05737 -5- Pages within 12 hrs

Chapter 7 Enacting elections Richard Heffernan

1 Introduction 271 1.1 Structure of the chapter 274 1.2 Aims of the chapter 274

2 Collecting and generating evidence about elections 276 2.1 Social science and general elections in the UK 278

3 Explaining participation and non-participation in elections 282 3.1 Explaining turnout in elections 284

4 Social science theories of voting decisions 289 4.1 The class-based theory: a sociological explanation

of voting 290 4.2 The party-identification theory: from a sociological to a

social-psychological explanation of voting 292 4.3 The valence-politics theory: an issue-based explanation

of voting 294 4.4 Description and understanding in the evaluation of

theories of voting 297

5 Enacting elections: informing and shaping public debate 300

6 Conclusion: making sense of participation 303 6.1 Chapter summary 303

References 305

1 Introduction

1 Introduction

This chapter introduces the idea that social science not only describes and understands social worlds, but it also helps to enact them – by seeking to inform public debates and policy making, social science helps to shape the ways in which processes and practices work. The chapter explores this idea by using the example of how social science investigates one particular type of participation; that is, participation as a citizen in elections. Voting is often thought of as a basic form of participation in the political life of the communities where people live. Voting in an election is a means by which people participate in two ways: by making their voice heard, expressing their support for one candidate or party, making demands for certain policies or actions; and also as a way of expressing membership of a political community – a locality or a nation, perhaps – as a civic duty undertaken as a citizen.

Figure 7.1 Political party tellers observing and monitoring the counting of votes cast in the Ceredigion constituency at the General Election, 6 May 2010

This chapter looks at the ways in which social science is used to make sense of two related aspects of elections:

. first, the use of social science to explain why some people vote and why others do not

271

Chapter 7 Enacting elections

. second, the use of social science to try to account for why people vote in the ways they do.

In both of these respects the chapter will show how social science understandings of participation in elections help to inform public debates about politics and shape efforts to enhance participation levels.

Elections are a focus of social science investigations partly because they are, of course, important. They are a means by which governments are chosen, and in turn held to account. They are also one of the largest rituals of participation in the modern world – they are occasions when thousands, hundreds of thousands, or millions of people all engage in the same event, often on the same day, to help decide on who will represent them. Voting in an election is a particular form of participation. In democratic elections people vote in private, but each person’s individual vote is made into an expression of the public’s will by being combined with all the votes cast. In this way, each individual voter becomes part of a collective. By having their own vote counted as part of the overall vote, the individual act of each voter becomes meaningful by being aggregated alongside others. Elections, by enabling people to participate in political processes, are understood as expressions of the ‘popular’ will and one way of measuring public opinion. David Butler, one of the best-known social scientists to have studied elections in the UK, goes so far as to suggest that ‘history used to be marked off by the dates of kings. Now it is marked by the dates of general elections’ (1998, p. 454).

Elections generate lots of data – the results of an election take the form of numerical outcomes, tabulating how many people voted for each candidate or party. The results of an election are therefore determined by the voting decisions of lots of people. But as to what elections mean – why some parties win, why others are defeated, why some candidates are or are not successful, beyond the mere fact of winning or losing – the results of an election don’t tell us this. Social science is one of the ways in which the background processes that led to particular outcomes are explained, interpreted and made meaningful. Social scientists seek to explain why some people vote and others don’t, and why those people who do vote cast their ballots in the way they do. They use social science methodologies such as surveys and interviews to find out these things, and generate theories to make sense of elections.

This chapter looks at the history of social science explanations of voting in general elections in the UK since 1945. These are the elections that

272

1 Introduction

determine the composition of national governments in Britain. The formal study of elections is called psephology, and social scientists who study elections are often called psephologists. The term originates from the Greek word psephos, which means ‘pebble’, because the ancient Greeks used pebbles as ballots to enable them to cast votes to choose a candidate for public office. As you will see in this chapter, elections are studied by social scientists from a number of disciplines – political scientists, sociologists, psychologists, geographers and others. They seek to understand various issues involved in elections: for example, how citizens make their electoral choices and cast their vote for one party rather than another; or how the outcomes of elections might vary depending on just how individual votes are aggregated. You will see that, over time, social scientists have developed different theories as to why people participate in elections, and why those who do vote in the ways they do. And you will see that these different theories can be evaluated in relation to evidence and concepts.

Psephology is the formal study of elections.

The chapter will show how social science informs and shapes public debate about the meaning of elections and voting. The focus is primarily on academic social scientists who investigate electoral processes. But academic social scientists are not the only people trying to make sense of elections. Polling organisations and pollsters seek to map, measure and assess public opinion before, during and after elections. Polling organisations are commissioned, both by parties and especially by news organisations, to take regular measures of public opinion and, using combinations of sophisticated statistical tools and qualitative methods like focus groups, to predict how electors will vote and to try to call the outcome of an election in advance of it happening. So social science plays an important role in the activities of pollsters too. These two forms of social science, academic and non-academic, feed into wider public debates that include pundits and journalists, politicians themselves, and members of the general public – all of whom have a variety of reasons for wanting to have better knowledge of what voters are thinking, who particular segments of the population might be more or less inclined to vote for, or why particular elections produced the outcomes they did. It is as part of this wider public conversation about the meanings and motivations behind voting decisions and election outcomes that social science, as used by both academics and non-academics, helps to enact the meaning of this form of political participation.

273

Chapter 7 Enacting elections

The chapter illustrates the role of social science in describing, understanding and enacting participation in elections by focusing on two sets of questions.

. The first relates to how election analysts seek to uncover and explain levels of electoral participation. Here they ask: who is entitled to vote? And who actually does vote, why and when?

. The second set relates to how social scientists seek to describe who votes for different parties or candidates, and to understand why people chose to vote for the party or candidate which they did vote for.

1.1 Structure of the chapter

Using the UK as a case study, Section 2 outlines how social science considers electoral behaviour, and looks at the generation of data about elections. The data about elections are the basic means through which social science descriptions of political participation are generated. Sections 3 and 4 then address two issues that social science has attempted to explain by collecting evidence, developing concepts and proposing theories. The first issue, examined in Section 3, is the question of why some people don’t vote in elections – the issue of turnout. The second, considered in Section 4, is the question of why people vote in the way they do. This section looks at different theories of electoral behaviour which social scientists have used to explore electoral outcomes in order to better understand them, and perhaps to predict future behaviour based on evaluations of past behaviour. Section 5 elaborates on the way in which social science descriptions and understandings of elections also help to enact social worlds by informing and shaping public debates about politics, participation and public life.

Turnout refers to the number of people who vote in an election as a proportion of the total number of people who are eligible to vote.

