Home | About Us | Contact Us | Site Map | FAQs

^aDepartment of Government, University of Essex, Colchester, Essex CO4 3SQ, UK. E-mail: sanders@essex.ac.uk

It is widely believed that the Conservatives can significantly improve their chances in the next general election by 'moving to the centre of the political spectrum'. This paper uses data from the 2005 British Election Study to simulate both the direct and indirect consequences of such a move. The simulations show that the Conservatives would undoubtedly benefit from moving to the ideological centre-ground of British politics. However, they also indicate that a move to the centre, on its own, would probably add only 5 percentage points to the Conservatives' share of the vote in 2009/10 ¾ clearly insufficient to ensure a Conservative victory. One way in which the Conservatives could garner the additional votes they need is to target the large number of public sector workers who have been alienated from Labour by its emphasis on public sector micro-management.

British Politics (2006) 1, 170-194. doi:10.1057/palgrave.bp.4200015

conservatives; general elections; ideological centre-ground; public sector reforms

Introduction

The Conservatives' third successive general election defeat, in May 2005, was a reflection of their failure to make any real electoral progress since the ERM crisis of September 1992. The fracturing of the Conservatives' reputation as the party of economic management competence in the wake of that crisis dealt them an electoral blow from which they have yet to recover. Three changes in leadership failed to make any serious impression either on the Conservatives' medium-term opinion poll ratings or on their electoral performance. To be sure, part of the explanation for the Conservatives' difficulties derives from Labour's continuing popularity, and in particular from its success in managing the macro-economy. However, the Conservatives' failure in 2005 to capitalize on the misfortunes of an incumbent government that could muster only 36 percent of the popular vote suggests a deeper malaise within the party that requires serious attention. A key problem that the Conservatives face, of course, is the one that besets any opposition party. It tends to be governments that lose elections rather than oppositions that win them. In this sense, the Conservatives ¾ like the Liberal Democrats ¾ must simply wait for the government to make enough policy errors for the electorate to lose confidence in Labour's ability to manage the country's affairs. What an opposition party can and must do, however, is to make enough voters believe that it is a plausible government in waiting. At the time of writing, this is something that the Conservative Party has persistently failed to achieve.

This article explores some of the ways in which the new Conservative leader, David Cameron, might be able to reposition his party in order to render it 'electable' by the time of the next general election. Some of the repositioning proposed is based on nothing more than anecdote and impressions that have been obtained from talking to people who work in different parts of the UK public sector. Most of what is suggested, however, is based on publicly available survey data that were collected as part of the 2005 British Election Study. These data enable us to analyse the sources of Conservative support; to identify those sources that are amenable to manipulation by the Conservative leadership; and to suggest what is likely to happen to Conservative support if certain repositioning strategies are adopted. Part 1 of the paper discusses how the Conservative Party could make electoral capital out of the current government's handling of public sector reforms. Part 2 examines the changing ideological landscape of UK public opinion. It shows that contemporary UK voters are not distributed along a single left-right continuum, and explores how the Conservatives can benefit from the new multi-dimensional ideological space that the electorate now inhabits. Part 3 develops a simple model of Conservative voting in 2005. It shows how Conservative support can be increased if the Conservatives are prepared to: abandon the claim that they are 'the party of tax cuts'; go into an election with a leader who is not actively disliked by the electorate; and focus even more strongly on issues related to citizens' concerns about their own personal, and national, security.

The Size of the Task and the 'Public Sector Opportunity'

It is tempting to conclude on the basis of the Conservatives' performance in 2005 (a 33 percent vote share in Great Britain: only 3 points behind Labour) that they simply require 'one more heave' in order to return to power in 2009/10. However, the geographical distribution of votes is such that, under current electoral boundaries, it would require in the region of an 8-point swing from Labour to the Conservatives to produce a Tory majority in the next parliament (Whiteley, 2005). Although the work of the Electoral Commission in redrawing constituency boundaries during the course of the present parliament is likely to reduce the size of the task somewhat, the Conservatives are likely to require at least a 7-point swing in their favour if they are to stand any chance of forming a majority government after the next general election.

One obvious approach for any party that is seeking to increase its support base is to contrive new ways of 'reaching out' to groups to which it has not previously had much appeal. The Conservatives have undoubtedly suffered since the mid-1990s from Labour's ability to make itself more attractive to middle class voters. And in both 2001 and 2005, the Conservatives appealed less to men than to women. (In 2001, according to the British Election Study, Conservative support was 2 percentage points higher among women than men; by 2005, this 'gender gap' had grown to 5 points.) While it is obvious that the Conservative Party would benefit if it could appeal more strongly to the middle class and to men, it is by no means clear what sort of policies it could articulate in order to appeal directly to these groups without risking a commensurate loss of support among working class and women voters. This is the endemic problem of political marketing strategies that focus attention on the interests of particular socio-demographic groups of voters. What is gained on the swings can be lost on the roundabouts.

In fact, however, there is one 'in-group' that the Conservatives could target to its advantage without risking the electoral opprobrium of the corresponding 'out-group': people who either work in the public sector or who have family members who work in the public sector. This group now represents over 40 percent of the electorate.

