Methodology

How the axes are scored

This page is the whole method: what each axis measures, how a click becomes a number, why half the statements are worded backwards, what the match percentage actually says, and — set out plainly — the things this test cannot tell you about yourself.

If any of it looks wrong, that is a useful thing to tell us. A scoring rule you cannot inspect is not a scoring rule, it is a claim.

The two axes

Each axis runs from −100 to +100 and is scored from its own forty statements. No axis borrows an answer from another, which is what lets them come apart — and coming apart is the entire reason for having two instead of one.

Economic

How much of collective life works better coordinated, and how much through exchange between individuals?

−100 · Shared provision

Outcomes are heavily shaped by structure, so the things everyone needs should be arranged collectively rather than sold.

+100 · Open markets

Competition and voluntary exchange allocate resources better than any planner can, and coordination has costs people underrate.

Example statement · forward-keyed

“Most services improve when providers have to compete for customers.”

Agreeing moves you toward open markets.

Example statement · reverse-keyed

“Some needs are basic enough that access to them should not depend on ability to pay.”

Agreeing moves you toward shared provision.

Stability over time. The more stable of the two. It rests on generalised evidence from lived circumstance, so it moves when circumstances change rather than when arguments do.

State

When individual freedom and coordinated order pull apart, which do you reach for before the details arrive?

−100 · Personal liberty

No central body holds enough information to decide well on an individual’s behalf, and every constraint should have to justify itself first.

+100 · Collective authority

Many outcomes people want are only reachable together, and "together" requires the capacity to bind those who would opt out.

Example statement · forward-keyed

“A rule that lowers a shared risk is worth the freedom it costs.”

Agreeing moves you toward collective authority.

Example statement · reverse-keyed

“A restriction should have to prove its case before it is imposed, not after.”

Agreeing moves you toward personal liberty.

Stability over time. The most situational. Perceived threat pulls almost everyone toward authority, and a stable period lets them drift back — measurable within months.

Item design

Why half the statements are worded backwards

People agree more than they disagree. Put a confidently phrased sentence in front of someone moving at speed and there is a small, persistent, well-documented pull toward saying yes to it. It is called acquiescence bias, and it is not a character flaw — it operates on careful people too.

Now imagine an axis where every statement is phrased so that agreeing means “markets”. An agreeable respondent finishes that test pegged at one pole, having told you nothing except that they are agreeable. The instrument has measured its own wording.

Reverse-keying fixes it structurally rather than statistically. Exactly half the statements on each axis — ten of twenty — are written so that agreement points the other way. The agreement tendency now pushes in both directions at once and cancels, and what survives is the part of your answers that reflects the actual disposition.

It costs something, and we should say so: reversed items demand closer reading, and a respondent skimming quickly can misread one. That is a real source of error. It is a much smaller source of error than an axis that measures agreeableness.

Worked example · state axis

Someone who agrees with everything, on an axis with twenty items:

If all twenty ran one way: Twenty agreements × one direction = a score pinned at +100. Reported as a strong authority lean. Actually measured: agreeableness.
With ten forward, ten reversed: Ten agreements push toward authority, ten push toward liberty. They cancel exactly, and the score lands on zero — which is the honest answer for someone whose responses carried no directional information.

The same logic runs on both axes. A person who answers without reading gets a result near the centre rather than a confident, flattering and entirely invented one.

The arithmetic

From a click to a position

1. Each answer becomes a number from −2 to +2
Strongly agree is +2, agree +1, neutral 0, disagree −1, strongly disagree −2. Five points rather than a slider, because people cannot reliably distinguish more than about five levels of their own agreement — a hundred-point scale adds noise, not resolution.
2. Reversed items are flipped back
Every statement carries a direction of +1 or −1. The response is multiplied by it, so that after this step every item on the axis points the same way and can be added together honestly.
3. The axis total is normalised to −100..100
The directed responses on an axis are summed and divided by the largest total that axis could produce. The result is a proportion of the maximum, scaled to a hundred — which is why the number is comparable across axes even though it is not a percentage of anything in the world.
4. Skipped and neutral answers dilute rather than distort
A neutral contributes zero. It pulls your score toward the centre rather than in a direction, which is the correct behaviour: an answer that carried no information should not manufacture one.

Both axes, plotted

The compass view shows economic against state, which is the whole of the result: two axes are exactly what a two-dimensional chart can hold, so nothing is reported alongside it that the picture leaves out.

Reading the number

What the match percentage means

Your two axis positions are a point on a plane. Each named profile is a region of that same space with a centre. We find the centre nearest to you, hand you its name, and then — this is the part most tests leave out — tell you how far away you were.

The match percentage is that distance, inverted and scaled. It answers the only question that matters about a label: how well does this one actually fit me?

85–100%

You sit close to the centre of the profile. The description should read as though it was written about you, and most of it will be.

70–85%

A good fit with real exceptions. Expect a couple of paragraphs that land squarely and one that does not — the report says which is which.

Below 70%

You are between descriptions. Reading only the named one will mislead you — your coordinates are the finding here, and the label is barely a summary of them.

A low match is not a failed test. It is the instrument declining to overstate itself, which is the behaviour you should want from it.

Limitations

What this test cannot tell you

A paid report is only defensible if the free claim is accurate. So here is the list, in plain terms, of what 50 statements across two axes genuinely cannot establish about a person.

How you will actually behave

Dispositions are one input into behaviour. Circumstance, relationships, what you know at the time and plain self-interest are the others, and in any specific situation they frequently dominate. A position on an axis is not a prediction.

Whether you are right

The instrument has no view on which positions are defensible. It measures where you sit. Nothing in the scoring rewards one pole, and no result page tells you that your answers were good ones.

How much you have thought about any of it

A position held carefully for twenty years and one assembled during the test produce identical response patterns. There is no signal in the data that separates them, and we do not pretend to find one.

Who you will be in five years

One of the two axes responds to circumstance within months. A result is a reading taken on a particular day, not a permanent property of you.

Anything clinical

This is not a psychological assessment, a diagnostic instrument or a screening tool of any kind. It has not been validated against clinical criteria because it is not trying to measure anything clinical.

What you should do about it

The test describes; it never prescribes. No statement in it references a party, a candidate, a campaign or an election, and no result tells you how to act on what it found.

When we change the instrument

Items get replaced when they stop discriminating — when nearly everyone answers the same way, or when responses to an item correlate more strongly with a different axis than with its own. Both are signs the statement is measuring something other than what it was written for. When an item changes, the scale it feeds changes with it, so scores from before and after a revision are not strictly comparable. We would rather say that than quietly pretend a five-year-old result and a new one sit on the same ruler.

Further reading on the reasoning behind all of this: what a spectrum test measures, compass versus typology and reading your result honestly.

You have read the rules

Now see what they produce.

50 statements, two axes, about 10 minutes — scored exactly the way this page describes.

Take the political spectrum test Browse the insights

Free to take. Full results require a plan.