Background and Context: Gender research shows that women are better at reading comprehension. Other studies indicate a lower tendency in women to choose STEM professions. Since data modeling requires reading skills and also belongs in the areas of information systems and computer science (STEM professions), these findings provoked our curiosity. Objective: Comparing student modeling errors across different error categories; Finding out whether learning the notations of one type of modeling, class diagram (CD), leads to fewer modeling errors in the next type, entity-relationship diagram (ERD); Finding out whether the reading advantage of women would manifest within a STEM activity of database modeling. Method: We conducted a within-subject design experiment with college students. We manipulated the modeling type, which is a CD or an ERD, and observed its impact on the type and number of errors in modeling tasks. Findings: Certain modeling aspects that are at a higher level of abstraction are much more prone to errors. Male students made more errors. There is a learning effect from one type of model to the other. Female students had much more significant learning from one model to the other. Implications Educators should invest time and effort into applying modeling activities where the error rate is higher. Since female students have an advantage over male students in several modeling aspects, it is further proof that female students' lack of propensity for the STEM professions is based on a perceptual barrier regarding their ability to succeed and not on an inferiority in these domains.