MORE THAN A QUARTER of a century ago, Maurice Duverger began the preface to his now classic work, Les Partispolitiques, with this observation:
This work starts from a basic contradiction: it is at the present time impossible to give a valid description of the comparative functioning of political parties; yet it is essential to do so. We find ourselves in a vicious circle: a general theory of parties will eventually be constructed only upon the preliminary work of many profound studies; but these studies cannot be truly profound so long as there exists no general theory of parties. For Nature answers only when questioned and we do not yet know what questions this subject demands. The example of America is cogent: studies of political parties abound; they are based upon considerable and serious observation; they are often of great value; not one, however, throws any light upon problems like the evolution of party structures, the number and reciprocal relations of parties, the part they play in the State, for all these studies are conceived within the framework of America alone, they deal with problems that are specifically American and do not refer to general questions. Yet how can one refer to general questions when for the most part they are still undefined?
Duverger's brilliant effort sparked a shower of activity in the comparative study of political parties in the 1950s and early 1960s that promised to fulfill his conjecture that "in fifty years' time perhaps it will be possible to describe the real workings of political parties." But the passage of time has dulled the spur of Duverger's stimulus to the study of parties, and the field seems to have relapsed into somnolent circularity. The vast numbers of detailed studies produced since Duverger's work have not promoted the theoretical development he anticipated.
I believe that a major factor retarding theoretical development in the comparative study of political parties is the failure to embrace comparative analysis in a comprehensive sense. The dominant tradition within comparative politics has been overly conservative. Students have been cautioned to limit their comparisons to "things that are similar" rather than to things that are "different." By concentrating attention on what Przeworski and Teune call "the most similar systems design" rather than designing research to encompass the "most different systems," students of political parties were not forced to cope with diversity, were not challenged to be broad in their conceptualizations. Instead of promoting the integration and synthesis of knowledge, comparative parties research tended toward channelization within cultural-geographical areas. Students of parties in Western democracies, for example, produced a literature largely distinct from that on communist party states. Asian scholars rarely compared parties in their area with those in the emerging states of Africa. Latin American scholars treated parties in their countries as sui generis. While detailed studies of "similar" parties capitalized on the researchers' expert knowledge of domestic politics in certain countries, the price paid for descriptive accuracy was narrowness of conceptualization.
If the goal of parties research is general theory about political parties, then we must expand our scope to include the universe of countries and parties. We must study parties where they are weak as well as where they are strong. We must study parties that restrict competition or engage in subversion as well as those that compete openly for office. We must study parties that are new as well as those that are old. Whether one's theoretical orientation employs parties as dependent variables, independent variables, or organizational settings for political behavior, we should begin our studies with a view toward capturing the variance necessary to provide explanatory power in any model. If we are less than comprehensive in our scope, we have no hope for developing general theory.
Raoul Naroll used the term "holonational" to refer to cross-national studies that employ worldwide representative samples. Holonational studies of conflict and other attributes of nation-states have brought new in-