Observations and Exits: The Relationship Between Observational Measures of Teacher Effectiveness and Teacher Retention

• In Event: Exploring the Implementation and Effects of Various Levers in Multiple-Measure Teacher Evaluation

Abstract

Recent research has shown that teachers with lower scores on value-added measures (VAMs) experience lower rates of retention than do higher performing teachers (Goldhaber & Hansen, 2010). However, one challenge to the usefulness of VAMs as an evaluation and workforce management tool is the widespread skepticism among many educators, principals, and policymakers with respect to their reliability and specificity. This suggests that if a measure of human capital consistent with VAMs and more trusted by practitioners and policymakers were identified, it could prove a useful and effective addition to the teacher evaluation process.

In this paper we examine the Los Angeles Unified School District’s (LAUSDs) new teacher evaluation system, the Teacher Growth and Development Cycle (TGDC). The TGDC is based primarily on principals’ observations of teachers’ practice, scored according to a Danielson-based rubric, the Teaching and Learning Framework (TLF). Teachers receive a binary final rating (meets or below standard), as well as 4-point scores on 15 separate elements of the TLF. In this study, we assess the extent to which the TGDC, is used and trusted by principals in their evaluation of teachers. We further examine the consistency of TGDC ratings with other measures of human capital (including VAMs), and whether TGDC ratings predict teacher retention both within their schools and within LAUSD the following year. In addition, we examine the extent to which talent management practices might differ in LAUSD based on which measure of human capital were used to identify teacher quality.

The use of TGDC and its relationship to teacher retention has not been examined previously and fills a more general void in the literature on the impact of principal observation-based evaluation systems on subsequent teacher retention. We combine longitudinal administrative teacher-level data with teachers’ binary TGDC rating and average overall score (from the focus elements, and upon which the final rating is primarily based), and principal survey data from (response rate = 63%). These data are taken from the 2013-14 and 2014-15 school years, when TGDC was implemented district-wide. In addition, we use two years of qualitative data from school case studies and system-level interviews to help provide color and nuance to our quantitative results.

Using multinomial logistic regression predicting teacher retention within the same school and within LAUSD the year following TGDC evaluation, and multiple teacher quality metrics, this paper is the first of its kind to demonstrate that in Los Angeles, teachers who receive lower TGDC evaluations are more likely to leave their school and the district the following year relative to higher scoring teachers. We also find that TGDC ratings and VAM scores appear to capture similar variation in teacher quality. Given this, TGDC-style evaluations -- rigorously designed frameworks populated through principal observations -- may prove to be a useful supplement to VAMs for monitoring and improving quality in the teacher labor market, particularly in environments in which VAM scores may be viewed as either too controversial or too blunt to be valuable as labor management tools.

Authors

• Katharine Omenn Strunk, University of Southern California

• Julie A. Marsh, University of Southern California

• Edward Cremata, University of Southern California