Chien Tsair-Wei, Shao Yang, Jen Dong-Hui
Department of Medical Research, Chi-Mei Medical Center, Tainan, Taiwan.
Department of Hospital and Health Care Administration, Chia-Nan University of Pharmacy and Science, Tainan, Taiwan.
Health Qual Life Outcomes. 2017 Oct 27;15(1):216. doi: 10.1186/s12955-017-0784-8.
Many quality-of-life studies have been conducted in healthcare settings, but few have used Microsoft Excel to incorporate Cronbach's α with dimension coefficient (DC) for describing a scale's characteristics. To present a computer module that can report a scale's validity, we manipulated datasets to verify a DC that can be used as a factor retention criterion for demonstrating its usefulness in a patient safety culture survey (PSC).
Microsoft Excel Visual Basic for Applications was used to design a computer module for simulating 2000 datasets fitting the Rasch rating scale model. The datasets consisted of (i) five dual correlation coefficients (correl. = 0.3, 0.5, 0.7, 0.9, and 1.0) on two latent traits (i.e., true scores) following a normal distribution and responses to their respective 1/3 and 2/3 items in length; (ii) 20 scenarios of item lengths from 5 to 100; and (iii) 20 sample sizes from 50 to 1000. Each item containing 5-point polytomous responses was uniformly distributed in difficulty across a ± 2 logit range. Three methods (i.e., dimension interrelation ≥0.7, Horn's parallel analysis (PA) 95% confidence interval, and individual random eigenvalues) were used for determining one factor to retain. DC refers to the binary classification (1 as one factor and 0 as many factors) used for examining accuracy with the indicators sensitivity, specificity, and area under receiver operating characteristic curve (AUC). The scale's reliability and DC were simultaneously calculated for each simulative dataset. PSC real data were demonstrated with DC to interpret reports of the unit-based construct validity using the author-made MS Excel module.
The DC method presented accurate sensitivity (=0.96), specificity (=0.92) with a DC criterion (≥0.70), and AUC (=0.98) that were higher than those of the two PA methods. PA combined with DC yielded good sensitivity (=0.96), specificity (=1.0) with a DC criterion (≥0.70), and AUC (=0.99).
Advances in computer technology may enable healthcare users familiar with MS Excel to apply DC as a factor retention criterion for determining a scale's unidimensionality and evaluating a scale's quality.
许多生活质量研究已在医疗环境中开展，但很少有研究使用Microsoft Excel将克朗巴哈α系数与维度系数（DC）相结合来描述量表的特征。为了展示一个能够报告量表效度的计算机模块，我们对数据集进行了处理，以验证一个可作为因子保留标准的DC，以证明其在患者安全文化调查（PSC）中的有用性。
使用Microsoft Excel的应用程序可视化Basic来设计一个计算机模块，用于模拟2000个符合拉施评分量表模型的数据集。这些数据集包括：（i）两个呈正态分布的潜在特质（即真实分数）上的五个双相关系数（相关系数分别为0.3、0.5、0.7、0.9和1.0）以及对其各自1/3和2/3长度项目的回答；（ii）20种从5到100的项目长度情景；以及（iii）20种从50到1000的样本量。每个包含5点多分类回答的项目在难度上在±2个对数单位范围内均匀分布。使用三种方法（即维度相互关系≥0.7、霍恩平行分析（PA）95%置信区间和个体随机特征值）来确定保留一个因子。DC指用于通过灵敏度、特异性和接受者操作特征曲线下面积（AUC）指标检验准确性的二元分类（1表示一个因子，0表示多个因子）。为每个模拟数据集同时计算量表的信度和DC。使用作者制作的MS Excel模块，用DC展示PSC实际数据，以解释基于单位的结构效度报告。
DC方法呈现出准确的灵敏度（=0.96）、特异性（=0.92），DC标准（≥0.70）以及AUC（=0.98），均高于两种PA方法。PA与DC相结合产生了良好的灵敏度（=0.96）、特异性（=1.0），DC标准（≥0.70）以及AUC（=0.99）。
计算机技术的进步可能使熟悉MS Excel的医疗用户能够将DC用作因子保留标准，以确定量表的单维度性并评估量表质量。
