Predicting Party Sizes: The Logic of Simple Electoral Systems

The seat product is the product of assembly size (S) and average district magnitude (M). Taagepera derives that the effective number of parties winning seats is the sixth root of the seat product: Ns = (MS)^(1/6) [citation:16][citation:19]. From this, the largest party's seat share is predicted as the inverse eighth root of the seat product, and cabinet duration follows from the effective number of parties . The model predicts long-term world averages, not individual election outcomes, and serves as an 'ignorance-based' benchmark in the absence of other information [citation:19].

Seats and Votes (1989) identified district magnitude as the key variable and developed formal models linking votes to seats. Predicting Party Sizes (2007) is described by Taagepera as 'a very different book' because of marked advances in understanding the connection between institutional inputs and political outputs [citation:12]. The earlier work provided the foundational framework; the later book makes specific quantitative predictions that agree with world averages and can tell practitioners by how much to change electoral rules to achieve desired outcomes [citation:1][citation:9].

Taagepera calls his models 'ignorance-based' because they start from sparse observations about how mechanical features of electoral systems set bounds on quantities like the number of parties [citation:16]. The predictions are made 'in the absence of any further information' about social cleavages, political history, or campaign dynamics [citation:19]. These predictions are not intended to be perfect; they provide a baseline expectation against which actual outcomes can be measured. Taagepera argues this approach is preferable for institutional design because it relies only on variables that are in principle subject to redesign [citation:16].

A primary limitation is the restriction to 'simple' electoral systems — First-Past-The-Post and List PR [citation:3]. Taagepera himself warns that for complex multi-tier systems, 'all bets are off regarding consequences' [citation:8]. A second criticism is that the models predict long-term averages, not specific election outcomes, limiting their applicability for practitioners interested in a particular contest [citation:3]. Reviewers also note the exclusion of social cleavages as a conditioning factor [citation:16]. Some of the formulas, such as the number of pertinent parties, are described by Taagepera himself as 'corazonadas' (hunches) lacking full theoretical grounding [citation:11].

Taagepera's work belongs to the 'Duvergerian agenda' — the search for predictable relationships between electoral systems and party systems [citation:6]. However, he departs from Duverger in a crucial respect: he argues that the key variable for predicting the number of parties is not the electoral formula (plurality vs. PR) but district magnitude and assembly size [citation:6]. The seat product model replaces Duverger's qualitative law with a quantitative formula that specifies not just the direction but the magnitude of the relationship between electoral rules and party system fragmentation [citation:1].

Taagepera states that while previous models told only the direction in which to change an electoral system, his models tell 'by how much' it must be changed to obtain a desired change in the average number of parties and cabinet duration [citation:1][citation:9]. Each chapter opens with a bulleted list 'For the practitioner of politics' offering concrete advice — for example, cutting multi-member constituencies into smaller districts or reducing assembly size to double the mean cabinet duration . The model can also be used to allocate seats among countries in the European Parliament on a logical basis [citation:9].