Uncovering Variability in Children’s Classroom Behavioral Regulation Using Wearable Accelerometers

• In Event: Leveraging Wearable Sensors to Access Processes of Children’s Everyday Behavioral and Physiological Regulation

Abstract

Preschool children sometime struggle to inhibit impulses and choose behaviors appropriate for the classroom. Although between-person differences in behavioral regulation are well documented, little work has examined within-person variability. One reason may be measurement constraints. Traditional assessment paradigms such as adult reports of behavior excel at capturing broad individual characteristics but were not designed to observe short-term fluctuations in children’s behavior. Researchers may be able to use wearable sensors to collect multiple observations of children’s behavior in their everyday environments. A recent feasibility study demonstrated that wearable accelerometers can detect children’s classroom behaviors related to behavioral regulation (Koepp et al., 2022). However, those findings emerged from a small study and it is not yet clear whether the accelerometer method is suitable to detect within-child differences. The goals of this study are to 1) further validate this approach in a new sample and 2) use the approach to explore variability in children’s behavior across classroom activities and school days.

Participants were 65 three- to six-year-old children (Mage (SD) = 4.5 (0.84), 53% male, 66% Non-Hispanic White) enrolled in a university laboratory school in the southwest United States. Parents provided informed consent for their children’s participation. Children wore accelerometers (ActiGraph BT3GX devices; ActiGraph, Pensacola, Florida) around their waists each school day for one school week. We used an algorithm developed by Pate and colleagues (2006) to determine when children were non-sedentary. Following Koepp et al., (2022), we identified the number of times a child initiated movement in their classroom during indoor, center-based play. To report children’s general impulsivity, teachers responded to six items from the Child Behavior Questionnaire Short Form (CBQ-SF; Rothbart et al., 2001; alpha = .85; sample item is “Usually rushes into an activity without thinking about it”). We used the Flanker task of the NIH Toolbox (Zelazo et al., 2013) to assess children’s inhibitory control.

After accounting for children’s age, the mean number of times a child initiated movement across days showed a significant association with teacher-reported impulsivity (β = .31, p = .004) but not with children’s inhibitory control on the Flanker task (β = .16, p = .231). Planned analyses will examine validity against other parent and teacher reports of behavior and other assessments of the NIH toolbox cognitive battery. The accelerometers also showed promise for detecting within-child differences in behavior. For each study day (N = 238), we calculated the number of times children initiated movement during center-based play. Intra-class correlations from a multi-level model (Raudenbush & Bryk, 2002) showed that 30% of variance in behavior was accounted for at the child level (ICC = .30). Planned analyses will examine variability in children’s behavior across other early childhood classroom activities (i.e., large group, small group, cleanup, snack, and transitions).

Through these preliminary and planned analyses, this study will examine the suitability of wearable accelerometers to study behavioral regulation in the classroom and whether and how children’s behavioral regulation varies across classroom activities and days.

Authors

• Andrew E. Koepp, University of Texas at Austin
  Presenting Author

• Elizabeth T. Gershoff, University of Texas at Austin