3 ms·
Anyone else find it strange that the author used SAT scores as a metric for intelligence? I think it is kind of absurd to think that a test that is specifically
by godelski 9y ago
Anyone else find it strange that the author used SAT scores as a metric for intelligence? I think it is kind of absurd to think that a test that is specifically not designed to test intelligence and designed to give trick or misleading questions is a good metric. Not to say there isn't a decent correlation between intelligence and test scores, but a significant amount of people fall through the cracks. From what I understand, even the academic communities don't place a lot of weight on tests like SAT, ACT, GRE, etc. I think it is fine to use this method to measure intelligence, but not in isolation.
- skybrian 9y agoThe survey asked for both SAT and IQ scores, but both questions were optional.
- dragonwriter 9y ago> Anyone else find it strange that the author used SAT scores as a metric for intelligence Since SAT scores are very highly correlated with IQ test scores, and they are often more available, no, not at all. I'm much more concerned with how the sample (if it can even be called that) was generated, which kind of makes the whole thing worthless. The pointers to peer-reviewed studies (even though almost certainly cherry-picked) are more useful than the original “research” reported.
- godelski 9y agoLike I said, there is a correlation, but I definitely would not argue that it is "very highly correlated". I found one paper by Frey [1] that discusses a significant correlation, but you will see that it is definitely not the "very highly" that you claim. I only gave a quick glance of the article, but I'm not even sold on its results because their methods suggest they might be over fitting the data (them just trying different fit methods to yield better results feels sketchy. Maybe they back it up better than I saw over the quick glance). That all being said, correlation itself is not a de facto proof nor does it mean it is a good metric. There is a 99.8% correlation (which I'd call "very highly correlated") between US spending on science/space/technology and deaths by strangulation/hanging/suffocation[2]. I think we can all see that it would be laughable to use one as a predictor of the other. Yeah, this example is ludicrous and obvious, but the difficult part is when you expect things to be dependent (which is what we are seeking for a good metric, not correlation), but aren't. It is easy to find data to match the results you are looking for, but that doesn't mean they are good results. [1] https://www.psychologicalscience.org/pdf/ps/Frey.pdf?origin=publication_detail https://www.psychologicalscience.org/pdf/ps/Frey.pdf?origin=... [2] http://www.tylervigen.com/spurious-correlations http://www.tylervigen.com/spurious-correlations
- savanaly 9y agoI'm confused what a "sample" group means in this context. Take people who have one and only one sibling: For the studied population (people who chose to take the SSC survey) the % of people who were the older sibling was some number. The "sample" to compare that to is...what? The human population as a whole? Then we don't need a sample group, we know it has to be 50%. So if the number he got in his study is much different from 50% that says something about the population in the study being different.