van Breukelen G J, Berger M P
Universiteit Maastricht, capaciteitsgroep Methodologie en Statistiek.
Ned Tijdschr Geneeskd. 2000 Aug 12;144(33):1580-4.
In medical research the responses of patients to a number of items are often combined into a summary measure or total score for a more general patient characteristic, such as mobility. Clinical and statistical criteria exist for determining the weight of each question in the total score. The choice of weights may affect the differences between patients. The total score is often analysed statistically as if it had interval level of measurement, thereby allowing conclusions about the size of differences between patients, but it has merely an ordinal level. The incorrectness of the interval assumption may also affect differences between patients or patient groups. Finally, differences between patients are less dependent on the weighting of the items and on the assumption of an interval scale as the total score is based on more homogeneous items. Standard statistical software contains statistical methods for determining the homogeneity as well as the weights.
在医学研究中，患者对多个项目的反应常常被合并为一个综合指标或总分，以反映更一般的患者特征，如活动能力。存在临床和统计标准来确定总分中每个问题的权重。权重的选择可能会影响患者之间的差异。总分通常会进行统计分析，就好像它具有区间测量水平一样，从而能够得出关于患者之间差异大小的结论，但实际上它仅仅具有顺序测量水平。区间假设的不正确性也可能影响患者或患者组之间的差异。最后，患者之间的差异较少依赖于项目的加权以及区间量表的假设，因为总分是基于更同质的项目得出的。标准统计软件包含用于确定同质性以及权重的统计方法。
