2015 UK election opinion poll errors – what does this mean for Australian politics?

UK 2015 general election polling error
In the lead-up to the UK General Election held on 7 May 2015 nearly every opinion poll rated the contest to be a very close race between the Conservatives and Labour with the real prospect of a hung parliament (www.telegraph.co.uk/news/general-election-2015/11374181/latest-poll-tracker.html). The polls were proved wrong because they underestimated the Conservative vote by 4.2 percentage points (Booth, R. 15 May 2015, http://www.theguardian.com/politics/2015/may/14/why-did-the-election-polsters-get-it-so-wrong). The Conservatives secured a 7 point lead and clear majority with the capacity to govern in their own right (www.dailymail.co.uk/news/article-307427/Why-DID-polls-wrong-Respected-pollsters-failed-account-shy-Tories-embarrassed-reveal-true-allegiances-writes-GUY-ADAMS.html).

This significant error across the polls caused the British Polling Council to establish an independent inquiry to look at the causes and make recommendations for future polling (www.bbc.com/news/uk-32652104).

A number of reasons have been suggested for the polling errors.

‘shy Tory’ voters who concealed their allegiance until they voted (see for example, Porter, T., 8 May 2015, http://www.ibtimes.co.uk/election-2015-five-reasons-why-pollsters-got-it-so-wrong-1500357).

Lower voter turnout than implied in most polls causing the opinions of non-voters to be included in polling results (Nardelli, A. 9 May 2015, http://www.theguardian.com/politics2015/may/09/election-polls-made-three-key-errors).

People lying to pollsters, for example, claiming to vote Labour but voting differently or not voting at all (www.dailymail.co.uk/news/article-307427/Why-DID-polls-wrong-Respected-pollsters-failed-account-shy-Tories-embarrassed-reveal-true-allegiances-writes-GUY-ADAMS.html).

Fewer people owning landline telephones which has been a standard method for contacting voters in the past. (Australian research also points to sampling problems that landline surveys do not pick up mobile only households and potential demographic bias incurred because younger people are more likely to be mobile only households, older women are most likely to answer the phone and telephone interviews conducted during the day are more likely to reach unemployed persons, stay-at-home parents or retirees (Macreadie, R. 2011, ‘Public Opinion Polls’, Research Service, Parliamentary Library, Parliament of Victoria, http://www.parliament.vic.gov/publications/research-papers/2007-public-opinion-polls/download).

Internet polls based on a self-selected sample not constituting a random and representative sample and respondents more likely to lie online (see, for example, http://www.dailymail.co.uk/news/article-307427/Why-DID-polls-wrong-Respected-pollsters-failed-accounty-shy-Tories-embarrassed-reveal-true-allegiances-writer-GUY-ADAMS.html; http://www.ibtimes.co.uk/election-2015-five-reasons-why-pollsters-got-it-so-wrong-1500357).

Prospect of the Labour Party only being able to govern in coalition with the Scottish National Party may have put some voters off voting Labour because it was regarded as too risky (www.theguardian.com/politics/2015/may/14/why-did-the-election-pollsters-get-it-so-wrong).

Asking the wrong questions? A suggestion there is a difference between asking questions about attitude in relation to party choice and questions directly asking for choice of party at the ballot box (www.dailymail.co.uk/news/article-307427/Why-DID-polls-wrong-Respected-pollsters-failed-account-shy-Tories-embarrassed-reveal-true-allegiances-writes-GUY-ADAMS.html).

It was also suggested that the betting market would be a more reliable indicator because the money at stake causes people to give more thought to their judgement. As a case in point, compared to the opinion pollsters prediction of a ‘tie’ the Conservatives were 5-1 on favourite to win on Betfair (www.mytimes.com/2015/05/09/world/europe/uk-british-election-polling-html?_r=0).

What are the implications of this issue for Australian politics?

