Poll Analysis: Redfield & Wilton 31st of March – 1st of April 2023

Third in a flurry of polls to drop after Humza Yousaf’s election as First Minister is the second entry in Redfield & Wilton’s new Scotland tracker (link to tables). This is supposed to be a monthly tracker, which would give the most consistent and frequent Scotland-only polling. In theory, that should be something I’m thrilled by, given the best we’d had was the (excellent) pairing of Savanta and the Scotsman, though that didn’t have the set frequency that this does.

I’ll get onto exactly why when we actually turn to the voting figures, but my initial excitement at having a consistent, periodic Scotland tracker has kind of turned to ashes in my mouth. A lot of folk on Twitter were taken aback by just how annoyed I was by (what I consider to be) a glaring deficiency in this poll. It’s easy if you’re not as engaged in this as I am, or if you’re based down south and you’re used to getting basically a new major poll every weekday, to shrug such deficiencies off as “well, it’s only one pollster”. It is only one pollster, but it’s the only one that looks set to do frequent, periodic Scottish polling.

We already get so few polls in Scotland that rather than being able to do something like a rolling average over the month or taking only the most recent per agency, my tracker page operates on an arbitrary “last five polls” basis. We can go months at a time without a poll usually, and in that context a regular tracker will pop up multiple times in the average. If it’s getting stuff wrong, then it skews everything. Even if removed from or adjusted in averaging, as I may very well do if this continues, I’m just one little nerd covering these. Thousands upon thousands of people will see the figures reported uncritically and get a frankly incorrect picture of Scottish politics. That’s just not good, and that’s why I’m so annoyed by this situation.

The previous Redfield & Wilton covered the 2nd – 5th of March. Changes are shown as (vs that poll / vs last election).

Regional Vote

Hoo boy. Right. First things first – as usual, a lot of the movement here is within margins of error, but it’s nonetheless jarring at the moment to see the SNP up and Labour down. At a time when Labour have just had some bests, this is their lowest polled list figure since December. With the Conservatives down too, versus gains for the Lib Dems, Greens and Reform UK, it’s apparently a good day for the smaller parties, in contrast to some weaker polling elsewhere.

This is also quite a rare poll at the moment to show the Lib Dems ahead of the Greens, as the last R&W did too. Although I’m about to get on to my deep reservations about this poll, it’s important to point out this aspect of it is quite reasonable. Given their respective polling spreads at the moment, it shouldn’t be a surprise to see the occasional poll where the Lib Dems squeak ahead. It goes against trend, yes, but not in a way that’s at all hard to believe.

Now we turn to the problem – 13% for the Lib Dems. That is extremely hard to believe. After last month having said a high Lib Dem figure didn’t mean “the poll is bad”, to be honest this month I’m saying the poll is bad. With no disrespect meant to the Lib Dems, they are not on 13% of the vote in Scotland. The 11% last month was remarkably high, but to then find another 2% above that is taking the absolute biscuit. I know some won’t like that assessment, but the point of these pieces is not simply to report polls, but to analyse them. There are solid, evidenced reasons for me to think this is complete nonsense.

Firstly, it’s so much an outlier that it’s not even touching margin of error. The three other post-leadership election polls had the party on 6%, 7% and 7%. The average of all 12 non-R&W polls this year is 7.4%. Although we often report polling error as 95% chance of being within 3%, that’s only true at 50%, as I explain here. At 7.4%, the margin is 1.6%. So everyone else currently thinks that at best likely guess, the Lib Dems could be on 9%. Last poll’s 11% was high but at a real stretch in that context but just about within bounds that could be expected, whereas 13% is in relative terms massively out of kilter – it’s nearly twice as high as everyone else!

Whilst other parties can have similar spreads between different pollsters, the emphasis there is on “spread”. The Greens for example can over a period vary from 8% to 14%, but at each extreme, they have closely corroborating figures. Panelbase have long found lower Green figures, as did Opinium when they polled, now joined by Survation. And when Savanta was finding 14% for them, there’d been Survations with 12% and Ipsos with 13%.

Indeed, the Greens may just be particularly hard to get right because poll weighting is generally done via Westminster and Holyrood Constituency vote, rather than list vote, and that leads to the wide spread. This Lib Dem figure is unique in just how much of an outlier it is, that no other pollster is coming close, and FPTP vote based weightings work for them. You might well be inclined to say “well, maybe it’s good to avoid herding”, but that’s where my next points come in.

Secondly, Scotland is not England. That seems like an obvious statement, but at the risk of sounding like a chippy Scot, it’s one some commentators based down south often forget. In England, there are plenty of Conservative held seats where even polling as they are now, Labour are not realistically in with a chance, but the Lib Dems are. That could believably help boost their vote, as people scunnered with the Conservatives turn to the best option to unseat them locally. In Scotland, that just doesn’t apply.

There’s only one constituency in each parliament the party don’t already hold that they have any prospect of picking up. They aren’t in any way the obvious alternative for voters who’ve been put off the SNP. Although this is the regional vote rather than constituency, it’s still the case that the Scottish Lib Dems have experienced a dramatic withering far beyond their English counterparts. In last year’s local elections, their gains were primarily in Edinburgh, Highland and North East Fife. They were pretty stagnant in other historic strongholds like Aberdeenshire and Argyll and Bute, and continued to slip into near oblivion in Glasgow and most of its surrounds. That pattern of support easily explains 7-9% shares, allowing for decent growth since 2021. It just isn’t enough to give 13%!

