Method
The three axes
- Ownership
- Who owns and directs productive capital, and how output is allocated. This axis is not a measure of how much government exists. State ownership is not treated as collective ownership here: an enterprise owned by a ministry and run on hierarchical wage labour is not worker control, and a centrally directed economy that subordinates workers to an administration is not communal ownership either. Questions about how much power an authority holds belong to the Authority axis instead. 0: capital owned and directed collectively (by workers, communities, or commons) with output allocated toward need and use 100: capital owned and directed privately, with output allocated through markets, prices, and managerial or owner authority
- Change
- Orientation toward social change, inherited norms, and the scope of the moral community. This is not an economic axis: ownership, taxation, and the size of the state are measured elsewhere. It concerns what is valued as normal and legitimate, and how far inherited arrangements should be conserved or revised. 0: inherited norms, institutions, and particular attachments treated as authoritative and worth conserving 100: inherited arrangements treated as revisable, with weight given to individual self-determination and to widening the scope of who counts
- Authority
- How much concentrated, unaccountable power over the individual is acceptable, whether held by a state, a party, an employer, or a platform. This axis is about the structure of power rather than its ideological colour, so it applies equally to public and private authority. It does not measure economic policy or cultural values. 0: concentrated decision-making, obedience, and compliance with hierarchy treated as legitimate or necessary 100: dispersed decision-making, consent as the basis of authority, and enforceable limits on concentrated power
Scoring
A graded response model for agree/disagree and true/false items; a position-based model for multiple choice, where each option is itself a point on the axis. Estimates are posterior means under a standard-normal prior, with intervals of ±1.96 posterior standard deviations. Skipped questions are excluded rather than counted as centrist.
Limits
The item parameters are provisional. Calibration needs response data from a large representative sample, and none exists for this instrument. The discrimination and threshold values encode an explicit judgement about what each statement measures and how strongly, and are published in full in src/engine/items.js.
The instrument also reports less precisely near the extremes of each axis, and on the right pole less precisely than the left. The extreme difficulty band holds 12 of 230 items, so a score beyond roughly 80 should be read with a wider interval than the number alone suggests.
Every coordinate on the figures page is an editorial estimate, not a measurement. No one listed there took this test.
Privacy
No accounts, no cookies, and nothing that identifies you. Scoring, rendering and export all run in your browser. A shared link carries scores in the URL fragment, which browsers never send to a server.
One thing is sent, and it is worth being precise about. When you finish a test, your answers are added to an anonymous tally so the questions can be improved. Nothing identifying is included: no account, no cookie, no IP address, no device information, no timestamp, and no link between one submission and another. Each answer increments a separate counter, so what is stored is a count per question per answer, not a record of anybody's result.
The reason is that the item parameters are judgement values that have never been fitted to real answers, which is the main weakness of this instrument. Aggregate counts cannot fix that; a response matrix can. If you would rather not contribute, the test works identically either way, because scoring is entirely local and does not wait for the tally.
There is also a plain visit counter, shown in the footer. It is a count of visits rather than of people, because counting people honestly would need an identifier and this site keeps none. A flag in the tab's session storage stops a refresh adding to it, and that flag disappears when the tab closes, so nothing persists and nothing carries over between your visits.