How These Personality Tests Are Built, Scored and Limited

Every item on this site was written for this site. No commercial test booklet was licensed, copied or reproduced. The MBTI-style items were written against the four preference pairs described in the public type literature — E/I, S/N, T/F, J/P — and each item states one preference so the visitor can pick the side that fits better.

The SBTI items were written for the themes the SBTI system covers: self-view, emotional style, attitude to rules, decision and execution style, and social boundaries. CSPI follows the same pattern for online social behaviour, and the Enneagram pages use a short self-report set.

Wording is deliberately concrete ("I value personal space in any relationship") rather than abstract, because a concrete statement is easier to answer consistently. The trade-off is real: two people with the same underlying preference can read the same sentence differently.

MBTI: each option carries exactly one letter. Each dimension is scored by counting answers on each side, and the higher side becomes the letter. Ties go to E, S, T and J. Four letters in order give the code, for example INFJ.

The full MBTI set has 56 questions, 14 per dimension. Some language versions still ship a shortened 16-question set (4 per dimension) because those translations are incomplete. The number shown above the start button on each test page is the actual number for that language — trust that number, not a marketing round figure.

SBTI: each option scores 1 to 3. Answers are averaged per sub-dimension (15 in total); an average above 2.3 becomes H, above 1.7 becomes M, anything else L. The 15 sub-dimensions roll up into 5 main dimensions, which produce a five-letter code such as HLHMM that maps to one of 27 archetypes.

There is no weighting, no machine-learning model and no tuning over time: the same answers always produce the same code. You can verify that by submitting identical answers twice.

We do not maintain a norm group. Your result is not compared against a reference population, so a type label describes the pattern of your own answers and says nothing about how common or rare it is.

We do not collect answers for research. Answers stay in your browser — local progress is kept in localStorage so a refresh does not lose your place — and traffic is measured with Google Analytics in aggregate. We publish no sample size, because there is no research sample to describe.

The only population figures on this site are the estimated type frequencies published by the Myers & Briggs Foundation for the general US population. They are external estimates, rounded, shown with a leading "~", and every page that prints one also prints the caveat that they differ by country, age and sample.

Cronbach's alpha measures whether the items of one scale agree with each other. Test-retest reliability measures whether the same person gets the same result twice. Criterion validity measures whether the score predicts something outside the questionnaire. This site has run none of the three, so it publishes none of the three.

Earlier versions of a few articles quoted an internal-consistency coefficient and a behavioural-consistency rate. Neither had a sample, a calculation or a date behind it, so both were removed in October 2026 instead of being re-used. The table below records what happened to each claim, including the removed ones.

What can be stated without a study: the instrument is deterministic (same answers, same result), the scoring rules are exactly the ones described above, and no answer is silently weighted or discarded. That is a statement about how the tool works, not about how accurate it is.

The MBTI-style types come from a published typology with decades of use and a substantial critical literature. The 16 descriptions on this site follow that public literature; the population figures come from the foundation that holds the trademark.

SBTI is a house system built for this site: 15 self-report dimensions grouped into 5, producing 27 named archetypes with a deliberately playful tone. It has no academic source and no external population data, and it does not claim to measure anything beyond how you answered. It exists to be entertaining and to give you vocabulary for digital habits — not to classify people for professional purposes.

CSPI and the Enneagram pages keep the same distinction. CSPI is a house system. The Enneagram is a public typology that this site summarises; it is not a clinical instrument here, and nothing on those pages is a diagnosis.

Self-report bias: answers track mood, energy and circumstances. The same person can land on a different code a month later, and a forced choice between two options you feel equally about gets resolved arbitrarily.

Barnum effect: type descriptions feel accurate partly because they are written to be broadly agreeable. A description feeling "very me" is not evidence that the classification is correct.

No diagnostic use: nothing here is a psychological diagnosis, medical advice, a hiring or admission screen, or a verdict on a relationship. If you are dealing with distress, talk to a qualified professional rather than to a website.

Language coverage: translations differ in length and nuance, and the shortened 16-question versions are less stable than the full set. Compare results within one language, not across two.