interrelationships to be discovered. If the observed interrelationships among the concepts conform to the predicted ones and if the variables seem to be measuring what they purport to measure, then the conceptual framework can be said to be "validated" in these senses.

In practice, we find that one of these approaches to validation is somewhat more difficult to pin down than the other. Standard techniques of correlational analysis can be used to determine whether observed interrelationships among variables conform to expected ones. But how does one determine whether the variables are measuring what they are supposed to be measuring? The measurement literature proposed several procedures, linking them with different "types" of validity. McGaw and Watson (1976, pp. 212-213) cite four as the most common:

Face validity consists of little more than the subjective determination by the researcher that the indicator actually measures what it purports to measure ....

Content validity involves the determination that the content of the measure is an adequate and representative sample of the content universe of the property being measured ....

Predictive validity involves the determination that one indicator can be used to predict accurately a particular value, criterion, or position on some other indicator ....

Construct validity involves the determination that the indicator relates to other indicators consistent with theoretically derived hypotheses concerning the concepts (constructs) that are being measured.

McGaw and Watson go on to say that "construct validity is achieved by testing hypotheses that derive from a theory and that predict relationships between variables or indicators" (1976, p. 214). We see that this type of concept validity, by requiring the testing of hypotheses derived from theory, further blurs the already fuzzy distinction between theory validation and concept validity. Our attempt at validation the conceptual framework of the ICPP project will utilize all these procedures to varying degrees. Face validity, which involves only the subjective judgment of the adequacy of the measurement procedures, is the weakest test of concept validity, and no special emphasis will be placed on it in the validation. But face validity is important, and readers who form negative opinions of the conceptual framework's face validity are unlikely to accept its ultimate validity--regardless of the empirical results. In judging face validity by the items proposed as conceptual indicators, readers will also be assessing content validity, deciding whether the samples of items are sufficient for measuring the concepts. Again no explicit arguments will be made on behalf of content validity, but the number and range of the variables in the framework constitutes an implicit argument for content validity.

With face and content validity serving as background. criteria, special emphasis in the validation effort will be on construct validity and, where possible, on predictive validity. Because both procedures involve the tests of expectations concerning relationships among the variables, the process of validating the conceptual framework inevitably involves measurement evaluation in concert with some theoretical explorations.

Additional issues in validation procedures will be confronted in the course of the validation effort. We shall take up each concept from the framework in turn, beginning with a party's external relations with society and the concept of "institutionalization."

EXTERNAL RELATIONS

Institutionalization. Parties vary in their life-spans and in their stability during their life-spans. We attempt to tap this dimensions of variation with the concept of institutionalization, defined as the extent to which the party endures and elicits stable patterns of behavior from those who interact with it or within it. Seven basic variables were originally proposed to tap this concept. The posited relationships are as follows:

1.01 Year of Origin. The more recent the founding of the party, the less institutionalized the party.

1.02 Name Changes. The greater the magnitude, frequency, and recency of name changes, the less institutionalized the party.

1.03 Organizational Discontinuity. The greater the magnitude, frequency, and recency of splits and mergers, the less the institutionalization.

1.04 Leadership Competition. The greater the leadership change through overt processes, the more the institutionalization.

1.05 Legislative Instability. The more fluctuation in legislative representation, the less the institutionalization.

1.06 Electoral Instability. The more fluctuation in votes won, the less the institutionalization.

1.07 Number of Pages Indexed. The more literature written about a party, the more institutionalized it is.

This last variable, number of pages indexed, was soon eliminated from further consideration on the grounds that it more directly measures some aspect of scholarship than of political parties, and it was never involved in item analyses to determine the scalability of the institutionalization items. Scalability was assessed by

go to page 144