1.2 Aims of the chapter

This chapter aims to:

. introduce the social science investigation of elections, and its role within wider public debates about politics and participation

. consider the ways in which social scientists describe and understand non-voting, helping to inform and shape debates about participation

274

1 Introduction

. consider the ways in which social scientists describe and understand the individual motivations and social variables affecting voting decisions and electoral outcomes

. show how social science is used by academics and non-academics to inform and shape public debates about elections, helping to enact social worlds.

275

Chapter 7 Enacting elections

2 Collecting and generating evidence about elections

As we have already seen, one reason that social science investigates elections is because they are important processes. But there is also something about elections that makes them particularly well suited to social science investigation. In one sense, the outcome of any election is normally easy to establish – the results take the form of lots of numerical data.

Table 7.1 shows the results of the general election held in the UK in 2010. As you look at it, consider the different types of data used to describe the outcome of this election.

Table 7.1 describes the results in different ways. For example, the Conservative party gained 10,726,614 votes in total; the Labour party 8,609,527. It also shows that this represented 36.1 per cent and 29 per cent of the total votes cast, respectively. The table also has information about the increase in votes, both in total and as a percentage share of the vote since the previous election in 2005, as well as the numbers of seats won by different parties.

But the table doesn’t tell us much, if anything, about why voters cast their ballots in the way which, when aggregated, generated this overall picture. That is the thing about elections: they produce lots of numbers, but the information provided by these numbers, the election results, is actually rather ‘thin’ – it records only who millions of different individuals voted for, but not why. It doesn’t even tell us who voted for different parties – for instance, how many women voted for the Conservatives, how many middle-class professionals voted for Labour? You can’t tell this from looking at the results of an election on their own.

276

2 Collecting and generating evidence about elections

Table 7.1 The results of the 2010 UK general election

Party Seats Gain Loss Net1 Votes %2 +/-%3

Conservative 307 100 3 +97 10,726,614 36.1 +3.8 Labour 258 3 94 -91 8,609,527 29.0 -6.2 Liberal Democrat 57 8 13 -5 6,836,824 23.0 +1.0 Democratic Unionist Party 8 0 1 -1 168,216 0.6 -0.3 Scottish National Party 6 0 0 0 491,386 1.7 +0.1 Sinn Fein 5 0 0 0 171,942 0.6 -0.1

Plaid Cymru 3 1 0 +1 165,394 0.6 -0.1 Social Democratic & Labour Party 3 0 0 0 110,970 0.4 -0.1 Green 1 1 0 +1 285,616 1.0 -0.1 Alliance Party 1 1 0 +1 42,762 0.1 +0.0 UK Independence Party 0 0 0 0 919,546 3.1 +0.9 British National Party 0 0 0 0 564,331 1.9 +1.2 Ulster Conservatives and Unionists – New Force

0 0 1 -1 102,361 0.3 -0.1

English Democrats 0 0 0 0 64,826 0.2 +0.2 Respect–Unity Coalition 0 0 1 -1 33,251 0.1 -0.1 Traditional Unionist Voice 0 0 0 0 26,300 0.1 Christian Party 0 0 0 0 18,623 0.1 Independent Community and Health Concern

0 0 1 -1 16,150 0.1 +0.0

Trade Unionist and Socialist Coalition 0 0 0 0 12,275 0.0 Scottish Socialist Party 0 0 0 0 3,157 0.0 -0.1 Others 1 1 1 0 321,309 1.1 0.0

1. Change in number of seats from the previous election.

2. Percentage share of the votes cast.

3. Percentage change in share of vote since previous election.

Source: BBC, 2010

Elections are ‘informationally thin’ in this way because voting is undertaken in secret. The secret ballot is actually quite a new practice in the long history of elections, dating back to the nineteenth century. But it is now understood to be a quite basic requirement of democratic elections that people get to cast their ballots in secret, free from observation by or influence from other people. One result of this is that the data that are generated as an output of an election – the results – tell us very little about the inputs that led to those results. The task of social science in describing and understanding elections lies in trying to get behind the results, finding out further facts about who voted and

277

Chapter 7 Enacting elections

who didn’t, and who voted for whom, and, perhaps above all, to try to explain why people voted in the ways they did. And it is by doing this, by filling in the blanks behind election results, that social science helps to make sense of electoral processes as public events.

2.1 Social science and general elections in the UK

The UK has a parliamentary political system. At general elections, its citizens elect Members of Parliament (MPs) to take seats in one part of the legislature, the House of Commons. The government emerges by being in possession of a majority of seats there. So, for example, in 2010 the government was formed by two parties, the Conservatives and the Liberal Democrats, agreeing to set up a coalition that could command a majority in the Commons by combining the seats they had won between them – giving them 364 seats, out of the overall total of 650 seats in the Commons. The prime minister, rather than being directly elected by UK voters, is the party leader who is able to ‘lead’ the Commons majority.

In the UK, general elections are just one sort of election in which citizens are able to participate. For example, a resident of London could, in a period of five years, participate as a voter in elections to choose the following representatives:

. an MP

. a Member of the European Parliament (MEP)

. the London mayor

. a London Assembly member

. local councillors, who are usually elected in multi-member wards.

Meanwhile, an elector in Scotland, Wales or Northern Ireland has the opportunity to vote for:

. an MP

. in Scotland, two Members of the Scottish Parliament (MSPs); in Wales, two Assembly Members (AMs); in Northern Ireland, up to six Members of the Legislative Assembly (MLAs)

. MEPs

. local councillors.

278

2 Collecting and generating evidence about elections

So in the UK voting is a means of participating quite regularly, about once a year, in political life. But how do social scientists go about studying all these elections?

In the UK, the systematic study of elections dates from the period following the end of the Second World War in 1945. This was the year of publication of the first of the Nuffield Election Studies, a series of scholarly works on British general elections. The series was founded by Robert McCallum, the social scientist who popularised the term psephology. For many years, up until 2005, it was co-authored by David Butler, the social scientist specialising in electoral analysis who was quoted above. Over time, Butler became a public intellectual, regularly appearing on television around elections.

Figure 7.2 Psephologist David Butler on election night, 1974

Each Nuffield study explores both the election outcome and the election campaign, and surveys the chronology of the previous Parliament before the election. The first aim of the election studies set up by McCallum and developed by Butler was to be descriptive – to provide a historical record of elections. But its second aim was to be

279

Chapter 7 Enacting elections

explanatory – to answer questions such as ‘why did voters vote as they did?’ (McCallum, 1954). This aim of developing better understandings of elections is a more complex matter. Butler’s particular contribution to the series was the development of rigorous statistical analysis, in which the relevance of different variables in shaping electoral behaviour was assessed. And election studies of this sort also had a third aim – to inform and shape public debate:

Not only is the study of elections valuable for historical record, it may also guide and clarify the ideas of the active politician. It is a time for stocktaking, for reviewing methods and policies. To the extent that an election survey is accurate and objective and informed with some degree of political imagination, the active worker and political journalist may profit by it.

