“Who still trusts pollsters, though?” is the inevitable response from Labour campaigners when faced with the Conservatives’ current poll lead over Labour.
After the 2017 General Election results—when a hung parliament surprised so many pundits—it’s not hard to see their case. Although the polls throughout the campaign were relatively accurate about the vote share for the Conservatives and the Liberal Democrats, they under-estimated Labour almost uniformly and over-estimated the performance of the SNP in Scotland and Ukip.
So can you trust them?
How trustworthy are polls?
Firstly, it’s important to note that the big pollsters, like YouGov, are members of the British Polling Council, which enforces rules as to how polls can be conducted.Not every poll is like this, however. Loaded questions—and, more often, loaded write-ups—are not uncommon, particularly if someone has paid for a poll and had input in producing it.
That being said, surveys funded by a particular paper or campaign group are not necessarily biased. A common misconception is that a certain group of people—say, Times readers, or middle-class respondents—will be over-represented in results.
Pollsters weight their samples to mitigate effects such as this, and you can usually find the sampling and poll tables on their website to find out more.
How to read opinion polls
One common technique is to look at a “poll of polls.” The BBC, for instance, includes a “trend line” as well as individual poll results on its party support tracker.
From this, we can see that Labour’s performance is rising, but they are still behind the Conservatives. The Liberal Democrats have fallen to 17 per cent.
It is also worth noting that national polls which look at vote share may not account for the complexities of our first past the post electoral system. Some pollsters therefore also have seat projection models; last time, for instance, YouGov’s pointed to a hung parliament.
You can find out more about different opinion polls, including dedicated pages looking at Scotland and Northern Ireland (where different parties stand compared to England), at Electoral Calculus.
How good are polls at predicting the election result, then?
Veteran pollster Peter Kellner reminds us that in most countries “where poll results can be compared with actual results … well-designed polls are usually accurate to within 3 per cent.”
“However, it is true that in a closely-fought election, a polling lead (in a sample of 1-2,000) of less than 5 per cent for one candidate or party over another cannot be regarded as a certain indicator of who is ahead at the time the survey was taken.”
He also points out that while polls act as a “snapshot” of current public opinion, they cannot guarantee who will remain ahead.
Thinking again of the question of voter volatility, we can imagine how a poll conducted today could show a different picture to one conducted just before polling day.
____________________________
Read: Welcome to the age of the political swinger
____________________________ While the political parties will try to guide campaign messaging, there can always be unexpected twists—and while many are billing this as “the Brexit election,” it is not yet certain whether voters’ concerns over Brexit will necessarily guide their hand on polling day, especially if other issues emerge as key stories over the course of the campaign.
Opinium, for instance, has found that 59 per cent of poll respondents identified health and the NHS as a key issue facing the country, above the 52 per cent who named Brexit.
Other pollsters, however, disagree.
So… in short, we don’t really know?
We can’t, no. But we can understand all the ways we do and don’t know.
Great, thanks.
