SCALING AND TEST THEORY1•2

The chapter heading. "Scaling and Test Theory," is new to the Annual Review, but the topic itself is not. In previous volumes, authors of the chapters on statistics have routinely devoted up to half of their space to material in this field. Emphasis in the past, of course, has been on those aspects of psychological measurement most closely related to statistics. As a result, much of the more experimentally oriented work on scaling has been covered only casually. In the present review the emphasis is re­ versed. Several books of more than usual inter est have been published during the past year. Churchman & Ratoosh (17) edited a volume of papers origi­ nally presented at a five-session symposium on measurement held during the 1956 meetings of the American Assoc iati on for the Advancement of Science. The volume is divided into four parts: meanings of measurement, theories of measurement, problems in the physical sciences, and problems in the social sciences. Except, perhaps, for the physical sciences section, each part contains papers directly relevant to psychological measurement. Gullik­ sen & Messick's Psychological Scaling: Theory a1ld Applications (51) is a series of pape rs presented at another conference on measurement. This one, held at Princeton in 1958, cov ers new developments and ideas in nearly all branches of psychological scaling. Chapters of both volumes are dis­ cussed separately in appropriate sections of this review. Thurstone's The Measurement of Values (129) is another important volu me of pape rs on psychological measurement. The volume brings to­ gether in a single source Thurstone's many original contributions to the theory and methodology of scaling. Most of the chapters are concerned with the development, extension, and applications of a single general model to different experimental procedures and to different content areas. Thur­ stone's judgment model gave us a single rationale for relating the methods of paired comparisons, rank order, and category rating and sorting. It has been used for measurement of psychophysical attributes, attitudes, values, and preferences. Essentially the same general notions have also served as a basis for multidimensional scaling models, detection theory, and one of the general models for measurement of ability (130). Luce's Individual Choice Behavior (82) presents us with an alternative general model. In this volume, Luce develops his simple, but powerful, 1 The survey of the literature pertaining to this review was concluded in