6 ms·

> My initial assessment is that there isn’t any correlation between IQ and test scores This is completely false - I'm not sure how you got that assessment. Ma

> My initial assessment is that there isn’t any correlation between IQ and test scores This is completely false - I'm not sure how you got that assessment. Many test scores (e.g. ACT) are heavily IQ/g-predictive. SAT used to be better at predicting IQ/g but it's still pretty good. g-factor is the single largest predictor of academic success (which, these days, is mediated mostly by test scores).

austincheney 6y ago
> Many test scores (e.g. ACT) are heavily IQ/g-predictive. Is there evidence of that? Many university recruiters at some of the most prestigious schools believe otherwise.
- comp_throw7 6y ago
  Yes, the evidence is quite strong: https://randomcriticalanalysis.com/2015/06/18/on-sat-act-iq-and-other-psychometric-test-correlations/ https://randomcriticalanalysis.com/2015/06/18/on-sat-act-iq-...
  - austincheney 6y ago
    > The SAT – ASVAB correlation is quite strong and it is probably weakened by range restriction effects. People with lower IQ scores and/or lower HS GPA are much less likely to sit for the SAT. That's a bullshit guess. More importantly that study did not account for performance disparities versus preparation.
    - comp_throw7 6y ago
      Putting aside the question of the selection effect (which, frankly, seems like it could only realistically go in one direction, unless you want to propose a plausible mechanism by which people with higher than average IQ are _less_ likely to sit for these standardized exams)... You could also spend 5 seconds Googling the question and find that that this is not exactly an untrod ground in academic literature. The relationship is strong and well-established, far more so than almost any other result from the social sciences. Also, yes, test results, can, in fact, be improved by studying (or other things!). This does not mean that they are not strongly correlated to IQ. Reality is generally not monocausal. I don't see how this contradicts the original claim.
      - austincheney 6y ago
        > This does not mean that they are not strongly correlated to IQ. It does not mean they are related or that they aren’t, which only suggests you are missing the point: selection bias. > You could also spend 5 seconds Googling the question and find that that this is not exactly an untrod ground in academic literature. It has been studied exhaustively. The relation between IQ and academic performance is poor according to research from the APA in addition to the personal opinions of people who perform college selection professionally. That correlation is better than a random guess but isn’t anything close to alignment. People, despite the evidence, need this to be true. After all it is a money making industry. Standardized college selection tests are like tax preparation, an industry that can be easily eliminated by a more simplistic unbiased and more effective tool. With enough effort to the contrary the arguments begin to sound like a self-sustaining fraud built around a fictional necessity and a bunch of bandwagon fallacies.
        comp_throw7 6y ago
        We aren't talking about the relation between IQ and academic performance (although that is also clearly established to be significant). We're talking about the relationship between IQ and ACT/SAT scores. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6963451/ https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6963451/ https://www.psychologicalscience.org/pdf/ps/Frey.pdf https://www.psychologicalscience.org/pdf/ps/Frey.pdf I certainly don't "need" it to be true - things are either true or they aren't, and needing or wanting differently doesn't change that. "That correlation is better than a random guess but isn’t anything close to alignment." Do you want to make a more specific claim? Again, the original claim you're attempting to rebut is "Many test scores (e.g. ACT) are heavily IQ/g-predictive." How strong does the effect need to be? Since, again, nobody is claiming that it's 100% predictive, merely a significant factor. Do you want to give me an effect size? Keep in mind the typical social psychology effect size comes out to r usually between 0.15 and 0.3: https://psyarxiv.com/2epc4/ https://psyarxiv.com/2epc4/ https://jenni.uchicago.edu/Spencer_Conference/Representative%20Papers/Richard%20et%20al,%202003.pdf https://jenni.uchicago.edu/Spencer_Conference/Representative...
        6y ago
- FabHK 6y ago
  > Many university recruiters at some of the most prestigious schools believe otherwise. Is there evidence for that? As has been pointed out, some national Mensas accept standardised tests as a substitute for an IQ test. For references, see here eg.: https://www.sciencedirect.com/science/article/abs/pii/S0160289606000535?via%3Dihub https://www.sciencedirect.com/science/article/abs/pii/S01602... Wikipedia: Some measures of educational aptitude correlate highly with IQ tests – for instance, Frey & Detterman (2004) reported a correlation of 0.82 between g (general intelligence factor) and SAT scores; other research found a correlation of 0.81 between g and GCSE scores, with the explained variance ranging "from 58.6% in Mathematics and 48% in English to 18.1% in Art and Design". https://en.wikipedia.org/wiki/Intelligence_quotient#Social_correlations https://en.wikipedia.org/wiki/Intelligence_quotient#Social_c...
  - austincheney 6y ago
    The first paragraph of that Wikipedia article affirms my opinion: > The correlation between IQ scores and grades is about .50. This means that the explained variance is 25%. Achieving good grades depends on many factors other than IQ, such as "persistence, interest in school, and willingness to study" (p. 81). While that paragraph specifically mentions school grades instead a more narrow subset of certain specific tests the point still stands as affirmed by the APA. The numbers you quoted from the third paragraph allow are only about 80% aligned. That is a tiny bit better than half between perfection and a random guess. The unfortunate reality is that performance on those tests is primarily a direct correlation of time and money spent on preparation specifically for those tests. A major differentiator between those standardized tests and many general IQ tests is centrality versus assessment distribution. You wouldn't for example test creativity with a multiple choice question with only one correct answer the same for everybody. Look, most people who are from a higher income population want to believe that standardized tests are a fair reflection of intelligence, because with enough practice and preparation they can score higher and thus believe they are more intelligent than the population at large. If the population at large spent less effort preparing for the test then its value as a tool of performance compared to other people is less valid. That is selection bias.

