college | median sat (reading + math) | number of u.s. freshman | college rank of median student out of 4.413 million 18-year-olds | median iq if iq and college attended correlated perfectly | median sat iq equivalent | predicted mean wechsler iq assuming 0.72 correlation between the sat and wais |
caltech | 1525 | 224 | 112th (top one in 39,402) | 161 (U.S.. norms) | 147(U.S. norms);146 (U.S. white norms) | 134 (U.S. norms);132 (U.S. white norms) |
harvey mudd & princeton | 1500 | 1322 | 885th (top one in 4,986) | 153 (U.S. norms) | 145 (U.S. norms);143 (U.S. white norms) | 132 (U.S. norms);130 (U.S. white norms) |
yale | 1495 | 1214 | 2153rd (top one in 1,794) | 149 (U.S. norms) | 145 (U.S. norms);143 (U.S. white norms) | 132 (U.S. norms);130 (U.S. white norms) |
harvard & mit | 1490 | 2828 | 4174th (top one in 1,057) | 147 (U.S. norms) | 145 (U.S. norms);143 (U.S. white norms) | 132 (U.S. norms);130 (U.S. white norms) |
university of chicago | 1485 | 1337 | 6,257th (top one one in 705) | 145 (U.S. norms) | 144 (U.S. norms);142 (U.S. white norms) | 132 (U.S. norms);130 (U.S. white norms) |
columbia university | 1475 | 1139 | 7495th (top one in 589) | 144 (U.S. norms) | 143 (U.S. norms);141 (U.S. white norms) | 131 (U.S. norm);129 (U.S. white norms) |
washington university in st. louis & notre dame | 1465 | 3464 | 9796th (top one in 450) | 143 (u.s. norms) | 143 (U.S. norms);141 (U.S. white norms) | 131 (U.S. norm);129 (U.S. white norms) |
SAT (reading + math) was converted to IQ using the formula IQ = (post 1995 SAT)(0.081) + 23.835 explained here.
Reasoning is not only or especially iq. The priorities are reversed here, what matters most is the cognitive intimate of living beings. It begins by being and ” spread ” of thoughts and especially for literal actions throughout space and time.
You guys take actions within a specific context (iq, school, work) as a cognitive or operational centrality of human beings, which certainly is not in the purest truth. The IQ measures how well a person can do in their life cycle as an worker, but not as a human being.
I would fully believe in the cognitive tests that the vast majority of people who score high in these tests were nearly infallible in the basic pattern recognition or vital priorities.
The hbd want to build a sort of IQ-cracy, but it is interesting to note that an unusually high proportion of people who score higher on cognitive tests, often are not spectacular in the (supposed) simple ability to think rationally.
We have a lot of people who are very (organically or potentially) intelligent and throughout their lives they construct personal confused systems of facts or are negligent in relation to its systems, or even worse, build factoids systems.
It’s no use having a ” great potential ”, you also need to know how to use it.
It is believed that if you are born more intelligent, you will be so also in behavior, and I’m not just talking about the successful worldly behavior as you guys love to emphasize (the life cycle of worker).
What matters most is not JUST the potential, but the capacity or reasoning to use it properly and especially in philosophical terms, because after all, we’re all here discussing some truisms, hidden, anonymously, precisely because individuals who are completely devoid of morality , took the power (remember that the conservative past was not much different than today, only more stable, but stability is not always positive) and forced us to act this way.
The modus operandi in America does not operate on logic or honor, but the quest for power, status and/or money. Most people attend these schools in order to demonstrate their superiority over others via material wealth.
The SAT / IQ equivalents seem like a gross overestimation. In most states, the large majority of HS graduates take the SAT or ACT and, besides Caltech, the median test scores associated with the most elite schools generally gravitate around the top 1%.
It is far more accurate to say that the average Harvard student has (SAT-assessed) cognitive ability around 1 in 100 persons – very likely a wee bit higher, but not substantially so.
This is very different from what you’re asserting, that Harvard kids center around the 1 in 1000 persons mark, the truth of the matter exaggerated by a factor of about 10.
Iirc, a 145 IQ would actually be very close to the hard roof the the test, certainly not discernibly below (assuming a 1 to 1 correlation).
The SAT / IQ equivalents seem like a gross overestimation. In most states, the large majority of HS graduates take the SAT or ACT and, besides Caltech, the median test scores associated with the most elite schools generally gravitate around the top 1%.
It’s assumed by Charles Murray, Ron Hoeflin and others, that only about a third of U.S. 17-year-olds take the SAT, but virtually all of the most academically capable do, so the top 1% of the SAT population is perhaps the top 0.33% of the U.S. population (roughly IQ 140 as measured by the SAT)
What year is the data from the table from?
It’s obvious that many more people in general are now attending college (and graduating, for that matter), than was the case 2-3 decades ago.
This is just off the top of my head (distant memory of a statistic), but I think around 80% or so of highschool graduates end up attending some kind of postsecondary institution. Taking these admissions test (while not required for community college) is a streamlined and ubiquitous behavior at this point. From personal experience, it seemed like virtually everyone intended to go to college unless they were especially daft, or decided to capitulate once highschool performance and admissions test scores were properly reckoned. College is now a racket for the at-risk majority to attend, then drop out – for a price.
