SAT scores and IQ

[Anarchist420]

Who here thinks SAT scores (1996 until the writing part was added) correlate with IQ?

According to the proposed formula I read about, the verbal component barely correlates with IQ while the math component correlates quite a bit with IQ. Why is that?

 

[slayer202]

Math is more analytical.

 

[Atomic Playboy]

No, it correlates with what you've studied in school. I started taking the SATs in 7th grade because my mother was paranoid about me getting into a good college. I got around 1080 or so my first time around and steadily increased up to the 1400 range by junior year. If you're using a reputable IQ test, there won't be a 30% fluctuation in a 4 year period based on whether you remember the quadratic formula or know what gregarious means.

 

[SP33Demon]

No correlation. SATs measure learned ability, IQ tests measure ability to learn.

 

[Arachnotronic]

i'd like to think SAT correlates with IQ...but mainly due to my perfect score on the darn thing (third try, though).

 

[Gigantopithecus]

Of course there's a correlation. The real question is how tight it is.

 

[Dr. Zaus]

Standardized tests are games.

If you don't know how to play at all then the smarter you are the faster you'll catch on to the basic strategy.

At some point though, anyone that studies enough will have all of the basic strategies down.

But then along the top-edge of play you have intelligence becoming a factor once-more because only the most intelligent AND dedicated to learning the game will be at the top.

 

[Anarchist420]

Of course there's a correlation. The real question is how tight it is

True. I should've been more precise in my OP.

 

[Dr. Zaus]

True. I should've been more precise in my OP.

The level of disparity between IQ and SAT score is, itself, a bell curve. With very-low-levels and very-high-levels of training correlating most closely with IQ.

 

[tynopik]

i think most studies showed that sat prep courses only offered a minimal improvement, so even after you knew all the 'tricks', a good part of it just comes down to ability.

i don't know how well it compares across different groups, but from observing the scores in my class, it was pretty darn accurate

 

[Arachnotronic]

The trick to the SAT isn't to try to "learn tricks". It's to just get genuinely good at reading comprehension and math.

 

[Baked]

SAT is just regurgitation of P-SAT. Now, if you can get a perfect score without doing a million P-SATs, going off just what you learned in school, then yeah. With that said, repetition of things you know doesn't equal high IQ.

 

[Dr. Zaus]

i think most studies showed that sat prep courses only offered a minimal improvement, so even after you knew all the 'tricks', a good part of it just comes down to ability.

i don't know how well it compares across different groups, but from observing the scores in my class, it was pretty darn accurate

I went from the 60th percentile to the 98th percentile on the GRE by training myself; and I wen from the 70th percentile to the 99.8th percentile on the LSAT with training.

Not the SAT, clearly, but similar standardized test games.

 

[tynopik]

http://www.answers.com/topic/testing-impact-of-test-preparation-programs

The average effect of test preparation programs on the verbal section of the SAT is probably between 5 and 15 points (.05 and.15 of a standard deviation); the average effect on the quantitative section is probably between 15 and 25 points (.15 and.25 of a standard deviation). The largest effects found in a published study of commercial SAT preparation reported estimates of about 30 points per section.

http://www.marketplace.org/topics/life/report-sat-prep-courses-get-bad-score

the report pulled together academic studies about the effectiveness of SAT preparation. It found that the average score increase as a result of the prep courses is 30 points.

 

[dave_the_nerd]

The final test is a take home. You pass when you realize these things don't matter very much.

 

[Dr. Zaus]

The final test is a take home. You pass when you realize these things don't matter very much.

:thumbsup:

http://www.answers.com/topic/testing-impact-of-test-preparation-programs



http://www.marketplace.org/topics/life/report-sat-prep-courses-get-bad-score

I once taught a prep-course. I had NO trouble upping a 500 math to a 600 math when people worked on it, did the work assigned and practiced as I told them.

It's the emotional issues surrounding character and dedication that have lead to that mediocre performance in the first place that keeps people from improving. Not that the test is un-teachable.

