LONDON — New research from behavioural science experts suggests that the “shy Tory” effect some pollsters have reported in previous elections is growing, with a growing embarrassment factor attached to identifying as a Conservative voter.
As a result, the firm is warning that most of the polling ahead of Thursday’s general election on Thursday has underestimated the number of people likely to vote Conservative.
In a survey carried out by research consultancy Decision Technology (Dectech), only 26% of Brits said they plan to vote for a Conservative candidate on Thursday.
However, when using a counting technique that protects anonymity, that figure rises to 31%. This gives an “embarrassment factor” of 1.16 times, which is higher than the 1.11 times seen at the 2015 general election when the same experiment was carried out.
After most major pollsters failed to predict the Conservative majority in 2015, Peter Kellner, chairman of polling firm YouGov, suggested that “shy Tories” could explain the margin of error. The theory goes that Labour support had been significantly overstated because it was seen as more socially acceptable to indicate support for a left-of-centre party.
A subsequent report published by British Polling Council and the Market Research Society argued that there was only “trivial” evidence of “shy Tory” factors in 2015, and said polling error was incorrect for other reasons, most importantly unrepresentative samples.
The report added that it “does not mean that [“shy Tory” factors] might not cause more substantial problems in the future.”
Political polling is ‘fundamentally inaccurate in its current form’
The technique which grants respondents anonymity is called the Unmatched Count Technique (UCT). It relies on the usual self-report approach that pollsters rely on but involves showing respondents a series of statements asking how many are true for them, but not which ones. This reduces the instances of lying due to social taboos while allowing measurement of the incidence rate of these taboo topics.
Dectech says the research indicates the need for pollsters to radically reform their methodologies and introduce the UCT technique to their fieldwork.
Election after election, we have seen results that ran counter to predictions from all major pollsters
Henry Stott, Dectech director, said: “Given the predictions of a landslide Tory victory when the general election was first called, some were suggesting that the taboo around voting Conservative had decreased and even speculated that we may see a ‘Shy Labour’ effect this year. But our research demonstrates that this is far from true — although there may be fewer people voting Tory than in previous years, the embarrassment factor has grown.
“Political polling is fundamentally inaccurate in its current form. Election after election, both here in the UK and overseas, we have seen surprising results that ran counter to predictions from all the major pollsters. It is clear that there are certain voting behaviours that are seen as too socially taboo to admit to in a direct survey response — even in exit polls — but are not considered so taboo that it actually changes how people vote.
“While pollsters have tried to learn lessons from the past and have made some changes to their methodologies, particularly in terms of voter sampling, they are not going far enough. Our research demonstrates that the Unmatched Count Technique could provide more accurate results. A radical overhaul is long overdue to account for people’s embarrassment when asked to self-report on their behaviour.”
Business Insider Emails & Alerts
Site highlights each day to your inbox.