Early school leaving has profound implications at a socio-economic level and planning effective prevention programs within school is crucial for contrasting it. Based on the hypothesis that the dropout is the last step of a process that ends in the student's decision to leave school, we studied the interplay between multiple risk factors of school dropout and the intention to not continue studies, mapping the plausible causal relationships by exploiting a Bayesian network analysis approach. The main aim of this work is to understand how the intention of dropping out of school is affected by contextual and self-perception variables, identifying the variables that play a key role in the network of the risk dropout factors in order to plan a successful dropout intervention program within schools. We analysed data from national standardized tests, collected by INVALSI, of students at the last year of the lower secondary school in Italy, to monitor the students' transition to upper secondary school. The use of network analysis shed light on the factors to work on within the schools to design effective dropout prevention programs.