NEWS11 May 2015
UK — The British Polling Council, supported by the Market Research Society, has announced it is setting up an independent enquiry to look into possible causes of bias in the pre-election polling.
The council said in a statement: “the final opinion polls before the election were clearly not as accurate as we would like, and the fact that all the pollsters underestimated the Conservative lead over Labour suggests that the methods that were used should be subject to careful, independent investigation.”
The enquiry will be chaired by Professor Patrick Sturgis, professor of research methodology and director of the ESRC National Centre for Research Methods and will make recommendations for future polling.
In terms of average pre-election polling the Conservative vote was underestimated by 4.2%, Labour vote was overestimated by 2.4% and Lib Dem was overestimated by 0.9%. No polling company consistently showed a Conservative lead to the extent that was finally recorded.
Over the weekend pollsters started putting out statements and opinions analysing the pre-election polls with the final result.
Anthony Wells, director at YouGov said: “Every couple of decades a time comes along when all the companies get something wrong. Yesterday appears to have been one such day.”
YouGov’s final poll put Conservatives on 34% (the election result was 37.8%), Labour 34% ( 31.2%), Lib Dem 10% ( 8.1%), UKIP 12% ( 12.9%) and Green 4% ( 3.8%).
“The all-important margin between the Conservatives and Labour was significantly off. When something like this happens there are two choices. You can pretend the problem isn’t there, doesn’t affect you or might go away. Alternatively, you can accept something went wrong, investigate the causes and put it right,” said Wells.
He pointed to the poll difference not being a random sample error but some deeper methodological failing and said it wasn’t down to mode effects but that “one potential cause of error may be the turnout models used”.
Andrew Cooper, founder and director at Populus pointed to the 1992 election and a similar over-estimation of the Labour vote, and subsequent change to methodology.
“At the next four general elections the polls have been right. There are likely to be variety of reasons behind the difference in the polls and the final outcome. Very late swing to the Conservatives, polling weightings, polling methodology and claimed propensity to vote will be just some of the factors that are likely to be discovered once an investigation is completed.”