GUIDES FOR UNDERSTANDING
Online IQ tests: How to check their quality
A polished page, a score with decimal places or a downloadable certificate can look persuasive. None of those features establishes the quality of an assessment. Before sharing information or interpreting a result, look for evidence that lets you check how the test was developed and which uses have actually been investigated.
Start with the claim being made
Ask what the test claims to measure: Visual reasoning, knowledge, memory or a broader ability. Then check whether its documentation supports that particular claim. References to general intelligence research do not show that the exact questions used on a website have been evaluated.
International Test Commission guidelines call for empirical evidence supporting interpretations of test scores. They also address the selection of suitable instruments, the competence of test users and clear communication of results. These are useful questions to ask of a provider, rather than treating professional-looking presentation as proof.
Find the version, sample and conditions
Useful methodology identifies the test version, the people who were assessed and the procedure followed. Look for the country, ages and study date. When norms are used, it should be clear which group they represent. A translation does not automatically turn research in one population into a local validation.
Check the permitted devices and administration conditions too. A small screen, interruptions or changes to the response format can alter the experience. Do not assume that an assessment retains its properties after its questions are shortened, its time limit changes or its delivery is redesigned.
Separate precision from validity
Consistency of measurement and evidence supporting an interpretation address different questions. A high reliability coefficient does not by itself establish that a test measures everything it advertises. It is not the percentage of people whose IQ has been guessed correctly.
If a website displays a range or percentile, check how it was calculated. Such values need an appropriate method and an identifiable reference group. A technical-looking number cannot replace that explanation, even when it appears in an attractive chart or next to a certificate issued by the website itself.
A worksheet for comparing two websites
Open the methodology and record five things: Exact name and version; interpretation promised; research matching that version; reference population; and administration conditions. Beside each item, note where you found it. If a link only opens a general homepage or a claim has no identifiable support, mark it unresolved instead of treating it as established.
Consider this fictional comparison. Website A displays a number called IQ but identifies neither tasks nor sample. Website B reports correct answers, explains its scoring and acknowledges having no norms. The second communicates its limits more clearly; that does not turn it into a validated IQ assessment. Choose according to your purpose, rather than the more flattering score.
What you can check in this project
This project uses a selection from OMIB and separates the original research from its implementation. The bank was studied using a procedure different from this website’s 22-task pilot. We do not transfer the reliability coefficient of the researched forms to this shortened version. Checking that software counts answers correctly is a technical check; establishing their psychological interpretation requires additional work.
The available mode shows your result on screen, without requesting email or requiring a purchase. We disclose the material source and scoring rule. We do not yet provide IQ, our own population norms or a diagnosis. This helps you decide whether an exploration of reasoning meets your purpose.
Frequently asked questions
Does free mean unreliable and paid mean validated?
No. Price and interface quality do not replace evidence about the version, interpretation and population in which it is used.
Does a reliability coefficient of 0.92 mean 92% accuracy?
No. A reliability coefficient summarises a property of scores under a particular procedure. It is not an IQ guessing success rate and does not automatically transfer to an adapted test.
Sources you can check
Educational content by Sator editorial; it is not a professional assessment or clinical review. Commercial examples are not hands-on reviews. How we prepare these guides.