Exploration of Predictors for Korean Teacher Job Satisfaction via a Machine Learning Technique, Group Mnet

Abstract

Despite the high academic achievements of Korean students in international comparison studies, their teachers’ job satisfaction remains below the Organization for Economic Co-operation and Development (OECD) average. As job satisfaction is one of the major factors affecting student achievement as well as student and teacher retention, the identification of the most important satisfaction predictors is crucial. The current study analyzed data from the OECD 2013 Teaching and Learning International Survey (TALIS) via machine learning. In particular, group Mnet (a penalized regression method) was employed in order to consider hundreds of TALIS predictors in one statistical model. Specifically, this study repeated 100 times of variable selection after random data-splitting as well as cross-validation, and presented predictors selected 50% of the time or more. As a result, 18 predictors were identified out of 558, including variables relating to collaborative school climates and teacher self-efficacy, which was consistent with previous research. Newly found variables to teacher job satisfaction included items about teacher feedback, participatory school climates, and perceived barriers to professional development. Suggestions and future research topics are discussed.