Polling inquiry highlights communications challenge for industry

The findings, available here, outlined the initial thoughts of the inquiry panel around the general causes of the failure of almost all pollsters to correctly predict a Conservative majority last May. 

Ruling out factors such as turnout weighting and late swing, the discussion focused on the idea that unrepresentative samples — specifically underrepresentation of older voters and overrepresentation of the politically engaged — were the most likely explanation of the error. Early recommendations from the panel focused on improving sampling, either by obtaining a more representative sample of people within levels of current weighting variables, or by improving weighting by using more or better weighting variables. 

However Professor Patrick Sturgis, director of the National Centre for Research Methods and chair of the panel, insisted that there was "no silver bullet" to address the issue. "The risk of polling misses in the future can be reduced, but not removed," he said. 

According to ComRes’ director of political polling, Tom Mludzinski, this awareness of the limitations of polling is key to moving forward. "I think that’s part of the communications challenge for the industry – being realistic about what we can and can’t do," he said.

"In some ways what happened during the election has strengthened our hand in what we’re trying to do – that is in being more realistic, but also being cleverer. We shouldn't just be trying to go for one-off findings, but also trying to find out what’s behind the data: what’s underlying it and what’s causing reactions and perceptions and where it might be going."

The process has also been, said Mludzinski, a useful exercise in hearing what’s happening across the industry. "We feel we’re on the right lines with the work we’ve done already," he said. "It’s positive for all the agencies to be looking within themselves and it’s in our competitive nature to be as accurate as possible. I’m sure we’ll all take away a number of learnings."

A full report of the inquiry, commissioned by the British Polling Council and the Market Research Society, will be published in March. Interested parties can submit evidence for consideration by the inquiry panel here.