In December 2020, six years ago, I noted YouGov polls reporting small Yes leads of 1 and 5% while all around were reporting on average a 7.5% lead. https://talkingupscotlandtwo.com/2020/12/03/dont-fear-the-next-yougov-poll-on-indepedence/

I wrote:

It’s all down to their sampling bias.

and suggested this might be a factor in preferring that sampling bias over those six years:

YouGov was founded in the UK in May 2000 by Stephan Shakespeare and future MP Nadhim Zahawi (above) at the time both active in the British Conservative Party. In 2001 they engaged BBC political analyst Peter Kellner, who became chairman, and then from 2007–2016, President.

https://en.wikipedia.org/wiki/YouGov#:~:text=YouGov%20was%20founded%20in%20the,from%202007%E2%80%932016%2C%20President

Since then I have returned several times to look at their weighting/skewing of their samples in line with the 2014 referendum result and in 2024 and wrote How YouGov make sure they get a NO lead and stronger opposition to the SNP: https://talkingupscotlandtwo.com/2024/05/21/how-yougov-make-sure-they-get-a-no-lead-and-stronger-opposition-to-the-snp/

In March 2025, I wrote YouGov find No voters more readily than other pollsters to cast doubt on their honesty: https://talkingupscotlandtwo.com/2025/03/27/yougov-find-no-voters-more-readily-than-other-pollsters-to-cast-doubt-on-their-honesty/

In July 2025, I wrote Want a No lead? Ask Survation. Want a crazy big No lead? Ask YouGov. Want a Yes lead? Don’t fiddle with the sample: https://talkingupscotlandtwo.com/2025/07/05/want-a-no-lead-ask-survation-want-a-crazy-big-no-lead-ask-yougov-want-a-yes-lead-dont-fiddle-with-the-sample/

In November 2025, I wrote Only YouGov can dig out a big NO prediction regardless of surrounding events and only Survation can never ever find a YES lead: https://talkingupscotlandtwo.com/2025/11/13/only-yougov-can-dig-out-a-big-no-prediction-regardless-of-surrounding-events-and-only-survation-can-never-ever-find-a-yes-lead/

Three more posts in 2026, found the same phenomenon, adjusting their samples to mirror the 2014, 55:45 ratio and predispose No leads, 12 years later and half a million voters passed away.

Then in September 2026, YouGov’s naked bias still using the thousands who have passed since the 2014 to make you think No could still win: https://talkingupscotlandtwo.com/2026/09/11/yougovs-naked-bias-still-using-the-thousands-who-have-passed-since-the-2014-to-make-you-think-no-could-still-win/

In the above, they had weighed their sample to produce a ratio, based on memory of voting back then, of 38/48 (385/478) Yes/No, a 10 % lead for No and manage to squeeze out 5% lead for No, amid 6 other pollsters finding Yes leads of between 7 and 7%

Then yesterday, The Times had Why polls on Scottish independence are all over the place.

Key points from the Times piece:

• In late spring (around the Holyrood elections), Norstat (for The Sunday Times) found 52% Yes and 43% No (Don’t Knows excluded), while YouGov showed the opposite trend (Yes behind by several points).
• The Times analysed 12 years of YouGov polls (which weight for the 2014 referendum) versus Ipsos (which does not). Ipsos consistently recorded higher Yes support, with the gap ranging from about 2 points (May 2026) to 13 points (October 2023).
• The most recent Sunday Times/Norstat survey (late October 2026) showed Yes with a modest 3-point lead (around 48% Yes / 45% No including undecideds in related reporting).
• Of roughly 30 bona fide independence polls in 2026, 19 had Yes ahead, 10 had No ahead, and one was tied; half of the No leads came from YouGov.

The article attributes much of the divergence to methodological differences, especially how pollsters handle weighting for the 2014 referendum and younger voters. https://www.thetimes.com/uk/scotland/article/why-polls-on-scottish-independence-are-all-over-the-place-kn5t28gn2