(McCallum, 1954, p. 511)

The Nuffield Election Studies are one example of how social science has sought to describe and understand elections in the UK. Data from surveys, along with data from other sources such as censuses, are used to identify patterns and correlations between electoral outcomes and different variables. Elections studies use information and evidence collected before, during and after elections, as well as across successive elections.

Much of the evidence now used to analyse elections in the UK is produced by the British Election Study (BES). This was first set up by Butler and the political scientist Donald Stokes in preparation for the expected 1964 election. They conducted the first national survey of voters in the early summer of 1963. Two more election surveys were carried out at the elections of 1964 and 1966. This research helped prompt the theories of electoral behaviour discussed in the next two sections of this chapter. The BES is now carried out routinely around each general election.

The BES is based on a large-scale post-election interview survey, which is conducted among a representative sample of voters. Surveys are conducted among the same respondents after successive elections. In this way, the BES generates data about people’s motivations for voting, and about changing patterns of voting over time. The BES is the longest-running survey of electoral behaviour of its kind in the world. It is actually one of the longest-running social science surveys of any kind.

280

2 Collecting and generating evidence about elections

It is funded by the Economic and Social Research Council (ESRC), the public body responsible for funding and promoting social science research in the UK. The goal of the BES is ‘to understand why people vote, and why they vote the way they do’ (Economic and Social Research Council, 2010). Each BES examines long-term trends in British voting behaviour; explains the election outcome, party choice, and turnout; and examines the consequences of elections for the operation of democracy more generally (ibid). Having accumulated data for each general election held over a period of more than four decades since 1964, the BES is an example of social science data being used to facilitate the study and analysis of long-term changes in voting behaviour. Above all, it is used to describe and understand the factors which influence the outcome of elections, and to understand long-term changes in the nature of participation and voter choice. The BES informs further research and analysis conducted by academics, polling organisations and news journalists. In this way, it provides the basic reference point of evidence and data for developing understandings of the dynamics of voter participation and voter choice in the UK. Public debates about the meanings of any given election, in terms of what has changed over time and what factors have influenced outcomes, are fundamentally dependent on the information generated by the BES and other social science data sets. We shall now move on to look at each of these issues in turn, in Sections 3 and 4.

281

Chapter 7 Enacting elections

3 Explaining participation and non­ participation in elections

In a democracy such as the UK, the level of participation in elections is a vital feature of the legitimacy of their outcomes and of the governments formed on the basis of the results. Democracy is meant to be a system in which all the people (or at least a majority of them) have a say in how governments are formed. If only small numbers of people actually vote, then this condition is not being met through elections. Furthermore, if some groups systematically don’t vote, or if fewer of their members vote compared with members of other groups, the inclusive, democratic functions of elections might be called into question. For example, in the UK turnout has tended to decline most markedly over time among young people. Does this constitute a problem, a challenge to the legitimacy of the electoral system?

The issue of voter turnout, then, is vital to the value ascribed to elections as part of democratic government. In most countries voting is a voluntary act, although in some (Australia, for instance) it is compulsory; citizens are expected by law to participate and incur a fine if they do not. Where voting is voluntary, as in the UK, the first step is for individuals to be registered to vote, by being included on an electoral register. In some countries, such as the USA, it is up to the individual to place themselves on the register; in others, such as the UK, local government workers proactively seek to register electors. Voter registration is the first stage at which issues of differential participation in elections arise – some groups tend to be systematically under-represented on electoral registers, for a variety of reasons. For example, social science research shows that relatively mobile population groups – including students and those living in temporary rental accommodation – are less likely to be included on local electoral registers in the UK (Electoral Commission, 2010).

When social scientists mention turnout, they are referring to the number of people who actually do ‘turn out’ to vote in a particular election as a proportion of the total number of people who are eligible to vote. Table 7.2 shows turnout at different types of elections in the UK, excluding Northern Ireland, in the period between 2005 and 2011, including local elections, general elections and elections to the European Parliament.

282

3 Explaining participation and non-participation in elections

As you look at the table, you will see that the figure ranged from a low of 34.5 per cent (in the European Parliament elections of 2009), to a high of 65.1 per cent (in the general election of 2010). Can you identify any correlation between voter turnout and type of election across the table?

283

Table 7.2 Turnout in UK elections, 2005–2011

%

General election 2005 61.3 English locals 2006 36.6 Scottish Parliament 2007 53.9 Welsh Assembly 2007 46.4 English locals 2007 38.3 English and Welsh locals 2008 35.6 London Mayor 2008 45.3 English locals 2009 39.3 European Parliament 2009 34.5 General election 2010 65.1 Scottish Parliament 2011 50.4 Welsh Assembly 2011 41.4 English locals 2011 39.2

Source: adapted from Denver, 2011

One thing social scientists have argued is that turnout rates vary according to different types of elections. Table 7.2 appears to provide some evidence for this. Elections can be categorised as either ‘first­ order’ or ‘second-order’ elections: this is a way of indicating the relative importance to voters of the different elections they might be able to vote in. In the UK, general elections that determine the composition of the House of Commons, and therefore elect the national government, are considered to be first-order elections. They have higher turnouts than other UK elections, such as those for devolved government in Wales and Scotland, for local government and for the European Parliament. These all appear to be second-order elections.

First-order elections such as the general election of 2010 (with a turnout of 65.1 per cent) have a very high public profile. These, being national events, are fiercely contested by the parties and attract overwhelming news media and public attention. We have already noted that the lowest turnout in Table 7.2 was for the much less important European

Chapter 7 Enacting elections

Parliament elections in 2009 (34.5 per cent), and indeed these elections have always had lower turnout rates than other kinds of elections.

So it seems that one thing that affects turnout, or in other words participation in elections, is the relative degree of importance ascribed by people to different elections. Clearly, then, electors do not consider every election to be equally significant. The likelihood of voting increases according to the perceived importance of the body that is being elected.

But participation in elections also varies across social groups at the same types of elections. Some groups of people are more likely to vote than others; and, furthermore, there appears to be evidence of a long­ term decline in the overall level of voter turnout in the UK. How do social scientists seek to understand these processes?

3.1 Explaining turnout in elections

There is a widespread perception, in the UK and elsewhere, that there is a crisis in political participation in liberal representative democracies. One reference point for this perception is the declining levels of turnout in elections. For example, Table 7.3 shows the level of turnout in general elections in the UK since 1945.

Can you identify any trend in turnout at general elections below?

