ABSTRACT
Response generative modeling (RGM) is an approach to psychological measurement that involves a “grammar” capable of assigning a psychometric description to every item in a universe of items and is also capable of generating all the items in that universe (Bejar & Yocom, 1991). Such an approach to measurement, if feasible, could have at least three important implications. First, the interpretation of scores from a generative instrument would be greatly facilitated because the process for generating the item is explicitly stated. Second, the possibility of generative modeling implies that we have a complete understanding of the underlying response process. Such knowledge might allow us, in turn, to abandon the multiple-choice format in favor of open-ended formats, a long-standing desire of psychometricians (e.g., Frederiksen, 1990), but without the expense associated with scoring open-ended responses. In other words, the same knowledge base that is used to create items can be brought to bear on the scoring of open-ended responses. Third, the ability to assign a psychometric description to an item is the key ingredient in what might be called intelligent test development aids. Job aids, in general, are rapidly becoming the key to increased productivity in many fields (e.g., Harmon, 1986). In a testing context, test development job aids might become essential if bills to outlaw pretesting succeed in becoming law (because it is through pretesting that test developers estimate the difficulty of an item before the test is administered in a final form), especially in light of growing statistical theory designed to allow equating tests “with little or no data” (Mislevy and Sheehan, 1990). Some speculations on the future of job aids for test development can be found in Bejar (1989); a discussion of open-ended assessment from a generative perspective, with special emphasis on certification testing, can be found in Bejar (in preparation); see also Baker (1988) and the Summer 1989 issue of the Journal of Educational Measurement.
