Klint Kanopka, New York University – Social-Emotional Skills in Education

On New York University Week:  Not all learning in schools comes from the curriculum.

Klint Kanopka, assistant professor of applied statistics, delves into social emotion learning.

Klint Kanopka is an applied statistician who uses computational models to solve problems in educational and psychological measurement. His recent work includes the development of a mixture item response model that accounts for effects caused by variation in item position, innovations in model comparison methods for polytomous item response models, and the use of flexible machine learning models to combine behavioral data and item response data.

Social-Emotional Skills in Education

When we think about our time in school, many of the most important lessons relate to how we navigate social situations and develop into emotionally mature adults. This kind of development requires the acquisition of a ton of different skills, and there is a push to measure and monitor the development of these social-emotional skills in students. Unlike academic skills measured with standardized tests, we measure these social-emotional learning (or SEL) constructs with self-report surveys. We were curious, “are changes in these SEL constructs over time reliable indicators of other academic outcomes?” and “do these relationships vary by demographic characteristics?”

To do this, we looked at almost 50,000 middle school-aged students across six of the largest public school districts in California. These districts conduct yearly surveys of four SEL constructs: growth mindset, which is the belief that your abilities can improve with effort; social awareness, which is the ability to consider the perspective of others; self-efficacy, which is your belief in your ability to succeed and accomplish tasks, and self-management, which is the ability to regulate your emotions, thoughts, and behaviors. Alongside these SEL measures, we focused on two academic outcomes: math and English language arts (or ELA) standardized test scores. Key to our methodology is that we modeled three different relationships: how SEL levels predict changes in each outcome, how changes in SEL predict changes in each outcome, and how changes in SEL predict changes in changes of each outcome.

It is often assumed that changes in self-reports are unreliable, but we find robust evidence that when SEL improves, so do academic and behavioral outcomes for students. Additionally, these relationships are roughly equal across demographic groups. SEL improvements are related to gains in test scores, regardless of the starting level of SEL. Finally, we find that for students with low test scores, growth mindset and self-management improvements are associated with the largest gains, which can range between the equivalent of eight and 45 days of learning.