원문정보
초록
영어
The analytical focus of this study centers on how to treat the missing values in public administration research. Examining the mechanism of missing values in data gathering, the paper constructs a regression model for ordinal categorical data of the survey by the KOSSDA numbered A1-205-0040. Applying four solutions for missing values such as complete case analysis method, EM technique, regression methodology, and multiple imputation, this paper tries to approach the focus of the research. Results confirm that the complete case analysis method can not be an appropriate solution for missing values in reality. The higher the percentage of missing values in data set, the riskier the application of this solution. On the contrary, the other three solutions turn out to be the appropriate remedies for missing values in general. Especially, the usefulness of EM technique and regression methodology can be highlighted. In short, future research should accommodate the necessity/accessibility of scientific techniques as well as investigate the mechanism of missing values in a given data set.
목차
I. 서론
II. 결측치의 발생기제와 유형
1. 완전무작위 결측치(MCAR)
2. 무작위적 결측치(MAR)
3. 무시할 수 없는 결측(NI)
III. 범주형 자료의 결측치 보완기법
1. 완전사례분석기법(Complete Case Anaysis Method)
2. EM기법
3. 회귀분석기법(Regression Method)
4. 다중대체기법(Multiple Imputation Method)
IV. 보완 방법들의 실제 적용
1. 분석모형의 구성
2. 실증분석의 결과 해석
V. 요약과 함의
참고문헌