as late as 1961. It was included in the study by relaxing the requirement that a party demonstrate a minimum level of strength and stability by winning at least 5 percent of the seats in two elections during our time period. Kabaka Yekka won 26 percent of the seats in the 1962 election and was admitted into the sample as part of an effort to accommodate the developing party situation in the new nations of Africa.

Close study of Table 3.1 invites some interesting observations about the growth of parties over time, but these will not be pursued beyond the brief inquiry in Chapter 15. Substantive discussions of these and subsequent coding results would deter us from the main task of this volume, which is to present the data with explanations only as necessary. Substantive data analysis is reserved for other publications of the ICPP project.

Changes in party names are assumed to be made for the purpose of making new links with the electorate, and they are also assumed to result in at least momentary confusion about the party's identity within the citizenry at large. Both factors recommend "name changes" as another indicator of institutionalization, or rather the lack of it. The greater the change in magnitude, frequency, and recency, the less the institutionalization. Concerning magnitude, a change from Socialist to Radical Socialist is not considered so great as a change from Socialist to Workers'. While it might be possible to assess the magnitude of changes by analyzing the meaning of different names, this approach is judged to be too problematic, and the assessment is limited to the determination of terms that match in both names. A minor change is defined as one that involves the repetition of one or more terms in both names. A complete change repeats no terms.

The frequency of name changes is handled in a straightforward way as the count of the minor and complete changes over a period of time.

Finally, the recency of the name change appears to be an important component of the institutionalization indicator. For example, it seems that name changes before 1941 (World War II) are a poor indicator of institutionalization for parties studied in 1950-1962. Therefore, the period during which the changes were made needs to be introduced into the concept. No changes before 1541 are counted. Changes between 1941 and 1949 are weighted less than those occurring between 1950 and 1956, and those between 1957 and 1962 are assigned the greatest weight.

Operational Definition. A party is assigned a value on the variable "name changes" according to the sum of its scores obtained by application of Table 3.2, which incorporates the magnitude, frequency, and recency of name changes. The maximum possible score is 18, which is the sum of the values in the last row of the table. Thus a party that changed names from Catholic to Christian Democrat in 1946 and switched to Democrat in 1959 would be scored 2 + 7 = 9 on this variable.

TABLE 3.2 Scheme for Coding Name Changes
Name Changes	1941-1949	1950-1956	1957-1962
No name changes	0	0	0
One minor change	1	4	7
Two minor or one complete change	2	5	8
More than one complete change, including one minor and one complete	3	6	9

Coding Results. As indicated in Table 3.3 for BV102, we were able to offer a code for every party concerning the magnitude, frequency, and recency of its "name changes." The distribution of codes is even more highly skewed than that of BV 101, with almost three-fourths of the parties experiencing no name changes during our time period. The high mean value for AC 102 reveals that the confidence with which these judgments were made tended to be extremely high. In fact, an AC value of 9--our highest adequacy-confidence code--was assigned to more than 100 of our coding judgments for name changes. Scholars who produced the parties' literature in our file tended to mention changes in parties' names if any occurred but to ignore the variable if the parties did not change their names. Thus our coders were apt to assign high AC codes if a substantial body of literature about a party failed to mention any name changes. As a partial result, there is a significant correlation of - .26 between the value of the code assigned to the variable and the extent of our confidence in our assignment. The correlation between BV102 and AC102 means that we are more certain about the coding value that should be assigned to parties that did not change their names since 1941 than we are about those parties that did.

go to page 23