JP writes:

“IQ has been measured more accurately and more tightly defined than any other social science or medical construct, but the implications of the vast differences in individual intelligence are not palatable to many people. So, now and then, someone not very well informed or ideologically possessed invents a new king of  intelligence, and a large audience seizes upon the idea as something that will save them from admitting to what they find conceptually unacceptable. A decade after this happens, the new kind of intelligence disappears from view, as its lack of utility reveals itself.

EQ was invented by a journalist, not a psychologist; most of the research he reviewed described prefrontal cognitive function, not emotion. Furthermore,  EQ scales measure the Big Five trait agreeableness, insofar as they measure anything, and agreeableness is already very well defined, and does not need any help from EQ. Agreeable people are warm, empathic, compassionate and polite, but they can be taken advantage of quite easily. For this reason, EQ is negatively correlated with managerial performance, rather than positively correlated (although the correlation is low).

Howard Gardner publicized ‘multiple intelligences’ in 1993 — although the word ‘talent’ was already working fine. None of his claims stood the test of time; his intelligences could not be measured. Robert Sternberg tried something similar with ‘practical intelligence’; the same fate befell his work.

Here is what IQ is: imagine you have an infinite bank of questions that require abstract reasoning to solve. Imagine that you pick out ten sets of ten questions from this bank, and give all ten sets to each of a thousand people. People who score highly on one set will tend strongly to score highly on all the rest. In other words, the rank order of performance will remain constant across all sets.

That rank order will also predict phenomena such as rank order of creative achievement, school grades, speed of learning, and competence at complex jobs — and it will do it remarkably well (with r’s of about .5, which is a larger effect than that reported in 95% of social science studies). It will also predict such things as brain size, thickness of neuron, and reaction time consistently, as well, but with lower power (r = .10-.15).

Furthermore, that rank order will be singular, in that it will not be decomposable into different abilities, particularly at lower levels of the rank.

Finally, IQ cannot be merely dismissed as a ‘statistical artifact.’ Although it is generally derived from factor analysis, at least in its purest form (g), which is the first factor derived from such analysis, a given person’s average score on ten sets of ten questions requiring abstract reasoning to solve will correlate very highly with that factor.

In other words, average performance across sets of questions that require abstract reasoning to solve is IQ, and IQ is as real as any average” (JP).