ABSTRACT
What does come through clearly from the committee's statement is that validity, or rather validation, refers not to a measure in question but to inferences made on the basis of scores obtained on it. In short, "One validates, not a test, but an interpretation of data arising from a specified procedure" (Cronbach, 1971, p. 447). (For a symposium on various aspects of the validation process, see Wainer & Braun, 1988.)
Accordingly, inferences may be more or less valid (appropriate, meaningful, useful), depending on the purpose, the respondents, and the circumstances for which they are made. For example, using scores on a given vocabulary test may afford more valid inferences about individuals' academic achievement than, say, predictions as to how well they will do in college or on a job. The validity of predictions will itself vary, depending on the type of college program or the type of job being considered. It is also clear that the same vocabulary test may be more or less valid, depending on respondents' age, ethnicity, race, educational background, to name but some factors. To minimize ambiguity, it is essential to specify, at the very least, for what, for whom, and under what circumstances are inferences from a set of scores being made.
