10 May 2011

How the SNP "broke" AMS...

Since Thursday's results in the 2011 Holyrood election, there has been much talk about the electoral system and how the SNP "broke" the additional member system, foiling its attempts to secure some measure of proportionality between votes cast and parliamentarians elected. Nationally, the SNP attracted 45.4% of constituency ballots cast and 44% of regional votes, but managed to secure 69 out of 129 Holyrood seats: over 53% of them. Taking the level of support at its highest, a bluntly proportional distribution of seats would have seen 58 SNP MSP elected. Yet AMS afforded the Nationalists some 11 more. As a kindly intercessor recently reminded me, in a leaner Holyrood of 73 MSPs, limited to its first-past-the-post constituencies, on last Thursday's vote the Nationalists would hold 53 seats or just over 72% of the whole chamber.

There are a number of threads to this - a raft of constituency wins implicated above all. However, I thought it might be instructive to provide a worked example of how the "top up" of additional members from a regional list signally failed to secure proportional representation. Handily, the North East of Scotland region furnishes a stunner. 

To the Northeast! It is not, I admit, territory I'm terribly familiar with and have only jaunted up that way once in my short life. Ten constituencies make up the full region, running across Dundee, Angus, Aberdeen and Aberdeenshire to the Buchan coast. Post 2007 boundary changes added a constituency and rejigged the edges of others. This is Nationalist territory par excellence, the SNP holding eight constituencies on the night and making off with two others, defeating Mike Rumbles and snapping up the seat Nicol Stephen relinquished for the red benches of the Lords. Labour's Lewis MacDonald similarly failed to regain Aberdeen Central, notionally an SNP seat after boundary changes. In total, therefore, the SNP took all of the North East's ten seats. But soft! On what percentage of the vote? The results are pretty astonishing. Joe FitzPatrick secured 57.6% in Dundee City West, Shona Robison polled at 64.2% in Dundee City East while Stewart Stevenson  secured 67.2% of votes cast in the Banffshire and the Buchan Coast constituency. Undoubtedly impressive, but nevertheless, even where the winning Nationalist candidate was supported by a crashing majority of those voting, just under a third of voters supported somebody else.

One rationale for the regional vote is to afford these disappointed electors some compensation, "topping up" first past the post results in the constituencies and securing a measure of proportionality in the final parliament. Yet on Thursday night, despite totally dominating the north eastern constituencies and securing ten MSPs, the Nationalists also managed to acquire another on the regional list. How did it happen? Remember your d'Hondt calculations. You have seven North Eastern regional seats to distribute. In the first instance, you work out your dividers.  These are calculated by taking the number of MSPs a party has already seen elected - and then adding 1 to this total. In the North East in 2011, this meant that the SNP started the regional list allocation with a divider of 11 (their 10 constituency MSPs +1) while every other party's total regional votes were divided only by 1. This also explains why the regional list results are only declared once all constituencies in the region have been allocated.

North East of Scotland
Total votes cast
SNP: 140,749 
Labour: 43,893
Tory: 37,681
Liberal: 18,178
Green: 10,407

The SNP took 52.7% of the North Eastern list vote, while their nearest rivals, Labour, polled only 16.4%.

Round 1
SNP: 140,749/11 = 12,795.4
Labour: 43,893/1 = 43,893
Tory: 37,681/1 = 37,681
Liberal: 18,178/1 = 18,178
Green: 10,407/1 = 10,407

Labour MSP elected

Round 2
SNP: 140,749/11 = 12,795.4
Labour: 43,893/2 = 21,946.5
Tory: 37,681/1 = 37,681
Liberal: 18,178/1 = 18,178
Green: 10,407/1 = 10,407

Tory MSP elected

Round 3
SNP: 140,749/11 = 12,795.4
Labour: 43,893/2 = 21,946.5
Tory: 37,681/2 = 18,840.5
Liberal: 18,178/1 = 18,178
Green: 10,407/1 = 10,407

Labour MSP elected

Round 4

SNP: 140,749/11 = 12,795.4
Labour: 43,893/3 = 14,631
Tory: 37,681/2 = 18,840.5
Liberal: 18,178/1 = 18,178
Green: 10,407/1 = 10,407

