Investigating the use of negatively phrased survey items in medical education settings: common wisdom or common mistake?

Background: Attitude surveys in medical education often combine negative items with positive items, a "common wisdom" strategy to counteract response sets. A body of research in other fields has demonstrated that negatively phrased items affect reliability and validity by introducing measurement artifact into scores. The authors investigated the effect of negatively phrased items in the Medical School Learning Environment Survey (MSLES) with data from six medical student cohorts at the University of Texas Medical Branch, Galveston, Texas.

Method: This study describes the impact of negatively phrased items in the MSLES through analysis of item and scale means and comparisons of coefficient alpha values.

Results: Findings indicate that negatively phrased items performed differently than the positively phrased items. Negatively phrased items were associated with lower scale reliability.

Conclusion: The authors conclude, as did earlier studies, that negatively phrased items introduce an artifact into attitude measurement. The "common wisdom" practice of routinely including negative items should be employed with care.