Perhaps some more recent data would clear this up, if anyone could lend a hand.
I haven’t looked into super recent data, but Charles Murray (pg 184 of his book Real Education claimed that in 2005, only 47% of U.S. high school grads, and 35% of U.S. 17-year-olds took the SAT.
Here’s some more recent data:
a record 1.7 million students from the class of 2015 took the SAT, while 1.67 million did the same from the class of 2014 and 1.65 million from the class of 2011. Nearly 33 percent of SAT test-takers from the class of 2015 were underrepresented minorities, compared to 29 percent from the class of 2011.
http://www.huffingtonpost.com/entry/2015-sat-results_us_55e751c6e4b0c818f61a56ce
So assuming about 4.4 million 17-year-olds a year, about 39% take the SAT. It seems that for the past several decades (at least since the 1980s and perhaps the 1960s), roughly the top third of late teens have been taken the SAT. Of course not every brilliant teen takes the SAT, but as Ron Hoeflin noted, whatever shortfall there might be is more or less negated by the number of foreign test takers.
But having said that, the SAT IQ equivalents might be overestimates because presumably, a lot of elite students took the SAT more than once and only showed their best score.
Love your blog; very interesting stuff that no one else seems to be interested in (in spite of the fact that IQ is number one in practically all avenues of life [like it or not]).
This post is confusing, however; the second-to-last column indicates an “IQ,” while the last indicates (more specifically) WAIS. Why did you further reduce the IQ generated by your SAT conversion formula to attain WAIS estimates, and what “IQ” is referenced by column second-to-last? 🙂 Is your point that the formula (as linked to) isn’t, in fact, accurate with which to determine *actual* IQ (WAIS) (I would assume that the number generated is far too high; with an SD of say 10, 75% of Caltech students would be “geniuses”)?
Further, Mensa disavowed SAT in 1994 due to lack of sufficient correlation. On that alone, I’d bet that a correlation of .72 is *far* too high (I trust Mensa’s analysis, though I don’t belong [I do belong to a 99th+ society that I won’t name here]). You recently estimated an IQ of 122, as well, which would indicate a correlation far lower than .72.
What’s the bottom line? 🙂 (If I had to guess, I’d bet that the WAIS scores generated here are more-or-less accurate, meaning that SAT/WAIS correlation is, in fact, well below .72 (likely around .5, in reality). 122 is obviously too low, and 145+ is ridiculously high.)
Thanks for the work–keep it coming!
Sorry; correction:
To add to this, ~1533 was apparently the (converted from 2400) 99th percentile SAT score cutoff in 2014.
If the top third of people take SAT, .33% attain this score, so 1533 = IQ ~140, SD15, assuming a correlation of 1.0. If that’s the case, the result of your WAIS calculation would indicate a correlation of ~.85, which simply can’t be accurate. Right? Am I going full retard, here? What’s being missed on my end? What two numbers are you correlating to arrive at .72? Thanks.
…I guess what I’m wondering is what the assumed correlation is that was used (or arrived at) re column penultimate. Unless it was 1.0, SAT/WAIS correlation wouldn’t be .72; .72 would be the correlation between WAIS and the <1.0 correlated IQ number displayed within column two (34/47 = .72). Correlation between pre-correlated 147 (correlation 1.0 equivalent, unless that's what we're saying 147 is meant to be) and WAIS: well below .72. Right? Man–I really need to stop trying to think in the midst of caffeine withdrawal. Forgive me if I've gone full retard.
Not sure I understand all your question, but I’ll clarify some major points,
The third last column is “median iq if iq and college attended correlated perfectly”. That column is simply what the median IQ at the college would be if 100% of the 224 smartest 18-year-olds in America all went to Caltech, and then if 100% of the next smartest 1322 18-year-olds in America went harvey mudd & princeton etc. In other words, if college attendance were 100% determined by IQ (i.e. all the smartest people went to the smartest colleges and all the dumbest people went to the dumbest colleges or no college at all), that would be the median IQs the colleges would have
The second last column is simply the ACTUAL median SAT of each college converted to IQ using this formula:
IQ = (post 1995 SAT)(0.081) + 23.835
And the last column is simply a statistical prediction of how the median student at each college would score on the WAIS based on an assumed 0.72 correlation between the SAT and WAIS.
So Caltech students, who have a median IQ of 147 on the SAT (that number might be inflated if Caltech allows multiple attempts on the SAT), then their expected IQ on the WAIS would be:
(147 – 100)(0.72) + 100 = 134
Thanks; so again, column second-to-last assumes SAT/IQ correlation of 1.0? What correlation arrives at 147, given that .72 of 147 arrives at 134 (presumable actual)? The ultimate correlation of SAT/WAIS therefore has to be lower than .72 (likely near .5, as you’d earlier estimated). Further, if it were .72, Mensa would very likely still accept it for admission (I’d have to assume).
My ultimate point: SAT is no means sufficient with which to gauge intelligence. 🙂
Thanks; so again, column second-to-last assumes SAT/IQ correlation of 1.0?
That’s right. It is assuming the SAT itself is an IQ equivalent test, and thus correlates perfectly with IQ by definition. But that doesn’t mean it correlates perfectly with OTHER IQ tests. Even official IQ tests (WAIS, Binet, Raven) don’t correlate perfectly with one another.
