How voters decide

VOTING has been described as the 'last religious act in a secular age' (Moscovici cited in Himmelweit et al., 1985). Despite falling turnouts, the majority of the UK electorate still cast their votes in a general election – but down to 59 per cent in 2001. For a social psychologist, voting is of especial interest. Everyone has to make the decision on the same day and the options are the same for everyone. The decision also has to be made repeatedly, so that each person builds up a repertoire of past decisions. Hence, it is possible to compare voting behaviour over substantial periods of time. But can such research help us to understand and predict the 2005 general election?

The consumer model of voting

How Voters Decide was the title of the most substantive psychological analysis of voting in the United Kingdom. In a longitudinal study, Himmelweit and his colleagues (1985) followed the same group of 178 men from the time they cast their first vote in 1959 at age 21 through to their sixth vote in October 1974. The generality of their findings was further tested through additional data drawn from six separate British election surveys (between 1970 and 1983), and involving more than 15,000 men and women.

From this research, Himmelweit et al. put forward what they called 'a consumer model of vote choice'. They proposed that the same principles hold in voting as in purchasing consumer goods. This was not to say that the voter thinks of parties merely as leading brands of detergent, distinguished only by their packaging. Rather the voter searches for the best fit – or the least misfit – between his or her views and preferences and the parties' platforms. In making their decision, the key link is the set of beliefs a person holds concerning the parties' policies on particular issues. Thus, in deciding how to vote in 2005, it would not be enough to be in favour, for example, of increased health spending; people need to know which party is most likely to implement such a policy.

In addition, Himmelweit et al. distinguished between what they called attitudinal shift and attitudinal fit. Attitudinal fit was measured by comparing the attitudes of voters with the mean profile of attitudes held by voters for the Conservative, Labour and Liberal parties respectively: the closer the attitudes, the nearer the attitudinal fit. Attitudinal shift was measured in their longitudinal sample by comparing attitudinal fit between successive elections: the larger the difference, the greater the shift in attitudes. A third factor of voting history was also included (how their longitudinal sample had voted at previous elections since 1959).

Based on these three factors, a discriminant analysis showed that the sample's votes could be classified with a high degree of accuracy (for example, 81 per cent in the October 1974 general election). Attitudinal fit was the most powerful factor, followed by attitudinal shift, then voting history. But omitting voting history altogether reduced the accuracy of classification by only 4 per cent. The habit of voting for a party Himmelweit et al. took to represent identification with a political party. In this respect, their findings contrasted with the traditional view in political science, that people's votes are primarily influenced by long-standing 'tribal' allegiances to specific political parties (e.g. Campbell et al., 1960, 1966). This suggests political parties are not wasting their money on campaigns: the potential for major voting changes is there.

The concept of attitudinal shift is of particular importance, because it suggests a causal role for attitudes: voters who changed their views didn't just keep them private, they acted on them and were more likely to vote for a different party on the next occasion. Thus, according to the consumer model, political attitudes are the most important factor in understanding how people vote. Party political identity – as measured by voting habit – is only of marginal significance. So if in 2005 the Iraq war has resulted in significant disaffection with Labour (attitudinal shift), then this could result in a major electoral upset.

The consumer model in the modern age

To have analytic and predictive value, a model of voting needs to work in different historical contexts. Himmelweit et al. claimed that their model did just this, by allowing for the role of different parameters to vary at different elections. Each election they likened to a new shopping expedition, in which both new as well as familiar goods are on offer. Not only can the role of issues vary from election to election, so too can the relative importance of past habit or party identification vary with the party preferred. Notably, membership of political parties has declined significantly in the modern era (Whiteley & Seyd, 2002); if this reflects a decline in party identification, the consumer model would still seem highly relevant to our understanding of the 2005 election.

But the past two decades have also seen other more dramatic changes in British politics. These have highlighted important factors not accounted for by the consumer model, and we turn to these factors now.

Competence - This is the most important factor. It is not enough that there should be a close fit between a voter's attitudes and those of a particular political party. Voters also need to believe that a political party can deliver in those policy areas that they most care about. According to research in political science, since 1987 the average voter has sat closer to Labour and the Liberal Democrats/Alliance than to the Conservatives on a whole range of policy issues – yet the Conservatives won in both 1987 and 1992 (Clarke et al., 2004). In terms of the consumer model, there is no obvious explanation for this. But arguably, the key to those Conservative successes – and also to Labour's subsequent victories in 1997 and 2001 – was that the loser was tainted with the image of managerial incompetence (Clarke et al., 2004). From this perspective, disaffection with Labour in 2005 over the Iraq war no longer becomes the only issue. It is the voters' judgement of managerial competence over a whole range of policy issues that will determine the election.

