Are polls ruining elections?

It seemed like you couldn’t escape a media discussion in the run up to the UK general election this month without an official poll of some sort underpinning it. In fact, all throughout the duration of a government’s tenure, there are popularity polls coming out very frequently.  

It’s not a new phenomenon, in fact, in the 1960s there were calls for election polls to be banned for fear of how they may influence voting behaviour. Brushed off by the political establishment at the time, they have since been a mainstay of British elections ever since.

So, how did the 20-odd pollsters perform this time round? With a resounding win for the Labour Party, which even the UK bookmakers knew was nailed on, the interesting thing to learn is how close were they to the final voting break down.

Albeit broadly right on the vote share, the pollsters certainly overplayed Labour and Reform UK’s voting share and underestimated how much the Conservative Party would eventually achieve.

And there lies the question. Did the polls themselves play a role in the results of a general election?

According to research conducted by The University of Warwick, it seems the answer is yes. Analysing all general election results since 1983 (excluding 2024), academics believe “the margin of victory identified in national poll predictions significantly impacts voters’ willingness to show up at the ballot.”

Moreso, the effect of polls on behaviour is also more distinct the nearer the poll is to the date of the election.

Although not conclusive, it could explain why the Conservative Party changed its attack line late in the run up to the election. When it seemed clear they were on their way out of government, they begged the electorate to prevent a Labour “super majority” by voting for them.  This was presumably a last gasp pitch to get shy Tory voters to the ballot box. And it may well have been successful.

For clarity, a super majority in UK politics is not actually a thing, but there may be an element of voters who chose to vote in this way convinced by the argument that a political party with a huge majority is not necessarily very healthy.

Either way, what certainly is true is the poor election turnout. Only 60% of the eligible voting public put an X against a candidate. This is the third lowest number since 1918. The highest was in 1950, where more than eight in ten chose to vote.

A new element to the 2024 election that should need to looked at is the requirement for voter ID at a polling station, the first time it has been required for a general election. There have been reports that up to 400,000 people may have been prevented from voting because they failed to provide appropriate ID.

Whether 400,000 extra votes could have been enough to shift the election meaningfully seems a little unlikely, but it would certainly have had a bearing on the results.

Even more than the polls themselves? Who knows…

Previous
Previous

Getting a-breast of the data

Next
Next

An introduction to Companies House and its data