Abstract
Construct validation is as important for the measurement of school outcomes as of constructs in personality or human abilities. A multifaceted inquiry is called for, bringing both psychological theory and empirical findings to bear upon the meaning of achievement test performance. It is proposed that achievement constructs be described in both psychological and behavioral terms, and this procedure is illustrated for the construct of functional literacy. Psychological models of specific skills taught in school are reviewed, and implications for testing are drawn. Achievement tests can be constructed by sampling a faceted item domain embedded in a broader domain of intended but untested instructional outcomes. Studies probing the relation of test performance to this broader domain are then carried out as part of the construct validation. This strategy can yield tests more closely linked to the curriculum, provide information required for defensible standard setting, and rationalize criterion-referenced interpretations for sound research and wise educational policy
Get full access to this article
View all access options for this article.