|
|||||||||||||||||||||||||||
|
|||||||||||||||||||||||||||
|
This can be illustrated with reference to two variables in the involvement cluster. Variable 11.02, membership participation, was coded for only 63 percent of the parties overall, one of the lower rates in the study. A separate breakdown shows that 74 percent of the Western parties were coded in comparison with 56 percent of the non-Western (parties in all areas outside of Anglo-America, Western Europe, and the Scandinavian countries). The other variable--11.04, purposive incentives--was scored for an impressive 85 percent of the parties, but the mean AC code was the lowest in the study, only 3.8. Again more Western parties were coded (92 percent), but a large majority of the non-Western (80 percent) were also scored for reliance on purposive incentives. Moreover, the quality of our scoring judgments (as reflected by the AC codes) was actually slightly less for the Western parties, which were attributed more complex mixtures of motivations and thus were harder to score for purposive incentives. In short, the conceptual framework does tend to be more applicable to empirical research on Western parties than to nonWestern parties, but the differences are minor, and substantial numbers of non-Western parties were scored on most of the variables. How can one "validate" a conceptual framework? In what sense can it be said that a conceptual framework has "validity"? The root term "valid" figures in at least two different concepts that have relevance for our purposes. One comes from the philosophy of science and refers to the distinction between the context of "discovery" (the creation of theory) and the context of "validation" (the testing of theory once created). Validation thus involves comparing theoretical predictions with empirical evidence and results in decisions concerning the acceptance or rejection of hypotheses derived from a theoretical framework. In this sense, a "valid" theory is one whose predictions conform closely to "reality" as manifested by the observations employed in testing the theory. Accordingly, it is nonsense to speak of validation in the absence of a developed theory. Clearly, the framework proposed for the comparative analysis of political parties is conceptual rather than theoretical. That is, it accommodates the major recurring concerns of scholars in analyzing individual parties (as opposed to party systems) within a hierarchical structure headed by ten major concepts, but it does not elaborate upon expected patterns of relationships among those concepts. While not explicitly theoretical, the framework is distinctly pretheoretical in that it views the formulation of an explicit conceptual framework as a necessary beginning for the construction of valid and fruitful party theory.[8] Nevertheless, one wonders how the philosophy of science notion of validation applies at all to a conceptual framework. The other relevant sense of "valid" as a social science term is less directly linked to theory testing. The measurement literature uses validity to signify that a procedure measures what it purports to measure and that a valid measure is thus one that accurately reflects some designated aspect of reality. Validity in this sense can be achieved quite independently of explicit theory. Success in measuring voters' behavior in an election, for example, can be assessed in the absence of theory that explains their behavior. It seems that this concept of validity does pertain to a conceptual framework. Although the philosophy of science notion of "validation" and the measurement idea of "validity" seem quite different, they share the ontological assumption that reality can be observed and measured. To the extent that observation and measurement blend into one another, one is hard pressed to draw a distinction between the two concepts. When the issue is pushed, one may also find it difficult to defend a distinction between a conceptual and a theoretical framework. Considering the case at hand, we realize that the framework does indeed harbor expectations about empirical relationships--the hallmark of "theory." In general, the basic variables within a cluster are expected to correlate highly,among themselves, for they are viewed as indicators of the abstract concept under which they are subsumed. These concepts, in turn, should be relatively independent of one another, for they are supposed to represent distinct dimensions of variations among parties across nations. So it appears that the validation of a conceptual framework can be approached along two fronts. Following the usage of validity in the measurement literature, we can attempt to demonstrate that the variables are measuring what they are supposed to measure. Following the sense of validation in philosophy of science, we can test expectations about empirical relationships among the variables embodied in the framework. In this connection, it is important to point out that the proposed framework grew out of my reading of the literature on comparative politics and preceded the collection of a single datum in the ICPP project. Thus, it was not devised a posteriori to fit interrelationships discovered among the data but a priori to account for |
|||||||||||||||||||||||||||
|
|||||||||||||||||||||||||||