3 ms·

First its worth baring in mind that actual IQ test requires a specialist and takes a long time, so most will use something that correlates with IQ scores instea

First its worth baring in mind that actual IQ test requires a specialist and takes a long time, so most will use something that correlates with IQ scores instead. Simple counter argument is that IQ tests are not domain specific enough and not causative of domain specific capability even where complex reasoning is involved. Therefore, you should test closer to the role. E.g. for developers, a ranking of candidates by LeetCode scores will not be the same as the ranking of candidates by Raven progressive matrices -- they may correlate, but the differences are important and -- arguably -- one is more job specific than the other.

kypro 3y ago
> First its worth baring in mind that actual IQ test requires a specialist and takes a long time, so most will use something that correlates with IQ scores instead. This isn't an issue with IQ testing, it's an issue with testing generally. You'll find most objections to IQ testing can be applied to practically every form of testing. Also, IQ tests don't require a specialist, nor do they need to take a long time. Only the most accurate IQ testing would generally require these things. Normally when hiring you're just trying to get a rough sense of a candidates abilities. > Simple counter argument is that IQ tests are not domain specific enough and not causative of domain specific capability even where complex reasoning is involved. Therefore, you should test closer to the role. I agree. For most roles I wouldn't be in favour of just measuring IQ – and I did touch on the limitations of IQ in my comment. But regardless of its limitations it's still true that individuals with higher IQs tend to learn faster and be more intellectually capable in general. I could say something like, "LeetCode scores just selects for software engineers who can write a certain type of code, but doesn't measure one's ability to play well as a team or one's ability to view problems at higher levels of abstraction". The right thing to test depends entirely on what kind of abilities you're selecting for. If you need a software engineer to be adaptable and broadly capable then IQ is probably a good metric to consider (among others). And in my opinion instead of implicitly creating a selection process which favours the next white guy with a degree and 10+ years experience as a software engineer, it would be more equatable to consider people who show strong potential but perhaps have less experience.

kypro 3y ago

> First its worth baring in mind that actual IQ test requires a specialist and takes a long time, so most will use something that correlates with IQ scores instead. This isn't an issue with IQ testing, it's an issue with testing generally. You'll find most objections to IQ testing can be applied to practically every form of testing. Also, IQ tests don't require a specialist, nor do they need to take a long time. Only the most accurate IQ testing would generally require these things. Normally when hiring you're just trying to get a rough sense of a candidates abilities. > Simple counter argument is that IQ tests are not domain specific enough and not causative of domain specific capability even where complex reasoning is involved. Therefore, you should test closer to the role. I agree. For most roles I wouldn't be in favour of just measuring IQ – and I did touch on the limitations of IQ in my comment. But regardless of its limitations it's still true that individuals with higher IQs tend to learn faster and be more intellectually capable in general. I could say something like, "LeetCode scores just selects for software engineers who can write a certain type of code, but doesn't measure one's ability to play well as a team or one's ability to view problems at higher levels of abstraction". The right thing to test depends entirely on what kind of abilities you're selecting for. If you need a software engineer to be adaptable and broadly capable then IQ is probably a good metric to consider (among others). And in my opinion instead of implicitly creating a selection process which favours the next white guy with a degree and 10+ years experience as a software engineer, it would be more equatable to consider people who show strong potential but perhaps have less experience.