Polit Denise F, Beck Cheryl Tatano
Humanalysis, Inc., Saratoga Springs, NY, USA.
Res Nurs Health. 2006 Oct;29(5):489-97. doi: 10.1002/nur.20147.
Scale developers often provide evidence of content validity by computing a content validity index (CVI), using ratings of item relevance by content experts. We analyzed how nurse researchers have defined and calculated the CVI, and found considerable consistency for item-level CVIs (I-CVIs). However, there are two alternative, but unacknowledged, methods of computing the scale-level index (S-CVI). One method requires universal agreement among experts, but a less conservative method averages the item-level CVIs. Using backward inference with a purposive sample of scale development studies, we found that both methods are being used by nurse researchers, although it was not always possible to infer the calculation method. The two approaches can lead to different values, making it risky to draw conclusions about content validity. Scale developers should indicate which method was used to provide readers with interpretable content validity information.
量表开发者通常通过计算内容效度指数（CVI）来提供内容效度的证据，该指数基于内容专家对项目相关性的评分。我们分析了护士研究人员如何定义和计算CVI，发现项目层面的CVI（I-CVI）具有相当高的一致性。然而，有两种计算量表层面指数（S-CVI）的替代方法，但未得到认可。一种方法要求专家们达成普遍共识，但一种不太保守的方法是对项目层面的CVI进行平均。通过对量表开发研究的目的样本进行反向推断，我们发现护士研究人员同时使用了这两种方法，尽管并非总是能够推断出计算方法。这两种方法可能会得出不同的值，从而在得出关于内容效度的结论时存在风险。量表开发者应指明使用了哪种方法，以便为读者提供可解释的内容效度信息。
