|
|||||||||||
|
quences if a party did not participate in elections: (1) it received a high electoral instability score (the value depending on the number of elections missed), and (2) it was coded with the lowest AC score. Because nearly 20 percent of the parties experienced this combination of codes, the correlation between BV106 and AC106 is extremely high at -.82, with low data quality again associated with high instability. No other variable in the study correlates with its AC code at a comparable strength, which underscores the pronounced relationship between code and quality built into this variable. Those who are not satisfied by this procedure for calculating instability scores for parties that did not participate in any elections may wish to select out all cases above 3 for AC106 and conduct their analyses with the remaining 80 percent of the parties coded. It can be argued that the more identifiable the party, the more attention it will be paid by scholars and other political observers. Hence, we proposed as a measure of a party's identifiability over time (and thus its institutionalization) the sheer number of pages referring to it in our microfilm information files. Clearly, this indicator is different in kind from the others, for it is not a direct measure of a party property. Although only an indirect indicator, it does say something about a party's existence and is thus offered as the last of the series of basic variables tapping the concept of party institutionalization. Operational Definition. This variable is simply the count of the pages in our microfilm files that were tagged with the party's code number for purposes of retrieval. It is the value entered under "No. of Pages" in Table 1.3. Coding Results. As noted in Table 1.3, not all 158 parties were coded by relying on an extensive microfilm information base. Of the 145 parties for which we did have microfilm files, the average party was mentioned on nearly 400 pages, as reported in Table 3.9. But, as shown by the breakdown of pages into 100 page categories, this high figure is misleading, for it owes to a few heavily documented parties mentioned on thousands of pages. It is sobering to note that one-third of the parties were coded from an information base consisting of no more than 155 pages. As mentioned in Chapter 2, we have tried to deal with the problem of varying quantity and quality information when scoring parties on our variables by assigning an accompanying adequacy-confidence code to each scoring judgment. The smaller the number of pages indexed, for a given party, the lower its AC codes and also the less likely we were able to code the party the basic variables in the framework. (See Chapter 14 for an assessment of coding success.)
|
|||||||||||
