A reminder that hypothetical polling questions are a bit rubbish

I wrote about this in my last post - exploring what, if anything, we could tell from the polling about whether Boris Johnson would get the blame if Brexit did indeed end up being delayed past the 31st October.

With the government now pushing for an election in December the issue has now arisen again, with lots of people dragging out a ComRes poll from the 16-17th October that asked how people would vote in an election if Britain had NOT left the European Union on 31st October, showing Labour one point ahead. Some people are sharing it with excitement, others with dismay. Both should probably calm themselves.

As a general rule, you can only usefully ask people a polling question if they actually know the answer... and most of us aren't actually very good at predicting how we will respond to hypothetical situations. If you take this specific question, it was asking people to imagine quite a lot. How had the delay come about? Had the government fought it, or gone along with it? How had the government explained and reacted to the delay? Given the dates of the fieldwork, many respondents wouldn't even have known about the deal. All of these things will impact how the public react and whether they blame the Conservatives or not...

but were impossible for respondents to know.

In short, polls measure current public opinion. They can't predict the future. While you can ask respondents to predict their own future opinions, they aren't necessarily very good at it.