Collaborative learning strategies are widely used in higher education to deepen learning, promote team-building skills and achieve course learning objectives. Using peer evaluation is an important strategy to ensure that engaged and active students are rewarded for their efforts, and to discourage loafing within groups. However, less is known about what biases may influence students' peer evaluations. In this paper, we investigate what variables students may (consciously or unconsciously) use to evaluate their peers. We explore the role of sex, race, course performance and group leadership on peer evaluation. We also investigate whether these variables correlate with students' final course grade. We found that students who reported being leaders in groups were evaluated higher than peers who reported being followers, and that course performance positively correlated with peer evaluations. White students received higher peer evaluations than students of colour. This difference reflects trends in group leadership and course performance, with more white students than students of colour reporting being leaders in groups and receiving higher grades.