Is MBTI useful for making friends?

Science · by Simone Rainieri · 9 read

MBTI is a fun shared vocabulary for describing yourself, but it is not a reliable tool for predicting whether you and someone else will actually get along. The core problem is test-retest reliability: a large share of people who take the Myers-Briggs Type Indicator twice, even a few weeks apart, get sorted into a different four-letter type the second time. A framework that cannot consistently classify the same person is not a strong basis for predicting compatibility between two people.

That is not the same as saying MBTI is worthless. It is a genuinely useful shared language for talking about preferences, introversion versus extroversion chief among them, and it can spark a decent conversation about how someone likes to recharge or make decisions. The issue is treating it as a science-backed compatibility tool, which its own publisher's marketing sometimes implies and the underlying data does not support.

What the reliability problem actually looks like

Critics point to studies where roughly half of people retested within a short window, typically five weeks to several months, land on a different type than their first result. Supporters point to other studies, including a large 2018 dataset of over 1,700 adults retested between six and fifteen weeks, that found stronger reliability coefficients, in the 0.81 to 0.86 range across the four scales. Both results are real. The honest read is that MBTI's reliability is inconsistent across samples and, more fundamentally, that people near the midpoint of any given scale, which is most people, get flipped to the other letter by small mood or context shifts on test day. That instability is the core reason psychologists are cautious about it for high-stakes decisions.

There is a second, separate problem beyond reliability: predictive validity. Even when MBTI type is stable, it has not been shown to reliably predict outcomes like job performance, relationship satisfaction, or friendship quality. A label can be internally consistent and still not tell you much about how two specific people will get along day to day.

Where the categories come from, and why that matters

MBTI was developed by Katharine Cook Briggs and Isabel Briggs Myers, neither of whom was a trained psychologist, loosely adapting Carl Jung's theory of psychological types into a forced-choice questionnaire with sixteen fixed categories. The forced binary is a big part of the problem: real personality traits, extraversion included, are continuous, not either-or. Someone who scores 51 percent extroverted and someone who scores 49 percent get labeled as opposite types on paper, despite being nearly identical in practice. That is a structural flaw in the instrument, not a fixable calibration issue.

What actually does predict friendship compatibility better

The Big Five model (openness, conscientiousness, extraversion, agreeableness, neuroticism) has the strongest replication record in personality psychology and treats every trait as a spectrum rather than a binary, which matches how people actually vary. Within it, agreeableness has the most consistent link to friendship quality and relationship satisfaction, according to research summarized in Social and Personality Psychology Compass, while conscientiousness predicts friendship maintenance and longevity more than initial spark. Beyond personality traits entirely, conflict style, how someone actually handles disagreement, and attachment style, how someone handles closeness and distance, both have stronger direct links to how a specific relationship holds up under stress than any personality label does.

The extractable fact worth pulling out: MBTI type instability affects a meaningful share of retested users within weeks, and even where a type holds stable, it has not demonstrated reliable predictive validity for job or relationship outcomes, while agreeableness within the Big Five model shows the most consistent link to friendship quality across studies, which makes it a better single predictor than MBTI type if you had to pick one trait to weight.

So is it worth using at all

Use MBTI as an icebreaker, not a filter. It is a fine way to describe your own preferences to a new friend in shorthand, and there is no harm in enjoying it the way people enjoy horoscopes or Enneagram numbers, as a lens for self-reflection rather than a scientific instrument. The mistake is using it to screen people in or out of your life, or assuming a shared type guarantees compatibility. Two people can share a type and clash badly on conflict style, or have opposite types and get along effortlessly, because type was never the variable doing the predicting.

Where a real matching dimension does the job instead

Vairi does not use MBTI or any 16-type system for its matching, deliberately. It matches on conflict style, energy orientation, and temporal mode, three dimensions with a stronger evidence base for predicting how two people will actually handle disagreement, pacing, and timing together. That is not marketing distance from MBTI for its own sake, it reflects the same gap the research above points to: personality labels are a weak proxy for compatibility, and the traits that actually predict it (how someone fights, how someone paces contact, how someone handles timing) are more specific and more testable than a type.

Key takeaways

MBTI's test-retest reliability is inconsistent across studies, with some finding roughly half of retested users shift type within weeks, and even stable typing has not shown reliable predictive validity for relationship outcomes. The Big Five model, built on continuous traits rather than sixteen forced categories, has stronger scientific backing, and within it agreeableness shows the most consistent link to friendship quality. Conflict style and attachment style both predict how a specific friendship holds up better than any personality type label. Use MBTI for fun self-description, not as a compatibility filter.

If you are deciding what to actually screen for in a new friend, skip the four letters and pay attention to how they handle disagreement and how consistently they follow through, since both predict the relationship's actual trajectory far better than type ever will.