Path: Janda: Political Parties, Home Page > Part 1: Table of Contents > Chapter 5
Chapter 5: Social Attraction, Concentration, & Reflection (pp 41-52), this is p. 46
(you can navigate to other pages by clicking on page numbers below)
p. 41
p. 42
p. 43
p. 44
p. 45
p. 47
p. 48
p. 49
p. 50
p. 51
p. 52

TABLE 5.2 Adequacy-Confidence Codes for Social Support Data
AC Code
Category Label
Interpretation and Operationalization


The cultural differentiator is not relevant as a major basis of political division within the society.


Inadequate: no data

No survey data could be located on party support and no information exists in the informamation file for the country.


Inadequate: disagreement

Some information is available, but it is so contradictory or confusing that no judgments can be offered.


Barely adequate: lowest confidence

Either (1) the percentage of support figures used in the aggregation or articulation measures are estimated from imprecise literature references to party support, with constraints for the estimates imposed by data on the society's composition and party strength,1 or (2) the percentages come from a sample survey in which the relevant number of cases 2 is less than 15.


Adequate: low confidence

The percentages come from a sample survey in which the relevant N is greater than 15 but less than or equal to 30.


Adequate: low to medium

Based on survey data: relevant N more than 30 but less than 51.


Adequate: medium

Based on survey data: relevant N more than 50 but less than 101.


Adequate: medium to high

Based on survey data: relevant N more than 100 but less than 201.


Adequate: high confidence

Based on survey data: relevant N more than 200 but less than 351.


Adequate: highest confidence

Either (1) based on survey data with the relevant N more than 350, or (2) based on census

1Constraints were imposed to ensure the internal logic of the estimation procedures when forced to guess at the distribution of party support across social groupings in the absence of survey data or other breakdowns. Typically, the researcher would confront an empty table, with the social groups arrayed across the top and the parties along the side. From available data on the country, the researcher would enter these marginal values-the percentage distribution of the groups in the society across the bottom and the percentage strength of the parties along the side. Operating under these constraints, the researcher would then choose the cell entries that best conformed to his or her assessment of the sources of parties' strength and the nature of the parties' composition.
2 The "relevant" N for the calculation of attraction AC codes is the total number of cases for each subgroup of the cultural differentiator that constitutes the base for computing percentages of support for each party. The mean AC code over all subgroups in a given table is assigned to each party in the table as the final attraction AC code. The relevant N for concentration and reflection AC codes is the total number of party supporters in the survey that constitutes the base for computing percentages of contributions to the party composition.

The marginal totals for the rows of this matrix were then fixed at the percentage distribution of the parties' strength, which was often taken from election returns. The marginal totals for the columns of this matrix were then fixed at the percentage breakdown across the cultural differentiator, which was commonly obtained from population census data. For a hypothetical country, the marginal row totals might be 30-50-20 percent representing the relative strength of its three parties, whereas the marginal column totals might be 20 percent Catholic and 80 percent Protestant for its breakdown on religion as a cultural differentiator. These column and row percentages then constrained our guesses of the party support pattern as we estimated internal cell entries, which had to total to the fixed row and column marginals. In the case of the less developed countries with sizable proportions of the population completely uninvolved in political matters, we often had to add to our party groupings a "nonpartisan" or "noninvolved" category to permit reasonable allocations of support among the parties within the constraints of the population distribution across the cultural differentiator.

Our guesses as to the appropriate internal cell entries were informed by statements in the literature-often quite vague-about the party "being backed by the Catholic church" or "receiving the votes of most of the Protestants." We culled the literature in our files for such statements of party support and then tried to translate them into quantitative terms that, when viewed in relationship to one another within the party system, would satisfy the marginal constraints reflecting the political and social characteristics of the society. No doubt, few if any of our specific figures for party support coded in this impressionistic manner are exactly correct. The objective of our scoring, however, is not so much the accuracy of specific estimates as faithfulness in reflecting the support patterns of parties relative to one another within a country. For those who wish that we had not been so "creative" in our coding of party support combining impressionistic judgments with hard

go to page 47