This note evaluates the scrambled questions penalty using multiple choice tests taken by first-year undergraduate students who follow a microeconomics introductory course. We provide new evidence that students perform worse at scrambled questionnaires than at logically ordered ones. We improve on previous studies by explicitly modeling students individual skills thanks to a fixed effects regression. We further show that the scrambled questions penalty does not differ along gender but varies along the distribution of students’ skills and mostly affects students with lower-intermediate skills.