Panelbase, 1st-5th May 2020

Although Coronavirus is taking up most of the country’s time and attention, some amount of politics as normal is rumbling on in the background. Another Panelbase poll dropped a full week ago, but to my frustration (and also, I think, in breach of British Polling Council Rules…) they still have not published the full tables. I’ve got fed up of waiting for them and felt the lack of an analysis post was a glaring gap here on BBS, so I finally decided just to go ahead with those and come back and update later if necessary.

Well, turns out it was necessary! For the first time since starting BBS, this was a case where reported figures had some inaccuracies. It was only the Green and Lib Dem Holyrood Regional Vote shares that were wrong. They were 7% for the Greens, versus the 5% reported, and 6% for the Liberal Democrats, versus the 7% reported. This meant the Greens were on 5 seats, not 3, and the Lib Dems on 5, not 7, in the seat projection.

Display format;

Holyrood Voting Intention and Projection (Tracker)


  • SNP – 48% (nc / +6)
  • Conservative – 22% (-4 / -1)
  • Labour – 15% (+2 / -4)
  • Green – 7% (+1 / nc)
  • Liberal Democrat – 6% (nc / +1)

This one keeps to the current norm of very high SNP figures, with them holding steady at the same heights as they were in the March poll. Their lead over the second party has widened though, as the Conservatives have dropped a relatively weighty 4% here. There are minor gains for the Greens and, rarely, for Labour, who are still down on their 2016 result but up somewhat from their recent rock-bottom. That spares them another “lowest polling figure” note here.


  • SNP – 53% (+2 / +6)
  • Conservative – 23% (-3 / +1)
  • Labour – 15% (+1 / -8)
  • Liberal Democrat – 5% (-1 / -3)
  • Green – 3% (nc / +2)

As ever, the expected similar pattern for the Constituency vote. That uptick for the SNP gives them their second best figure for this ballot since the 2016 election, with only that recent YouGov placing them higher. There’s another relatively sizeable dip for the Conservatives here, a small bounce for Labour, and for some “worst result” variety, the Lib Dems are on their joint-worst figure for this vote since the last election. They polled 5% once in each of 2016 and 2017.

If we do the usual projection into seats, it might look like;

  • SNP – 73 (+3 / +10)
  • Conservative – 29 (-5 / -2)
  • Labour – 17 (nc / -7)
  • Green – 5 (+2 / -1)
  • Liberal Democrat – 5 (nc / nc)

Unsurprisingly, given their polling figures, this is another SNP majority, and is in fact their joint highest seat count under my calculator since 2016. The first time was shortly after the 2016 election itself. Their constituency dominance is projected to hand them all but 5 of the 73 constituencies (usual caveats apply) which would be remarkable. They gain seats at almost everyone else’s expense, with only the Lib Dems static despite projected constituency losses.

Westminster Voting Intention (Tracker)

  • SNP – 50% (+2 / +5)
  • Conservative – 26% (-1 / +1)
  • Labour – 17% (+1 / -2)
  • Liberal Democrat – 5% (nc / -5)
  • Green – 2% (-1 / +1)

This is only the third Scottish Westminster poll since the December election, and it has the SNP on half the vote, which would match their 2015 peak. As there’s been a bit of a realignment of voting patterns across the country since then, I wouldn’t expect that to lead to a repeat of their 56 of 59 seats from that election, with the Conservatives likely to hold onto a small handful.

Constitutional (Independence Tracker)

Usual Independence Question;

  • Yes – 46% (nc)
  • No – 46% (-1)
  • Don’t Know – 7% (nc)

Excluding Don’t Knows;

  • No – 50% (-1 / -5)
  • Yes – 50% (+1 / +5)

True to the current Scottish form, this is an absolute dead-heat, with even the tables showing precisely the same number of respondents leaning Yes as No. Whatever your position is on this particular question, I’m sure we can all agree that would be a result we’d really want to avoid in any future referendum!

I haven’t done the usual for fun projection here as by this point they just repeat themselves – if you want to see what an even split looks like with simple Universal Swing, there’s one in the post for the November Panelbase poll.


I’ve been adding the output from my alternative Scandinavian Style Model to these posts for a good while now, and with another bit of “hypothetical” output added, plus a third one under development now complete, I’ve decided to pop them into their own little section at the end. That should help make even clearer that these don’t represent our current electoral systems, but are instead interesting little alternative scenarios provided for comparison.

Anyway, that explanation out the way, we can look at the output from that more proportional Scandinavian model for the Scottish Parliament (changes vs AMS projection);

  • SNP – 63 (-10)
  • Conservative – 29 (nc)
  • Labour – 20 (+3)
  • Green – 9 (+4)
  • Liberal Democrat – 8 (+3)

As this is more proportional than AMS, the SNP don’t quite make a majority in this model. The ten seats they are down versus AMS are distributed pretty evenly between the Greens, Lib Dems and Labour, reminding us that particularly in the case of smaller parties the entirely regionalised nature of proportionality under AMS can have a negative effect. Note: As this was corrected in mid-June, be aware that this is effectively the first version of this calculator which switched from D’Hondt to Sainte-Laguë allocation for directly elected seats. The reasons for that change are touched upon in the June Panelbase Poll post.

The other wee bit of fun is to compare a range of different tweaks and reforms to the Scottish Parliament electoral system, basically all of which are more proportional than the current setup but most aren’t as radical as the total reform above.

The third and final hypothetical is to sort of do something people always ask me for – a seats projection from a Westminster poll! Just, y’know, not one for First Past the Post, as amongst other reasons (outlined here), pure FPTP is absolutely murder to try and project. Instead, this applies a Scandinavian Style model similar to the Holyrood version to Westminster. As with its Scottish Parliament counterpart, the intent here is to show what seats might look like if they actually reflected how people voted, rather than being somewhat distorted by boundaries and disproportionality. This would come out at;

  • SNP – 30
  • Conservative – 16
  • Labour – 10
  • Liberal Democrat – 3

The key thing here compared to the current situation and likely outcome of an FPTP election with those vote shares is that Labour are appropriately ahead of the Lib Dems in terms of seats, rather than trailing or equal. Although they are marginally short of a majority of votes, the SNP do win a majority of seats here, as the 2% for the Greens isn’t enough for the latter to cross the electoral threshold in this model, plus there’s another 2% going spare for “other” parties.