austincheney 6y ago

> Many test scores (e.g. ACT) are heavily IQ/g-predictive. Is there evidence of that? Many university recruiters at some of the most prestigious schools believe otherwise.

comp_throw7 6y ago
Yes, the evidence is quite strong: https://randomcriticalanalysis.com/2015/06/18/on-sat-act-iq-and-other-psychometric-test-correlations/ https://randomcriticalanalysis.com/2015/06/18/on-sat-act-iq-...
- austincheney 6y ago
  > The SAT – ASVAB correlation is quite strong and it is probably weakened by range restriction effects. People with lower IQ scores and/or lower HS GPA are much less likely to sit for the SAT. That's a bullshit guess. More importantly that study did not account for performance disparities versus preparation.
  - comp_throw7 6y ago
    Putting aside the question of the selection effect (which, frankly, seems like it could only realistically go in one direction, unless you want to propose a plausible mechanism by which people with higher than average IQ are _less_ likely to sit for these standardized exams)... You could also spend 5 seconds Googling the question and find that that this is not exactly an untrod ground in academic literature. The relationship is strong and well-established, far more so than almost any other result from the social sciences. Also, yes, test results, can, in fact, be improved by studying (or other things!). This does not mean that they are not strongly correlated to IQ. Reality is generally not monocausal. I don't see how this contradicts the original claim.
    - austincheney 6y ago
      > This does not mean that they are not strongly correlated to IQ. It does not mean they are related or that they aren’t, which only suggests you are missing the point: selection bias. > You could also spend 5 seconds Googling the question and find that that this is not exactly an untrod ground in academic literature. It has been studied exhaustively. The relation between IQ and academic performance is poor according to research from the APA in addition to the personal opinions of people who perform college selection professionally. That correlation is better than a random guess but isn’t anything close to alignment. People, despite the evidence, need this to be true. After all it is a money making industry. Standardized college selection tests are like tax preparation, an industry that can be easily eliminated by a more simplistic unbiased and more effective tool. With enough effort to the contrary the arguments begin to sound like a self-sustaining fraud built around a fictional necessity and a bunch of bandwagon fallacies.
      - comp_throw7 6y ago
        We aren't talking about the relation between IQ and academic performance (although that is also clearly established to be significant). We're talking about the relationship between IQ and ACT/SAT scores. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6963451/ https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6963451/ https://www.psychologicalscience.org/pdf/ps/Frey.pdf https://www.psychologicalscience.org/pdf/ps/Frey.pdf I certainly don't "need" it to be true - things are either true or they aren't, and needing or wanting differently doesn't change that. "That correlation is better than a random guess but isn’t anything close to alignment." Do you want to make a more specific claim? Again, the original claim you're attempting to rebut is "Many test scores (e.g. ACT) are heavily IQ/g-predictive." How strong does the effect need to be? Since, again, nobody is claiming that it's 100% predictive, merely a significant factor. Do you want to give me an effect size? Keep in mind the typical social psychology effect size comes out to r usually between 0.15 and 0.3: https://psyarxiv.com/2epc4/ https://psyarxiv.com/2epc4/ https://jenni.uchicago.edu/Spencer_Conference/Representative%20Papers/Richard%20et%20al,%202003.pdf https://jenni.uchicago.edu/Spencer_Conference/Representative...
        6y ago
FabHK 6y ago
> Many university recruiters at some of the most prestigious schools believe otherwise. Is there evidence for that? As has been pointed out, some national Mensas accept standardised tests as a substitute for an IQ test. For references, see here eg.: https://www.sciencedirect.com/science/article/abs/pii/S0160289606000535?via%3Dihub https://www.sciencedirect.com/science/article/abs/pii/S01602... Wikipedia: Some measures of educational aptitude correlate highly with IQ tests – for instance, Frey & Detterman (2004) reported a correlation of 0.82 between g (general intelligence factor) and SAT scores; other research found a correlation of 0.81 between g and GCSE scores, with the explained variance ranging "from 58.6% in Mathematics and 48% in English to 18.1% in Art and Design". https://en.wikipedia.org/wiki/Intelligence_quotient#Social_correlations https://en.wikipedia.org/wiki/Intelligence_quotient#Social_c...
- austincheney 6y ago
  The first paragraph of that Wikipedia article affirms my opinion: > The correlation between IQ scores and grades is about .50. This means that the explained variance is 25%. Achieving good grades depends on many factors other than IQ, such as "persistence, interest in school, and willingness to study" (p. 81). While that paragraph specifically mentions school grades instead a more narrow subset of certain specific tests the point still stands as affirmed by the APA. The numbers you quoted from the third paragraph allow are only about 80% aligned. That is a tiny bit better than half between perfection and a random guess. The unfortunate reality is that performance on those tests is primarily a direct correlation of time and money spent on preparation specifically for those tests. A major differentiator between those standardized tests and many general IQ tests is centrality versus assessment distribution. You wouldn't for example test creativity with a multiple choice question with only one correct answer the same for everybody. Look, most people who are from a higher income population want to believe that standardized tests are a fair reflection of intelligence, because with enough practice and preparation they can score higher and thus believe they are more intelligent than the population at large. If the population at large spent less effort preparing for the test then its value as a tool of performance compared to other people is less valid. That is selection bias.