Application of the Weighting Method in Incomplete Data for Factor Analysis：the Example of Student Data of the Second Wave Survey in TEPS

• Main Authors: YAO,WEI-CHE, 姚偉哲
• Other Authors: WANG,HONG-LONG
• Format: Others
• Language: zh-TW
• Published: 2009
• Online Access: http://ndltd.ncl.edu.tw/handle/73962425382665875051

碩士 === 國立臺北大學 === 統計學系 === 97 === Our study is to apply weighting methods in dealing with the problem in incomplete data analysis under factor analysis, and compare with other methods for missing data. The data we used were got from the second wave Taiwan Education Panel Survey (TPES), and used structure of original missing data to make missing data of various missing proportion randomly to compare with the effects of different weighting methods among different missing proportions. Be aimed at weighting method, this study discussed numerous ways to weigh data which comprised weighting by demographic data, mental health data, and the feature of missing data under demographic data. To comparison with weighting method for each missing proportion, we observed presentation of number of factors, eigenvalue, main factor variables, and factor loading. Based on weighting method, we compared with other common method of missing data procedures as well, and weighed mental health questions in questionnaire, and then proceed with factor analysis.