What We Should Know About Principal Turnover: The Need for a Machine Learning Approach

• In Event: Principal Leadership: Influence, Success, STEM, and International Perspectives

Abstract

Principal turnover has become a vital concern globally. Previous research has often identified drivers of principal turnover through linear regression frameworks, and nonlinear patterns may have gone unnoticed. Thus, we need a good method to study principal turnover. Classification and Regression Trees (CART) remains an underutilized tool that enables researchers to model nonlinear patterns and provide understandable visualizations of results. Using a machine learning approach, the study, for the first time, aims to explore the drivers of the two types of principal turnover, leaving the principalship and changing schools in the United States (N = 129) and Singapore (N = 151) from an international dataset (TALIS 2018). Implications and cross-cultural comparison of results are discussed.

Authors

• Jasmine Shi, Michigan State University
  Presenting Author

• Alex J. Bowers, Teachers College, Columbia University
  Non-Presenting Author