In the early 1980s, public sector workers tended to shy away from voting Conservative. As Table 1 shows, in the 1983 general election, the Conservatives gained the support of almost 47 percent of those who worked in the private sector, compared with only 37 percent of those who worked in (or had a spouse/partner who worked in) the public sector. As the table also shows, notwithstanding the general decline in Conservative support since the 1980s, by 2005 this 10-point public sector 'deficit' in Conservative support had fallen to under 2 points. This declining differential in private/public sector support for the Conservatives suggests that the Conservatives are clearly capable of attracting support from those who work in the public sector. To be sure, the public sector is not as large in the early 21st century as it was in the early 1980s. In 1983, according to the British Election Study, over 49 percent of eligible voters either worked in, or had spouses/partners who worked in, the public sector. By 2005, the equivalent figure was 40 percent ¾ a clear reduction but nonetheless still a sizeable proportion of the electorate.

In my view, public sector workers and their families represent an obvious target for a sustained campaign by the Conservatives. It would obviously be unreasonable to suppose that all, or even a majority, of the public sector workers who voted Labour in 2005 could be recruited to the Conservative cause in 2009/10. However, the Conservatives will only win power by making relatively small support gains in a number of different areas. Modest Conservative gains among public sector workers could add two or three percentage points to the Conservatives' vote share in the next election. On its own this would clearly not be sufficient to propel the Conservatives back into office ¾ but it would certainly provide a stronger platform for some of the additional changes that are suggested in the subsequent sections. The key point is that the Conservative Party can potentially increase its support among public sector workers without doing anything that might reduce its support elsewhere.

The reason that the public sector offers fertile ground for the Conservatives lies in the problem of low morale that besets large swathes of public sector employment. There is no direct systematic empirical evidence to support the claim that a crisis of morale exists. There is nonetheless some indirect evidence that people who work in the public sector are not as content with their lot as they were when New Labour came to power. Table 2 reports how public sector workers' economic perceptions have changed since 1997. In 1997, almost 40 percent of public sector workers were optimistic about the country's economic future. This compared with an equivalent figure of only 34 percent optimism among private sector workers. As the table shows, by 2005, after 8 years of New Labour public sector reforms, this 6-point 'optimism differential' had effectively disappeared: public sector workers were no more likely to be economically optimistic than their private sector counterparts. It is obviously important not to over-interpret this relatively modest over-time change. However, it is symptomatic of the frustrations that many public sector employees feel with regard to the treatment that Labour has meted out to them over the last 8 years or so. They are grateful that more resources have been allocated to public service provision. However, they are deeply frustrated that these extra resources are disproportionately spent not on improved resources but on responding to the new 'audit environment' that is threatening to overwhelm them. If the Conservative Party can properly understand why public sector employees have become increasingly disillusioned with Labour since 1997, it should be able to capitalize on that disillusionment to its own electoral advantage.

Part of the research for this paper involved a series of in-depth interviews with a number of middle- and high-ranking employees from the education, health and police services. A small number of middle-ranking civil servants were also interviewed. All interviewees were guaranteed anonymity. Almost without exception, they took the view that the culture of target, monitoring and audit that now pervades UK public service is both counter-productive in terms of 'real' service delivery and a massive waste of resources. It is possible that these interviewees were untypical of public sector opinion more generally, though I doubt it. They identified a number of important deficiencies in Labour's approach to public sector reform. The Conservatives could clearly benefit from a campaign that promised to remedy these deficiencies.

Setting absurd objectives and targets that distort real priorities

One well-known, and possibly apocryphal, story from the Soviet period is about a factory manager who found he could meet his production targets by concentrating all of his factory's efforts on producing left-footed boots. A senior civil servant at the Treasury used this story to illustrate the general principle that, once a performance indicator becomes a target, it ceases to be an effective indicator of real performance. This respondent suggested that the objectives, aims and targets of the Department of Constitutional Affairs (DCA) provided a good illustration of how current Treasury-inspired civil service practice produces targets that distort priorities. The DCA begins its own 'Aims' document thus:

DEPARTMENT FOR CONSTITUTIONAL AFFAIRS
AIM
Ensure effective and accessible justice, protect the rights of citizens, and modernize the law and constitution

Objective I: Ensure the effective delivery of justice.
Improve the delivery of justice by increasing the number of crimes for which an offender is brought to justice to 1.2 million by 2005-06; with an improvement in all CJS areas, a greater increase in the worst performing areas and a reduction in the proportion of ineffective trials.
Source: www.hm-treasury.gov.uk/performance/lcd.cfm

No-one could argue with either the stated Aim or Objective I. But a moment's reflection demonstrates the absurdity of the target associated with Objective I. Increasing the numbers of offenders brought to justice says nothing about the severity of the crimes involved or about the need perhaps to reduce the number of crimes that are committed in the first place. These targets eventually find their way down to real police officers who have to deal with the public. Faced with a need to 'meet the target in our CJS area', officers on the ground find themselves instructed to increase the number of 'effective prosecutions' no matter what the actual priorities are for fighting crime in their particular area. As a result, resources are shifted from what matters (violent crimes, particularly against women; long-running enquiries into organized gangs ¾ where successful prosecutions are difficult) to prosecutions in relatively trivial areas (such as motoring and non-violent drunkenness offences) where successful prosecutions are relatively easy to secure. A recently retired Borough Commander in the Metropolitan Police confirmed that this was exactly the sort of practice that he had been obliged to adopt in order to meet prosecution targets that his force had been set.

