입학사정관제에서 의사결정의 정확률 및 성공비율 추정을 위한 로지스틱 회귀분석의 적용

Applying Logistic Regression for Estimating Decision Accuracy and Success Rate for the Admission Officer System in Korea

이 연구에서는 첫째 입학사정관제가 도입됨에 따라 입학전형요소의 잠정적 합격점수를 입학사정관이 변화시켰을 때 대학입학 후 지원자가 대학에서 성공할 확률이 어느 정도이며, 입학사정관의 의사결정 정확률이 어느 정도인지를 산출하는 방법을 상술하였다. 둘째, 모 국립대학의 2005학년도부터 2007학년도까지의 대학수학능력시험의 수능환산점수와 대학 성적평균 자료를 이용하여 로지스틱 회귀분석으로 통계적 모형을 만든 후, 의사결정의 정확률, 증분된 정확률, 성공비율을 실증적으로 추정하고, 실제 적용상 이슈들에 대해 논의하였다.

The educational policy for introducing Admission Officer System in college level were publicized and many universities in Korea are planning for allowing admission officers to select applicants for universities. The purposes of this study were (1) to review statistical methods for estimating criterion related validity and elaborate on the procedures for estimating decision accuracy, incremental accuracy rate, and success rate as an admission officer changes the cutoff scores on college admission test scores, and (2) to provide decision accuracy, incremental accuracy rate, and success rates obtained from the real data of a national university in Korea. This study, particularly, used on an College Scholastic Aptitude Test (CSAT) as an independent variable because most of universities in Korea use it for admission process. The college GPAs of freshman, sophomore, and junior students were dependent variables and were dichotomized into 1 or 0 for logistic regression analyses. The results were provided and statistical and several practical issues including no equating in CSAT, extrapolation in logistic regression, sample size, and validity of the criterion variable were discussed.