Post by Stephanie Williams
What's the science?
The term “growth mindset” reflects the belief that abilities are not hard-wired into the brain, but can be developed and improved over time. Growth mindset interventions can be effective in changing the beliefs of students who think their academic abilities are fixed. Recently, experimental work in educational settings has investigated how interventions can use ideas related to a growth mindset to intervene in cases of academic underachievement. This week in Nature, Yeager, Hanselman, and colleagues demonstrate that a growth-mindset intervention can effectively improve grades and enrollment in advanced math courses in particular student populations and school environments.
How did they do it?
The authors analyzed the effects of a validated growth-mindset intervention in a nationally representative sample of high schools. Ninth grade students were randomized to a particular condition (intervention vs. control activity), and teachers, as well as researchers, were blind to the condition of each student. The intervention consisted of two 25-minute online modules. In 25-minute self-administered online sessions, students heard stories from both older students and adults, interacted with guided exercises, and were asked to reflect on their own learning. At the end of the second session, students participated in a math task with two options: an easy but low learning task, and a challenging but high learning task. The authors used the percentage of students in the control condition that chose the challenging problem as a metric of challenge-seeking norms for a particular school. The authors monitored the fidelity of the intervention implementation by analyzing the percentage of the module screens students viewed (97%) and analyzing the number of open-ended questions that students responded to (96%).
The authors were interested in identifying which contexts were most conducive to the growth mindset intervention. They analyzed treatment effect heterogeneity by looking at the types of school contexts (availability of resources, motivation to follow an intervention, etc.), and peer contexts (supportive vs. unsupportive of taking intellectual risks) The authors point out that schools on both ends of the resource spectrum could fail to see an effect of the intervention: schools with low-quality curricula don’t offer as many learning opportunities for students, and schools with ample resources may have less of a need for the intervention. The authors assessed the effect of their intervention on the mindset of students, as well as the effect on their GPAs and advanced mathematics course enrollment. They looked at GPAs of all core courses, and also at math and science GPAs in their secondary analysis. These subjects are of particular interest because of the commonly held belief that success in math or science is innate. The authors also re-analyzed the data using a Bayesian machine learning algorithm and compared the findings from both sets of analyses.
What did they find?
The authors found that lower-achieving students who received the intervention performed better than lower-achieving students who did not receive the intervention. This was true when the effect was quantified by assessing overall GPA as well as math-science GPA. The intervention was more effective in low and medium-achieving schools than in high-achieving schools. The authors found no significant difference in the treatment’s effectiveness in low vs. medium achieving schools, but point out that there is high variability in student performance in low-achievement schools. The intervention did still have an effect on students in high-achieving schools. For the top 25% of high achieving schools, the intervention increased the rate at which students enrolled in advanced math courses the following year by 4 percentage points. Students in school environments with peer norms that supported the adoption of intellectual challenges showed larger GPA benefits from the treatment. The authors interpret this result as suggesting that students in unsupportive environments may feel social pressure not to take intellectual risks in front of their peers. They also suggest that beliefs can affect how students react to ongoing academic challenges. The results from the Bayesian machine learning analysis confirmed these findings.
What's the impact?
A low-cost growth mindset intervention can have a substantial effect on grades and enrollment rates in advanced mathematics courses. The findings around context-specific treatment effects (e.g. school resources or peer environment) can be used to inform future types of interventions. Future interventions should target the messages students receive about learning ability from schools and assess additional challenges faced by adolescents, including social and interpersonal difficulties.
Yeager, D, Hanselman, P., et al. A national experiment reveals where a growth mindset improves achievement. Nature (2019). Access the original scientific publication here.