Table 7.3 Turnout in UK general elections, 1945–2010

Year % Year %

1945 72.6 1974 (Oct.) 72.7

1950 83.6 1979 75.9

1951 81.9 1983 72.6

1955 76.7 1987 75.3

1959 78.7 1992 77.7

1964 77.1 1997 71.2

1966 75.8 2001 59.1

1970 71.9 2005 61.3

1974 (Feb.) 78.8 2010 65.1

Source: adapted from Denver, 2011

284

3 Explaining participation and non-participation in elections

You can see that in 1945, turnout was 72.6 per cent. In 2010, however, it was 65.1 per cent. More than 24 million people voted in 1945; slightly fewer than 30 million in 2010. More people voted in the latter year, but turnout was down, because in 2010 a smaller proportion of those people eligible to vote actually did so.

If you look at Table 7.3 a little more closely, you will see that turnout rates at general elections were actually quite steady between 1945 and 1992, averaging 75 per cent, and peaking in 1950. But since then they have plummeted. A post-war low of 59.1 per cent in 2001 was followed by only a slight increase to 61.3 per cent in 2005. Even the 2010 election, a closely fought event where the outcome was always in doubt (a factor which ought to increase turnout), only saw an increase to just 65.1 per cent.

Figure 7.3 A polling place in Glasgow during the 2012 local council elections

So the evidence suggests a complex picture of very high turnouts in 1950 and 1951 followed by a stable trend, only for turnout to slump dramatically in 2001 and then remain below the average for the post­ war period. How have social scientists accounted for this decline, and how do they seek to explain non-voting? We can identify six types of explanation.

285

Chapter 7 Enacting elections

1 The first is that election turnouts reflect the competitive nature of the election. Uncompetitive elections, where it is expected – from polling data – that one party is going to win easily, tend to have lower turnouts than elections where a close outcome is predicted. Many commentators suggest that this explains the fall in turnout in both 1997 and 2001, when Labour, as was long expected, swept the electoral board; and in 2005, when Labour was expected to be re­ elected with a smaller Commons majority. This did not happen, however, in 1983, when the Conservatives won a long-expected substantial majority on a respectable turnout of 72.6 per cent. However, in 2010, at an election in which no single party won an overall majority of seats, the result proved too close to call throughout the election campaign. But turnout rose by only 3.8 per cent over 2005 to 65.1 per cent (still some 10 per cent below the average for 1945–92). Such a fact, for many commentators, calls this explanation into question.

2 The second explanation suggests that the propensity of an individual to vote is a reflection of their perception of the personal and social benefit, or utility, they will derive from voting. Turnout will be greater when electors incur fewer ‘costs’ and receive greater ‘benefits’ from voting. But this explanation actually has trouble explaining why people do vote. On its own assumptions, since each individual vote is highly unlikely to make a material difference to the electoral outcome, it is not clear why any individual would bother to vote at all. So this approach has limited value in explaining why some people do vote and others do not.

These two explanations tend to explain turnout in a narrow way, either focusing on the nature of elections and campaigns, or thinking of voters as individuals who make ‘rational choice’ decisions, similar to the approach you came across in Chapter 4.

3 The third explanation emphasises the social circumstances of individual voters. Social scientists have shown that turnout varies in different constituencies at the same election. For example, in the 2010 general election turnout ranged from just over 44 per cent (in Manchester Central) to almost 78 per cent (in East Renfrewshire). Geographers have demonstrated that Britain has long been divided into constituencies with relatively high turnouts (such as rural or suburban, middle-class and relatively affluent places) and those with lower turnouts (such as urban, working-class and poorer areas) (Johnston and Pattie, 2006).

286

3 Explaining participation and non-participation in elections

These variations reflect differences in the socio-economic circumstances of different people. Levels of education, access to information, having time – all these factors tend to determine people’s participation in civic activities of all sorts, including voting in elections. Because these resources are unevenly distributed across different social groups, this explanation helps to suggest why certain types of people are more likely to vote than others.

In the terms used in Chapter 4, this explanation emphasises the structural factors that shape people’s likelihood of voting.

4 The fourth explanation focuses on the political ‘supply side’, by suggesting that electors respond both favourably and unfavourably to the ways in which parties seeking votes try to engage with and mobilise them. From this perspective, variations in turnout can be explained by the extent to which the parties are perceived as offering significantly different choices (Clarke et al., 2004, pp. 261–74). Voters are more likely to vote when they think that very different policies would be pursued by the different parties, especially when the outcome of a close election is considered to be in doubt.

None of these four explanations helps, however, to completely account for the long-term trend of declining voter turnout.

5 A fifth explanation suggests that declining turnout is because feelings of apathy and alienation play an important role in people being either not motivated enough to vote, or alternatively actively motivated to not vote – in other words, to abstain. Electoral apathy is as old as the hills; bear in mind that even back in the ‘good old days’, when some 75 per cent of electors turned out to vote – as was the case between 1945 and 1992 – this means that around 25 per cent did not. Compared with apathy, voter alienation can be seen to be both positive and assertive. Alienation – if reflected in turnout falling below the established average – testifies to the fact that more electors than ever before respond unfavourably to the ways in which parties seeking votes try to engage with and mobilise them. Dealignment is a

concept used in political science to describe the trend whereby voters abandon previous attachments and affiliations to particular parties.

287

6 A sixth, related explanation holds that there is a declining association between voters and political parties. This phenomenon, known as ‘partisan dealignment’, refers to the fact that electors are far less likely than in the past to vote automatically and relatively unthinkingly for a particular party. If you strongly support a party you will be more likely to turn out to vote, because partisans are more likely to vote than those with no alignment. Given that voter

Chapter 7 Enacting elections

partisanship has been declining for some time, this argument can help explain why turnout has trended downwards over a long period of time. However, the evidence demonstrates that while the strength of electoral partisanship declined from the mid-1970s onwards, turnout did not begin to fall until 1997 and fell markedly only in 2001, thereafter rising somewhat in 2005 and 2010. This explanation is related to the theory of valence-based voting, discussed further in Section 4.3.

Valence-based voting is a model of voting which suggests that the strongest factor influencing voter choice is judgements about the competence and performance of politicians.

You can see, then, that the explanations that social scientists have presented to account for declining electoral participation and falling turnout are quite varied. Some focus on specifically electoral issues, some on broader issues of political behaviour, some more broadly on patterns of social difference as a whole. Different approaches support different programmes to address the matter. In Section 5 we shall return to these issues to see how the different approaches to understanding turnout can inform attempts to enhance participation and reverse the decline in voter turnout.

