Defining Constructs and Assessment Design

This chapter describes the evolution of constructs and illustrates current construct issues that affect the design of second language assessments. To understand the term construct better, the chapter begins with historical overviews presented from two perspectives. First, educational and psychological measurement ideas are reviewed in terms of the Standards for Educational and Psychological Testing (the Standards). The Standards have represented a consensus among American scholars and practitioners on what information is most helpful and important for guiding the development and use of tests. Second, language testing ideas are reviewed using the Standards as a backdrop. In both overviews, we can see shifts in meaning that have led to our current state. The second part of the chapter discusses four inter-related issues that affect current and future designs of second language assessments: the relationship among measure, construct, and theory; a move away from the 1999 Standards view of construct validity and toward an argument-based approach; the way one's worldview affects the construct definition; and the way the inferences we want to make from test scores are influenced by the test developer's view both of the world and of validity. By better understanding the various meanings of construct in second language assessment, test score users, students, and researchers alike should be alerted to look beyond the term itself, and examine the nuances of its use. As test developers adopt an argument-based approach to validity, they should be aware that constructs are not always necessary in assessment design, but if they are included there is a responsibility to define them in relation to a model or a theory, and to examine the predicted relations. Keywords: Assessment evaluation; Assessment methods; Northern America