turely terminated due to lack of resources. The explanation for this unfortunate development is given under "coding results" for BV601 through BV605. The rationale for excusing this particular set of variables from the coding, however, is slightly different from that advanced for BV601-BV605, BV611-BV616, and BV621-BV626. Those three sets of variables were eliminated because they were highly correlated respectively with BV600, BV610, and BV620--the three corresponding strategy variables. While BV641 through BV644 have no counterpart strategy variable within the goal orientation cluster, there is a counterpart variable under "autonomy," the final concept in the "external relations" group of the ICPP conceptual framework. BV704, which will be considered in the next chapter, expresses the extent of a party's "relations with domestic parties." Expectations of strong correlations between BV704 and the individual variables BV641 through BV644 led to the decision to cease the more detailed coding of party tactics for "allying with other parties." Before arriving at that decision, we had coded about 30 parties for BV614-BV644, but the sample is too sparse to justify reporting summary statistics, which would be too unreliable for estimating population parameters .⁴

Clearly we have surrendered some important information about political parties by not coding them for their activities in electoral competition, restricting competition, subverting the political system, and allying with other parties. But this loss has to be judged against the potentially greater loss of not completing the project because of loss of resources and energy. In all, 21 variables in this cluster were eliminated from coding, which amounts to about 20 percent of the variables in the conceptual framework. By dropping these variables at an early state of the project, we were able to transfer resources to the completion of other parts of the project resources which would have been consumed in coding 33 variables for the goal orientation cluster alone. As it was, coding the parties even with the reduced set of variables took about four years. That was long enough.

This last set of variables in the goal orientation cluster also pertains to indirect tactics that a party can use in furthering whatever strategy it employs. The distinction between direct and indirect tactics under BV631-BV634 is relevant here. Some prominent tactics of "providing for members' welfare" are

6.51	Providing food, clothing, or shelter
6.52	Running employment services (not including patronage jobs)
6.53	Interceding with government on citizen's behalf
6.54	Providing basic education, not primarily political education
6.55	Providing recreational facilities or services

Operational Definition. We use the same three-point scale discussed under the operational definition for BV601 -BV605. In brief,

0	Party never or virtually never engages in the activity. Used also if literature does not specifically state that the party does not engage in the activity but discussion of the activity is conspicuous by its absence.
1	Party occasionally performs the activity but it is not considered to be a common practice of the party.
2	Party frequently performs the activity which is considered to be common practice for the party.

Coding Results. Tables 7.8 through 7.12 report the results of our coding BV651 through BV655, respectively. Although we continued to try to code these variables throughout the life of the project, we found the information inadequate for coding more than half the parties. In line with the argument developed under "coding results" for BV631-BV634, many of the parties uncoded on these variables probably did not, in fact,

go to page 88