In this section we have seen how social science attempts to use concepts of different sorts to explain why people do or do not vote at elections, and how these explanations are evaluated in relation to the evidence on turnout that is available. In the next section we move on to see how social science seeks to explain why people who do vote, do so in the particular ways they do.

288

4 Social science theories of voting decisions

4 Social science theories of voting decisions

As with voter turnout, discussed in Section 3, social scientists have developed different theories to understand why people vote as they do in elections. In this section we look at three influential explanations, or theories, of electoral behaviour and examine the forms of evidence and evaluation that are used to support or criticise them.

We have already seen that the formal and systematic study of elections in social science is a relatively new endeavour. Only since the 1940s have social scientists had the means to collect and analyse survey data, and to make use of other aggregate data from censuses or other databases, in order to develop theories of voter decision making. For example, the sort of data generated by social science research programmes such as the BES enable social scientists to do two things:

1 to describe voting patterns, by establishing who voted for which party and by making comparisons over time about the changing nature of those who vote for different parties

2 having described the voters who support different candidates or parties, to develop explanations of why people vote as they do, and what influences their voting decisions.

In pursuing these two aims, social scientists build models that help them to hypothesise different causal relationships between variables and voting decisions. The explanatory theories that social scientists develop enable the conceptual exploration of the phenomenon under study, allowing the better understanding of the phenomenon that they have described, and also making it possible for them to predict future behaviour based on their evaluation of past behaviour.

In the following sections, we look in turn at three different theories which have been developed to explain voting decisions:

. a sociological theory, which examines the relationship between social characteristics such as class and voting

. a social-psychological theory, which examines the relationship between voters’ partisan self-images and voting, and is sometimes known as the party-identification theory

. an issue-based voting theory, also known as the valence voting model, which suggests that voters make their choices based on their

289

Chapter 7 Enacting elections

perception of issues, party appeal and, increasingly, their estimation of party leaders.

These three theories have been used to better understand and predict electoral outcomes. They have formed the basis both of social science analysis of elections, and of the ways in which wider public debates frame the meanings of electoral results and the process of electoral politics. But as we look at them in turn, we shall also see that each theory can be subjected to evaluation in terms of its conceptual coherence, its empirical adequacy, and the degree to which it is comprehensive in addressing all possible factors and variations.

4.1 The class-based theory: a sociological explanation of voting

In election studies in the UK after 1945, the predominant theory of voting was a sociological model that suggested that class-based variables were the most important factors shaping electoral behaviour. In this sort of theory, electoral choice was understood as a function of a voter’s individual position within various social structures – the class position of their parents, or their own occupational status, level of educational attainment, or form of housing tenure. On the basis of these factors, it was said, people develop long-term identifications with particular ideologies represented by political parties.

In the post-war period in the UK, this sort of explanation appeared to have some validity. It suggested that by being working class, an elector would more likely than not vote Labour; by being middle class, the elector was more likely to vote Conservative. In the two-party system that dominated in this period, relatively few votes were cast for parties other than Labour or Conservative. The theory of class-based voting was championed by Butler and Stokes (1974), who argued that the links between class and partisanship – that is, strong identification with a particular party – were strongly supported by the evidence. Class voting, perhaps tinged with other forms of social identity, was the dominant electoral theory in the UK from the 1940s through to the 1960s. It led the political scientist Peter Pultzer to suggest that ‘class is the basis of British party politics; all else is embellishment and detail’ (1967, p. 98). The class-based theory of voting has three core ideas.

The first is that relatively fixed social characteristics, such as social class, work through long-term socialisation processes to predispose

290

4 Social science theories of voting decisions

individuals to support one party rather than another. Second, the results of these socialisation processes can be either reinforced or undermined by the individual’s social context – for example, by the kind of work the individual does, the sort of neighbourhood he or she lives in, and the informal groups to which he or she belongs. The third idea is that social characteristics and social contexts produce a distinctive set of affiliations and senses of solidarity around parties. Over time, and again largely as a result of socialisation, most individuals tend to develop stable and enduring attachments to particular political parties. These party identifications serve both to colour the ways in which individuals interpret political information and to predispose them to vote for the same party in successive elections.

(Clarke et al., 2004, p. 5)

This theory explains voting primarily with reference to structural factors – people’s positions within various social structures lead to stable identifications that are reflected in their electoral choices:

The approach never claimed that all variation in voting preferences would be determined by social structure. Other contingent factors, such as the condition of the economy, specific issues, perceptions of party leaders, or events, would cause some people to switch their votes between elections. But, most people would continue to vote according to their structural positions and their party identifications and this, in turn, would be associated with relative electoral stability.

(Clarke et al., 2004, p. 5)

In later theories, the relationship between people’s structural position, the specific contexts in which they live (geographical location, for example) and party identification is understood to be much more fluid than in this sociological theory. Over time, evidence showed, for example, that significant numbers of working-class people did vote for the Conservative party, and that the Labour party drew support from significant sectors of the professional middle classes. In terms of the theory, these were ‘deviant voters’, because their behaviour deviated from the pattern expected by the voting theory. The predictive ability of

291

Chapter 7 Enacting elections

the theory came increasingly to be questioned, and it was also pointed out that the model lacked widespread application – it seemed very much shaped by the UK pattern of class, whereas in other parts of Europe, for example, religious affiliation has often been a predominant factor shaping electoral behaviour.

The decline of the class-based theory reflected, in part, changes in the social structure itself, as the traditional industrial working class shrank, and white-collar and service sector employment grew. As the nature of class in the UK changed, so too did the idea that class completely underpinned electoral behaviour. Social scientists describe this process as ‘class dealignment’, whereby both the form and the nature of social class changed at the same time as class failed to retain the close association with voter choice which social scientists had previously suggested that it had.

4.2 The party-identification theory: from a sociological to a social-psychological explanation of voting

As we have just seen, the primary claim of the sociological theory of class-based voting is that the relationship between class and party identification is a predictable and a stable one. The notion that electors identified with a certain party was seen to be the explanation for the fact that significant numbers of voters consistently voted for the same party across several elections. Voters with a weak or no party attachment were considered to be among the ‘swing’ voters whose votes would decide a tight election between the two main competing parties, Labour and Conservative. In this theory, it was argued that:

the voter’s choice of party was largely based on class and party identification … Most people were brought up to think of themselves as supporters of one party or another (with upwards of 40 per cent of the electorate describing themselves as very strong party supporters) and thus developed an enduring loyalty to a party which they would support, for the most part, through thick and thin. The emphasis in this explanation is on stability – class and party identity didn’t change much and so most people always

292

4 Social science theories of voting decisions

voted for the same party – rather than volatility. Electoral change would be slow and gradual.

