Path: Janda's Home Page > Workshop > Canonical Analysis > Return to Opening Page
Political Studies (1998), Vol. 46, pp. 61l-632
Effects of Party Organization on Performance during the 'Golden Age' of Parties
Summary and Conclusion

We believe that our cross-national study offers two main contributions to the parties literature. First, we demonstrate that party organization features do correlate significantly with indicators of performance, even in a macro analysis at the national level. Most of our six hypotheses were supported in their broad outlines.[62] Concerning the concept of 'party performance', we argue for looking at more than electoral success, and our empirical findings substantiate the case for measuring different aspects of performance. In fact, electoral success was less well explained by organizational conditions than were legislative cohesion and breadth of activities, two other aspects of performance. While this may strike some observers as disappointing, the predominant finding in the scarce literature on sub-national electoral districts is that organization variables provide only weak explanations of electoral success.[63] Recall also that we omitted all variables usually thought to explain why parties win votes, e.g., the state of the economy, civil unrest, political scandals, politicians' popularity, and so on. In retrospect, our findings that complexity, centralization, and involvement are better explanations of a party's legislative cohesion and breadth of activities than its electoral success, seem intuitively reasonable.

Second, our use of canonical analysis introduces an empirical method for identifying party syndromes, clusters of interrelated organizational and behavioural traits. Scholars often loosely characterize parties as packages of attributes. Duverger excelled in this, and more recently Epstein described 'responsible parties' as 'organized, centralized, and cohesive'--using precisely our language.[64] We demonstrate that parties we call 'doctrinaire' experienced high degrees of centralization and involvement with low levels of complexity and factionalism in their organizations and fared poorly at the polls while maintaining strong cohesion in the legislature and engaging in many activities. On the other hand, parties that we call 'mobilizing' tend to do well in elections while engaging in several non-campaign activities although showing little legislative cohesion. Organizationally, mobilizing parties tend to be highly complex but not very centralized, and they have little factionalism and low levels of membership involvement.

These findings on party organization and performance reflect arguments in Duverger's Political Parties. This is with good reason since the data come from the 'Golden Age' of political parties--the time of his writing. But what Duverger theorized, we can support empirically.[65] Whether or not such findings would hold today is questionable. Presumably if parties have moved more toward 'electoral professional' or 'cartel party' models, the more society- oriented variables of involvement and breadth of activities would diminish in significance while complexity and electoral success might increase. As this study has supported Duverger's thinking for his time period, we hope to test contemporary theories of party with more current data in a future crossnational study.

Path: Janda's Home Page > Workshop > Canonical Analysis > Return to Opening Page