Path: Janda: Political Parties, Home Page > Part 1: Table of Contents > Chapter 14
Chapter 14: Validating the Conceptual Framework (pp. 135-161), p. 161
(you can navigate to other pages by clicking on page numbers below)
p. 135
p. 136
p. 137
p. 138
p. 139
p. 140
p. 141
p. 142
p. 143
p. 144
p. 145
p. 146
p. 147
p. 148
p. 149
p. 150
p. 151
p. 152
p. 153
p. 154
p. 155
p. 156
p. 157
p. 158
p. 159
p. 160

be explained by "party" characteristics. This in itself is a remarkable theoretical point. However, Harmel also found that the conventional system factors which the literature offers as explanations of party centralization have relatively low correlations in a bivariate sense and, even when taken together in a multiple regression model, explain only about 34 percent of the variance. This means that current theoretical factors account for only about half the variance that we know inheres in system characteristics. Clearly this problem requires additional theoretical development, but it would not even have been seen in this light were it not for the availability of conceptually grounded data to test the extant theory.

In the second example, Kirchheimer's view of the catchall party was that it exchanged "effectiveness in depth for a wider audience and more immediate electoral success" (1966, p. 184). In contrast to the study of party centralization, there has been some rigorous comparative research testing Kirchheimer's expectations of the catchall party. Rose and Urwin's analysis of 76 parties in 17 Western countries (1969, p. 23) concludes that the electoral strength of a party varies inversely with its cohesiveness in accordance with Kirchheimer's hypothesis. However, other aspects of Kirchheimer's theory--such that catchall parties were more likely to be less concerned with ideology-were not adequately tested by Rose and Urwin, "because social scientists have done surprisingly little work in measuring the programmatic content of parties" (1969, p. 28). Using earlier data from the ICPP project, Gillies and Janda were able to test both of Kirchheimer's propositions for a broader set of parties (1975). Employing a measure of ideological extremism based on the economic leftism scale, they found that party diversity was associated with low ideologies, as Kirchheimer hypothesized, but only in Western countries--to which Kirchheimer limited his remarks. While Gillies and Janda also found the predicted correlations between social diversity and electoral success, later-work by Gillies (1979) using a richer set of variables under alternative causal models has questioned the direction of the causal arrow underlying those correlations. Gillies found that electoral success is as likely to produce social diversity as the other way around, which is a fundamental challenge to a widely accepted theory.

The point of these examples of theoretical applications of ICPP data is the familiar one that good data helps the development of good theory. More than 25 years ago, Duverger noted the "vicious circle" that required a general theory of parties to be founded on profound studies but that studies could not be profound without a general theory (1954, p. xiii). As Duverger tried to break out of the circle by developing a preliminary general theory of parties, I began more than ten years ago to iterate his effort from the data side. Now it is time for another iteration from the theory side, and soon the circle will no longer be "vicious" but one of positive feedback.


A conceptual framework for comparing political parties, proposed nearly a decade ago, was reexamined and tested with data collected on 158 parties operating during 1950-1962 in 53 countries representing all cultural-geographical areas of the world. The ten major concepts in the framework subsumed 111 basic variables pertaining to party characteristics and activities. These concepts were formulated, and the interrelationships among the basic variables were specified in advance of any data collection, much less any data analysis. The attempt to validate the conceptual framework consisted primarily of analyzing the cross-national parties data to test for predicted relationships among the basic variables.

The observed relationships among the variables largely conformed to expectations of the conceptual framework. Eight of the ten variable clusters performed mostly as expected, another was not properly tested due to shortage of research funds, and only one cluster of variables completely failed to support the initial conceptualization. A test for discriminate validation, however, cast doubt on the independence of the involvement and degree of organization scales, and this question needs to be considered in subsequent research. But, from the standpoint of convergent validation (with the question of discriminate validation still to be resolved), the expectations of the conceptual framework were largely supported by the data analysis. Over 75 percent of the variables included in the empirical tests were incorporated into scales linked to their parent concepts in the original framework.

In all, a total of eighteen scales and subscales were formed to tap the concepts in the framework. The reliabilities of these scales ranged from .69 to .96 and averaged .82. Scholars who wish to utilize the data collected by the International Comparative Political Parties Project, reported in Part Two, and deposited with the Inter-University Consortium for Political and Social Research should find it helpful to know that the basic variables available do tend to interrelate as originally conceptualized. These variables can be used to measure such things as party institutionalization, governmental status, social diversity, issue orientation, goal orientation, degree of organization, centralization of power, coherence, and involvement. Of course, those who operate with alternative conceptual frameworks may use the basic variables in other ways according to their own theoretical expectations.

return to Table of Contents