1. Britain Says NO: Voting in the 2011 AV Ballot Referendum Paul Whiteley Harold Clarke David Sanders Marianne Stewart

2. Britain Says NO: Voting in the AV Ballot Referendum Paul Whiteley, Harold D. Clarke, David Sanders and Marianne C. Stewart Parliamentary Affairs, November 2011, 1-22 The purpose of this paper is to explain why voters made the choices that they did in Britain’s Alternative Vote (AV) referendum on 5 May 2011. The paper utilises four alternative theoretical models to analyse individual voting behaviour. They are described as the cost–benefit, cognitive engagement, heuristics and mobilisation models. The explanatory power of these models is investigated using a large survey data set gathered in the AV referendum study conducted in conjunction with the British Election Study. Multivariate analyses show that all four models contribute to explaining why some people voted in favour of electoral reform, with the cost–benefit model exhibiting particularly strong effects. The conclusion discusses public reactions to the referendum and possible implications of the decisive rejection of electoral reform after a campaign characterised by disaffection and disengagement.

3. 2011 BES AV Ballot Survey National Internet Survey – Fieldwork YouGov Campaign and Post-Referendum Waves Campaign Wave N = 22,124 Post-Referendum Wave N = 18,556 (83.9% retention rate) Campaign Wave – 30 Daily Replicates, Average N = 732 All Respondents Previously Interviewed in 2010 BES CIPS or BES CPS Surveys Data: http://bes2009-10.org

4. How Did it Begin? Early YouGov Poll June 14, 2010

5. Referendum Vote Intention Dynamics: January – May 2011

6. Referendum Vote Intention Dynamics: April 5th – May 4 th 2011

7. Interest in the Referendum in the Run-Up to the Vote

8. AV Referendum Vote Shares

9. Theoretical Perspectives on Referendum Voting Merits of the Case – Costs and Benefits of the proposal Cognitive Engagement – political involvement & knowledge (lack of knowledge promotes status quo voting) Heuristics – Leader Images and Partisanship (follow the leaders you like) Mobilization – parties & others try to bring supporters to polls

10. AV Voting Model Specification Merits of Case – (1) cost-benefit scale; (2) democracy scale; (3) traditionalism scale Cognitive Engagement – (1) interest in referendum; (2) political knowledge scale; (3) attention to politics (4) media – AV news Heuristics – (1) party leader images; (2) party identification Mobilization – (1) campaign contact Controls – Age, Education, Gender, Income, Country (Scotland, Wales)

11. Cost-Benefit - Opinions About AV and FPTP

12. Principal Components Analysis of AV Indicators Communalities Initial Extraction AV is fairer 1.000.751 AV too much influence for small parties 1.000.543 AV better reflects opinions 1.000.777 AV makes MPs work harder 1.000.549 FPTP is a British Tradition 1.000.735 FPTP easier responsibility attribution 1.000.395 AV hard to understand 1.000.298 Extraction Method: Principal Component Analysis.

13. Total Variance Explained Component Initial Eigenvalues Extraction Sums of Squared Loadings Total % of Variance Cumulative % Total % of Variance Cumulative % 14.04857.833 57.833 4.048 57.833 57.83 2.86712.388 70.221 3.6809.717 79.938 4.5417.726 87.664 5.3955.647 93.311 6.2894.123 97.434 7.1802.566 100.000 Extraction Method: Principal Component Analysis.

14. Component Matrix(a) Component AV is fairer -.867 AV gives too much influence to small parties.737 AV better reflects opinions -.881 AV makes MPs work harder -.741 FTPT is a British tradition.857 FTPT is easier attribution of responsibility.629 AV is hard to understand.546 Extraction Method: Principal Component Analysis. a 1 components extracted.

15. The AV Cost-Benefit Scale (Note: Don’t Knows are recoded so they end up being close to the mean)

16. Opinions About Constitutional Reform

17. Political Knowledge Scores Number of Correct Answers to Eight Questions

18. Yes Voting by Level of Political Knowledge

19. Party Leader Images 2010-2011

20. Yes Voting by Party Leader Images

21. Campaigning: A slanging Match?

22. Evaluations of the Yes and No Campaigns

23. Total Positive & Negative Evaluations of Yes & No Campaigns

24. Binomial Logit Model of Voting Yes in the Referendum Bs.e. Cost-Benefit Scale 2.43***0.05 Democracy Scale 0.11***0.03 Traditionalism Scale-0.25***0.03 Interest in Referendum 0.16***0.03 Political Knowledge 0.14***0.02 Attention to Politics 0.010.02 Media Consumption of AV News-0.000.02 Cameron Image-0.17***0.02 Miliband Image 0.08***0.01 Clegg Image 0.15***0.02 Labour Identification-0.22**0.09 Conservative Identification -0.81***0.10 Liberal Democrat Identification 0.37***0.12 Other Party Identification-0.080.11 Yes to Fairer Votes Campaign 0.25**0.09 No to Av Campaign-0.17*0.08 Gender (Male) 0.21***0.06 Age 0.0010.002 Annual Family Income 0.010.01 Education 0.15***0.02 Scotland-0.22**0.09 Wales-0.190.13 Constant-2.86***0.22 McKelvey R-Squared.75 Percentage Correctly Classified87.9

25. Changes in Probability of Voting Yes Associated With Changes in Significant Predictor Variables

26. Possible Consequences of the Decisive No Vote Issue of electoral reform likely will not be reopened for many years Spill-over effects for constitutional reform – widespread support for some reforms but they may not happen ‘Paradox of Referendums’ – A lack of trust in regular political process -> turn to referendums. But if issues are complex & remote the status quo bias will be strong. This bias is likely to be strengthened by lack of trust in parties & politicians & widespread disengagement AV Referendum was not a positive exercise in democratic governance – low turnout, widespread lack of interest, inter- campaign bickering, negative knock-on consequences We waited 35 years for a second national referendum – we may have to do the same for the next one.