or "members" to determine magnitude of splits or mergers.
A hypothetical example may help in using Table 3.4. Suppose that a party founded in 1900 had undergone three mergers with other parties before 1941 but then maintained its organizational integrity until 1953, when it suffered a major split, and 1959, when it experienced a major merger. The party's score on this variable would then be calculated as follows: 0 points for the mergers before 1941; 5 points for the major split in 1953; and 8 points for the major merger in 1959--for a total score of 13.
Coding Results. We were able to code virtually every party on BV103, but, as the mean AC code indicates, we did so with less confidence than in coding either year of origin or name changes. As Table 3.5 shows, this variable has bimodal tendencies in its distribution. Almost 30 percent of the parties experienced no splits or mergers from 1941 through the end of our time period-and thus received a score of 0 on BV103. But approximately 15 percent of the parties stood at the other end of the distribution with the maximum score of 19--meaning that they suffered the ultimate discontinuity by terminating during our time period. The assignment of such a large value for termination is not entirely satisfactory, and improvements in operationalizing this variable need to be explored for the purpose of recoding the terminated parties in particular. As in our coding of BV 102, there is again a relationship between the codes for BV103 and the quality of the coding as reflected in our AC scores. But this time the relationship is reversed, and the correlation between BV 103 and AC 103 is .22. The coders tended to be more certain of instances of high discontinuity than low. I suspect that the correlation is inflated primarily by the coincidence of the highest AC codes with the highest discontinuity codes, for coders were nearly always certain of a party's termination during our period.
An institutionalized party, it is assumed, features (1) unambiguous identification of the legitimate party leadership at the national level and (2) change of personnel in the top leadership position over time. Moreover, when leadership change occurs, it comes about through an overt process involving (a) criticism, challenge by an aspirant for the incumbent's position, and formal meeting of party members on the issue of leadership change, or (b) a prescribed succession of leadership on the death or disability of the national leader.
Identification of the legitimate national leader may 6e problematic for some parties, where the leadership structure is so diffuse that many individuals-or no individuals-claim the post. The important fact here is not "effective" leadership but "legitimate" leadership. In the United States, for example, the presidential candidates of the two major parties can be considered as the symbolically legitimate party leaders. Even after the election, the winning candidate may continue to supply legitimate leadership although his effectiveness suffers. The losing candidate usually offers only symbolic leadership (consider his designation as the "titular head" of the party), but he is frequently the only individual who claims any legitimate authority.
In the absence of sample survey data on participants' views of the party, one test of the existence of the party as an "abstract entity," apart from its leaders, would be its existence and operation under the leadership of different individuals. The fact of change is judged to be more important than the frequency of change, and the recency of change is also judged to be more important than the frequency of change. A party that has had the same leader since its inception is judged to be less institutionalized than one that has changed leaders; a party that has changed leaders recently is judged to be more institutionalized than one that has had the same leader for years.
Leadership changes, when they occur in an institutionalized party, follow some publicized procedure that culminates in a formal choice between leader aspirants.