Transparency
Methodology
How we turn scattered national polls into one comparable global dashboard — using only verified readings.
Editorial Scope & Focus
This terminal is strictly dedicated to tracking the electoral momentum, structural growth, and vote share of right-wing, conservative, and populist political movements globally. By applying one consistent standard across different electoral systems, we provide a unified data environment for comparing these political coalitions on equal terms.
The Verified Poll Average
Every headline figure on this terminal is a verified poll average: the mean of the last verified voting-intention readings (at most five) collected for that party, or the single latest verified reading where only one is on record. Each reading is a published survey figure with a named pollster, fieldwork date and — where published — sample size; the last readings are listed row by row on each party page.
Collection runs off public opinion-polling tables (primarily Wikipedia's per-election polling pages, one per country). Our scraper reads the party's column, takes the most recent valid figures together with the pollster, fieldwork date and sample size from the same rows, and writes them to the dataset. The overview table and weekly brief show that week's average; the party page lists the underlying recent polls and plots the verified history — every dot is a verified scrape, no interpolated weeks, no generated institutes.
Why We Average (a Little)
Polling firms differ in sampling mode (phone, online, mixed), weighting schemes, and turnout models — so-called house effects that can push individual readings several points above or below the underlying trend. Averaging a handful of recent verified readings damps single-poll volatility: one anomalous survey moves a five-reading mean by only a fraction of its deviation.
What we do not do: invent institute breakdowns. Where the record holds one verified reading, the site shows one — the average then equals that reading, labelled as the latest verified poll. Smoothing beyond the verified points (trend interpolation, generated weekly closes) was removed: the chart shows stored records only.
Sourcing Standards & Inclusion Criteria
Tables enter the pipeline against three structural criteria: genuine national coverage, a publicly documented methodology (sampling mode, sample size, weighting), and a regular publication rhythm. Readings that look like seat counts, calendar-event rows, or malformed cells are rejected by the scraper's validation locks rather than published.
To keep cross-border comparisons honest, readings are standardised to decided voters (excluding undecideds and non-voters) wherever pollsters publish that basis. Every party page names its verified source (e.g. ORC, KONDA, MetroPOLL for Turkey) with figure and fieldwork date — so any reader can check the number against the original table.
Update Cadence
The average is recomputed whenever the scraper verifies a new published reading — typically weekly. Each update is date-stamped on the relevant party page (`lastUpdate`), and the new record is appended to that party's verified history. Quarterly change compares the current average against the previous verified record, so readers can distinguish a fresh move from a standing figure.
(Note: Party logos are shown for identification under fair use and belong to their respective owners.)
Limitations — Read This Before Citing Us
Polls are estimates, not predictions — and an average of a few polls is a smoothed estimate, not a vote forecast. Typical margins of error (±2–3 points), turnout modelling, and late-deciding cohorts all limit precision, especially in multi-party proportional systems. Our collection index is public polling tables: if a table is stale or miscopied upstream, our reading inherits the error until the next verification pass.
A single survey wave can reflect sampling volatility; consecutive verified moves in the same direction constitute trend confirmation. Readers should focus on the trajectory across several records — rather than treating any single print as a projected election-day result. Figures reflect the latest verified voting-intention estimates and are subject to revision by the original publishers.
Corrections Policy
If a pollster revises a figure or we misrecord one, we correct the record promptly, and the correction flows through to every derived display automatically — including the average. To report an error, please include the country, pollster name (e.g. ORC), and publication date via our Contact page so the claim can be verified against the original source.
Figures are compiled from publicly reported national polls. Headline values denote the mean of the last verified readings — not an election forecast.