What correlation arrives at 147, given that .72 of 147 arrives at 134 (presumable actual)?
No correlation was used to arrive at that number, although implicitly, a perfect correlation was assumed. The formula for converting SAT to IQ equivalents was based on the actual frequency of different SAT scores.
So if only 1% of American 17-year-olds can obtain an SAT score of X, then X is assigned an IQ of 135, because BY DEFINITION, IQ 135, is the top 1% of the IQ distribution.
In other words, the numbers in the second last column were not derived from the numbers in the third last column (they were arrived at independently). However the numbers in the last column WERE derived from the second last column using regression
The ultimate correlation of SAT/WAIS therefore has to be lower than .72 (likely near .5, as you’d earlier estimated)
No. it’s the correlation between college attended and WAIS that would correlate near 0.5. That’s why you see such massive regression from the third last column to the last column. But you see much less from the second last to the last.
Further, if it were .72, Mensa would very likely still accept it for admission (I’d have to assume).
Mensa used to accept the SAT, and post 1995 SATs correlate with pre-1995 SATs about as well as parallel forms of pre-1995 SATs, so I suspect Mensa dropped the test because the college board denies it’s an IQ test, not because the psychometric properties changed
My ultimate point: SAT is no means sufficient with which to gauge intelligence.
It’s not as good as a gold standard IQ test like the WAIS, but it seems to correlate about as well with official IQ test as many official IQ tests correlate with one another…but I’d be happy to be proven wrong.
Ahhh–OK. I was merely trying to determine the legitimacy of 147 (e.g.); based on your response, by itself, it’s of limited value (given impossible correlation of 1.0). I simply didn’t assume that any formula attempting to gauge IQ by way of SAT would assume/produce correlation of 1.0, but there it is. (147 SD24 might be accurate.) 🙂
Haha–I’m not out to prove you wrong; the work you’re doing is invaluable (to those who don’t deny the validity of IQ in utter futility, that is). That said, I may be an extreme outlier, but I scored 140 WISC-III (so I’m told; too young to remember, no score sheet), 99th AFQT (missed perfect score by one question [fact; score sheet]; perfect = 99.9th percentile, so I’ve read), but only 1190 on SAT post-’95 (because I’d basically dropped out of school due to boredom circa 9th grade). 🙂 It drives me bonkers that SAT is given more merit as a supposed proxy of “intelligence” than membership in a 99th+ IQ society; I deal with it every day. “No; I’m actually ‘smarter’ than the average student at any school in America.” “Yeah, but you went to public school.” (…and so forth.) 🙂
p.s. Mensa categorizes post-’95 SAT as “no longer correlates.” No idea if their determination (as you indicated, it may not have been theirs) was quantitative or subjective, but there it is. 🙂
“N/A These tests no longer correlate with an IQ test. Note that the acceptance date applies to the date you took the test, not the date you join Mensa. You can still join Mensa by using older scores. ”
http://www.us.mensa.org/join/testscores/qualifyingscores/#NA
I simply didn’t assume that any formula attempting to gauge IQ by way of SAT would assume/produce correlation of 1.0, but there it is.
Virtually every chart you find converting SAT to IQ (except for Frey & Detterman) assumes a 1.0 correlation . They’re not saying the SAT correlates perfectly with official IQ tests, they’re simply asserting that the SAT is statistically equivalent to an IQ test.
For example, when you took the WISC-III and scored 140, nobody said, well the WISC-III only correlates 0.9 (or whatever) with the Binet, thus we’ll reduce you WISC III IQ to
0.9(140 – 100) + 100 = 136
Because the WISC-III is considered a valid IQ test in its own right,
Similarly, many assume the SAT is a valid IQ equivalent test in its own right
But it’s a completely arbitrary decision
But the Wechsler scales are much more accurate than the SAT, so if ever someone shows a large gap between the two, I tend to side with the Wechsler.
“Similarly, many assume the SAT is a valid IQ equivalent test in its own right”
…despite its obvious fundamental differences from WAIS, e.g. 🙂
“Assumption.” The problem perfectly summed, in but one word. 🙂 (See: unproven, and frankly, without a trace of justification given the nature of SAT vs. any other typical IQ test.)
“they’re simply asserting that the SAT is statistically equivalent to an IQ test.”
…but it clearly isn’t equivalent, statistically. The “assertion” is “assumption.” If they’re utilizing (God knows why) a correlation of 1.0, then they are claiming that the resultant IQ is accurate. Else, why provide it? (…because they don’t actually know SAT/IQ correlation, and for some reason, logic be damned, assume 1.0…?) 🙂
“But it’s a completely arbitrary decision”
Couldn’t agree more–the bottom line. Anyway, I like your WAIS estimates within the chart; I’d *believe* (assume…) that they’re spot-on based on prior research I’ve seen (including yours) of Harvard and MIT IQ, etc.
“But the Wechsler scales are much more accurate than the SAT, so if ever someone shows a large gap between the two, I tend to side with the Wechsler.”
