Interesting that the changes introduce more cultural bias.

the true reason is pretty obvious: the made-over test was designed to narrow the male-female gap at the elite score level

Is there evidence (from e.g. field tests) that the new test actually has these results? The only bit I could find on subpop groups said that "Based on the pilot tests, there is no indication that subgroup performance will be substantially different on the revised General Test than on the current General Test."

How does all this relate to the over-representation of females at elite liberal arts colleges, in some cases approaching a 60/40% split. What is causing it? High School grade point averages (since girls mature faster and are more conscientous students)?

When I asked the woman in charge of admissions at Kenyon College about this problem (when she came to Atlanta) by voicing my concern as to the social and cultural ramifications of such biological imbalance, (since colleges are places where men and women interact, often find mates, etc) she reacted angrily, almost biting my head off, as if I were somehow guilty of anti-female bias.

When I asked the woman in charge of admissions at Kenyon College about this problem (when she came to Atlanta) by voicing my concern as to the social and cultural ramifications of such biological imbalance, (since colleges are places where men and women interact, often find mates, etc) she reacted angrily, almost biting my head off, as if I were somehow guilty of anti-female bias.

Maybe she was having trouble finding a mate? ;)

You increasingly see a similar thing in med school, where single female med student/resident/doctor friends of mine moan about there not being enough men around. My advice about finding a 'taxi driver with a heart of gold' never seems to go over well.

Another thing I've noticed among the high powered women I know is that the normal inter- race gender preferences are nullified. For example, East Asian male doctors, lawyers, CEOs that I know can easily date and marry white women at a similar professional level who they meet in a social circle situation. In a more middle, non-elite level, the WM/EAF:EAM/WF ratio is easily 9 to 1 in my experience. Probably an example of limited choice leading to limited hypergamy, with professional status mitigating empirically observed race preferences.

The normal rules, of course, still apply in the online and club setting for reasons that have already been discussed in other threads.

A lot of this got settled out for me years ago when a guy who'd looked into ETS (and had met a lot of people there) said to me, "Look, just remember that the people who make the SAT score lower on it than you do."

Is there evidence (from e.g. field tests) that the new test actually has these results?

Well, to be 100% certain, they'd have to release info on the male and female SD, or the % of females above 750, etc. Notice that they phrased it in a way that avoided this issue; you can only conclude that the means wouldn't change much, which is what I assume.

The inference of the post is pretty safe, though, given what was gutted. Again, although analogies are far more highly g-loaded than reading comp, males do better at analogies vs females doing better at reading, so guess which were junked?

Male advantage in analogies
(ctrl F "analogies")

Also note that the proportion of geometry questions was to be decreased -- there goes the male visuospatial advantage. Because who needs to visualize things in the sciences?

The GRE may not predict gradute performance perfectly, since you need more than just high IQ, but the GRE measures this part pretty durn well. The only changes that would be worth the trouble would make it more g-loaded, not less. They can dress up the dumbing down however they want (more "real world relevance"), but they can't get around the fact that just defining vocab words is a far more reliable way of detecting the smarties than giving them a reading comp task.

Reading comp is necessary for solid reporting; analogies are necessary to expand further into conceptual space, using our understanding of more tangible concepts as scaffolding.

Re the need to have means and SDs by sex (or in other cases, race). This info is available for a number of past years for both verbal and math SAT. ONE can't necessarily accurately compare male and female means because substantially more females took the test, which is usually an indication of an increase in weaker students. OTOH, the math scores from 2004 above 600 are male biased, with boys above750 outnumbering girls about 21K to 9 K.

The desire of elite universities to make the SAT match the world as they want it to be is certainly real. One can argue that Dick Atkinson (former Chancellor for the whole University of California system, and a former psych prof at Stanford) was the driving force against verbal analogies. His public complaint was that kids in private schools and elite public schools were practicing them in school, making this artificial and directly $$ biased.

Atkinson further argued, only slightly more subtly than I will present it, that the SAT I (verbal and math) is an IQ test and not a test of what a student can really accomplish or what he or she has learned. Thus, it is obviously not as important as something that tests what a student has learned. Thus, UC has downgraded the weight it gives to the SAT I and made up for it with heavier weighting on the SAT II (subject tests). They have also pestered the SAT people to make the SAT I more a test of learned material than of intelligence. One might infer from Agnostic's post that Atkinson was was using one prevarication (questions of bias resulting from socioeconomic status etc and the test) to hide a second unstated problem, sexual dimorphism.

