The science of intelligence
How is intelligence measured?
Short answer
Intelligence is measured with standardized tests: the same tasks, given the same way, scored against a large norm sample of people the same age. The result is a deviation IQ, set to an average of 100 and a standard deviation of 15. Factor analysis checks what the tasks share, and reliability studies check how much the score wobbles.
Want your own number? The test is 25 visual puzzles, about 12 minutes, no sign-up.
Step 1: a fixed set of tasks
An intelligence test is a sample of problems chosen to draw on reasoning, knowledge, memory and speed. Some use words or numbers and require specific cultural knowledge, like vocabulary. Others use shapes or designs (Gottfredson, 1997). What makes a test standardized is that everyone gets the same kinds of task, with the same instructions, time limits and scoring rules. That sameness is what lets one person's score be compared with another's.
The idea goes back to Alfred Binet and Theodore Simon, who published the first practical intelligence test in 1905 (Wikipedia: Intelligence quotient). The guide to named IQ tests covers the major batteries in use today.
Step 2: a norm sample
A raw score (how many items you got right) means little by itself. Test publishers first give the test to a large sample of people chosen to represent each age group. Their results become the norms. Your raw score is then placed against the norms for people your own age (Wikipedia: Intelligence quotient). See standardization sample for how those groups are built.
Step 3: the deviation IQ
The early IQ really was a quotient: mental age divided by age, times 100. In 1939 David Wechsler replaced it with the deviation IQ (Wikipedia: WAIS). Scores are standardized within each age group to a mean of 100 and a standard deviation of 15 (Wechsler, 2008). Because the scale is built on the normal curve, about 68% of people score between 85 and 115 and about 95% between 70 and 130 (the 68-95-99.7 rule). The step-by-step arithmetic is in how IQ is calculated.
Try one · Spatial Reasoning
Which option is the same shape, only rotated?
Tap the option you think is right, or show the answer.
Show the answer
Option 3 is the original figure turned a half turn.
How to solve it
- Pick a landmark: the marked cell or the end of the longest arm.
- Turn the figure in your head and watch where that landmark goes, and which side the other cells end up on.
- Turning the figure a half turn gives option 3.
- Each of the other options is a mirror image: flipping it over would match, but no turn does.
That was 1 puzzle. The full test has 25.
Step 4: factor analysis
How do researchers know a test measures one thing rather than a grab bag? They give many tests to many people and use factor analysis to see how scores cluster. Charles Spearman used this logic in 1904 to propose a general factor, g (Wikipedia: g factor). John Carroll applied it to more than 460 data sets and found a hierarchy of abilities, with general ability at the top (Carroll, 1993). That structure, covered on the CHC theory page, is why a full-scale IQ can combine very different subtests into one number.
Step 5: reliability and validity
Two more checks decide whether a test is any good.
- Reliability is how consistent the score is. Clinical full-scale IQs carry a standard error of a few points, so manuals report a confidence interval, often around plus or minus 5 points at 95% (Wechsler, 2008). Retesting also tends to raise scores by about a quarter of a standard deviation, roughly 4 points (Hausknecht et al., 2007).
- Validity is whether the score relates to things it should. IQ correlates about .50 with school grades (Neisser et al., 1996), and in about 70,000 English children, intelligence at 11 correlated 0.81 with exam results at 16 (Deary, Strand, Smith & Fernandes, 2007).
What measurement leaves out
The 1994 mainstream statement was explicit that intelligence tests do not measure creativity, character or personality, and are not intended to (Gottfredson, 1997). A score also reflects one session on one day, and every score carries measurement error. And even a strong correlation such as .50 with grades accounts for only about 25% of the variance (Neisser et al., 1996). A test measures a slice of intelligence well. It does not measure a person.
How an online test differs
Our free test uses the same logic in a smaller form: standardized item types, a scale centered on 100, and a confidence range instead of a single number. It is not a clinical instrument, and it reports an estimated IQ from visual reasoning items only. How it works explains the scoring.
Sources
- Gottfredson, L. S. (1997). Mainstream science on intelligence: An editorial with 52 signatories, history, and bibliography. Intelligence, 24(1), 13-23. doi.org/10.1016/S0160-2896(97)90011-8
- Intelligence quotient. Wikipedia. en.wikipedia.org/wiki/Intelligence_quotient
- Wechsler Adult Intelligence Scale. Wikipedia. en.wikipedia.org/wiki/Wechsler_Adult_Intelligence_Scale
- Wechsler, D. (2008). WAIS-IV Technical and Interpretive Manual. Pearson.
- 68-95-99.7 rule. Wikipedia. en.wikipedia.org/wiki/68%E2%80%9395%E2%80%9399.7_rule
- g factor (psychometrics). Wikipedia. en.wikipedia.org/wiki/G_factor_(psychometrics)
- Carroll, J. B. (1993). Human cognitive abilities: A survey of factor-analytic studies. Cambridge University Press. doi.org/10.1017/CBO9780511571312
- Hausknecht, J. P., Halpert, J. A., Di Paolo, N. T., & Moriarty Gerrard, M. O. (2007). Journal of Applied Psychology, 92(2), 373-385. doi.org/10.1037/0021-9010.92.2.373
- Neisser, U., Boodoo, G., Bouchard, T. J., Jr., Boykin, A. W., Brody, N., Ceci, S. J., Halpern, D. F., Loehlin, J. C., Perloff, R., Sternberg, R. J., & Urbina, S. (1996). Intelligence: Knowns and unknowns. American Psychologist, 51(2), 77-101. doi.org/10.1037/0003-066X.51.2.77
- Deary, I. J., Strand, S., Smith, P., & Fernandes, C. (2007). Intelligence, 35(1), 13-21. doi.org/10.1016/j.intell.2006.02.001