Setting targets because targets have to be set and setting only achievable targets

All observers of the Chancellor of the Exchequer's recent efforts to redefine the starting point of the current economic cycle so that public borrowing can be said to conform to his 'golden rule' will be familiar with the idea that target-setting is an intrinsically corrupt activity. An illustration of the 'setting only achievable targets' principle can be seen from the evidence that the DCA provides to show that it is likely to meet its target of bringing 1.2 million offenders to justice by 2005/06. The DCA document highlighted above provides evidence that purports to show that the stated target is being met:

By the end of June 2004, 35 of the Criminal Justice Areas had improved their performance in terms of the number of offences brought to justice over the baseline year of 2001/02 (my emphasis).
Source: http://performance.treasury.gov.uk/T123_I0370.pdf

This seems fair enough. But if we go to another of the DCA's related web-links, http://performance.treasury.gov.uk/T123_I0366.pdf, we find a graph that shows that the number of offences brought to justice was noticeably higher in 1999/2000 than it was in 2001/02. This begs the obvious question as to why the DCA defines 'the baseline year' as 2001/02. The equally obvious answer is that it was easier to set an achievable target that would show an improvement from 2001/02 than from 1999/2000. I cannot demonstrate that all of the targets that government departments and quangos set for themselves exhibit the same degree of calculation that I have described here. And I really do not wish to single out the DCA for distinctive treatment. However, it was clear from talking to all of my interviewees that this is an excellent illustration of what is a ubiquitous phenomenon across the UK public sector. For any given organization in the UK public sector, and for any given sub-unit within that organization, the setting of achievable targets is primarily a defensive public relations exercise, not a real exercise in providing the best possible service that can be provided given available resources.

The injunction to gather performance-related information typically under-estimates the transactions costs involved

Labour's approach to public policy seems to assume that the extra information that is needed in order to set targets, to monitor progress towards targets, to evaluate performance in the light of those targets and to revise strategy accordingly, can be achieved without substantial transactions costs. According to my respondents, this supposition is horribly wrong. Strategies for collecting, processing, retrieving and evaluating performance-related information are incredibly expensive. They require that information systems be continuously updated as the information technology environment evolves. They require new appointments in new areas of expertise. They require huge amounts of senior staff time in order to ensure that the organization does not score badly in any of the performance league tables that are published by government of its agencies. All of these things drive resources away from front-line services. Worse, front-line service staff find themselves enmeshed in a mountain of electronic form-filling every time they engage in any sort of action which provides a real service to the public. The form-filling inevitably reduces the amount of time that they can spend in actual service delivery. One of my health service respondents ¾ a GP with over 30 years' experience ¾ commented:

For every five minutes I spend with a patient, I have to spend another five minutes on my computer so that I can show I've met all my targets and complied with all the codes of practice that the government imposes on me.

Another senior health service manager made a similar point:

Everything you read in the tabloids about nurses not being able to get on with the job because of paperwork is true. They have to update their patient plans every time they take someone's temperature. If someone has to have their bedpan changed, it has to be logged¼. Heaven knows what use anyone makes of most of this information.

In the education sector, the problem is equally pronounced. A senior Special Needs teacher in a large comprehensive school told me:

All that matters as far as external evaluations of Special Needs provision in this school are concerned is for our paperwork to be in order. We have to show that we have up-to-date personal development plans for each child with Special Needs and that we have mechanisms for ensuring that we know how to update these plans in accordance with the child's needs, achievements and abilities. We spend as much time on these plans as we do on looking after, and trying to teach, the children. If we didn't have the burden of paperwork, we could make far more progress with the children.

Higher education experiences a similar plight, according to my respondents. A former Pro-Vice-Chancellor from a leading research university described the process of teaching evaluation in the following terms.

The irony of so-called teaching evaluation exercises in UK universities is laughable as well as fraudulent. The only thing that is properly evaluated is the supporting paperwork. As long as you have an appropriate strategy document and can show that you have right procedures in place to ensure that your stated teaching objectives are met, you come out of the exercise very well¼. the one thing that is never seriously observed or evaluated is the actual teaching in lectures, seminars or tutorials.

The unintended creation of perverse incentives

The favoured outcome of performance management schemes based on targets, monitoring and evaluation are performance 'league tables'. Such tables now abound throughout the UK public sector. The assumption that underlies the publication of these tables is that no institution wishes to be the bottom of any league table and most will wish to improve their league table position over time. On the face of it, this seems quite sensible: the competition engendered by league table publication will drive up the performance of the sector as a whole. Unfortunately, this approach also has a negative side. Not all can win prizes. If some move up the league table, others must by definition move down. Going up in a table is great for internal institutional morale; but going down is equally and obviously damaging ¾ and can lead to an institution going into a spiral of decline. The problems of league tables, however, extend beyond this simple logical point. The obvious example where league tables have produced perverse incentives for institutions has been in the health service. A senior NHS consultant in an East Anglian hospital told me that:

Trying to meet government requirements on bringing down waiting times has in some instances almost doubled the number of consultations that we have had to provide. To meet the target, we have a short preliminary meeting with the patient. We make a quick assessment and diagnosis and then make another appointment for a few weeks or months later. But the crucial thing is that we've met the target. If we don't, it's bound to crop up in some league table or other and the hospital will get bad publicity¼.The worst thing, potentially, is mortality rate data. What better incentive do you want not to admit really sick people to your hospital?

