Or, as Stuart Richie says:
Cool new PNAS paper about potential educational effects on IQ.
Previous work: control for age-11 IQ, still find edu-IQ correlation later in life.
This paper: control for age-20 IQ, correlation is gone. Suggests education has limited influence after age 20.
IQ measurement doesn’t interest me much, but IQ development or change over time does, for ego-driven reasons. As long-time readers know, I have a very high IQ (I qualified and participated in a research study for 3SD+), but my spatial abilities are very weak and I was stymied in advanced math (past algebra 1) until, in my early 40s, I learned how to compensate using logic. I also was late to learn how I learned; my brain won’t acquire new information unless it’s tagged with all sorts of meta-data. Learning new concepts was so laborious that in my teens, I simply assumed I was incapable of learning; not until I was faced with job-related challenges did I learn how I learned. My verbal skills are extraordinarily high, although it’s hard for me to compare to others because my particular combination of smarts would have required a more thorough classical education, which I don’t have. I read 1000 wpm and can acquire extraordinary amounts of information through inference, which of course can sometimes lead me astray.
So. In 2001, I took the GRE and got 790V, 640Q, 690A. It was the last time the Analytical section was included. 640 quant was the 65th percentile that year, 690 analytical was in the 85th percentile–some logic games are brutally spatial. Anything over 700 Verbal is in the 99th percentile. I was very proud of that quant score.
In 2008, I took the GRE and got 780V, 800Q. (I’m still annoyed by the 780; if I’d focused in more, I might have gotten a double 800.) 800 Q is just the top 4% in any given year, but it’s probably more accurate to call it a top 10% score.
According to this GRE IQ estimator, my original GRE V+Q of 1340 is 99.452nd percentile, and my second GRE V+Q of 1580 is 99.993. But (forgive me IQ estimators), any IQ based on combining V+Q makes no sense, because an 800V-640Q is considerably more difficult to achieve than an 640V-800Q, so how can they be identical, IQ wise? Plus, the decimal point specificity is just goofy.
Looking at my quant score alone, in seven years from the age of 39 to 46, I jumped from just above average in math to pretty close to 2SD.
Did I get smarter, or did I simply learn how to use my existing intelligence?
The quant section of the old GRE was extremely g-loaded. I used to tutor for the test and ran into dozens of people who’d majored in college in math, knew more calculus and all that nonsense about vectors and matrix determinants and ordered ring fields than I’ll ever know, yet scored in the high 600s. Which is not to say that it wasn’t relatively easy, just that lots of smart people would occasionally miss questions because they were more about g than math competency. The new GRE combines both. I can’t find an online GRE practice test, but I did the problems on the ETS site, and I think I’d still get in the top 10%.
The GRE Math Subject test is what used to be called an “achievement” test. God, testing lingo has changed so dramatically in so little time. g is involved in the sense that a certain level of intelligence is required to learn the material. But a 120 IQ who’s taken calculus and number theory would outscore a 145+ IQ who has not.
I got 13 right out of 56. A 390. I wonder how many people get an 800 on the GRE General Quant and a 390 on the GRE Math? That’s a terrific illustration, really, of the difference between achievement and aptitude. I knew none of the number theory, only some of the stats, none of the integral questions, but all of the limit and derivative questions, and random other stuff.
Every single one of the questions I answered correctly was using math I’ve learned in the past fifteen years. Had I taken the test in 2005, I would have gotten zero correct. I took AP Calculus as a senior, remember none of it. All the math I know today is from my tutoring days or my time in teaching.
Did I get smarter, or did I simply learn how to use my existing intelligence? Here, it seems clearer than in the first case. The GRE Quant (old form) is definitely an aptitude test, which makes my big score jump odd. But acquiring new knowledge isn’t the same as having a higher IQ. Right? (asking seriously). I could do much better on this test if I studied up on integrals and 3-dimensional systems. Hold that thought.
To contrast, I took the GRE English Literature Subject Test, all 230 questions. For me, this test is diametrically opposite the GRE Math Subject test. The latter requires actual math knowledge. But the English Lit test is about 70% interpretation, 10% terminology (literary terms) and 20% content knowledge (knowing the plot of Ben Jonson’s plays, or familiarity with Matthew Arnold’s poems). I missed 56 questions, scoring a 650, in the 86th percentile (although I’ve always distrusted the scoring on English lit tests). Two of the misses were analysis and both were careless errors I’d never make in a real test. All the other missed problems were content knowledge–not anything I’d forgotten, but things I’d never learned. My English degree wasn’t terribly rigorous, but what I learned thirty five years ago, I remembered. I recognized Shakespeare’s writing in a sonnet I’d never read before–ditto Donne and Milton. I even guessed my way through Derrida and Foucault. But Wiliam Caxton, Nikki Giovanni–eh. Never heard of them. I read a lot of Faulkner short stories, but avoided his novels. And so on.
Most of the high difficulty questions (less than 30% answered correctly) were literary analysis, and I nailed them. The only hard questions I missed were three content knowledge (obscure authors) and one grammar question. The rest were in the 45-65% range, which is typical when the test is covering a broad range of material and no one knows everything. I think I could probably learn my way to a 700, but higher than that would require more interest in literature than I have.
So the GRE Math subject test requires specific knowledge, while the GRE Literature subject test allows people with high aptitude to do very well, even if their specific literature content knowledge is weak.
There aren’t many forty-something folks taking GRE Subject tests, but doesn’t it seem likely that it’s more common for someone to develop math content knowledge later in life than it is to suddenly develop excellent reading skills? Which suggests that reading comprehension, verbal ability, is more hard-wired to cognitive ability than math is. That might explain why math test scores have improved more than reading scores, generally. For all the wailing about math achievement, we do better at teaching students to improve their math abilities than we do at making them better readers.
From the study abstract: “Education does improve cognitive ability”
It does? That seems backwards to me. Cognitive ability improves educability. If all we had to do was educate people to make them smarter, I wouldn’t have this blog.
Does education actually make people smarter, or does it just teach them how to use their existing intelligence?
I have no answers, so I’ll stop here.