How Breenter works
The aim is to summarise evidence without implying more certainty than the source material supports. Every output can be traced to a dated poll, source or explicit calculation rule.
Poll selection
The headline series includes only polls of UK or Great Britain adults that ask a substantially equivalent referendum voting-intention question: whether respondents would vote to rejoin the EU or remain/stay outside. One result is used per poll.
Support/oppose questions, “right or wrong to leave” questions, and questions about closer cooperation, the single market or the customs union are stored under separate categories. They are not combined with the headline series.
- Records must include original wording, fieldwork end date, publisher and source URL.
- An average-eligible record must include a sample size and both Rejoin and Stay Out values.
- Incomplete or conflicting records are flagged for review and excluded.
- Duplicates are identified using pollster, fieldwork end date, question wording and sample size.
Headline polling average
The average uses eligible polls whose fieldwork ended in the 90 days before the calculation date. Each result is weighted by both sample size and recency.
The square-root transformation and 2,500-person cap reduce the chance that one unusually large poll dominates. The 45-day half-life means a poll’s recency weight halves every 45 days. Rejoin, Stay Out, Don’t Know and other responses are averaged separately from the published values; undecided respondents are not silently removed.
If no poll qualifies, no average is shown. If only one poll qualifies, Breenter labels that fact and does not describe the result as a stable trend.
Rejoin Momentum
Rejoin Momentum is a 0–100 indicator, not a probability and not a forecast. It combines four components:
| Component | Weight | Rule |
|---|---|---|
| Polling average | 60% | The current Rejoin percentage. |
| Six-month trend | 20% | 50 is neutral; each percentage-point change adds or removes 5 points, capped 0–100. If data are insufficient, neutral 50 is used and disclosed. |
| Major-party support | 10% | A source-verified score based on official national-party positions. Until that dataset is configured, the component is held at a neutral 50 and disclosed rather than inferred. |
| Political developments | 10% | Average of the three most recent verified event impact scores. Events are dated and sourced. |
The final calculation is deterministic: 0.60 × polling + 0.20 × trend + 0.10 × parties + 0.10 × developments. Missing trend or party-position evidence defaults to neutral 50. Party and event inputs involve editorial judgement, which is why the result must not be presented as a precise probability.
Sources and update frequency
A Vercel Cron request runs once daily at 05:17 UTC. It acts as a deterministic update agent: configured providers are checked, every record is validated with Zod, duplicates and conflicts are flagged, an import log is stored, and the aggregate is recalculated.
YouGov provides public result tables, often as PDFs, and What UK Thinks provides a useful public archive. Neither has been wired as an unattended live source because a stable, documented polling API or machine-readable feed was not verified. The bundled series contains eleven manually verified, source-linked referendum-voting-intention records. An approved JSON endpoint can be connected without changing the calculation layer. Until then, the daily agent safely imports zero records rather than scraping or inventing data.
Limitations
- The bundled data are verified and useful for historical context, but not a complete polling archive.
- Polls carry sampling and non-sampling error; small changes may be noise.
- Question wording, response options and population coverage can change results.
- UK re-accession would depend on political decisions and an Article 49 accession process, not public opinion alone.
- Momentum does not predict whether or when the UK will rejoin the EU.