How the score works

The test

The test has 24 puzzles of four kinds (grids, number series, picture sequences and odd-one-out), six of each, ordered from easy to hard. Every puzzle is original: it is generated by our own code from a fixed starting number, so everyone who takes this version (v1) gets exactly the same puzzles. That is what makes results comparable. We do not use puzzles from Raven's, Mensa or any published test.

You have 20 minutes. The clock sets the limit; it does not change the score.

Step 1: the raw score

Each right answer counts one point, so the raw score runs from 0 to 24. A wrong answer costs nothing, so a guess is always better than a blank. Every puzzle counts the same, whatever its difficulty.

Step 2: who you are compared with

Every finished test is logged anonymously. An attempt joins the comparison group only if:

• it is the person's first attempt on this device (retakes score higher because the puzzles are familiar),
• it took at least 3 minutes, and at least 20 puzzles were answered (to leave out people clicking through),
• it is one of the first three attempts from the same internet connection that day.

The comparison group is people who chose to take an online IQ test. They are not a random sample of the population, so your score says where you stand among them, put on the IQ scale. It is not the same as a clinical IQ score from a test such as the WAIS.

Step 3, with enough people: your rank, put on the IQ scale

Once at least 500 attempts are in the comparison group, we work out what share of them scored below you (people with the same raw score count as half below, half above). That share is your percentile. We then find the point on the normal curve with the same percentile, on a scale with an average of 100 and a standard deviation of 15, which is the scale IQ tests use. For example, a result better than 84% of the group comes out at about 115.

Step 3, before that: the provisional estimate

With fewer than 500 attempts, a rank would jump around too much to be useful. So the score uses an assumption instead: that the average test-taker gets 12 of 24 right, and that 4 puzzles is one standard deviation. The formula is 100 + 15 x (raw score - 12) / 4. This is a starting guess, not a measurement, which is why the result page labels it provisional.

Why scores stop at 55 and 145

A 24-puzzle test cannot tell apart scores far from the average: near the top, one more right answer is the difference between several points on the scale. So scores are limited to 55-145. A score of 145 means "145 or above".

Your score does not change later

The score is fixed when you finish, and your result records which method was used and how many people you were compared with. The paid report adds a live rank among everyone who has taken the test so far, which does change as more people take it.

What we log

For each finished test: the raw score, which puzzles were right, the time spent on each puzzle, the date, and the country the connection came from. We do not log names, email addresses or IP addresses. To count attempts per connection we keep a scrambled code of the connection for two days, which cannot be turned back into an address. See the privacy page.

Limits

• Any test with 24 puzzles has measurement error: the same person can score a few points differently on another day.
• Tiredness, distractions, a small screen or a slow connection can lower a score.
• Reasoning puzzles cover one part of what IQ tests measure. Clinical tests also measure vocabulary, memory and processing speed, given one to one by a psychologist.
• This is not a diagnosis of anything, and it should not be used for decisions about school, work or health.

As results come in, we will check how hard each puzzle really is and publish what we find on this page. See also the IQ scale and average IQ.