Why Follow the Polls?

The Institute for Government held a panel discussion yesterday (Tuesday 16th April) on the role of polling in shaping political strategy in the run-up to an election. The panellists and audience were largely convinced of the value of political polls, but outside Westminster many people are sceptical.
So, why should we listen to what the weekly polls are saying?

Polling errors tend to be picked up by media, and the public, especially where pollsters return similar results, and collectively turn out to be wrong come election-time. Yet this masks the reality that polls are usually excellent predictors of national opinion, within the acknowledged margins of error, and have become more accurate over time. Only, we must be clear on what a poll is and is not doing.

In 2015 and 2017, when margins were so tight, the polling margin of error – typically 2% - was the very difference between a slim majority or a hung Parliament, hence the lack of consensus among pollsters in both elections. Today, Labour leads decisively in the polls, but translating a 20-point lead to Parliamentary seat-share is still a complex process.

The standard question asked is how you would vote if an election were called tomorrow, but this is not the same question as how people will ultimately vote. Come election day, there is no knowing how many ‘Don’t knows’ might opt for the Conservatives or Labour as a ‘safe’ bet, or how many supporters of Reform UK will decide to vote more tactically in unwinnable seats.

Similarly, each pollster models voting intention in a slightly different way, depending on how they define, or weight for, a ‘politically representative sample’. We each use slightly different ‘squeeze questions’ to push voters into saying which way they would most likely jump. All of these factors, and others, all influence what poll-watchers call the ‘house effects’, in other words the inherent tendency of each polling firm to model outcomes in one direction or another by virtue of their treatment of the raw data.

So keeping an eye on the polls is valuable but, except for the final eve-of-election polls, they are a snapshot in time rather than an attempt at predicting the actual outcome, especially this far out. Polls are invaluable then at telling us what is happening now – how political strategies and messages are landing with voters – but they must always be interpreted through the prism of the longer-term trends and underlying demographic shifts.

Jacob Anderson