Opinion Polls and the General Election – a case study from 1950

[Note: originally posted on a different blog on the day of the 2015 General Election, 7th May, 2015]


[Conservative Central Office staff recording the results of the 1950 General Election as they come in]

As we near the end of polling on Election Day 2015 with the Conservatives and Labour neck and neck and each predicted to win 34% of the vote, psephologists around the country will be anxiously awaiting the final test of their opinion polling techniques. Opinion polls have been a major feature of this election campaign, but rarely have they been completely accurate in predicting the outcome of an election, and typically the polling methodology used has received almost as much analysis as the election result itself.

The February 1950 general election proved no less of a headache to pollsters. After 5 years of a Labour government following a strict austerity programme, the Conservatives under Churchill were confident of victory. Demonstrating the growing interest in scientifically-based opinion polling, Conservative Central Office set up its own Public Opinion Research Department (PORD) at the end of 1948. The PORD, whose surviving records form part of the Conservative Party Archive at the Bodleian Library, produced summaries of public opinion based upon intelligence reports sent in by Area Agents around the country who reported on reactions to Party broadcasts and local election rallies, as well as opinion polls published in the Press. Its monthly summaries were circulated to Party officers, shadow Ministers and MPs, while a ‘Confidential Supplement’, which contained much more sensitive information, had a more restricted circulation of the Chief Whip and a select group of shadow Ministers, including Churchill.

In its first ever report on public opinion in January 1949, PORD identified the key issues affecting the electorate at that time, all mostly related to austerity:

user comment

user comment

[Public Opinion Research Department: Public Opinion Summary No. 1, Jan 1949: Shelfmark: CCO 180/2/1]

Though noting in this report that 44% of those polled by Gallup in December 1948 advocated the formation of a National Government, PORD consistently predicted a good result for the Conservatives in the year prior to the general election.

As Election Day drew closer, PORD’s forecasts reduced the Conservatives’ presumed lead only slightly, despite noting in December 1949, that,

Conservative defeatism and despondency is spreading, and with it there has been a rise in the Socialists confidence of coming victory’ and, ‘Although there is no vestige of evidence in any of the public opinion polls to support the expectation of a Socialist victory, the City of London continues to reflect the Conservative pessimism. Wall Street is apparently following the City’s lead, and extending its view across the United States.

[Source: Public Opinion Research Department: Confidential Supplement to Public Opinion Summary No. 12, Dec 1949: Shelfmark: CCO 4/3/250]

On 11th November, 1949, it predicted 49.3% for the Conservatives and 41.4% for Labour, which would have resulted in a comfortable majority of 109 MPs in Parliament. This changed to 49.1% and 41.4% respectively by December, but in its last report on the evening before the election it dropped this substantially further to 46% and a 41-seat majority. The result, when it came, was 43.5% for the Conservatives, 46.4% for Labour, and a Labour majority of 5.

Analysis of the results, and the poor polling, was quick to follow:

user comment

user comment

user comment

user comment

[Public Opinion Research Department: General Election Analysis No. 1, Mar 1950: Shelfmark: CCO 180/2/1]

PORD blamed the defeat primarily on Labour’s success in reeling-in the majority of undecided voters at the last minute, and the ‘violent blasts of all-out propaganda from the Socialist machine’. Indeed, some of intelligence reports received by PORD testify to some of the tactics employed locally by the ‘Socialist machine’, such as the ruse by Labour supporters in Darlington to keep Conservative canvassers occupied:

CCO 4-3-250(7) - Darlington

[Public Opinion Research Department; Area Agents’ Intelligence Reports, Feb 1950: Shelfmark: CCO 4/3/250]

But PORD also acknowledged the Party’s failure to capitalise on the Women’s vote, which was due ‘undoubtedly [to] the predominating political influence of the menfolk’ and a ‘widespread and continuing disbelief in the secrecy of the ballot.’

Despite the setback, the Labour majority was only 5, and the 1950 General Election saw an overall swing to the Conservatives of 3.8%. Indeed, internal Party communications immediately following the election were surprisingly upbeat, focusing on the inevitability of another general election in the near future (which took place on 25th October, 1951), and one could almost be forgiven for believing that the Conservatives had won the election:


[Editorial of the Conservative Party Newsletter, Tory Challenge, Mar 1950: Shelfmark: PUB 214/3]


[Introduction to Tory Challenge, Mar 1950 by Party Chairman Lord Woolton: Shelfmark: PUB 214/3]

Opinion polling has come a long way since 1950, although even in 2010 most pollsters under-estimated the final result achieved by both Labour and the Conservatives, and the 1992 election suffered probably the worst predictions of modern times. The PORD barely survived its dismal predictions at the 1950 election, being wound up in 1953. Its work continued though, being taken up by the Conservative Research Department, and records of 60 years’ of opinion poll analysis survive and are available for research in the Conservative Party Archive.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.