https://scholars.lib.ntu.edu.tw/handle/123456789/40170
標題: | 點數對評定量表信度與效度影響之模擬研究(II) | 作者: | 翁儷禎 | 關鍵字: | 評定量尺;李克特式量尺;量尺點數;題目分配;效標關聯效度;相關係數;rating scales;Likert-type scales;number of response categories;item distributions;criterion-related validity;correlation coefficient | 公開日期: | 2002 | 出版社: | 臺北市:國立臺灣大學心理學系暨研究所 | 摘要: | 本研究以模擬研究方法探討量表特徵與受試者特性對評定量表效標關聯效度的影響。操弄變項包括量表與效標各題真分數的相關、量表題數、同量表內題目間的共變數、量表點數、題目分配、受試者判斷的正確度(即個別題目之信度)。結果發現,題間共變、量表題數與受試者判斷正確度對效標關聯效度的影響較大,量表點數與題目分配則影響不大。當量表與效標各題真分數的相關偏低時,量表設計難以提高效標關聯效度。 The research examined the effects of scale properties and respondent characteristics on criterion-related validity of Likert-type rating scales by Monte Carlo method. The independent variables manipulated include correlation between true score of items from the scale and the criterion, the number of response categories, the inter-item covariance, the number of items, respondents’ judgment accuracy (or reliability of individual item), and the distribution of scores of the rating scales. The results suggested that inter-item covariance, number of items, and judgment accuracy affected criterion-related validity. Number of response categories and score distributions had little effect on validity coefficient. When the correlation between true scores of test and criterion is low, it is unlikely to increase criterion-related validity by manipulations of scale design. |
URI: | http://ntur.lib.ntu.edu.tw//handle/246246/9698 | 其他識別: | 902413H002022 | Rights: | 國立臺灣大學心理學系暨研究所 |
顯示於: | 心理學系 |
檔案 | 描述 | 大小 | 格式 | |
---|---|---|---|---|
902413H002022.pdf | 99.8 kB | Adobe PDF | 檢視/開啟 |
在 IR 系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。