tural or organizational aspects of party power, we would expect it to correlate with the degree of organization scale more strongly than was expected of the personal power scale, which should signify the absence of organization.
The findings completely support the expectations. The correlation between structural power and degree of organization is .34, while that between personal power and degree of organization is -.21. Despite the positive relationship between the two power subscales, they were differentially related to degree of organization even to the difference in sign. This pattern of relationships lends credence to the measurement of the structural and personal dimensions of party power while supplying additional evidence for the measurement of degree of organization.
Coherence. The concept of coherence is defined as the degree of congruence in the attitudes and behavior among party members. This concept was intended to embrace studies of party "cohesion" and "factionalism" and is tapped by six variables.
The last variable in this cluster presented conceptual difficulties from the beginning, when it was proposed as "party splits." The original paper reasoned as follows: "Splits within a party appear to be a clear indicator of lack of coherence, but it is also possible that the departure of dissidents leads ultimately to greater coherence. While this variable is being tentatively included in the coherence cluster, its contribution to coherence is problematic at present" (1970, p. I11). Further thinking led to its reformulation as party purges, indicating involuntary rather than voluntary departures from the party. Although there were conceptual problems with this formulation as well, the parties were duly scored for their recent history of party purges.
The intercorrelations among these six indicators reflect the confusion concerning the placement of this last variable in the set. It clearly had little empirical relationship to the other five measures, correlating on the average of only .08 in the expected directions. These empirical results settled the matter convincingly, and party purges was dropped from the set of coherence indicators. The expectations of relationships among the other five items were consistently upheld, however. Legislative cohesion correlated negatively with each of the four factionalism items, which in turn intercorrelated positively with one another. The average correlation among the set had a magnitude of .34, the first factor accounted for 47 percent of the variance, and the reliability of the resulting "coherence" scale was an acceptable, but not overpowering, .72.
Involvement. The last concept in the ICPP conceptual framework is "involvement," defined as the intensity of psychological identification with the party and as the commitment to furthering its objectives by participating in party activities. This concept was tapped through these indicators:
Conceptualizing involvement as "the intensity of psychological identification with the party" meant for the sake of the party as an organization or social force. This stands opposed to motivations for personal material advantage (variable 11.03) and for the sake of a charisma-