Methodology & Sources

What this site argues

That the only Holyrood majority which has ever produced an independence referendum was the 2011 SNP majority — and current polling for the 2026 election does not suggest a comparable mandate. The site shows the data plainly and lets readers judge the fit.

The 2011 baseline

All comparisons are anchored to the 2011 SP election: SNP won 69 seats (4 above the 65-seat majority threshold), 45.4% of the constituency vote, and 44.0% of the regional list vote. This single-party majority was treated by the UK Government as a democratic mandate for the 2014 referendum, in which Scotland voted 55.3% No on a 84.6% turnout.

No subsequent SP election has matched 2011. SNP slipped to 63 seats in 2016 (45.5% turnout) and 64 seats in 2021 (63.5% turnout). Both governments required Green confidence-and-supply to legislate.

Where the 2026 numbers come from

• Projected SNP seats and vote share are pulled live from the same projection model that powers seatwatch.scot: recency-weighted poll average (14-day half-life), house-effect correction, proportional swing onto 2026 boundaries, D'Hondt allocation. Methodology detail at seatwatch.scot/methodology.
• Per-constituency 2011 SNP shares for the change map come from the 2011 actual results, transposed onto current 2026 boundaries via postcode-weighted aggregation.
• Historic SP results (2011, 2016, 2021) for the over-time bars are the official Electoral Commission figures.
• 2014 referendum result is the official Electoral Commission total: 44.7% Yes, 55.3% No.

The "if a referendum were held today" panel

Per-constituency Yes/No predictions come from a logit-additive swing model. For each SPC, the calibration step measures the swing between the SNP's constituency-vote share at a chosen Scottish Parliament election (2016 by default) and the 2014 referendum's Yes share, both transposed onto current 2026 boundaries via postcode-weighted aggregation. That swing is then applied additively in logit space to the SNP's projected 2026 share to produce the Yes prediction:

swing_logit_SPC = logit(Yes_2014_SPC) − logit(SNP_calib_SPC)
Yes_pred_SPC    = expit(logit(SNP_target_SPC) + swing_logit_SPC)
No_pred_SPC     = 1 − Yes_pred_SPC

Logit space keeps predictions strictly in (0%, 100%) and treats a 5 pp shift at low support as a bigger multiplicative move than at 50% — the right behaviour for a bounded binary outcome. The national aggregate is turnout-weighted by 2014 referendum per-SPC turnout; SPCs missing from any of the three input runs (the target SP projection, the calibration SP run, or the calibration indyref run) are silently dropped from coverage.

This is a structural model, not a poll. It assumes the relationship between SNP constituency vote and Yes vote that held in 2014 still holds in 2026 — which it need not. Read the prediction as "what 2014's coalitions would say about the current Holyrood projection", not as a forecast of how Scotland would actually vote in a referendum held tomorrow.

What this site is not

It is not a Unionist campaign site. It is not affiliated with any party, candidate, or campaign. It commissions no polling. It draws on the same independent polling sources used by the BBC, Wikipedia, and the academic SP forecasting community.

The argument is narrow: that calling a second independence referendum in this Parliament would not have the democratic backing the 2011 majority provided for the first. Whether independence is desirable, whether Scotland has a right to a referendum on different grounds, and what the long-run polling trend looks like, are separate questions this site does not address.