What IQ Tests Actually Measure

IQ is one of psychology's most studied and most misunderstood constructs. Its scores predict school and work outcomes better than most psychological measures, and the concept is also politically contentious. Knowing what IQ tests actually measure, and where their limits are, is useful whatever your score.

What is IQ?

IQ is a standard score that places your performance relative to a reference group. Clinical tests are scaled so that their norm sample averages 100 with a standard deviation of 15. On that scale:

• Score of 115: about the 84th percentile (1 standard deviation above average)
• Score of 130: about the 98th percentile (2 standard deviations above; 1 in 44)
• Score of 85: about the 16th percentile (1 standard deviation below average)

The score is relative, not absolute. It tells you where you fall in a distribution, not how much "intelligence" you have in some objective unit, and it only means as much as the reference group behind it.

What IQ tests measure

General ability (g)

Since Charles Spearman's work in 1904, psychologists have known that performance on very different mental tasks is positively correlated. People who do well on verbal reasoning tend also to do well on spatial, memory and speed tasks, more than chance would predict. Spearman called the shared factor "g" (general intelligence). That is why IQ tests include several varied subtests and combine them: the average across many tasks is a better estimate of g than any single task.

Broad abilities

Beyond g, tests capture broader abilities that are partly separate. The main adult and child batteries, the WAIS-5 (2024) and the WISC-V, report five indexes:

• Verbal Comprehension: vocabulary and reasoning with language
• Visual Spatial: mentally assembling and rotating shapes
• Fluid Reasoning: finding rules in new problems
• Working Memory: holding and manipulating information
• Processing Speed: carrying out simple operations quickly and accurately

Other major batteries, such as the Stanford–Binet Fifth Edition and the Woodcock–Johnson tests, report their own sets of index scores alongside the overall IQ.

What IQ tests predict

The predictive validity of IQ is substantial by social-science standards, but often overstated:

Outcome	Correlation with IQ	Source and notes
School grades	about .54	Roth et al. (2015), meta-analysis
Years of education	about .56	Strenze (2007), longitudinal studies
Occupational level	about .43	Strenze (2007)
Income	about .20	Strenze (2007); explains roughly 4% of income differences
Job performance	~.51 → ~.31	Schmidt & Hunter (1998); revised by Sackett et al. (2022)
Mortality	lower risk	about 24% lower risk per standard deviation of youth IQ (Calvin et al., 2011)
Creativity	about .17	Kim (2005); whether a "threshold" exists is contested

The job-performance figure deserves a word. Schmidt and Hunter (1998) estimated ~.51; a 2022 reanalysis (Sackett et al.) lowered it to ~.31, still useful, no longer the clear leader, and ranked below structured interviews. In Schmidt and Hunter's own figures, conscientiousness was about .31, so "twice as predictive as personality" was never quite right either.

What IQ tests do not measure

This is where popular understanding most often goes wrong:

Rational thinking. A high IQ does not prevent poor decisions. Keith Stanovich coined "dysrationalia" for the gap between intelligence and rational judgment: people can reason well on a test and still fall for predictable biases in everyday decisions (Stanovich, 1993).

Emotional intelligence. The relationship depends on how EI is measured. Ability-based EI tests correlate modestly and positively with IQ, mostly through verbal knowledge; self-report EI questionnaires correlate only weakly with it and overlap more with personality (Joseph & Newman, 2010).

Creativity as a whole. IQ and creativity are correlated, but only modestly (Kim, 2005). Some studies find the link weakens above a certain IQ (Jauk et al., 2013); others do not, so treat any fixed "threshold of 120" as a hypothesis, not a fact.

Moral character. Intelligence has no built-in relationship with ethics.

Practical know-how. Robert Sternberg argues that "practical intelligence", knowing how to get things done in real settings, is partly separate from what IQ tests measure. The claim is debated, but the everyday point stands: tests sample reasoning on set problems, not judgment in your own life.

Fixed potential. An IQ score measures current performance, not a ceiling. Each additional year of education raises IQ by roughly 1 to 5 points (Ritchie & Tucker-Drob, 2018), and average scores in many countries rose by roughly 3 points per decade through much of the 20th century (the Flynn effect).

Online IQ tests: what to know

Online IQ tests vary enormously in quality. Warning signs of a poor one:

• Claims to measure your "true" or "certified" IQ in a few minutes
• Nearly everyone who takes it scores above 120
• No word on who your score is compared with
• The result is blurred or locked until you pay

Signs of a better one:

• It says what its scores are compared against and how big that sample is
• It reports a range, not a single exact number
• It covers several kinds of reasoning, not only pattern matrices
• It states clearly that it is not a clinical assessment and gives you the full result without a paywall

The controversy: group differences

IQ research is contentious because average score differences have been documented between socioeconomic, ethnic and national groups. What the evidence supports, following the American Psychological Association's task force report (Neisser et al., 1996):

• Average differences between groups on many tests are real as measurements; their size has changed over time.
• Environmental factors such as schooling, poverty, health and test familiarity clearly matter, and no direct evidence supports a genetic explanation of group differences.
• Heritability within a population (roughly 50–80% in adulthood in twin studies) says nothing about the causes of differences between groups.

Any responsible use of IQ data has to respect that distinction.

How accurate is any of this for you?

Accuracy depends on the test. An individually administered battery given by a psychologist is very reliable; a short online test is a rough estimate. For a fuller look at reliability, norms and what affects a score, see how accurate IQ tests are on our sister site, My IQ Path, which runs the same test as this one.

Taking our IQ assessment

My Path's IQ test has a 12-item short form and a 60-item full form, both free, covering six areas: fluid, verbal, numerical and spatial reasoning, working memory and processing speed. You receive:

• Your percentile among people who took the same form, its range, and that position re-expressed as an estimated IQ (≈ N, with range): an estimate, not a clinical IQ
• A profile showing which of the six areas are relatively stronger or weaker
• A note on what the percentiles are compared against: people who sat the same form on My Path, not a representative norming sample, and not age-normed
• An optional AI-written narrative connecting the profile to career areas (it uses AI credits)

Take the free IQ test →

For a fuller picture of your psychological profile, pair the IQ results with the Big Five (personality traits) and EQ (emotional intelligence) assessments.