Path: Janda: Political Parties, Home Page > Part 1: Table of Contents > Chapter 14
Chapter 14: Validating the Conceptual Framework (pp. 135-161), p. 137
(you can navigate to other pages by clicking on page numbers below)
p. 135
p. 136
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
p. 161

8. Degree of Organization
9. Centralization of Power
10. Coherence
11. Involvement

Accepting for the moment these conceptual labels in lieu of definitions given in subsequent parts of the paper, we can inquire into the framework's empirical applicability for the study of political parties. Two questions arise. (1) What range of the observable party phenomena does this set of concepts encompass? (2) How well can we meet its demands for data on these phenomena?

The first question can best be answered with reference to Table 14.2, which surveys all the "basic variables" (representing observable phenomena) that are subsumed under the final ten concepts. Each basic variable within a cluster constitutes a specific indicator of the more abstract concept that subsumes it. The expected relationships of the basic variables to their parent concepts and to one another are specified in subsequent sections of this chapter, where these expectations are tested with data. For now, it is appropriate to note only that the conceptual framework accounts for 111 different party properties. There seems to be no alternative conceptual framework in the literature that rivals it for coverage of party properties. Although it obviously does not exhaust the set of conceivable party phenomena, it does embrace a wide range of party characteristics and activities.

The second question really has several aspects to it. What is the amount and quality of available information on the conceptual indicators for parties throughout the world (i.e., how many parties can be scored and how confidently), and how reliably can the framework be applied in interpreting the observable phenomena (i.e., what is the intercoder reliability in scoring parties on the indicators)?

The number and diversity of variables subsumed within the conceptual framework made heavy demands for information about the parties in the study. The ICPP project sought this information from library-type sources--books, articles, newspapers, party documents, government reports, and so on. As explained in Chapter 1, more than three years were required just to complete the bibliographic searches, collect the textual material, and index information for retrieval and research. Over 60,000 pages of material on party politics in our 50 countries were derived from more than 3,000 papers and publications. The collected material varied from a high of 4,582 pages on party politics in India to only 122 pages on party politics in the Central African Republic. Such differences in the quantity (and quality) of information in our files signaled differences in our ability to code the parties on the variables in the conceptual framework. A special microfilm and computer system was devised to manage the information collected while imposing quality control on the research conducted.[4]

Substantial effort was devoted to assessing data quality throughout the project. As explained more thoroughly in Chapter 2, each scoring judgment for every basic variable was accompanied by an adequacy-confidence (AC) code indicating the "adequacy" of the information in the file and our "confidence" in making the judgment.[5] The AC codes were assigned according to a nine-point scale, with 1 for no information in the file to score the party of the variable and 9 for at least three different sources supported the coding decision and no disagreement in the literature.

After the variable was scored and adequacy-confidence codes were assigned to each coding judgment, the researchers had to record their comments on the substance of each coding judgment for entry into a separate information management system.[6] Slowed considerably by these quality-control procedures, the coding took five years for completion after the conceptual framework was completed in 1969. As a further check on the quality of our codes, the records of the parties' codes and coding judgments were sent to more than 40 outside area and country experts who (over a period of two additional years) checked our facts and interpretations of events against their own knowledge and judgments. While by far most codes passed their scrutiny, our consultants made countless minor and numerous major corrections in our research, thus improving the quality of our data immeasurably. Despite three years of bibliographic work, five years of coding, and two years of outside review, our data are surely still not perfect. Nevertheless, they represent a lengthy and careful attempt to meet the demands of the proposed conceptual framework.

Just how well they meet those demands can be judged by reference to the three columns of figures in Table 14.2. The first column reports the total number of parties scored for each of the 111 variables in the conceptual framework. The second gives the percentages of the

4. See Kenneth Janda, A Microfilm and Computer System for Analyzing Comparative Politics Literature," in George Gerbner et al., eds., The Analysis of Communication Content (New York: Wiley, 1969), pp. 407-435.
5. See Kenneth Janda, "Data Quality Control and Library Research on Political Parties," in Raoul Naroll and Ronald Cohen, eds., A Handbook of Method in Cultural Anthropology (Garden City, N.Y.: Natural History Press, 1970), pp. 962--973.
6. See Kenneth Janda, A Worldwide Study of Political Parties,' in Benjamin Mittman and Lorraine Borman, eds., Personalized Data Base Systems (New York: Wiley, 1975), pp. 129--137.

go to page 138