Classroom assessment research in the United States has shifted away from the examination of teacher-made tests, but such tests are still widely used and have an enormous impact on students' educational experiences. Given the major shifts in educational policy in the United States, including the widespread adoption of the Common Core State Standards, I argue that researchers should examine the tests and quizzes that teachers create and administer in order to determine whether those policies have had the intended impact of teachers' assessment practices. Furthermore, these investigations should be grounded in discipline-specific conventions for developing and demonstrating knowledge. I then propose a research-based framework for analysing mathematics exams that focuses on alignment with learning goals, cognitive complexity, variety of task formats, attentiveness to culture and language, and clarity of expectation. This framework is meant to be used formatively, helping researchers, administrators, and teachers identify strengths and areas for growth.