Election issues , political party performance and geography

Introduction Voters make decisions on whether to vote and who to vote for in places. Places are crucial because they not only provide the context for interaction and social identification but they are the containers in which parties operate and seek representation by socialising and mobilising voters to support them. Longstanding theories of party choice stress how such places provide the space for group conditioning and environmental influences on individual behaviour. They also highlight how experiences shape an individual’s identification with a political party which seeks to represent their interests. For a long time it was assumed that such partisan identifications remained intense and stable throughout the lifecycle following repeat exposure, mobilisation and participation in the political process. Aside from being both enduring and reasonably resistant to changes in the political environment, these partisan attachments were shown to have indirect effects on electoral behaviour by influencing responses to party candidates and their policy platforms. Yet in the British context growing scepticism exists not only about the traditional strength of class as an explanatory variable of vote choice but the weakness of partisan ties.


Introduction
Voters make decisions on whether to vote and who to vote for in places. Places are crucial because they not only provide the context for interaction and social identification but they are the containers in which parties operate and seek representation by socialising and mobilising voters to support them. Longstanding theories of party choice stress how such places provide the space for group conditioning and environmental influences on individual behaviour. They also highlight how experiences shape an individual's identification with a political party which seeks to represent their interests. For a long time it was assumed that such partisan identifications remained intense and stable throughout the lifecycle following repeat exposure, mobilisation and participation in the political process. Aside from being both enduring and reasonably resistant to changes in the political environment, these partisan attachments were shown to have indirect effects on electoral behaviour by influencing responses to party candidates and their policy platforms. Yet in the British context growing scepticism exists not only about the traditional strength of class as an explanatory variable of vote choice but the weakness of partisan ties. Such scepticism has led to a renewed interest in the individual rationality decision making models, party-issue linkages and performance based evaluations to explain political choice.
A key element of the political game in Britain is the way in which parties compete with each other in places to obtain representation, in the hope of securing enough seats in Westminster to form the government. Parties are rational in their approach to targeting resources in those places that are hotly contested and in developing highly strategic personal messages to key voters who they believe can swing the election in their favour.
Often such targeted messages stress the party's policy platform on particular issues that they consider salient to the voter. Each party will place different levels of importance on different election issues and voters will assess the relative importance of those election issues when deciding who to vote for.

Perceptions
of each party's performance on salient election issues will vary among different sections of the electorate, and hence each party will try to use credible information to stress past performance, competence and the viability of future initiatives to bolster support from those voters who decide the election outcome. Such comparative judgements about party performance on the issues that matter shape, at least in part, the likelihood that a voter will support that party. Yet given the importance of place in the electoral arena, it is unclear whether differences in voting patterns reflect spatial variations in the perceived importance of election issues or spatial variations in the perceived performance of political parties on those election issues.
Here Lancaster's (1966) characteristics model is used, which assumes that people derive utility from the characteristics embedded in a service or product rather, than simply the service or product itself; for instance, it is not the presence of a road network that is important but the quality of the road network and the efficiency of using it to get to your destination. Lancaster's demand theory helps justify the existence of brands, and managers will seek to differentiate their brand from competitors by adjusting characteristics. Political parties can be viewed as competing to provide a range of similar services but with each party providing a different performance and with varying importance given to each service. This paper extends Lancaster's model to preferences for political parties in a general election, and tests the usefulness of the model using a unique individuallevel dataset, collected across voters in three parliamentary constituencies, where each constituency had, at least in the recent past, elected representatives of different political parties.

Background
It is sensible to assume that every politician's primary objective is to gain power (Downs, 1957), but whether that is for self-gain, altruism or because they are ideologically driven is unclear. To achieve their goal, candidates and parties develop a policy platform, which, during an election, is put to voters through manifestos, speeches, public appearances and local and national campaign activity. Parties may tailor policies to garner support from those groups who have a longstanding identity with them. Yet voters often convey messages about their policy preferences to parties and as such parties may seek to lure voters by moving their policy positions to reflect this in order to maximise their support. Political rivalry is illustrated through parties positioning themselves along an ideological political spectrum, but while this allows parties to communicate their aggregate position to prospective voters it is not particularly useful when parties are attempting to inform prospective voters about their positions on specific election issues.
According to Clarke et al. (2004), one of the consequences of weakening partisanship and the dwindling importance of the partyclass linkage is that more voters are open to persuasion, exposed to short term influences and salient events operating in the electoral arena. The onset of de-alignment in Britain (Sarlvik and Crewe, 1983) has led to a renewed focus on rational decision making and individual utility maximising strategies to explain political choice. This neoclassical framework favoured by Downs (1957) is based on the premise that people are selfish, rational actors who conduct cost-benefit calculations based upon indicators like unemployment, price stability or economic performance indicators such as the government deficit. For Downs (1957), the left-right ideological continuum motivates party competition with voters rationally seeking parties they perceive to be closest to their own preference (to maximise utility) resulting in parties ultimately adopting policy positions that lures as many voters to them as a possible.
Yet Stokes (1992) and others stress that certain issues or 'valence issues', where public opinion on achieving such desirable policy outcomes is uniformly shared and heavily skewed, matter more. Individual voter and party differences along policy dimensions are not as important as how voters perceive a party's ability to solve existing national problems or issues. Within this valence framework what matters are comparative evaluations of parties' managerial capabilities, their competence and performance in solving these salient issues and achieving positive outcomes.
As such, voters will seek to maximise their utility by selecting the party they believe will best deliver positive results on those issues that concern them most.
More broadly, Lancaster's neoclassical model is adopted and applied here because of its flexibility to deal with different option choices (i.e. which political party to vote for) and combinations of characteristics (e.g. election issues), and allows the identification of how these issues combine to determine party support.
The standard Lancaster (1966) model calibrates the axes to correspond with the characteristics embedded in goods and services, such as boot capacity and horse-power embedded in different cars (Gwin and Gwin, 2003). Here this is applied to political voting where the axes correspond to different election issues, such as immigration and the economy, see Figure  1. The more an individual feels a particular election issue is important then the further along the axis the voter's response will be. When two election issues are represented on a twodimensional figure then it can be illustrated using a single ray. When one election issue is considered more important relative to another then the ray will be closer to the axis that represents that election issue. Figure 1 presents a situation when two election issues are considered together, immigration and the economy, and two political parties, A and B. Party A is perceived by a voter to perform better than party B on the issue of immigration but worse on the issue of the economy. Whether a political party receives the person's vote will depend on how well the voter perceives the party performs on different elections issues but also the willingness of the voter to trade off one election issue for another. The trade-off between election issues is portrayed in this model by the slope of an indifference curve (IC). The rational consumer then selects the party that maximises their satisfaction, with party A getting the vote if they are willing to trade off the economy for immigration, and therefore they perceive immigration to be more important (IC1), or party B receiving the person's vote if they are willing to trade off more immigration for less of the economy (IC2).

