Russell E W
J Clin Exp Neuropsychol. 1987 Aug;9(4):376-92. doi: 10.1080/01688638708405058.
Scaling of neuropsychological test variables requires knowledge of test parameters. This is a study of the parameters for 12 tests in the Rennick Index of the Halstead-Reitan Battery, using 732 subjects. z-score distributions were plotted. Great variability among test distributions indicated that, for brain-damaged subjects: the range is much greater than for normals; maximum z scores ranged from 3.8 to 27.3 for different tests; the form of the distributions was highly variable, ranging from normal to highly skewed distributions. Thus, the same z scores indicated quite different amounts of impairment on different tests. This raises questions regarding the applicability of z and T scores to a brain-damaged population. The second part of this study demonstrated how scaled scores, indicating equal amounts of impairment, could be constructed. This method used a reference scale produced by averaging control subjects' z scores on 12 tests. Individual scaled scores were obtained through a linear regression prediction.
神经心理测试变量的标准化需要了解测试参数。本研究使用732名受试者，对霍尔斯特德-雷坦成套测验的伦尼克指数中的12项测试参数进行了研究。绘制了z分数分布。测试分布之间存在很大差异，这表明，对于脑损伤受试者：范围比正常人的要大得多；不同测试的最大z分数范围从3.8到27.3；分布形式高度可变，从正态分布到高度偏态分布。因此，相同的z分数在不同测试中表明了相当不同程度的损伤。这就引发了关于z分数和T分数对脑损伤人群适用性的问题。本研究的第二部分展示了如何构建表示同等损伤程度的量表分数。该方法使用通过对12项测试中对照组受试者的z分数进行平均而产生的参考量表。通过线性回归预测获得个体量表分数。
