But I'm amassing beaucoup SAT factoids, so I'll put some up as I go.
For starters, here are the basics on the 16-year decline in SAT scores:
- SAT scores declined from 1963 to 1980
- Verbal mean: dropped from 475 to 425
- Math mean: dropped from 500 to 470
- By 1990, the Math mean has gone back up to 475, but Verbal mean has not budged
- By 1980 College Board perceives a “definite need” to “realign the verbal and math scales”
- SAT V and SAT M averages are 50 points apart
- “there was a clear need to repopulate the top end of the score scale, especially for SAT V”
And get your SAT I Score Equivalents right here:
CONVERT INDIVIDUAL AND MEAN SCORES FROM THE ORIGINAL SCALE TO THE RECENTERED SCALE
I've seen that converter before and have never figured out how I lose 10 points on the math under recentering (I would drop from a 700 to 690). I would go from a 750 up to a perfect 800 on the verbal, but I'm not sure that would've been enough to make up for a sub-700 score on the math.
ReplyDeleteThe specialness of my 800 in math, obtained in 1981 when scores were at their lowest, is diluted under recentering since a bunch of formerly lesser scores are now considered to be equivalent to 800. At least it's not as bad as the situation for verbal, where an old 730 is now an 800.
ReplyDeleteThe SAT is a horrible test for distinguishing the people on the right tail of the curve, even more so after the recentering, especially when the problem is compounded by today's massive amounts of test prep.
Yes; even in 66, when I took the SAT as a junior, my verbal 790 was pretty special - now 730 has been magically transformed into 800. I knew kids from all over my state, from my extracurriculars, and I never heard of anyone who did any test prep, let alone taking a prep course. I don't even remember seeing prep books in libraries and bookstores - and I spent LOTS of time in libraries. Perhaps it was different in the affluent areas of big metros, but many of my college classmates were from those areas and prep was never mentioned.
ReplyDeleteWhat is saddest about this is that there was NO SUCH NEED TO REPOPULATE THE HIGH END OF THE SCORE SCALE.
ReplyDeleteRealistically, the bottom half of the scale is pretty unimportant for college admissions: The population that scores much below a 1,000 on M+V is, on average, very much *NOT* ready for college (and probably never will be). If the whole point was to get more useful signal, then re-centering so that the new average was 400 on each of the sub-tests would have made a lot more sense.
I'm pretty sure that the math folks who do this for a living know this, so there has to be a reason that they chose to make the tests provide *less* information.
-Mark Roulo
@ Mark, they didn't want to create too much backlash. If confronted with the reality of how badly scores had fallen, people would have started to protest against the SAT en masse. They'd much rather keep their blinders on than deal with how poorly kids are being educated. And yes, it's very, very hard to work with someone scoring below 500. The skills are missing at so many levels and from so many angles that I can barely begin to even deal with what the test is actually testing.
ReplyDelete"@ Mark, they didn't want to create too much backlash. If confronted with the reality of how badly scores had fallen, people would have started to protest against the SAT en masse."
ReplyDeleteBy changing the test the way they did, they had to deal with the backlash anyway: "Hey, the scores have dropped over time, and we want to raise the scores back to the old average (without being able to raise the performance)."
They could just as easily have said: "The average is down, but there are so many *good* students at the top end that we want more range to show how well they are doing." It would be a lie, but so what? The re-centering was already misdirection, if not an outright lie.
*Or* they could have gone to a wider range while re-centering: 200 - 1000. Move the new average to 500, but add 200 points on the high end [NOTE: Better would have been to go to a 20 - 100 or 2000 - 10,000 range to make it harder to mix pre/post re-centering scores. One of the better computer benchmarking organizations does this when they revamp their benchmarks].
-Mark Roulo