Sources & methodology

These are two different things, and the difference is the most common thing people ask about.

Where you sit is your exact score, shown as a position along the line. Your range is one of five bands, and each band covers a stretch of scores. The words, the range name, and the description all come from the band. The position on the line comes from your score.

So two things can happen that look odd at first:

• Two people sit some distance apart on the line and still get the same words, because both scores land inside one range.
• Two people sit close together and get different words, because a boundary happens to fall between them.

It shows up most on Extraversion, where the middle range covers 10 points — 25 to 34, or about a quarter of the whole scale. Someone near the bottom of that range and someone near the top look clearly different on the line and still read the same description.

Neither view is the “real” one. The line is better for comparing people. The range is better for describing one person, because a description has to be written for a stretch of scores rather than for every possible number.

The five ranges are not equal slices of the scale. They are drawn from where people score, so each one holds roughly the same share of people no matter which trait you look at: about 10% at each end, about 20% in each of the next two, and the rest in the middle.

That means the same raw number means different things on different traits. A score of 36 is in the upper range on Extraversion, the middle on Agreeableness, and the lower-middle on Openness. A single set of cut-offs across all five would be wrong.

Range labels are calibrated against a large public reference sample, not against selfwyse's own users. A range tells you roughly where you sit compared with other people who answered the same questions. It is not a category you belong to, and it is not a measure of ability.

Each trait runs between two named ends — Introversion and Extraversion, Directness and Warmth, and so on. Both ends are real things a person can be glad to be. Neither is the healthy one.

That is why you will not see “high” or “low” anywhere in your result. Those words carry a verdict these scales cannot support. Each range has its own name per trait instead.

There is also no total score. The five traits are measured separately and do not add up to anything — a bigger number across all five would not mean a better result, so we do not print one.

Scores run from 10 to 50 on each trait: ten questions, each answered on a 1–5 scale. “People here” is the share of the reference sample that falls in each range.

Shared and circle reports mark each trait as either In sync or Worth a conversation. That is a description of the gap between two scores, not a grade.

“Worth a conversation” means the gap is bigger than most pairs’ gaps on that trait — we set each threshold so that about 30% of randomly matched pairs reach it. The thresholds differ per trait because the traits spread differently: extraversion 15 points, agreeableness 12 points, conscientiousness 12 points, emotional stability 14 points, openness to ideas 10 points. A flat number would flag Extraversion far more often than Openness purely because Extraversion scores are more spread out.

Two things worth being straight about there. The 30% target is a product judgment about how much of a report should be difference rather than agreement. It is not derived from the data. And these rates describe randomly paired people. Couples and close friends tend to resemble each other more than chance, so real reports should flag somewhat less often.

Two people marked “worth a conversation” are not doing worse than two people marked “in sync”. There is no compatibility score anywhere in these reports, and that is on purpose: the research does not support one, and a number like that would end a conversation rather than start one.

The 50 questions are the IPIP Big-Five Factor Markers, 50-item version, taken from the International Personality Item Pool. The items are used unmodified, in the published order, with the published scoring key.

The original scale was developed by Lewis R. Goldberg: Goldberg, L. R. (1992). The development of markers for the Big-Five factor structure. Psychological Assessment, 4, 26-42.

IPIP is a public-domain item bank, so there is no legal requirement to credit it. We do it anyway, because “built on decades of published, open research” is only worth saying if you can check it. The item text and the scoring key are here:

• https://ipip.ori.org/new_ipip-50-item-scale.htm — item text and administration order
• https://ipip.ori.org/newBigFive5broadKey.htm — scoring key

Item text verified against those pages on 2026-07-22. License: Public domain (International Personality Item Pool). Items unmodified.

No affiliation or endorsement. SelfWyse is independent. It is not affiliated with, sponsored by, or endorsed by the International Personality Item Pool project, Dr. Goldberg, or the Oregon Research Institute.

Two fair questions get asked about any test like this: does it give a steady answer, and does it measure the thing it claims to.

A steady answer. The usual check is whether the ten questions behind a trait agree with each other. It is reported as a number between 0 and 1, where above 0.70 is generally treated as acceptable and above 0.80 as good.

