation class can boost results, and that language skills for the Verbal section will be affected by the quality of English spoken in one's home. But in the push to raise high school graduation rates and college attendance rates, graduating from high school or even attending college is no guarantee that a job applicant has basic literacy and math skills, much less advanced ones. It's not surprising that employers are looking for substitute signals.
Not surprising, but that doesn't mean that the SAT is necessarily the best test for the job. It is related to intelligence, and it does predict first-year college grades. However, that doesn't mean that it is necessarily predictive of success on the job once one has a college degree, given that most of those with college degrees will have a more limited range of SAT scores (limiting the range of a variable necessarily limits the correlation of that variable with other variables). Any company wanting to use SAT score as any sort of selection variable would do well to collect data showing that SAT scores were in fact correlated with productivity, or performance reviews, or some such measure of job "success" in that environment.
What's more, if employers plan to use the test for hiring or promotion purposes, they're going to come under fire from the same people who use the score gap as a means to bash objective college standards. Colleges already feel the pressure to lower standards for minorities, allegedly because the test is "biased." Companies who use the SAT will most likely see the same score gaps (it would be interesting to know if they didn't), and they'll suffer the same pressure to not use the test because of those gaps. I wonder if they have any idea of the abuse they'll suffer if they turn down someone with a decent college GPA who had a low SAT score and was admitted to college under affirmative action.
Did anyone see the original article, or have a subscription to the online version?
Update: Apparently it's not in the online version. One helpful reader who did read the article mentioned a standard of 1350 as the SAT score for at least one company. I'd be very surprised if companies were using scores this high as absolute cutoffs. As this chart shows, a 1350 is at the 93rd percentile. Any company using that cutoff is almost certainly ruling out some qualified applicants.
What's more, take a look at this chart. It doesn't give composite percentile ranks, but we can easily come up with an example that illustrates the problem. Let's say the most likely combination of Verbal and Math that produces a 1350 score is 700 Verbal, and 650 Math. Let's compare the percentiles by score and race:
5% of White examinees score above a 700 in Verbal, while only 1% or so of black or Hispanic examinees do so.
19% of white male examinees score above a 650 in Math, and 11% of white females do. However, only 2% of blacks and 4% of Hispanics do this.
See the problem? The score gap exists throughout the SAT score scale, but at the high end of the scale, the number of black and Hispanics examinees drop off dramatically. Any company that requires a 1350 SAT score for hiring purposes is in effect doing what the College Board tells colleges not to do, which is the use of a strict cutoff score so high that it will eliminate a large proportion of minorities. Companies that use this high a cutoff won't have to waste much money on "diversity training", because they won't have that diverse an employee population.
Companies are, of course, perfectly free to make the decision to use the SAT, but I hope they're prepared for the lawsuits to come.
Posted by kswygert at October 29, 2003 01:34 PM