ABSTRACT
One broad goal of this chapter is to encourage more work towards an appropriate balance of both objective and subjective issues. In broader terms, the contemporary statistical models which have used here can be related to more classical issues of construct validity. These concepts include: sampling bias as a form of ecological validity, selection bias as a form of discriminant validity, prediction bias as a form of external validity, and measurement bias as a form of internal validity. The kind of models used to address these issues are relatively complex, but they all have some technical and conceptual benefits for the kinds of test-bias problems discussed. The substantive problems discussed in this research are timely for several reasons. Specifically, the colleges of the NCAA are currently engaged in heated debates on issues of eligibility and academic performances. These questions concern appropriate use of information from high school transcripts and from standardized test scores.