You and no one else, unfortunately. 🙂 (Except me. And maybe a handful of other intelligent, informed people who understand that IQ is the bottom line in all walks of life.) 🙂
Anyway, my ultimate point is that SAT and WAIS, SB, Cattell, Mensa test, etc. have very little in common. It the latter four are “IQ tests” (and indeed they are, though I’m not a fan of Mensa’s ludicrously overvalued memory section), SAT is likely not; I see no justification for it to be considered an “intelligence” test, but rather, as indicated, a test of “aptitude” (whatever that means). (…how well one stayed on track through college-prep high school courses, in essence.) 🙂 No IQ society accepts post-’95, to my knowledge, so it stands to reason that .72 (otherwise perfectly acceptable, one would think) is simply higher than experts believe it to be (though this likely used to be the case). All I’m saying. 🙂
(Interestingly, GMAT is accepted. I could argue the logic of this all day, night, and year; it’s even worse as a proxy of intelligence. 🙂 )
(If it were up to me, IQ societies [and colleges] would accept nothing but actual IQ tests as measures of intelligence [which, let’s face it, is precisely what they’re trying to determine]. No politically correct pseudo-IQ tests allowed.)
(Intertel accepts 99th percentile [760]. Triple Nine accepts 746 [not 99th percentile]). (Not sure about Mensa.) Wonder who’s wrong. Obviously, GMAT to IQ conversion is *widely* contested amongst experts, then. Anyway, I’ve gone off on a tangent, here; my apologies. 🙂
Anyway, I’ve gone off on a tangent, here; my apologies
No, you’re making good points. I think I’ll devote a post to this very subject.
Opinion (if interested):
SAT’s ultimate value (forget GMAT…): an arguably substandard, more-politically-correct indicator of *presumable* “intelligence,” *assuming* (eye rolling: commence) that a certain (subjective) amount of mental ability is required to stay on track through college prep courses sufficiently to realize the “learning” of that programmed material on SAT (which indicates what, precisely; the presumed ability to continue to do so through college? Do we want programmed learners of professors’ “teachings,” or actual *thinkers* within the highest echelon of “education?”). 🙂 (The actual value of said “learned material” gauged via SAT, real-world, is practically nil [as is the case with GMAT].) 🙂
(I applaud people who properly prepare for [over years] and sufficiently game SAT/GMAT, but my company [SAT, not IQ, is used by companies to gauge “intelligence,” of course–the source of my hostility toward it] would “hire” by IQ–not “aptitude.” My company, amongst the likely one percent or less of companies who actually practice this, would almost certainly win. 🙂 )
<> (Probably the last.) 🙂
(end rant) (Probably the last.)
The party line used to be that the SAT measures “developed reasoning” meaning that if you mastered the very basic math and verbal skills required, the rest is just reasoning ability and they (the college board) won’t admit it, but that reasoning ability really does come down to IQ. The very basic math and verbal skills required are things every high school student learns and things that are required by colleges.
As for the WAIS: A long time ago, I took the WAIS III and got a 128 overall, but that hides a lot of variation (performance was 118 and verbal 137). For instance, I maxed out on block design and matrix reasoning (very g-loaded items) but did terribly on “complete the picture” type things and only OK on things like comprehension. The charitable interpretation is that I am very good on things requiring raw intelligence (matrix reasoning, block design, vocab) and badly on things measuring implicitly learned social intelligence (comprehension) and impulsiveness (making a story out of a series of pictures by putting them in the proper order). I only did OK on things measuring working memory because I had a send-off party the previous day and perhaps took it under less than ideal condistions. In my opinion, the fact that I lived in many different countries can explain why I am not so familiar with the rules of behavior in the US. IN MY OPINION, these things don’t count. Does that mean such things are not components of intelligence? No, even though they aren’t the most g-loaded sections of the WAIS, they are valid. That’s the reality of even the gold standard test: it has things in it that measure how well you fit in with a particular culture and how conscientious you are.
SAT requires “learning,” in addition to “intelligence.’ This is precisely why it’s not a valid indicator of intelligence alone (and shouldn’t be being used, as it clearly is, as a supposed proxy; as stated, the actual material measured by SAT [including relatively advanced mathematics and vocabulary, contrary to your statement above] is all-but-worthless in the real world). The same is true of GMAT (basically the same test, though even more game-able and often terribly inaccurate of GMAT-taking ability on first attempt). 😉
Culture-fair IQ testing (e.g. WAIS, SB, Cattell) does not allow for bias re: culture, behavior, or how “conscientious” one is; with all due respect, this is an excuse used, rather frequently, by people who don’t score as highly as they always thought they might. WAIS/WISC is the real deal; SAT is a sad, real-world (practically) worthless, second-tier estimator of “intelligence” (that doesn’t measure intelligence very well). In my opinion, it should not be being used, again, if a school’s goal is to acquire intellectual capital (and it very well should be). (Again, companies use it as a proxy of intelligence. This is incontestably wrong, given other viable [e.g. 12-minute IQ tests] alternatives.)
The SAT might require advanced vocabulary, but so does the WAIS. The WAIS has a subtest that literally requires you to just define words. Even the SAT does not have a section that just requires you to define words (on the SAT, the words are used in context). This WAIS subtest is actually one that has the most correlation to the overall WAIS score. If you claim that the SAT measures “learning” because it uses advanced vocab then you have to claim the WAIS also measures “learning” because it, too, measures advanced vocab or at least advanced enough that it can distinguish between people of 130 IQ and 155 IQ. Have you actually taken the WAIS? The algebra is very basic algebra I material (lately they’ve included algebra II) but it is something that even some middle schoolers learn in the US. The knowledge required is minimal.
