I would like to suggest that the polling industry should not agonise too much about getting elections wrong. Error may be good for you; an occasional disaster can be good for the industry and good for the public. If polls were to have too long a run of always getting it right, it would raise FALSE expectations for your trade. It would discourage voters and politicians from fighting elections to the end and it would make pollsters too un-self-critical - and then they would be riding for a fall.
In 1945 on July 5th, the day of Britain's post-war general election the News Chronicle published a poll which said that the Conservatives would get 41 per cent of the vote and Labour would get 47 per cent. Now, everybody in Britain knew that Mr. Churchill was going to win the post-war election and everybody in Britain who saw that poll knew that it was one of these new-fangled gimmicks which was obviously wrong. And the News Chronicle itself said, "While we have no reason to disbelieve these poll figures, the British electoral system is such a lottery that you cannot in the least predict the outcome of the election from these polling figures."
As the result of a current study of British political attitudes we have been led to examine the accepted measures of class and in particular the point at which the line is usually drawn between the "middle" and "working" classes. In the summer of 1963 we collected interviews from a stratified random sample of just over 2 ,000 British electors and seventeen months later we managed to reinterview just on three-quarters of these. Our main findings will be written up elsewhere but, since our approach to class may be of help to people working in quite different fields, we are publishing this immediate report.