I’m skeptical of this story. Even taking for granted that this was when the test was still normalized to 1600 as the max, if one looks at even a mediocre state school a total of 790 would be clearly in the very bottom. Note that in this data, the bottom 1 percent for both is slightly over a 400 for both sections. So someone scoring in that range is possible but extremely unlikely. This is around the 15th percentile for anyone taking the test, but the very bottom don’t generally go to real colleges at all.
The average SAT score for a men’s basketball player at that school is 916, for football it is 926, over 250 points lower than the average of non-athletes. Consider that there are about 100 football players per school, and not all excel at athletics enough that admission departments change their standards for them equally. If 50 of them average 1050 (about bottom 20th percentile), the other 50 would have to average 790 for the average for all of them to be as low as 920. If 90 average as high as 940 (about bottom 5th percentile), the other ten would have to average 790 for their collective average to be 925. A single student, who might or might not only be a marginal football player, who scored 1140 (not an outlandishly high score, 40th percentile at that school) would raise the football average about two points.
Considering that average football player SAT scores are tracked and schools desire their admissions standards to be perceived as high, both as part of the NCAA certification process and to justify their money-making programs, Goodhart’s law should probably be applied an additional time. Not only are SAT scores imperfect proxies for intelligence, average SAT scores for a sport are imperfect proxies of their admission standards, which are probably even lower than implied. This means it is very likely that some individuals have far less than the average program SAT score.
I’m skeptical of this story. Even taking for granted that this was when the test was still normalized to 1600 as the max, if one looks at even a mediocre state school a total of 790 would be clearly in the very bottom. Note that in this data, the bottom 1 percent for both is slightly over a 400 for both sections. So someone scoring in that range is possible but extremely unlikely. This is around the 15th percentile for anyone taking the test, but the very bottom don’t generally go to real colleges at all.
The average SAT score for a men’s basketball player at that school is 916, for football it is 926, over 250 points lower than the average of non-athletes. Consider that there are about 100 football players per school, and not all excel at athletics enough that admission departments change their standards for them equally. If 50 of them average 1050 (about bottom 20th percentile), the other 50 would have to average 790 for the average for all of them to be as low as 920. If 90 average as high as 940 (about bottom 5th percentile), the other ten would have to average 790 for their collective average to be 925. A single student, who might or might not only be a marginal football player, who scored 1140 (not an outlandishly high score, 40th percentile at that school) would raise the football average about two points.
Considering that average football player SAT scores are tracked and schools desire their admissions standards to be perceived as high, both as part of the NCAA certification process and to justify their money-making programs, Goodhart’s law should probably be applied an additional time. Not only are SAT scores imperfect proxies for intelligence, average SAT scores for a sport are imperfect proxies of their admission standards, which are probably even lower than implied. This means it is very likely that some individuals have far less than the average program SAT score.
That’s an excellent set of points. I clearly underestimated the chance of such an event occurring.