Teacher assessments versus standardized tests : is acting "girly" an advantage?

We study if Italian teachers do apply gender discrimination when judging students. To this aim, we use a difference-in-differences approach that exploits the availability of both teachers (non-blind) and standardized test (blind) scores in math and language that Italian students receive during the school year. Using data for all sixth graders, descriptives show that in both scores girls are better than boys in the language scores, while in math boys perform better than girls in the blind test. Moreover, our analysis suggest that boys are always discriminated by teachers in both subjects. This result holds also when we control for class fixed effects, students noncognitive skills, gender specific-attitude towards cheating and possible cultural differences towards gender attitudes in math or language.