An introduction to ERGM models with an application to the study of gender bias in high school

Abstract: The seminar will be centred on Exponential Random Graph Models (ERGM), i.e. a class of statistical models of social network data which can be useful to test hypothetical social processes underlying the formation of social and cognitive relationships. First, the structure of the model will be presented, along with its underlying assumptions. Then, a brief description of estimation methods will be offered with a brief presentation of the available tools. In the second part of the seminar, the application of ERGMs will be showcased by presenting a study on the effect of gender stereotypes in the formation of perceptions regarding others' abilities among Italian secondary school students. We asked a sample of students to choose the four best candidates from their classmates for three hypothetical inter-class competitions in reading, mathematics and science. By modelling the network of nominations through ERGMs, we found that females were more likely to be nominated for the reading competition than males, while the opposite occurred for science, regardless of the students' actual school performance. The study can be found at: https://doi.org/10.1016/j.socnet.2022.09.001.