In many respects the foregoing testimonies only touch the surface of some of the difficulties associated with the present government's approach to public service delivery. As indicated previously, I cannot claim to have evidence to show that these views are representative. However, I suspect that they do reflect a genuine disquiet that exists across large areas of public service. Several of my respondents referred to their sense that the government was actively trying to de-professionalize them. Government was not prepared to trust them to exercise their professional judgement and to nurse/doctor/police/teach/research to the best of their ability simply because they thought it was their professional duty to do just that. Rather, under the guise of the entirely laudable aim of trying to ensure that public money was spent properly, government was seeking to micro-manage them by micro-incentivising every aspect of their activity. However, the practical effect of this micro-management was to divert resources ¾ and crucially effort, enthusiasm and energy ¾ away from the delivery of front-line services. The in-depth interviews also indicated that government policy has, either wittingly or unwittingly, put the key decision-makers in almost all public service institutions in an invidious position. Regardless of their own reservations about the burdens of target, monitoring and audit, key decision-makers cannot afford to admit in public that the system is seriously flawed for fear that their own institutions would immediately become targets for increased monitoring and attention by government- or quango-appointed auditors.

Labour's pronounced and prolonged failure to trust the professional judgement of public sector employees provides a clear political opportunity for the Conservatives. Workers in this sector are ripe for conversion to any party that promises ¾ with conviction ¾ to reduce the target/monitoring/audit burden that Labour has inflicted upon them. There are innumerable ways in which this burden could be reduced, and a variety of justifications that could be provided as to why such a reduction is necessary. David Cameron and his team have at least two years to work out a plan for reducing the burden and providing a justification for doing so which fits in with their other policy plans. I strongly commend it to them, in the interests of both party competition and the proper operation of public services in the UK.

The Declining Relevance of a Single Left-right Ideological Dimension

The idea that a left-right continuum can be used to describe the ideological positions of both parties and voters has received wide currency in political debate and analysis for at least a century. Indeed, contemporary political commentators frequently refer to the need for the Conservatives to follow Labour's example and 'move to the centre ground of British politics, where most of the voters are'. In contrast, others argue that the Conservatives need to differentiate themselves clearly from Labour; that the party should remain 'firmly on the right'; and that it could increase its vote share by persuading voters to relocate further to the right. Of these two positions, the first is the only one that offers any prospect of electoral success. While Labour, in the 1980s, could plausibly be characterized as a genuinely socialist party, it was possible for the Conservatives to pose as an aggressively right wing party and still win elections. With Labour firmly located in the centre, 'clear blue water' equates to permanent opposition for the Conservative Party. Put simply, the Conservatives have no real choice but to move to the political centre if they wish to return to government in the foreseeable future. At the same time, it is important to stress that the term 'ideological centre-ground' is by no means unambiguous. Electoral researchers have shown that there are least two major ideological dimensions that underpin most people's political views (see, for example, Heath et al., 1991). These involve a traditional 'left-right' economic dimension, which reflects the extent to which the state should be involved in economic activity ¾ where more equals left and less equals right; and a 'liberal-authoritarian' dimension, where 'liberal' is associated with tolerance towards minorities and unconventional lifestyles and 'authoritarian' is associated with severe penalties for those who transgress laws and/or established social conventions. One of the clever features of New Labour was its recognition that these two dimensions are much more independent than the single notion of 'left-right' implies ¾ in statistical terms, they are orthogonal or uncorrelated. This meant that New Labour could be both 'progressive' in terms of state intervention and authoritarian in terms of being 'tough on crime' ¾ positions that were both popular with voters and which, inter alia, contributed to Labour's three successive general election victories.

How does the ideological space appear in 2005? Table 3 reports the results of a factor analysis of a number of attitudinal measures derived from the 2005 post-election British Election Study survey. Note that all the individual attitude variables are ones that most people would typically associate with a single 'traditional' left-right scale. For example, 'left leaning' people would typically be the ones who reject the idea that immigrants increase crime rates; are opposed to capital punishment; believe that strong trade unions are needed; and reject the idea that men are better suited to politics than women are. By putting all of these measures together in a factor analysis, we can determine empirically what sort of political attitudes cluster together.

The results of the factor analysis show that there are four distinct (and, by definition, statistically uncorrelated) clusters of attitudes ¾ factors ¾ rather than a single underlying 'left-right' dimension. The first factor measures people's tolerance (or intolerance) of difference. Respondents who think that immigrants increase crime rates (factor loading, =+0.745) and that failed asylum seekers should be sent home (=+0.786) score highly on Factor 1. Conversely, respondents who think immigrants are good for the economy (=-0.822) and that people should be more tolerant towards unconventional lifestyles score low on Factor 1. The second factor is clearly a liberal/authoritarian dimension that focuses primarily on respondents' attitudes towards criminals. Respondents who believe that lawbreakers should have longer sentences (=+0.614) and that violent criminals should forfeit their human rights (=+0.703) score highly on this dimension. Respondents opposed to the death penalty (=-0.592) in favour of rehabilitation (=-0.580) have low scores on this dimension.¹ The third factor corresponds to traditional economic notions of left-right. People who think that there is no need for strong trade unions (=+0.737), that private enterprise best solves economic problems (=+0.684), and that ordinary people get their fair share of the nation's wealth (=+0.652) all score highly ('right') on this dimension; those who disagree with these ideas ('left') score low on it. Finally, there seems to be a distinct anti-feminism dimension, which relates to attitudes to women in politics and women's roles in the workplace and the home. This dimension is highly skewed statistically, with only around 15 percent of respondents adopting an anti-feminist stance. This makes it of less substantive interest than the other dimensions, because it offers very little traction to any party that is seeking to increase its vote share.

