‘Coachability’ does not refute IQ

I saw this going viral:

Under the timed conditions of an IQ test, eliminating false patterns or noise is necessary, so I don’t see how these are mutually exclusive. They both matter. All too often, these criticisms of IQ or of ‘IQ tests being defective’ are already accounted for by existing IQ tests or proxies, or can be controlled for. There’s nothing new. These IQ ‘owns’ are easily refuted.

A common misconception is that IQ and environment must always be at odds with each other, or that it’s one or the other. As I discuss in the post Trace – Freddie Enrichment and IQ Debate: Why Trace is Wrong, the influence of environment does not refute IQ; instead, it only heightens or reveals innate differences of ability, as it’s yet another variable that is controlled for.

For example, a lot of high schoolers are coached for high-stakes math competitions, so this becomes yet another variable which is controlled for, so the smartest competitors still end up winning because the effect of coaching is simply held constant. Even in top private schools in which SES is very high across the board, there is still huge variance of individual academic ability. If the mean IQ of an elite private school is 110 or so–you still have a Bell Curve! It never goes away.

Another criticism is that IQ tests are coachable, as supposed evidence that IQ is malleable, is not innate, or even does not exist at all. But even if IQ tests or IQ proxies, such as the SAT, GRE, or the LSAT are coachable, it still can discriminate top IQ. Coaching has the effect of raising the mean, but this is mitigated by re-norming or ‘curving’ the test (like on the LSAT) to account for coaching, and redesigning the test (e.g. adding or changing some questions to raise the ceiling) to be harder.

If we assume that people are getting better at taking the LSAT due to coaching, then to get a 170 may entail only getting 10 questions wrong instead of a more forgiving 15-20, but you still have a Bell Curve in terms of the number of questions right/wrong (the raw score), which is then converted to give the point score (120-180). So, coaching raises the raw score, but then this is mitigated by converting to the point system. Thus, despite the LSAT test prep industry, which has ballooned over the past decade, scores above 170 are uncommon and predictive of high IQ.

The downside is this puts people who are not coached at a disadvantage, but LSAT coaching and prep has become ubiquitous now, so this is not as a valid of a criticism anymore. It possibly also makes the LSAT less effective at discriminating top talent, as guessing or luck may play a greater role if fewer mistakes are tolerated to allow for a top score. This means having to periodically redesign the the test, usually to make it harder or less coachable in other ways.

This is also why the Flynn Effect means an IQ score of 100 today may be equal to 120+ half a century ago; the modern environment may have the effect of making people better at IQ tests, requiring re-norming and re-designing of the tests to be harder.