Polls apart: do the general election predictions really stack up?
This election has been punctuated with u-turns, lack of preparation and the constant reminder of the power of statistics and the danger of getting them wrong.
History continues to repeat itself but just like brands, politicians and pollsters appear not to listen and learn.
Pollsters have flubbed spectacularly in the last few elections. Having got it spectacularly wrong regarding the Tory victory in 2015, Brexit and Trump, they are understandably nervous that Labour has been closing the gap on the Tories.
The campaign trail has provided a roller coast ride.
Theresa May promised solemnly on several occasions not to call a general election until 2020 “to allow a period of stability”.
Politicians are, however, in the business of getting re-elected.
It is therefore not surprising that with polls suggesting the weakest opposition in living history, a u-turn snap election proved too tempting.
When Theresa May called the election on 18 April, the Tories were on a nine-year high A survey by YouGov based on a sample of 1,727 GB adults online suggested Con 48%, Lab 24%, Lib 12%, UKIP 7%, Other 9%
Within the first few days of campaigning The Tories hit their highest vote share since 1991: Con 50%, Lab 25%, Lib 11%, UKIP 7%, Other 8% (ComRes – sample 2,074 GB adults online).
U-turns, tax and pension fears, care costs, the NHS and security have all ensured we are kept on the edge of our seats.
YouGov’s shock poll this week predicting a hung parliament on 8 June is brave and the Times’ decision to splash it on its front page on 31 May even braver. The end of May indeed!
The controversial poll suggested the Conservative Party could be in line to lose 20 seats and Labour gain nearly 30 in next week’s general election, according to new modelling.
YouGov’s first constituency-by- constituency estimate of the election result predicts that the Tories would fall short of an overall majority by 16 seats, leading to a hung parliament.
With the delicious fusion of tragedy and comedy worthy of his namesake, Stephan Shakespeare, YouGov’s chief executive admits that it is only a central projection that “allows for a wide range of error” and “.. these are just the midpoints and, as with any model, there is leeway either side. The Tories could end up with as many as 345 and Labour as few as 223, leading to an increased Conservative majority.”
The Bard goes further and says that “it would take only a slight fall in Labour’s share and a slight increase in the Conservatives’ to result in Mrs May returning to No 10 with a healthy majority.” The Times points out that the projection means the Tories “could get as many as 345 seats on a good night … and as few as 274 on a bad night”. That is a massive margin of error!
Having been so badly bruised and with public confidence in the industry at an all-time low, pollsters have been putting careful caveats on their results commentary, emphasising they are merely snapshots of public opinion. This estimate, however, is clearly being presented by the Times as a prediction of next week’s result.
The pound fell against the Euro as investors feared an increased possibility of a high-tax Labour government or a hung parliament.
But just like fraudulent mediums, throw out enough predictions and just remind people of the hits….
Pollsters really don’t know if they have a representative sample.
Polling’s dirty little secret is that most people who are approached by a pollster refuse to answer. The more unpopular a view may be perceived to be, the less likely someone will admit to it. It is for this reason that I believe the pollsters were so wrong about Trump.
As a proud Member of The Inner Magic Circle with Gold Star as well as a lawyer, predictions and understanding human behaviour come with the territory.
For several years, I have successfully predicted each of the election results around the world.
I predict that Theresa May's decision to call a snap election will backfire. For once YouGov with their controversial new model poll will get it right. The Tories will not win a landslide victory rather there will be a hung parliament. Allowing the usual statistical margin of error of plus or minus 2% -3% applied to each party's vote and subject to my being wrong, I predict:
Con 318 Lab 261 SNP 35 LD 13.
Statistics have been the undoing of many politicians on the campaign trail from undisclosed caps on healthcare to Diane Abbott on the Nick Ferrari show agonising over the cost of 10,000 extra police officers.
With the digital dexterity of president Jacob Zuma, Jeremy Corbyn’s car crash interview on Woman’s Hour made him the latest Labour figure to struggle with the costings of the Labour manifesto. Desperately trying to find information on his iPad, the Labour leader was unable to say how much his party’s plan to introduce free childcare for 1.3 million children would cost.
As home secretary and Theresa May stand-in Amber Rudd pointed out to Corbyn during last night’s TV debate, there is no magic money tree.
Throughout this election politicians have wriggled out of giving clear answers, fluffed figures and fallen victim to failure to prepare and pollsters have thrown out widely varying predictions.
As a result, many times throughout the media coverage we have been tempted to respond in the same way as the audience member in Battle For Number 10 with what must be the new election meme:
If there are particular stories you feel should be subjected to a pressure test to find out whether they really stand up to serious scrutiny or you want help to avoid the predictable errors and omissions of others get in touch...