Frequently asked questions

DataRadio is a continuous global public-opinion poll. It asks people in every country the same forced-choice questions about economics, social norms, sovereignty and the use of force, weights the answers to each nation's population, and publishes the result nation by nation. Answering is free and requires no account.

Every item is a pair of written statements — side A and side B — rather than a claim to agree or disagree with. Both sides are written to be endorsable by a reasonable person, and which side appears first is randomised per respondent, so position cannot bias the result. A respondent's score on an axis is the weighted mean of their answers, rescaled to a range of -100 to +100.

Economic (market versus state allocation of capital and risk), Social (how fast a society revises inherited norms), Sovereignty (whether binding authority sits with the nation-state or above it), and Force (when coercion — sanctions, arms, troops — is legitimate). Left-and-right fuses these four; separating them is the point.

Only when that country has at least 100 responses on the question. Below that threshold no national number is published, and the page says how many more answers are needed. Any national figure attributed to DataRadio below 100 responses did not come from DataRadio.

Published national readings are raked to each nation's adult population on age, gender and education using UN World Population Prospects and UNESCO attainment data, and countries are post-stratified by adult population for world figures. Tallies labelled 'unweighted' are raw counts and are labelled that way wherever they appear.

No, and DataRadio does not claim otherwise. A self-selected web sample skews young, online and male. Weighting corrects the composition it can measure — age, gender, education, country — and cannot correct for who chose to visit at all. Every figure carries an effective sample size and a margin of error so the uncertainty travels with the number.

Cite it as: DataRadio, 'Global public opinion', dataradio.org, accessed on the date you read it, with the n and margin of error shown beside the figure. Because polling is continuous, a figure without a date is not reproducible.

Answering the poll and reading every published result is free to everyone, permanently and without an account. Paid tiers exist — Supporter and Pollster — but no analytical feature is behind them. Charging for persistence would bias the long-run panel toward people who hold payment cards, which is the exact distortion this site measures.

IP addresses are never stored; they are read to determine country and to rate-limit, then discarded. There is no device fingerprinting. Answers are stored against a random identifier held in a cookie, never against a name, and an account is optional.

Anyone can propose a topic, free of charge. Proposals are reviewed for whether both sides are stated fairly before entering the field. Members on the Pollster tier have their proposals fielded, but a fielded member question carries no axis weight — it is asked and reported without moving anyone's score, so the axes themselves cannot be bought.

The interface exists in 41 languages. Those catalogues are currently machine-translated and labelled as such on every page that uses one. Survey items are held to a stricter standard — machine, translated, reviewed, adjudicated — and only publish at the top of that ladder, because a question's wording is the instrument.