IQ Scores Explained: How to Read an IQ Test Report

1 · HOW THE NUMBER IS MADE

From ticked answers to 117, in six steps

Follow the sample person through scoring. Numbers are illustrative; real tests use the publisher's norm tables at every lookup.

Raw score34of 47 items correct on one subtest

Each subtest is scored right/wrong or with partial credit, often stopping after a run of failures. The raw score is just the points earned.

Why it matters: Raw scores mean nothing alone: 34 is excellent for a 7-year-old and ordinary for an adult.

2 · WHAT THE SCORE IS MADE OF

Every subtest mixes four ingredients

Run a factor analysis on any set of cognitive tests, as people have since Spearman, and a general factor turns up running through all of them. Below it sit broad abilities, which the Cattell-Horn-Carroll model names as verbal knowledge, fluid reasoning, working memory and speed among others. Below those, skills tied to a single task. Below that, plain measurement error.

Adding many subtests together cancels much of the specific and error parts, which is why the FSIQ is more reliable and more g-saturated than any single subtest.

Spearman's g, 1904 →

55%20%15%10%

General (g)Broad abilitySpecific to the taskError

Vocabulary: a strong measure of g.Shares are illustrative round numbers in the range reported for Wechsler batteries, not values from a specific study.

3 · HOW STABLE IS IT?

Fairly stable for groups, looser for one person

Retest correlations are high over weeks and stay substantial over a lifetime: Ian Deary and colleagues retested people who had sat the Scottish Mental Survey at age 11 in 1932 and found a correlation of about .63 with their scores at 77. A high correlation still leaves plenty of individual movement.

Or set r: 0.63

MOVE MORE THAN 10 POINTS BETWEEN TESTS44%

SOMEONE AT 130 THE FIRST TIME AVERAGES119

Assumes both tests have mean 100 and SD 15. The pull toward 100 for high and low first scores is regression to the mean, not anyone getting less clever.

4 · WHAT IT PREDICTS

Real predictions, and how often they get one pair wrong

Pick an outcome. The last number answers a concrete question: take two random people; how often does the higher IQ belong to the person with the better outcome?

CORRELATION0.56Strength of the straight-line relationship.

VARIANCE SHARED31%r squared: the share of differences in years of education that line up with IQ.

PAIRS ORDERED RIGHT69%Of random pairs, how often the higher-IQ person also has more years of education. 50% would be a coin flip.

Strenze (2007) meta-analysis of longitudinal studies: IQ measured in youth vs later educational attainment. Concordance for a random pair is ½ + arcsin(r)/π, assuming both variables are roughly normal.

5 · NORMS GO STALE

A score is only as current as the people it was compared with

James Flynn showed that raw performance on IQ tests rose through the 20th century, about 3 points a decade in the US. A test normed long ago compares you with a weaker reference group and flatters you. That is why publishers re-norm every 10–20 years.

It does not hold everywhere. Bratsberg and Rogeberg worked through Norwegian conscript records in 2018 and found the gains going into reverse for cohorts born after the mid-1970s. Similar slowdowns have turned up elsewhere since. Stale norms now cut in either direction.

Years since the norms were collected: 15

SCORE ON OLD NORMS117

ON CURRENT NORMS ≈113

With 15 years of gains at 0.3 points a year, the old norms flatter this score by about 5 points.

6 · HOW MUCH TO TRUST A SCORE

Tick what's true of the test you took

2 OF 7Entertainment, mostlyWithout representative norms and standard conditions the number may not be an IQ at all, even if it looks like one.

REFERENCES

Wechsler, D. (2008). WAIS-IV Technical and Interpretive Manual. Pearson.

Deary, I. J., Whalley, L. J., Lemmon, H., Crawford, J. R., & Starr, J. M. (2000). The stability of individual differences in mental ability from childhood to old age: Follow-up of the 1932 Scottish Mental Survey. Intelligence, 28(1), 49–55.

Strenze, T. (2007). Intelligence and socioeconomic success: A meta-analytic review of longitudinal research. Intelligence, 35(5), 401–426.

Schmidt, F. L., & Hunter, J. E. (1998). The validity and utility of selection methods in personnel psychology. Psychological Bulletin, 124(2), 262–274.

Sackett, P. R., Zhang, C., Berry, C. M., & Lievens, F. (2022). Revisiting meta-analytic estimates of validity in personnel selection. Journal of Applied Psychology, 107(11), 2040–2068.

Flynn, J. R. (1984). The mean IQ of Americans: Massive gains 1932 to 1978. Psychological Bulletin, 95(1), 29–51.

Bratsberg, B., & Rogeberg, O. (2018). Flynn effect and its reversal are both environmentally caused. PNAS, 115(26), 6674–6678.

Carroll, J. B. (1993). Human Cognitive Abilities: A Survey of Factor-Analytic Studies. Cambridge University Press.

McGrew, K. S. (2009). CHC theory and the human cognitive abilities project. Intelligence, 37(1), 1–10.

IQ scores explained: read the whole report, not just the number.