Association for Public Policy Analysis & Management

Responding to the need to look beyond test scores to measure school quality, an increasing number of school districts are striving to incorporate socio-emotional learning (SEL) measures in their accountability policies. Growth mindset – believing that intelligence and talent can change, as opposed to being fixed (Dweck, 2000) – is one of these measures. Experimental research has found that developing a growth mindset can improve academic achievement (Good et al., 2003; Aronson et al., 2005; Blackwell et al., 2007; Yeager et al., 2014; Paunesku et al., 2015, Yeager et al., 2016) and that schools can affect students’ mindset (Kraft and Grace, 2016). However, there is no information on how mindset varies across and within American school districts or whether large-scale measures of mindset would behave consistently with what previous, highly-controlled, experimental literature predicts. This study fills this gap by using data from five school districts in California that measure growth mindset for all students in 3^rd to 8^thgrade to assess the extent that students with stronger growth mindset learn more in a given year than those without.

We first describe the characteristics of students and schools that predict a growth mindset. We find, consistent with previous research (Claro et al., 2016; Kraft and Grace, 2016) that disadvantaged populations tend to report lower levels of growth mindset than their peers. These gaps are greater at schools with higher academic achievement or lower concentration of students with ELL status.

We continue by analyzing the effect of student mindset on academic achievement. We address issues of self-selection into schools and potential reference frame bias (Duckworth and Yeager, 2015) by using schools and grade fixed effects. In addition, we estimate models increasingly saturated by controlling for one and two previous years of achievement, not only linearly but also as quadratic and cubic, as well as for student characteristics and their responses to survey items that measure other facets of social-emotional development (i.e., self-efficacy and self-management). Findings show that, on average, students with a mindset measure that is one standard deviation more growth-oriented increase their test performance by approximately 0.02 to 0.03 standard deviations in math and English language arts. We find no evidence that the relationships between mindset and learning vary by ELL status, gender, FRPL status or racial subgroup, but they do vary per previous achievement scores and are greater in higher grades. Mindset is more predictive of learning for students who begin at lower levels of performance. We conduct a variety of additional analyses in order to test the robustness of our primary estimates.

Our results provide evidence that beliefs students hold on the potential to increase their skills and intelligence predict students’ learning independently from other SEL measures and their previous achievement. This presents initial evidence of the benefit of tracking and supporting the developing of growth mindset. Going forward, it will be critical for researchers and policy makers to identify the reliability of growth mindset measurements and the distortions that could occur when measuring growth mindset with stakes attached.