This is p. 179

favourable to farmers and rural interests than governments without an Agrarian party . . . (Assumptions 2,3(c).)

(v) The direction of policy in specific areas will be influenced by party control of the competent ministries (Assumptions 1,2,3; Table 2.2 Assumption 3). (pp.50-51)

Budge and Keman tested their theory with data on twenty western governments and found that it fit the post-war experience. Their analysis of coalition government concluded that "parties exert a strong, and even determining, influence on government decision-making," and swung the evidence back to the "party matters" side (1990, 158).

Nevertheless, governments, not parties, remained the real units of analysis for Budge and Keman. Studying the policy impacts of individual parties requires examining their platforms or manifestos to see what particular parties propose and what they really pursue when given the opportunity. There are such studies that demonstrate party impacts in the United States (Pomper and Lederman 1980) and in Britain (Rose 1984), but there is little cross-national research on this complicated problem. However, Hofferbert was involved in similar studies of party impacts in two countries--the United States (Budge and Hofferbert 1990) and Germany (Hofferbert and Klingemann 1990)--using data from the ECPR party manifesto project. Both studies demonstrated specific impacts of party programs on government policy, with party influence in the U.S. coinciding most strongly with occupancy of the presidency and party influence in Germany working mainly through possession of cabinet ministries. Hofferbert and Klingeman warned, "Comparative policy studies, with their general focus on blunt indicators of party control are clearly in need of refinement and amendment." The party patterns "are but poorly revealed by contrasting the mere incumbency of one set of collective actors (parties of the 'left' or of the 'right') against one another and then seeking policy consequences of that contrast" (Hofferbert and Klingeman 1990, 300-301).

Although research on political parties is related to that on party systems, the two bodies of literature employ different concepts and theories. Duverger reflected this difference by dividing his classic Political Parties into "Book I: Party Structure" and "Book II: Party Systems." Sartori's Parties and Party Systems (1976) is similarly divided into two parts, with that on party systems more than twice as long as that on parties. Whereas parties are studied by specialists interested in them as political institutions, party systems attract the larger group of scholars interested in politics across nations. Especially in the U.S., more articles have been published on cross-national studies of party systems than comparative analyses of parties. For the most part, the literature on party systems focuses on western democratic nations, but there have been some efforts to build typologies of Latin American party systems (Werz 1987 and Collier and Collier 1991, 498-505).

Although the difference in the unit of analysis produces a different set of analytical concepts, party concepts and system concepts might seem to be identical in a one-party system. In a penetrating analysis, Sartori reasoned that parties make a system only when there are multiple parties (1976, 42-45). One party can relate to the state in a "party-state system," in which the party's properties figure in analysis, but there cannot properly be a one-party system. Despite the frequent usage of the terms "one-party" or "single-party" systems, Sartori's view holds in the literature, which requires at least two parties to make a party system.

In the first edition of their book, Lane and Ersson (1987b) carefully reviewed scholars' positions on definitional, conceptual, and analytical issues involved in studying party systems. All scholars agreed that a party system, like any system, consists of parts and relationships among those parts, such that the system is more than the sum of its parts. Accordingly, Lane and Ersson defined a party system as "a set of political parties operating within a nation in an organized pattern, described by a number of party-system properties" (1987b, 155). Scholars differ on what system properties are important concepts in theories of party systems, and their emphases are reflected in their research. Using Lijphart's set of important properties (1984), we find emphasis on (1) minimum winning coalitions (Dodd 1976; Laver and Schofield 1990), (2) government durability (Dodd 1976; Grofman 1989), (3) effective number of parties (Taagepera and Grofman 1985, Herzog 1987, Molinar 1991), (4) number of issue dimensions (Powell 1987; Listhaug, Mcdonald, and Rabinowitz 1990), and (5) electoral disproportionality (Rae 1971, Lijphart 1990). Others would add party competition (Laver 1989, Strom 1989, Ware 1989) and volatility (Pedersen 1983).

To establish the important dimensions of variations in party systems, Lane and Errson factor analyzed a set of fourteen different indicators (along the lines of those proposed by Lijphart) for 272 elections in 16 European countries from 1920 to 1984. (The factor analysis was only reported in the first edition of their book, not in the second.) They concluded that the European party systems had five major dimensions: