Matthew Goodlaw, Statistical Supervisor, Educator Quality Division, New Mexico Public Education Department
Olivia Lippincott, JMP Senior Associate Systems Engineer, SAS
Teacher effectiveness is quantified in New Mexico using multiple measures including, though not limited to, value-added scores and ratings from classroom observations of teachers' practices and behaviors. Essentially, the value-added estimate is the average of the differences between the students' actual and predicted test scores. The classroom observation ratings are generated by principals and other certified observers. In order to improve the internal certification process, we used JMP Pro to analyze the data. With ordinal scales, such as observation rubrics, there are different ways to analyze the data and account for variation. We examine and compare the traditional factor analysis model with correspondence analysis, and item response theory models. We also examine the variance in the classroom observations and explore patterns between the textual evidence collected by the observer and numerical score they assign using Text Explorer. Finally, we used the mixed models to explore the extent that the quantitative information extracted from the Text Explorer accounts for variance in the relationship between classroom observation scores and value-added scores.