Analysis of Undergraduate Teaching Evaluations in Computer Science

Undergraduate teaching evaluations are widely believed to be biased. We analyzed the evaluations from our department for all classes from 2000 through 2008, looking for specific biases. We looked for correlations between evaluation numbers and whether the course was required; how much time students spent on the course; student classification (freshman, sophomore, etc.), and expected grades. While all of these show correlations with teaching evaluation responses, the effect of each factor is not consistent over all questions and varied strongly depending on the evaluation question and the course level.