A natural experiment dramatically reveals that scores on a difficult standardized mathematics aptitude test predicts achievement in mathematics over a 2 or 3-year period considerably better than the judgment of the students' present mathematics teacher. (Author/AM)

Two competing definitions of test fairness and their differing implications are explained and illustrated, and data from a number of published studies on test bias are reanalyzed; the focus is on unfair bias that may exist in tests that are approximately equally valid for both majority and minority groups. (Author/JM)