Return to: Search Page or to: Table of Contents Vol. 16, Issue 6

Jeremy J. Albright, "The multidimensional nature of party competition," Party Politics, 16 (November, 2010), 699-719. [Available at ]

First paragraph:
In models of party behavior and voter-elite linkages, comparative political scientists often prefer to constrain analysis to a single policy dimension in order to maximize the comparability of measurements across countries and minimize problems of cyclical social choice preference orderings (Huber and Powell, 1994). Consequently, empirical tests of models relating to coalition formation (Laver and Schofield, 1990), campaign strategies (Budge, 1994), voter-party congruence (Powell, 2000), and extremist parties (Adams et al., 2006) often assume a single ideological dimension. However, it is not hard to imagine that collapsing the complexity of political competition to one axis can obfuscate a great deal of relevant phenomena. Surprisingly, very little work exists that attempts to document just how much information is lost.

Figures and Tables:
Figure 1. Scree plots for comparative manifestos project data
Table 1. Results from principal components analysis of CMP data
Figure 2. Count of issue references: pooled CMP data
Figure 3. Count of issue references by country
Table 2. Linear spline models of left-right estimates with knot at 1975
Figure 4. Left-right scores by year: social democratic and non-social democratic parties
Table 3. Linear spline models of uncertainty estimates with knot at 1975

Last Paragraph:
Finally, none of this is to say that left-right is meaningless. On the contrary, the scale's durability speaks to its intuitive appeal and its helpfulness for providing rough assessments about the positions of political actors. The purpose of this paper has been to show what gets lost when one limits analysis to a single dimension. The complexity of comparing the strategies of political actors in different contexts makes definitive statements about the dimensionality of competition difficult. Questions about the number of relevant dimensions are decades old, and with questionable progress having been made toward a definitive answer, it is perhaps doubtful that a concrete, time-invariant conclusion will ever emerge. Thus, empirical studies of political behavior and democratic institutions will need to carefully justify the decisions they make concerning the policy domains assumed to be relevant and why more fully specified models are not appropriate.

Last updated October 2010