FAQ

What is this measuring?

Two things at once. The horizontal axis is economic: how much you want wealth and industry held in common versus left to markets and private owners. The vertical axis is social: how much authority, tradition and national loyalty you are willing to impose, versus how much personal freedom you want left alone. They are separate questions, which is why a single left-to-right line cannot hold both.

Why is there no neutral option?

Because a neutral button is where people hide. Offered one, a large share of answers land there regardless of what the person actually thinks, and the result says nothing. Forcing a lean costs you a little precision on any single proposition and buys back far more across the whole test. The propositions are balanced, so being forced off the fence does not push your score in a particular direction.

Why do the questions change depending on my answers?

The test is adaptive. After each answer it re-estimates where you sit and then picks whichever remaining proposition would tell it the most about you. Someone who answers as a committed socialist gets asked about worker ownership; someone near the centre gets asked milder things. It stops once both axes are pinned down closely enough, rather than marching everyone through a fixed list. Answers that are clear-cut settle quickly; less settled patterns take more of the set. You may also notice propositions arriving in rough opposing pairs. That is deliberate: it cancels out the common habit of agreeing with statements, whichever way they point.

How are the coordinates calculated?

Each proposition carries a position, roughly how far out you have to be before agreeing with it becomes natural, and a weight for how sharply it separates people. Around that position sit three thresholds, one for each step of the answer scale, so strongly agreeing counts for more than agreeing, and both count differently again from the two ways of disagreeing. This is the graded response model, the standard way of scoring ordered answers like these. Your position on each axis is the value that makes the whole set of your answers most likely, so agreeing with a moderate statement moves you a little and agreeing with a far one moves you a lot. One honest limit: where each proposition sits on the scale is judged against a broadly Anglo-American political centre, so readers elsewhere should expect the middle of the map to feel slightly displaced from their local one.

Can I actually reach the corners?

Yes, and that is deliberate. Every direction has propositions running out to the edge of the scale, and the scoring applies no pull toward the middle, so answering consistently at the extremes will place you at the extremes. A test whose corners cannot be reached is drawing a map with regions nobody can occupy. The cost is that the propositions at the far end state positions many people find repellent; they are there to be measured against, not endorsed.

Where do the party and public figure positions come from?

They are editorial estimates: readings of public positions, manifestos and voting records, placed on the same scale as the propositions so the comparison is at least consistent. They are not measurements, nobody sat these people down with the test, and reasonable people will disagree with individual placements. Treat them as a rough map for orientation, not a verdict on anyone.

What is the Overton window box?

The dashed rectangle marks roughly the range of positions that count as ordinary mainstream politics, the span you could hold and still be electable. It moves between countries and it moves over time. Sitting outside it does not mean you are wrong; it means your views are not currently represented by that mainstream.

Is any of this stored or sent anywhere?

No. There is no account, no database, no analytics and no tracking. Every proposition, every calculation and every line of commentary happens in your browser. The only things the page ever requests are the portrait images, and those are served from this site itself, never from a third party, so no outside company sees that you were here. Nothing about your answers is transmitted at any point. Closing the tab destroys the result. If you want to keep it, use "save image" or "copy link".

Where do the portraits come from?

Wikimedia Commons, restricted to files whose licences permit reuse. Every file, its author and its licence are listed on the image credits page. Where no freely licensed image exists for someone, the site shows a lettered badge rather than borrow a picture it has no right to use.

How does the share link work if nothing is stored?

Your coordinates are written into the part of the link after the #. Browsers never send that part to a server, so the numbers travel inside the link itself and nowhere else. Anyone opening it sees your position rendered from those two numbers, not your answers, which never leave your device. For the same reason, a shared result is a claim, not a certificate: anyone can put any numbers in a link. The test measures the positions a person expresses, not whether they mean them.

How sure is the test about my placement?

Exactly as sure as the percentages say. Each axis carries a certainty figure: the probability, given your answers, that your true position lies within one point of the mark. It starts low, rises as your answers agree with each other, and falls when they pull in different directions. It never reaches 100, because certainty about a mind is never total. The same figures move live during the test, so you can watch the reading settle.

How accurate is my result?

Simulated against respondents whose true position was known, the test recovers that position with a typical error of well under a point per axis and no systematic pull toward the centre or the edges. The certainty figures are honest too: checked in simulation, positions claimed at a given certainty were right at least that often. Do not read much into the difference between −4.1 and −4.5. The quadrant, and roughly how far out from the centre you sit, are the parts worth taking seriously.

How is this better than other political tests?

Four ways, each checkable. It is adaptive: it asks statements chosen for what they will teach it about you, never finishing below 80% certainty, instead of marching everyone through the same long list. It is honest about certainty: it reports a percentage that starts low, can fall when your answers conflict, and never claims 100, where most tests print a dot as if it were fact. It is provable: the how-it-works page derives the method from stated premises, the figures on it are computed by the same code that scores you, and the results page lets you delete any answer and watch how little the mark moves. And it collects nothing: no account, no analytics, no stored answers.

Your questions were written by hand. Why should I trust the percentages?

The statement positions are authored judgment and the site says so plainly; that is premise one on the how-it-works page, and it is the same footing every political test stands on, stated or not. What is not judgment is everything downstream: the arithmetic is checked in simulation, where positions claimed at a given certainty were right at least that often. Treat the percentage as a well-constructed estimate, not a laboratory measurement.

Two axes cannot capture politics.

Correct, and the test says so when it matters. Views that do not reduce to two numbers show up as lower certainty and a plain sentence telling you to read the score as a region. Two axes is the deliberate ceiling because the output must be readable at a glance; models with more dimensions exist and are better suited to research than orientation.

A friend's placement of some figure looks wrong to me.

Figure and party placements are estimates of public records held to the same scale as the test, with mainstream officeholders placed inside the range of mainstream politics of their era. Disagreement about individual placements is expected and legitimate; the placements carry no authority over your own result.

I disagree with my result.

The test can only reflect the answers you gave it. If the result feels wrong, retake it and notice which propositions you hesitated on, because that hesitation is usually the interesting part. It is also worth remembering that this measures stated principles, not what you would do in office, and the two come apart constantly.