Ottenbacher K J
State University of New York, Buffalo.
Am J Ment Retard. 1993 Jul;98(1):135-42.
The empirical literature examining the interrater agreement for visual analysis of single-subject data was examined using the technique of meta-analysis. Aggregated study results revealed an overall interrater agreement value of .58 for 14 studies representing a total of 789 raters. Additional analyses revealed that the interrater agreement values did not vary across raters based on previous experience or level of expertise. Characteristics of the graphs, including whether the data were hypothetical or real, or whether designs were the simple AB format or more complex (ABAB) designs, did not significantly influence the interrater agreement values. The only graphic characteristic related to increased interrater agreement was the presence of a trend (regression) line. Trend lines appeared to improve interrater agreement if the raters were familiar with their method of computation and interpretation. Simply including a trend line in the graphs produced only a marginal improvement in interrater agreement.
本研究运用元分析技术，对考察单受试者数据视觉分析中评分者间一致性的实证文献进行了审视。汇总的研究结果显示，14项研究（总计789名评分者）的评分者间一致性总体值为0.58。进一步分析表明，评分者间一致性值不会因评分者先前的经验或专业水平不同而有所差异。图表的特征，包括数据是假设的还是真实的，或者设计是简单的AB格式还是更复杂的（ABAB）设计，均未对评分者间一致性值产生显著影响。与评分者间一致性增加相关的唯一图表特征是存在趋势（回归）线。如果评分者熟悉趋势线的计算和解释方法，趋势线似乎会提高评分者间的一致性。仅仅在图表中加入趋势线，只会使评分者间一致性有微小的提高。
