The body of work regarding gender bias in academia shows that female instructors are often rated lower by students than their male counterparts. Mechanisms are complex and intersectional and often associated with role congruity theory. Little research has examined parallel patterns in graduate teaching assistant (TA) evaluations. In research institutions, TAs make up a large portion of teaching teams. Identifying bias and working to remove it is critical to shifting the already-well-documented gender imbalance in higher education. To evaluate gender-associated perceptions of graduate TAs' teaching skills, we analyzed Likert-scale, mid-semester survey data using ordinal logistic regression models for PhD TAs in five (pre-COVID) semesters in the College of Engineering at Cornell University, a large R1 institution in the United States. We also regressed scores for each survey question against the overall TA quality rating for male- and female-identifying TAs to compare the strength of those relationships and explore potential differences in student expectations associated with gender roles. A subset of narrative comment data were coded into themes, analyzed, and triangulated with other observed patterns. Male TAs had a higher likelihood of receiving a better rating than female TAs for all survey questions in which students rated performance. Statistical evidence of different slopes of relationships between particular questions and overall TA quality rating suggested that female and male TAs were "valued" more for behaviors/attributes congruent with roles ascribed to that gender in broader society. Female TAs received a higher proportion of positive comments for communication skills and more comments regarding supportiveness than male TAs. Males received more comments about their overall value as TAs, however all comments regarding overall quality as TAs were positive regardless of gender. The amount and proportion of comments that were positive or negative for knowledge, enthusiasm, preparedness or fairness were the same for male and female TAs. Gender-based disparity is occurring in TA evaluations and aligns with patterns observed in research on teaching evaluations for faculty. Correlation between overall TA ratings and scores for specific survey questions and narrative responses indicate that role congruity influences traits that students perceive as important and positive in TAs of different genders.