The great irony in this is that the SAT I was originally designed by the elites to open up top universities by identifying students who had real intellectual potential but did not have the background (either social or via access to top K-12 schooling(socially loaded as well)) to necessarily stand out. Now, they want to get rid of the 'real' SAT since it probably does spot talent, but not of the diverse mix wanted.

One point not mentioned about the GRE: It has already been weakened. Years ago, it was just Verbal and Math. Then somewhere along the line, they added an 'Analytical' test, which I gather was based on logic problems. This came with clear 200-800 scores. Recently this has been changed to 'Analytical Writing'. This is certainly not the same as a logic problem test and obviously has a major subjective grading component, with grades in half unit increments from 1-6.

One reason the U. of California required and heavily weighted the SAT II Subject tests so heavily was to give a freebie to students who grew up speaking Spanish. They're expected to ace the Spanish SAT II, which means more can get into UC, which placates the sizable Latino Caucus in the state legislature, which has been threatening to cut UC's funding unless more of their people get in.

Many of the recent changes in the SAT have been driven by Prop. 209, which outlawed ethnic quotas in California.

From above we find:

Men have 8% more brain mass than women, even after controlling for differences in height and body mass. However, women's brains are better organized for language; women use their brains more symmetrically. These differences are not solely attributable to biology: they are partly a product of differences in experience, such as play choices in childhood that promote different kinds of brain development.

Except, of course, that play choices are clearly dictated by biology. Normal females recognize and prefer other individuals that exhibit female-oriented play styles, while normal males recognize and prefer other individuals that exhibit male-oriented play styles.

See, for example: The Two Sexes: Growing Up Apart, Coming Together

The NYT is so insensitive!

Quote 1:

In 2002, for instance, the testing service discovered that some people in China, Taiwan and South Korea had taken the test, memorized questions and answers and posted them on Web sites, allowing other students to log on and see the questions in advance.

Quote 2:

“It certainly is a further chink in their armor ...”

Isn't it going to be a bit tricky to bias a test against, for example, Chinese, if they are good both at mathematical reasoning AND memorisation?

In that Maine site mentioned above they say:


• This study finds a "substantial and systematic gender gap" in high school graduation rates. Nationally, the high school graduation rate in 2001 was 72% for female students and 64% for males. In Maine , 74% of female students graduated, compared with 67% of males.


This does not seem unreasonable given that males have greater SDs than females. We would expect to find more males below the cutoff IQ below which you cannot be expected to graduate from highschool. Possibly also because males mature more slowly than females.

Regarding the perhaps snide remark about the intelligence of people at ETS -- I used to work there and knew plenty of people there who scored 800s on there GRE V or Q. One research scientist I knew there was reputed to have an IQ of 200, and I suspect that was true both because he was the leading person in the world in his field and all I could tell from my conversations with him was that he was way our of my league.

Regarding GRE changes, I am fairly sure the primary reason they withdrew this is what they said, but I will be asking colleagues who work on the program about this at a conference next week. The bigger issue is why they computerized the test to begin with. My focus is on the why, not the computer. It appears to have been approached as a technology in search of a problem rather than the other way around.

I posted above anonymously, but unintentionally.

AJ -
Female medical students/residents claim there aren't enough men around? That sounds very odd to me. Aren't medical schools still predominately male?

"Regarding the perhaps snide remark about the intelligence of people at ETS -- I used to work there and knew plenty of people there who scored 800s on there GRE V or Q."

I took the GRE twice and got a full score both times (2400/2400 & 1600/1600). Does it mean something? I never assumed it meant much as it seemed too easy and I didn't really prepare for it(a week or so).

I have never seen myself as being exceptionally intelligent, though I manage to get by(no false modesty, this is my honest opinion of myself). I am from India and pretty much all my friends got an 800 on the Q section without preparation. This may have something to do with the intensive math training in schools in India (nature vs nurture?). The V, I admit, was somewhat more challenging.
But overall I can't see what the big deal is about the GRE. It seems like you guys take it really seriously, but it didn't appear to be a good measure of exceptional intelligence to me.

Edited By Siteowner

When you take away the discriminating power of a test, be it SAT or GRE, at the highewst levels; you basically give the discriminating power to the admissions committeees with their agenda's and biases. Not a good thing.

SOB:

There's also an incentive difference, since I think more men than women can find some kind of strong-back-weak-mind job that lets them make a living. Though this is probably harder to find every year.

FWIW, the guy I met who made the crack about scoring higher than the people who create the SAT wrote about ETS, and met a lot of people there.