The results shown in Table 3 have important implications for a possible Conservative strategy over the next 3 years or so. As noted above, Blair's New Labour could be both progressive economically and authoritarian socially (recall the mantra from 1997: 'tough on crime; tough on the causes of crime') because the economic left-right and liberal-authoritarian dimensions were orthogonal. This meant that Blair could make Labour more authoritarian than it had been in the 1980s without substantially changing its position on the economic left-right dimension ¾ and still be popular with enough voters to win elections. Table 3 shows that 'economic left-right', 'liberal versus authoritarian attitudes towards criminals', and 'tolerance/intolerance of difference' are also orthogonal, that is, uncorrelated. In non-technical terms, this means that voters make clear distinctions in their own minds about their attitudes to these three things. They see no contradiction, for example, between being authoritarian in dealing with criminals (traditionally regarded as a 'right-wing' position) and being sceptical of the unambiguous virtues of private enterprise (traditionally regarded as a 'left-wing' position). This in turn means that the Conservatives can ¾ indeed should ¾ adjust their positioning on all three axes in order to maximize their vote share in 2009/10. Crucially, they can do so without giving voters the feeling that the party is somehow being 'inconsistent' in the values that it is espousing.

'Moving to the centre', therefore, potentially means adjusting the Party's position on all three of the axes. Note, however, that the Conservatives may decide, for example, that they do not wish to adjust the relatively tough stance that they currently take towards criminals. After all, as the 'percent agree' figures in Table 3 show, most voters support tough law and order policies, so there would be very little benefit in the Conservatives' moving away from the ground that they (and New Labour) currently occupy. The crucial point, however, is that the Conservatives can maintain their current, aggressive, stance against criminals and still move towards greater tolerance of difference. In the (outdated) language of traditional left and right, they can simultaneously become more progressive in the way that they approach minorities of various sorts and maintain their strong (non-progressive) anti-criminal stance on law and order without contradicting the way that voters look at the world. Voters make a clear distinction between dealing with criminals and how people (including policymakers) should behave towards others who are different from them. The Conservative Party could make considerable electoral progress over the next few years by showing that it has genuinely become more progressive towards ethnic minorities, refugees, gays and women. Critically, there is no logical or empirical reason why it cannot do so while simultaneously maintaining a strong position on law and order.

A Model of Conservative Voting in 2005 and its Implications for Conservative Strategy

It is obvious that the possibilities for Conservative success in 2009/10 are intimately bound up with the reasons for the party's failure in 1997, 2001 and 2005. This section develops a model that reflects recent more formal analyses of UK voting behaviour but which also focuses on factors that the Conservatives themselves have the capability to change. The model assumes that there are six sets of variables that exert important direct effects on Conservative support, as follows.

Party identification: There is a long-standing dispute in the scholarly literature as to whether party identification is best characterized as an 'enduring affective attachment' to a particular party (Campbell et al., 1960) or as a 'cumulative running-tally of retrospective evaluations' (Fiorina, 1981). Whichever interpretation is correct,² however, there is probably not a great deal that the Conservatives can do in the medium term to influence the extent to which voters 'identify' with each of the major UK parties. Thus, although it is important to include terms for identification with the major parties in a model of Conservative support for control purposes, the actual magnitudes of any party identification effects are not really of substantive interest. The model simply expects, unsurprisingly, that Conservative identifiers will be more likely to support the Conservatives and that Labour and Liberal Democrat identifiers will be less likely to vote for them.

Managerial competence: It was noted above that the 1992 ERM crisis inflicted serious damage on the Conservatives' reputation for managing the economy. Recent research on 'valence politics' suggests that voters' perceptions of the general managerial competence of rival leadership teams is of absolutely critical importance in their party choice decisions (Clarke et al., 2004). A term is accordingly included that summarizes voters' assessments of the Conservatives' ability to handle six key issues ¾ the economy, health, terrorism, crime, asylum and taxation. The simple expectation is that the more highly that voters rate the Conservatives on this 'competence' index, the more likely they are to vote for them.

Leader images: People who are not particularly interested in politics often tend to use 'heuristics' ¾ cognitive shortcuts ¾ in order to arrive at their voting decisions. One of the most obvious heuristics that is deployed involves the images of the various party leaders. People assess the integrity and judgement of rival leaders and make an overall assessment of their suitability for office. The model taps this aspect of voter perceptions here by considering the extent to which voters express a liking for each of the three main party leaders. It is expected that the more people like the Conservative leader, the more likely they are to vote for the Conservatives; the more they like the Labour and Liberal Democrat leaders, the less likely they will be to vote Conservative.

