Government and Politics


Valence and volatility

Do voters choose a party because of the party leader, because of their allegiance to the party or because of the party’s particular policies for achieving general goals for the good of society? David Denver looks into the results of opinion polls to see how voters make their party choices.

Local elections and parliamentary by-elections still have the capacity to excite considerable media interest and attract the attention of the wider electorate. This was seen in May 2008 when the Conservatives captured the Crewe and Nantwich constituency from Labour on a swing of 17.6%, Boris Johnson toppled Ken Livingstone from his position as London mayor, and the Conservatives led Labour by 44% to 24% in the local elections outside London.

Trends in opinion polls

Important as these events were, however, they were time-specific and geographically specific. For those with an ongoing interest in politics and the fortunes of the parties the monthly opinion polls provide a more regular and reliable picture of support for the parties across the country as a whole. Trends in support from the 2005 general election to June 2008, shown in Figure 1, are based on the monthly average figures for voting intentions in all published polls. The figure tells a fascinating story. After the general election Labour received a short-term fillip in the polls. Support for the party that wins a general election normally increases immediately afterwards. Indeed, the pattern is so well established that it is known as the ‘post-election honeymoon’.

Within a few months support for Labour began to decline, and this decline continued more or less steadily for almost 2 years. During this period, in December 2005, David Cameron became Conservative leader. Soon afterwards, for the first time in many years, the Tories established a clear and sustained lead over Labour in voting intentions.

In June 2007, however, things changed. Gordon Brown became prime minister and Labour’s popularity soared, but this sharp improvement was shortlived. In the last few months of 2007 support dropped back. Matters grew worse in 2008. Following a poorly received budget in March and the disastrous elections mentioned above, support for the government dropped like a stone, giving Cameron’s Conservatives a huge lead.

The trends for the Liberal Democrats and ‘others’ are interesting too. The popularity of the former declined quickly after the 2005 election (as it usually does after elections), then wobbled about a bit before declining further during 2007. In December of that year Nick Clegg replaced Menzies Campbell as party leader, and the party recovered a little (though still scoring less well than it had done in the previous election). For most of the period ‘others’ (including the SNP Plaid Cymru, the Green party, UKIP and BNP) hovered around 10% of voting intentions. Ten years previously this level of sustained support would have been unimaginable.

Voter volatility

Although the level of support for ‘others’ is unusual, there is nothing particularly new about voter volatility between elections. Indeed, over the years observers have come to expect a regular cycle between elections in which the governing party loses a lot of support only to recover as the next election approaches. Specialists in ‘economic voting’ were able to relate the ups and downs of support for the parties to the performance of the economy or to voters’ expectations about the economic outlook.

Since 1992, however, the cycles of popularity between elections have not conformed to expectations. As the 1997 election approached there was no upsurge in support for the incumbent Conservatives, even though the economy was improving. Between 1997 and 2001 Labour never once lost its lead over the Conservatives in the monthly averages. After the 2001 election Labour support declined somewhat and did not recover very much in the run-up to the 2005 election. In all these cases economic performance and the level of economic optimism among voters was only weakly related to patterns of party support.

It is apparent that volatile inter-election trends cast doubt on the classic explanation of party choice offered in the 1960s by David Butler and Donald Stokes (Butler and Stokes 1974) — that party choice was largely based on class and party identification. Most working-class people voted Labour and most middle-class people voted Conservative. This was bolstered by the fact that most people thought of themselves as party supporters, with upwards of 40% of the electorate describing themselves as very strong party supporters. The emphasis in this explanation is on stability — most people always voted for the same party — rather than volatility. Electoral change would be slow and gradual.

Valence politics

If the Butler and Stokes model is now inadequate, how are we to explain party choice at the start of the twenty-first century? The short answer is ‘valence politics’ (Clarke et al. 2004), but the meaning of this is not immediately apparent and requires some explanation. In their earlier work, when discussing the impact of voters’ issue opinions on their choice of party, Butler and Stokes made a distinction between what they called ‘position’ issues and ‘valence’ issues. Position issues are those on which people can take positions (for or against public ownership of industries, for example). Valence issues are those on which nearly everyone takes the same side (against crime, for a strong economy, for example). Politics, it is argued, is increasingly about valence issues — the differences between the parties on position issues have become relatively small. For voters deciding which party to support, then, the question is not which party takes ideological or policy positions that they share but which party is likely to be most competent at achieving the goals that are widely shared (such as reduced crime, low inflation, a well-run health service). Making judgements like this is not straightforward and demands some knowledge of what parties might do or have done in the various policy areas. Voters, therefore, tend to use a convenient short cut. They make judgements about the party leaders. This is a much simpler task. We don’t need to know much about policies or politics to decide whether we like or dislike the party leaders that we see often enough on television. We can all have opinions about people without necessarily knowing very much about them.

