Confidentially is not enough: framing effects in student evaluation of economics teaching

Contrary to previous research we show lack of anonymity is associated with large positive shifts in student evaluation of teaching. The results are consistent with the simple observation that due to higher expected future earnings economics and business students have more at stake in terms of potential retaliation by an instructor. The observed positive bias is strongest for international students. Our analysis is based on both a comparison of distributions and ordered probit multi-variate regression. These methods overcome the statistical problems associated with previous studies which looked at differences in means for ordinal responses.