The Effects of the Partnership Model on Teacher Turnover in Michigan’s Lowest Performing Schools

• In Event: The Intersection of K-12 School Accountability and Human Capital

Abstract

The Every Student Succeeds Act (ESSA) requires states to improve their lowest-performing schools (i.e., Comprehensive Support and Improvement, CSI), by providing supports and resources that aim to rapidly increase achievement. However, CSI designations frequently have unintended side effects, such as increased teacher turnover, which can disrupt the intended improvement efforts. That said, recent work suggests that some turnover may actually be beneficial for teacher working conditions (e.g., collaboration). Michigan’s program to support low-performing schools, called the Partnership Model, has identified three separate cohorts, beginning in the spring of 2017. The most recent cohort, Cohort 3, was identified using test score data from the 2021-22 school year and planned their improvement strategies during the 2022-23 school year. Full implementation then began during the 2023-24 school year. We aim to understand how this model of support may impact teacher retention as prior research suggests turnaround may increase staff turnover. Therefore, we ask the following research question: What is the effect of the Michigan Partnership Model on teacher turnover?

We use Michigan state administrative records for the 2023-2024 school year and the list of CSI schools identified in the 2021-2022 school year. We restrict our sample to include only teachers who worked in a school that received an Michigan State Index Score (as required by ESSA) based on the 2021-22 testing cycle. Further, as we are interested in the effects of the Partnership Model, we dropped those teachers in schools that were designated as CSI but are not a member of Cohort 3. This creates an analytic sample of 59,340 teachers across 2,498 schools.

We estimate the effects of Partnership on teacher turnover through a sharp regression discontinuity design through the following form: where represents the outcome for teacher i in school s for each respective turnover pathway. Specifically, we operationalize turnover in three mutually-exclusive ways: 1) within-district transfer, 2) between-district transfer, and 3) leave Michigan education. We also create a fourth turnover outcome that represents a teacher leaving their school for any reason, which is all inclusive of the three previous turnover outcomes. is a running variable of the centered CSIIndex Score, as calculated and provided by the Michigan Department of Education – centered at 23.88. is an indicator of treatment, where schools with a centered CSI Index Score equal to or below 0 are considered treated and those above 0 are considered not treated. and are vectors of teacher-level and school-level covariates, respectively, and is an idiosyncratic error term clustered at the school level.

Preliminary findings suggest that, on average, Partnership increases the likelihood of within-district transfer by about 7 percentage points. Further, we do not find any evidence of Partnership’s effects on between-district transfer or on leaving Michigan education. Before the APPAM conference, we intend to unpack this finding with interview and survey data of teachers and principals across the treated schools. Additionally, we intend to run a fuzzy regression discontinuity model, as we find small amounts of treatment non-compliance.

Authors

• Alexander Moran, University of Maryland
  Presenting Author

• Erica Harbatkin, Florida State University
  Non-Presenting Co-Author