Class Absences and Grades
With my move to Seattle University, the opportunity has arisen to re-examine my attendance and preparation policy for class. During the past two years, I required students in all of my courses to show up on time to class and to be prepared to discuss the assigned material. If they were tardy, absent or unprepared, I deemed them absent for that class session. My rule was to withdraw a student from the course who ended up being deemed absent for more than 25% of the scheduled class sessions. Having amassed attendance and grade data for 5 courses (2 first-year courses and 3 upper-class electives) and 223 students, I couldn’t resist the temptation to figure out whether class absences affected my students’ final grades. I’ve been crunching the numbers the past few days with Stata and have been surprised by what I found.
By way of background, approximately 80% of the students enrolled in my courses were deemed absent for three class sessions or less (and this figure includes the 17% who had perfect attendance). The remaining 20% were deemed absent for anywhere from three to seven class sessions. It’s worth noting that I did not begin taking attendance at the first scheduled class session and generally waited for a couple of class sessions before I began doing so, especially for upper-class electives given the shopping period provided to 2Ls and 3Ls. Keeping this in mind, and the fact that tardiness and unpreparedness counted as absences for my courses (although, based on my recollection, the majority of absences were “true” absences), students in my courses were deemed absent, on average, for approximately 9% of the class sessions for which attendance was kept. (Excluding those with perfect attendance, this figure jumps to 11%.) Given that no penalty was triggered until a student was deemed to be absent from 25% of the class sessions for which attendance was kept, I’m generally pleased by these figures.
But what effect, if any, did class absences have on final grades in my courses? With my first cut at the data, I’ve run a linear regression of final grade (which, for all courses, was based solely on an anonymous, in-class exam) on class absences. It seems as if class absence did influence a student’s final grade in my courses (i.e., it had statistical significance). That said, the variable has limited explanatory power: It accounts for only 5% of the variation in grades among students. Reference to the regression coefficient for class absences further suggests that the variable had only a slight influence on final grade: According to the model, a student’s final grade decreased by approximately .08 quality points (on the 4.0 grading scale) with each absence. Put another way, a student’s final grade dropped by nearly a third of a step (e.g., from B+ to B) with every four absences.
So what do I make of the data at this point? By requiring students to have attended at least 75% of class sessions, perhaps I reduced the adverse effect class absences could theoretically have on final grade. It strikes me that students with an excessive number of absences (e.g., 50% and upward of scheduled class sessions) would be most affected in their exam performance. But students who took my exams didn’t fit that profile, and maybe that’s why class absence accounts so little for the variation in grades in my past courses. On the other hand, maybe it’s the case that the material I presented in class didn’t add much to what the students were getting from studying on their own. I really hope that’s not true! Ideally, I’d like all students to have had the same amount of classroom instruction going into the final exam. If my policies encourage students to attend all or the overwhelming majority of my classes, and if, in turn, that minimizes the effect of attendance as a determinant of final grade, then perhaps that’s one good reason to continue my policy.