Saturday, 3 January 2015

The SNP conundrum: a Glasgow experiment

Early this morning, I lay awake in bed thinking about Glasgow constituencies.  I appreciate that this is not a conventional thing to do, but that's how I roll. 

The problem

What was troubling me was the difficulty of working out how the rise of the SNP would work through to the May 2015 general election results, so I started thinking about a methodology for dealing with this.  We know that the SNP have risen sharply in the polls.  However, we have no clear idea of how this will play out.  Opinion polls are snapshots and not predictions. 

Uniform national swing is always dangerous to use when a rise is as sharp as the one that the SNP seem to have got (it is inherently unlikely that a large increase will be uniform) and on this occasion we can be fairly confident that the increase is not uniform, because it seems to have been driven by the referendum on independence.  Since Yes did much better in some areas than others, we can expect the SNP to be increasing support disproportionately in the areas where Yes did best.

I'm not the first to have been looking at this problem.  Stephen Bush had a bash in a Telegraph blog here:

And John Curtice has tried interpreting the most recent ICM Scotland opinion poll here:

Getting to Yes

So I started thinking about how I might approach this.  Put bluntly, I decided to make a series of guesses informed by such data as we have.

The great advantage of lying in bed without access to any data is that you can start from first principles without trying to fudge the assumptions to get the answer that you want.  My starting point is to look at how we got from the 2010 results to the referendum result in Glasgow.  I decided to proceed as follows.  I would allocate all the Conservative votes in 2010 to No and all the SNP votes in 2010 to Yes (I appreciate that this is not strictly true in either case, but I'm looking for a rule of thumb - there are more heroic assumptions coming up so I don't need to stress too much about being too accurate).  Next, I would allocate the 2010 Lib Dem votes 60:40 to No.  As for all the other parties other than Labour, I decided to treat them as dividing 50:50 between Yes and No.

We next need to recognise that many referendum voters did not vote in 2010.  I decided to allocate these 60:40 to Yes.  That leaves the 2010 Labour voters, who I decided to treat as making up the balance on each side. 

There are multiple assumptions in this (for example, that departing and dying voters were exactly replaced by new voters).  But it's a start.

Putting theory into practice

How does that work out?  There are seven general election constituencies in Glasgow, and they totalled as follows:

Lab: 128818 (56.2%)
Lib Dem: 31403 (13.7%)
SNP: 39702 (17.3%)
Con: 17482 (7.6%)
Others: 11818 (5.2%)

Total: 229223

Turnout: 54.9%

The referendum results are here:

As you can see, it was a 53.5% Yes vote on a turnout of 75% (so 40.125% Yes, 34.875% No, 25% Didn't Vote).

These two results are not exactly comparable, because the referendum result for Glasgow covered a slightly larger area than the Westminster seats.  But there are plenty more assumptions flying around, so let's use those figures for now.

Using my assumptions above, that gives us a Yes vote comprised as follows:

SNP: 17.3 x 54.9%
Lib Dems: 13.7 x 0.4 x 54.9%
Others: 5.2 x 0.5 x 54.9%
Non-voters: 0.6 x 20.1%

26% Non-Labour voters

So Labour voters: 14.125% (ie 45.8% of 2010 Labour voters)

And a No vote comprised as follows:

Conservative: 7.6 x 54.9%
Lib Dems: 13.7 x 0.6 x 54.9%
Others: 5.2 x 0.5 x 54.9%
Non-voters: 0.4 x 20.1%

18.2% Non-Labour voters

So Labour voters: 16.675% (ie 54.2% of 2010 Labour voters).

Back to bed, back to reality

Now I would like to invite you to return with me to my bed.  Without any knowledge of the above calculation, I then started thinking about how the referendum might affect voting intention.

I started by assuming that 2010 SNP and Conservative voters would each remain loyal to their cause.  I figured that the SNP would pick up 80% of Yes voters who had not previously voted for them but who had voted in previous elections.  I assumed that the Lib Dems would lose three quarters of their vote in total from 2010, and to the extent that it did not go to the SNP that lost vote would be split evenly between Labour, other parties and Don't Vote.  Labour would keep all their 2010 voters who did not defect to the SNP.

What of those who voted only in the referendum?  I figured that the experience of voting might be catching, for some at least.  I guessed that half would vote again in May 2015, and that this would be split 3% Labour, 6% SNP, 1% to others - I gave the SNP a share disproportionate to the Yes vote to reflect its much greater ability to run a ground game.

So let's feed that through into the above figures:

Labour: 37%
SNP: 47.6%
Lib Dems: 3%
Conservatives: 6.5%
Others: 5.9%

Turnout: 63.7%

What does that mean?

If that was indeed the result, it would be carnage for Labour in Glasgow.  There would be a swing of just under 25% from Labour to the SNP in Glasgow.  The only seat that it would retain if the swing was uniform across Glasgow would be Glasgow North East - and that only just.

Meanwhile, the SNP are favourites in only one seat in Glasgow: Glasgow North, and then only at 8/11.  If my guesses are anywhere near right, the SNP would have a majority of 15% or so in Glasgow North.
Can I extend this to other Scottish constituencies?
Well, yes I can, and it will work well in Labour-held constituencies, since these are essentially two body problems.  When I have the time and inclination, I shall do so.

I don't see this technique working as well in Lib Dem or Conservative held seats though.  There are additional variables at play in such seats.

How confident am I in my guesses?

Not very.  But the errors should mostly cancel themselves out.  I've made limited allowance for the SNP's greater ground resources and I've given Labour nearly as big a chunk of former Lib Dems as the SNP.  I've made no allowance for any increase in UKIP tallies, which is far more likely to damage Labour than the SNP.  Set against that, I'm not at all sure how the likely increase in turnout will play out and I've made no allowance for the Greens to get any uplift.  I may have been too assertive in how many former Labour supporters who voted Yes will switch to the SNP but I've made no allowance for any former Labour supporters to drift to abstentions. 

And of course, we may have more events.

For now, however, I'm using these calculations as a rough and ready way of taking the impact of the referendum into account.  It would not stand any kind of academic analysis.  But it will do for now, in the absence of better information.  If you disagree, at least this post may help you make your own assumptions.

1 comment:

Alastair said...

I decided to back my guesses with some cash. Cumbernauld, Kilsyth & Kirkintilloch East would be an SNP win on this basis, so I've taken the 2/1 on this with Ladbrokes.