The picture arrangement and comprehension subtests are thought to be related to social skills (social intelligence). See this http://paws.wcu.edu/mccord/pdf/campbell-mccord-jpea-1996.pdf
SAT uses verbal analogies dependent on vocabulary knowledge, does it not? It’s of greater weight than it is within WAIS, is it not? Your further comments precisely address my point. WAIS does not require substantial material to have been learned in advance, and this alone is why it is objectively superior. I have taken WISC, as stated above (twenty years ago; “forgot” the memory section, likely because it was of minor relevance).
Send Mensa an e-mail asking why SAT isn’t accepted, post-’95. Their statements will echo what I’ve stated here (insufficient correlation for the reasons stated).
“The picture arrangement and comprehension subtests are thought to be related to social skills (social intelligence). ”
I don’t need to click on that document to know that that’s utterly nonsensical. 🙂
By OK I mean my working memory index was something like 124 but I felt like I could have done so much better. I felt very scattered and easily distracted. I couldn’t focus enough during the test to do what the questions asked of me in an optimal manner.
That may be, but IQ testing exists precisely to overrule what you “feel like” your intelligence is. 🙂
Let me tell you a story: I took Mensa’s home test and scored 79/80. When I took the real test, which featured 1/3rd memory questions, I hadn’t a clue that the story I was being read before the test was supposed to be being remembered, given that the home test makes no mention of a memory component within the actual test. As a result, I answered (presumably) each of the forty memory questions incorrectly (minus some lucky guesses) (yet still scored 90th percentile, overall). 🙂
Mensa’s test should also be abolished and replaced with WAIS (no memory component; in my opinion, memory is not a valid component of “intelligence,” and if it were, it wouldn’t be the most essential, as their test indicates). Why Mensa feels WAIS to be unworthy of Mensa is unknown (ego?).
(Sorry; WAIS does have a memory section. It’s of minor importance.)
(I apologize for not reading your prior comment in full. 🙂 Still, what would the difference have been, overall? 130? Take Mensa’s test, if your goal is admission.)
In your defense, I agree that the memory section (as indicated) is problematic (impact to overall WAIS-IV FSIQ: I don’t know, but probably more than it should present). I don’t believe that the abbreviated form of the test includes this, though I might be mistaken. I’m of the opinion that a 12-minute test, which certainly contains no memory section, is perfectly sufficient with which to reasonably gauge IQ. If I were HR, that’s what I’d be issuing to all prospective candidates.
I believe you can take AFQT free of charge (on taxpayers), if Intertel admission is your goal (Mensa doesn’t accept it anymore–they’re too cool). 🙂 Try that one. If you score 98th percentile, my advice would be to take Mensa’s silly $50 test (just “remember” that it’s mostly memory). (…if society membership and/or reconfirmation of your IQ are your goals, that is.)
I went back and read more of your posts, 99th+ Percentile. Maybe for someone who basically stopped his education at 9th grade, the SATs are not a good measure of intelligence. Nevertheless, I stand by the assertion that they are, or at least used to be, a good measure of intelligence for people who have mastered the required material in high school. This doesn’t apply to you, 99th+ Percentile, but I think it’s somewhat valid for the average college student. For the WAIS-IV, the information, vocabulary, and arithmetic subtests are testing mostly “learning” (3 out of 10, that is). Similarities could be argued to be biased in favor of learning as well. For the SAT, once you’ve mastered the basic material, what differentiates a high score from a mediocre or low one is how you use that learned knowledge to solve problems. Mensa doesn’t use the SAT anymore because the college board doesn’t claim the SAT measures intelligence (but the reason they don’t claim it measures intelligence is because that would be a PR nightmare… it other words, it might actually be a disguised IQ test). Mensa uses the GMAT because the makers of the GMAT, the GMAC, claims it measures “higher-order reasoning skills” which kind of sounds like IQ. Pearson, the makers of the other graduate school exam Mensa accepts, the MAT, explicitly call the MAT a “high level mental ability test”. I am OK with my 128 on the WAIS-III and I took it for personal reasons — also, I have since qualified for Mensa but I’m on the fence about joining. I only mentioned that I took the WAIS III because it shows I have experience with one form of the WAIS (so I know what I’m talking about) and because I am trying to convey that even though I did badly on sections of it, I’m not blaming it on the test by claiming those sections are flawed. In other words, to be blunt, I am trying to convey that I am being objective and that the WAIS is a good IQ test (although it’s not culture fair – matrix reasoning tests tend to be more culture fair). That said, I think IQ estimates obtained from the SAT tend to be too high. Nevertheless, I don’t think they’re useless.
Here’s a document on the interpretation of Weschler tests taken from C.L. Nicholson & C.L. Alcorn, (April, 1993). Interpretation of the WISC-III and Its Subtests. Presentation at the Annual Meeting of the National Association of School Psychologists, Washington D.C. (if you have a problem with it, talk with the National Association of School Psychologists):
Look, I don’t want to be mean, but this comment is invalid on many levels. I’m beginning to repeat earlier statements and time is being wasted; in the interest of my time, this will be my last reply to you. If you seemingly disprove an argument or two of mine that don’t take away from the central point that SAT is not (in fact) a comprehensive intelligence test and should never be used as a proxy of IQ, great! 🙂
Let’s begin:
“I went back and read more of your posts, 99th+ Percentile. Maybe for someone who basically stopped his education at 9th grade, the SATs are not a good measure of intelligence.”
SAT should never be used as a proxy of IQ, for this reason. It isn’t a measure IQ, as I’ve arguably proven. You seem hell-bent on convincing me that SAT doesn’t require skills learned above ninth grade, but obviously this is wrong. Else, the test would be taken by ninth-graders, and/or there would be no advantage at taking the test when it’s typically taken vs. taking it during the ninth grade. (To be clear, I have multiple Master’s degrees from top-tier schools. SAT and GMAT didn’t get me in [hard work did].) 🙂
“Nevertheless, I stand by the assertion that they are, or at least used to be, a good measure of intelligence for people who have mastered the required material in high school. This doesn’t apply to you, 99th+ Percentile, but I think it’s somewhat valid for the average college student.”
The test doesn’t measure intelligence. It measures mathematical and verbal knowledge and to a lesser extent, reasoning ability (one of many components of IQ). The majority of IQ components, as tested by WAIS, SB, and Cattell, are missing entirely and cannot be otherwise assumed (so let’s not do so, shall we…). “Somewhat valid?” I agree–as in correlation .4-.5. 🙂 “Average” college student? Meaning what? If you’re mocking my moniker (can’t really tell), it isn’t boasting. It’s statement of fact as determined by WAIS and AFQT, both of far greater validity as measures of comprehensive intelligence than SAT. Technically, I’m a “genius.” How’s that for subjectivity? 🙂 (In my opinion, that subjective label should never be applied to anyone under 160, SD15.)
“For the WAIS-IV, the information, vocabulary, and arithmetic subtests are testing mostly “learning” (3 out of 10, that is).”
How did you determine that these are “mostly learning,” and not tested at a lesser grade-level than SAT?
“Similarities could be argued to be biased in favor of learning as well. For the SAT, once you’ve mastered the basic material, what differentiates a high score from a mediocre or low one is how you use that learned knowledge to solve problems.”
Yes, and the level of material required is higher than that within an actual IQ test, because SAT actually tests the extent of material learned in high school college-prep courses, not intelligence as gauged by IQ tests.
“Mensa doesn’t use the SAT anymore because the college board doesn’t claim the SAT measures intelligence (but the reason they don’t claim it measures intelligence is because that would be a PR nightmare… it other words, it might actually be a disguised IQ test).”
Again, Mensa classifies SAT as “N/A” (see earlier post and link). “Insufficient correlation.” Meaning, it’s invalid for use as an assumed measure of IQ (because that isn’t what SAT measures). College Board cannot claim that SAT is an intelligence test because SAT isn’t an intelligence test (though it may *used* to have been more aligned with IQ testing). It may require a certain (unknown) level of intelligence, but it predominantly requires college-prep learning (which is what it primarily tests).
“Mensa uses the GMAT because the makers of the GMAT, the GMAC, claims it measures “higher-order reasoning skills” which kind of sounds like IQ. Pearson, the makers of the other graduate school exam Mensa accepts, the MAT, explicitly call the MAT a “high level mental ability test”.”
Ridiculous. Mensa et al. would never use a test simply because it’s conceiver/marketer calls it something that may or may not qualify it as some form of intelligence test. Had you read *all* of my earlier posts, you’d have seen that IQ societies vary *widely* re IQ/GMAT conversion. Why? GMAT isn’t an intelligence test. Why Mensa et al. accept the second-tier GMAT (in one of my programs, there were a handful of people who scored 150 points higher on take two. It’s thusly easily game-able and otherwise worthless…), I simply do not know, especially as it clearly cannot be converted to an IQ figure with even reasonable accuracy. Political reasons? We can all speculate.
“I am OK with my 128 on the WAIS-III and I took it for personal reasons — also, I have since qualified for Mensa but I’m on the fence about joining.”
How did you “since qualify?” (Just curious.) p.s. don’t join Mensa (or at least don’t advertise it). 🙂
“WAIS is a good IQ test (although it’s not culture fair – matrix reasoning tests tend to be more culture fair).”
Does the cultural “unfairness” you claim to perceive truly affect the bottom line? Again, this is generally an excuse used by many to justify lower-than-expected overall scores. It’s also used, en masse, by IQ deniers (who, ironically, tend to have relatively high IQs).
“That said, I think IQ estimates obtained from the SAT tend to be too high. Nevertheless, I don’t think they’re useless.”
147 vs. 134, as demonstrated here, is nearly an entire standard deviation. They’re useless, and to assume a correlation of 1.0, for any purpose, is entirely pointless (stemming from extreme arrogance). Garbage in, garbage (147) out.