(Denver, 2011, p. 86)

The validity of this theory was called into question in a two-step process. First, by the late 1970s, in spite of some residual linkage, social class was no longer a very good predictor of party support. The process of class dealignment, whereby the link between social class and voting began to fracture, was first noted in the early 1970s and continued unabated in the 1980s and 1990s (Clarke et al., 2009).

From the 1970s onwards electoral analysts began to play down the explanatory power of sociological factors like class structure. Instead, a new theory was advanced, one which argued that people were linked to parties for reasons other than class identity. This social-psychological theory still held that there was a firm and enduring linkage between electors and a particular political party – it held to the idea that party identification was the key factor in electoral behaviour. But it suggested that this identification was shaped by other factors rather than social class alone.

The development of this theory reflected in part a response to evidence, but also a conceptual shift in focus. Rather than assuming that a person’s social position (in the labour market or housing market, for example) would determine their voting behaviour, this approach sought to explain why individual party identifications, where they existed, proved so resilient over successive elections. It was suggested that, once established, a person’s identification with a particular party formed a stable and enduring emotional or psychological bond. Party identification therefore sought to explain why electors had internalised a strong, stable and long-term attachment to one party rather than another. It also explained why an elector would consistently vote for that party, because it posited that they were committed to what was described as their ‘normal vote’. Casting a vote was, in this view, akin to supporting a football team: someone formed an attachment and it was rare for them to abandon this attachment and change allegiance, even when their ‘structural’ circumstances changed.

In this account, then, a particular conceptualisation of the importance of personal, psychological identity is considered the key explanatory factor in determining people’s voting behaviour over time.

293

Chapter 7 Enacting elections

Second, having weakened the link between class and voting by deepening the understanding of the process of party identification, social scientists were subsequently led to revise this new social­ psychological theory in light of further evidence which suggested that, in fact, party identification was lessening in importance over time. Electoral surveys showed that party identification declined markedly from 1964 onwards. Writing in 1974, Butler and Stokes found that 45 per cent of electors ‘strongly identified’ with either the Conservative party, the Labour party, or, to a lesser extent, the Liberal party. And they found that these identifications had remained relatively stable over several election cycles and would usually harden with age. However, by the late 1990s it was found that only 9 per cent of electors would consider themselves to be ‘strong party identifiers’ (Clarke et al., 2009). So, if the social-psychological theory had weakened the link between class and identification, in turn the importance of party identification based on internalised attachment was called into question.

Social scientists now tend to view the UK electorate as decoupled from class and partisan bases, and consequently as having become more volatile and unstable – more difficult for politicians to take for granted, and more difficult for social scientists to predict. As a result electoral behaviour is no longer consistent and stable. Voter support is now often considered to be less a form of partisanship and more an expression of temporary association. Today, far fewer voters can be classified as strong partisans, firmly committed to supporting one party over another at every election. And it was strong, stable partisans, not weak occasional ones, that lay at the heart of the sociological and social­ psychological models of voting behaviour. Over time, then, the number of floating voters, those unattached to a specific party by either social class or partisanship, has been seen to have substantially increased. This is in part a shift driven by the generation of new data and evidence about voter behaviour and characteristics. But it is also a conceptual shift – away from structural explanations of voting, towards theories which ascribe much more agency to voters in explaining electoral behaviour. This shift is indicated in the third theory discussed in this section, known as the ‘valence-politics’ theory of voting.

4.3 The valence-politics theory: an issue-based explanation of voting

Valence politics refers to the idea that voters make their decisions about who to support based on their evaluative judgement of the competence

294

4 Social science theories of voting decisions

and potential competence of different parties. The development of this approach reflects a shift away from theories in which electors and parties are linked together by strong ties of attachment, towards a theory in which voters make electoral decisions based on how they, their families and perhaps their neighbourhoods are affected by major contemporary conditions and their assessment of how successful or not the government has been in either creating or tackling those conditions. This theory, one in which people vote on issue-based grounds, argues that voters are more likely to vote for a party of government when it has been economically successful and has been able to deliver prosperity to them as individuals.

In this theory, it is assumed that people broadly agree about what the government should be elected to do: it should effectively and carefully manage the economy; secure economic growth; achieve and maintain low rates of inflation and unemployment; ensure national and local security and the rule of law; and provide a wide range of well-funded, efficient and effective public services: ‘Voters, because they focus on such “valence issues”, vote for the political party (but also, increasingly, the party leader or prime minister) that they consider can do the job of government best and so better “deliver the mix of policy outcomes that are widely seen as good things”’ (Clarke et al., 2009, p. 5).

It has been suggested that this theory focuses too narrowly on economic issues, neglecting other key issues of importance to electors, and therefore to parties seeking votes. Nevertheless, the idea that elections are determined by valence politics – by voters’ judgements about issues and competence – is now widely accepted in electoral studies.

The valence-politics approach represents a significant shift in how electoral politics are conceptualised by social scientists. It suggests that elections are primarily determined not by ‘position’ issues, on which different people are likely to have very different opinions, but by issues where people tend to be in agreement with regard to the overall aim – issues like crime, or corruption, or economic growth. These latter are valence issues because their importance is agreed on – so the key factor in voter choice becomes a matter of judging which party is best equipped, most competent, to actually deliver the sought-after ends. The idea of valence voting has strong similarities with the theory of individualisation discussed in Chapter 4. Both argue that agents increasingly make judgements and choices that diverge from the

295

Chapter 7 Enacting elections

predictable patterns once associated with traditional positions of class, gender, occupation, or other structures of social relations.

From this perspective, voters support one party rather than another because of the evaluation they make of a party’s ability – or that of its leader – to deliver on the policies they value. They can also punish those parties – usually the party in government – that have not delivered on these policies. Voter identification can be either positive or negative. And it can, obviously, change over time in response to political, economic and social trends.

The emergence of this theory is in part a reflection of a change in how social scientists conceptualise voting behaviour. As the geographers Charles Pattie and Ron Johnston observe, a valence-politics approach ‘radically reconceptualises party identification, seeing it not as long term psychological attachment to a party, but as a running tally of a government’s performance’ (2009, p. 468). According to this theory, governments lose elections when electors feel that their delivery has not been satisfactory and when the opposition is considered to offer a credible alternative – when ‘potential’ wins over ‘performance’ because the government’s ‘performance’ is considered by enough voters to be inadequate.

One important implication of the shift to theories of valence politics is the argument that a key determinant in voter behaviour is the assessment of party leaders:

Leadership is a valence issue in that few of us would want to elect a prime minister with no leadership skills. Perceptions of leadership competence have independent effects on party choice, even after controlling for factors such as underlying party identification and evaluations of the economy. Leaders matter, and can be an electoral asset for their party (as [Tony] Blair was for Labour in 1997) or a liability (as, arguably, Gordon Brown rapidly became [for Labour] in early 2008). Other valence issues of importance in influencing party choice include voters’ evaluation of competence in handling major public services such as the NHS and education, benefits such as pensions, and public problems such as crime and security.

