lishing party identity for the purpose of determining its origin. To handle this problem as it occurs, we must have recourse to our conceptualization of a political party as an organization that pursues the goal of placing its avowed representatives in government positions.
The term "organization" in our definition implies a set of persons who interact in pursuit of a common goal, displaying division of labor and role differentiation in their interactions. While organizations, and therefore parties, may differ in extent of division of labor and role differentiation, we conclude that some authority pattern is always present, which means that every party has leaders, active supporters, and more distant followers-members.
At any point in time, the set of persons comprising the party organization is apt to differ from that at any other point, with changes possible in any or all of the categories of leaders, supporters, and followers. If we were to adopt a strict set-theoretic conception of the political party as a unit of analysis, we would be forced to acknowledge the existence of a different party if only one element of that set were lost, gained, or even replaced by another. But, for practical considerations, we adopt a far looser conception and say that a party at time t is equivalent to one at time t+1 year, if there is a large degree of overlap between the two sets of activists at both time points--with "activists" defined as leaders plus their active supporters. This conception puts a premium on the continuity within the party of those participants whose behavior is most directly associated with its activities, while allowing for sudden and major changes in the party's membership or electoral following. This conception also permits the party to change its issue orientation, its structure, and its method of operation--and still be regarded the same "party" at times t and +1. Dramatic changes in these party properties would not constitute a change in party identity, but a comparable change in party activists between t and t +1 would, for our purposes, signal the emergence of a new party.
The set of activists that defines our unit of analysis openly distinguishes itself from rival sets of activists through the use of a unique party name and related symbols. The party name or label acquires a special significance in establishing the identity units of analysis. It is assumed that party labels are adopted for a purpose and that a party label comes to have symbolic value for the organization. A party may consider changing its label if it no longer seems to suit the party's purpose of if a better label can be used. But changes in party labels are more likely to come shortly after the party's formation rather than later, for the symbolic value of the label tends to increase with time. Possession of the party label is often an issue in the case of party splits or mergers, and the name tends to be retained by the majority group in the case of splits and incorporated into a composite name in the case of mergers.
We recognize the value of party label in establishing the identity of a political party. Therefore, we are prepared to regard as equivalent units those national parties that possess identical names at time t and at time +1 year, this length of time taken as a practical unit for research. Thus, a party with any given name in 1956 is automatically regarded as equivalent to the party with the same name in 1957--unless there is evidence of a significant change in the composition of the sets of activists between time t and t+ 1. In that case, we follow a set-theoretic decision process to establish party identity as shown in the decision diagram reproduced as Figure 3.1.
Operational Definition. The "year of origin" refers to the year in which the activists organized for the purpose of pursuing their goal rather than the year in which the party first presented candidates for an election--although the two may often be the same. Year of origin is coded by entering the last two digits of the year; for example, if 1925, then "25" is entered. Parties originating before 1900 are coded with negative values; for example, if 1890, then "-10" is entered. (The year itself is easily reconstructed by adding 1900 to any code.) Thus the larger the value, the younger the party and the lower its score on this indicator of institutionalization.
Coding Results. The pertinent statistics summarizing our coding of BVl0l are contained in Table 3.1, which shows that we were able to assign a "year of origin" to every one of the 158 parties in the study. The mean value of AC101--the adequacy-confidence code