The opinion polls generally provide a good picture of voting intention but in 1970, 1992 and with the Liberal Democrat vote in 2010 most did not get it right.

This was further exemplified in the 2015 General Election, UK Referendum on membership of the European Union (although there was less data to match against) and overseas the election of Donald Trump as U.S President.

The overall pattern is important as any one poll can deviate because of sampling error and papers, who buy a stream of polls, may just pick the one that gives the story that they want, but even the overall pattern is hardly exactly right.

The polling industry has exploded in the last few years with Sky News ticker-tape showing at least 5 different polls at any one time.

The predictions for the 2015 General Election were particularly damming and led to some heads of polling companies being brought before a Parliamentary Select Committee.

In 2015 the Conservatives won an overall majority to form a government by themselves. The polling companies for weeks had suggested that this would not happen, it was too close to call and the likely scenario was another coalition. Indeed the Conservative Party during the campaign capitalised on the fear of the Labour Party going into coalition with the Scottish Nationalist Party.

There can be no doubt that when some people went to vote these polling figures were influencing their choice. Even at their basic form people may be influenced to vote for the under-dog (very British) or not feel the need to turn out to vote after a long day at work because it looks like their favourite has victory secured.

In addition, variation in swing between constituencies can be considerable. ukpollingreport.co.uk is very good on the latest polls and the problems of interpreting them and has an historic archive of national polls back to 1970.

Previous Next