A theory of spatial competition with biased voters: Party policies viewed temporally and comparatively

British Journal of Political Science

Volumn 31, Issue 1, 2001, Pages 121-158

  Adams, James ^a  

^a UNIVERSITY OF CALIFORNIA   (United States)

Author keywords

[No Author keywords available]

Indexed keywords

EID: 0035624975     PISSN: 00071234     EISSN: None     Source Type: Journal    
DOI: 10.1017/S0007123401000060     Document Type: Article

References (48)

1. The Manifesto Research Group, under the direction of Ian Budge, commenced work in 1979, and has published the following reports: Ian Budge, David Robertson and D.J. Hearl, eds, Ideology, Strategy, and Party Change in 19 Democracies (Cambridge: Cambridge University Press, 1987); Michael Laver and Ian Budge, eds, Party Policies and Government Coalitions (London: Macmillan, 1992); H-D. Klingemann, R. I. Hoffbert, Ian Budge et al., Parties, Policies, and Democracy (Boulder, Colo.: Westview Press, 1994).
2. The Manifesto Research Group, under the direction of Ian Budge, commenced work in 1979, and has published the following reports: Ian Budge, David Robertson and D.J. Hearl, eds, Ideology, Strategy, and Party Change in 19 Democracies (Cambridge: Cambridge University Press, 1987); Michael Laver and Ian Budge, eds, Party Policies and Government Coalitions (London: Macmillan, 1992); H-D. Klingemann, R. I. Hoffbert, Ian Budge et al., Parties, Policies, and Democracy (Boulder, Colo.: Westview Press, 1994).
3. The Manifesto Research Group, under the direction of Ian Budge, commenced work in 1979, and has published the following reports: Ian Budge, David Robertson and D.J. Hearl, eds, Ideology, Strategy, and Party Change in 19 Democracies (Cambridge: Cambridge University Press, 1987); Michael Laver and Ian Budge, eds, Party Policies and Government Coalitions (London: Macmillan, 1992); H-D. Klingemann, R. I. Hoffbert, Ian Budge et al., Parties, Policies, and Democracy (Boulder, Colo.: Westview Press, 1994).
4. In two-party systems, vote-and office-seeking goals are typically compatible, in that parties maximize their chance of holding office by maximizing votes. In multiparty systems, in which attaining membership in the governing coalition typically revolves around post-election negotiations, there is no necessary connection between electoral support and the likelihood of inclusion in the governing coalition. For a development of these points, see Kaare Strom, 'A Behavioural Theory of Competitive Political Parties', American Political Science Review, 34 (1990), 565-98.
5. See Ian Budge, 'A New Theory of Party Competition: Uncertainty, Ideology, and Policy Equilibria Viewed Comparatively and Temporally', British Journal of Political Science, 24 (1994), 443-67.
6. David Chapman, 'Models of the Working of a Two-Party Electoral System - I', Papers in Non-Market Decision-making (1967), 19-38, and 'Models of the Working of a Two-Party Electoral System-II', Public Choice, 1 (1968), 19-38.
7. David Chapman, 'Models of the Working of a Two-Party Electoral System - I', Papers in Non-Market Decision-making (1967), 19-38, and 'Models of the Working of a Two-Party Electoral System-II', Public Choice, 1 (1968), 19-38.
8. See pp. 458-67 in Budge, Robertson and Hearl, eds., Ideology, Strategy, and Party Change, for complete details of these coding procedures.
9. See Table 1 in Budge, 'A New Theory of Party Competition'.
10. With respect to policy differentiation, the central point is that under the standard Downsian model, for any number of parties the left-and rightmost parties always converge towards the position of their nearest competitor. However, when four or more parties compete, then if an equilibrium exists the left-and right-most parties will be differentiated from each other, as Gary Cox shows in 'Electoral Equilibria under Alternative Voting Institutions', American Journal of Political Science, 31 (1987), 82-109. However, in 'The Principle of Minimum Differentiation Reconsidered: New Developments in the Theory of Spatial Competition', Review of Economic Studies, 41 (1975), 27-49, Curtis Eaton and Richard Lipsey show that such multiparty equilibria will exist only under extremely unusual or contrived circumstances.
