This paper grew out of a presentation made as part of a panel discussion on the Bakke case, and is second in a series of Institute for the Study of Educational Policy documents on important educational issues and problems. Many of the factors that impact test performance or scores are examined. The primary thesis of the paper is that it is psychometrically legitimate and practically sound to supplement test scores when tests are used as selection devices in a decision-making process. Limiting dependence on tests is a practice that will result in greater fairness to more individuals. The Bakke case is used as the source of basic data. A discussion of the factors affecting test scores is presented, and the issues of validity and selection are discussed. Specific areas of concern are: measurement for selection; origins and identification of test bias; bias resulting from improper use; using regression analysis in determining predictive validity; and the issue of overprediction. Recommendations supporting the paper's thesis are offered. (Author/GSK)