(Pattie and Johnson, 2009, p. 470)

296

4 Social science theories of voting decisions

Figure 7.4 In the run-up to the 2010 UK general election, the leaders of the three main parties appeared in a televised debate for the first time

The valence-politics theory thus posits a significant shift away from fixed or strong identifications towards active evaluation as a core consideration in how electors make choices and cast their votes. In turn, this means that election campaigns matter more than ever. Social science descriptions and understandings of the increasingly fluid and unpredictable nature of the voting process have helped inform a shift in the broader public understanding of how elections are won and lost. This is a key form of social science enactment – evidencing change over time and influencing the behaviour of parties seeking votes, who increasingly recognise that they have to ceaselessly campaign for votes by emphasising their appeal and criticising that of their opponents (see Figure 7.4).

4.4 Description and understanding in the evaluation of theories of voting

We have seen how social science has shifted from one theory of voting behaviour to another, depending on the degree to which a particular theory is able to predict or explain electoral outcomes and patterns of

297

Chapter 7 Enacting elections

change across successive elections. See Table 7.4 for a summary of the main features of each of the three theories of voting discussed here.

Table 7.4 Three social science theories of voting

Theory of action

Basis of voting decision

Nature of voter– party relationship

Strength of attachment over time

Sociological theory

Structural explanation

Rational utility – class interests

Partisan Stable and predictable

Social­ psychological theory

Structure and agency

Emotional attachment – identification with party

Partisan Stable and predictable

Valence-politics theory

Agency Judgement of competence – issue­ based

Dealigned, weakened partisanship

Fluid, temporary, unpredictable

You can see that the three theories can be differentiated along four dimensions:

. the theory of action which underlies their accounts of voting

. the basis of voting decisions they each identify

. the nature of the relationship between voters and parties

. the assumptions made about the strength of the attachment between voters and parties over time.

The shifts from one approach to another have not been driven simply by testing theories against evidence. The changes are also related to important conceptual shifts. For example, consider this quote from Johnston and Pattie, describing the view of voter behaviour that is now widely accepted under the valence-politics approach:

Voters are increasingly open to alternative claims for their support, and identification with individual parties has become weaker, a process called partisan dealignment. Their decisions are largely influenced by evaluations of current conditions (dominantly but not solely economic) and the perceived performance of parties and candidates in government, by their expectations regarding the future, and by their assessments of the potential governance offered by rivals.

(Johnston and Pattie, 2004, p. 47)

298

4 Social science theories of voting decisions

One thing to notice about this description of valence-based voting is that it still assumes, like the class-based theory which was ascendant previously, that economic factors play a crucial role in explaining electoral outcomes. But the difference lies in how these are conceptualised. In the class-based approach, the assumption is that people’s economic interests are determined structurally, by their position within stable patterns of class, employment, housing and education. In the valence-politics approach, economic factors are considered to be just as important, but they are now understood in terms of ‘issues’ in relation to which voters make judgements regarding parties’ relative performance and competence.

So whereas class-based and party-identification theories both emphasise voters’ stable structural position and associated loyalties, valence theories see voting as a much more fluid process, with voters making short-term judgements about past performance and future potential. This shift in theory, then, is partly due to empirical shifts in voter behaviour – for example, the observable decline of strong, partisan identification in the UK since the 1970s. And it is partly to do with how these new descriptions of voter behaviour have in turn led to new theories of the agency of voters.

299

Chapter 7 Enacting elections

5 Enacting elections: informing and shaping public debate

In Sections 3 and 4 we have seen how social science investigations of elections generate descriptions and understandings of voting that fill in, as it were, the blank spaces that the raw materials of an election result produce. By digging beneath election results, comparing electoral data with other data sets, conducting surveys and doing interviews, social scientists seek to make sense of elections and voting decisions – to make sense of them as indicative of changing expressions of the popular will, or as reflections of changing patterns of socio-economic division, or as indicative of new fissures or identities.

Even in this simple sense, then, social science helps to enact elections – social science allows election results to be placed into narratives, not least by providing comparative and historical perspective. This is just one way in which social science informs and shapes how elections are thought about and conducted. Social science research is used, for example, to inform campaigns by parties to better target those people most likely to vote for them. But it’s also used in campaigns to increase participation – or voter turnout – across the board, as a generally agreed civic duty, by helping to target groups who are highly represented among non-voters. And as already suggested, it is not only academics who use social science to make sense of, and to describe and understand, elections – it is also used by non-academic polling organisations, by journalists and by pundits, and social science in turn informs and shapes public debates about the meanings of elections and of political processes more broadly.

Social science analysis of voting and elections is interesting not least because it helps us see that the enactment of social worlds does not have to take the form of direct influence on policy making. There is more to ‘the uses of social science’ than having a direct, observable ‘impact’. Electoral studies demonstrate a more diffuse contribution to public debates. For instance, by evaluating the formal theories of electoral behaviour discussed above, social scientists have demonstrated that political parties in the UK now face two significant challenges:

. the changes in the ‘sense of belonging’ among established electorates, from which parties draw support – something reflected in declining levels of party identification

300

5 Enacting elections: informing and shaping public debate

. the rise of evaluative judgemental voting based on valence politics – which obliges parties to pay ever closer attention to leaders, images and issues.

These changes reflect the conditionality with which voters now cast their ballots. The idea that issue-based competition has become more important is based in part on social science analysis of evidence; but it also relates to a shift in how voting behaviour is conceptualised. The idea that voter evaluations of party policies and of the quality of party leaders play an ever larger part in voter choice has become widely accepted – not only among social scientists, but among politicians, pollsters and pundits, and among voters themselves (Heffernan, 2009).

Social scientists are, then, part of a broader public conversation about politics, voting and election. Sometimes they inform and shape public debates indirectly – by influencing other expert and elite actors, such as journalists or policy makers. Social scientists often act as experts who are consulted by government and are invited to serve on commissions of inquiry. Election experts, for instance, are often asked by government to advise on ways in which voting may be made easier.

