“Journal of Marketing Education” (from Sage Publishing) has published a study by David Ackerman of California State University Northridge and Christina Chung of Ramapo College of New Jersey about the presence of bias in online student evaluations of their professors. Their research compared actual on-campus ratings versus online ratings for the same professor and class and found that the latter tend to reflect whatever tone early ratings give. Thus, if a professor gets a couple early very bad ratings, later ratings are typically not so good. The opposite also holds true. They theorize that a kind of peer pressure, once removed, may be at work. But I would just call it “framing” in the Kahenman-Tversky sense.
What I like best of all about their work is their characterization of “Rate My Professor” and similar forums as the online equivalent of bathroom stall walls–a place to vent one’s frustrations.
Now what we need is more research into on-campus ratings in terms of their accuracy.