Marco Morucci and Sally Symington test the ‘Shy Tory’ hypothesis that has been mooted as the explanation for the polls’ failure to predict the election outcome and find it unconvincing.
The release of exit polls on 7 May was a moment of shock and awe for political scientists, pollsters and forecasters across the UK: polls and subsequent predictions had grossly underestimated both the vote share for the Conservative party as well as their share of seats.
Hungry for a quick and simple explanation of the phenomenon, the mainstream media and commenting classes were quick in bringing up the ‘Shy Tory’ hypothesis. The adage dates back to 1992 and goes something like this: right-wing voters felt cornered by the adversarial and negative propaganda directed at them by the left wing, prompting them to feel safer in withholding their voting intention on surveys by either answering they’re undecided or won’t vote.
The theory has received little real scrutiny nor been critically evaluated despite the self-reinforcing coverage it has been given since the election, to the extent that it has now morphed into ‘Lying Tories’. However, some pollsters and experts have already manifested their doubts on its value. It presents a number of flaws that discourage its adoption as a principal explanation for the polls.
First, if Shy Tories are a thing, then why aren’t there ‘Shy Voters’ in general? There is little evidence of the phenomenon occurring anywhere else in modern, western democracies. When this happens it is in highly divided ethnic contexts, or in post-authoritarian countries. Practically, voters have to feel physically threatened to deem it worthwhile to lie on anonymous polls. As dreaded as a few raised eyebrows around the pub table might be, it’s safe to say that Tories are unlikely to have equated it with physical harm. In addition, if the hypothesis were true, then we would have to explain why voters that allegedly lied on the opinion polls felt perfectly fine disclosing their actual preference in the exit poll.
Second, even if the Shy Tory factor was real, polling firms have been systematically adjusting for it since 1992. A number of studies in later elections have confirmed the effectiveness of these methods. Moreover, if we adopted the hypothesis, then we would have to accept the existence of ‘Boisterous Greens’, which caused the over-estimation of the party’s vote share by most polls.
More to the point as the Shy Tory hypothesis goes, right wing voters should answer either ‘undecided’ or ‘won’t vote’ in polling surveys. If this were to have any effect on turnout prediction, it should lead to an underestimation; in fact polls significantly overestimated voter turnout instead. Clearly, the actual turnout figures indicate that many people who had said beforehand that they would vote, chose not to on the day.
Adding to this, the gap between estimated and actual turnout was negative in the case of Labour won seats, positive in the case of Conservative (Table 1). This suggests the shortfall in turnout occurred amongst Labour voters and perhaps explains why Labour vote share was overestimated, but equally not why Conservative share was underestimated, and explicitly goes against the conclusions of the Shy Tory hypothesis.
An alternative explanation for the failure of the polls can be found in the structure of the campaign run by the Conservatives. Guided by the established campaigning strategies of Australian election guru Lynton Crosby, the Conservatives ran a pragmatic campaign, strongly localized and almost exclusively aimed at target constituencies. This centred around a ‘de-capitation’ strategy aimed at Lib-Dem marginals; and almost all of them showed an above average swing to the Conservatives on election night (Table 2).
The (almost always) regionally balanced sampling strategy adopted by pollsters could very well have failed to pick up the effects of this localized, and regionally unbalanced campaign, as more than 50% of these target are in South West England. As far as we know, Conservative-targeted constituencies counted as much as all the other ones in the polls. This means the effect of the campaign might have been overshadowed by an abundance of observations from places the Tories weren’t actually targeting.
It is illustrative to look at the seats which changed hands between the Conservatives and Labour. Notably the share of votes between the two main parties is very similar (in line with the ‘raw’ data produced by pollsters, but with the Conservatives requiring a much lower threshold percentage of votes to win the seat.
Significantly, the collapsing Liberal Democrat vote fragments differently across constituencies. Notably the Conservatives gained seats from Labour when a significant share of the Lib Dem vote went to UKIP, whereas Labour gained seats by increasing their share of the vote to a far greater extent than UKIP. Furthermore, the median marginality of the 2010 results for seats gained by the Conservatives was 2.3% compared with the median marginality of seats gained by Labour of 3.8%.
Adding to this, the demographics-based weighting system adopted by some of the pollsters failed to capture the key Tory supporting demographics: in the case of ICM this weighting shifted the predicted share three points away from the Conservatives and towards Labour, thereby yielding a 6 percentage point gap in the wrong direction before other adjustments were made.
Crucially, this also explains why seat shares were not predicted well. As Benjamin Lauderdale already noted in his forecasting post-mortem, mistakes in national share polling, rather than constituency-level, were the main cause of wrong seat predictions. Moreover, the targeted campaign explains also why Tories have actually won the election: they have gained more seats than 2010, while maintaining virtually the same national-level share of votes (Labour has even gained votes).
About the Authors
Marco Morucci is a Research Assistant for Parliamentary Candidates UK. He holds a B.A. in Political Science and International Relations from LUISS Guido Carli University of Rome and a MSC in Comparative Politics and Democratisation from LSE. His research interests include politics and government in developing countries; democratisation in Southeast Asia and quantitative and computational methods for the Social Sciences.
Sally Symington is a Researcher for Parliamentary Candidates UK. She completed a MSc in Public Policy (part-time ) at UCL in August 2014. The subject of her dissertation was the effect that both gender quotas, in the form of All Women Shortlists, and term limits would have had on the levels of female representation in the UK House of Commons. Sally has an undergraduate degree in physics and she worked as an investment analyst and fund manager for ten years. Following a career break, she returned to voluntary work. She has experience of local democracy having been a parish councillor for the last seven years, of which four have been as chair.