Women are underrepresented at multiple levels of physics education. One avenue for understanding the classroom experiences that perpetuate underrepresentation is physics identity, defined using the three dimensions of recognition, performance, and competence. Existing literature suggests that women tend to have a much weaker physics identity than men and that women tend to report a lower sense of competence in the form of self-efficacy than their male peers. This study examined confidence and self-efficacy as an aspect of physics identity in an AP Physics 1 class using a mixed-methods, sequential explanatory design. The quantitative data consisted of students' actual and predicted scores on in-class assessments, which showed no statistically significant difference in the accuracy of students' self-assessments by race or gender. To identify classroom activities that impacted self-efficacy, we collected responses to an open-ended prompt and conducted student interviews. Labs emerged as having both a positive and a negative impact on self-efficacy on many students, regardless of race or gender and male students were more likely to discuss peer-to-peer interactions as a source of self-efficacy. Boys also described figuring out how to apply concepts from labs to problem sets as an experience that contributed to their self-efficacy, while the only girl who mentioned problem sets described them as a negative experience. When describing evidence their teacher believed they are good at physics, boys focused on assessments where they had high scores, while girls focused on the feedback on assessments where they had low scores.