calling for "a general theory." What would a general theory look like? We can try to envision this mythical object by speculating on its form and scope.
Must a general theory of political parties meet the rigorous requirements of a hypothetico-deductive system, complete with axioms, postulates, and mathematical expression? If so, rational choice theory applied to parties ( Greenberg and Shepsle 1987; Holler 1987; Laver and Shepsle 1990; Baron 1991; Coram 1991) would seem to qualify. Judging by the scholarly style of some who have called for general theory (e.g., Sartori and von Beyme), this requirement is too demanding. For many, a general theory could consist of tightly organized verbal statements of theoretical propositions that were empirically testable. If so, examples can be found in Duverger, Katz ( 1980), Panebianco ( 1988), Budge and Keman ( 1990), Laver and Schofield ( 1990), Strom ( 1990), and Schlesinger ( 1991).
In discussing what it means to build party theory, Budge and Keman ( 1990) advanced "a series of hypotheses or propositions each of which can be examined and compared with what the collected information on that point actually tells us. Taken together, such propositions constitute the 'theory' or 'explanation' of why and how the parties act as they do, and if upheld should also enable us to anticipate what they are going to do in government, under given circumstances" (pp. 7-8). An important way to achieve theoretical clarity "is to put all this in mathematical form," but that narrows the focus because of restrictive and simplified assumptions (p. 9). They chose to favor broad coverage and practical relevance, relying on verbal expression of tightly reasoned propositions. Their work and other examples indicate that party theory need not be mathematized to be general.
A truly general theory should also have a broad scope of application to political parties. There are two aspects to the scope of a theory: the types of parties to which it applies and the range of party phenomena that it explains. The scope of application depends on the definition of party that underlies the theory. The issue is whether the definition is narrow and theoretically exclusive, or broad and theoretically inclusive. If based on a broad definition of party, as used in this review, party theory should be applicable in principle to political parties everywhere, within the parameters of the theoretical conditions. We are left with the range of party phenomena encompassed by the theory.
What would a truly general theory of parties explain? For example, would a general theory about parties encompass voting behavior? As Schlesinger ( 1991) argues, that is excluded by defining parties as organizations; voters are choosers among parties, not the parties themselves. Although it is common to speak of parties as coalitions of voters, none are coalitions in the proper sense of the word. "None represent conscious, explicit agreements by members of these categories to pursue joint action" (p. 8). He states that this exclusion is a critical point. "Much of the difficulty political scientists have had in developing a theory of political parties has come from not knowing what to do with the voters" (p. 8). Excluding voting behavior from a theory of parties does not mean that party theory cannot include propositions about voting behavior. It simply means that party theory has no responsibility for explaining voting behavior, for its theoretical target is party organization and behavior.
Must a general theory of parties explain everything about parties? Furthermore, must it be a single integrated theory? If so, we will lack a general theory until we have a body of interrelated propositions that -- at a minimum -- explains (1) covariation among organizational characteristics within individual parties; (2) how and when parties form, change, and disappear; and (3) the effects of political parties on political life -- on its institutional and personal aspects. If this is the goal of general theory, then it is virtually certain that there will never be a general theory of political parties. Nor will there be a general theory of voting, or of legislatures, or of any other political institution or behavior, for we demand too much.
We should not expect a single general theory to explain everything about parties. Instead, we should acknowledge the existence of coherent bodies of explanation for broad segments of party phenomena as general theory. Our task should be to discover islands rather than look for continents. An analogy with voting behavior may be helpful: There are bodies of theory that explain voting turnout and bodies that explain voting choice. Turnout and choice theories share some points of contact, but they are essentially separate bodies of propositions. In the parties field, as I hope to have demonstrated in this review, we already have some bodies of theory that address distinct but important party phenomena.
In his recent and comprehensive review of literature on American parties, Crotty ( 1991) observed, "Political parties research, then, is more a uniting of a