Exploring Predictors of Highly Effective Teachers Through Multilevel Model

Abstract

This paper explored significant teacher and school level predictors of the status of highly effective teachers based on Florida teacher evaluation system employing a two-level multilevel analysis approach known as hierarchical generalized linear model (HGLM). The analysis used 3,895 teachers who taught students in Kindergarten to Grade 12 and 210 schools from one of the largest urban school districts in United States of America. In this study, teacher and school level data are used at level-1 and level-2 models, respectively. At teacher level, the results showed significant effects of percentage of student level demographic, academic, and disciplinary variables aggregated at teacher level as well as teacher’s experience and educational degree level. At school level, percentages of Black and percentage of Hispanic students showed significant effects on the status of highly effective teachers. The percentage of variance explained at school level is found 19.5% with an effect size of 0.44 which is determined as a “medium” size of school effect.