TABLE 8.5: Social Sectors as Primary Sources of Party Leaders
Code	Sector	1st Half	2nd Half
1	Labor	18	17
2	Farmers (not landed aristocracy)	6	6
3	Education/scientific/professional	47	57
4	Business and commerce	22	26
5	Large landowners/nobility/traditional rulers	15	19
6	Church	8	9
7	Military	8	9
8	Other domestic parties	3	3
9	Foreign governments or parties
10	Government bureaucracy	6	6
11	Party bureaucracy	3	3

funds" in Table 8.2. Because more parties were coded for sources of leaders than sources of funds, the frequencies are generally greater in Table 8.5. But, allowing for that, the relative differences in frequencies by social sectors are striking. The business sector, which looms as the most important source of party funds, is a distant second to the educational and professional sector in supplying party leaders, while the latter almost never serves as primary source of party funds. Similarly, the military never emerges as the primary source of funds but is as likely to supply party leaders as the church and even more so than the farmers. From the standpoint of measuring autonomy, however, what does it mean that the leaders of approximately one-third of the parties are largely representatives of the educational / scientific / professional sector of society? Because this sector has such diverse interests, it probably restricts a party's autonomy less than if any other social groups were to dominate its leadership ranks. In retrospect, this category probably should have been elaborated into subcategories to permit more refined analysis.

Duverger provides an extensive discussion of the conceptual aspects of interparty relationships (1963, pp. 330-351). He first distinguishes among electoral, parliamentary, and governmental "alliances," with alliances being understood as similar to coalitions, except that coalitions are "temporary" agreements while alliances are more lasting. Electoral alliances, he notes, vary according to the specifics of the electoral laws and system. Two important types of variation are whether alliances are (1) tacit or explicit and (2) local or national. For our interest, we note that explicit alliances involve more sacrifice of autonomy than tacit ones, as do national alliances over local ones.

Parliamentary alliances (or coalitions) can be prompted either in support of the government or in opposition to it. Alliances in opposition to the government obviously exclude cabinet coalitions. Alliances in support of the government may or may not include cabinet coalitions, for parties may lend support to organize a government yet refuse to take responsibility for the government.

Duverger suggests that electoral alliances, especially tacit local ones, are easier to form than parliamentary alliances: "It is easier to join forces to win seats than in order to exercise power: the former type of alliance only requires a negative agreement against an opponent, the latter a positive agreement upon a programme which demands a more deep-seated similarity. In some cases electoral alliances cannot be transposed to the parliamentary level because they are self-contradictory: the allies in each constituency are not identical" (1963, p. 334).

Finally, Duverger comments on the power relationships within alliances, noting that "every alliance is unequal, and the only worthwhile question is that of the degree of inequality" (1963, p. 344). He identifies three factors in assessing inequality: respective strength of the parties, their "position on the political chessboard," and their internal structure. For our purposes in measuring autonomy, we are especially interested in the extreme situation of alliance domination by a party that overshadows its partners on all three factors. When that occurs, the dominant party appears to have lost little of its autonomy by virtue of its association or alliance with weaker members.

Operational Definition. The ideas expressed in the conceptual definition are incorporated into the following scale, which ranges from complete dependence to complete autonomy. To Duverger's typology, derived from the classic occidental model, there must be added a fourth cooperative pattern: the nonelectoral alliance, which finds parties collaborating to subvert the political system (see variable 6.20). Such an alliance can entail the surrender to another party of at least a measure of independence; hence it should be coded along the same

go to page 96