By Jeremy Blackwell
You’re lucky this time you won’t have to go through the ‘A few words about me’ and the ‘Hits and misses’ parts. You already got them. So let’s go directly to the heart of the matter.
How does my model work?
Like all election projection models, my model works mostly on ‘Uniform National Swing’ (UNS). With UNS you have all seats moving in exactly the same direction as the national polling average—which never happens in real life.
So I have a second algorithm on top of the first one which I call ‘relative swing’. It factors in not just the usual UNS, but the way parties’ vote shares change relative to each other.
For example, you might start with the SNP on 50%, Labour on 24%, the Conservatives on 15%, and the Liberal Democrats on 8%; then later have the SNP on 45%, the Conservatives on 30%, Labour on 15%, and the Liberal Democrats on 5%. With UNS you simply subtract 5% from the SNP vote, 9% from the Labour vote, 3% from the Liberal Democrat vote, and add 15% to the Conservative vote, in every seat. (And get negative votes for some parties sometimes.)
But the truth is that it’s more complex than this. That’s what ‘relative swing’ takes care of, by taking into account the ‘multiplier effect’. With the above numbers you multiply the SNP vote by 0.9, the Labour and Liberal Democrat votes by 0.625, and the Conservative vote by 2.0 in every seat. (And sometimes get a vote total over 100%.)
So my model is a mix of both, and currently tuned on 80% UNS and 20% ‘relative swing’.
The first step as described above will sometimes deliver negative votes in a few cases. And in some seats the vote total doesn’t add up to 100%. The second step is to eliminate all the negative votes. This is done simply by automatically switching from the model parameters to ‘100% relative swing’. This being a multiplier, it guarantees there can’t be any negative results. Third and last step is to deal with seats where the vote total is not 100%. This is also done automatically by recalculating votes proportionally from what the second step delivers.
Is the model foolproof? Of course not. No statistical model is, especially when you consider tactical voting. That’s why we can get wrong projections even with accurate polls (see ‘Hits and misses’ in my previous article). But I’ve compared my results with those you can get using Electoral Calculus or ScotlandVotes and they’re pretty similar. Not identical, but pretty close (more on that later). So I guess their underlying algorithms are pretty similar to mine.
How do I feed the model?
The source data are all the polls that I can find. Ideally I would rely on full Scottish polls (that is polls fielded in Scotland only and with a sample size of over 1,000). There were plenty such polls before the 2015 General Election and they proved accurate, unlike UK-wide polling. In the graph below the small dots are individual polls. The large dots at both ends are the actual 2010 and 2015 results. The trendlines show the evolution of vote shares and how polls correctly predicted the SNP landslide.
We are not that lucky this year. The snap election took everyone, including the pollsters, by surprise and there are many fewer full Scottish polls: nine so far and only five in 2017.
I therefore also have to rely on Scottish subsamples of UK-wide polls. A subsample size is typically 100 to 150, so they have a much larger margin of error (MOE) than full polls. This is the reason why many people think they should be discarded from any analysis. In principle I agree with this point of view, but this year subsamples are the bulk of the data we have. And there is a simple way to deal with the larger MOE and still get valid projections when interpreting the data. Which I will explain in the next section.
What does the model give in return?
First of all, below is a table of projected vote shares for all 59 seats individually. This is simply a projection of what the vote shares are supposed to be with current polling and the algorithm described above. I’m taking a calculated risk here. This table will be online after the election and of course what’s in there will be compared to the actual results, seat by seat. Some will be spot on; some will be way off; most will be somewhere in between. And the misses could be held against me. But I believe in transparency, so here you have it:
Then you have the trendlines. It’s a basic function in Excel graphs so there’s nothing imaginative in this. Here is what you get right now from Scottish polling, set on 15-point rolling average:
Trendlines show what was already apparent from the raw polling data. The SNP is down from 2015, while the Conservatives are up. The safe result for the SNP is a 25% lead, but recent polling points to 15 to 20% only.
Then we have the rolling average I use for the projection. I include the 12 most recent polls (full polls and subsamples) to determine it. But it is a weighted average. This means the raw results are weighted by sample size. Since full samples are typically 1,000 to 1,100, and subsamples are 100 to 150, it means a full poll has 6 to 8 times more weight than a subsample. That way I take care of most of the ‘outlier effect’ of subsamples.
