The science
How we score
The number isn’t a stretched percentage of right answers. It comes from a statistical model used in large-scale exams, and we show the margin of error and the test’s limits alongside it.
Original questions with known difficulty
The questions are ours: the figures are generated from rules, in code, and each one is checked to have a single right answer. The wrong options are built so the right one can’t be found by elimination without working out the pattern. There are 40 questions, and 37 count toward the estimate.
| Area | Questions that count |
|---|---|
| Matrices (visual patterns) | 10 |
| Number and letter series | 5 |
| Shape rotation | 5 |
| Logic | 5 |
| Verbal reasoning | 5 |
| Memory (digit sequences and recall) | 7 |
The other 3 questions are about you. Your age changes the comparison: you are compared with people in your age group. Education and self-rating don’t change the result; education may only be used, anonymously, to correct the norm when it is recalculated.
Item Response Theory
The result is calculated with Item Response Theory, using the three-parameter model. Each question has three properties: its difficulty, how well it separates people with more ability from people with less, and the chance of getting it right by guessing.
Your ability is the value that best explains your pattern of right and wrong answers. That's why a hard question counts for more than an easy one, and guessing doesn't produce a high number.
The 95% range
Every test has measurement error, and a short, unsupervised online test has more of it than a full test given by a psychologist. So next to the number we show the range your result is likely in, with 95% confidence. With 37 questions, the standard error is about 5 to 6 points in the middle of the scale, which gives a range of roughly 10 points either way.
Age comparison, with a provisional norm
The value is converted to the usual IQ scale (mean 100, standard deviation 15), comparing you with people in your age group. Solving new problems changes over adult life, and the age comparison accounts for that.
Today’s norm is provisional. It uses parameters published in scientific research, mainly ICAR (the International Cognitive Ability Resource), a public-domain battery taken online by almost 97,000 people (Condon and Revelle, 2014). Because people who take part in these online studies score higher than the general population, the parameters were adjusted so as not to inflate your result; the age curve follows the literature (Salthouse, 2009, and the age tables of classic tests). The norm is the same in all three languages of the site. It will be recalculated with the answers of people who take the test, separately for each language, and the report always says which norm was used.
Limits we state up front
- The test has a ceiling and a floor: for adults aged 20 to 29, it can’t tell scores apart above about 138 or below about 57. At other ages these limits shift, and the report shows yours. If you get everything right, the report says your result is at least that value.
- Response time doesn’t change the number: it is only used to detect answers that came too fast. Education and self-rating don’t change the number either.
- Answers too fast to have been read, or a very unusual pattern of right and wrong answers, mark the result as unreliable, and we offer a free retake.
- We don't give an IQ per area: with few questions per area, the error would be too large. We show how many you got right and whether the area came out above, at or below your overall result.
- We do not use a "national average" or country rankings. The historical figures in the report are not an IQ comparison: they are retrospective Cox (1926) estimates, shown as a curiosity (see below).
Historical figures: the Cox (1926) estimates
In 1926, psychologist Catharine Cox, part of Lewis Terman’s team at Stanford, read biographies, letters and school records of about 300 historical figures and estimated what IQ they would have shown in childhood and in youth (ages 17 to 26). Nobody took a test: these are raters’ estimates based on what the sources said, and Cox herself measured how reliable each person’s data were — often, not very. The scale is not the one our test uses either: Cox worked with the mental-age idea of her time, and her numbers do not compare point by point with a modern IQ. We show the value she published for each person’s youth, not the higher “corrected” values that circulate online. The list reflects who the books of 1926 considered eminent: almost only European men. The comparison is just historical curiosity.
In the report, we show a few historical figures whose Cox estimate is within 10 points of yours. Never as "you have the IQ of…", with no ranking and no badge: it is historical curiosity. If your estimate is below 110, we show no names, because the study has no figures in that range.
Cox (1926) estimates: retrospective approximations, not real measurements.
References
- Condon, D. M., and Revelle, W. (2014). The International Cognitive Ability Resource: Development and initial validation of a public-domain measure. Intelligence, 43, 52–64.
- Cox, C. M. (1926). Genetic Studies of Genius, Vol. II: The Early Mental Traits of Three Hundred Geniuses. Stanford University Press.
- Embretson, S. E., and Reise, S. P. (2000). Item Response Theory for Psychologists. Lawrence Erlbaum.
- Salthouse, T. A. (2009). When does age-related cognitive decline begin? Neurobiology of Aging, 30(4), 507–514.
An educational estimate, with a margin of error. In Brazil, where IQNorth is based, psychological intelligence tests used in clinical, educational or occupational settings may only be used by psychologists (Law No. 4,119/1962 and CFP Resolution No. 31/2022). This is not one of them, and the result must not be used for hiring, school admission or any decision about a person.