Below are the published figures for this instrument, next to the ones we calculated ourselves on the 695,290 people in our reference sample. We ran our own because a scoring key applied even slightly wrong shows up here first — numbers close to the published ones are good evidence we are scoring the questions correctly.

“Average score” and “spread” are the mean and standard deviation of the 10–50 total on each trait, in that same sample. Scores were recomputed a second time through independently written code on a sample of records; the two methods agreed exactly.

Measuring the right thing. The five-trait model is one of the most studied frameworks in psychology, and these items were written and tested specifically to measure it. That is the record we are standing on.

To be straight about the limit of that: the consistency checks above are ours, but we have not run a study of our own testing whether the scores predict anything outside the questionnaire. That work exists in the published literature on this model, not in our data. If we ever do run our own, it will be on this page.

The boundaries come from 695,290 people in the Open Psychometrics Project’s public IPIP dataset (IPIP-FFM-data-8Nov2018, collected 2016–2018). Respondents consented to their responses being used for research. The boundaries sit at the 10th, 30th, 70th, and 90th percentiles of that sample, worked out separately for each trait.

We use where people scored rather than assuming a bell curve, because real trait scores lean to one side and bunch up near the ends. Working from where people are describes them better than working from where a curve says they should be.

What we left out, and why. The raw file holds 1,015,341 responses. We kept 695,290 of them, 68.5%:

A record was kept only if all 50 items were valid, so the same 695,290 people contribute to all five traits.

An older, smaller Open Psychometrics file of 19,719 cases was deliberately not used. IPIP's own norms page flags it as unreliable — ambiguous direction on the Emotional Stability scale and a non-standard scoring method. Every score here was recomputed from the raw item responses rather than trusting any pre-computed totals.

What that sample is, and isn’t. These are people who chose to take a free personality test online. That group skews younger, more online, and more curious about personality than people in general. It is very large and it is real, but it is a volunteer sample. So the ranges describe your position among people like that, not among everyone.

• The intended lower bound was 16, matching the age range the instrument is validated for. This dataset records no age at all, so that filter could not be applied.
• The test was administered in English, so respondents self-selected for English proficiency regardless of where they were.
• Two people either side of a boundary are more like each other than either is to the far end of their own range. Boundaries are lines we drew to make description possible; nothing changes about a person when they cross one.

The full computation is a single script run against the published dataset. It applies only the filters listed here, and anyone with the public dataset can re-run it and get these numbers. Computed 2026-07-27.

If you have met the Big Five before, you may expect the fourth trait to be called Neuroticism. It is the same ten questions, read in the opposite direction: here, a higher score means steadier rather than more reactive. This is IPIP’s own labeling of the scale.

Because the direction is part of the scoring rather than just the name, it was a real choice. We went with Emotional Stability because Neuroticism is clinical vocabulary, and it carries a judgment the other four trait names do not — nobody reads “you scored high on Neuroticism” neutrally. Both ends of this scale, Sensitivity and Steadiness, describe something a person can reasonably be glad of.

If you have met the Big Five before, you may expect the fifth trait to be called Openness to Experience. That name comes from a different questionnaire. The fifty questions here are Goldberg’s Big-Five Factor Markers, and IPIP’s own name for this scale is Intellect or Imagination.

We did not use Intellect, because it sounds like a score for how smart you are, and it is not one. We did not keep Openness to Experience either, because these ten questions do not ask about experience. They ask about ideas: whether abstract ideas interest you, whether you have a vivid imagination, whether you are full of ideas. Not one of them asks about travel, art, or trying new things. Openness to Ideas is what the questions measure, so that is what we call it.

selfwyse helps you reflect and connect — not to diagnose, treat, or predict any medical, psychological, or mental health condition. It is a self-report questionnaire: it records how you answered 50 questions on one particular day, and people’s answers do move over time.

It is not for hiring, promotion, performance review, medical or mental health decisions, legal proceedings, or anything else with serious consequences for you or anyone else. The full version of that is on the disclaimer page.