University of Cambridge Cognitive Ability Test

Its an modern open-source and collaborative resource of IQ style questions for research studies. IQ has a lot of flaws, not the least of which is its usage outside of academic research, as a measure of job aptitude and hiring potential (spoiling its utility as a metric since the questions can be trained... setting aside its dubious usefulness as a hiring tool anyway).

It seems like ICAR is spending a-lot of effort to remain scientific, and i feel like a website like this goes against that by spoiling the test utility for future potential participants.

>i feel like a website like this goes against that

what do you mean "a website like this", HN? or the destination of the link at the top of this discussion?

The link for this discussion goes to the test on the same site that you link to.

Are you saying people need to make their way to that test from the front page of the site following particular breadcrumbs? that people from HN shouldn't go there till they're ready to participate in a scientific manner? i just don't understand your point...

IQ is limited because it only looks at one facet and it generalizes a lot. Tests that look both into ability and aptitude, and especially those who split them by domain are much more relevant and applicable to real world scenarios.

These tests serve no purpose outside of academic studdies. Distributing them for people to self-score sully their scientific value, and perpetuate their use for unsound usecases.

IQ does actually measure quite widely, but only one of many parts (the visual pattern matching part) is widely know outside of academic use. But even then it should not be used for hiring or intelligence scoring people against one-another.

If you actually want to test people for hiring, using domain specific interview questions and tests is a-lot more reasonable and usefully correlated with real-world job performance.

so you're saying you have an agenda and you're pushing it.

Psychologists are scientists and the replicability of IQ testing is extremely high and repeatedly confirmed. And despite how much psychosocially challenged nerds here like to complain about psychology, they are in good company: psychology itself is not normie opressors, psychologists are also psychosocially challenged nerds.

before you say anything else, the statistical methods we use today across medical testing were first applied and developed to psychometric testing, so if you are going to attack that, you are attacking all of medical science.

>IQ has a lot of flaws, not the least of which is its usage outside of academic research, as a measure of job aptitude and hiring potential

Isn't IQ one of the best predictors of job training success, across both civilian and military, blue collar and white collar?

It's also one of, if not the single most generalizable predictors that we know of right now, even more so than nationality, race, gender, SES (socioeconomic status), parental SES, you name it. It predicts just about everything - from hard biological measures like reaction time and brain mass to lifetime odds of being in a car accident (distinct from causing a car accident - higher IQ people are statistically less likely to be hit by another driver), divorce rates, lifetime income, longevity, the list goes on and on. IQ is not the strongest predictor for every one of these, but every stronger predictor for any one of those fails to predict as many things as IQ does. Parental SES, controlling for IQ, provides no predictive power for your reaction time, for instance, despite predicting educational attainment better than IQ does.

The critique that IQ is an imperfect proxy for g is totally valid.

The self-assuaging fantasy that g itself doesn't exist is a classic example of a psychological defensive mechanism of rejection, one rooted in a need to defend a worldview that holds all people as inherently equal, when we're measurably, biologically not.

IQ is meant for population statistics. For that purpose it's a great scientific tool.

> socioeconomic status

No. A real IQ test tries to cancel out educational level from the score by comparing people in buckets of age, education and a few other importance factors.

A systematic deviation of education quality in the same "level" is not possible to cancel out, making IQ indirectly measure socioeconomic status. Hence why the us governments banned IQ on the basis or racism for government hiring.

You cannot measure two prople with 1/4th of a IQ test (logic puzzle is only one part of a full test) and make any useful statistical conclusions. A domain specific interview question or aptitude test a much clearer value to hiring prococess.

Im honestly not sure what you're trying to get across here. IQ is a dated but good scientific tool to measure cognitive differences across large samples of humans.

I never said they're not scientific. They're invaluable statistical tools. I'm saying their use outside of academia for hiring (among other things) is a fundamental missuse of both the tests and statistics.

Online versions of ICAR tests like these serve little purpose for anyone, and at worst encourage people to abuse ICAR in a similar way.

> The self-assuaging fantasy that g itself doesn't exist is a classic example of a psychological defensive mechanism of rejection, one rooted in a need to defend a worldview that holds all people as inherently equal, when we're measurably, biologically not.

I think this statement conflates two different senses of the word “equality”. Equality of abilities is different from moral equality. It is perfectly coherent to accept that people aren’t equal in terms of their abilities but are still morally equal. For example, just because Person A is smarter than Person B it does not follow that the interests of Person A matter more than those of Person B, or that the suffering of Person A matters more than the like suffering of Person B, etc. So the view that g is real and people have different IQ scores is consistent with the idea that all people are inherently equal. Because in most contexts the concept of inherent equality is not a biological or psychological concept but a moral concept.

