Incorporating Multiple Measures into Teacher and Principal Evaluation Models

Three new reports from the Regional Education Laboratories (RELs) West and Midwest examine a new multiple-measure teacher evaluation system in Arizona and investigate whether student and teacher survey measures should be added to a principal evaluation model in the Midwest. Principal and teacher perceptions of implementation of multiple measure teacher evaluation systems in Arizona This study describes how multiple-measure teacher evaluations were put into practice in a set of ten volunteering local education agencies (LEAs) in Arizona. After a key shift in state policy, five “pilot” LEAs implemented the new Arizona Department of Education teacher evaluation model in the 2012/13 school year, while five other “partner” school districts developed their own local models aligned with the new state requirements.  Secondary analyses of survey and focus group data from the pilot and partner LEAs indicated that teachers and principals tended to more favorably view performance assessments (observations of teachers) that have traditionally comprised evaluations, and were more skeptical about incorporating results from student assessments and stakeholder surveys.  Study participants had mixed perceptions about the new evaluations’ initial outcomes, and raised concerns about the time burden involved, inter-rater reliability, and the need for ongoing training and support. Properties of the multiple measures in Arizona's teacher evaluation model This study explored the relationships among the components of the Arizona Department of Education’s new teacher evaluation model, with a particular focus on the extent to which ratings from the state model’s teacher observation instrument differentiated higher and lower performance. The study used teacher-level evaluation data collected by the Arizona Department of Education from five participating pilot LEAs during the 2012/13 school year.  The study relied primarily on descriptive statistics calculated from the results of the different component metrics piloted in these LEAs, as well as analysis of the correlations among these components. Results indicated that teachers’ observation item scores tended to concentrate at the Proficient level (the second-highest score on a four-point scale: Unsatisfactory, Basic, Proficient, and Distinguished), with this level accounting for 62 percent of all observational item scores. In addition, while the strength of the correlation between results from observations and the state’s student academic progress metric was generally low, the correlation varied significantly between high- and low-performing teachers, as well as between certain teacher subgroups. The utility of teacher and student surveys in principal evaluations: An empirical investigation This study examined whether adding student and teacher survey measures to existing principal evaluation measures increases the overall power of the principal evaluation model to explain variation in student achievement across schools. The study was conducted using data from 2011-12 on 39 elementary and secondary schools within a midsize urban school district in the Midwest.  The research team used the results of the district’s Tripod student and teacher surveys to construct six school-level measures of school conditions that prior research has shown to associate with effective school leadership. The study finds that adding the full set of six survey measures as a group results in statistically significant increases in variance explained in mathematics and composite value-added outcomes, but not in reading.  A stepwise regression procedure identified two measures – instructional leadership and classroom instructional environment – as an optimal subset of the six measures. This evidence indicates that student and teacher survey measures can have utility for principal performance evaluation.