How Accurate Is The MBTI Test? What The Research Says

Feeling understood can make a description useful to you. It does not, by itself, establish how a test measures preferences or predicts an outcome. Use these four questions to read a claim more precisely.

The publisher’s United States (North American English) supplement to the MBTI Manual for Global Step I and Step II Assessments, copyright 2018, reports the following Global Step I retest results. All four rows refer to the same US English retest subsample: n = 236, interval ≤15 weeks. The larger US English reference sample was N = 3,578; it is a different denominator. Collection dates are not specified in the supplement.

These are publisher-authored results, not independent external validation. Printed page 7 (PDF page 8), Table 11 reports scale correlations and letter agreement separately. A correlation of .88 is not “88% of people correctly typed.”

In the same US English retest subsample (n = 236, ≤15 weeks), Table 12 reports that 51% received all four preferences unchanged. That is full-code agreement, not a true-personality accuracy rate. Do not derive it by multiplying the four per-letter percentages.

Hypothetical illustration, not participant data: suppose a questionnaire assigns one letter above a score boundary of 50. A score of 49 on one occasion and 51 on another would produce a changed letter despite a small score difference. This invented boundary explains classification; it is not an official MBTI scoring rule or a finding about any individual.

A correlation describes a sample’s scores. It cannot tell you why your result changed, or promise that your code will remain unchanged. Check the actual report: official preference indicators and another quiz’s trait percentages are not interchangeable.

Reliability concerns consistency. Validity asks whether an assessment supports its intended interpretation or use. A consistent score alone cannot establish a model’s categories, identify a uniquely true type, or predict every outcome.

The publisher supplement also examines relationships with the Adjective Check List and CPI 260 as convergent and discriminant evidence. Those analyses are not studies showing the best career, intelligence, compatible partners or hiring success for each type. The Foundation’s validity page presents its position; read it as the model steward’s explanation, not an independent verdict.

The broader US English reference study recruited through a market-research firm with Census demographic targets and included adults aged 18–86. Those details do not make this 2018 document a current population survey, and you should not assume every reference-sample characteristic describes the smaller retest subsample.

Use this original record-keeping method without taking another test first. It is a way to organize your questions, not a validated typing procedure.

The Myers & Briggs Foundation’s professional ethics guidance says the official instrument should not screen job applicants and practitioners should not advise a career or personal relationship solely from type. This is attributed professional guidance, not a statement of employment law.

A result can give you language for discussing preferences. Career choices and relationships still need concrete information about skills, conditions, interests, communication and the other people involved.

Our 20-question MBTI-style tool is an independent reflection tool, not the official MBTI assessment. The official-study numbers above do not apply to it. We do not publish a validation-derived accuracy percentage for this tool.

The definitions, source table and record-keeping steps in this guide are usable without a test. If you choose our tool, treat its output as a starting point for reflection, not confirmation of a true type or of the research above.