Instructor Attire and Student Output: Evidence from Randomized Class-Section Assignment

This paper examines whether instructor attire affects not only standard academic outcomes but also students' voluntary written output. The analysis uses data from six first-year introductory "Information (Data Science)" classes at Chiba University of Commerce. All classes were taught by the same instructor under a common syllabus and grading scheme, and instructor attire was randomly assigned at the class-section level, with three sections in the suit condition and three in the casual condition. To reflect this design, the paper reports student-level regressions with class-clustered standard errors together with randomization-based inference. The results show no significant effect of suit attire on final exam scores or attendance. By contrast, suit attire is associated with fewer reaction-paper submissions and fewer words written. The estimated effect on the sentiment score is positive but imprecisely estimated. Overall, the evidence suggests that instructor attire matters less for standard academic performance than for students' willingness to produce written output.