IQ is one of those numbers everyone knows without quite knowing what it stands for. People quote it to impress, compare it with friends, and sometimes worry about their own. But what does an IQ test actually measure, and what is a score obtained in a few minutes online really worth?
A school invention
The first modern intelligence test was not designed to rank geniuses. In 1905, French psychologist Alfred Binet and physician Théodore Simon published a "metric scale of intelligence" at the request of the French Ministry of Education. The goal was very practical: to identify children who needed extra support at school.
Their method compared a child with children of the same age, which gave a "mental age". In 1912, German psychologist William Stern suggested dividing this mental age by the child's actual age and multiplying by 100: the intelligence quotient was born. A 10-year-old who performed like an average 12-year-old scored 120.
That formula breaks down in adulthood (nobody gains one "mental year" per year at 40), so today's tests use a different principle.
100 is the average, by design
Today, IQ is no longer calculated, it is ranked. American psychologist David Wechsler, whose scales (the WAIS for adults, the WISC for children) are the most widely used in the world, introduced the "deviation IQ": your score shows where you stand compared with a large sample of people your age, known as the norming sample.
Scores are set to follow a bell curve with a mean of 100 and a standard deviation of 15. In practice:
- about 68% of people score between 85 and 115;
- about 95% score between 70 and 130;
- a little over 2% score above 130, the threshold often used to describe "giftedness" (it is also the entry bar for Mensa, which requires scoring higher than 98% of the population).
An IQ of 100 is therefore neither "good" nor "bad": it simply means you are right in the middle of your age group. And a score only makes sense with its margin of error, usually around 5 points either way.
What the test measures
A test like the WAIS doesn't ask one kind of question but runs a series of very different tasks: defining words, finding what two ideas have in common, copying patterns with blocks, completing matrices, repeating strings of digits, crossing out symbols as fast as possible… In its fourth edition, these tasks are grouped into four indexes:
- verbal comprehension (vocabulary, reasoning with words);
- perceptual reasoning (visual logic, spatial skills);
- working memory (holding and manipulating information in your head);
- processing speed (how fast you complete simple tasks).
The fact that these abilities are linked, so that people who are good at one tend to be somewhat good at the others, was noticed as early as 1904 by British psychologist Charles Spearman. He called it the g factor, for "general intelligence". The full-scale IQ is an estimate of it.
Among the best-known tasks are Raven's Progressive Matrices, created in 1938: a grid of patterns with one missing piece, which you complete by working out the rule. They use no words at all, which makes them a good tool for measuring "fluid" reasoning, the kind you use when facing a new problem.
What it doesn't measure
IQ predicts school performance fairly well, and job performance to a lesser extent. But it leaves out many things that matter in life:
- creativity and originality;
- social and emotional skills;
- motivation, perseverance and curiosity;
- knowledge and expertise built up in a field;
- practical wisdom, common sense, humour.
A score also depends on testing conditions: tiredness, stress, language skills, familiarity with this kind of exercise. That is why a serious assessment is always carried out by a psychologist, who interprets the results task by task rather than stopping at the overall number. A full assessment often takes more than an hour.
What about online IQ tests?
An online test, even a well-designed one, cannot replace that assessment, for several reasons:
- No representative norms. To say you did better than 80% of the population, the same test must have been taken, under the same conditions, by thousands of people chosen to represent that population. Online tests rely at best on their own players, who are not a representative sample.
- Uncontrolled conditions. You can be interrupted, look for help, or retake the test and improve through practice.
- Too few questions. Fifteen or so items cannot match the precision of a full test battery.
Does that make them pointless? Not if you take them for what they are: a logic game inspired by real tasks that gives you an idea of how comfortable you are with this kind of reasoning. That is the spirit of the Viberankly IQ test: 15 non-verbal questions (Raven-style matrices, number and shape sequences, mental rotation), an estimated score and a range, presented as a playful estimate, not a diagnosis.
A number that changes over time
One last surprise: population IQ scores rose sharply over the 20th century, so much that tests have to be re-normed regularly. This phenomenon, known as the Flynn effect, shows that IQ also measures habits of thinking shaped by school and society. We wrote a whole article about it: the Flynn effect, or why IQ scores rose for a century.
In short
- IQ places a person relative to others their age: 100 is the average, and two thirds of people score between 85 and 115.
- It mainly measures reasoning, working memory and processing speed, and leaves out creativity, emotions and motivation.
- Only a psychologist using a normed test can measure IQ. Online tests give an estimate, best taken as a game.