The findings and conclusions from research on predictive validity among markedly different groups are discussed. Empirical findings using the regression line neither support nor contradict the existence of bias or the hypothesis of differential validity. Conclusions drawn from the research are questionable. Elimination of bias in a test, contrary to published reports, is shown to be beneficial to the low scoring group. As a result of these considerations, it has been determined that the main issue is bias in the test, and the related issues of differential validity and differential prediction cannot be addressed until bias has been convincingly detected and eliminated from the test. Ten tables and figures illustrating the findings are appended. A 12-item bibliography is also included. (Author/BW)