Methodology

The principle

A public instrument will be challenged, usually by somebody unhappy with the result. The only defence that holds is the methodology published beforehand, not the explanation given afterwards.

So the method note is written in parallel with the instrument and published with it.

How the items are written

We start from the literature on the subject and, for electoral compasses, from the candidates’ manifestos and public statements.

Statements are worded neutrally and tested on a small group before launch.

Any that fail to differentiate answers are cut. An item everyone answers the same way wastes the participant’s time.

How the match is calculated

Answers are scored on axes defined before collection, with weights also set beforehand.

The match with a candidate or a profile is a distance computed on those same axes, not a classification adjusted after we have seen the results.

The formula and the weights are published, so anyone can redo the calculation.

What we do not claim

An online instrument does not produce a representative sample just by having many respondents.

We publish representativeness indicators and state explicitly which conclusions cannot be drawn from the data collected. In practice this is the most important section of any report we deliver.

Independence

We do not work for campaigns that ask us not to publish the methodology or to adjust items in somebody’s favour.

Candidate positioning is based on their public statements, and every candidate is given the same opportunity to confirm their positions.