Electoral Systems and Party Systems: A Study of Twenty-Seven Democracies, 1945-1990

The effective threshold is Lijphart's key independent variable, merging district magnitude and legal electoral thresholds into a single measure of the actual barrier small parties face. He argues these two variables are functionally equivalent: they both set limits to the representation of small parties [citation:14]. The higher the effective threshold, the higher the disproportionality of electoral outcomes and the more frequent the creation of parliamentary majorities [citation:5]. Lijphart follows Taagepera and Shugart in using this composite measure, though he acknowledges that the formula produces a problematic 50 percent threshold for single-member districts, which he considers too high [citation:14].

Lijphart finds that the electoral system's influence on the effective number of elective parties is weaker than Duverger's Law would predict, and the link to the effective number of parliamentary parties is only somewhat stronger [citation:5]. The most important conclusion is that multipartism or two-partyism is not caused only by proportional representation or plurality systems; many intervening variables exist, including Duverger's psychological factor, effective thresholds, and the mutual relationship between electoral and party systems [citation:5]. This qualifies rather than refutes Duverger, showing that the electoral system matters but is not the sole determinant.

Alberto Penadés argues that Lijphart's conclusions about the impact of electoral systems on party systems are neither convincing nor substantiated by his data. He contends that both the formulation of hypotheses and the empirical testing are flawed, though he concedes that the conclusions concerning the determinants of proportionality appear better founded [citation:3]. Lijphart responded directly to this critique, defending his decision to merge district magnitude and electoral threshold into the effective threshold, while acknowledging that his formula for translating district magnitude into effective threshold was 'not ideal' [citation:14].

Lijphart decided to consider every significant change in electoral system—a change in formula or a change of at least 20 percent in one of the other dimensions—as a new electoral system [citation:17]. This methodological choice produced a large number of cases (70 systems from 27 countries) that he treats as units of analysis. One strategy focuses on the consequences of electoral system changes within individual countries, allowing him to examine what happens to party systems when the rules change [citation:4]. This within-country approach complements the bivariate and multivariate analyses of all cases.

The final chapter, 'Electoral Engineering: Limits and Possibilities,' addresses the practical implications of his findings. Lijphart ends the book with a series of recommendations for 'electoral engineers,' but he is careful to note the limits of institutional reform [citation:8][citation:18]. His recommendations warrant close reading not only for newly democratizing countries of Eastern Europe but also for reformers in Italy, Japan, and New Zealand and would-be reformers elsewhere [citation:18]. The book's empirical finding that electoral systems are neither as diverse nor as complex as is often assumed provides grounds for cautious optimism about reform while warning against overestimating what institutions alone can achieve.

Electoral Systems and Party Systems represents the definitive empirical treatment of a subject Lijphart had explored in numerous earlier contributions [citation:17]. It provides the detailed evidence for the institutional arguments developed in Democracy in Plural Societies (1977) and later synthesized in Patterns of Democracy (1999). The book is part of the Comparative European Politics series and was published by Oxford University Press in 1994 [citation:1][citation:7]. It won the American Political Science Association George H. Hallet Award in 2010, recognizing its lasting contribution to the field [citation:8].