How the 12 Axes Test Works

Choose your level of detail.

All three versions show twelve axes. Quick provides an initial picture with fewer questions. More questions cover more of the source pool; this is not a guarantee of scientifically established accuracy. Changing test length starts a new test.

What the axes explore.

Explore all twelve dimensions and both ends of each spectrum in our guide to the twelve axes. The guide brings the definitions together in one place; this page explains the question selection and scoring.

How questions are selected.

The question pool comes from the open-source politicaltests 12axes project. Full uses all 288 statements. Quick selects three questions per axis; Standard selects five.

All three versions include the same sixteen original statements for historical comparison: two in each of its eight dimensions, one in each scoring direction. Quick and Standard draw the remaining statements from different source groups, with either a two-to-one or one-to-two direction split in Quick, and a three-to-two or two-to-three split in Standard. Source groups are not independently validated topic categories.

Your question set and order are fixed when you start. Refreshing or returning to a previous question does not draw a different set. Longer versions provide more statements for your twelve-axis result; they do not guarantee a more accurate historical match. This site's shared-statement sampling is an extension of the original questionnaire.

How scoring works.

Five responses run from Strongly agree to Strongly disagree, with values 2, 1, 0, −1 and −2. Each answer is multiplied by the statement's scoring direction. For each axis we calculate:

50 + 25 × mean(direction-adjusted answers)

The displayed left score is that value; the right score is its complement to 100. Scores use the actual number of questions answered on that axis and are rounded to one decimal place. Editing an answer replaces its contribution.

Saved tests from the earlier seven-response version keep their original choices and answers. Earlier result links remain readable, and historical reference readings retain their recorded evidence ranges.

Neutral answers produce 50 / 50. A centered result can also come from opposing answers cancelling out. Percentages are positions on this test's scale, not probabilities, confidence levels or percentages of people who agree with you.

What your result does—and does not—say.

These readings summarize your answers to a limited set of statements. Wording, interpretation and the selected questions can affect them. They are not a diagnosis or a permanent identity.

Your twelve-axis result remains the main reading. The historical comparison is an additional perspective; this version does not assign you a country or an ideology. There is no single combined political score.

Saving and sharing.

Progress is saved in this browser where storage is available. You can continue, clear it or start again. Your result link contains scores and a test version, not individual answers. Anyone with the link can read the result. Opening another person's result does not replace your unfinished test.

Read our privacy information for the details.

How historical comparison works.

We reconstruct responses to sixteen shared statements from dated writings, speeches and historical records. The references include different political perspectives. Each profile identifies a period and whether it describes public positions or institutional practice.

Comparison uses eight dimensions, each with equal weight, and the same sixteen statements in every test length. These shared-statement readings are calculated separately from your twelve total scores. A profile needs evidence for at least fourteen statements and at least one in every comparison dimension. This is an editorial evidence requirement, not an accuracy guarantee.

Unknown statements stay unknown. They widen the possible reading rather than becoming neutral answers. We order references conservatively by the largest distance allowed by the recorded evidence ranges. Less evidence cannot improve a comparison by dropping an inconvenient statement. Ties favor narrower evidence ranges, then fuller evidence. Where ranges overlap, the order is approximate.

When fewer than four comparison dimensions contain a non-neutral response, the preview does not force a person onto the result. This is a conservative editorial rule. The preview first shows up to three nearest references from an editorial selection of familiar figures. That subset is labeled separately; expand the full-library view to see the overall nearest three, including specialist thinkers. Recognition does not change scores, weights or the full ranking. Cards identify the person, period and evidence, with actual shared and different responses and links to sources. It is a relative comparison within this library, not proof that anyone is close to you in every respect. We do not report a similarity percentage or a scientifically validated identity. More questions broaden your twelve-axis result but do not guarantee a more reliable historical comparison. Versioned v2, v3 and v4 links use their saved reference libraries. Old links without shared-statement answers cannot generate this comparison.

Sources and credits.

12Axes is an independent implementation using the politicaltests question pool, under its MIT license. This build corrects question-count descriptions and normalizes scores by actual per-axis counts. Quick is an added sampling mode.

Homepage animation adapted from Unwoven by Clément Grellier. See portrait sources and animation dependency notices. We are not affiliated with other 12Axes sites or with Adobe.

Compare 8values and 9axes to explore different approaches.