Tory MSP elected

Round 5

SNP: 140,749/11 = 12,795.4
Labour: 43,893/3 = 14,631
Tory: 37,681/3 = 12,560.33
Liberal: 18,178/1 = 18,178
Green: 10,407/1 = 10,407

Liberal MSP elected

Round 6

SNP: 140,749/11 = 12,795.4
Labour: 43,893/3 = 14,631
Tory: 37,681/3 = 12,560.33
Liberal: 18,178/2 = 9,089
Green: 10,407/1 = 10,407

Labour MSP elected

Round 7
SNP: 140,749/11 = 12,795.4
Labour: 43,893/4 = 10,973.25
Tory: 37,681/3 = 12,560.33
Liberal: 18,178/2 = 9,089
Green: 10,407/1 = 10,407

SNP MSP elected

In fairness, therefore, the Nationalists' regional gain was in the seventh round in a seven-seater region, but such was the bigness of the SNP's regional vote - and the comparative smallness of their competitors' share - that even the whacking great divider of 11 couldn't trim back the SNP lead far enough to give the Greens the opportunity of a north eastern seat.

There are other lessons elsewhere. Being a Nationalist in Glasgow has not always been a particularly satisfying experience. Well-used to massive Labour dominance in the constituencies, since 1999 the regional list system (insofar as it was able to do so) performed its intended function, affording SNP-minded Glaswegians the chance to return sympathetic members of parliament. Before the election, a number of folk pondered the possibility of tactical voting on the list, suggesting that Glasgow Labour voters could comfortably engage in second vote manoeuvres. We thoroughly misplaced our prescience on that one. Our misplaced confidence was premised on the assumption that the (almost) clean sweep for Labour we've been used to would continue, ad nauseum, ad infinitum, pereat mundus. If, as happened, the constituencies are wrestled from Labour control, Labour tactical voting on the list becomes an awfully risky endeavour for the Labour supporter. Similarly, I dare say there were SNP folk on the list whose hopes were high of being returned - but who were disappointed by their colleagues' triumphs in the constituency ballot.

Perhaps the most melancholy figure of the night was Shirley-Anne Somerville. The former Edinburgh list MSP was defeated by Labour's Malcolm Chisholm in Edinburgh North and Leith, but her fellow candidates' victories in the rest of Edinburgh ratcheted up the d'Hondt divider on the region to nine in the first round, reducing their 110,953 regional votes to only 12,328, well behind the Liberal Democrats on 15,588, who nonetheless also failed to secure a seat on the Lothian list. As a result, Somerville was defeated while everywhere else, her party was winning. No doubt a decidedly depressing existential pinch to be in, that.  Since we're talking about regional list allocations, and further to my earlier post on the results from my own constituency of Glasgow Southside, it is worth following the list allocation through its various rounds - if only to enjoy a proper understanding of just how far George Galloway fell short of his (nonchalantly) coveted new seat...

Glasgow region 

Total votes cast
SNP: 83,109 (39.8%)
Labour: 73,031 (35%)
Tory: 12,749 (6.1%)
Green: 12,454 (6.0%)
George Galloway: 6,972 (3.3%)
Liberal Democrat: 5,312 (2.5%)

Round 1
SNP: 83,109/6 = 13,851.5
Labour: 73,031/5 =14,606.2
Tory: 12,749/1 =  12,749
Green: 12,454/1 = 12,454
George Galloway: 6,972/1 = 6,972
Liberal Democrat: 5,312/1 = 5,312 

Labour MSP elected

Round 2
SNP: 83,109/6 = 13,851.5
Labour: 73,031/6 =12.171.8
Tory: 12,749/1 =  12,749
Green: 12,454/1 = 12,454
George Galloway: 6,972/1 = 6,972
Liberal Democrat: 5,312/1 = 5,312

SNP MSP elected

Round 3
SNP: 83,109/7 = 11,872.7
Labour: 73,031/6 =12.171.8
Tory: 12,749/1=  12,749
Green: 12,454/1 = 12,454
George Galloway: 6,972/1 = 6,972
Liberal Democrat: 5,312/1 = 5,312

Tory MSP elected

Round 4
SNP: 83,109/7 = 11,872.7
Labour: 73,031/6 =12.171.8
Tory: 12,749/2=  6,375.5
Green: 12,454/1 = 12,454
George Galloway: 6,972/1 = 6,972
Liberal Democrat: 5,312/1 = 5,312

