One of the difficulties of opinion polling is that it lives and dies on the ability of the people you are asking to predict the future. Asking someone how they’ll vote, particularly when the timing of the next election is unknown, is a lot like asking if you’ll buy a coffee or a tea tomorrow.
You may have a strong preference one way or another, but that doesn’t mean that your preference is a useful guide to what you will end up buying. I myself would much rather have tea than coffee, but I despise most café-bought cups of tea, which tend to be either too strong or too weak. If I can’t have a pot of tea, I’d much rather have a coffee.
And if neighbours on my block have had a loud party or argument that has kept me up half the night, even if I had the choice of a pot of tea, I might opt for a strong coffee, even though I don’t even like coffee that much.
As you can see, it’s very difficult to model an opinion survey that would accurately tell you even something as superficially simple as what I’ll buy from a café tomorrow, let alone how I’ll vote in an election at some point in the non-specified future, in a constituency as yet unknown to me, with candidates still to be decided. And that’s just one person’s decision!
The People’s Vote campaign is trying to do something still more difficult with its polling about what happens if or when Brexit takes place: predicting not just how people will feel about the major parties, but the shape of an election campaign that takes place afterwards. That’s not just “which will, you choose, coffee or tea?” but “Which will you choose, coffee or tea, on a Thursday, when it is raining heavily, and you are far from home and have just had bad news?”
So that’s the important thing to understand right off the bat about any poll, including the 25,000-person mega-poll released by the People’s Vote campaign this week, that seeks to tell us what will happen to the Conservatives or to Labour after Brexit: it’s a really big ask and it may be impossible to carry off with any degree of accuracy. Their answer: if Labour facilitates the passage of Brexit, the party will go down to epochal defeat.
The advantage of size is that it allows us to have a more accurate sense not only of how the whole country will vote but specific regions and demographics. Disappointingly, unlike Survation’s mega-poll for Channel 4, the People’s Vote campaign haven’t done a specific breakout for ethnic minority voters, one of the crucial swing blocs in the referendum. Indeed it’s difficult to see what value they’ve really got out of this mega-poll at all, though one assumes that they will have paid for additional questions that will inform their strategic decisions but have not been released to the public (and therefore YouGov has no obligation to publish).
What it doesn’t do is make the answers any more robust in of themselves. A statistically relevant sample is only one part of getting useful answers: the other part is the questions you ask.
To return to the tea-and-coffee analogy: it’s very easy to construct a survey that produces your preferred answer, whatever that is. If you want me to say I want tea, tell me I can get it by the pot. If you want me to opt for coffee, tell me I can buy tea in a cup or not at all. It’s a lot harder to construct a poll that gives you the correct answer.
So we know that in the event that Labour facilitates Brexit, then the party will pay an electoral price among Remainers and that if it blocks Brexit it will pay an electoral price among Leavers if (and this is the crucial part) the next election is about Brexit. What we don’t know is what the next election will be about.
Think about the last election: when pollsters asked what would happen if the Conservatives and Labour went into an election promising to implement Brexit and the Liberal Democrats, SNP, Greens and Plaid Cymru went into an election promising to stop it, they got results pointing to a heavy Labour defeat.
What pollsters couldn’t know to ask, but would almost certainly have got a more accurate answer to is: what happens if at the election, the Liberal Democrat campaign becomes a referendum on whether homosexual sex is sinful, the SNP go into an election promising another independence referendum, and the Greens and Plaid Cymru’s messages are essentially shut out of the national debate altogether? The answer, it turned out, was that Labour gained both seats and votes.
The problem is that elections are complex events and that arguing for what political parties should do about Brexit based on the next election is a losing position for anyone advancing it, because any honest conversation has to start with a confession that no one really knows enough to even ask the right questions.