Originally published 06/01/09

Are customer satisfaction surveys a fair and unbiased tool for for assigning employee bonuses?

No, according to a study published in the Academy of Management Journal, An Examination of Whether and How Racial and Gender Biases Influence Customer Satisfaction. According to lead author David Hekman, assistant professor of management at the University of Wisconsin, surveys “are highly reliable—but they are reliably wrong.” The authors believe that customer satisfaction surveys are biased because they are “anonymous judgements by untrained raters that usually lack an evaluation standard.”

Hekman conducted several experiments to measure customer satisfaction. In one experiment, subjects watched videotaped interactions between a bookshop sales clerk and customers, and were asked to imagine they were the customer and to rate the bookshop’s service performance. Three actors played the part of the sales clerk—a white male, a black male, and a white female. All used identical settings and scripts.

The subjects shown the white male clerk rated the bookshop’s service 19% higher than subjects who viewed the other two actors.

If we use customer satisfaction measures to assign bonuses, can we assume we’re not compounding race and gender bias? If this assumption isn’t correct, how can we make survey-influenced compensation systems fair? Could other attributes not measured in Hekman’s study—for example, weight, age, and appearance—create similar results?