Party leaders

The key to party popularity nowadays, therefore, is the popularity of the party leaders. It is worth noting that this formulation is the exact opposite of what the Butler–Stokes model implies. It used to be that people’s party identification determined what they thought about party leaders. Labour supporters liked Labour leaders and disliked Tory leaders, and vice versa for Conservatives. Now that party identification has withered, evaluations of party leaders seem to influence strongly the voters’ choice of party. Unlike voting based on a class or a party identification largely inherited through the family, evaluations of leaders are subject to rapid change. Prime Minister Thatcher was thought by the voters for a long time to be ‘resolute’ and ‘determined’. As she became unpopular she was seen more as ‘stubborn’ and ‘inflexible’. Tony Blair was initially considered sincere and trustworthy, but by the end of his premiership large majorities of the electorate thought that he had lost touch with ordinary people and could not be trusted. It is no coincidence, then, that in discussing the trends in party support from 2005, I referred to changes at the top of the parties. Menzies Campbell was ditched by the Liberal Democrats because the party was languishing in the polls. The election of Nick Clegg appeared to bring about a slight improvement, initially at least. David Cameron’s predecessors as Conservative leader — Michael Howard and, before him, Iain Duncan Smith — had been unable to make much of an impression on the electorate. Cameron has been more successful, and under his leadership the Conservatives have generally received better poll ratings. When Gordon Brown became prime minister the change in Labour’s fortunes was immediate.

Brown versus Cameron

Figure 2 illustrates how closely the level of support for the major parties is associated with evaluations of party leaders. It shows Labour’s ‘lead’ (which is mostly negative) over the Conservatives in voting intentions from March 2006 and also the lead that Blair and Brown had over Cameron in being thought the best person for prime minister. (Unfortunately, there are no ‘best prime minister’ figures for October to December 2007, because the Liberal Democrats were electing a new leader.)

For most of the period the prime minister was more popular than his party and this was especially true of Gordon Brown in the summer of 2007. In the spring and early summer of 2008, however, Brown was even less popular than his party. More generally, however, it is striking how the two trend lines track one another. The party that people intend to vote for is strongly related to the electorate’s judgements about which party leader would make the best prime minister.

Similar patterns can be observed if we look at evaluations of the party leaders individually. As an example, Figure 3 shows satisfaction with how the leader of the opposition is doing his job compared to the Conservative share of voting intentions. Levels of satisfaction shot up when Cameron replaced Howard, and Conservative support also increased, although much more modestly. Overall, opinions about Cameron are clearly much more volatile than support for the party. Cameron had a particularly bad spell in the late summer and autumn of 2007 after getting much negative publicity for going on a foreign visit while severe flooding affected some parts of Britain. Nonetheless, the fortunes of the party appear to be strongly related to how the public evaluates the performance of the leader. The fact that in both graphs the line measuring party support and the line measuring reactions to leaders tend to go up and down together is not proof that one (judgements about who would be the best prime minister or satisfaction with David Cameron) causes the other (decisions about which party to support). The two matters are obviously closely connected — by voting for a particular party people are in effect voting for a particular person to be prime minister — but it is possible that both are caused by something else, such as the performance of the economy. In addition, the poll figures show only the aggregated responses to the relevant survey questions and tell us nothing about the extent to which individual voters base their choice of party on their evaluations of leaders.

Nonetheless, there is plenty of supporting evidence in the specialist electoral studies literature that in the modern British electorate this is indeed what happens (Denver 2007).


What has been presented here is a very simplified account of valence politics, and it has to be said that it is not the whole story as far as understanding voting and elections in Britain is concerned. There remain large numbers of voters who identify with parties and who support their party through thick and thin. There are assuredly also some voters who have clearly worked out positions on central (or even not so central) policy issues and make their decisions in elections on that basis. There may even be a few voters who still think of politics in terms of class conflict. However, there seems little doubt that evaluations of, or simple reactions to, party leaders have become increasingly used by voters as a shorthand way of making a decision about which party to support in elections. This has been encouraged by the rise of television as the primary means of political communication and the intense focus of the media on party leaders. This makes for a tough life for party leaders. The fortunes of their parties are largely riding on their shoulders and, these days, the penalty for failure is swift demotion. Another consequence is that, in choosing leaders, parties increasingly have to pay attention to the electoral appeal of likely candidates (rather than ideology, standing in the party or even ability, for example). Whether this is a welcome development is a matter for debate.

David Denver is Professor of Politics at Lancaster University.

Reproduced by permission of Philip Allan Updates