Does Size Matter? Searching for Rhyme or Reason in Course-End Student Surveys in a Large, Eclectic Physics Department

Over the past eight years, surveys of student opinion have been collected (near term-end) in all courses at Utah State University using the same survey instrument. The instrument consists of 25 questions, each of which can be responded to by choosing an integer ranking between 1 (``very poor'') and 6 (``excellent''). The University reports a statistical summary of all surveys each term in which all responses are treated equally irrespective of class size (a factor University administrators have asserted is negligible). Discussions of survey content at USU usually focus solely on two items: ``rate the course'' and ``rate the instructor.'' To some extent faculty tenure, promotion, and salary are based on these two aggregated data. Because of their possible impact on faculty careers, I have examined all responses in all surveys collected over the years in my department. Typical of social data, these results exhibit substantial variability and are highly non-normal. Appropriately treated, however, they reveal a significant class size dependence on the two ``rate the...'' global items. This fact harbors potentially important policy considerations for departments (like mine) that have a broad range of class enrollments.