Weird Internet Ideas: How Smart is Vox Day?

And do we actually care? Not really but it is an interesting lead in to look at some utter piffle said about IQ scores.

In the world of the alt-right, the IQ has a kind of magical quality. The quasi-academic research into inter-country IQ scores and the toxic legacy of the race/IQ debate taps into a lot of the alt-right’s areas of interest and also provides a handy smoke screen of numbers for them to assert as if they had scientific rigour.

I don’t think IQ is nonsense – if anything I’m probably less sceptical about IQ than many on the left but that doesn’t mean that it does anything that the alt-right claims for it. However, talk about IQ often plays an interestingly ironic role – it frequently shows up a degree of ignorance and innumeracy on the part of the person doing the talking.

Frequent examples among the alt-right observable in the wild at Vox Popoli, are references to how many ‘standard deviations’ a person is than others. Typically this is cited as put down of a supposed SJW and trumpeting of the superior intelligence of the alt-right minion.

There are some delicious examples in a recent comment thread. The context is Vox Day being upset because a more knowledgeable person elsewhere on the net disputed his analysis of a battle scene on Game of Thrones. Now note, the question at hand is not measures of emotional maturity but ones of intelligence. The quote from Vox is this: “I’m now convinced that his IQ doesn’t come within 50 points of mine”.

I’m not convinced of that – in fact, I’m quite confident that the claim is not really meaningful.

Modern IQ tests work by centring on a median score of 100 and using standard deviations to map out a scale of sorts. It isn’t a measurement/interval scale as such, a one point movement on one part of the scale does not necessarily represent the same change in intelligence/whatever as another point. However, it does describe relative positions across the notional population it is normed against. Typically 15 IQ points represent one standard deviation. So “50 points” represents more than THREE standard deviations.

Some things to note before we continue:

• IQ scores are scores derived from and defined by tests
• IQ scores are limited to what those tests can deliver
• IQ scores, like any test score, have a degree of error

IQ scores gain a degree of validity in part by their correlations with other things. However, the further you go away from the median score (100 points) the fewer people any such correlations are being compared with. In addition, in a purely statistical-measurement sense, the error margins for any test score increase round very low or very high test scores. In simplistic terms, the information a test gives is based on the mix of right versus wrong responses. One extra right or wrong answer for a person near the median score can have less impact on their IQ score than one extra right or wrong answer for somebody close to 0 correct answers or close to 100% correct answers. Finally, the compatibility of IQ tests is less good at extremes – a score of 100 is pretty much intended to be the same thing on any reputable test (~ ish) but the further you get away from that, the less those scores mean the same thing.

Looking at the top end of an IQ score and you’ll find some variation between bands which tests regard as meaningful ( https://en.wikipedia.org/wiki/IQ_classification ). However, picking Stanford-Binet or something like MENSA’s test and it is fair to say that being as generous as possible to the claims of IQ testing an IQ score of 160 is really pushing it in terms of an upper score of any kind of meaningfulness.

For an interesting point of comparison, all round very, very smart person Richard Feynman once scored an IQ of 125 points on a school IQ test. You’ll find online various discussions of why that may be (specifically that the test may have emphasised verbal skills rather than mathematical ones – which is a whole other kettle of fish). However, it isn’t a particularly odd finding. All the various kinds of error around IQ testing, from classic measurement error to the number of people the score is being compared against, to the fundamental relation between the content of the test and what it is claiming to be testing, increase sharply the further away from the median score you get. Additionally, the higher or lower your IQ score is compared with that median, the less TYPICAL you are. Atypical people aren’t going to have particularly valid IQ scores.

In Vox Day’s case, his claim is this: the difference in IQ score between man-who-made-Vox-grumpy (MWMVG) and Vox is >50 IQ score points. If we assume the MWMVG is at least in the average range (90-109) Vox is claiming an IQ score of >140 and possibly >159. Note that the upper end of just ‘average’ IQ has Vox claiming to be pretty much at the limit of meaningful IQ scores on the most generous reading of IQ and even at the lower end well above the boundary which most reputable IQ test stop bothering to classify (around 130 IQ points). An informed (and presumably smart) person shouldn’t make a claim any more precise than ‘greater than 130’ – beyond that the figure as some sort of intrinsic property of a person that would be consistent across multiple methods of quantification doesn’t make sense EVEN ASSUMING IQ MAKES MUCH SENSE ANYWAY.

Put let’s take that figure of 130. Let’s say Vox is taking a more grounded view of his own IQ and is seeing himself as 130. A 50 point difference would put the MWMVG at an IQ of below 80. For comparison, an IQ of below 70 is used diagnostically as evidence of intellectual disability. An IQ of 80 to 70 is likely to represent somebody who would struggle with school and many cognitive tasks (assuming the score was representative). Which would be an odd thing for Vox to claim – after he is attempting to write a point-by-point rebuttal of what the MWMVG and struggling to do so, claiming that he is struggling to counter an argument from a person with an IQ lower than 80 would be tantamount to claiming he really doesn’t know what he is talking about.

Of course, there is a simpler analysis. IQ is about general cognitive ability. Arguments about how a fictional battle works on a TV show does involve a range of cognitive ability but only somebody utterly clueless about IQ would think it was a sensible way of judging a disparity between IQ of the two participants (except of course in extreme cases).