 

[mugs]

No correlation. SATs measure learned ability, IQ tests measure ability to learn.

It stands to reason that your IQ will affect how much you have learned (and how well you have learned it) by a specific point in your life. That is what the SAT measures.

I have to give my IQ credit for my SAT score, because it certainly was the result of effort.

 

[tynopik]

I once taught a prep-course. I had NO trouble upping a 500 math to a 600 math when people worked on it, did the work assigned and practiced as I told them.

what percentage of people worked on it, did the work assigned and practiced as you told them?

unless you kept really good, objective records, this is something that is almost impossible to determine. If someone 'fails', it's easy to point to the things they did wrong while ignoring what they did right and if they 'succeed' it's easy to point to what they did right while glossing over what they did wrong.

It's the emotional issues surrounding character and dedication that have lead to that mediocre performance in the first place that keeps people from improving. Not that the test is un-teachable.

Those character traits aren't unrelated to IQ. It's just as likely that the first score was aberrant for them and they would have done just as (or almost as) well on the retest whether they took a prep course or not

 

[Dr. Zaus]

what percentage of people worked on it, did the work assigned and practiced as you told them?

~60% of three classes that I taught over the course of a year, each with about 50 people to start and 40 at the end.

what percentage of people worked on it, did the work assigned and practiced as you told them?

unless you kept really good, objective records, this is something that is almost impossible to determine. If someone 'fails', it's easy to point to the things they did wrong while ignoring what they did right and if they 'succeed' it's easy to point to what they did right while glossing over what they did wrong.

I was collecting data on improvement for a study on the efficacy of the program; It got me an A in my multiple-regression quantitative methods course. I couldn't publish the paper because some ass-hat at the IRB said I couldn't and I didn't fight it because Pedagogical scholarship is poorly rewarded anyway. I hadn't even thought about the paper until this thread and I seem to remember that its been lost to a hard-drive reformat before drop-box and Ill try to remember what I can of it:

There's the confound of self-report bias and people who might no turn in the homework assigned but did it (as it wasn't for a grade, but for feedback). The adjusted R^2 for the total model of the factors 'outside study hours', 'home works turned in', and 'practice tests completed' came out to ~.5 with significant weights on all three factors (I don't recall the weights) but I do recall that on median, blocking for weighted factors of the above in a construct I call 'trying' the average score went from the low 400s to the mid 500s while the 'slack-asses wasting both my time and theirs' went from the upper 300s to the lower 400s (I assumed this was a test of regression toward the mean).

So half of the overall improvement was correlated with the model and in the end I estimated about a 100point improvement for people that were 'trying'.

Potential confounds:
Self reports for two of the three variables.
(though this should reduce the impact of the model, not improve it)

Third-variables not accounted for.
(maybe self-starters, the assumption that the attributes of the class were causal is not tested)

Limited range. My students were, to put it politely, on the low-end. the scores were not normally distributed, but rather were leptokurtic and positively skewed . It looked more like a no-tail chi^2 distribution, really (with a few out-liars on the high-end that I dropped from the model).

I dropped 'attendance' post-hoc as it was multicolinear with most of the other factors and insignificant when they were entered into the model. (vif > 10)

Those character traits aren't unrelated to IQ. It's just as likely that the first score was aberrant for them and they would have done just as (or almost as) well on the retest whether they took a prep course or not

experimental causality can not be determined by the data I collected, nor can the impact of survivor bias; but what could be determined was that those worked on it, did the work assigned and practiced as I told them improved while those that did not do the work (but stayed in the class) kept sucking it. The improvement for my sample group was not significantly correlated with their original score (but I blame this on the very limited range of the students).

I would have expected more regression toward the mean, but I didn't see it. I think this is because the GMAT (which i was teaching) is based on a computer-adaptive-test an regression toward the mean was adjusted for over the course of the test.


So that's what I got:

I improved significantly, other people seem to improve significantly when they try. But overall, the same factors that make someone crappy at the test make them unlikely to work hard.

