A political compass with its working shown

Sextant

Every political compass decides its axes before anyone answers a question. This one found them: five dimensions, measured in the responses of 37,583 people, not assumed in advance.

Fifty questions, about six minutes. You get a position on each axis with its error bar, because a dot claims a precision no questionnaire has.

One of these is not like the others. Institutional trust turns out to be close to statistically independent of both ideology and party — two people with identical politics can sit at opposite ends of it. A two-axis compass cannot see it at all. That is the main reason this exists. Close to, not exactly: within a single survey year the correlation with self-reported ideology runs from −0.17 to +0.09, and its sign follows whoever holds the presidency. That is a tenth of what the redistribution axis scores, and it is not zero.