Deborah Jones Merritt (Ohio State) has published Bias, The Brain, and Student Evaluations, 82 St. John’s L. Rev. 235 (2008).  Here is the abstract:

Student evaluations of teaching, a common fixture at American law schools, harbor surprising biases. Extensive psychological research demonstrates that students’ assessments respond overwhelmingly to a professor’s appearance and nonverbal behavior. For example, student ratings of professors’ abilities after viewing just thirty seconds of silent videotape footage correlate strongly with end-of-semester evaluations from the same professors’ pupils. Because the nonverbal behaviors that influence teaching evaluations are rooted in physiology, culture, and habit, there is a potential for characteristics like race and gender to affect evaluations. The current method of gathering evaluations, moreover, allows social stereotypes to filter students’ perceptions and increases risks of bias because it taps students’ intuitive cognitive processes—what psychologists have referred to as “System One” judgments.

To improve the accuracy of student evaluations and reduce the taint of bias, schools must design new evaluation systems. Systems based on facilitated group discussion, for example, would awaken the more reflective, deliberative judgments that psychologists term “System Two” decisions. This article illuminates the cognitive processes that underlie all facets of the legal system as it draws upon research in cognitive decision making to present the compelling case for reforming the current system of evaluating professors’ classroom performance.

Here are some quotations from the article that should give any law professor pause (courtesy of Conglomerate):

By relying on the current student evaluation system, law schools implicitly endorse an inflexible, largely stylistic, and homogeneous description of good teaching. Rather than encouraging faculty to use nonverbal behaviors to complement excellent classroom content, organization, and explanations, the present evaluation system largely eliminates the "dog" of substance, leaving only the "tail" of style to designate good teaching. …

[T]he ratings that students award through the present evaluation system bear little relationship to objective measures of learning. The current system of student evaluations, in other words, rewards and penalizes faculty according to relatively trivial indicia, rather than what they accomplish in the classroom.