Ideological distance: The analysis presented in the previous section suggested that there are probably at least four dimensions of contemporary ideology ¾ traditional economic 'left/right'; liberal versus authoritarian approaches to criminals; tolerance versus intolerance of difference; and pro/anti-feminism. Constraints of data availability mean that we can explore the explanatory role of only two of these dimensions. This is because, in order to properly consider the effects of ideology on voters, we need to know (a) where the voter locates her/himself along any given ideological dimension; and (b) where s/he locates each of the main alternative political parties. The 2005 BES established this sort of voter- and party-positioning along two relevant 'trade off' scales. The first was a tax/spend scale. Respondents were asked to indicate where they placed themselves and the parties in terms of their preferences for 'higher taxes and spend more on public services' as opposed to 'lower taxes and spend less on public services'. The model uses the responses to this scale as a surrogate for voters' estimates of their own and the parties' positions on the traditional economic left-right scale. It is assumed that a preference for higher taxes and more public spending corresponds to a left-wing position; a preference for lower taxes and reduced spending to a right-wing position. The second scale relates to the trade-off between 'reducing crime' and 'protecting the rights of the accused'. Respondents were again asked to locate both themselves and the main parties on this scale. A preference for reducing crime is assumed to connote an authoritarian position; a preference for protecting the rights of the accused, a liberal one. An illustrative set of positions on the liberal/authoritarian scale is shown in Figure 1. The crucial feature of the figure is the distances between the respondent's own position and where s/he places each of the parties. The simple predictive principle, derived from Downs' rational actor model (Downs, 1957), is that the voter will be more likely to vote for a party to which s/he is close than one which is distant. In short, the larger (smaller) the distance between the respondent and a given party, the less (more) likely s/he is to vote for it.

Issue priorities: It is sometimes said that a particular party 'owns' a particular issue. For example, it used to be said that Labour 'owned' the health and education issues while the Conservatives 'owned' law and order. What this meant was that people who prioritized health and education tended to vote Labour because Labour had traditionally taken a distinctive ('spend more') stance on these issues. By the same token, people who prioritized law and order tended to vote Conservative because the party was traditionally associated with tougher policies towards suspected criminals. The position is undoubtedly much cloudier today than it was in, say, the 1980s. Labour has become more authoritarian, seeking to encroach into Conservative territory on law and order. The Conservatives, for their part, have rediscovered the importance of a firm commitment to high-quality public services. Indeed, political parties are engaged in a constant struggle to ensure that the issues they believe best serve their electoral interests are at the forefront of voters' minds. The 2005 BES asked respondents (in an open-ended question) what they thought the most important issue facing the country was. The responses were wide-ranging but the most frequently-cited issues covered four main areas: the economy, public services (health and education), 'security-related' (crime, terrorism, asylum and immigration), and 'Europe-related'. Terms are accordingly included for each of these issue-priority areas in the model developed. Since Labour focused its campaign on its economic record and its delivery of 'improved' public services, it is expected that people who emphasized these issues would be less likely to support the Conservatives. Since the Conservative campaign focused on security-related issues, it is expected that people who prioritized these issues would be more likely to vote Conservative.

Economic perceptions: A considerable amount of research suggests that economic performance has a powerful effect on the electoral fortunes of the incumbent government. Although Labour's stewardship over an extended period of economic stability and growth may have reduced the impact of economic considerations, the available evidence suggests that voters' economic perceptions still play a non-trivial role in determining voting decisions in the UK (Sanders, 2005). The model accordingly includes a term for voters' economic optimism/pessimism about their own financial circumstances. The expectation is that people who are optimistic (pessimistic) about the economic future will be more likely to wish to maintain (change) the political status quo that has created their optimism (pessimism) and will accordingly be less (more) inclined to vote for the Conservatives.

Demographic controls: Following standard practice in statistical models of party choice, a limited number of demographic control variables are included, in order to reduce the chances that the model is mis-specified. These include controls for age, gender, ethnicity, manual/non-manual occupational class, and country (in the latter context, residents of Scotland and Wales are combined into a single 'celtic' grouping).

The formal model estimated is

where the ratings scores for Howard, Blair and Kennedy are 0-10 likeability ratings, the distance measures are absolute distances on 0-10 scales; economic optimism is a 5-point scale measuring household financial prospects, e_t is a random error term, and all other terms are as defined in the text.

Table 4 shows the results of estimating equation (1). Estimation is by logistic regression. The results show that some of the variables outlined above exert strong effects on Conservative support while others fail to achieve statistical significance. The party identification and leader-liking terms, unsurprisingly, all behave exactly as expected. Conservative identifiers and people who liked Howard were significantly more likely to vote Conservative. Identifiers with either of the other major parties and people who liked Blair and/or Kennedy were significantly less likely to support the Conservatives. Similarly, people who rated the Conservatives' managerial capabilities highly were more likely to vote Tory. In terms of ideological distances, only the tax/spend dimension proved to have significant effect on Conservative support. The negative coefficient for this term is correctly signed: controlling for all the other terms in the model, people who placed themselves a considerable distance from the Conservatives on this dimension were significantly less likely to vote Conservative. Note, however, that the same is not true for the reduce crime/protect rights scale. This term yields a non-significant coefficient, indicating that people were no more (or less) likely to vote Conservative if they were close to or far distant from the Conservative position on this dimension. The non-significance of this term is probably testimony to Tony Blair's success in making Labour equally attractive to the Conservatives in terms of the stance he has taken on law and order. With regard to the 'most important issue' terms, none actually achieves statistical significance, implying that voters' issue priorities had no direct bearing on Conservative support in 2005, over and above the effects of the other variables in the model. (Note that this includes the European issue area, which some party activists have wrongly claimed as a vote-winner for the Conservatives.) The economic optimism term is significant and (correctly) negatively signed. Optimists in 2005 were less likely to support the Conservatives, the corollary of their complementary tendency to support the incumbent Labour government. Finally, only two of the demographic control variables achieved statistical significance. Manual workers and people living in Scotland and Wales were significantly less likely to support the Conservatives. Given the appalling showing of the party both in inner city areas (where manual workers tend to be concentrated) and in Scotland and Wales, these findings are unsurprising.

