MEASURING THE LIMITS: WHEN DOES LONGITUDINAL TEACHER ATTRITION BECOME PROBLEMATIC?

Abstract

Researchers have long examined the magnitude and causes of teacher turnover, emphasizing its variability across settings, and observed potential effects on schools and students (Gershenson et al., 2018; Goldhaber & Theobald, 2021; Holme et al., 2017; Ingersoll, 2001). While teacher attrition is often framed as excessively high and problematic (Dias-Lacy & Guirguis, 2017; Neason, 2014; Seidel, 2014) and past scholars have documented harm to students based on exposure to attrition (Henry & Redding, 2020; Ronfeldt et al., 2013), attrition is not fully avoidable. As scholars have begun to attend to the longitudinal nature of attrition within schools including episodic (temporary high attrition), chronic (sustained high attrition across multiple years), and cumulative attrition (accumulation of loss over time), it is often assumed that elevated patterns over the long term will necessarily be even more harmful to students. Yet, existing research offers little insight into when turnover shifts from typical and manageable to atypical and harmful. This study builds on the existing literature (Holmes et al., 2019) by conducting descriptive and inferential analyses to identify policy-relevant measures for episodic, chronic, and cumulative attrition. Using 12 years of statewide data from Maryland public schools, this dissertation confirms that schools experience a wide range of attrition patterns over time, with the average school experiencing a 1-year episodic attrition rate of 14.5%, a chronic 3-year average attrition rate of just under 15%, and a cumulative 3-year attrition rate of 37%. While some schools do experience 1-year attrition rates above 60% and 3-year average attrition rates above 50%, these schools represent outliers, and only a small subset of schools lose more than 30% repeatedly. Further, while chronic and cumulative attrition have been theorized as distinct (Holme et al., 2017), this study demonstrates that schools with above-average rates of 3- to 5-year average attrition also tend to be schools experiencing above-average rates of 3- to 5-year cumulative attrition. This finding suggests these patterns reinforce each other rather than occur independently and challenges the utility of analyzing them separately for policy and practice. Moreover, nonlinear regression analyses reveal minimal association between 1- to 3-year average attrition and student achievement when attrition is below 30%, but there are significant declines in ELA and math performance when rates exceed 30%. Thus, while high chronic attrition rates (3-year average attrition rates) are highly uncommon, they are associated with severe consequences for students and may be driving the effects previously observed in past studies using linear models robust. These patterns are robust to alternative model specifications, including differing functional forms and the inclusion of school year fixed effects. This study refined the measurement of chronic attrition and supports the use of nonlinear models to examine their effects. It also offers key policy insights by highlighting both the potential severity and relative rarity of the most harmful attrition environments as findings support shifting policy focus from generalized turnover concerns to monitoring persistent, high-level attrition driven by deeper organizational challenges.