Leadership - Modern elections are fought out primarily through the mass media, and especially television. Where political leaders were once shadowy figures seen only through occasional appearances at public meetings, their image is now beamed into people's homes through television. Furthermore, they are personally subject to ever more intense media scrutiny. In these circumstances, the image of the leader is arguably an important electoral factor in its own right. Factor analyses of leader ratings show two distinct but interrelated dimensions: competence and responsiveness. Ratings of 'keeps promises', 'decisive' and 'principled' load on competence, ratings of 'caring', 'listens to reason' and 'not arrogant' load on responsiveness (Clarke et al., 2004). Whereas the Iraq war has undoubtedly damaged Blair in terms of responsiveness, what will be crucial is whether he still outstrips Howard on perceived competence. Perhaps this perception of damage to Tony Blair's responsiveness is behind the absence of his photograph in many Labour candidates' election addresses, whereas in previous elections he has generally been featured prominently.

Identity - According to the traditional view in political science, voters acquire social class and party political identities early in life. Once formed, these identities are stable and continue to exert powerful effects on voting in successive elections (Converse, 1969). Similarly, in the research by Himmelweit et al. (1985) party identity was operationalised in terms of voting habit: voting for the same party reflects party identity, changing allegiance reflects a lack of party identity. However, from a modern standpoint, both these perspectives seem somewhat dated. We now accept that identities can be transient and fluctuating: in the political context, party identities may be updated and modified in the light of new events and new information (Clarke et al., 2004). From this perspective, what matters in 2005 is not so much the minority of people who are paid-up members of political parties; rather, it is the effects of the last four years' political events on the looser political affiliations and party identities of the electorate as a whole.

The valence voting model

All the above considerations have been incorporated into a new voting model, proposed by Clarke et al. (2004). They argue that voters are primarily concerned with what they call valence – the ability of politicians to perform in those policy areas that people care about most. Based on data from the 2001 British Election Study, they found that the best two predictors of voting for all three parties were leader evaluations and party identification (partisanship). For Labour voters, favourable assessments of the economy were a good predictor too. Also predictive of voting – but of less importance – were 'issue proximity' (judging the stance of a political party in relation to the voter's own policy position) and selecting a party as best able to handle the most important election issue.

Clarke et al. (2004) interpreted these findings as follows. Voters use current information about the performance of leaders and their parties as sensible, albeit rough and ready, guides to political action; they also reduce cognitive load by storing that information in updated partisan attachments which may change over time. Thus, partisanship and leader images may be seen as cognitive shortcuts to judgements about valence, and hence about which way to vote. These shortcuts should also be predictive of the outcome of the 2005 election, as undoubtedly will be voters' assessments of the economy.

Conclusions

According to the consumer model, political attitudes are the most important factor in understanding how people vote. According to the valence model, the stance of a political party in relation to voters' attitudes is not sufficient; voters are also influenced by evaluations of whether politicians can deliver in those policy areas of greatest concern. From this perspective, politicians have to convince voters they can deliver in policy areas of greatest concern; such electoral credibility will be the key to electoral success in 2005.

- Professor Peter Bull is in the Department of Psychology at the University of York. E-mail: [email protected].

References

Campbell, A., Converse, P., Miller, W. & Stokes, D. (1960). The American voter. New York: Wiley.
Campbell, A., Converse, P., Miller, W. & Stokes, D. (1966). Elections and the political order. New York: Wiley.
Clarke, H.D., Sanders, D., Stewart, M.C. & Whiteley, P.F. (2004). Political choice in Britain. Oxford: Oxford University Press.
Converse, P.E. (1969). Of time and partisan stability. Comparative Political Studies, 2, 139–172.
Himmelweit, H.T., Humphreys, P. & Jaeger, M. (1985). How voters decide (Revised edn). Milton Keynes: Open University Press.
Whiteley, P.F. & Seyd, P. (2002). High-intensity participation: The dynamics of party activism in Britain. University of Michigan Press.