Skip to content

The Boundaries Are A-Changin: Another Perspective

09/09/2010

Back in January, I analysed the boundary changes for the Scottish Parliament, and came up with a broad idea of what I thought the Chamber would have looked like, had the 2007 Election been fought on those constituencies. Now, Professor David Denver has completed his research into the matter.

The Professor has used the results of the 2007 Local Elections as his building blocks. This has two obvious problems. Firstly, as Denver himself accepts, the change to STV has left Council wards too large for all of them to be placed wholly within one constituency, as used to be the case in reviews. As he says:

In the very many cases where wards are split between two or more old constituencies and/or between two or more new constituencies estimating party support is difficult without local knowledge.

That’s issue one. Issue two is broader: we vote differently in local elections and Holyrood elections. We even vote differently between the Constituency and Regional component of the Holyrood vote. That’s because we’re making a different decision (who runs the Council versus who runs the Scottish Government) with different options. So the two likely choices for First Minister in 2007 were the incumbent – Jack McConnell – and the successful challenger – Alex Salmond. The choice was between a Labour-led administration, or one led by the SNP. But in places like Edinburgh and Aberdeen, for example, it was a choice between a Labour-led Council and a Liberal Democrat one. South Ayrshire was, in effect, a choice between Labour and the Tories. So the factors that might drive our voting habits were different on each ballot paper.

Moreover, the choice of candidates would obviously be different: the people who voted for the successful Green Council candidates couldn’t, for example, have voted Green on their Constituency ballot, except in Glasgow Kelvin. Parties and independents who were available on one ballot paper, wouldn’t have been available on all three. And of course, that’s before one factors in personal votes.

So there’s a lot of guesswork involved, so much so that Denver had to fall back on historic polling data – not ideal either considering that the figures would include the SNP’s 2003 nadir as a back up to its 2007 zenith. But even without that, the facts remain that he had to guess at the ultra-local vote breakdown. He had to bridge the gap between local factors and Scotland-wide ones. He had to make votes for one set of candidates fit not one but two entirely different sets. And that’s before you factor in the fact that tactical and personal votes interfere with calculations.

The worst part is that it needn’t have been this way: one good thing to come out of the debacle that was the 2007 election process was that the Scotland Office were able to give us a ballot box by ballot box breakdown of the election and its results. This gave us, with the exception of postal votes, the hyper-local data that Denver needed, and while it can’t do much about tactical and personal voting, it uses data from an election to the same body, so that clears a big issue up, and this is the dataset I used in making my calculations.

Of course, it’s not perfect: where Council wards have been split it’s entirely possible that polling stations close to a boundary have been mis-allocated, and I had to have some way of dealing with the postal votes (this being the one occasion where the local election ballot might have helped by at least giving us an indication of turnout), but as the majority of votes were cast in person, it makes sense to go with the best source of data – the Scotland Office.

But methodology is subjective: the postal vote split meant that I was a little bit suspicious of the low turnout I calculated in Glasgow Kelvin (offset by a surprisingly high turnout in Glasgow Maryhill & Springburn), which Denver appears to have sidestepped. However, there are a number of key differences.

I get Nicola Sturgeon winning Glasgow Southside by 183 votes. Denver gets Labour winning by 27. My SNP majority of 30 in Edinburgh Eastern becomes a 545-vote Labour lead to Denver, but my 606-vote Labour majority in Linlithgow is more than double Denver’s projection. I get Bruce Crawford having a 49-vote lead in Stirling; Denver has him 389 votes behind but he sees the SNP winning Aberdeen Central by 349 votes whereas I look at that seat and see Labour winning by 32. Finally, my 204-vote Labour lead in Dumfriesshire, should, in his eyes, by a 649-vote Tory lead.

Of course, there are other differences – none that affect the outcome of the hypothetical version of the 2007, but make things trickier when trying to extrapolate polls: I see Skye, Lohaber & Badenoch as far more vulnerable than Denver, for example. But the actual notional figures should have only one effect to the outcome: Denver and I both agree that the SNP would have five seats in Glasgow, but I believe one of them should be a constituency (Southside), and that allows the Greens a Regional seat. Denver believes that Labour should win that constituency, pushing all of the SNP’s representation in Glasgow on to the List, at the expense of the Greens. So where I see Labour losing two seats and the Greens breaking even, Denver should only see Labour losing one, and the Greens also losing one.

But he doesn’t.

And that’s because he’s miscalculated on his own figures for South Scotland. He projects that Labour should have 81,326 votes and three constituencies, while the Tories should have 62,972 and four constituencies. The picture is completed by the SNP’s 80,668 votes and two constituencies, and the LibDems’ 28,001 with no constituencies.

Denver says that should lead to 3 SNP Regional seats, 2 LibDem, one Labour and one Tory. But let’s walk through the D’Hondt calculations:

Lab: 81326 / (3 + 1) = 20331.5
SNP: 80668 / (2 + 1) = 26889.333
Con: 62972 / (4 + 1) = 12594.4
LibDems: 28001 / (0 + 1) = 28001

Seat 1: LibDem

SNP: 80668 / (2 + 1) = 26889.333
LD: 28001 / (1 + 1) = 14000.5

Seat 2: SNP

Lab: 81326 / (3 + 1) = 20331.5
SNP: 80668 / (3 + 1) = 20167

Seat 3: Labour

Lab: 81326 / (4 + 1) = 16265.2
SNP: 80668 / (3 + 1) = 20167

Seat 4: SNP

Lab: 81326 / (4 + 1) = 16265.2
SNP: 80668 / (4 + 1) = 16133.6

Seat 5: Labour

Lab: 81326 / (5 + 1) = 13554.333
SNP: 80668 / (4 + 1) = 16133.6

Seat 6: SNP

SNP: 80668 / (5 + 1) = 13444.667
LD: 28001 / (1 + 1) = 14000.5

Seat 7: LibDem

So that’s 3 SNP, 2 LibDems, but 2 Labour and no Tories – on Denver’s own figures.

And that means that the headlines about the Tories reaching that magic figure of 20 are bogus: the Tories are, on Denver’s own notional votes, only on 19 seats.

So he’s used a needlessly complex process of deriving the notional figures (one that might make working postal votes marginally easier, but ignores the ballot box-level data that would give him the local knowledge he needs), and miscalculates the outcome he gets from them.

The problem is, the no one in the press bothered to run the calculations themselves. Instead, they just blindly followed his report. I just hope that someone else is putting something together, and that they check and re-check everything as they go along.

Advertisements

From → Politics

3 Comments
  1. Great post Will. I’ve only just spotted this same error (a full two months after you did, but we can gloss over that)

    Hope you’re fact-checking anyone who mistakenly uses the incorrect figures.

Trackbacks & Pingbacks

  1. The error in the 2007 ‘notional’ Holyrood result « Better Nation
  2. #SP11 Region Watch – North-East Scotland « Better Nation

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: