Polling puzzle: how response times may explain election surprise
|
Version 0 of 1. In their mission to find out what went wrong in the general election in May, pollsters have been staring deeply into some unlikely places – including the brain. Martin Boon of ICM Unlimited has been talking to neuroscientists to uncover the telltale signs when respondents’ answers aren’t what they seem. “Brain science has come a long way since the days of crude lie detectors, which would look out for a little pick-up in pulse. It is now possible to put probes on the skulls of a respondent as they watch a speech, or a leaders’ debate, and gain direct insights into the emotional response,” says Boon. Related: Why opinion pollsters failed to predict overall majority for Cameron “In principle, it should also be possible to estimate the robustness of a respondent’s suggestion that they will, for example, vote Labour, by wiring them up while they complete the survey, and analysing what is going on in the brain while they tick the box.” However, requiring all participants to be set up with brain-scanning equipment and every response analysed by a trained neuroscientist would send polling costs rocketing and sample sizes would shrivel, rendering them less reliable. Fortunately there are other, simpler, signs that may betray a poll respondent’s true intentions. Boon is particularly interested in how long it takes respondents to answer a question. Psychologists suggest quick response is associated with certainty, while a delay could indicate an openness to changing one’s mind. After three-year old Alan Kurdi was found washed up on a Turkish beach this summer, ICM conducted an online poll to assess the public response. It confirmed that one family in 10 might welcome a stranger in need of a haven into their home – a strikingly high number. But when Boon analysed how many milliseconds respondents had taken to say yes or no to that question, he found that fewer than half of those who had said they would give up their spare room had answered with a speed that suggested they really meant it. By contrast, those who had indicated that they would keep their door closed had mostly pressed the button with less hesitation. Boon says: “If some people hesitate before giving the more ‘socially appropriate’ answer, while others plump more quickly for the ‘selfish’ choice, then that might well be a sign that your raw poll is going to understate the strength of ‘self-interested’ sentiment.” This is one way to explain why voters plumped for the Conservatives in the privacy of the ballot box on 7 May, in spite of what the polls said. If the research on response times proves fruitful, it could eventually lead to new weighting schemes, which take account of how fast punters click and how certain they are to stick with their declared choice. |