11. With respect to policy differentiation, the central point is that under the standard Downsian model, for any number of parties the left-and rightmost parties always converge towards the position of their nearest competitor. However, when four or more parties compete, then if an equilibrium exists the left-and right-most parties will be differentiated from each other, as Gary Cox shows in 'Electoral Equilibria under Alternative Voting Institutions', American Journal of Political Science, 31 (1987), 82-109. However, in 'The Principle of Minimum Differentiation Reconsidered: New Developments in the Theory of Spatial Competition', Review of Economic Studies, 41 (1975), 27-49, Curtis Eaton and Richard Lipsey show that such multiparty equilibria will exist only under extremely unusual or contrived circumstances.
12. The spatial modelling literature is voluminous, and cannot be fully reviewed here. However, five different approaches which attempt to account for policy divergence are those of John Aldrich, 'A Spatial Model with Party Activism', American Political Science Review, 41 (1983), 974-90, which considers the role of party activists; Thomas Palfrey, 'Spatial Equilibrium with Entry', Review of Economic Studies, 51 (1984), 139-56, who analyses the effect of allowing new parties/candidates to enter the race; Donald Wittman, 'Candidate Motivation: A Synthesis of Alternatives', American Political Science Review, 77 (1983), 142-57, who posits that candidates/parties have policy goals; David Austen-Smith and Jeffrey Banks, 'Elections, Coalitions, and Legislative Outcomes ', American Political Science Review, 82 (1988), 405-22, which explores implications of postelection bargaining over governing coalitions;
13. The spatial modelling literature is voluminous, and cannot be fully reviewed here. However, five different approaches which attempt to account for policy divergence are those of John Aldrich, 'A Spatial Model with Party Activism', American Political Science Review, 41 (1983), 974-90, which considers the role of party activists; Thomas Palfrey, 'Spatial Equilibrium with Entry', Review of Economic Studies, 51 (1984), 139-56, who analyses the effect of allowing new parties/candidates to enter the race; Donald Wittman, 'Candidate Motivation: A Synthesis of Alternatives', American Political Science Review, 77 (1983), 142-57, who posits that candidates/parties have policy goals; David Austen-Smith and Jeffrey Banks, 'Elections, Coalitions, and Legislative Outcomes ', American Political Science Review, 82 (1988), 405-22, which explores implications of postelection bargaining over governing coalitions;
14. The spatial modelling literature is voluminous, and cannot be fully reviewed here. However, five different approaches which attempt to account for policy divergence are those of John Aldrich, 'A Spatial Model with Party Activism', American Political Science Review, 41 (1983), 974-90, which considers the role of party activists; Thomas Palfrey, 'Spatial Equilibrium with Entry', Review of Economic Studies, 51 (1984), 139-56, who analyses the effect of allowing new parties/candidates to enter the race; Donald Wittman, 'Candidate Motivation: A Synthesis of Alternatives', American Political Science Review, 77 (1983), 142-57, who posits that candidates/parties have policy goals; David Austen-Smith and Jeffrey Banks, 'Elections, Coalitions, and Legislative Outcomes ', American Political Science Review, 82 (1988), 405-22, which explores implications of postelection bargaining over governing coalitions;
15. The spatial modelling literature is voluminous, and cannot be fully reviewed here. However, five different approaches which attempt to account for policy divergence are those of John Aldrich, 'A Spatial Model with Party Activism', American Political Science Review, 41 (1983), 974-90, which considers the role of party activists; Thomas Palfrey, 'Spatial Equilibrium with Entry', Review of Economic Studies, 51 (1984), 139-56, who analyses the effect of allowing new parties/candidates to enter the race; Donald Wittman, 'Candidate Motivation: A Synthesis of Alternatives', American Political Science Review, 77 (1983), 142-57, who posits that candidates/parties have policy goals; David Austen-Smith and Jeffrey Banks, 'Elections, Coalitions, and Legislative Outcomes ', American Political Science Review, 82 (1988), 405-22, which explores implications of postelection bargaining over governing coalitions;
16. Norman Schofield and Itai Sened, 'Political Equilibrium in Multiparty Competition, with Both Electoral and Coalition Risk' (paper presented at the Public Choice Meetings, New Orleans, 1998), which considers the role of electoral risk in party position-taking.
