医学研究前后测量设计资料差值的信度评价

Evaluation of the Difference Score Reliabilities of Pretest-posttest Design Data in Medical Studies

目的通过医学研究实例,评价医学研究前后测量设计资料差值的信度,探讨前后测量的相关性及其对差值信度的影响。方法收集13个前后测量设计资料的医学研究实例,信度评价指标采用重复测量差值的标准差、Cron-bachα系数和差值信度系数;计算采用可自动执行的SPSS11·0宏命令。通过一个具体实例说明影响差值信度的主要因素。结果13个实例中,26组的差值信度都比实测值的信度低,其中6组为0信度。结论医学研究前后测量设计资料,如果满足前后两次测量值的方差相等、信度不变,差值信度要低于实测值信度。因此,采用差值作为医学前后测量设计资料的反应变量,需评价其信度大小。

Objective Through medical research examples to evaluate the difference score reliabilities of pretest-posttest design data in medical studies and to explore the pretest-posttest correlation and its effect on the reliability. Methods 13 sets of pretest-posttest design data were collected. The reliability indicators for evaluation were the SD of difference scores, Cronbach a coefficients and reliability coefficients of difference scores. SPSS11. 0 Macro was used for automated data processing. An example was adopted to show the factors affecting the difference score reliability. Results In 26 groups of 13 examples, the difference score reliabilities exclusively decreased compared with the measurement scores. Among these 6 groups＇ reliabilities were zero. Conclusion If the pretest scores and the, posttest scores have equal variances and equal reliabilities, the difference score reliabilities are much less reliable than the measurement scores themselves. The difference score reliabilities should be evaluated when taking the difference sores as s response variable to describe the pretest-posttest design data,