Mark Diffley: Polls are a snapshot, not a prediction of the future

Polling companies may  not always get the right result, but polls are a good way of testing public opinion
Polling companies may not always get the right result, but polls are a good way of testing public opinion
Share this article
Have your say

As we all now know, making political predictions is pointless. Nobody ever gets them right so we’re best off all actively avoiding them, to save the humble pie we’ll inevitably have to eat afterwards.

And while we are at it let’s ignore polls in future, since they are always wrong and can no longer be trusted.

Well, sure the polls have, like the wider environment in which they operate, struggled with the recent fluidity of opinion shifts, but we need some perspective as well as some understanding of what polls are.

Recently I wanted to scream at the television during an edition of the BBC’s Daily Politics show, when Lord Digby Jones confronted an ex-
colleague about the performance of the polls at the election.

“People like you said it was going to be a 20 or 23 per cent lead,” slammed Jones, claiming this as the reason that he got his own prediction of the result wrong.

While it is true that not all the polls predicted the closeness of the eventual outcome, though some absolutely did, it is unhelpful to have polls misrepresented like this, not least since not one final poll had anything like a 20-point Tory lead.

Perhaps Lord Jones was thinking about the polls undertaken when the election was called in mid-April when the Conservatives did have a lead of around 20-points. But that is rather the point – polls are a snapshot of opinion at a point in time not a prediction of the future, and what this election has shown more than any other is that campaigns can change opinions, something which the polls all picked up as it progressed.

And the point about the polls in this election is that, unlike in 2015, they varied widely in their assessment of the gap between the two main parties. Pollsters themselves were transparent in saying that this was due to judgement calls they were making about likely turnout. In other words, the ‘raw’ data, before being adjusted for turnout, was consistent among all pollsters in showing a tight race, but the variation came when judgements about likely turnout were made.

This left some pollsters looking like geniuses, though those who made the wrong call about turnout have been the first to admit they got it wrong.

So, let’s stop seeing polls as tablets of stone, but rather appreciate them for what they tell us. Let’s look at them in the round rather than focussing on those that support a particular view. Let’s look behind the headline numbers on voting at data on policies and leaders which give clues as to what voters are thinking. And let’s not forget that they do not attempt to predict the future.

Understanding how public opinion is changing at the moment may feel like trying to nail jelly to a wall, but good polling remains the best means of doing so.

Mark Diffley, public opinion and political analyst and owner of Mark Diffley Consultancy and Research