Large-scale observational measures of classroom practice increasingly focus on opportunities for student participation as an indicator of instructional quality. Each observational measure necessitates making design and coding choices on how to best measure student participation. This study investigated variations of coding approaches that may be feasible in large-scale studies, and the ramifications of these variations for drawing inferences about instructional quality. Using data from classroom videos, we found that decisions about whether to keep track of individual students in the coding, observe multiple contexts in the classroom (e.g., whole-class and small-group discussions), and capture nuances of student participation changed the resulting characterizations of classroom practice. Most importantly, simplifying the coding approach did not fully capture and even misrepresented the level and nature of student participation in many classrooms.