Assessing the Quality of Instruction In University Economics Courses: Attrition as a Source of Self-Selection Bias in Mean Test Scores

This paper examines the hypothesis that the quality of instruction influences observed mean test scores through its effects on both individual student performance and on class attrition rates. Since these effects are expected to display opposite signs, the level and dispersion of these scores are biased downward. It is shown that hypothesis tests based on simple mean test scores may yield unreliable and possibly misleading results. Data from classes in the principles of economics are employed to test the hypothesis.