Bullying Research: Variance Estimates and Covariate Analyses to Improve Efficiency

• In Event: 34.087 - Evidence for Practices to Improve School Climate and Safety

Abstract

Estimating design parameter values for bullying and victimization provides more precise estimates of group variances and therefore better power analyses for experimental designs. This study leverages school climate survey data to estimate design parameters on outcome variables including continuous measures of student-reported instances of bullying and victimization during the past thirty days. The estimation is conducted by grade for two-level HLMs of students nested in schools for unconditional and conditional models. Our results show that unconditional ICC values for bullying are between 0.0030 and 0.0088 while unconditional ICCs for victimization range from 0.0033 to 0.0078. The values tend to decrease from 6 to 12 grade. Use of covariates and more precise measures are effective methods to improve efficiency.

Authors

• Carl Westine, University of North Carolina - Charlotte

• Tuba Gezer, University of North Carolina - Charlotte