Association for Public Policy Analysis & Management

Over the last several decades, the teacher effectiveness literature has coalesced around two key findings. First, teachers are the most important in-school factor for improving a range of short- and long-term student outcomes (Chetty et al., 2014b; Jackson, 2016; Nye et al., 2004; Hanushek & Rivkin, 2010). Second, teaching is multidimensional, such that teachers who are effective in one area of practice (e.g., raising test scores) often are not equally successful in others (e.g., supporting students’ social and emotional development; Kraft, 2019; Jackson, 2018).

As this body of work continues to grow, though, researchers have started to uncover nuances in results – both within and across studies – that raise critical questions about potential challenges and tradeoffs of “good” teaching. For example, in recent experimental work using the same dataset as the one used in this paper, Blazar (2018) identified a negative relationship between teacher effectiveness as measured by gains in students’ math test scores versus students’ reports of their engagement and happiness in class. This finding elicits the questions: Are there unintended consequences of raising test scores? How can teachers support students’ mastery of academic content in ways that also engage them in the classroom environment?

We examine this last challenge and tradeoff in the teaching profession by looking inside classrooms, an opportunity rarely afforded researchers on a broad scale. A key design feature of our work is use of an experimental dataset in which teachers were randomly assigned to class rosters within schools. With these data, we first document the relative importance of students’ self-reported engagement and happiness in class versus math test scores when predicting longer-term outcomes in middle and high school. We find that both measures independently predict middle school test scores, and that students’ engagement and happiness in class predicts high school self-reports of their beliefs on the importance of education and plans on attending college. The predictive validity of upper-elementary self-reports of engagement and happiness in class suggest that this measure is not simply noise and, rather, an important indicator of students’ longer-term success.

Second, using observations of teaching practice and data from a teacher survey, we examine whether specific classroom practices explain what produces strong math test scores versus strong engagement and happiness in class. We find that gains in math test scores are predicted by teaching practices (e.g., cognitively demanding math activities, clarity and precision of the mathematics, strong behavior management) in ways that closely align with theory. Comparatively, observation scores do not reveal clear mechanisms and teaching behaviors that support students’ engagement and happiness in class. Counter to intuition, we found that having a teacher who made more mathematical errors in class led to higher engagement and happiness in class.