인과적 타당성을 위한 도구변수 분석법: 교육 연구에의 활용 가능성 탐색

Abstract

Instrumental variable (IV) analysis is a statistical approach that enables valid estimation of causal effects, even in non-experimental settings where randomization is not possible. Despite its potential utility, IV analysis remains underutilized in Korean education research. The purpose of this paper is to introduce the fundamental principles of IV analysis and its applications. Specifically, the paper employs causal graphs to provide a more intuitive explanation of the key assumptions and mechanism underlying IV analysis. In addition to providing theoretical explanations, the paper explores the applicability of IV analysis in the context of education research through examining randomized encouragement designs, perceived construct studies, and treatment noncompliance. By doing so, this paper aims to facilitate a better understanding of theoretical foundations and practical considerations when applying IV analysis in education research.