BTW contentiousness is orthogonal to general intelligence.

If I was to guess, I would say that SAT/GRE/GMAT/LSAT is a function of contentiousness, social support and general intelligence. With the very high and very low ends being overwhelmed by general intelligence and social support being a necessary but not sufficient quality.

 

[tynopik]

interesting

 

[TecHNooB]

close thread, ban op

 

[Shadow Conception]

I think there is a strong correlation. The questions in the math section of the SAT aren't simple regurgitation questions -- they're designed to test how creatively you can solve a problem in addition to remembering principles.

 

[Juked07]

No correlation. SATs measure learned ability, IQ tests measure ability to learn.

Haha you just said learned ability and ability to learn are uncorrelated, nice.

 

[Juked07]

The level of disparity between IQ and SAT score is, itself, a bell curve. With very-low-levels and very-high-levels of training correlating most closely with IQ.

Except the SAT maxes out. So at the top of the SAT score curve, there is a wide range of IQs, and correlation falls back towards zero again. (It only takes moderate IQ to get a very high SAT score if you work hard, as others have mentioned).

 

[SKORPI0]

Interesting read.... :|

[FONT=Arial,Helvetica]SAT tests measure IQ[/FONT][FONT=Arial,Helvetica]
[/FONT]

[FONT=Arial,Helvetica]Psychologists find SAT tests to be strong indicator of general intelligence [/FONT][FONT=Arial,Helvetica]in a March 3, 2004 press release::[/FONT]
[FONT=Arial,Helvetica]While the SAT is generally a good predictor of a student's performance in the first year of college, a [/FONT][FONT=Arial,Helvetica]new study from researchers at Case Western Reserve University[/FONT][FONT=Arial,Helvetica] finds that, more than anything else, the SAT is a measure of overall intelligence. Meredith C. Frey, a doctoral student in psychology, and Douglas K. Detterman, a professor of psychology, examined the relationship between SAT results and general cognitive ability in two studies. They believe the results of their study mean researchers can fairly accurately estimate a person's intelligence without administering a lengthy IQ test.[/FONT]

[FONT=Arial,Helvetica]But is this SAT an IQ test? "It is in a sense an IQ test. The SAT and IQ test correlate very highly. Between the SAT and the IQ, they correlate almost as much as the SAT correlates with a second administration of the SAT, as much as it correlates with itself. So they're very similar tests in content." [/FONT]from[FONT=Arial,Helvetica] [/FONT]http://www.pbs.org/wgbh/pages/frontline/shows/sats/test/views.html

[FONT=Arial,Helvetica]Is the SAT an IQ test?
[/FONT][FONT=Arial,Helvetica]"According to people in the field--especially if they're sort of letting their hair down--they will say the SAT is essentially an IQ test, particularly the verbal portion is essentially an IQ test. I want to step back a little from the idea that the IQ test is a scientific measurement of intelligence. From the beginning, IQ tests essentially traffic in vocabulary items: antonyms, analogies, reading comprehension, it's a test of vocabulary fluency and accomplishment. So, the premise of an IQ test is that it is the same thing as intelligence. Maybe it is, maybe it isn't. It's measuring one specific thing. It's not a magical, mystical test. The SAT grew out of an IQ test and the verbal in particular takes the oldest chestnut IQ testing techniques and applies them to high school seniors. And you know very widely, including in the Bell Curve itself, the SAT verbal score is used as a proxy IQ score, or is used as interchangeable with IQ scores. It's the thing that--to the extent that there's a sort of secret about Educational Testing Service, which administers the SAT, the secret is that at least the test makers there know that what they're doing is administering a mass IQ test but the organization's very invested in denying that." [/FONT]from[FONT=Arial,Helvetica] [/FONT]http://www.pbs.org/wgbh/pages/frontline/shows/sats/interviews/lemann.html

[FONT=Arial,Helvetica]
[/FONT]