17. See Angus Campbell, Philip E. Converse, Warren E. Miller and Donald E. Stokes, The American Voter (New York: John Wiley, 1960). For applications to British politics, see David Butler and Donald Stokes, Political Change in Britain (New York: St Martin's Press, 1969); Richard Rose and Ian McAllister, The Loyalties of Voters: A Lifetime Learning Model (London: Sage Publications, 1990), and John Bartle, 'Left-Right Position Matters, But Does Social Class? Causal Models of the 1992 British General Election', British Journal of Political Science, 28 (1998), 501-29.
18. See Angus Campbell, Philip E. Converse, Warren E. Miller and Donald E. Stokes, The American Voter (New York: John Wiley, 1960). For applications to British politics, see David Butler and Donald Stokes, Political Change in Britain (New York: St Martin's Press, 1969); Richard Rose and Ian McAllister, The Loyalties of Voters: A Lifetime Learning Model (London: Sage Publications, 1990), and John Bartle, 'Left-Right Position Matters, But Does Social Class? Causal Models of the 1992 British General Election', British Journal of Political Science, 28 (1998), 501-29.
19. See Angus Campbell, Philip E. Converse, Warren E. Miller and Donald E. Stokes, The American Voter (New York: John Wiley, 1960). For applications to British politics, see David Butler and Donald Stokes, Political Change in Britain (New York: St Martin's Press, 1969); Richard Rose and Ian McAllister, The Loyalties of Voters: A Lifetime Learning Model (London: Sage Publications, 1990), and John Bartle, 'Left-Right Position Matters, But Does Social Class? Causal Models of the 1992 British General Election', British Journal of Political Science, 28 (1998), 501-29.
20. See Angus Campbell, Philip E. Converse, Warren E. Miller and Donald E. Stokes, The American Voter (New York: John Wiley, 1960). For applications to British politics, see David Butler and Donald Stokes, Political Change in Britain (New York: St Martin's Press, 1969); Richard Rose and Ian McAllister, The Loyalties of Voters: A Lifetime Learning Model (London: Sage Publications, 1990), and John Bartle, 'Left-Right Position Matters, But Does Social Class? Causal Models of the 1992 British General Election', British Journal of Political Science, 28 (1998), 501-29.
21. I note, however, that several recent papers on two-party competition incorporate the notion of voter bias. These include papers by Scott Feld and Bernard Grofman, 'Incumbency Advantage, Voter Loyalty, and the Benefit of the Doubt,' Journal of Theoretical Politics, 3 (1991), 115-37; Tim Groseclose, 'Character, Charisma, and Candidate Location' (unpublished); Mark Berger, Michael Munger and Richard Potthoff, 'Why Even Downsian Candidates Will Generally Choose Different Platforms' (presented at the Annual Meeting of the Public Choice Society, New Orleans, 1999); Stuart Elaine Macdonald and George Rabonowitz, 'Solving the Paradox of Nonconvergence: Valence, Position, and Direction in Democratic Politics', Electoral Studies, 17 (1998), 281-300. A major difference between these papers and my own is that each of them assumes that all voters are biased towards the same party or candidate, while here 1 assume that different voters are biased towards different parties.
22. I note, however, that several recent papers on two-party competition incorporate the notion of voter bias. These include papers by Scott Feld and Bernard Grofman, 'Incumbency Advantage, Voter Loyalty, and the Benefit of the Doubt,' Journal of Theoretical Politics, 3 (1991), 115-37; Tim Groseclose, 'Character, Charisma, and Candidate Location' (unpublished); Mark Berger, Michael Munger and Richard Potthoff, 'Why Even Downsian Candidates Will Generally Choose Different Platforms' (presented at the Annual Meeting of the Public Choice Society, New Orleans, 1999); Stuart Elaine Macdonald and George Rabonowitz, 'Solving the Paradox of Nonconvergence: Valence, Position, and Direction in Democratic Politics', Electoral Studies, 17 (1998), 281-300. A major difference between these papers and my own is that each of them assumes that all voters are biased towards the same party or candidate, while here 1 assume that different voters are biased towards different parties.
