True story
IQ is largely a pseudoscientific swindle
111–120 of 400 posts
Re: IQ is largely a pseudoscientific swindle
#112I have no way to navigate this, I've heard psychologists claim that IQ is the most well proven concept in the social sciences. Now this guy comes along saying the complete opposite. Anyone have any advice on how to weight this up and figure out what the truth is?
The thing to remember is that psychology is a field that has a well-worn history of being driven by heavy misinterpretation of flawed experiments, and psychometrics in particular tends to vary from the awful to the truly execrable in this regard. IQ and the Big-5 personality tests are the best of the field, but we're not talking Standard Model-level of verification--these theories are the nearsighted one-eyed kings of the land of the blind.
The current consensus among psychologists is that intelligence is not driven by one single factor. Rather, there are separate kinds of intelligence (e.g., crystallized and fluid intelligence), and there may be a master throttling general intelligence factor called g. As psychology came around to this viewpoint, IQ tests have been retooled to measure g... except g can't be directly measured. So g is teased out of intelligence tests by statistical means as the primary factor of correlation, and the fact that there does exist a mildly strong correlation is taken as evidence that g exists. (And if you're saying to yourself that correlation is not causation and that this argument is circular, well, there are plenty of psychologists, if a distinct minority, who would agree with you on that point).
Most psychologists, I expect, would prefer to talk about intelligence in terms of the subcategories such as crystallized and fluid intelligence, for which there is stronger evidence and more readily accessible metrics. (Most IQ tests also spit out these scores as well, but it's the IQ part that is the most widely broadcast).
Re: IQ is largely a pseudoscientific swindle
#113Earlier quoted context omitted.
The problem is not whether the answer is binary. The problem is that the author of the test needs to have at least the same level of intelligence (and probably even higher) than what the test is measuring. Otherwise how can you tell if the answer is correct? So to get a good IQ score one needs to be able to figure out what a test designer (with average IQ) thought the correct answer should be.
> The problem is that the author of the test needs to have at least the same level of intelligence (and probably even higher) than what the test is measuring. What makes you believe that?
I gave an example above. Imagine you ask second graders to write math tests for college. They might extrapolate from their experience like this "We started working with single-digit numbers, like 2+3 or 4*2. Then we learned to work with two-digit numbers, 23+17 or 17-73. High school math should probably be three-digit numbers, college will be four-digit numbers, and math professor must be able to work with eight-digit numbers."
This is actually a reasonable approach (extrapolating from known data), but do you think this test will separate math ability well among college students?
Re: IQ is largely a pseudoscientific swindle
#114Earlier quoted context omitted.
Taleb is barely this side of being a troll: he has plenty of interesting tidbits to say and is obviously a smart guy, but read enough of his stuff and he's clearly more focused on generating heat instead of light. If you read through the article, he's not quite making the claim that one might assume from the headline: which is basically that IQ isn't much of a useful or well-defined metric.
It did seem like a lot of the article was making claims about the motives of IQ supporters. That seemed like a bit of a red flag to me, that this was more about politics than science (not that either exists without the other). Thanks
In both politics and science, attacking your opponents' motivations is fallacious. The merit of a position should be judged independently of the personal failings of whoever puts it forward.
Re: IQ is largely a pseudoscientific swindle
#115I have given dozens of ravens-progressive-matrices variant tests to friends of mine of various intellects and the results are basically never surprising.
I guess it's bullshit though.
Third world country girl who struggled trying to learn computer science and gave up and became an accountant? 95
Kind of slow guy that can't do any algebra in his head, bronze 1 at league of legends? 98
Genius guy that is master tier in league of legends and constantly outwits everyone in banter? 143
Asperger sharp awkward guy that is astonishingly good at math - diamond 3 league of legends? 133
Me: I think I'm an idiot sometimes and smart other times. Platinum 1 at league of legends? 122
My older brother (1): Obviously smarter than me but not by that much seemingly, and definitely a lot physically healthier than I am: 130
My oldest brother (2): National bridge champion and obviously very intelligent, masters in math, been an alcoholic for 5 years and got noticeably less sharp overtime: 135
Another guy who was challenger tier at league of legends and also obviously very intelligent in conversation? 138
Smart younger guy that is kind of whacky and out there, diamond 3 at league of legends: 120 (deviates from the norm here but still lends evidence to the correlation)
Extremely competent native american guy that is 2000 ELO in Chess and was diamond 2/3 in league of legends? 120 (while high though, and same caveat as the above example as the other example just above)
hMmmMmmmMmMMMMMM I guess it is just confirmation bias.
I found it interesting how well it correlated with rank in league of legends. There are about 5-6 other samples I just don't recall the specifics anymore.
