BECAUSE SOCIAL CLEAVAGES tend to be manifested in political cleavages, the social bases of political parties have been a major concern of political analysis. The ICPP project employs three concepts pertaining to the patterns of support that political parties receive from significant social groupings. Although these three--social attraction, social concentration, and social representation--are regarded as separate concepts in the "external relations" portion of our conceptual framework, their common concern with support patterns dictate that they be treated together in this chapter.

Any society can be analyzed for social cleavages along several dimensions. As explained later, there are six dimensions that seem especially relevant to a cross-national comparative analysis of party support. These are (1) economic status, (2) religion, (3) ethnicity (including language and race), (4) region, (5) urbanization, and (6) education. With respect to each of these dimensions, three separate questions need to be answered. (1) How evenly does the party attract support from various subgroups along the dimension? (2) How heavily is the party's support concentrated within any particular subgroup? (3) How accurately does the party reflect the overall composition of society on that dimension? The concepts of attraction, concentration, and reflection address, respectively, each of these questions.¹

Basic to the distinctions among these concepts is a visualization of two alternative methods of assessing the social bases of party support. Suppose that sample survey data are available for a given country in the form of a crosstabulation of party preference by major groupings of a social variable-for example, occupation. How should any given party's social support be assessed? Two alternatives present themselves immediately to the analyst. (1) Assess the party's support in terms of the proportions of the groups' preferences it receives, or (2) assess the party's support in terms of the proportions of its total preferences that come from each group. These alternatives are perhaps more easily explained graphically than described verbally. Table 5.1 depicts the hypothetical situation. Calculating percentages by columns conforms to method 1; calculating percentages by rows conforms to method 2.

Voting studies have tended to analyze data such as those in Table 5.1 by columns, reporting the percentages of a given group--for example, unskilled laborers--that support each of the parties. This mode of analysis conforms to the major interest of voting studies in predicting voting choice of individuals. The percentage that prefers a given party can be regarded as an estimate of the probability that a member of that group would support that party. Students of voting behavior have been less apt to calculate percentages by parties (by rows in the preceding example), because the resulting percentages indicate nothing about the probability of voting choice of individuals--as the party rather than the group of individuals becomes the unit of analysis. Thus, the voting studies are more likely to produce information on the percentage of blue-collar workers voting "leftist" in Western democracies than on the proportion of the "leftist" vote that comes from blue-collar workers.

From the standpoint of parties research, both methods for calculating party support are important, for they can be interpreted as component values in distinct measures of social "attraction" and social "concentration." Based on the proportions of the groups' support given to a party, a measure can be devised for the attractiveness of the party to various social groups. Based on the proportions of the party's support drawn from given groups, a measure emerges for the concentration

---

1. In earlier work (see Janda 1970; Gillies and Janda 1975), the approach to the analysis of party support was conceptualized differently. The present concepts of social attraction and social concentration were respectively labeled social aggregation and social articulation and were used as surrogates for Almond's concepts of interest aggregation and interest articulation (Almond and Coleman 1960; Almond and Powell 1966). This required the theoretical assumption that the processes of interest aggregation and articulation follow from the underlying structures of social support for the parties. It seems wiser to provide for independent testing of this linkage by keeping these pairs of concepts analytically distinct and labeling those pertaining to party support in a more denotative fashion, which we attempt to do with "attraction" and "concentration." A similar argument holds for the replacement of "social representation" (in the former usage) with "social reflection." The term "representation" has many connotations that are avoided by "social reflection," which denotes more directly the quality of mirroring the society in the composition of the party.

go to page 42