After reading this passage, among numerous others such as Frey/Detterman's 2004 piece, I gave into the fact that the SAT (at least the ones I took in 1998) does, in fact, measure intelligence. It seems that when someone's score is high, he/she probably wouldn't criticize it because it was a verification of intelligence. But since my composite was a modest 1090, I just dismissed the results as 'due to nerves' or whatever else made me feel better.

Looking back, I realize that having taken one of those waste-of-time-and-money Kaplan prep courses (GUARANTEED to raise your score 150 points!) actually verified, not discredited, the whole SAT/IQ association. We tried and tried to get my math score up and it went down 30 points. For the 3 times that I took it, there was only a 10-20 point variation on each composite score. The instructor was baffled why my math didn't go up, due to the prep and the advanced math courses I'd taken throughout high school (none of which I scored higher than a 'C' in). Well, it's probably because my mathematical abilities are just 'average.' And I think the SAT proved that. And numbers don't lie either. While correlations (in this case SAT and 'g') may or may not be real in all circumstances, this, I'm sorry to say, is. I think everybody could save a lot of time and money by just acknowledging their weaknesses (but also their strengths) and accept the bottom line conclusion that this pain in the ass test measures something...and something pretty static.

And yet there is substantial circumstantial evidence that the prep courses can raise scores in many cases - and practice will even help on straight IQ tests.

Possibly there's only so much improvement possible, though, and if you've already scored at the top of your potential, no amount of practice will help.

I am disinclined to accept that explanation when dealing with people who consider themselves "bad at math". I suspect most of them have the potential and never developed it due to improper teaching methods and self-labeling.

"Good at math" is best understood within a certain environment, just as genes for greater height or intelligence. It may be that you've the ability to do some good work in math, but that your learning style isn't all that well-matched to the way many mid-level college math courses run.

My sister considered herself lousy at math, till her interests switched from anthropology to evolutionary biology to public health. Then she ended up getting an MPH in epidemiology, and becoming the person all her classmates asked for help with their complicated math problems. I think she just didn't get it intuitively until she had an interesting way to use it.

Unlike Mr. anonymous of the 2400 score, I studied for the GRE, mainly learning tricks to be faster at the analytical and memorising words for the verbal. I got 2380. I lost 20 points in the verbal. My IQ is not more than 130-4 (Stanford-Binet). I am very skeptical about the value of the GRE to discriminate between people who have IQ's ranging from 125-145. I gave the GRE more than 5 years ago. Unless someone can explain to me otherwise, I believe that people are too credulous to believe in the GRE results and I think it can be gamed.

Then there are goofball/outliers like me. I have no gift whatsoever for math. Hate numbers, despise charts, have a hard time paying attention, just wish it would all go away, marvel that anyone can enjoy and have fun with and juggle the damn things. Why don't they speak plain English instead? Yet I did pretty well on the math SAT. Riddle me that one. And no sweet-talk about how how it wasn't measuring whether I like math, it was measuring math talent and math aptitude. I truly have no math talent or aptitude.

To Caledonian: I agree, definitely. I think I hit a peak and no amount of practice, etc. would substantially change the results. I understand that many people do improve significantly on these tests with courses, etc.; however, I had read that significant improvements (ie 50+ points on each subtest) did not happen very often. It may seem like more but I think it's because these isolated cases are the ones more likely to "chime in" on a comment forum such as this to say "hey guess what!! ...."

As for self-labeling and mental conditioning, I would say yes, it's quite probable that many people had their potential sundered by external factors, such as bad teachers or influence. But for me, I was always told I was great at math and strived to do well in school. In hindsight, though, I think I was misplaced in some of those classes. I'm not "bad" at math per se, I'm simply "average". Nothing tragic. :-)

Michael -- my guess is that you took the SAT before it had a larger "book smarts" component in Math. If you're as bad at math as you say, you're like the person who scores high on the Verbal test but who couldn't write an impressive essay, is ignorant of literary devices, etc. "Aptitude" is a pretty good term -- at least as far as reasoning using numbers is concerned, you're suited to further studies in math, although you'd have to put in the time & effort to "know math." In contrast, a person with an average SAT / GRE math score would struggle with college math, even with high motivation.

Michael, my recollection of the math GRE is that the level of material was rather basic... like, I don't think there was any math on the test that was more advanced than the trigonometry I took in 10th grade.

This may explain something of your experience: if you had algebra, geometry, and trigonometry, even if you hated them, but remembered them, you'd probably do OK on the test.

I do find your assertion that you really have no aptitude rather odd. Test says you do! So either you do, or you guessed a lot and got lucky... or it really doesn't test what they think it does.