How your picture
takes shape.

01 / The questionnaire

One shared core. Optional room to clarify.

Everyone starts with the same 48 statements, interleaved in four rounds across twelve dimensions. Each dimension has two statements oriented toward each pole, with equal weight. The revised wording edition keeps the same structure, weights, and scoring as the earlier balanced edition. Wording can affect responses, so the two editions are not assumed empirically equivalent. The earlier 36, 60, and 240-statement versions are retired; their saved progress is preserved, and old result links remain readable.

Four statements per dimension is a provisional balance between policy coverage and burden, not an established optimal length. This edition received an item-by-item wording review; participant pretesting remains necessary.

After the core, you can take up to 24 additional statements. A “not sure” response offers the uncertainty pair for that dimension. Otherwise, responses scoring at least 75 toward opposite poles offer its mixed-response pair. Mixed answers can reflect legitimate distinctions between policies; they are not labeled contradictions. No single left- or right-leaning answer chooses a branch.

Each clarification pair contains one statement for each pole. If you accept the round, both are asked. The follow-up plan is made from the whole core, is bounded to one pair per dimension, and does not recursively generate more questions. Changing a core answer keeps pairs that still apply and removes retired pairs. You can finish with completed balanced pairs, exclude a partial pair with an explanation, and add remaining eligible pairs later.

02 / Scoring and coverage

Neutral and unsure mean different things.

Strongly agree, agree, neither agree nor disagree, disagree, and strongly disagree use response values 1, 0.75, 0.5, 0.25, and 0. “Not sure / need more context” has no numeric value and is excluded.

For a statement oriented toward the left reference pole, the response stays as given. For a statement oriented toward the right pole, it is reversed. The score is 100 times the average of the two groups’ mean values. Each group contributes half, even if uncertainty leaves unequal numbers of rated statements.

A dimension needs at least one numeric response from each orientation to be shown. This is an operational coverage rule, not an established reliability threshold. Results list rated, unsure, neutral, and mixed responses. A midpoint can reflect neutral or opposing views; an unrated dimension is left blank. Catalog matches and Duel comparisons require coverage in all twelve dimensions.

The short core is sensitive to individual answers. One adjacent response change moves an axis by 6.25 points when all four statements are rated; the effect can be larger with fewer responses. Each dimension now shows its arithmetic step based on rated responses in each orientation. This calculation measures how the scoring formula reacts, not a margin of error. Clarifications add information, but do not establish scientific precision.

The closest-match illustration shifts one dimension at a time up and down by its larger adjacent-answer step, bounded to 0–100. It holds the other eleven fixed and re-ranks the ideology catalog. At a boundary, a zero-movement scenario is excluded. These score scenarios do not reconstruct actual answer edits or estimate likelihood. Tied displayed indices follow catalog order.

03 / Catalog interpretations

A nearby idea is a prompt to explore.

Ideologies, countries, regimes, and figures have existing editorial twelve-axis coordinates. Their numeric coordinates have not yet received a documented source assessment. Every profile has a source-context panel distinguishing identity, description, timeframe, image interpretation, and numerical coordinates. A checked name or membership reference does not validate the description as a whole or the political scores. Unreviewed entries are explicitly identified. Descriptions and numeric positions should be examined critically; a figure’s beliefs may change over time, and one number cannot represent a whole country.

The similarity index is 100 − √(mean squared axis distance). It gives all twelve dimensions equal weight, ranges from 0 to 100, and is rounded to one decimal place. It is a distance-based index, not the chance you belong to an ideology or agree with a person. Nearby ideologies appear in strict score order, including related variants. Close scores can change order after small response changes.

Catalog medians describe the profiles in this collection, not a population survey. Matching a government or historical figure never implies endorsement of its actions. Catalog coordinates are retained for continuity; this release does not pretend that rewritten questions make those interpretations validated.

04 / Privacy and editions

Your answers stay on your device.

Progress and your latest completed answers are stored locally when browser storage is available. Shared result links contain twelve scores, edition identifiers, and per-dimension response counts and flags. They do not contain individual statement responses. A recipient cannot open your private answer review. Opening a shared profile keeps your completed assessment; using it as the saved Duel profile requires an explicit action. Editing completed answers keeps a separate copy from any active quiz draft.

Both balanced wording editions retain their own coverage and original statement definitions. Resuming or editing an earlier completed assessment keeps its wording; a fresh start uses the current edition. Legacy and vector-only links show scores without coverage. Original response evidence cannot be reconstructed from a vector. Similarity rankings on those links use the current matching method, and may differ from earlier releases.

05 / What still needs research

Better measurement needs people, too.

The next research steps are cognitive interviews about how statements are understood, expert review of construct coverage, diverse pilot samples, repeated assessments, and analysis of reliability, redundancy, and sensitivity. Each catalog coordinate also needs a dated source, rationale, and review record. Until then, this remains an exploratory model.

Questionnaire design references: Pew Research Center’s question-writing guidance ↗ and AAPOR’s survey best practices ↗. These references inform design principles; they do not validate this questionnaire.

Clear comparisons, with context.

Results open with shared linear scales for comparing positions. The optional shape view offers a pattern overview; neither chart implies quality or certainty. Duel uses the same similarity equation, a captioned twelve-row table, named reference poles, and an ordering control. Its small, middle, and wide gap counts are declared display thresholds, not scientific groups.

Presentation guidance: W3C’s guidance on complex images ↗, accessible tables ↗, and GOV.UK’s mobile and readable layout guidance ↗. These guide the interface; they do not authenticate catalog claims.

Explore the statements and their scope.

The labels describe this edition’s scope. Open a dimension to read its context and all four core statements.