Panel Paper: Using Student Behavior As a Proxy Measure of Self-Control: Evidence from Positive Behavior Interventions and Supports (PBIS) Systems

Thursday, November 8, 2018
Lincoln 3 - Exhibit Level (Marriott Wardman Park)

*Names in bold indicate Presenter

Martin R West, Katherine Klenk and Ethan Scherer, Harvard University

A growing number of schools use Positive Behavioral Interventions and Supports (PBIS) systems to manage students’ behavior and support their social-emotional development. One byproduct of PBIS implementation is the generation of detailed daily data on students’ behavior in school, but these data are rarely captured and to our knowledge have not been systematically analyzed. In this study, we use a unique dataset comprising individual-level PBIS data on student behavior, annual survey-based measures of social-emotional (soft) skills, and administrative records from three charter management organizations (CMOs) serving 1,500 students in grades 6-12 across 6 campuses. We use this dataset to explore the potential of PBIS data to serve as an “unobtrusive measure” of students’ ability to control their behavior, a key social-emotional skill typically tracked via student surveys.

Our descriptive analyses reveal that, on average, students in these schools receive 334 merits and 175 demerits annually, confirming that students receive frequent feedback on their behavior. Perhaps surprisingly, the number of merits and demerits a given student receives are positively correlated, a pattern that could reflect efforts to ensure that students receiving large numbers of demerits receive sufficient positive feedback. Merits and demerits accrue at a fairly constant weekly rate throughout the school year. Further, the number of merits and demerits specific students receive are highly correlated across semesters, suggesting that they are a reliable indicator of student behavior.

We supplement this descriptive analysis with a series of regressions designed to shed light on the relationship between PBIS feedback, survey-based measures of social-emotional skills, and academic outcomes. We show that the number of demerits and (conditional on the number of demerits) the number of merits students receive are both strong predictors of student self-reports of their self-management skills gathered at the end of the prior, concurrent, and subsequent academic year. Relationships with other social-emotional skills are in the expected direction but weaker, providing some evidence of discriminant validity. We also show that demerits and merits are strong predictors of students’ course grades in subsequent years, with a combined predictive power exceeding that of survey-based measures of self-control.

While most research on the development of social-emotional skills has relied on measures based on self-report surveys, these measures may suffer from such problems as social desirability bias and reference bias due to differences across school contexts. The likelihood of gaming renders self-report surveys inappropriate for use in situations in which stakes for educators or students are tied to their results.

Our results suggest that PBIS data could complement self-report surveys for the purpose of tracking the development of self-control in school settings. In particular, PBIS data may be useful for identifying students in need of support and in evaluating the success of strategies aimed at improving student skills even within a single school year. While our conclusions are limited to the charter schools in this study, the fact that an estimated 25,000 schools nationwide report that they are implementing PBIS implies that they could have broad applicability.