TraitAtlasHow TraitAtlas works, and where it stops.
Every report should be explainable. Here is exactly what we measure, how scores are calculated, where the questions come from and what results can't tell you.
Our principles
- Measure only what a test is designed to measure.
- Keep scoring deterministic, transparent and version-controlled. The same answers always produce the same scores.
- Separate personality, interests, preferences, wellbeing, ability and demonstrated competence.
- Explain uncertainty and missing evidence instead of hiding it.
- Never use these results alone for hiring, promotion or pay decisions.
The tests
| Test | What it measures | Evidence level |
|---|---|---|
| Four Styles 40 items, about 6 min | Discover how you show up in daily life. | Introductory framework |
| Big Five Quick 20 items, about 4 min | Your Big Five snapshot in under four minutes. | Research-based (public domain) |
| Big Five Personality 50 items, about 10 min | The most researched model of personality, in plain language. | Research-based (public domain) |
| 16 Personality Types 32 items, about 6 min | Your four-letter type, with a character to match. | Introductory framework |
| Career Interest Explorer 60 items, about 8 min | Find the kinds of work you'd actually enjoy. | Research-based |
| Values & Motivation 28 items, about 6 min | What actually keeps you going at work. | Introductory framework |
| Work Style & Preferences 24 items, about 6 min | How you like to plan, collaborate, focus and communicate. | Introductory framework |
| Wellbeing Check-in 10 items, about 3 min | How you've really been feeling lately. | Research-based |
Where the questions come from
- Big Five uses the 50-item IPIP Big-Five Factor Markers (Goldberg, 1992). Big Five Quick uses the Mini-IPIP (Donnellan, Oswald, Baird and Lucas, 2006), 20 of those kinds of items. Both come from the International Personality Item Pool, which places its items in the public domain.
- Career Interest Explorer uses the 60 work activities of the O*NET Interest Profiler Short Form from the U.S. Department of Labor, Employment and Training Administration (USDOL/ETA), used under the O*NET Tools Developer License. The activities, the five-point like-to-dislike scale and the scoring are unchanged; a score of 0 to 100 here is the standard 0 to 40 raw score on a wider range. The report and India-specific career ideas are ours. TraitAtlas has modified all or some of this information; USDOL/ETA has not approved, endorsed or tested these modifications. O*NET® is a trademark of USDOL/ETA.
- Wellbeing Check-in combines the WHO-5 Well-Being Index (World Health Organization) and the Satisfaction With Life Scale (Diener, Emmons, Larsen and Griffin, 1985), in their original wording and scoring. The WHO-5 is © World Health Organization and used under the CC BY-NC-SA 3.0 IGO licence, so this check-in is free and never sold; WHO does not endorse TraitAtlas. The Satisfaction With Life Scale is used with credit, as its authors permit.
- 16 Personality Types is an independent type framework. Most of its statements come from the IPIP and relate to four of the Big Five factors; the rest are TraitAtlas statements in the same style. It is not affiliated with any other type test.
- Four Styles follows the classic four-temperament tradition with original words. Values & Motivation and Work Style are designed by TraitAtlas and are in early validation.
How scores are calculated
Rating-scale tests
Each statement is rated on a fixed scale: five points for most tests, 0 to 4 for career interests, and 0 to 5 and 1 to 7 for the two parts of the Wellbeing Check-in. Some statements are worded in the opposite direction and are reverse-scored, which reduces the effect of agreeing with everything. A scale's score is the average of its items, shown from 0 to 100 on the scale range, and bands (lower, balanced, higher) describe where you sit on that range. For the WHO-5, the 0 to 100 score is the standard raw score multiplied by four.
If you skip too many items on one scale, we mark it as "not enough answers" rather than guessing.
Types and codes
16 Personality Types: each of the four preferences gives one letter, from the side you score on, and the bar shows how clear that preference is. Near the middle means you use both sides comfortably. Career Interest Explorer: your interest code is your top three areas, in order.
Word choice (Four Styles)
In each of 40 groups of four words, 20 about strengths and 20 about weaknesses, you pick the one most like you. Each word maps to one style. Your profile shows the share of your picks per style, overall and separately for strengths and weaknesses, so each set adds up to 100%. Picking among equally positive or equally negative words makes the result harder to fake than "agree or disagree" questions.
Characters
Characters such as The Banyan Tree are memorable names for a result, drawn from everyday Indian life. They come from fixed rules (your main style, type, interest area or strongest scale) and never from AI. They add nothing to the score.
Norms
Scores are positions on the scale, not percentiles against other people. As anonymised data builds up, we will publish Indian norm groups and say clearly when a report uses them.
Versioning
Every test has numbered versions. Once a version is published, its questions, scoring and interpretation rules are frozen and fingerprinted. Your report always states the version you took, and older reports keep the rules they were created with. Translations change wording only, never scoring.
Where AI fits
AI never scores you, sets your character or changes a report. It is used only for optional, clearly labelled AI writing: Atlas Profiles, manager's guides and team summaries, which are generated when someone asks. These read your fixed scores, cite the tests each point comes from, and follow written rules: no clinical, hiring or pay claims. They are stored as versions so you can compare them over time. AI can still be wrong, so treat its writing as a starting point for reflection.
Compatibility reports
Comparisons use fixed rules: how far apart two people are on each scale of the same test, plus each person's own report. They describe similarities and differences and suggest how to work well together. They do not predict whether a relationship will succeed.
Reliability and fairness
For our own questionnaires we are collecting consented, anonymised data to report internal consistency (Cronbach's alpha), test-retest stability and differences across groups such as gender and language. We will publish these results, including any items we change as a result.
Team insights
Team patterns are distributions, not rankings. They are hidden until at least five people complete (an organisation can raise this threshold). A missing test is never treated as a trait. The Wellbeing Check-in is never offered through an organisation.
Questions or a disputed result?
If you think a report is wrong or unfair, use "Dispute a result" in your account, or write to support@traitatlas.local. A person will review it.