Re: IQ is largely a pseudoscientific swindle
#116The title doesn't seem to match up with the article. The point seems to be that IQ is good at predicting intellectual deficits at the low end but bad at predicting 'success' above that; further it indicates that IQ is mis-used in this range. This doesn't seem to indicate that IQ is 'pseudoscience', just that it's a misused statistic. Not a real shocker there, almost every statistic is misused by some group.
Taleb is saying that if IQs above 100 have no predictive power, then what’s the point in measuring them? It would be like taking a BuzzFeed quiz that tells me my Hogwarts house: a real measurement yielding a reproducible result, sure, but a pointless one and any use of it beyond “Huh, that’s interesting” would be pseudoscience.
Re: IQ is largely a pseudoscientific swindle
#117I have no way to navigate this, I've heard psychologists claim that IQ is the most well proven concept in the social sciences. Now this guy comes along saying the complete opposite. Anyone have any advice on how to weight this up and figure out what the truth is?
The thing to remember is that psychology is a field that has a well-worn history of being driven by heavy misinterpretation of flawed experiments, and psychometrics in particular tends to vary from the awful to the truly execrable in this regard. IQ and the Big-5 personality tests are the best of the field, but we're not talking Standard Model-level of verification--these theories are the nearsighted one-eyed kings o…
>Most psychologists, I expect, would prefer to talk about intelligence in terms of the subcategories such as crystallized and fluid intelligence, for which there is stronger evidence and more readily accessible metrics. (Most IQ tests also spit out these scores as well, but it's the IQ part that is the most widely broadcast).
When talking about IQ, i.e. 'g', this incorporates both crystalized and fluid intelligence. So to say that there's stronger evidence for these vs IQ doesn't seem to make any sense.
Re: IQ is largely a pseudoscientific swindle
#118Missing a bit of sociological historicizing here.
IQ and SATs were essentially designed to keep undesirables, notably Jewish, people out of academic circles. Eugenics and these tests marched hand in hand.
> Armed with this so-called objective methodology, 3 American eugenicists advanced a straw-man rationale for large-scale testing.14 They reasoned that society needed to identify, segregate, and sterilize the "feeble-minded,"' 5 initially defined as those with mental disabili- ties' 6 but later extended to include any "unfit" person of low intelli- gence, character, or ethnicity.17 In both Germany and the United States, persecution of the "feebleminded"'18 hastened a broader eu- genic campaign against immigration, miscegenation, and other pro- fessed threats to Nordic ascendancy.1 9
https://via.library.depaul.edu/cgi/viewcontent.cgi?referer=h...
https://www.businessinsider.com/the-ivy-leagues-history-of-d...
https://twitter.com/ft_variations/status/1075782671320657920...
Re: IQ is largely a pseudoscientific swindle
#119Earlier quoted context omitted.
Wiki says that the average IQ of MDs, JDs, and PhDs is 125.[1] That is pretty far away from 70. [1] https://en.wikipedia.org/wiki/Intelligence_quotient#Other_ac...
From the article: > Note that IQ tests correlate with SAT scores! That is (in part) Nassim Nicholas Taleb's point: we use IQ to select for success, then say IQ is correlated with success. It is circular reasoning. The more interesting measure is not jobs for which IQ is the main selection criteria to qualify for the training (your examples of MDs, JDs, and PhDs), but things where people create their own impact, lets…
> > Note that IQ tests correlate with SAT scores!
IQ tests (of which there are a ton) are regularly calibrated, internationally comparable (there are some reasonably culturally neutral test protocols) and stable while SAT is a US-only metric.
Maybe SAT is calibrated to correlate with IQ tests?
Re: IQ is largely a pseudoscientific swindle
#120Earlier quoted context omitted.
From the article: > Note that IQ tests correlate with SAT scores! That is (in part) Nassim Nicholas Taleb's point: we use IQ to select for success, then say IQ is correlated with success. It is circular reasoning. The more interesting measure is not jobs for which IQ is the main selection criteria to qualify for the training (your examples of MDs, JDs, and PhDs), but things where people create their own impact, lets…
If the SAT test wasn't selective in some meaningful way then people admitted through affirmative action should be expected to do approximately as well as the people that was admitted purely through merit. The outcome says otherwise [1], or if prefer a video [2]. Mismatching people to schools based upon an ideal hurts those people when they come unprepared relative to their peers. Many diverse people admitted through…
Focussing on "affirmative action" feels uncomfortably close to the exact kind of dog whistling this article criticized up front.
As I understand it, most colleges have been using some form of linear regression model on many factors to filter admissions in a way that maximize returns (i.e. alumni donations) for decades now. Surely there is other evidence which shows that if SAT scores were actually a poor heuristic of potential success, schools would have stopped relying on them by now. (without having to bring race into the conversation)