What goes with IQ

IQ is glosso IQ: the IQ test's own scoring of its reasoning puzzles, wherever on glosso they were answered (the IQ test, the Cognition Genre survey, the daily drip, and the retired Raven's drip puzzles), plus its small adjustment for answering promptly. Only people with or more scored puzzles count, so each score is reasonably precise. 100 is the median glosso taker, not the general population; glosso takers likely sit above the population average.

Scales (agree–disagree, never–always, turn-off–extremely) are correlated with IQ directly. Typed-in numbers use rank correlation, since they run long-tailed. Choice questions are split into their options: each option is picked-or-not against IQ, and the list shows the option that goes with IQ most strongly.

r runs from −1 to 1. Around 0.1 is small, 0.3 is moderate and clearly visible in the averages, 0.5 is large. The whiskers are 95% confidence intervals. Reliable means a false-discovery rate under 5% after correcting for the tests run here (q < .05), so a handful of the reliable links are still expected to be flukes.

Hold sex constant compares each person with others who gave the same gender answer, since men score about points higher here on average. For most questions it barely changes anything.

Each question was answered by a different subset of people ( at least), so they can't be compared as if from one sample, and none of this says what causes what. The breakdown under each question shows the average IQ for each answer with a 95% interval; answers given by fewer than people are left out.

Everything here is an aggregate. No individual's answers or scores are shown, and the page rebuilds daily from the live data.