Representative samples are an issue for the pollsters – but so are respondents who lie

The British Polling Council recently published their report about what went wrong with the polls in the 2015 general election. The report dismissed problems associated with voter registration, question wording, postal voting and mode of interviewing and opted for unrepresentative samples as the key factor in explaining what happened. Here, Paul Whiteley and Harold Clarke make the case for examining another factor: respondents who don’t tell the truth.