As a general rule prime ministers are more likely to maintain their position so long as they and their party retain electoral popularity. In the Australian political context speculation about a leadership challenge for a sitting prime minister takes hold when polling results find the prime minister is very unpopular and the party in government would lose an election based on the predicted two-party preferred vote. It should be borne in mind, however, that Australian pollsters’ practice of assigning preferences based on the previous election result when calculating the two-party preferred figure does not take account of shifts in the party landscape that have occurred in the interim and which could possibly affect the accuracy of the result.

Poor opinion polls were central to the poll-axing of Australian prime ministers Kevin Rudd and Julia Gillard (Singleton, G. et al 2013, Australian Political Institutions 10th edition, pp. 222-224). Prime Minister Tony Abbott’s leadership was challenged in the government party room in January 2015 when his popularity slumped to 22 per cent and the government’s two-party preferred vote fell to a low of 44.5 per cent at which level of support the government would have lost an election (Singleton, G. Blog January 2015, ‘Abbott leadership under challenge’, https://singletonauspol.wordpress.com/tag/abbott-leadership-under-fire).

Abbott survived the challenge but the circumstances that brought it on remained an issue. Subsequent poor poll results with the Coalition continuing to trail Labour in the two-party preferred vote provided the breeding ground for further comment that the prime minister’s power base remained fractured (Matthewson, P. ‘New polls but same old pain for Abbott’, 13 April 2015 http://www.abc.net.au/news/2014-04-13/nmatthewson-new-polls-but-same-old-pain/6388070).

The Abbott government’s May 2014 budget was a major contributing factor to its poor polling performance. Its 12 May 2015 Budget faired a little better. A ResearchTEL poll taken the day after the budget was brought down (www.reachtel.com.au/blog/7-news-national-poll-13may2015) which reported a slight increase in the Coalition’s PRIMARY vote of 1.3 per cent saw headlines by some media as an ‘Abbott lift’ (Perth Now) and a ‘much-needed poll boost’ (thenewdaily.com.au; skynews.com.au). There was media speculation that the response to the 2015 May budget would be the catalyst for an early election but this was not confirmed by the prime minister (abc.net.au/news/2015-05-13/budget-2015-tony-abbott-sidesteps-possibility-of-early-election/6465520) and for good reason. The Coalition’s vote may have increased slightly but the two-party preferred figures from the ResearchTEL poll taken on 13 May 2015 showed the Coalition at 47 per cent trailing Labor’s 53 per cent. On those figures the Coalition would lose a federal election. Abbott’s personal approval rating increased to 33 per cent but this is still a very poor level of electoral support for his leadership.

Australian politics, it may be argued, are ‘poll obsessed’ (Glenday, J. 2013, ‘Opinion polls explained: How to read them and why they matter’, Off the Hustings, http://www.abc.net.au/news/federal-election-2013/off-the-hustings). Even small movements in the polls down or up, despite the accepted 3 per cent margin of error, are watched very closely by the parties who respond accordingly with changes to policy and/or a change of leadership if it is thought this would enhance the party’s chances of improving their position in the polls always with an eye on winning the next election. Inaccuracies in the polls, such as occurred in the UK general election, could have policies and leaders placed in the firing line without a substantive reason. The question is whether Australian parties and their leaders would be willing to take the risk.

Informative articles on issues relating to opinion polls include: Macreadie, R. 2011, ‘Public Opinion Polls’, Research Service, Parliamentary Library, Parliament of Victoria, http://www.parliament.vic.gov/publications/research-papers/2007-public-opinion-polls/download; Bonham, K. 2013, ‘A Field Guide to Australian Opinion Pollsters’, http://kevinbonham.blogspot.com.au/2013/08/a-field-guide-to-australian-opinion.html; Glenday, J. 2013, ‘Opinion polls explained: how to read them and why they matter’, abc.net.au/news/federal-election-2013/off-the-hustings.

Dr Gwynneth Singleton
15 May 2015

Advertisements