Predictors of Assumed “Bad” Behavior: A Classification and Regression Tree Analysis

ABSTRACT Bias influencing teachers’ classroom management is increasingly clear, but the circumstances that influence the likelihood of relying on those biases are less understood. This study employed Classification and Regression Tree (CART) analysis, resulting in four models examining how teachers’ appraisals of first‐grade students’ externalizing problem behaviors are influenced by students’ demographics and teachers’ assessments of objective and subjective socioemotional skills. Results from a sample of 3305 teachers in the United States indicated that gender and racial biases are more likely to be relied upon when teachers are assessing subjective skills rather than objective ones. The potential inaccuracies in teachers’ subjective assessments of socioemotional skills and behaviors highlight the need for enhanced preservice teacher training and increased school support for socioemotional learning. These measures are critical to improving the accuracy of teacher evaluations and minimizing the influence of bias.