23. I note, however, that several recent papers on two-party competition incorporate the notion of voter bias. These include papers by Scott Feld and Bernard Grofman, 'Incumbency Advantage, Voter Loyalty, and the Benefit of the Doubt,' Journal of Theoretical Politics, 3 (1991), 115-37; Tim Groseclose, 'Character, Charisma, and Candidate Location' (unpublished); Mark Berger, Michael Munger and Richard Potthoff, 'Why Even Downsian Candidates Will Generally Choose Different Platforms' (presented at the Annual Meeting of the Public Choice Society, New Orleans, 1999); Stuart Elaine Macdonald and George Rabonowitz, 'Solving the Paradox of Nonconvergence: Valence, Position, and Direction in Democratic Politics', Electoral Studies, 17 (1998), 281-300. A major difference between these papers and my own is that each of them assumes that all voters are biased towards the same party or candidate, while here 1 assume that different voters are biased towards different parties.
24. I note, however, that several recent papers on two-party competition incorporate the notion of voter bias. These include papers by Scott Feld and Bernard Grofman, 'Incumbency Advantage, Voter Loyalty, and the Benefit of the Doubt,' Journal of Theoretical Politics, 3 (1991), 115-37; Tim Groseclose, 'Character, Charisma, and Candidate Location' (unpublished); Mark Berger, Michael Munger and Richard Potthoff, 'Why Even Downsian Candidates Will Generally Choose Different Platforms' (presented at the Annual Meeting of the Public Choice Society, New Orleans, 1999); Stuart Elaine Macdonald and George Rabonowitz, 'Solving the Paradox of Nonconvergence: Valence, Position, and Direction in Democratic Politics', Electoral Studies, 17 (1998), 281-300. A major difference between these papers and my own is that each of them assumes that all voters are biased towards the same party or candidate, while here 1 assume that different voters are biased towards different parties.
25. Donald Stokes, 'Spatial Models of Party Competition', American Political Science Review, 57 (1963), 19-28.
26. For results supporting these propositions see Tables 2 and 5 in Guy Whitten and Harvey Palmer, 'Heightening Comparativists' Concern with Model Choice: Voting Behaviour in Great Britain and the Netherlands', American Journal of Political Science, 40 (1996), 231-60.
27. In recent years much work on spatial theory has relaxed the assumption of perfect information by introducing probabilistic voting - i.e., situations in which parties cannot predict voters' choice with certainty, but can only attach probabilities to various outcomes. The central theoretical result for probabilistic voting models is that party equilibria are likely to exist in which all parties converge to identical sets of policy positions. In addition, these studies typically do not consider temporal patterns of party competition. Many of the fundamental results on probabilistic voting are presented in James Enelow and Melvin Hinich, The Spatial Theory of Voting (New York: Cambridge University Press, 1984). For an extension to situations involving three or more candidates/parties, see Tse-min Lin, James Enelow and Han Dorussen, 'Equilibrium in Multicandidate Probabilistic Spatial Voting', Public Choice, 98 (1999), 59-82.
28. In recent years much work on spatial theory has relaxed the assumption of perfect information by introducing probabilistic voting - i.e., situations in which parties cannot predict voters' choice with certainty, but can only attach probabilities to various outcomes. The central theoretical result for probabilistic voting models is that party equilibria are likely to exist in which all parties converge to identical sets of policy positions. In addition, these studies typically do not consider temporal patterns of party competition. Many of the fundamental results on probabilistic voting are presented in James Enelow and Melvin Hinich, The Spatial Theory of Voting (New York: Cambridge University Press, 1984). For an extension to situations involving three or more candidates/parties, see Tse-min Lin, James Enelow and Han Dorussen, 'Equilibrium in Multicandidate Probabilistic Spatial Voting', Public Choice, 98 (1999), 59-82.
29. Ironically, Downs himself, in pp. 107-13 of An Economic Theory of Democracy (New York: Harper and Row, 1957), argued that parties are constrained to present consistent policy proposals in order to develop a reputation for trustworthiness, without which voters would not support them in future elections. However, most subsequent spatial models ignore this prohibition against policy change.
30. Two other stylized features of this distribution are that it includes no independent (unbiased) voters, and that it overstates the proportion of LD sympathizers in the British electorate. However, I show below that the heuristic arguments I develop for this stylized distribution extend to empirical data on British voters' preferences.
