How the Values Assessment works

The 48 values and four dimensions

The assessment presents 48 values organized as four dimensions of twelve. The current list is a 2026 revision of a list that ran on our previous platform for years. The original derivation predates the current codebase and was not formally documented — we won’t reconstruct a tidier origin story than the record supports. The 2026 revision was theory-driven: several values were retired, and Loyalty, Tradition, Stability, and Vitality were added specifically to cover the conservationregion of Schwartz’s theory of basic values, which the earlier list under-represented. The four-dimension structure guarantees that every respondent considers values from all four regions of life, not just the region they reflexively identify with.

Statements, not labels.During discovery you never see the value names. Each value appears only as a first-person, gut-feel statement — “Earning a respected place among your peers” rather than the word “Status” — because single-word labels invite social-desirability responses: people pick the word they believe they should want. Names are revealed only at the results stage. This is a design rationale, not a formally tested claim about our instrument specifically.

How the assessment works

1. Selection. Four screens, one per dimension, each showing twelve statements. You select up to three per dimension — the cap reduces choice overload and stops anyone filling their shortlist from a single region of life.
2. Narrowing. Your selections are shown together and you narrow to exactly five — the first forced trade-off across dimensions.
3. Pairwise ranking. The five finalists are ranked through a full round-robin of all ten head-to-head pairs (not an adaptive subset), with the pair order arranged so no value appears in consecutive match-ups.
4. The ranking. Values rank by number of pairwise wins; ties (including circular preferences) resolve deterministically by your earlier picks. You can reorder the result by hand at the reveal — and whether you did is recorded with the result, so adjusted rankings are distinguishable in the data.

Why forced trade-offs, not ratings

Asked to rate values on importance scales, most people rate most values as important: values are, by nature, things people endorse, so rating data shows severe ceiling effects and weak differentiation. Values only become informative when they compete. The assessment therefore never asks “how important is X?” — only “X or Y?”, at three escalating levels: within-dimension selection, cross-dimension narrowing, and head-to-head ranking.

The design draws on a research lineage we state as lineage, not as validation: Schwartz’s theory of basic values (list coverage), Self-Determination Theory and Kasser’s aspirations research (the intrinsic/extrinsic lens), and Rokeach’s terminal/instrumental distinction (the means/end classification). No existing psychometric instrument was adapted, licensed, or replicated, and the assessment has not been validated against any of them. For the wider history of the field, see our short history of values research.

Archetypes and motivation styles

Eight archetypes (Harmonizer, Connector, Explorer, Creator, Guardian, Leader, Giver, Sage) and five motivation styles (Connection-Driven, Achievement-Oriented, Freedom-Seeking, Meaning-Centered, Stability-Seeking) are assigned by the same engine: pure, deterministic arithmetic. No AI is involved in scoring — an AI writes narrative text around your result afterwards, but it cannot change the result.

Each of the 48 values carries hand-calibrated loadings onto each archetype and style. Your top five values are weighted by rank, adjusted by distinctiveness multipliers computed from more than 11,000 completed assessments — very common picks (like Family) are damped and rare, distinctive picks are lifted, so near-universal values don’t dominate every profile — and scored against each archetype with a normalization that stops heavily-loaded archetypes from winning by default. When the top two scores are close, the result is flagged as blended and presented that way. Identical inputs always produce identical outputs, and the scoring runs under automated snapshot tests, including sixteen constructed exemplar profiles that each must land on their expected archetype.

Calibration status, stated plainly: the loading matrices reflect expert judgment plus consistency testing. They were not derived from factor analysis or clustering, and no inter-rater or external validation of archetype assignments has been conducted.

The well-being model

Nine factors in three pillars — Energy (Vitality, Movement, Rest), Connection (Positivity, Empathy, Support), and Mindset (Focus, Clarity, Presence) — all answered on the same 1–7 scale. Five factors come from the free assessment; four are asked in a deeper Premium survey. There is no composite index: answers are stored and shown as your raw self-ratings, banded for display. The nine-factor structure is our own model, designed for this product; it is not an adaptation of an existing validated well-being scale and has not been independently validated.

Your inner/outer orientation is computed from your values, not your well-being answers: each value carries an intrinsic-to-extrinsic orientation assigned by expert judgment grounded in Self-Determination Theory, and your lean is the rank-weighted average across your top five. It is presented as a reflective lens, not a verdict — and the well-being numbers and values pattern never modify each other. They sit side by side on the premise, borne out correlationally in our dataset, that knowing your values without the capacity to live them is where strain shows up.

The dataset and what we claim from it

The Global Values Report 2026 analyzes 9,656 people who completed the assessment between June 2025 and April 2026, spanning 114 self-reported countries of residence (the largest groups from the UK, US, Australia, the Philippines, and Canada; roughly two-thirds female). Of those, 9,327 completed the pre-assessment check-in and 6,231 the nine-factor well-being module. More than 50,000 people have taken the assessment across all platform versions since it first launched; the analyses on this page and in the report use the current, verifiable datasets.

All analyses are descriptive and correlational — frequencies, cross-tabulations, and composite proxy indicators. No inferential statistics were applied, because the sample is self-selected and non-probabilistic, and no causal claims are made. The central finding, the values-living gap, is the recurring pattern that clarity about your values does not imply the capacity to live them: one in four respondents reporting high values clarity also report high stress, and self-rated purpose clarity shows a strong monotonic association with lower stress and higher fulfillment.

Career directions

Career directions (a Premium feature) are generated by a large language model working from your top five values, archetype, and orientation, under a versioned prompt with strict rules: the frame is fit, never fate — directions whose day-to-day reality feeds your values, with honest cautions on each — and never predictions, aptitude claims, or advice to quit anything. This is editorial matching within human-set constraints, not a labor-market model or a validated career-interest inventory, and the product says so.

Version history

The assessment ran for several years on its original platform, evolving to the instrument that collected the Global Values Report dataset. In July 2026 the app was rebuilt from scratch with the scoring algorithm frozen as a versioned package under snapshot tests; the rebuild introduced the full round-robin ranking, the revised value list, the fifth motivation style, the distinctiveness multipliers, and the statement-first discovery format. Scoring has not changed since the freeze, and any future change will increment the algorithm version stored with every result.