A Simulation Study of Missing Data Imputation by Using the Bootstrap Method
碩士 === 靜宜大學 === 財務與計算數學系 === 102 === Missing data are a common occurrence in almost all research. A direct approach to missing data is to simply omit those cases with missing data and to run the analyses on what remains. However, the deletion of missing data could result in a substantial decrease in...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | zh-TW |
| Published: |
2014
|
| Online Access: | http://ndltd.ncl.edu.tw/handle/37300099135920528743 |
| Summary: | 碩士 === 靜宜大學 === 財務與計算數學系 === 102 === Missing data are a common occurrence in almost all research. A direct approach to missing data is to simply omit those cases with missing data and to run the analyses on what remains. However, the deletion of missing data could result in a substantial decrease in the sample size and cause the loss of information. Rather than removing observations with missing data, an alternative approach is to impute missing values. Efron (1994) proposed nonparametric bootstrap methods to estimate the parameter of interest in a missing data situation. In this study, we applied the nonparametric bootstrap methods proposed by Efron to do the missing data imputation. We conducted simulation studies to explore the use of the proposed bootstrap method. In addition, we applied this approach to impute missing values in a real data. The results are reported and discussed.
|
|---|