General Election 2024: Why are different polling companies getting such different results?
With general election campaigns in full swing, a wave of polling data is emerging, fueling the curiosity of those eager to predict the outcome in six weeks. However, even at this early stage, a notable disparity exists between the findings of different pollsters regarding voter intentions.
On the closer end, JL Partners reports Labour leading the Conservatives by 12 points, a three-point tightening from their previous month’s results. Contrasting sharply, YouGov’s poll for Sky News shows Labour with a 27-point advantage over the Conservatives.
What accounts for this significant divergence? The primary factor lies in how different pollsters handle the “don’t knows”—voters who have yet to decide how they will cast their ballots.
Professor Will Jennings, Sky’s polling expert, highlights this as the main source of discrepancy. Polling companies vary in their treatment of undecided voters:
- Some pollsters exclude respondents who say “don’t know,” basing their results only on those who have chosen a party.
- Others, referred to as “squeezers,” follow up with undecided voters to determine which way they are leaning.
- A third group, including Opinium and JL Partners, reallocate the “don’t knows” using various methods.
Opinium assumes voters will follow their previous voting patterns, which tends to favor the Conservatives, as many undecided voters previously supported them in 2019. JL Partners employs a complex approach, using responses to other questions about leaders and policies to predict the likely choice of undecided voters, often resulting in higher numbers for the Conservatives.
Each method has its advantages and drawbacks. For instance, some undecided voters may be uncertain for specific reasons that these methods might not account for.
In response to underestimating the Conservative vote in 2015, many pollsters adjusted their calculations, but this led to over-correcting and subsequently underestimating Labour support in the 2017 election. These adjustments contribute to the uncertainty about which poll data most accurately reflects current sentiment.
Another concern is the composition of the panels used by pollsters. While some, like YouGov, maintain their own panels and can track opinions over time, others rely on purchased respondent pools or “river sampling,” which recruits participants through online advertisements or prize draws.
It’s crucial to remember that polling provides a snapshot of sentiment at a given moment. It should be used as one of several tools to gauge public opinion rather than a definitive forecast of election outcomes.
