The Big Five, for almost every purpose. It measures five continuous traits with published reliability figures and a long record of predicting real outcomes, and its free versions are as good as the paid ones. The Myers-Briggs Type Indicator sorts people into sixteen types by splitting four continuous scales down the middle, which is why a large share of people who retake it get a different type, and why it is not used in personality research. The MBTI is better at one thing: giving people a memorable label and a vocabulary to talk about difference, which is a real use, just not a measurement.
Side by side
| Big Five | MBTI | |
|---|---|---|
| What it measures | Five continuous traits, with facets | Four preferences, each split into two, giving sixteen types |
| How it is scored | Sum of items per scale, compared with a norm sample as a percentile | Each scale split at the midpoint into one of two letters |
| Reliability | Scale alphas typically .8 to .9; test-retest .7 to .9 over months | Scale scores are moderately reliable; the type is not, because people near a midpoint flip letters on retest |
| Predictive validity | Predicts work, health and relationship outcomes with small to moderate effects (Roberts and others, 2007) | Little evidence that type predicts outcomes beyond what its underlying continuous scores predict |
| Relation to each other | The four MBTI scales correlate with four of the Big Five: extraversion, openness, agreeableness and conscientiousness (McCrae and Costa, 1989); the MBTI has no neuroticism scale | |
| Who owns it | Public domain versions (IPIP) are free for any use | A trademark of the Myers-Briggs Company; the official test is paid and administered by certified practitioners |
| Where it is weak | Self-report, non-representative norm samples, and five broad traits miss much (see below) | Type instability, no neuroticism, and a theory of fixed categories the data do not support (Pittenger, 2005) |
Where the Big Five is weak
Honesty requires saying so. The Big Five is a description, not an explanation; it tells you where you sit, not why. It measures self-image, which can be shaded or faked. Its norm samples, including the ones on this site, are people who took a test online, not a cross-section of any country. Five broad traits leave out honesty and humility, which the six-factor HEXACO model adds, and they say nothing about interests, values, abilities or what a person is like on a particular day. And a percentile against a sample of strangers is a comparison, not a judgement of worth.
Which to take for which purpose
For understanding yourself with figures you can trust, for research, for comparing yourself with other people, or for seeing whether you have changed: the Big Five, and on this site the 120-item version with its facets. For a shared vocabulary in a team workshop, where the point is conversation rather than measurement: the MBTI does that job and the Big Five does not. For hiring: neither, and this site does not help with that.