I don't disagree with moral equality one bit - the golden rule should absolutely apply to everyone and we should all strive to look at each other with compassion, tolerance, empathy, understanding, grace, humility, and goodwill - but there is a loud, vocal subset of people that truly believe in absolute biological equality - not just between people with different IQ's, but even between different biological sexes and all other categories of humans; "tabula rasa" proponents who argue that ALL differences in outcomes along the lines of categorical differences (e.g. sex, race, gender, ethnicity, nationality, religion, culture) are exclusively and irrefutably explained strictly by discrimination and discrimination alone, which is a patently absurd assertion that should be refuted. Discrimination is real and should be confronted vocally, but the idea that it's the only factor explaining differences in outcomes between groups is a harmful myth.

The fact of the matter, relating back to the original discussion, is that sex/race/gender/ethnicity/nationality/religion/culture-blind IQ testing is not only a strong predictor of job performance, it is perhaps one of the best tools we have for eliminating discrimination based on sex/race/gender/ethnicity/nationality/religion/culture in hiring, as it explicitly controls for differences along these lines by exclusively targeting an assessment of g in abstract ways that are explicitly stripped of cultural, religious, racial, and gendered biases.

Pseudonymized hiring that relied exclusively on IQ tests, with zero indications of race/sex/gender (e.g. legal name), stripped of proxies for SES and/or parental SES (e.g. which university was attended, if any) would be significantly less biased than current hiring practices. Throw in job-specific pseudonymized skill evaluations (so, no voice calls, no video calls, just direct assessments to candidates) and you've got a system to dramatically reduce hiring discrimination along protected classes.

No, I think he's closer to the science than this comment is. IQ has validity for its intended purpose of as a diagnostic tool for cognitive deficits. It is in fact not a valid tool for casually ranking ability based on individual tests, both because of test-retest consistency and because ordinality of IQ breaks down as scores get higher.

I think you could have written a reasonably strong argument that the parent commenter overstated their case, but you didn't do that: you just overstated the opposite pole of the argument and then declared the discussion off limits due to politics.

If IQ was one of the best predictors of job training success, it would be used everywhere. Instead, very few companies use it. There's a persistent myth that IQ isn't more widely used because of legal concerns, but that can't be right, because IQ and general cognitive tests are used by several of the largest companies (with the deepest pockets for discrimination suits) in North America --- and, further, the Griggs v. Duke jurisprudence that roots the myth doesn't have any force outside the US, where... IQ testing is not generally used in employment.

I don't know what the "self-assuaging fantasy" is supposed to mean, but you can read Cosma Shalizi to see how any set of tests structured like IQ tests are necessarily give rise to a "g" fact, even if you randomly generate them. I feel like I don't have to assuage myself too much that math works.

... or you can just discard all the hokum and only use job-specific pseudonymized skill evaluations --- a concept more widely known as "work samples" --- and be using the gold-standard tool for candidate selection, supported by research going all the way back to Deming.

... no it isn't? IQ tests were explicitly designed as individual diagnostics. They're clinical tools.

From where did you get the idea that IQ was meant for population statistics?

>ordinality of IQ breaks down as scores get higher

ordinality of any sample from a Gaussian process breaks down as the thin part of the population density is approached, but that will only affect those at least three sigmas above or below, so it's an edge case that doesn't matter for the fat part of the curve.

you are cherry picking pieces of arguments that might threaten your "nobody is better than anybody else" anti-merit ideology. "From those who have the ability to those who have the need" is a merit-based sickle, comrade, and IQ is its hammer, is my argument.

>I think you could have written a reasonably strong argument that the parent commenter overstated their case, but you didn't do that: you just overstated the opposite pole of the argument and then declared the discussion off limits due to politics

I don't think you really think any of that. You are disingenuous if you claim that your (collective (in more than one sense of the word)) side wants to actually discuss the science of IQ; you simply want it removed from the discussion because it does not support your politics. If the economy rewards high IQs, and people have social mobility, you will see an economic sorting from top to bottom, and you don't like that. If IQ has genetic factors (which it seems to) that's a bigger disaster because once the economic sorting takes place, social mobility goes away (the latter half of the 20th century) A true scientific socialist would not try to imagine/censor these problems away in a central committee, but would admit them and try to figure out better ways.

I talked about the science of IQ in my comment something that you did not actually engage with. the correct wording would have been for you to say "g does not exist" and provide some basis for that belief

One way I know that I didn't say "nobody is better than anybody else" is that isn't something I actually believe. I think you have me confused with someone else.

(you edited your comment and added a paragraph to it after I wrote this)