The purpose for which a test is used and the examinees' stage of learning are two anchor points that are incorporated into a suggested framework for conducting construct validation studies for tests of students with English as a second language (ESL) or English as a foreign language (EFL). The framework includes the use of generalizability theory, construct definition, and factor analysis. Two procedures are discussed in terms of the means by which they confound reliability and validity. Difficulties with correlational evidence to support construct valid test score interpretations are also discussed. Problems with product-moment correlations, the Rasch model, and a construct validation technique proposed by G. H. Henning are singled out for particular criticism. (Author/TJH)