The final step is projecting the seats by SLLM rating. SLLM stands for ‘Safe, Likely, Lean, Marginal’ and is the likelihood each party has of winning any given seats. Below is what we have right now:
As a by-product, the model also projects the winner’s margin in all 59 seats. A graph is an easy way to see how close some are and also how safe some others are. Below is what we get right now. I think it clearly identifies the danger zones for the SNP, most of which were already identified in 2016.
And what do I project and predict?
First of all, there is a big difference between projection and prediction.
Projection is when you just run your model and publish the results as they are. That’s what I did in the previous section.
Prediction is when you say ‘here I’m going to take a risk and say the statistics are wrong’. That happens when you are aware of local factors that go against statistical trends. Or when the model projects an almost perfect tie in any given seat. Then I take the risk to predict that seat differently from what the model says.
The first table below has the result of my projection for all 59 Scottish seats in the third column, compared to what other models deliver (Electoral Calculus or ScotlandVotes) when fed with the exact same data. And the fourth column is my prediction which already differs from the projection.
The second table is the summary (total number of seats for each party). I will update both on Thursday, and then every Thursday until the morning of Election Day.
My changes between projection and prediction are not that risky at the moment. I switched Aberdeen South back to the SNP column because the model projects it as a tie. But the SNP gained the seat by 15% in 2015. The SNP also did well in the almost overlapping Scottish Parliament constituencies in 2016 (a 43% average in Central and South). It will be closer than in 2015 (and with Conservatives second this time) but I think Callum McCaig will hold this one for the SNP.
I also switched Moray and Perth and North Perthshire back to the SNP column because they are part of the ‘historic five’ that the SNP has held continuously since 1997 through boundary changes. And they are currently held by two major SNP figures (Angus Robertson and Pete Wishart). Both are seats that the SNP can’t afford to lose. So I expect them to allocate resources there to the level needed to hold both.
What are the main issues at hand and what could influence the outcome?
I will not discuss this in detail here. It is not really my point. And I expect everyone to be fully informed and aware of the latest developments. Let’s just say the Prime Minister wants this election to be about Brexit and Brexit only, while there are many other issues to debate (rape clause, pensions triple-lock, fracking, education, NHS, benefits cap). I say also election fraud. This is probably why the Prime Minister doesn’t want to take part in any debate. But that’s me campaigning so I will stop here.
In Scotland several parties will try to turn this GE into another independence referendum, just like they’re trying to do already with Council elections. I just hope voters won’t fall for it, and vote on the real issues in each election. And I think one of the questions that should be asked is: which devolved powers would the Prime Minister take away from the Scottish Parliament if she wins this GE? But that’s me campaigning again so I will again stop here.
With so little time left before the election I can’t see any major event influencing the outcome. Even a proven terrorist threat and North Korea did not have any significant impact on the French Presidential election, so I think it would be the same here. Right now I don’t foresee anything that could alter current trends.
Unless someone has an ‘EdStone Moment’ in mind. If so, please drop the idea—it wasn’t even funny the first time around.
The easy part is that this General Election will definitely be SNP-Conservatives one-on-one, and that the SNP will win it. Labour and the Liberal Democrats show a remarkable lack of ambition, seriously contesting only three seats each as per BBC reports. They have conceded they are now only minor players in Scotland.
The difficult part is to predict by how wide a margin the SNP will win. Which is precisely my purpose here. I think the SNP already clearly know where the danger zones are (they have been identified in the 2016 SP elections) and will do what’s necessary. What everybody must also have in mind is that even if the SNP were to lose 8 or 9 seats to the Conservatives (as Professor Curtice so imprudently considered possible on the basis of just two polls), they still will be the first party in Scotland with a huge majority of the Scottish representation in Westminster. And a mandate. But that’s me going partisan, so I will leave it at that.
Much of the outcome will be decided by turnout. In 2015 Scotland had the highest turnout of all four nations. I hope this will be the same in 2017 and that we will all make our best informed choice. You know mine. Hope you will make yours too in a reasoned way.
And by doing so we will once again prove we are a strong nation.
Saor Alba gu bràth.
Jeremy Blackwell, 30 April 2017
Jeremy Blackwell is an analyst and statistician living and working in Edinburgh. You can follow him on Twitter at @WeAreThe59.