John Curtice and prediction

Rather a lot of heat, rather than light, has been generated in the last couple of days by the Sunday Herald picking up on a paper by the statisticians’ psephologist, Professor John Curtice of Strathclyde University published by the Electoral Reform Society. You can find the offending article here – http://www.electoral-reform.org.uk/sites/default/files/files/publication/The-2016-Scottish-Election-Briefing.pdf – though describing it as “offending” is a bit tough on the Prof, who has done little more than run the numbers and pose a few questions.

What seems to me to be the main issue comes from the data of Table 3 in Curtice’s article, which sets out how the allocation of seats from the regional list vote in Lothian would play out. To do this, he calculates the variation in support for parties since the last election in 2011 and applies this over the country as a whole. This allows him to predict results on a constituency by constituency basis and then, with a forecast of regional votes, as well as the allocation of constituency seats, how Regional List seats would be allocated. The outcome of this is a forecast that the SNP will win 70 out of 73 constituency seats, but only two Regional list seats, in the Highlands & Islands. The Lib Dems would win one constituency (probably Shetland) and the Conservatives would win two. Curtice does not divulge where these are, but it seems most likely to be in South Scotland. Thus in the other five regions – Central Scotland, Glasgow, Lothian, Mid Scotland & Fife, and West Scotland, it is likely that the SNP will have won all the constituency seats.

Curtice however focuses on one of the regions where the SNP has won all the constituencies – Lothian (where there are 9 constituency seats), but the same analysis could be done for the other four regions set out above.

The playing out of the allocation of Regional List seats in Lothian demonstrated by Curtice in this table, which shows that on the one hand, the SNP Regional List vote is 41.2%, which is more than the three Unionist parties’ Regional List votes put together. Yet the SNP win no seats, while the Unionist parties win six – three Labour, two Conservative and one Lib Dem. The Greens take the other Regional List seat.

The operation of d’Hondt works against the SNP, since their 9 constituency members means their Regional List vote is divided by ten from Round 1, meaning they go into the first round of allocating Regional List members with a vote of not 41%, but 4.1%. There would have to be ten rounds (and thus ten Regional list members) before the SNP would take a seat.

The section in Curtice’s paper on which there has been all the comment is this “That this situation could arise in a number of regions, given the SNP’s current standing in the polls, has led to speculation that nationalist supporters might be wise on the second ballot to vote tactically for a different party, such as the Greens or the left-wing RISE grouping, both of which also support independence. That way their vote might contribute to the election of another independence supporting MSP rather than apparently being wasted.”

On the basis of this analysis this seems to me to be a reasonable observation, though it most certainly does not justify the Sunday Herald’s headline that “Independence supporters should not cast second vote for SNP at Holyrood election, says study”. I think it is clear that Curtice is saying no such thing. He is pointing to an issue – that if the SNP dominate constituency outcomes as much as the polls (and Curtice’s own forecast) suggest, the operation of the Regional List allocation of seats makes it difficult for the SNP to do well there. Might SNP members do better to use their Regional List vote otherwise than voting for the SNP? What is an SNP voter to do, is basically what, I think, Curtice is enjoining us to think about?

One noticeable feature from the above table is that the party that just loses out on round 7 are the Greens, who on 5.8% are just a fraction behind Labour. To get in front of Labour they need a further divided vote of just 0.3% to have 6.1%. Taking that back to the opening vote (what Curtice calls Stage 1) this would be a vote of 12.2%, or 0.6% more than they are forecast to get. Thus if that proportion of the SNP Regional List vote were to vote Green instead, then instead of Sarah Boyack (3rd on the Labour list) being elected for Labour, Andy Wightman (2nd on the Greens list) would be elected. Seems a decent swap to me.

Thus, Curtice is posing a very a real question. If the SNP are so dominant on the constituency vote, the electoral system we use makes it very hard for them to win more than a handful of Regional List seats. Taking the Lothian example, it means that 41% of the Regional List vote elects precisely no one, and given its relative allocation among the other parties, allows the Labour Party to pick up 3 seats with just under 18% of the vote, but with a small addition to their vote the (independence supporting) Greens could have had 2 seats and the Labour Party would have been reduced to two seats .

 

 

 

Advertisements

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