Green MSP elected

Round 5
SNP: 83,109/7 = 11,872.7
Labour: 73,031/6 =12.171.8
Tory: 12,749/2=  6,375.5
Green: 12,454/2 = 6,227
George Galloway: 6,972/1 = 6,972
Liberal Democrat: 5,312/1 = 5,312

Labour MSP elected

Round 6
SNP: 83,109/7 = 11,872.7
Labour: 73,031/7 =10,433
Tory: 12,749/2=  6,375.5
Green: 12,454/2 = 6,227
George Galloway: 6,972/1 = 6,972
Liberal Democrat: 5,312/1 = 5,312 

SNP MSP elected

Round 7 
SNP: 83,109/8 = 10,388.6
Labour: 73,031/7 =10,433
Tory: 12,749/2=  6,375.5
Green: 12,454/2 = 6,227
George Galloway: 6,972/1 = 6,972
Liberal Democrat: 5,312/1 = 5,312

Labour MSP elected

Glasgow Totals:  7 SNP MSPs (5 constituency, 2 list); 7 Labour MSPs (4 constituency, 3 list), 1 Tory (list), 1 Green (list).

As you'll notice, the Liberals had a snowball's chance in hell of retaining their Glasgow seat of 2007. If the region had more than seven MSPs to allocate, the 8th seat would have gone to the Nationalists, the 9th to Labour - and only on the 10th would George Galloway been elected. I wondered what would have happened if Bill Kidd had not defeated Labour's Bill Butler by seven votes in Glasgow Anniesland, but a quick calculation demonstrates that the loss would have been supplemented with a third nationalist list member, won at the expense of the Labour Party. If nothing else, the results in the North East and in Glasgow should remind us to be awfully cautious about our assumptions.

8 comments :

  1. I find myself unhealthily obsessed with this stuff... but thanks!

    ReplyDelete
  2. I'm in the process of writing a post on this - on how you lot unpicked the lock. If anyone designed a system to confine me, I'd spend all my time trying to find a way round the barrier.

    Your campaign was genius in many ways, although the ineptitude of Labour and our travails meant that it had much less close scrutiny than it demanded.

    The genius bit was that you turned the whole thing on its head.Rather than FTPT constituencies and a top up list, it seems that you took the list as your starting point and then, once you'd ramped up a colossal list vote through your air war - telling people the list vote was to elect Alex Salmond - you then unleashed that on the constituencies in a huge tidal wave with the "both votes SNP" thing in the last few days. That wasn't about getting the constituency vote onto the list. It was, as I now see in hindsight, the other way round.

    Pure. Bloody. Genius.

    Labour clearly lost this election, and we were not in the best position but you lot got it spot on and I expect campaigners the world over will be looking carefully at all you did to learn for themselves.

    ReplyDelete
  3. Thank you, LPW, for the clearest explanation of d'Hondt it has been my pleasure to read.

    I don't think we should blame Labour's campaign strategy for their losses; their campaign did exactly what the SNP campaign did - reflected very faithfully the record, team and vision of their party - and the electorate made their choice.

    ReplyDelete
  4. Me too Jim! And there is a whole nation's worth of fresh data which one could plough through...

    Caron,

    Will look forward to it! Am musing myself on our collective failure to see this coming (and why!) May post something anon.

    birnie,

    Glad you found it helpful. AMS is actually pretty simple and how it works in practice is easily worked through with little more than pen and paper. Despite its apparent obscurity, there's no great magic behind it.

    ReplyDelete
  5. Of course, it helps that the SNP's total votes for the list seats were the most garnered for list seats in a Holyrood election (876,421 beating Labours record of 786,818 set in the 1999 parliamentary elections).

    ReplyDelete
  6. Allan,

    Quite so. In the North East, it was the combination of outrageously high regional votes and comparatively low voting for other parties "wot won it".

    ReplyDelete
  7. It broke? It seems to me that the system worked really well for the first time.

    ReplyDelete
  8. To say that the result reflects a breakage in the system is obviously a rather laden observation, hence the "scarequotes" I deployed in the piece. That said, it seems to me beyond contestation that the 2011 Holyrood election, with its combination of high SNP votes and relatively low opposition support beyond Labour, comprehensively subverted the intended consequences of the electoral system.

    ReplyDelete