31. Of course, one might explain parties' policy shifts by arguing that parties reach successive policy equilibria, but that these equilibrium positions change over time in response to shifts in the distribution of voters' policy beliefs. However, this explanation has difficulty accounting for historical instances in which parties dramatically shifted their policy priorities despite relative policy stability in the electorate. Some examples are the rightward shifts of the US Republicans (during the 1980s) and the British Conservatives (during the late 1970s), as well as the French Socialists' shift towards the centre in 1983. In each case, public opinion polling suggests that while these parties' policies changed dramatically, voters' underlying preferences did not. For evidence on the relative stability of the British, French and American publics' left-right orientations during this period, see Russell Dalton, Citizen Politics in Western Democracies (London: Chatham House, 1996), Figure 6.4, pp. 137.
32. One might think that Labour could do better by leapfrogging both rival parties, thereby locating to the right of the Conservatives. However, the fact of voter biases precludes such a strategy. To see this, note that because all voters to the right of 6.5 are Conservative sympathizers, Labour would have to differentiate its policies from those of the Conservatives in order to attract these voters' support - the same dilemma that prevents the Liberal Democrats from leapfrogging. Specifically, with the Conservatives located at 6.0, Labour maximizes its support on the right by locating near 7.0, thereby attracting support from most voters to the right of 7.0. However, this position is less electorally attractive than the location 3.6. For a further discussion of this point, see James Adams, 'Partisan Voting and Multiparty Spatial Competition: The Pressure of Responsible Parties', Journal of Theoretical Politics, 10 (1998), 5-31, at pp. 12-13.
33. Specifically, if the Liberal Democrats shift in the direction of the Labour party, they will lose more votes to the Conservatives on the right than they gain from Labour on the left, because voters are much more heavily distributed near the Conservatives, who are located nearer to the centre than is Labour.
34. For a summary of the World Values Survey data that illustrates the centrism of Western publics, see Dalton, Citizen Politics, p. 137. This conclusion is also supported by studies that use alternative methodologies to estimate voters' spatial locations. See, for instance, the spatial maps of voter positions in eight democratic polities presented in Norman Schofield's 'A Comparison of Majoritarian and Proportional Electoral Systems Based on Spatial Modeling and "Rational" Politicians', presented at the Conference on Constitutional Issues in Modern Democracies, Messina, Italy, 1997. These maps, which were generated via multidimensional scaling analyses of voter survey data, confirm that in all cases the voter distributions are centrist, not polarized, but that voters are not overwhelmingly concentrated at the centre.
35. For a summary of the World Values Survey data that illustrates the centrism of Western publics, see Dalton, Citizen Politics, p. 137. This conclusion is also supported by studies that use alternative methodologies to estimate voters' spatial locations. See, for instance, the spatial maps of voter positions in eight democratic polities presented in Norman Schofield's 'A Comparison of Majoritarian and Proportional Electoral Systems Based on Spatial Modeling and "Rational" Politicians', presented at the Conference on Constitutional Issues in Modern Democracies, Messina, Italy, 1997. These maps, which were generated via multidimensional scaling analyses of voter survey data, confirm that in all cases the voter distributions are centrist, not polarized, but that voters are not overwhelmingly concentrated at the centre.
36. Many analysts argue that respondents outside the United States interpret the party identification question as an enquiry about their current vote intention, an interpretation that invalidates this item as an independent influence on the vote. Other analysts argue that, for surveys in which the party identification question has a similar wording to that used in American surveys (as is the case for the French survey employed here), this item taps longer-range dispositions independent of current vote. See, for instance, the exchange between Christopher Fleury and Michael Lewis-Beck, 'Anchoring the French Voter: Ideology Versus Party', Journal of Politics, 55 (1993), 1100-09, and Philip Converse and Roy Pierce, 'Comment on Fleury and Lewis-Beck', Journal of Politics, 55 (1993), 1110-17
37. Many analysts argue that respondents outside the United States interpret the party identification question as an enquiry about their current vote intention, an interpretation that invalidates this item as an independent influence on the vote. Other analysts argue that, for surveys in which the party identification question has a similar wording to that used in American surveys (as is the case for the French survey employed here), this item taps longer-range dispositions independent of current vote. See, for instance, the exchange between Christopher Fleury and Michael Lewis-Beck, 'Anchoring the French Voter: Ideology Versus Party', Journal of Politics, 55 (1993), 1100-09, and Philip Converse and Roy Pierce, 'Comment on Fleury and Lewis-Beck', Journal of Politics, 55 (1993), 1110-17
38. The principal investigators for the 1992 BGES were Anthony Heath, Roger Jowell, John K. Curtice, Jack A. Brand and James C. Mitchell. The data analysed here were made available through a computer file produced by Economic and Social Research Council Data Archive, Colchester, and distributed through the Inter-university Consortium for Political and Social Research, Ann Arbor, Mich. The other issues included in the 1992 BGES relate to national defence, women's rights, the European Community, taxation, tradeoffs between unemployment and inflation, welfare and redistribution. In addition to the analyses reported here, I analysed party policy strategies for the latter four policy variables, which presumably tap the left-right economic dimension. Results for the redistribution and taxation variables were quite similar to the results I report for nationalization; results for unemployment/inflation and for welfare were different, presumably because survey responses were skewed sharply to the left on these items. Because the distribution of British voters' left-right orientations is presumably not skewed sharply to the left, the results for nationalization, redistribution and taxation are probably more indicative of what an analysis based upon left-right would show. These analyses are available from the author upon request.
