Student evaluations of teaching (SETs) are frequently used to assess college teachers. However, education research has shown that there is potential for bias in SETs, especially based on instructor variables. Aside from Amy Dayton's 2015 work on assessment that advises using SETs only in concert with other measures, English studies scholars have not extensively studied SETs in many decades. This project codes 1,074 questions from 55 SET forms at different higher education institutions by institution, question theme, content, subject, and sentence subject as well as thirteen categories based on Herbert Marsh's SEEQ form, which serves as the basis for contemporary SETs. Our analysis reveals that instructors are often placed as grammatical subjects in questions despite their lack of agency over some areas such as participation and student devaluation of alternative teaching practices. Such question formation skews students toward evaluating instructor factors that are, at best, uninformed, and, at worst, biased. We recommend that SET questions align with student learning and/or engagement rather than teacher performance, rewriting questions to put students in subject positions and cue them on best practices as a better reflection of rhetorical theorizations of classroom agency.