Validity of Scales
Validity of Scales
复制标题
量表的效度
DOI：
--
复制
发表时间：
2011
期刊：
影响因子：
--
通讯作者：
E. Svensson
中科院分区：
文献类型：
--
作者：
E. Svensson
评价量表和多量表问卷测量质量的概念是效度和信度。定量数据（区间和比率数据）的相应概念是准确度和精密度。如果评定量表测量了具体研究中预期测量的内容，则该量表有效。主观现象自我评价的有效性是相对的，不能绝对地评价。一个量表的效度是研究特有的，每次选择该量表或问卷进行新的研究时，都必须考虑它的效度。因此，有各种各样的有效性概念，每一个都针对一种特殊类型的质量评估。主要概念是标准、结构和内容效度，但使用了大量的子概念。这些概念的含义并不明确，取决于应用和研究范式。标准效度是指量表与真实状态或金标准的一致性，根据研究目的，将使用临床、预测和同期效度等子概念。结构效度是指在缺乏真实状态或金标准的情况下，具有相同理论定义的量表之间的一致性。研究中使用了聚合、描述、判别、发散、析因、翻译效度和平行信度等子概念。生物学效度是指在特定人群中与其他测量方法进行比较时，量表评估与假设期望的接近程度。区分等级量表用于区分个体或群体，当没有外部标准时，则要评估区分效度。平行信度是指量表的可重复性。概念内容效度是指量表或多量表问卷在覆盖重要领域方面的完整性。使用了面子、生态、决策、共识、抽样有效性、全面性和可行性等子概念。对评级量表的评估生成的有序数据仅具有等级不变的特性，这意味着响应指示等级顺序而不是数学值。当重新标记有序响应时，数据的统计处理结果不得改变。评价效标效度和结构效度的合适的统计方法是指比较量表之间的顺序一致性或关系。在视觉模拟量表（VAS）和语言描述量表（VDS-VAS）上对感觉到的背痛进行双盲配对评估的散点图，具有五个排序的分类反应，见图10。从图中可以明显看出，评估之间存在很大的重叠。e配对观测值（X，Y）中的不一致概率，
e concepts of quality of measurements made by rating scales and multi-scale questionnaires are validity and reliability. Corresponding concepts for quantitative data (interval and ratio data) are accuracy and precision. A rating scale is valid if it measures what it is intended to measure in the specic study. e validity of self-estimated subjective phenomena is relative and cannot be assessed absolutely. e validity of a scale is study specic, andmust be considered each time the scale or the 7questionnaire is chosen for a new study. erefore there are various concepts of validity, each addressing a specic type of quality assessment. emain concepts are criterion, construct, and content validity, but a large number of sub concepts are used. e meaning of these concepts is not univocal and depends on applications and research paradigms.Criterion validity refers to the conformity of a scale to a true state or a gold standard, and depending on the purpose of the study sub concepts like clinical, predictive and concurrent validity will be used. Construct validity refers to the consistency between scales having the same theoretical denition in the absence of a true state or a gold standard. Sub concepts like convergent, descriptive, discriminant, divergent, factorial, translation validity and parallel reliability have been used in studies. Biologic validity refers to the closeness of scale assessments to the hypothesized expectation when comparing with other measures in a specic population. Discriminative rating scales are used to distinguish between individuals or groups, when no external criterion is available, then discriminant validity is to be assessed. Parallel reliability refers to the interchangeability of scales. e concept content validity refers to the completeness of the scale or multi-scale questionnaire in the coverage of important areas. Sub concepts like face, ecological, decision, consensual, sampling validity, comprehensiveness and feasibility have been used. Assessments on rating scales generate ordinal data having rank-invariant properties only, which means that the responses indicate a rank order and not a mathematical value. e results of statistical treatments of data must not being changed when relabeling the ordered responses. Appropriate statistical methods for evaluation of criterion and construct validity oen refer to the order consistency or to the relationship between the scales of comparison. e scatter plot of paired assessments of perceived back pain on a visual analogue scale (VAS) and on a verbal descriptive scale (VDS-) having ve ordered categorical responses is shown the Fig. . As evident from the plot there is a large overlapping between the assessments. e probability of discordance in paired observations (X,Y),