模糊語意量表與傳統李克特式量表之信度模擬比較分析
| Autor: | 顏乾明 |
|---|---|
| Rok vydání: | 2003 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 91 In our daily life, we often get to access various kinds of questionnaire or surveys, and these surveys were commonly designed and used the traditional system of Likert scale, which records the data using binary logic, but there was constantly hesitation exist in our human’s thinking logic or attitude. If we just rely on the traditional binary logic result, it is often unproper. Because of the fuzzy principle describing human’s fuzzy logic, there were certain characteristics and advantages, so according to the extended fuzzy theory, which was the fuzzy linguistic scale. In this research, the fuzzy linguistic variable mainly uses triangular fuzzy number, focusing on different scale, including the number of category, number of questions, number of interviewee and the specialties under the interviewee, using MATLAB language testing result combinations programnize, then break up to compare the result of reliability in fuzzy linguistic scale and the Likert scale. As the result of this research, in any group, the data demonstrated that using fuzzy linguistic scale rather than Likert scale was much more reliable. Under the same type of calculation method, the reliability is much higher when interviewee’s latent trait is uniformly distributed. When there were more items of questionnaire, the reliability of fuzzy linguistic scale increase; the number of interviewee is not the factor concern that affects the reliability in fuzzy linguistic scale. The results of this research could prove some implications and references for theoretical and empirical study. Finally, some recommendations and suggestions are discussed for future research. Keywords: triangular fuzzy number, number of category, reliability, Likert scale |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|