Development of Software to Record Behavioral Issues and Prediction Using Bayesian Inference

Abstract

School-wide positive behavioral interventions and supports (SWPBIS) is an evidence-based approach aimed at minimizing behavioral issues among all students registered at a school. Although SWPBIS is increasingly being practiced in Japan, data collection systems that can be used in Japanese are expected to be developed along with support systems to promote transformations in teacher support behaviors based on data. In this study, we prototyped a software capable of efficiently recording office discipline referral (ODR) data for use in SWPBIS, and then adopted the novice expert ratio method (NEM) to evaluate its usability to reveal points of improvement in the design. We also examined statistical modeling using a Poisson distribution along with a derivation of predictive distribution for the number of behavioral issue occurrences using Bayesian inference as methods for teachers to transform support behaviors based on ODR data.