Cheap talk in the classroom

In this paper, I offer a theoretical explanation of the robust gender differences in educational achievement distributions of school children. I consider a shot cheap talk game with two different types of senders (biased teachers and fair teachers), two types of receivers ("normal" and "special" pupils) and uncertainty about the sender type on the side of the receiver. I demonstrate that the group of pupils who, in expectation, get either too much or too little encouragement will have less top achievers and a lower average achievement than the group of pupils who get a more accurate feedback message, even if the prior talent distribution is the same for both groups of pupils. -- Cheap talk ; Education ; Discrimination ; Gender