Of course, the significance, or otherwise, of the various independent variables shown in Table 4 tells us nothing about the practical prospects for the Conservatives in 2009/10. In order to make substantive sense of the results presented in the table, we need to consider two questions. First, which variables in the model do the Conservatives have the capacity to change between now and the next general election? Second, how much are they likely to be able to change them, and what consequences would this imply for their vote share in 2009/10?

Consider the question of capacity first. It seems pretty self-evident that in the short term the Conservatives cannot do much to change the extent either to which people 'identify' with other political parties or with the Conservative Party itself. Similarly, there is not an awful lot they can do to affect the extent to which people like or dislike the leaders of the other main parties. Equally, they are largely impotent people with regard to how people are likely to feel about their personal economic prospects. And they certainly cannot influence the demographic composition of the country as a whole. We also saw from Table 4 that, in 2005, the Conservatives derived no direct benefit as a result of individual voters' priorities in any of the four major issue-areas. All of this implies that there are only three aspects of the Table 4 model, which are amenable to improvement as a direct result of the Conservatives' own efforts: (1) voters' assessments of the Conservatives' managerial competence; (2) the likeability ratings of their leader; and (3) the distance between the party and most voters in terms of the higher taxes/improved services versus lower taxes/reduced services trade-off.

In a moment, we will consider the likely consequences for the Conservatives of changes in these three key variables. However, before a full set of simulations can be run, we need to consider an important subsidiary question. It is all very well to say that the Conservatives could or should improve their 'managerial competence' ratings, but how can this be done? In order to answer this question, we need a model of Conservative managerial competence. In trying to specify such a model, it has to be acknowledged that, potentially, perceptions of Conservative competence may be influenced by exactly the same things that influence Conservative voting itself. In order to accommodate this possibility, I use the logic of recursive causal models to specify an equation for voters' perceptions of Conservative managerial competence as follows:

where all terms are as defined in (1).

Table 5 reports the results of estimating (2). Estimation is by OLS as the measure of managerial competence is a 24-point scale. The results in Table 5 are broadly similar to those in Table 4. However, there are some important differences. Liberal Democrat identification makes no significant difference to people's assessments of Conservative managerial competence. Similarly, although the Howard leader rating term is significant and correctly signed, the Blair and Kennedy terms are both non-significant. Finally, both the ideological distance terms yield significant, correctly signed coefficients.³ Taken together, these findings suggest that perceptions of Conservative managerial competence are more strongly related to specific features of the Conservative Party itself ¾ in particular its leader and the party's ideological location ¾ rather than to the features of all the parties, as appears to be the case with Conservative voting.

These findings are important because they suggest that the leader's image and ideological distance exert a double effect on Conservative support. As Table 4 shows, leader image and distance on the tax/spend scale ¾ in addition to assessments of Conservative managerial competence exert direct effects on the Conservative vote. Table 5 shows that, in addition, people's perceptions of Conservative competence are also affected (positively) by the Conservative leader's image and (negatively) by the extent to which people distance themselves from the Conservatives on the tax/spend scale. Crucially, what this implies is that if the Conservatives can 'improve' their leader's image and reduce the gap between where voters place themselves and where they place the Conservatives on the tax/spend scale, then they can effect both a direct and an indirect boost to their electoral fortunes. In addition, they can also boost voters' assessments of Conservative managerial competence by reducing the gap between voters and the party on the reduce crime/protect rights scale.

What sort of change in these variables is feasible? And what are the likely consequences for Conservative support in 2009/10 if the right sort of changes can be made? Table 6 describes the average scores for self-placement and for the placement of the three main parties on the tax/spend and reduce crime/protect rights scales. On tax/spend, Labour and the Liberal Democrats are very close to the mean voter position. The Conservatives, however, have a much lower score ¾ indicating that they are perceived by voters as being much nearer the low tax/reduce services end of the 0-10 scale. The Conservatives would need to move closer to the mean position by a full-scale point in order to neutralize the electoral 'benefit' that Labour and the Liberal Democrats currently enjoy. On the reduce crime/protect rights scale, the Conservatives are on average seen as being closer than either Labour or the Liberal Democrats to where voters place themselves. This said, the Conservatives could still narrow the gap ¾ and thereby gain further electoral leverage ¾ by becoming even more clearly in favour of severe punishment for criminals.

If the Conservatives were to effect these or similar changes, what would the likely consequences be for their vote share? We can use the coefficients shown in Tables 4 and 5 to simulate what is likely to happen. Table 7 describes the position at the time of the 2005 general election and then makes different assumptions about how the key independent variables in Tables 4 and 5 might change in the run-up to the next general election. The Conservatives have some degree of control themselves over three of these variables ¾ the leader's ratings and the two ideological distance measures. The fourth variable in the simulation is average personal financial expectations, on the assumption that these expectations may decline if the economy falters between now and the next general election. The precise calculations involved in the simulations are shown in Table 7. The implications of the simulations are clear. As simulation 1 shows, if the Conservative shift their ideological positions on tax/spend and reduce crime/protect rights, and if the Conservative leader achieves an average ratings score of around 5.5, then the Conservatives can expect to push their vote share up to around 36 percent ¾ not enough to win an election, but an improvement on 2005. Simulation 2 shows that, if David Cameron can improve his leader ratings to 6 on the 0-10 scale ¾ a feat which, at the time of writing, does not seem entirely implausible ¾ the Conservatives could push their vote share up to 37 percent. As simulation 3 shows, a modest economic downturn could add a further percentage point to Conservative fortunes, in which case the Party could be making a serious bid for office in 2009/10. In short, these simulations suggest that, with appropriate ideological changes, the Conservative could be challenging Labour seriously at the next general election.