“Here’s a document on the interpretation of Weschler tests taken from C.L. Nicholson & C.L. Alcorn, (April, 1993). Interpretation of the WISC-III and Its Subtests. Presentation at the Annual Meeting of the National Association of School Psychologists, Washington D.C. (if you have a problem with it, talk with the National Association of School Psychologists):”
I don’t “have a problem” with any research, I don’t care about “interpretation,” and this is borderline ad hominem-like. I have a problem with specific research being made to appear as though it’s representative of the population mean (logically fallacious). I can cherry-pick, as well (but won’t). 🙂
Face it: as stated, SAT is an inferior gauge of intelligence. As previously posted, it includes ~30% of that which is measured, much more accurately, by WAIS, SB, and Cattell, amongst other *actual* intelligence tests (that require a far lesser degree of learning than SAT). SAT may well have *previously* been a reasonably accurate measure of intelligence, pre 1995. As Mensa (and all other IQ societies) agree, it no longer is (“insufficient correlation”). Again, I’m not an advocate of dumping testing for admission to schools (SAT should *never* be used by employers to gauge intellect [for that matter, GMAT shouldn’t, either], however). I’m an advocate of replacing SAT with even a 12-minute IQ test, or better, WAIS or its abbreviated version. Society would immediately benefit. 🙂
Runningsquares, to clarify, what level of education (grade) does SAT require (in your opinion)? I may have misinterpreted your statement, and simply wanted clarification. If not ninth grade, which? If taken by all eleventh-graders, say, would the playing field be level, or would those who’ve taken college-prep courses stand to score more highly?
http://blog.prepscholar.com/when-should-you-take-the-sat-or-act-best-test-dates
“(2) Content Readiness and Classes in School
Much of the SAT and ACT tests general concepts that you’ll learn in high school, and your scores will be higher once you’ve taken the classes that cover material on the tests. In particular, the best time to take the SAT or ACT is after you’ve taken Geometry and Algebra II in school. Both tests heavily feature algebra and geometry, and if you’ve never encountered these subjects, your knowledge foundation will be weak and you’ll do poorly on the math section.
As for the reading and writing/English sections, most students will have had enough exposure to English and grammar to form a baseline understanding of the concepts needed for the SAT and ACT. You will need to prep specifically for the SAT and ACT beyond what you’ve learned in school to excel on these sections.
Now, this doesn’t mean that just because you took Algebra II and English classes, you’ll do well on the test. The test questions themselves are very different from what you’ll encounter in school, and much of SAT and ACT prep involves learning these special types of questions and becoming familiar with the format of the test. But having the underlying content foundation will make you improve much faster.”
In other words, SAT is not an intelligence test.
The fact that you indicated that a ninth-grade education is insufficient, as you concur proves my point, does it not? Again, simply looking for your clarification (on this one point).
I should be allowed to make a point-by-point reply as well, right? I should note that there are many psychological studies that use the SAT as a proxy for an IQ test. Search articles citing the Frey and Detterman study (http://pss.sagepub.com/content/15/6/373.short).
“SAT should never be used as a proxy of IQ, for this reason [because they require some background in schooling].”
Arguably, IQ tests in general require some sort of background. In order to take an IQ test in English you need to be able to understand prompts given in English. I’m not sure why this is at issue. If you need a little bit more knowledge to do well, it doesn’t suddenly make the rest of the test’s discriminative power be solely on content mastery (as opposed to reasoning or intelligence). The basic material that one needs to know to do well on the SAT used to be (when I took it post-1995) algebra I and geometry (maybe trig). Whether you’re ready by ninth grade, I guess, depends on whether you’ve mastered the material in those classes by ninth grade. In that case the test would probably be taken by tenth graders after they took algebra I and geometry or sometimes in eleventh grade or senior year to give students more time to master algebra I (which they review in algebra II and trig classes).
It’s great that you got several graduate degrees and so forth. Congratulations. I’m not attacking you personally though.
“The test doesn’t measure intelligence. It measures mathematical and verbal knowledge and to a lesser extent, reasoning ability (one of many components of IQ).”
First to counter the idea that the test doesn’t measure intelligence: Frey and Detterman show a correlation (corrected for range restriction) of about 0.72 between the revised and recentered (post-1994) SAT and a Raven’s test (http://pss.sagepub.com/content/15/6/373.short). That’s a very high correlation and almost what you would expect between two IQ tests, like the SB and the Raven’s test. I’m not cherry picking. This article has been cited 393 according to google and is the subject of popular press http://www.boston.com/news/globe/ideas/articles/2004/07/04/the_sat_tests/?page=full, so it has been noticed. The mathematical knowledge required is not abstract algebra, complex functional analysis and multivariate calculus on non-Euclidean spaces. If that were the case, then maybe you would have a point, but it’s not the case. If you accept the validity of IQ, you have to accept the idea that you can summarize a person’s ability to do a variety of cognitive tasks with one IQ score. That means you don’t have to test all the many “components” as you say, which is why the BOMAT or advanced progressive matrices tests can claim to measure IQ even though they only test one component (matrix reasoning).
“How did you determine that these are “mostly learning,” and not tested at a lesser grade-level than SAT?”
I didn’t “determine” anything. What I was saying is that those WAIS subtests are measuring knowledge acquired through experience (implicit or explicit learning). Those subtests might be the vocab and information subtests, for instance. Grade level is really irrelevant. The test is meant to discriminate between adults, so the material is at an “adult” grade level. I mentioned the SAT being being valid as an IQ proxy for the “average college student” because “the average college student” took the traditional route and takes the SAT after algebra I and geometry. It had nothing to do with you personally.
