Abstract
The pattern of voting at the Brexit referendum has unsurprisingly attracted much attention from academics and commentators. Several academic papers published since that pivotal event in British political history, many using quantitative approaches, tell the same story with nuances reflecting the nature of the data deployed. Qualitative commentaries provide largely similar interpretations – it was the disenchanted with globalisation, with the country’s London-based political elite and party system, with immigration and with unaccountable ‘rule’ from Brussels that swung it – though not in Scotland and Northern Ireland.
Brexit tells the same story, but in greater detail, by three authors with different research orientations: Goodwin has a strong reputation for work on British right-wing political parties and as a commentator on contemporary British politics; his co-authors have substantial track records of quantitative analysis of survey data, and bring to this book monthly surveys conducted since 2005 that detail trends in attitudes towards Brexit and related issues. Although the book is a cooperative endeavour, the main authorship of each chapter is clear.
The chapters are ordered chronologically. The first three – ‘Brexit introduced’, ‘Campaign prologue’ and ‘Into battle’ – set the scene for the referendum and the detailed analyses that follow, providing a generally excellent overview of the issues, personalities and politics. There are some lacunae, however; little is said, for example, about registration and turnout – or the relative lack of it, especially among the relatively young – or about the use of social media in the campaigning. Differential registration and turnout by age were crucial to the referendum outcome and should have received more attention than they are given.
‘Attitudes to Brexit over time’, based on the survey data, argue that three factors influence how people think about issues such as the referendum: (1) calculation – evaluating the relative costs and benefits of EU membership; (2) community – how voters think of themselves, in this case as ‘British’, or ‘English/Scottish/Welsh’ (Northern Ireland is almost entirely ignored), or ‘European’; and (3) ‘cues’ – perceptions of various opinion leaders and their attitudes to the issue. This is merged into a continuing theme of Clarke and Whiteley’s recent analyses of voting at British general elections – valence politics; voters evaluated the benefits of EU membership according to their perceptions of its delivery against key concerns such as economic prosperity, provision of public services, state security and democratic accountability. The arguments are tested using relatively complex modelling of time series data, with terse explanations that general readers will not find particularly illuminating, concluding that a mix of factors affected people’s changing attitudes to the EU. Further analyses involve a wide range of variables but the interpretations are not always clear: we are told that people who rated the Conservative leader highly were more likely to be Eurosceptic – but David Cameron was for Remain!
The fifth chapter – ‘The people’s army’ – is largely based on a survey of UKIP members’ characteristics, attitudes and activities. Most were relatively old, male and with few qualifications; they supported the goal, especially the demand for reduced immigration, but only those with political ambitions were prepared to do much campaigning. The next chapter, on ‘The rise of UKIP’, examines both changes in its support over the 12 years preceding the referendum and its performance at the 2014 European Parliament and the 2015 UK general elections. UKIP’s support is linked to several main indicators in statistical models: attitudes to the volume of immigration, views on the provision of public services as exemplified by the NHS, feelings of relative deprivation and views about the party’s leader for most of the period – Nigel Farage; there is nothing about any possible role of the media in stimulating those attitudes and views. Pessimism about the first three all generated further support for UKIP, as did positive images of UKIP’s leader. As with many of the other statistical analyses, some of the results are difficult to interpret. They also omit important issues. For example, for much of the twenty-first century’s first decade it was generally believed that UKIP’s support came mainly from disillusioned right-wing Conservatives, but after 2010, many former Labour supporters at least temporarily switched their allegiance to the anti-EU party – a trend, with a clear geography and potential implications for post-2016 voting patterns that is nowhere mentioned, let alone analysed (Johnston et al., 2018).
The problems of interpreting statistical models also characterise the chapter on ‘Voting to leave’. A constituency-level analysis shows, as have other studies, that places with older, less qualified, lower income populations were more likely to vote Leave – though less so in Scotland and Wales than in England (differences within England are not explored). But individual-level data analyses then suggest that although these socio-demographic characteristics accounted for 89% of the variation across the 632 constituencies, they were largely insignificant when attitudinal and other indicators were introduced. This difference must surely be because the independent variables in the latter equations were not actually independent; older and less qualified people were more likely to feel relatively deprived, for example. In such situations the relative importance of the various suggested influences on the outcome is difficult to entangle because of what is statistically termed confounding (Johnston et al., 2017). This is clearly identified, for example, in a table reporting a regression with voting Leave as the dependent: there was a substantial and significant negative relationship showing that those who identified as Scottish were less likely to vote Leave than those who did not – but also a substantial positive, though statistically insignificant, relationship with SNP partisanship. But SNP partisans are much more likely to identify as Scottish than non-partisans: what are the true relationships with these collinear variables? The result of including a wide range of, almost certainly related, independent variables in such models is that the conclusions can only be general – everything, or more or less everything, is related to the pattern being analysed, but we cannot be sure which are the most important drivers because everything is confounded with everything else.
The final two chapters present post-referendum analyses. ‘The consequences of Brexit’ explores whether leaving the EU will significantly impact the British economy. Changes in GDP over the previous 65 years suggest that joining the EU in 1973 had no long-term effect on economic growth – nor did it for many other countries after they joined – casting doubts on Treasury forecasts, widely used by those promoting the Remain cause, of a negative impact, although ‘we cannot confidently forecast 15 years ahead’. Analyses of the likely impact of Brexit on immigration conclude rather tortuously that because Britain appears unable to control immigration from outside the EU ‘by implication Brexit will do little to control immigration from inside the EU’! Finally, ‘Beyond Brexit’ summarises the book’s findings, concluding that ‘strong public concern over the large number of migrants entering the country was front and centre to Leave securing victory’ – but there were relatively few immigrants (from either within or outwith the EU) in many of the areas where support for Leave was strong! It then asks whether the result might have been different by looking at possible outcomes if turnout had been complete by incorporating estimates for those who said they didn’t vote would have voted, including for those who didn’t know (but we are not told that it is almost certain to be the case that non-voters are underestimated in the surveys deployed and may not be representative of all non-voters). Allowing for uncertainty in those estimates, Remain was the more likely outcome in two-thirds of the 1 million simulations run. Finally, using European Social Survey data, analyses suggest few major differences in attitudes to the EU across 18 countries including the United Kingdom, with immigration a major issue in most countries.
Whether one of the book’s ‘puffers’ on its back cover is right in suggesting that it is ‘sure to be the standard reference on Brexit’ is challengeable, therefore – although it may be that nobody else essays such a task. That it provides ‘a wonderfully written history of the rise of UKIP’ is less challengeable – much of those sections of the book are authoritative although they lack the apparent insights of the books by non-academic commentators (Shipman, 2016) – than claims for the ‘explanations’ provided by the statistical analyses. The data on which these are based are informative, but the modelling sometimes obscures more than it reveals; perhaps it tries to do too much and a clear picture of the relative importance of influences on the outcome fails to appear.