Conclusions

In this paper, I have deliberately mixed fairly formal statistical estimation with impressionistic qualitative evidence derived from interviews with a non-representative set of informants from the public sector. The formal statistical analysis shows that it is clearly within the Conservatives' capacity to change themselves so that they can be more appealing to the electorate. They can certainly elect a more attractive leader ¾ and in David Cameron they have already succeeded in doing so. The statistical analysis provided shows that in order to convince voters that they are capable of managing government effectively, the Conservatives need to follow Labour's example in the late 1980s and 1990s and to adjust their ideological position. In the 1980s, Labour's 'extreme' ideological positioning led many voters to conclude that they were simply unsuited to running the country. It is likely that since the mid-1990s, the Conservatives have suffered from a similar affliction ¾ though at the opposite end of the spectrum. Most voters in Britain want good public services, and they are prepared to pay for them. The Conservatives, if they wish to be seen as managerially competent, must get closer to the 'average voter' on the crucial 'tax/spend' dimension of ideology. If they do, as the simulations show, they can expect to make considerable gains in terms of their vote share at the next general election. Crucially, however, the Conservatives should understand that by moving in a more 'progressive' direction on tax/spend, they do not have to abandon their traditional, 'strong' stance on law and order. On the contrary, they can afford to maintain a strong anti-criminal stance and continue to gain in electoral terms.

Because of a lack of suitable data, the simulations could not address the possible impact of the Conservatives becoming more inclusive and tolerant in terms of their position towards minorities and alternative lifestyles. However, as the factor analysis in Table 3 showed, voters clearly differentiate between tolerance/intolerance of difference and the need to deal firmly with criminals and threats to public security. Just as the Conservatives can benefit from moving closer to the average voter on tax/spend, it is highly likely that the party can also benefit from moving closer to the average voter by becoming more tolerant of difference. Crucially, because public attitudes to criminals and to minorities are largely independent of one another, it can do this without needing to change its strong law and order stance.

The key point of the simulations, however, is that they suggest the likelihood of a hung parliament after the next general election. The models on which the simulations are based make no specific assumptions about the respective 'costs' to Labour and the Liberal Democrats if the Conservatives were to increase their vote share by five percentage points. In fact, on almost any combination of assumptions (Labour lose 5 points, LibDems lose none; through to Labour lose no points; LibDems lose 5), a hung parliament remains the most likely outcome. What will make a difference both to the Conservatives and to the next election outcome is the Conservatives' ability to draw greater support from groups that they have traditionally tended to ignore.

It is in this context that public sector workers' discontent could be converted into an important electoral asset for the Conservative Party. The qualitative 'evidence' provided here in support of this claim has necessarily been limited. Nonetheless, the public sector failings and disillusionment outlined above offer a tremendous political opportunity for the Conservative Party. The simulations show that the Conservatives could reach around 38 percent of the popular vote in 2009/10 by changing themselves. If the party can add a further 2 or 3 percent to its vote share by a vigorous campaign that insists on the need to end the absurdities of targets, league tables and micro-management in the public sector, David Cameron could well find himself in Downing Street in four years' time.

Acknowledgements

I am indebted to the Journal's anonymous reviewers for their comments on an earlier draft of this paper.

Footnotes

¹Note that the fifth component variable of this dimension refers to suspected terrorists, a variable that also scores relatively highly on the intolerance of difference factor.

²Recent evidence suggests that the latter interpretation is probably more useful in the UK context (see Clarke et al., 2004).

³There are two other minor differences in the effects shown in Tables 4 and 5. First, in the Conservative/not equation in Table 4, none of the issue priority terms was significant. In the Conservative competence equation in Table 5, the 'economic issues' term is significant. However, this is not robust and is highly sensitive to the model specification. If the no-significant terms in the equation are dropped and a reduced form equation is re-estimated, the economic issues terms loses significance. All the other significant terms in Table 5 retain their significance in this reduced form. Second, there are minor changes in the significance of the demographics. Manual status and celtic location are significant in the vote equation; age and celtic location are significant in the competence equation.

Campbell, A., Converse, P., Miller, W. and Stokes, D. (1960) The American Voter, New York: John Wiley & Sons.

Figure 1 Illustrative Positions on the Liberal/Authoritarian Scale. Question: Some people think that reducing crime is more important than protecting the rights of people accused of committing crimes. Other people think that protecting the rights of accused people is more important than reducing crime. On the 0-10 scale, where would you place yourself on this scale? [And where would you place¼ the Conservative Party¼ the Labour Party¼ the Liberal Democrat Party].

Table 1 Vote by public sector employment status, 1982, 1997 and 2005

Table 2 Optimism/pessimism about the economy by public sector employment status, 1997 and 2005

Table 3 Factor analysis of selected ideological scale items from the 2005 British election study post-election cross-sectional survey

Table 4 Logistic regression model of conservative/not voting in the 2005 general election

Table 5 OLS regression model of conservative managerial competence ratings in 2005

Table 6 Average scores on tax/spend and reduce crime/protect rights scales, 2005

Table 7 Predicted Conservative vote shares in 2009/10