目標：簡述測量等值性(Measurement Equivalence/Invariance; ME/I)的本質、影響及檢測程序，檢視病安態度研究對該議題重視的程度並討論ME/I某些疑慮。方法：搜尋PubMed電子
資料庫，2000年1月至2010年4月英文期刊論文，檢索條件以“Patient Safety ttitude”or“Safety Culture”or“Safety Climate”in Keyword，同時以“Health Care”or“Hospital”in Title/Abstrat，以及“Safety Instrument”in Text，針對問卷調查為主的量性研究分析工具心理計量特徵，分析層次以及如何利用測量作比較。結果：41篇文獻內容分析顯示，效度衡鑑統計技術多以信度估計、相關分析、迴歸分析、因素分析、多特質多方法，以T-test, ANOVA, MANOVA, Chi-square test比較跨群差異，少有以ME/I檢視病安氣候因素結構在跨樣本或跨情境等值程度，僅報導適配度評定假設模型與觀測資料的共變異矩陣之接近程度。本文提出ME/I實作、未能通過ME/I檢定的原因及補救方法。結論：ME/I攸關著病安氣候構念的概化以及預測氣候是否改變，其重要性不容忽視，期以本文與國內學者共同關注這些議題的後續發展。(台灣衛誌 2011；30(4)：388-407)

Objectives: The aims of this study were to articulate what measurement equivalence/ invariance (ME/I) is, determine its impact, describe the procedural techniques in conducting ME/I, review the health-care literature on the climate and culture of patient safety, and discuss questions about ME/I testing. Methods: We looked at English language journal articles published between Jan 2000 and April 2010 by searching PubMed (free Medline) electronic databases using the keywords “patient safety attitude” or “safety climate” or “safety culture”, simultaneously with the terms “health care” or “hospital” in title/abstracts, and “safety instruments” in text. Our focus was on quantitative empirical studies adopting questionnaire-survey methods. A total of 41 articles
were reviewed to analyze the psychometric properties of instruments, the level of analysis and the use of this measure for comparisons. Results: We found that most studies demonstrated reliability and validity of the constructs using the statistical methods of reliability estimation, correlation, regression, exploratory factor analysis, confirmatory factor analysis, and multitrait-multimethod matrices, and tested group differences by t-tests, ANOVA or MANOVA. Evaluations of invariant factor structure across populations or contexts by applying ME/I testing were seldom addressed. The majority of these studies did not address the ME/I issue. We suggest that some analytical forms of ME/I testing, the triggers for not supporting ME/I, and the remedies for patient safety require further research. Conclusions: There are some questions about the sensitivity of ME/I analytical procedures and the susceptibility of these procedures to contextual influences. (Taiwan J Public Health. 2011;30(4):388-407)