「臺灣簡明版世界衛生組織生活品質問卷」之信度概化研究

Reliability Generalization Study on the Taiwan Version of the WHOQOL-BREF

與「簡明版世界衛生組織生活品質問卷」（WHOQOL-BREF）相同，「臺灣簡明版世界衛生組織生活品質問卷」已廣泛應用至國內各研究領域。雖分數信度為效度係數、效果量等統計量的基礎，然文獻上缺乏該量表之分數信度的研究。信度概化法（reliability generalization, RG）採用後設分析方法，以探討某一測量工具之跨研究分數信度的心理計量特徵，及其分數信度與研究特徵的關係，為檢驗分數信度的重要工具。因此，本研究採用RG以探討「臺灣版WHOQOL-BREF」之四範疇分數信度的心理計量表現。本研究從「臺灣博碩士論文知識加值系統」與「華藝線上圖書館」兩個資料庫，以「WHOQOL」或「世界衛生組織生活品質」作為關鍵字蒐集2000年至2015年8月使用「臺灣版WHOQOL-BREF」的博碩士論文，刪除不符合研究目的之論文後，共納入248篇。除分析其分數信度的描述統計，並採用多重方法如迴歸分析、相關分析、t考驗、變異數分析，以檢驗分數信度與研究特徵的關係。研究結果顯示，四範疇之跨研究的平均分數信度高於.70，且其標準差低於.09，顯示此四範疇的分數信度具跨研究穩定性。四範疇之量表分數標準差可顯著預測該範疇之分數信度，且部分範疇之分數信度受教育程度、宗教信仰、樣本年齡的影響。建議未來在評估是否使用「臺灣版WHOQOL-BREF」時，需考量這些研究變項，尤其是在生理健康與環境範疇上。

As the short version of the World Health Organization Quality of Life questionnaire (WHOQOL-BREF), Taiwan version of WHOQOL-BREF is widely used in various research fields. Although the score reliability is the foundation of statistics, such as validity coefficient and effect size, the study about score reliability of WHOQOL-BREF Taiwan version is rare. By using meta-analysis method, reliability generalization (RG) is a powerful approach for examining the psychometric properties of score reliability and assessing the study characteristics associated with score reliability. The present study therefore utilized RG to examine psychometric performance of the score reliability from four domains of the Taiwan version of the WHOQOL-BREF. This study used the keywords, WHOQOL and World Health Organization Quality of Life, to collect doctoral and master dissertations (from 1998 to the May 2019) which used the WHOQOL-BREF Taiwan version from two databases: National Digital Library of Theses and Dissertations in Taiwan and Airiti Library. After deleted dissertations which were incompatible with the purpose of this study, 248 dissertations were included. In addition to analyzing descriptive statistics of score reliability from these dissertations, variety of statistic methods (such as regression analysis, correlation analysis, t-test, and analysis of variance) were conducted to examine the relationship between score reliability and study characteristics. The findings showed that the mean score reliability of four domains of WHOQOL-BREF Taiwan version were higher than .70 and their standard deviations were lower than .09, indicating that WHOQOL-BREF Taiwan revealed consistency across studies. Moreover, the standard deviations of scores from these four domains could significantly predict their score reliability, as well as parts of their score reliability were influenced by study characteristics such as education level, religious beliefs, and age of study participants. We suggested that when assessing whether to use WHOQOL-BREF Taiwan version or not, researchers need to consider the influence of these study characteristics, especially for physical health and environmental domains.