Makerfield By Election Betting Guide: Has the Restore vote been exaggerated?

By-election history tells us to always question the polls

If the polls are to be believed, Thursday will see Andy Burnham triumph in Makerfield and return to Westminster with Downing Street firmly in his sites.

But should they be believed?

Polls of individual constituencies have a fairly poor reputation, ever since Lord Ashcroft funded numerous such polls ahead of the 2015 election which proved largely unreliable. More recently though, pollsters do seem to have improved. Let's look at all the by-election polls since the 2019 election.

It's a mixed bag but a fair bit better than the reputation of constituency polls suggests. It's worth noting that many of the big misses were early polls in seats, often before candidates were selected and the voters got to really hear the campaigns.

Big poll misses are common

These headline results do conceal some embarrassing misses on other candidates however.

In Batley & Spen the Survation poll might have been within 7% of the Lab/Con outcome but underestimated George Galloway by 16%. In Uxbridge & South Ruislip we saw JL Partners find Reclaim on the cusp of saving their deposit (5% in the poll) only to get less than half that at the ballot box.

And Survation (again) might want to forget their poll in Mid Bedfordshire finding the True & Fair Party (led by Gina Miller, of Brexit court case fame) on 4% only for them to score...0.2% on the day.

Polls often better on leading candidates

In fact, polls consistently do much better on the leading contenders than the also-rans. Generally, overestimating the latter by a fair margin.

In Gorton & Denton, three polls found the Tories on 8%, 5%, 6%. Actual result: Under 2%. In Runcorn & Helsby the Tories and Lib Dems were both overestimated by both polls by roughly 5% each. Survation's final poll in Mid Bedfordshire not only had the True & Fair cock-up but two other candidates clearing 5%, neither of whom did. Likewise JL Partners in Selby & Ainsty predicted five parties to clear 5% but only three did on the day. The list goes on.

Voters for minor candidates, even for major parties, fading away from poll response to ballot box, is the most consistent pattern in the past few years. It has occasional exceptions (particularly if the candidate is George Galloway), but much more often than not candidates polling 5-10% never show that support on the day.

Has Restore vote been overinflated?

So what does this mean for Restore in Makerfield? Three pollsters have them on 8%, 7%, 8%, and 5% - prime flattering to deceive territory. But journalists visiting the seat have often reported their vote seems solid. And in fairness, Restore did exceed expectations in Great Yarmouth at the local elections and clearly have a real ground game in Makerfield - something other 3^rd/4^th place campaigns often don't.

Nonetheless, I think it's touch and go whether they will show up. And the markets are far too confident that the polls have them measured correctly. You can get close to 9/25.50 on Restore to lose their deposit, or lay them at around 6/42.50 to miss the 5-10% bucket in either direction. I think the market has forgotten how hard it is to measure modest levels of support, and either of those are good bets.

Now read our Makerfield live blog with updated prices on this key by-election