Health Services Management Department, Guizhou Medical University, Guiyang, China.
Department of Social Welfare, Jeonbuk National University, Jeonju, Republic of Korea.
J Med Internet Res. 2023 Dec 13;25:e48838. doi: 10.2196/48838.
The eHealth Literacy Scale (eHEALS) was introduced in China in 2013 as one of the most important electronic health literacy measurement instruments. After a decade of development in China, it has received widespread attention, although its theoretical underpinnings have been challenged, thus demanding more robust research evidence of factorial validity and multigroup measurement properties.
This study aimed to evaluate the Chinese version of the eHEALS in terms of its measurement properties.
A cross-sectional survey was conducted in a university setting in China. Item statistics were checked for response distributions and floor and ceiling effects. Internal consistency reliability was confirmed with Cronbach α, split-half reliability, Cronbach α if an item was deleted, and item-total correlation. A total of 5 representative eHEALS factor structures were examined and contrasted using confirmatory factor analysis. The study used the item-level content validity index (I-CVI) and the average of the I-CVI scores of all items on the scale to assess the content validity of the dominance model. Furthermore, the validated dominance model was subsequently used to evaluate the relevance and representation of elements in the instrument and to assess measurement invariance across genders.
A total of 972 respondents were identified, with a Cronbach α of .92, split-half reliability of 0.88, and item-total score correlation coefficients ranging from 0.715 to 0.781. Cronbach α if an item was deleted showed that all items should be retained. Acceptable content validity was supported by I-CVIs ≥0.80. The confirmatory factor analysis confirmed that the 3-factor model was acceptable. The measurement model met all relevant fit indices: average variance extracted from 0.663 to 0.680, composite reliability from 0.810 to 0.857, chi-square divided by the df of 4.768, root mean square error of approximation of 0.062, standardized root mean squared residual of 0.020, comparative fit index (CFI) of 0.987, and Tucker-Lewis index of 0.979. In addition, the scale demonstrated error variance invariance (Δnormed fit index=-0.016, Δincremental fit index=-0.012, ΔTucker-Lewis index=0.005, Δcomparative fit index=-0.012, Δrelative fit index=0.005, and Δroot mean square error of approximation=0.005).
A 3-factor model of the Chinese version of the eHEALS fits best, and our findings provide evidence for the strict measurement invariance of the instrument regarding gender.
电子健康素养量表（eHEALS）于 2013 年在中国推出，是最重要的电子健康素养测量工具之一。在中国经过十年的发展，它受到了广泛关注，尽管其理论基础受到了挑战，因此需要更有力的实证研究来证明其因子有效性和多群组测量特性。
本研究旨在评估中文版 eHEALS 的测量特性。
在中国的一所大学进行了横断面调查。对项目的分布和地板效应和天花板效应进行了反应分布和地板效应的检查。采用 Cronbach α、分半信度、删除一个项目后的 Cronbach α 和项目总分相关系数来确认内部一致性信度。共检验了 5 种具有代表性的 eHEALS 因子结构，并采用验证性因子分析进行对比。采用项目水平内容有效性指数（I-CVI）和量表中所有项目的平均 I-CVI 分数来评估优势模型的内容有效性。此外，还利用经过验证的优势模型来评估工具中各元素的相关性和代表性，并评估性别间的测量不变性。
共确定了 972 名受访者，Cronbach α 为 0.92，分半信度为 0.88，项目总分相关系数范围为 0.715 至 0.781。删除一个项目后的 Cronbach α 表明，所有项目都应保留。I-CVI≥0.80 支持可接受的内容有效性。验证性因子分析证实 3 因子模型是可以接受的。测量模型符合所有相关的拟合指标：平均方差从 0.663 到 0.680，综合可靠性从 0.810 到 0.857，卡方除以 df 为 4.768，近似均方根误差为 0.062，标准化均方根残差为 0.020，比较拟合指数（CFI）为 0.987，Tucker-Lewis 指数为 0.979。此外，该量表还表现出误差方差不变性（Δ规范化拟合指数=-0.016，Δ增量拟合指数=-0.012，Δ Tucker-Lewis 指数=0.005，Δ比较拟合指数=-0.012，Δ相对拟合指数=0.005，Δ近似均方根误差=0.005）。
中文版 eHEALS 的 3 因子模型拟合度最佳，我们的研究结果为该工具在性别方面具有严格的测量不变性提供了证据。
