The opinion polls: The election they got (almost) right

Parliamentary Affairs

Volumn 58, Issue 4, 2005, Pages 28-42

  Crewe, Ivor ^a  

^a NONE

Author keywords

[No Author keywords available]

Indexed keywords

EID: 27144530648     PISSN: 00312290     EISSN: None     Source Type: Journal    
DOI: 10.1093/pa/gsi074     Document Type: Review

References (16)

1. See I. Crewe, 'A Nation of Liars? Opinion Polls and the 1992 Election', Parliamentary Affairs, 1992/4, pp. 475-95 and Market Research Society
2. The Opinion Polls and the 1992 General Election, July 1994.
3. An invaluable source of information about the polling companies' methods - and about UK opinion polls in general - is the excellent UK Polling Report website (http://pollingreport.co.uk). The measured comments by 'Anthony' in the Commentary section are always worth reading.
4. For example, in the MORI/Financial Times poll conducted on 1-3 April, 74% of Conservatives said they were 'absolutely certain' to vote, compared with 57% of Labour supporters. As a result Labour led the Conservatives by 38% to 33% among all respondents but trailed by 34% to 39% among the 'absolutely certains' - a ten point difference. In its final poll for the Financial Times, 81% of Conservatives reported that they were 'absolutely certain' to vote compared with 70% of Labour supporters. As a result, a 15% Labour lead among all respondents was reduced to a 10% lead among the 'absolutely certains' - a five point difference. See MORI's very informative 2005 general election website: http:// www.mori.com/election2005.
5. See Market Research Society, The Opinion Polls and the 1992 General Election, pp. 48-53.
6. See, for example, MORI, 'Polling on the internet' (http://www.mori.com/ mrr/2002/c020712.shtml)
7. and in defence of internet polling, P. Kellner, 'For the record', Guardian, 12.2.03 (Politics Special Report: http://politics.guardian.co.uk/ polls/comment/0,11030,893890,00.html.
8. Labour's lead fell from 12 to 3 points in the NOP/Independent poll, 8 to 3 points in the ICM/Guardian poll; 7 to 2 points in the Populus/The Times poll and from 0% to a 5% Conservative lead in the MORI/Financial Times poll.
9. See endnote 3.
10. P. Riddell, 'Conservative support reaches a record low', The Times, 4.5.05, p. 26.
11. 'Andrew Cooper Q&A', Times on Line, 6.5.05.
12. On the pattern of constituency swings in 2005, Labour would have retained eleven of the seats it lost to the Conservatives (and possibly Dundee East, lost to the SNP) if the actual share of the vote had been the same as the average final poll forecast.
13. ICM, YouGov and the British Election Study recorded little change over the same period, and MORI reported an increase in Labour's lead from 2% to 10%.
14. The Harris poll (based on fieldwork between 1 May and 4 May), with the actual result in parentheses, reported Conservative 15% (16%), Labour 39% (39%), Liberal Democrat 22% (23%), SNP 20% (18%), Others 4% (4%), a mean error of 0.8%.
15. The Daily Telegraph predicted a Labour majority of 130; the Guardian one of 90-110; The Times one of 'well above 100 and possibly up to 130', and the Independent, which published the highly accurate NOP poll, a majority of 90, but under the headline 'ONE QUARTER OF VOTERS UNCERTAIN AS LABOUR SAG AT THE FINAL POST'.
16. Which makes it surprising that the Chair of MORI, Sir Robert Worcester, who has frequently made this point in the past, could not resist the temptation of trumpeting the BBC/ITN's forecast of a 66 majority as a triumph for the exit poll, which MORI coconducted with NOP. (See 'Worcester's Weblog', MORI website, 6.5.05.) It wasn't really. It owed more to the skill - and good luck! - of the forecasters.