How accurate is an online IQ test?

A short online IQ test gives you a soft estimate of your reasoning. It can place you in a broad band, but it cannot give you a precise IQ number, and anyone quoting you a single score like 127 is claiming more than the test can know. Read the result as a range, and never use it to make a hiring decision.

What a full IQ assessment involves

A clinical IQ score comes from a very different process. Tests such as the Wechsler Adult Intelligence Scale (WAIS) are given one to one by a trained examiner and cover many subtests (vocabulary, working memory, processing speed, visual puzzles and more). The ten core subtests of the WAIS-IV (2008) alone typically take 60 to 90 minutes. Scores are compared against a large sample chosen to match the population by age and other factors, so "100" means the average of a known group.

An online test skips almost all of that. That does not make it useless. It makes it a different kind of measure, and it should be read that way.

Why short tests are less precise

Fewer items, more noise. Every test score contains some measurement error. With dozens of items across many subtests, errors tend to cancel out. With a handful of items, one lucky guess or one misread question shifts the result noticeably.

Unknown norms. A score only means something relative to a comparison group. Many online tests never say who that group is. If the people who take a free IQ test differ from the general population, which is likely, the scale drifts.

No proctor. At home you might be tired, interrupted, rushed or helped. You might have seen similar puzzles before. None of that is cheating, but all of it moves the number.

Effort and state. Reasoning scores depend on how much effort you put in on the day. A test taken at midnight after a long week measures that week as much as your reasoning.

One kind of reasoning. A short test samples a narrow slice of what full batteries measure. Your score reflects the item types it happened to use.

Why we report a range, not a number

Our reasoning tests are short and unproctored, so we report each result as a range on a common scale rather than a single score. The width of the range reflects how much one item can move the result. It is our honest way of saying "somewhere around here".

A range also changes how you read it. "Somewhere in the low 110s to high 110s" invites a sensible question: does this match how I handle new problems? A single number invites comparison and bragging, which the test cannot support.

What a short reasoning test can tell you

• A rough sense of how you reason with words, numbers and patterns.
• How you approach unfamiliar puzzles.
• Whether your verbal and non-verbal reasoning look similar, if you take both kinds of test. A gap between them is often more useful than either score. Our culture-fair IQ test explainer covers the non-verbal side.
• One data point to read next to your personality results. Someone high in openness and quick at pattern reasoning may enjoy work that keeps throwing up novel problems.

What it can't tell you

• Your worth or your potential. Reasoning is one ability among many.
• Expertise. Years of domain knowledge often beat raw reasoning on real work.
• Creativity, judgment or practical skill. Short tests do not measure them.
• A clinical picture. If you need an assessment for a learning difference, giftedness or any clinical question, see a licensed psychologist. An online estimate is not a substitute.

Why it should never decide a hire

Employers are tempted by IQ-style tests for a reason. Schmidt and Hunter (1998) reviewed 85 years of research and found that general mental ability, measured with proper tests, was among the strongest single predictors of job performance. Sackett, Zhang, Berry and Lievens (2022) later showed those older estimates were inflated by overcorrection and revised the validity of cognitive ability, and of most other selection methods, substantially downward.

That research used validated, supervised tests, not short online estimates. Cognitive tests also carry a known risk of adverse impact, meaning they can screen out protected groups at different rates. In the United States, the Uniform Guidelines on Employee Selection Procedures (1978, 29 CFR Part 1607) treat any selection procedure with adverse impact as discriminatory unless it has been validated. A short, unproctored online estimate meets none of that.

Our rule is simple. Reasoning scores are soft estimates. They never headline your report, and on their own they should never decide who gets hired. Ask the same of any tool you use on other people.

How to get the most accurate estimate you can

• Take the test when you are rested and can give it your full attention.
• Find a quiet place and remove interruptions.
• Do not look anything up or ask for help. The only person you fool is yourself.
• Take it once. Retaking the same items measures memory, not reasoning.
• Read the range, not the midpoint.

Where the reasoning tests sit in our assessment

Our assessment includes two reasoning tests among its eight: a classic test with verbal and visual items, and a culture-fair matrix test. Cognitive reasoning is one of the best-studied areas in psychology, so these sit in our validated evidence tier alongside the Big Five. Our versions are short and unproctored, which is why we treat the scores as soft ranges and weight them lightly in the report. For the general case on test quality, see are personality tests scientific?, and for keeping any score in proportion, how to read your results without fooling yourself.

Put your reasoning in context

A reasoning range on its own tells you little about how to work. Our assessment runs all eight tests in one sitting, takes about an hour, and compiles them into one report. Founder OS is live today at $9 and reads your reasoning alongside your personality, so the report is about how you operate, not a number.