김용남 교수

When randomized experiments are infeasible, quasi-experimental designs can be exploited to evaluate causal treatment effects. The strongest quasi-experimental designs for causal inference are regression discontinuity designs, instrumental variable designs, matching and propensity score designs, and comparative interrupted time series designs. This article introduces for each design the basic rationale, discusses the assumptions required for identifying a causal effect, outlines methods for estim

For misguided reasons, social scientists have long been reluctant to use gain scores for estimating causal effects. This article develops graphical models and graph-based arguments to show that gain score methods are a viable strategy for identifying causal treatment effects in observational studies. The proposed graphical models reveal that gain score methods rely on a bias-removing mechanism that is quite different to regular matching or covariance adjustment. While gain score methods offset n

완전매개는 주어진 매개변수를 거치지 않고서는 원인변수가 결과변수에 직접적으로 영향을 미칠 수 없는 특수한 구조적 상황을 가리킨다. 본 연구에서 지난 이십년 간 국내 교육학 분야 학술지에 출판된 연구물을 검토한 결과, 전체 901건의 매개분석 적용 논문 중 약 20%에서 완전매개를 보고하고 있었다. 원인-결과 간의 작동기제를 완벽하게 설명했다는 완전매개의 실질적 의미를 감안할 때, 이러한 비율의 타당성에 관한 합리적인 의문이 있다. 본 연구에서는 국내 매개분석 적용 연구에서 확인되는 통제변수의 소극적 활용이 거짓 완전매개 보고로 이어질 수 있다는 이론적 설명을 시도한다. 특히 본 연구는 인과그래프 혹은 인과구조모형 접근을 적용하여, 매개분석에서 교란효과의 통제실패가 초래하는 문제점에 대해 보다 직관적인 설명을 제공한다. 구체적으로 교란효과로 인한 편향의 증폭과 상쇄 현상이 소개된다. 이러한 주장에 대한 시뮬레이션 증거를 제공하고, 통제변수를 고려하지 않은 완전매개 결과의 타당성에 대해

Despite the long-standing discussion on fixed effects (FE) and random effects (RE) models, how and under what conditions both methods can eliminate unmeasured confounding bias has not yet been widely understood in practice. Using a simple pretest-posttest design in a linear setting, this paper translates the conventional algebraic formalization of FE and RE models into causal graphs and provides intuitively accessible graphical explanations about their data-generating and bias-removing processes

Suppression effects in multiple linear regression are one of the most elusive phenomena in the educational and psychological measurement literature. The question is, How can including a variable, which is completely unrelated to the criterion variable, in regression models significantly increase the predictive power of the regression models? In this article, we view suppression from a causal perspective and uncover the causal structure of suppressor variables. Using causal discovery algorithms,

The expression of leptin and OBR in breast cancer tissue may be not a prognostic factor for disease-free survival of breast cancer patients. In the future, further studies are needed to determine whether leptin expression could be a predictive factor for tamoxifen therapy in the postmenopausal obese subgroup among the early breast cancer patients.

The demand for a high-resolution metal-oxide-semiconductor (CMOS) image sensor has increased in recent years, and pixel size has shrunk below 1.0 μm to allow accumulation of numerous pixels in a limited area. However, shrinking the pixel size lowers the sensitivity and increases crosstalk because the aspect ratio is worsened by maintaining the height of the pixel. This work introduces a high-sensitivity pixel with a quad-WRGB (White, Red, Green, Blue) color filter array (CFA), spatial deep-trenc

Abstract Does reviewing previous answers during multiple‐choice exams help examinees increase their final score? This article formalizes the question using a rigorous causal framework, the potential outcomes framework. Viewing examinees’ reviewing status as a treatment and their final score as an outcome, the article first explains the challenges of identifying the causal effect of answer reviewing in regular exam‐taking settings. In addition to the incapability of randomizing the treatment sele

Despite the long-standing discussion on fixed effects (FE) and random effects (RE) models, how and under which conditions both methods can eliminate unmeasured confounding bias have not yet been widely understood in practice. Using a simple pretest-posttest design in a linear setting, this article translates the conventional algebraic formalization of FE and RE models into causal graphs and provides intuitively accessible graphical explanations about their data-generating and bias-removing proce

무선할당을 적용하기 어려운 교육연구에서 연구자들은 흔히 집단별로 사전점수와 사후점수를 측정하는 유사실험설계를 활용한다. 공분산분석(ANCOVA)과 차이점수 분석법(gain score analysis)은 이러한 사전-사후검사 자료를 분석하는 대표적인 두 가지 통계적 접근이다. 사전점수를 일종의 통제변수로 활용하는 공분산분석 접근이 ‘숨어 있는 교란변수가 존재하지 않는다’는 가정을 요구함에 반해, 차이점수 분석법은 교란변수가 존재함에도 불구하고 타당한 인과효과를 추정할 수 있다는 이론적 장점이 있다. 물론, 이러한 가능성은 차이점수 분석법에서 요구하는 인과적 가정의 타당성에 의존한다. 공통추세 가정은 ‘존재하는 교란변수가 사전점수와 사후점수에 미치는 영향력이 동일하다’고 해석할 수 있는 내용으로, 차이점수 분석결과에 대한 인과적 해석의 타당성을 결정하는 핵심가정이다. 반사실적(counterfactual) 진술에 의존하는 공통추세 가정은 본질적으로 그 경험적 검증이 불가능하기에, 연구자들

Whether examinees' answer changing behavior while taking multiple-choice exams is beneficial or harmful is a long-standing puzzle in the educational and psychological measurement literature. Formalizing the problem using the potential outcomes framework, this article shows that the traditional method of comparing the proportions of to and right to wrong answer changing patterns--a method that has recently been criticized by van der Linden, Jeon, and Ferrara (2011)--indeed correctly identify the