Files
Abstract
The purpose of this study was to model student growth percentiles in a Georgia school district in order to examine the variability of scores within classrooms and the impact of using this growth model as a measure of teacher effectiveness. This dissertation applied student growth percentiles based on Damian Betebenners Colorado growth model as a precursor to Georgias implementation. This study considered the variability among student growth percentiles at the elementary classroom level given teacher evaluation and compensation will soon be based on these outcomes due to the federal governments recent proposal to reauthorize the Elementary and Secondary Education Act along with the Race to the Top grant program. Both initiatives incorporate growth procedures as measures of teacher performance impacting evaluation and pay. With the 2014 deadline for implementation of Race to the Top, this study applied existing growth model research to student test scores longitudinally. Students were grouped with their academic peers based on their 2009 Grade 3 Criterion Referenced Competency Test scores. Using 2010 Grade 4 scores for the same students, student growth percentiles were assigned. Data were then disaggregated to the class level and variability within individual teachers classes was examined. This study modeled the application of Georgias new growth model on a small scale in order to consider the use of these scores in evaluating teachers across the state. The outcome of this study demonstrated large variability within classrooms. These results make using student growth percentiles to measure teacher effectiveness problematic due to the large dispersion of scores for individual teachers. The findings from this study support existing research that suggests value-added models should not be used as a singular measure of teacher effectiveness. The results of this study are applicable for stakeholders in Georgia education as state and federal policy move toward basing teacher evaluation and compensation on student growth percentiles.