Abstract: I study the manipulation of test scores in Italy. Using an experiment that randomly assigns external monitors to classrooms, I apply a new methodology to study the extent of the manipulation and propose a correction method. I find that the manipulation was associated with more correlation in the answers after one controls for mean test scores. It was concentrated in the south and islands region, and it tended to favor female and immigrant students. Finally, the correlation patterns between the amount of manipulation and the number of missing answers suggest that teachers were more responsible for the manipulation than students