39. For the empirical distribution, party sympathizers were defined as those who thought of themselves as Labour, Conservative or Liberal Democrat as well as those who reported feeling closer to one of these parties. Independents were those who answered 'No/none' to both questions. Respondents who replied 'Don't know' or who did not answer these questions are excluded from this analysis, as are respondents who identified with one of the smaller parties. The total number of respondents thereby excluded is fewer than a hundred.
40. See John Huber and Ronald Inglehart, 'Expert Interpretations of Party Space and Party Locations in 42 Democracies', Party Politics, 1 (1995), 73-111, Appendix 2. Note that these expert placements, which were based on a 1-10 scale, have been rescaled to reflect the 1-11 scale used in the British survey.
41. See, for instance, Ken Kollman, John H. Miller and Scott E. Page, 'Adaptive Parties in Spatial Elections', American Political Science Review, 86 (1992), 929-37.
42. In the case of the random policies model, the similarity of the parties' average positions over the course of their policy trajectories masks the fact that any given time period the parties may present quite divergent policies. However, this does not change the fact that under this model the parties display no systematic tendencies to locate in differing regions of the policy space.
43. Respondents' party placements were used because of the lack of expert ratings of the parties' positions for the 1988 election. However, the mean party placements used here are quite similar to the expert ratings (which were collected in 1991-92) reported on p. 97 of Huber and Inglehart's 'Expert Locations of Party Space'.
44. This spatial mapping results when the parties move in the sequence: Communist, Socialist, UDF, RPR, National Front. Alternative simulations with different party sequences produced comparable results.
45. For the simulations on the locally Downsian model I assumed that parties could vary their positions by a maximum of 1.5 units between elections, along the 1-7 left-right scale. Simulations using alternative restrictions on party policy shifts suggest that results for this model are quite sensitive to the degree of restriction placed upon parties. When parties were restricted to a maximum shift of 0.8 units or less between elections, then predictions for the locally Downsian model matched the parties' observed behaviour reasonably well, although the predicted degree of leapfrogging was unrealistically high. However, this restriction to shifts of at most 0.8 units appears unrealistically strong, given the actual party policy shifts as measured by the MRG.
46. Stokes, 'Spatial Models of Party Competition', p. 373. In this regard, I note that the MRG's coding is based largely around saliency ideas - i.e., parties compete by emphasizing ideas such as fighting crime and cutting unemployment, which their opponents cannot oppose directly - which are related to Stokes's notion of valence issues. However, other kinds of valence issues, such as voters' perceptions of party leaders' competence and integrity, are not incorporated into the MRG coding, and it is these sorts of judgements I have in mind here. I thank Ian Budge for drawing my attention to this point.
47. See for instance John Bartle, Ivor Crewe and Anthony King, 'Was It Blair Who Won It? Leadership Effects in the 1997 British General Election' (presented at the 1997 Annual Meeting of the American Political Science Association, Washington, DC) and Anthony King, 'Why Labour Won - At Last', in New Labour Triumphs: Britain at the Polls (London: Chatham House, 1997).
48. See for instance John Bartle, Ivor Crewe and Anthony King, 'Was It Blair Who Won It? Leadership Effects in the 1997 British General Election' (presented at the 1997 Annual Meeting of the American Political Science Association, Washington, DC) and Anthony King, 'Why Labour Won - At Last', in New Labour Triumphs: Britain at the Polls (London: Chatham House, 1997).