“Mensa et al. would never use a test simply because it’s conceiver/marketer calls it something that may or may not qualify it as some form of intelligence test.”
That is clearly not what I meant. Mensa (for the most part) accepts tests that are actually good IQ tests. However, if the maker of the test says that their test (call it test A) does not measure IQ and a person enters the society using his scores on that test (test A), then others can claim he didn’t qualify with a legitimate IQ test. To eliminate the possibility that someone can claim another Mensa member “cheated” his way into the organization, they don’t accept tests where the test makers clearly state that they are NOT intended to measure intelligence. The truth might be that their test does measure intelligence, but that doesn’t matter as far as Mensa is concerned.
“Does the cultural “unfairness” you claim to perceive truly affect the bottom line? Again, this is generally an excuse used by many to justify lower-than-expected overall scores. It’s also used, en masse, by IQ deniers (who, ironically, tend to have relatively high IQs).”
I didn’t have any ulterior motive in stating that the WAIS is not culture fair. I wasn’t trying to prove a point, it was only in response to your stating that it is culture fair. I’m OK with my 128 (although I’m pretty sure I could score higher on the WAIS-IV, how would that help me?). The WAIS is not culture fair. That doesn’t make it a bad intelligence test, it just means it assumes a particular cultural background. That’s why they adapt the WAIS for Australia and New Zeeland (South Africa, US, etc.). https://www.quora.com/Intelligence/What-are-the-biggest-criticisms-of-the-Wechsler-Adult-Intelligence-Scale-WAIS . I’m not claiming that there aren’t culture fair IQ tests, only that the WAIS isn’t one of them.
Lastly, the interpretation of low picture arrangement and comprehension subtest scores is one given by the psychologist (with a PhD) who discussed the test with me. Some people don’t like to interpret subtest scores at all, but that’s another story. It’s also an interpretation that you can easily find in the literature. Not everyone agrees, but it’s not a cherry-picked wacko idea by any means.
You sure can. Doesn’t mean I’m going to read it. SAT is not an IQ test. Let’s move on.
(It isn’t an intelligence test, either. Just wanted this to be clear. You know, IQ vs. “intelligence.”) 🙂
Oh, and as a side note: Even if the WAIS were given as an admission test for college, you would not want to take it before you’re at least 16-17 years old (a sophomore or junior in high school, the WISC is valid until an age of 16:11).
WAIS, WISC, whatever. I vote for abbreviated WAIS. As stated, even a 12 or 15-minute test would be of much greater value than SAT. 🙂
SAT is neither an intelligence nor IQ test. It is a test of higher-level learned material that may or may not have required a certain level of intelligence. Objective fact. See my many other posts, here, on the matter, if this remains elusive. Let’s move on.
Back to the original point, SAT’s material coincides around 30-40% with WAIS/WISC, based on the below (can’t imagine IV differs much). It’s less than half the story:
The WAIS-III, a subsequent revision of the WAIS and the WAIS-R, was released in 1997. It provided scores for Verbal IQ, Performance IQ, and Full Scale IQ, along with four secondary indices (Verbal Comprehension, Working Memory, Perceptual Organization, and Processing Speed).
Verbal IQ (VIQ)[edit]
Included seven tests and provided two subindexes; verbal comprehension and working memory.
The Verbal Comprehension Index (VCI) included the following tests:
Information
Similarities
Vocabulary
The Working Memory Index (WMI) included:
Arithmetic
Digit Span
Letter-Number Sequencing and Comprehension are not included in these indices, but are used as substitutions for spoiled subtests within the WMI and VCI, respectively.
Performance IQ (PIQ)[edit]
Included six tests and it also provided two subindexes; perceptual organization and processing speed.
The Perceptual Organization Index (POI) included:
Block Design
Matrix Reasoning
Picture Completion
The Processing Speed Index (PSI) included:
Digit Symbol-Coding
Symbol Search
Runningsquares, as this should show you, SAT and WAIS (and other IQ tests) have very little in common. SAT isn’t a comprehensive measure of intelligence.
(Not sure why I can’t reply to my own comment, as I cannot edit the original, either, but here we are.) Runningsquares, as stated above, I have multiple degrees from top-tier (top-twenty) schools (public, undergrad only). I competed (curved grading) against people who’d scored 1500, etc. on SAT, 750 on GMAT, etc., and frankly, pummeled them 50% of the time or more (because my IQ was likely higher). 🙂 Not boasting–providing further evidence that SAT/GMAT aren’t really “capability” (intelligence) tests. They’re substandard proxies, of a more politically correct nature, that shouldn’t be being used given far better/more efficient alternatives.
PumpkinPerson, we’re saying that Harvard’s WAIS mean would be 132, SD15? Or a different SD? I guess 132/15 seems reasonable; ~30% would be at 140 and up. Thanks! Very much looking forward to your GMAT/IQ post. Even better, how about GMAT v LSAT v MCAT v SAT v WAIS v abbreviated WAIS? 🙂 (Lol…imagine how much easier life would be if they’d simply give politically incorrect, much more accurate, much more efficient, actual IQ tests.) 🙂