Data and calibration
This study used a crosssectional questionnaire using a stratified sample from the South West of England region. Through a combination of Likert scales and open answer questions, the questionnaire focused on how important the public feels elections issues are, and how they view political parties perform on these election issues. For brevity, this paper only considers the public's perceptions of the Conservative, Labour and Liberal Democrat parties. The policy issues selected for analysis were the economy, education, immigration, antiterrorism and the National Health Service (NHS).
Questions were framed as retrospective evaluations in order to account for differences between voters; each respondent was asked "How well do you think the Conservative government handles the economy?", then proceeded to ask the question again but substituted "the economy" with each of the other policy issues, and then this process was repeated but with the Conservative party replaced with the Labour party and then the Liberal Democrats. where possession of more money can mean the purchasing of a greater quantity of particular attributes, the UK voter only has one vote; a final point along that ray therefore corresponds to what the voter perceives they will receive in terms of the two election issues in exchange for 'spending' their vote on a particular party. The angles of the rays indicate how the respondents rate the importance of one election issue relative to another election issue, and the length of the ray indicates how well a particular party performs relative to another ray which corresponds to a different political party.

Results
After combining all of the three constituency's results, Table 2 shows that the Conservative party received the highest scores for performance on the economy with a mean of 3.07 and on antiterrorism with a mean of 3.38. Labour received the highest mean scores for education (3.29) immigration (2.50) and the NHS (3.35). The Liberal Democrats scored the lowest mean scores for their performance on the economy, immigration and anti-terrorism, and had the highest deviations from the mean across all policy areas, suggesting that opinions of this party were highly varied. The Liberal Democrats may have lost vote share in the 2015 general election because voters' had relatively poor perceptions of their likely performance on contemporarily important election issues. Figure 2 presents a radar chart that illustrates the perceived relative performance of each political party on each election issue. It can be seen that if education, immigration and the NHS were considered by the electorate to be the most important election issues then the Labour party would have got into power. Also of note is that the respondents on average saw the Liberal Democrats outperforming the Conservatives on education and the NHS. If either the Labour or Liberal Democrat parties are going to gain power then they need to shift the political debate away from the economy and terrorism, and onto education and the NHS. The continuation of the political debate on issues of the economy and terrorism will continue to result in the Conservatives gaining votes. Similarly, it is in the   The results highlight that it is not only the perceived importance of election issues that varies across space but also the perceived performance of political parties on those election issues. This is of interest to analysts and political parties because it appears that the information provided to the electorate is either interpreted in slightly different ways across constituencies or that the effectiveness of that information in getting to the electorate varies across space.
What is behind these spatial differences? There may be a process that spatially sorts voters with certain preferences for election issues across constituencies. Perhaps people who have done well in the labour market, and who have located to a constituency with high house prices, value the economy more importantly than the NHS, and are therefore less willing to trade off the economy for an improvement in the NHS. Perhaps people who have suffered more with ill health, and therefore may have also done less well in the labour market, colocate in another area and collectively place greater emphasis on the NHS and would trade off improvements in the economy for greater improvements in the NHS.
Because there are spatial patterns related to labour market success it is likely to be the case that voting patterns relate to these issues. Another possible explanation is that there are local community peer group effects where locals source information from offline and online media forms, campaigning and discussions or interpretations of policy information in particular ways. A further possibility is that experiences accumulate over time and change or strengthen perceptions of the importance of election issues and that these may be associated with residential location preferences.
Currently, however, it is not known which of these or other factors dominate, whether, how and why their importance changes over time, how they interact with other contributing factors and how they in turn influence voting patterns. This will be the subject of future research.