News for hackers. Stuff that matters.
Re:SAT Scores Fall as More Students Take the Test by rjkennedy98 on Thursday September 26, 2019 @09:14AM (#21080603)
Is it all because more students take the test? Reports of falling IQ scores in many developed countries have been pretty well documented. Perhaps it is starting to happen in the US also.
Re:SAT Scores Fall as More Students Take the Test by skolos on Thursday September 26, 2019 @09:24AM (#21080670)
Could you please provide references to these "pretty well documented" studies?

Thanks.

Re:SAT Scores Fall as More Students Take the Test by losvedir on Thursday September 26, 2019 @09:34AM (#21080736)
I believe they're referring to the reversal of the Flynn effect[0].

[0] https://en.wikipedia.org/wiki/Flynn_effect#Possible_end_of_p...

Re:SAT Scores Fall as More Students Take the Test by kypro on Thursday September 26, 2019 @01:09PM (#21083012)
https://www.gwern.net/docs/iq/2016-dutton.pdf

https://www.nbcnews.com/think/opinion/iq-rates-are-dropping-...

This is fairly well documented. Average IQ scores in France have been declining a couple of points per decade if I recall correctly...

I believe mass immigration from low IQ countries and dysgenics are often sourced as two of the potential reasons. But I have no idea how true that is.

Re:SAT Scores Fall as More Students Take the Test by pmart123 on Thursday September 26, 2019 @09:24AM (#21080667)
I believe IQ scores are rising still for lower scores, but falling for the top 1-2%?
Re:SAT Scores Fall as More Students Take the Test by gwern on Thursday September 26, 2019 @01:25PM (#21083189)
The very most recent US Flynn effect study finds the opposite pattern, actually (falling for low end, rising for high end): https://www.gwern.net/docs/iq/2019-platt.pdf

Personally, I don't buy it. The Flynn effect is large enough, and the way the normal distribution works with tail effects, that if there were any Flynn effect in general (much less focused on the high end), that you should see dramatic effects like several-fold increases in the number of perfect SAT scores. You don't see anything remotely like that - even despite loads of countervailing factors like increases in SAT access and prep!

Re:SAT Scores Fall as More Students Take the Test by SamReidHughes on Thursday September 26, 2019 @01:38PM (#21083309)
Intelligence isn’t normally distributed though.

E: And even if it was, and was capable of “shifting” left or right, which is a nonsensical notion about populations, a shift of a few IQ points wouldn’t have a dramatic many-fold increase in perfect scores.

Re:SAT Scores Fall as More Students Take the Test by gwern on Thursday September 26, 2019 @04:29PM (#21085173)
You're right, intelligence, like anything else (eg height), is not exactly normally distributed: it has fatter tails than a normal distribution - so a shift in the mean would increase the density at a high threshold even more than you would calculate based on a normal.
Re:SAT Scores Fall as More Students Take the Test by SamReidHughes on Thursday September 26, 2019 @07:09PM (#21086611)
No, a fat tail would do just the opposite. A fat tail is where more points are far from the threshold.

But it's a moot point because an increase of 2 points among people in the 99th percentile implies little about those in the 99.99 percenteth quantile.

Some description of a fat tail from 60-99 percentile doesn't say much about the top one percent. At a certain point, the dating market tops out.

Re:SAT Scores Fall as More Students Take the Test by gwern on Monday September 30, 2019 @09:41AM (#21115217)
No, it would. Simulate it at a threshold like 1600 scores. The steeper increase in that part of the range yields the same phenomenon as normals.

> (Also, intelligence/aptitude/performance scales like IQ, SAT, or ASVAB scores are arbitrarily numbered, so I'm don't know what a "fat tail" means in that respect. Maybe there's a standard definition that isn't completely arbitrary.)

Er, you can look at plenty of cardinal-scaled things like vocabulary or digit span. Doesn't change the non-normality at the tails.

Re:SAT Scores Fall as More Students Take the Test by pmart123 on Friday September 27, 2019 @12:28AM (#21088427)
I believe it is the same case with height.

https://www.johndcook.com/blog/2008/07/20/why-heights-are-no...

It's close to a normal distribution one or even two deviations away from the mean, but real-life has way more samples appearing +5 deviations away from the mean. I haven't really looked into it ever, but this could be because of a positive skew versus a distribution with purely fat tails?

Re:SAT Scores Fall as More Students Take the Test by SamReidHughes on Friday September 27, 2019 @02:07AM (#21088935)
The difference is, it's easy to make a long-boned freak or a blithering idiot. But there's no hormone (I know of) that makes you a genius.

(Also, intelligence/aptitude/performance scales like IQ, SAT, or ASVAB scores are arbitrarily numbered, so I'm don't know what a "fat tail" means in that respect. Maybe there's a standard definition that isn't completely arbitrary.)