of indirect members by sectors.¹ For the record, however, labor unions accounted for most of the instances of indirect membership. Also for the record, there was no significant correlation between the codes for BV702 and the confidence of our coding judgments.

A party led by persons recruited from one of the main institutional sectors of society will develop special leadership links with that sector. If most leaders come from two institutional sectors, the conditions for compromising autonomy are not so propitious, and they fade almost entirely when leaders are recruited from three or more sectors, as party autonomy can be restored through competition among leaders from different sectors.

At the conceptual level, it is difficult to define "leaders" with distinct boundaries. In general, we refer here to national party leaders, although that, of course, includes representatives from subnational units. But the size of the national leadership group varies from party to party, and criteria for satisfactorily identifying this group are not obvious. To gain a measure of consistency in identifying the group of leaders and coding their social origins, we somewhat arbitrarily focus on the party's parliamentary representatives-or candidates for legislative office when the party has little success in actually winning seats. While not all the major party leaders may be included in the party's contingent in the national legislature, this category may be expected to incorporate many of them. At least, it identifies a distinct population of leadership-level party members that offers some basis of comparability across parties. In the case of ruling Communist parties, however, we depart from this focus and instead use members of the Central Committee as a better choice for comparability.

Operational Definition. Our coding scheme is similar to that used to score variable 7.01, "sources of funds." We also use the institutional sector code given in the operational definition of that variable to indicate the sector or sectors that contribute the party's leaders, usually defined as its parliamentary delegates. We coded the lowest applicable score.

1	Two-thirds or more of the party's leaders come from a single institutional sector of society.
2	Two-thirds or more of the party's leaders come from two sectors of society.
3	About half (between one-third and two-thirds) of the party's leaders come from a single institutional sector of society.
4	About half (between one-third and two-thirds) of the party's leaders come from two sectors of society.
5	Two-thirds or more of the party's leaders come from more than two sectors.

Coding Results. We came through with codes for about 85 percent of our parties on "source of leaders." According to our operationalizations of the variables, parties were more closely tied to specific sectors of the society for their leaders (BV703) than for their funds (BV701). The statistics in Tables 8.4a and 8.4b state that almost one-third of the parties drew most of their leaders from only one sector of the society while another third drew their leaders from only two sectors. The sectors that contributed the most to the leadership ranks of parties across the world are enumerated in Table 8.5.

It is instructive to compare the findings for "sources of leaders" in Table 8.5 with those for "sources of

*John C. Thomas assisted in writing this section.

go to page 95