General Election 2024: Why are different polling companies getting such different results? (2024)

With the general election campaigns well underway, a slew of polling is coming in, - feeding the ravenous appetites of those who want to know how the vote will turn out in six weeks' time.

But just a week in, there is already somewhat of a chasm between what different pollsters are saying in these voter intention surveys.

On the 'narrow' (close) side, JL Partners had Labour 12 points ahead of the Conservatives - three points closer than their last set of results from a month earlier.

General election latest: PM's pledges 'could send UK into recession'

At the other end of the spectrum is YouGov's poll for Sky News - their results show a gap of 27 points between the two main parties.

You can read more about how the Poll Tracker here.

So what is causing this massive difference? You could fit JL Partners' gap into YouGov's more than twice over.

Well, there are a number of factors, but a glaring one is how different groups interpret the "don't knows" - people who say they have not decided which way they want to cast their ballots.

"How people are treating people who say they are undecided seems to be the primary source of disagreement between the results," Professor Will Jennings, Sky's polling expert, says.

He explains the different ways in which polling companies treat this category of voters.

One type simply excludes anyone who replies "don't know" - and then just calculate their results off those who chose a party.

Read more about the general election:
What happens now an election has been called?
Find your new constituency and how it's changed
How boundary changes make Starmer's job harder
The MPs who are standing down

"Another group of pollsters 'squeeze' people who are undecided by asking them a follow-up question about which way they are leaning," Prof Jennings adds.

And then another group, such as Opinium and JL Partners, reallocate the 'don't knows' in various ways.

Prof Jennings says: "Opinium uses how people voted at the last election and assume that they will vote the same way - this tends to favour the Conservatives, because people who voted for the party in 2019 tend to be heavily represented in this group, as there are just more of them.

"JL Partners use a complex method that uses information from answers to other questions in the same poll - about leaders and policies - to identify the party that undecided voters are most likely to opt for.

"This tends to produce higher numbers for the Conservatives at the moment."

There are pros and cons to all these methods - including the fact that some voters are "undecided for a reason", the professor adds.

He says that, in response to not forecasting the 2015 Conservative election win, many pollsters changed their calculations to "correct for the historical under-estimation of the Conservative vote".

But these changes ended up over-correcting - meaning there was a big under-counting of Labour votes and slight over-estimation of Tory support in polls before the 2017 election.

These factors mean it is hard to know for certain which set of data is closest to the current sentiment.

There are also concerns about the makeup of the panels that pollsters use.

Some, like YouGov, have their own panels and they can track people's views back through time.

👉 Click here to follow Electoral Dysfunction wherever you get your podcasts 👈

But others purchase their respondent pools from companies, and others use 'river sampling', where they attract people using adverts on websites or prize draws.

It is important to remember that polling is a snapshot of sentiment at a point in time. It is at best a tool to use in conjunction with others to help inform a view, rather than a bulletproof forecast of how people will behave.