Clinical vs. statistical prediction is only half the problem-and the last half at that. The prior problem, largely neglected, is clinical vs. mechanical measurement-for data can be collected, as well as combined, in either way. Such neglect promotes an incomplete and mismatched dialogue where "clinical" and "statistical" may have different meanings to different persons. Examining clinical and mechanical methods of both measurement and prediction provides a broadened framework that defines the several possible "clinical" and "statistical" methods-and their combinations. Applying this framework to 45 studies shows an apparent superiority for mechanical modes of both data collection and combination, and also suggests that the clinician is more likely to contribute through observation than integration. Grossly uncontrolled differences, however, in clinical training, subjects, criteria, etc., prevent definitive conclusions. To achieve more adequate comparison requires certain specified methodological improvements.

This research (a) scales quantitatively the 30 cultural characteristics oj Murdock's (1957) World Ethnographic Sample, (b) correlates them over all 565 societies, (c) HE purpose of this research is to structure the large body of data pre-T sented by Murdock (1957) into a concise but comprehensive form that permits the efficient extraction of its major significance for cultural theory and cross-cultural methodology-and to illustrate the factor-analytic method that makes this possible. QUANTITATIVE SCALING OF THE WORLD ETHNOGRAPHIC SAMPLEThe data comprise 30 basic economic, ecological, social, and political characteristics for a sample of 565 societies ''. , . designed to be as representative as possible of the entire known range of cultural variation" (Murdock 1957: 664). Murdock rated nearly all the ethnographic materials himself and provided no assessment of reliability, though one later study (Ember 1963) shows moderate agreement with his ratings of political integration.Although, as Murdock acknowledges, some errors almost surely occurred in collecting, rating, and tabulating the data, such errors affect the present process of determining relationships over a large number of societies much less than they would the characterization of individual societies. (The data employed in this analysis include "addenda and corrigenda to January 20, 1959"-incorporated into a reprinted version [Murdock 19611 of the original article [Murdock 19571.) Murdock's ratings produced simply unordered categories (an average of about seven for each characteristic), but the data possess more regularity than such merely nominal scaling implies. It was in fact possible, within each characteristic, to order all categories on a single continuum, of up to four points. For social stratification, for example, Murdock's five original categories were grouped into three levels. The highest level-presence of social classesincluded the categories "three or more social classes" and "hereditary aristocracy or noble class differentiated from ordinary freemen." The middle level 708

Since the theory of games was first made widely available, with application to economic behavior, its use has been suggested in many other areas, from the global to the individual. Several correspondences between game theory and certain aspects of political process have been noted.The contribution of game theory to substantive knowledge in the empirical sciences, however, has been modest; Luce and Raiffa judge that its impact has been greater in applied mathematics. The area of political behavior—despite the apparent applicability of the notion of conflict of interest—is similarly lacking in studies, although a few notable exceptions exist.