Cutbot's Holyrood seat forecaster
April 6th, 2016
We have a new baby. Well, two new babies. The first one is the news that our public affairs monitoring service now covers Westminster as well as the Scottish Government and Scottish Parliament. So if your organisation has a need to track areas of policy you’re interested in, at either level or both, please do drop us a line. That’s the hard sell over.
The second one is we are today launching a Holyrood seat forecaster. You pop in the vote shares you think the parties might get, or which the latest poll suggests, and we’ll give you a forecast for how many seats each party could get. Now, such things do already exist, notably the one produced by our friends at Weber Shandwick. And, to be fair, theirs is prettier than ours. But there are a few wee things that we wanted a forecaster to do that theirs doesn’t.
First, what about RISE? Their polling isn’t great at the moment, but what if it picks up? We’re basing those predictions on the SSP’s 2011 results, where they did best (relatively) in West, then Glasgow. Our model suggests that a little over 3% would be enough to get them that first seat in West.
Second, what about the Greens? Specifically, they’ll have three constituency candidates this time, and two of those are in stronger areas for the Greens. Even without a surge in Green votes, that could make a difference, so we’re factoring them in. There’s a longer post available for the anoraks on how exactly we’ve done that (which also has more detail on the calculations for “others”, and about how the figures overall are generated). The short explanation: we’ve used a substantial amount of handwaving and the best finger-in-wind system we could think of to generate forecasts for those three Green constituency vote shares. It’s sketchy, but it’s better than pretending they’re not standing.
The final main thing we wanted to show is the actual constituency and regional results our national forecasts are built on. What might the most marginal constituencies and regional seats be? Who might narrowly lose out and who to? Where might the SNP win so many constituencies that they don’t get any list seats? So the model includes all that kind of information.
We also generate a custom URL for you, each time you do a prediction, so if you want to share that forecast by email or on social media it’s easy to do, and there’s a Tweet This button too, which will work if you’re logged into Twitter in the same browser. We also produce a link for you for each constituency and region so you can share specific links of that sort if there’s one you’re interested in. Perhaps you’re the candidate and our forecast shows you might win?
It would be nice do a shiny map like Weber Shandwick have done, and we may still do that. But for now your results will be a lot of tables and data. Other than a nice map, please do let us know if there are any features you would like.
Please bear in mind that polls are only a bit of fun, and so too is this. No forecaster like this can take account of local demographic changes, of the strongest campaigns in each constituency, the candidate everyone hates, or any of the countless factors that will influence every result next month. So we can say one thing for sure: even if you put the exact result in of the May 5th vote, we won’t get every last seat and region right. But hopefully we’ll be close enough.
Oh, and thanks to TNS, there’s a poll today to test it against. Here’s what their numbers look like when put through our forecaster. A slightly increased overall majority for the SNP, with the four most marginal seats all going their way by less than 700 votes.