Thirdly, and building on that point, if you look at the swing versus 2021 this would make the Lib Dems (+8) the biggest gainers, ahead of Labour (+6). The reality of Scottish politics right now is that with both the SNP and Conservatives weakened, the obvious and natural beneficiaries have been Labour. Across the core, urban, Central Belt seats that make up more than half the Holyrood total, Labour are the obvious contenders. People in the likes of Glasgow, North Ayrshire and West Lothian are not plumping for the Lib Dems. In a sense, Lib Dem figures here are something of a robbery from Labour!

Fourthly, 13% is not only the best Lib Dem figure of the term, it matches their best figure of the previous term which came at the height of Brexit drama in 2019. It’s actually better than any Holyrood list vote share they got before their collapse! Look, I’m turning into a broken record here, but that’s just not happening right now. That is not the political reality of Scotland. Looking across polling in general, things make sense for the most part, but this just doesn’t.

The SNP down? Their leadership contest was bruising and they’ve been in government for 16 years. The Greens likewise? They’ve gotten a lot of flak lately, and are part of government with the SNP. Conservatives down but not as badly as last year? Ongoing damage from their complete chaos, but that’s calmed down enough to allow some recovery. Labour up? They’re now the obvious “government in waiting” option in both parliaments, and with historic status as a dominant party. Lib Dems to break their Holyrood record from when they were at their most popular? Give me a break.

This has been a far longer screed than I usually write about a single poll, never mind one party’s share in one question. However, it’s so far out of the realms of possibility I feel it has to be robustly addressed. Frankly, if the next poll doesn’t show something more in line with other pollsters, I’m just going to remove Redfield & Wilton from my coverage entirely – if they’re getting this so wrong, what’s the knock-on elsewhere in their polling? In my view this isn’t simply an outlier, it’s suggesting a serious flaw in methodology.

Constituency Vote

The constituency looks very slightly more normal in that the SNP are down, even if that’s also with Labour dropping a point. Those plus Conservative losses are again balanced by Lib Dem and Green upticks. As with the list vote this is the best Lib Dem figure of the term and although it’s not quite as unbelievably high, it does nonetheless raise my eyebrow given the context. 

Seat Projection

Projecting that into seats might give us something like this:

Please see this page for how projections work and important caveats.

As this poll has Labour below where the other two recent polls have and the Lib Dems significantly higher, the SNP have a much chunkier lead over Labour, and the Lib Dems obviously have their best seat projection too. Not to flog a dead horse, but the Lib Dems are not getting seats in Glasgow and Central right now, that’s just silly. If we take 4% directly from the Lib Dems to Labour, which I think would be a more realistic poll, that swaps 7 seats between the two parties.

Lib Dem weirdness aside however, this is about on a par with the other pollsters, who also projected to a combined majority for the Pro-Union parties. Redfield & Wilton had been a bit of an outlier in finding that during the contest, with the first ordinary poll of the term to do so, but now on this front at least they’re more in line with what we’re seeing elsewhere.

One final point worth adding on this poll is of course that it doesn’t reflect any possible effects from the dramatic scenes on Wednesday where Peter Murrell, former CEO of the SNP and Nicola Sturgeon’s husband, was arrested then released without charge as part of an ongoing police investigation. Beyond the fact this relates to finances there is no other information available at the moment, and obviously I shan’t be speculating on that. However, it’s the kind of event that could further heap political trouble on the SNP that the next set of polls could pick up.

Were it not for, once again, the best Lib Dem result since the election (though at least jointly in this case), this would look even more normal still – SNP down a fair chunk, Labour up, with the SNP on their worst polled share for this vote since 2019. It does have that high Lib Dem share though and I do view it with some dubiety given my big grumble above.

With just 5% between the SNP and Labour, that’s the narrowest gap we’ve seen between the two for a very long time. They’d almost certainly be running one another pretty close in terms of total seat numbers at this point, given how the SNP’s vote is spread and Labour’s quite concentrated.

Over on the constitution, some very slight, margin of error shifts bring this a point up from what had been a replay of 2014 in the last poll. As noted with the two polls preceding it (from Panelbase and Savanta), this means Independence support is higher than support for the parties backing it. Whether that becomes an established trend or if it represents a level to which those parties soon bounce back to, we’ll need a few months to determine.


As ever, the last little bit of analysis concerns those hypothetical and more proportional voting systems that BBS likes to play about with. The use of pure FPTP at Westminster is an affront to democracy, and though Holyrood fares far better, AMS is still deeply imperfect. The examples here simply transpose the poll findings onto more proportional voting systems – the reality is that different systems would of course result in different voter behaviour.

When AMS as it actually operates is giving majorities for the Pro-Union bloc, these wee sections become a bit more pro-forma – it’s more proportional, so obviously that further erodes the SNP’s FPTP advantages.

Aaaand since this is yet more proportional again, it really boosts the Pro-Union bloc, to a chunky 74 vs 55 advantage.

Scandinavian Style Westminster

Given where the Westminster vote is, even under FPTP we likely would be seeing very close to a tie between the SNP and Labour in seats. Where this model would differ from that reality then is more seats for the Conservatives.

If you find this or other Ballot Box Scotland output useful and/or interesting, and you can afford to do so, please consider donating to support my work. I love doing this, but it’s a one-man project and takes a lot of time and effort. All donations, no matter how small, are greatly appreciated and extremely helpful.
(About Donations)