But sometimes social science can have a very visible public role in debates about politics. This is nowhere more the case than in the field of electoral studies. Perhaps surprisingly, given the sometimes highly technical, statistical nature of social science research on voting, psephologists have been among the best-known public intellectuals in British cultural life. We have already seen how David Butler helped to pioneer the study of elections in Britain, through the Nuffield Election Studies and then the BES. But from 1950 to 1979 Butler was also the public face of academic electoral studies, along with another academic, Robert MacKenzie. These two became crucial parts of the BBC’s election broadcasting as it pioneered the televised coverage and analysis of elections. They, and other academics, served as expert analysts, providing in-depth analysis and historical context, and identifying trends as results came in on election night. This type of expert analysis has now become a staple feature of news coverage of elections, but Butler and MacKenzie pioneered innovative ways of making their complex analysis and predictions accessible to the general public. Most famously, MacKenzie is associated with the ‘invention’ of the swingometer (see Figure 7.5), a simple graphic device that represented how a given swing in the percentage of votes cast for one party or another would translate into shifts in the number of seats held by each party in Parliament (BBC, 2011a; 2011b).

301

Chapter 7 Enacting elections

Figure 7.5 MacKenzie with the swingometer on election night, 1964

Butler and MacKenzie’s role as ‘media dons’ is a well-known example of a much more routine way in which social science enacts social worlds as well as describing and understanding them – by communicating their findings, interpretations and explanations in the public sphere. Writing in 2001, Butler himself helps us see this dimension of enactment, in a reflection on his own career in broadcast coverage of UK elections. Having studied each election since 1945, Butler observes that elections ‘have been transformed in the past half-century’. He goes on to suggest that ‘The real turning point came in 1959, with the coming of television and opinion polls. We’ve now moved to 24-hour electioneering from stately middle-of-the-day events’ (BBC, 2001).

Here Butler encapsulates two of the ways in which social science has informed and even shaped how elections are organised and made sense of – through the widespread use of social science methods such as opinion-polling, used by academics, news organisations and campaigns; but also by noting the importance of television in reshaping how elections are organised. Because if television has played an important role in shaping elections, then social scientists like Butler, MacKenzie and others have themselves been important actors in shaping how television has transformed the meaning and conduct of elections.

302

6 Conclusion: making sense of participation

6 Conclusion: making sense of participation

In this chapter we have seen how social science seeks to describe voting behaviour at elections through the generation of data about who votes and who they vote for, and also how it seeks to explain why some people vote and others don’t. You have seen that in relation both to turnout and to voting decisions, social science theories are evaluated in relation to evidence, but also in terms of changing concepts of what counts as a plausible explanation. For example, the shift from theories of class-based or social-psychological voting to more issue-based theories is in part a response to evidence of changing patterns of voter behaviour and identification. But it’s also a result of a conceptual shift away from theories focusing on stable, structural factors, towards theories which conceptualise voting in much more fluid and more agency-oriented ways.

This chapter has also shown how social scientists’ analysis of voting helps to inform and shape public debates about elections and politics more broadly. It does so in a variety of ways. Social science descriptions and understandings inform policy making and policy reviews around issues of improving electoral registration, increasing turnout, considering electoral reform and other matters. Social science is integral to how news organisations report elections and how parties devise strategies to win them. And electoral studies has, perhaps surprisingly, successfully made itself one of the most visible examples of social science in the public sphere – as expert analysts on television or radio or in newspapers, social scientists are routinely asked to predict, explain and interpret results, trends and patterns. If voting is an important way of participating in social worlds, then social science has a key role in informing and shaping the terms in which such participation is practised.

6.1 Chapter summary

In this chapter you have learned that:

. voting is a basic form of participation, in terms of expressing both partisan demands and shared identity as citizens

303

Chapter 7 Enacting elections

. social science generates data and evidence about voting and elections, in order to make sense of the large amount of ‘thin’ information that election results provide

. social science addresses the question of participation in elections through analysis of issues of voter turnout

. various explanations of turnout can be evaluated in terms of their empirical adequacy, conceptual coherence and scope of application

. social science explanations of why people vote as they do have shifted over time, from more structural approaches towards more agency-centred approaches

. the rise of valence-based theories of voting behaviour reflects both responses to new evidence about voting, and conceptual shifts in how voting is understood

. social science enacts voting and elections in various ways: by making sense of complex electoral outcomes; by providing expert opinion; and by contributing to public conversations about the meaning of elections, the determinants of voting decisions, and the trends of changes over time.

304

References

References

BBC (2001) ‘Sultan of swingometers’, 7 June [online], http://news.bbc.co.uk/1/hi/uk/1369188.stm (Accessed 28 March 2012).

BBC (2010) ‘Election 2010’ [online], http://news.bbc.co.uk/1/shared/election2010/results/ (Accessed 28 March 2012).

BBC (2011a) ‘Profile of David Butler’ [online], http://www.bbc.co.uk/archive/swingometer/11203.shtml (Accessed 9 January 2012).

BBC (2011b) ‘MacKenzie’s lore: the king of the swingometer’ [online], http://www.bbc.co.uk/archive/swingometer/11204.shtml (Accessed 9 January 2012).

Butler, D. (1998) ‘Reflections on British elections and their study’, Annual Review of Political Science, vol. 1, no. 2, pp. 451–64.

Butler, D. and Stokes, D. (1974) Political Change in Britain: Forces Shaping Electoral Choice, London, Macmillan.

Clarke, H., Sanders, D., Stewart, M. and Whiteley, P. (2004) Political Choice in Britain, Oxford, Oxford University Press.

Clarke, H., Sanders, D., Stewart, M. and Whiteley, P. (2009) Performance Politics and the British Voter, Cambridge, Cambridge University Press.

Denver, D. (2011) ‘Elections and voting’ in Heffernan, R., Cowley, P. and Hay, C. (eds) Developments in British Politics 9, London, Palgrave Macmillan.

Economic and Social Research Council (2010) ‘British election study’ [online], http://www.esrc.ac.uk/_images/BES_tcm8-2690.pdf (Accessed 16 November 2011).

Electoral Commission (2010) The Completeness and Accuracy of Electoral Registers in Great Britain, London, Electoral Commission.

Heffernan, R. (2009) ‘Political parties’ in Flinders, M., Gamble, A., Hay, C. and Kenny, M. (eds) The Oxford Handbook of British Politics, Oxford, Oxford University Press.

Johnston, R. and Pattie, C. (2004) ‘Electoral geography in electoral studies: putting voters in their place’ in Barnett, C. and Low, M. (eds) Spaces of Democracy, London, Sage.

Johnston, R. and Pattie, C. (2006) Putting Voters in their Place: Geography and Elections in Great Britain, Oxford, Oxford University Press.

McCallum, R. (1954) ‘The study of psephology’, Parliamentary Affairs, vol. 8, no. 4, pp. 508–13.

305

Chapter 7 Enacting elections

Pattie, C. and Johnston, R. (2009) ‘Voting and identity’ in Flinders, M., Gamble, A., Hay, C. and Kenny, M. (eds) The Oxford Handbook of British Politics, Oxford, Oxford University Press.

Pultzer, P.G. (1967) Political Representation and Elections in Britain, London, Allen and Unwin.

306