Simon Franzmann and André Kaiser, "Locating Political Parties in Policy Space: A Reanalysis of Party Manifesto Data," Party Politics, 12 (March, 2006), 163-188.

First Paragraph:
Empirical research in comparative politics is increasingly based on spatial concepts that make use of political parties' positions in policy space (Enelow and Hinich, 1984; Hinich and Munger, 1997; Laver and Shepsle, 1996; Macdonald et al., 1991; Powell, 2000; Tsebelis, 2002). Therefore, it is indispensable to have as much reliable data on policy positions to hand as possible - both for international and for inter-temporal comparisons. This information can be gathered from a variety of sources, including expert, elite and mass surveys, where observers are questioned about their perceptions of parties' policy positions, or roll-call data where data on issue preferences are derived from legislative behavior (Budge, 2001; Laver and Garry, 2000; Mair, 2001). However, all these methods come with considerable problems: how can we be certain that observers' perceptions are based on actual policy positions rather than on behavior which is itself already structured by the political process? And how do we make data available for as many party systems as well as issue dimensions and for as long a time series as possible?

Figures and Tables:
Figure 1. Twelve steps to extracting party position scores in an n-dimensional policy space
Table 1. Regression for welfare state expansion (per504) in Germany
Table 2. Regression for environmental protection (per501) in Germany 1949-98
Table 3. Regression for environmental protection (per501) in Germany 1949-80
Table 4. Regression for environmental protection (per501) in Germany 1983-98
Table 5. Issue character of the German party system
Figure 2. Germany's left-right party positions 1949-98
Figure 3. Germany's left-right party positions, CMP dataset
Figure 4. Germany's left-right party positions, CMP dataset smoothed
Table 6. Correlations of left-right scales with CMP dataset and mass surveys in Germany 1976-98
Table 7. Issue character of the Italian party system
Figure 5. Italy's left-right party positions 1946-98
Table 8. Issue character of the Swedish party system
Figure 6. Sweden's left-right party positions 1948-98
Table 9. Issue character of the British party system
Figure 7. Britain's left-right party positions 1945-97

First Paragraph of Conclusion:
Starting from a number of deficiencies in the CMP approach to studying left-right policy scales, we develop a new proposal, one that avoids an inductive investigation based on factor analysis. We ground our additive model on the distinction between position and valence issues and show how this can be applied in a way that takes account of the possibility that the content of the left-right dimension varies from country to country as well as over time. We accept the proposal in the literature that election programs are used by parties to signal policy changes to the electorate. However, in contrast to previous research we argue that because of the asymmetry of information between parties and voters regarding the exact policy positions of parties these signals overemphasize the extent of change. We therefore propose a smoothing procedure to measure positional scores. The approach is illustrated with respect to the British, German, Italian and Swedish party systems.