Here's a shocker: when dealing with a virtual instructor, people prefer an avatar that looks and acts like them
. A collaborative study between George Washington University and North Carolina State found that participants reacted more positively to virtual instructors who were the same gender and ethnicity as themselves. Oddly, they found that these factors alone did not improve responses. Gender had no noticeable effect on its own, and the being presented with an avatar of the same ethnicity alone actually made responses from participants worse. It was only when both factors were combined that participant engagement improved, though the effect was relatively small. The more important factor, though, was how the virtual avatars measured progress. Researchers found that the most effective virtual instructors
were those that customized their feedback style to a student's preferences (e.g. being measured against her own progress or against other students).