The unfulfilled promise of criterion-referenced measurement is that it would permit valid inferences about what a student could and could not do. To come closest to achieving all that criterion-referenced testing originally promised, tests of higher item density, with more items per amount of domain, are required. (SLD)

The term "criterion-referenced measurement" has been so frequently used that accumulated surplus meaning has sometimes swamped the fundamental concept. Stripping away extraneous interpretations affirms the contemporary relevance of the concept. Substituting the term "construct-referenced" underscores the need for better…

The instructional impact of criterion-referenced testing during the past three decades is reviewed, and a method of devising criterion-referenced tests is suggested that departs from the usual model. Adding verification along the lines of panel review that newly constructed test items do elicit appropriate information is suggested. (SLD)

Cohen and Hyman's response contains several misunderstandings of the original article by Mehrens and Kaminski. One frequently wishes to make inferences to a domain from a test, but teaching a specific performance and testing for that performance does not allow for a domain inference. (SLD)

The scales of the National Assessment of Educational Progress (NAEP), as constructed, do not yield meaningful criterion-referenced interpretations. Poorly defined NAEP goals and the present knowledge base do not allow the measurement of what examinees can and cannot do. Inappropriate interpretations of NAEP data are discussed, with specific…

A sample (N=37) of currently published achievement tests was surveyed as to the availability of five types of technical data: (1) item selection techniques; (2) standardization; (3) types of norms; (4) types of validating data; and (5) types of reliability data. Recommendations for publishers and cautions for test users are given. (BS)