Oppenheimer writes in favour of opinionated teaching.

Kelly-Woessner and Woessner find that this may not be the best route to strong teaching evaluations. Both perceived ideological and partisan differences between student and instructor reduce course evaluation scores, with partisan differences generally being stronger in both economic and statistical significance. They conclude:

Somewhat surprisingly, the results of this study suggest that, from a purely strategic standpoint, instructors who come across as political moderates are not necessarily safe from criticism. In a
class filled with liberal Democrats, even a political moderate would frequently find himself at odds with the students as a whole. Consequently, if the goal were simply to win the love and adoration of the students, clever instructors would merely pander to the median “voter.” By mimicking students’ views and reinforcing long-held beliefs, professors might score well on student evaluations, while
providing no useful information at all.

Ilya Somin asks if politicians pander. Do profs? Is the classroom equivalent of the gerrymander a strongly ideological first lecture inducing exit by those who’d otherwise wind up giving low ratings? We’d then expect pandering in core classes and gerrymandering in electives. I wonder if the Woessners’ data is rich enough to run lecturer-level fixed effects and look at whether the lecturer’s perceived ideology varies across core and elective courses.