This article (11 page PDF) contains a pretty good literature review and summary discussion on cheating. It "introduces a unique multiple choice exam design to observe and measure the degree to which students copy answers from their peers." What's fun is that it provides "a measurement of actual cheating frequency among students, as opposed to relying on reported cheating in anonymous surveys." The studied measured whether less cheating would occur if students were assigned seating randomly (thereby breaking up peer groups) and - surprise! - less cheating occured. Total N is 160, making it one of the larger studies in January's ijSOTL.