eTestor

How we build and score our tests

We would rather tell you what these tests can and cannot do than let you assume. This page covers where the frameworks come from, how your answers become a score, and what an unproctored online test genuinely cannot measure.

Last updated November 16, 2026

Where the frameworks come from

Our tests are built on published psychometric models — the Five-Factor Model, adult attachment theory, Holland's RIASEC interest model, Moral Foundations Theory, and matrix-style reasoning items in the tradition of Raven's Progressive Matrices. The sources are listed at the bottom of this page.

We write our own items rather than reproducing a copyrighted instrument. That is an important distinction and we want to be exact about it: our 16-type test is a Myers-Briggs–style questionnaire, not the official MBTI® instrument, and our IQ test is not a Wechsler or Stanford-Binet. Scores are not interchangeable with those instruments.

How answers become a score

Most tests use Likert items, where each response maps to a point value on one or more scales, with reverse-keyed items scored in the opposite direction so that agreeing with everything does not produce an extreme profile. Scale scores are summed, then expressed either as a percentage of the maximum, as a position on a spectrum, or as a percentile against our norm data.

Our IQ test is scored differently: items are difficulty-weighted, timed, and converted to the conventional scale where the mean is 100 and the standard deviation is 15 — so a score of 115 sits roughly one standard deviation above average, above approximately 84% of people, and 130 sits above approximately 98%.

Where our norms come from

Percentiles are calculated against the distribution of scores from eTestor users who have completed the same test, banded by age and gender where the underlying research shows those bands matter. This is a self-selected internet sample, not a nationally representative one, and people who choose to take a personality test online are not a random slice of the population. A percentile here tells you where you fall among people like our users, which is a narrower claim than where you fall among everyone.

What these tests cannot do

They cannot diagnose anything. They are unproctored, so we cannot verify who answered or under what conditions. They rely on self-report, which is affected by mood, recent events and how you want to be seen. Repeat testing produces practice effects — on reasoning tests, typically in the range of 3 to 5 points. And a result describes a tendency, not a ceiling.

If you need a verified assessment — for a clinical question, an educational accommodation, a gifted programme or any formal process — that requires a proctored instrument administered by a licensed professional.

Where we use AI, and why

Three parts of eTestor use AI (OpenAI's models), and nothing else does. First, before a test starts, the wording of its questions is adapted to the age band and gender you choose, so examples and vocabulary feel natural; the model is instructed never to change what a question measures, its direction or its strength, and the scoring keys stay fixed. Second, the written analysis in a full report is drafted by AI from your scores and answers, within rules we set (every section tied to a named scale, no clinical claims). Third, the questions of the English level test were drafted with AI, kept only after two blind AI reviews placed each one at its level, and checked by us; they are fixed and shown to everyone word for word.

Your score itself is never produced by AI: it is computed by our scoring code from your answers. The IQ puzzles are designed by hand, not by a language model. The test pages, guides and this page are written and fact-checked by us, with every study linked to its published record. The AI receives your scores, your answers (never on the political compass and moral foundations tests), your age band and gender, never your name or email.

Corrections

If you think something here is wrong, tell us and we will fix it or explain why we disagree. Our about page has contact details.

References

  1. Costa, P. T., & McCrae, R. R. (1992). Revised NEO Personality Inventory (NEO-PI-R) and NEO Five-Factor Inventory (NEO-FFI) professional manual. — The Big Five / Five-Factor Model.
  2. Goldberg, L. R. (1993). The structure of phenotypic personality traits. American Psychologist, 48(1), 26–34. — Why five factors keep reappearing across languages.
  3. Raven, J. (2000). The Raven's Progressive Matrices: Change and stability over culture and time. Cognitive Psychology, 41(1), 1–48. — The matrix-reasoning format our visual IQ items follow.
  4. Hazan, C., & Shaver, P. (1987). Romantic love conceptualized as an attachment process. Journal of Personality and Social Psychology, 52(3), 511–524. — Adult attachment styles.
  5. Brennan, K. A., Clark, C. L., & Shaver, P. R. (1998). Self-report measurement of adult attachment. — The two-dimensional anxiety/avoidance structure.
  6. Paulhus, D. L., & Williams, K. M. (2002). The Dark Triad of personality. Journal of Research in Personality, 36(6), 556–563. — Narcissism, Machiavellianism and psychopathy as separable traits.
  7. Holland, J. L. (1997). Making vocational choices: A theory of vocational personalities and work environments (3rd ed.). — The RIASEC model behind our career tests.
  8. Graham, J., Haidt, J., & Nosek, B. A. (2009). Liberals and conservatives rely on different sets of moral foundations. Journal of Personality and Social Psychology, 96(5), 1029–1046. — Moral Foundations.
  9. Horne, J. A., & Östberg, O. (1976). A self-assessment questionnaire to determine morningness-eveningness in human circadian rhythms. International Journal of Chronobiology, 4(2), 97–110. — Chronotype.