Intelligence: A Very Short Introduction is a short booklet by Oxford University Press that summarizes known results on intelligence. This is useful because so much of the discussion on IQ and mental abilities tends to be very politically motivated, and focus more on ideology than data.

Here are some conclusions from the book:

Interesting conclusions

  • All mental abilities tend to be highly correlated, looking at hundreds of different studies on IQ tests. This correlation is known as the general factor or “g” and accounts for about half of intelligence in any specific area. In other words, it is meaningful to speak of a single intelligence.
  • The vast majority of things that tend to affect cognitive abilities affect g uniformly. For example aging tends to affect all mental abilities evenly.
  • The best ways to stay smart in your old age are to not have a heart attack, live in an upperclass area, marry somebody smart, and do interesting/stimulating work.
  • There’s a moderate correlation between intelligence and reaction time.
  • When discussing “nature vs. nurture” there are really 3 factors. Genetics, shared environment (the environment that sibilings in the same family might experience), and unshared environment (the environmental factors unique to each person). IQ seems to depend overwhelmingly on genetics, followed by unshared environment, with almost no influence from shared environment.
    • Twins separated at birth have about as much variation in their IQ scores as the same person taking an IQ test multiple times! A .7 correlation if separated early, and a .8 correlation if raised together.
    • In contrast, adopted children raised by the same family have almost no correlation.
    • “Genetics” does not necessarily mean heredity. While the correlation between identical twins is very high, the correlation between children and parents is lower (though still significant). The correlation between children and birth parents is about .3, between children and adoptive parents is about 0.
  • IQ tests and work-sample tests are each highly correlated (~.5) to work performance. IQ tests matter more the more complex a role is. IQ and work-sample tend to overlap, so if you’re using two tests use IQ + Integrity or work-sample + integrity. In the USA IQ tests are illegal, so work-sample + integrity tests are the best predictors of job performance (much better than standard interviews).
  • Scores on intelligence tests have been rising rapidly over time all across the world. No one knows why-these data would suggest that an average person in 1945 would be considered mentally challenged today, which is hard to believe.

Fun stuff!