Everything we have written
Newest first. Two kinds of piece: the ones that explain a thing your brain does, and the ones that explain why two tests of the same thing give you different numbers.
This page exists because until today there was no way to see what we had written. An article was reachable only from its own pillar, and only two pillars are in the navigation — so three of these were effectively unlisted while being fully published. That is the same defect shape as everything else on the corrections page: the thing existed and the door did not.
Why your score will not match another site's
One per test. They are the most useful thing here, because a difference between two scores is nearly always a difference between two tasks — and almost nobody publishes enough about their task for you to find out which.
Is being good at mental math a sign of intelligence?
A sign of something real and narrow, and not of the thing the question means. Two factors share the word “maths”, and the one we measure is the one practice moves most.
Why can't you just ignore the word?
Stroop's 1935 experiment found the interference runs one way only — words wreck colour-naming, colours barely touch reading. Eight days of practice shrank it and never removed it.
What actually shows up at a brand-new website
Four days of edge logs from a site nobody has heard of. Almost everything that arrives is a machine, and most of them are trying doors rather than reading pages.
Does your vocabulary keep growing after twenty?
In the best study we found, the median rises steadily to seventy — about one new word every two days. Why the age tables online contradict each other.
Published estimates of adult vocabulary run from under 10,000 words to over 200,000, and the gap is mostly the definition of "word". Why we give you a count rather than a size.
Why your vocabulary score will not match another site's
Most vocabulary tests report how many words you know. Ours reports how many of twenty you got right. One is a projection from a model, the other is a count — and "language" turns out to be at least six separate factors in the reference kit.
Why your spatial score will not match another site's
The sharpest of these. The two tests are often not measuring the same ability at all — visualization, spatial orientation and spatial scanning are three separate factors with three separate sets of marker tests.
Why your symbol match score will not match another site's
Three versions of one substitution paradigm, taken by the same people, agree only partially with each other — and the window length alone moves the number.
Why your pattern score will not match another site's
What a grid task samples, what it cannot, and why the rule that makes a tile the odd one out is the whole difficulty.
Why your digit span will not match another site's
The same 763 people produced three different average spans in one sitting, depending only on which scoring rule was applied to the identical answers.
What each thing is
What a browser can and cannot measure about your reaction time
With the human removed and true reaction time fixed at 105 ms, a touchscreen returned 400 ms. The hardware added nearly four times more latency than the response it was measuring — which is why one of our tests has no clock.
What is actually being timed, and how much of a measured reaction time belongs to the machine rather than to you.
Is a pattern recognition test an IQ test?
No — and the reasons why are more interesting than the answer, because some of the tasks are genuinely shared.
What the term covers, what a grid task can and cannot sample, and why it is not a measure of intelligence.
There is an answer, and it comes with more conditions attached than the question expects. The scoring rule alone moves the average by nearly half a digit.
What it is, what genuinely moves it day to day, and what the evidence does and does not support about training it.
How our own numbers are built
What the Fingerprint number is
Every constant in the arithmetic, including the one we chose rather than derived — and why a number can exist at all when our reference sample is zero people.
The norming ladder, which rung we are on, and what a score at that rung is allowed to say.
The language we ruled out in advance, written while nothing was at stake so it could be honoured when something is.
Two pages that are not articles and are worth more than most of them.
Corrections — everything we have got wrong, with dates The instrument register — every test, its version, its rung and its reference sample