An online poll by ITV’s This Morning show has been circulating on social media, showing a massive lead for Jeremy Corbyn.
It puts the Labour leader on a staggering 68%, with Theresa May trailing behind at just 19%.
It’s been retweeted hundreds of times, with Corbyn supporters saying it shows that Britain is ‘waking up’ and it’s ‘looking great for Labour’.
One of the reasons it’s causing so much excitement is that it’s based on more than 165,000 votes. This is a far bigger sample than most political surveys, by polling companies like YouGov, ComRes or Survation. For instance, a recent YouGov poll on voting intentions used a sample of 1,727 UK adults.
But is the poll reliable?
Bigger doesn’t always mean better. If the sample is skewed, it won’t reflect national opinion – no matter how many people are polled.
This Morning’s survey was open for anyone to contribute to online. And it was left open for several days. So although it collected lots of votes, it’s not designed to reflect the UK electorate — nor does it claim to.
The former president of YouGov, Peter Kellner, told us: ‘A properly conducted poll attempts to create a sample that looks like the country as a whole. So if you’re doing a Britain-wide survey, you want the right proportion of people from the north and the south; men and women; young and old; rich and poor; and so on.
‘Then, when you’ve got the data in, it’s never quite spot on. You might get slightly too many men, or slightly too few Scots or something. So you then “weight” the data to fine-tune it. So the published results have a sample that – as far as you damn well can – looks exactly like the country as a whole.’
This Morning‘s poll does none of this.
This isn’t the only wildly unreliable election survey circulating on the internet. Another one by thepoliticalanalyzer.com puts the Labour leader on over 91%.
It’s not just that these polls don’t reflect the electorate – they may not even reflect the sample of people who voted in them, because it’s easy to vote more than once. Simply delete your cookies, refresh the page, and you can vote as many times as you like.
What’s more, polls like these are often circulated widely among supporters, making them even less likely to reflect the electorate as a whole. This Morning‘s poll was even shared by the ‘Jeremy Corbyn for PM’ Twitter page, which urged its 124,000 followers to ‘have your say’.
We’ve seen these type of dodgy online polls before in the US presidential election, when Trump supporters on Reddit and 4chan flooded a list of surveys in an attempt to change the media’s assumption that Clinton would win.
And even when there isn’t a coordinated attempt to skew the results, open online polls still reflect the types of people who are on the website.
Remember, too, that this isn’t the first time skewed statistics about Corbyn have been spread over the internet. Last year an infographic went viral that compared council election results in 1995, 2006 and 2016. But the elections in question were contested with a different set of seats, so were not comparable.
Are the established pollsters any better?
Many established polling companies have seen their reputation damaged recently, after a series of incorrect forecasts. Most of them wrongly predicted that Britain would vote to remain in the EU, and that David Cameron would fail to win a majority in 2015.
However, these were close-run races, so although they got the ultimate winners wrong the margin error wasn’t miles off.
On the 2015 election, Kellner said: ‘It was three points out – it wasn’t ten or twenty. So when the polls “get it wrong”, they’re not usually that far off. Usually the polls are pretty close.’
Nate Silver, the US statistician and political analyst, has said that UK polls tend to be around five or six percentage points out. That’s enough to make the wrong call about Brexit, which closed at a 52%-48% split. But it’s not enough to accidentally place Corbyn around 40 points behind where he actually is.
Opinion polls should always be interpreted with caution, but with this one we’re happy to dismiss it outright. Created as a diversion, many seem to have taken it seriously online.
‘It’s completely worthless,’ said Kellner. ‘They got just about everything you can possibly get wrong, wrong. This is not, in any meaningful sense, truthful information. It’s just preposterous.’