NUMBER OF SCALE POINTS AND THE RELIABILITY OF SCALES
NUMBER OF SCALE POINTS AND THE RELIABILITY OF SCALES
复制标题
DOI：
10.1177/001316446502500404
复制
发表时间：
1965-01-01
影响因子：
2.7
通讯作者：
GRAHAM, WK
中科院分区：
文献类型：
--
作者：
KOMORITA, SS;GRAHAM, WK
显然，采用两点量表的量表，如同意-不同意，正确-错误等，更短，更方便管理和评分。如果是这样的话，为什么那么多的研究者推荐并使用七点量表呢？选择量表类别数量的一个明显标准是受试者区分类别的能力。虽然类别较少的量表可能无法让受试者充分利用其辨别能力，但类别较多的量表可能超出受试者的辨别能力，从而增加测量误差。因此，使用大量类别的主要原因可能是为了增加量表的可靠性。在应用于等级量表时，Symonds（1924）在对这一问题的最早研究中，利用Kelley对相关系数的粗分组修正进行了理论分析，得出了使量表信度最大化的最佳类别数为7，而使用更多类别时信度的增加可以忽略不计的结论。然而，在这方面，
Clearly, scales which employ a two-point scale such as agree-disagree, true-false, etc., are much shorter and more convenient to administer and score. If so, why do so many investigators recom-mend and use the seven-point scale? One obvious criterion for the choice of the number of scale cate-gories is the ability of subjects to discriminate between the cate-gories. While a scale with few categories may not allow the subject to make full use of his capacity to discriminate, a scale with a large number of categories may be beyond the subject’s capacity to dis-criminate, thus, increasing errors of measurement. Thus, probably the main reason for using a large number of categories is to increase the reliability of the scale. As applied to rating scales, in one of the earliest studies of this problem, Symonds (1924) made a theoretical analysis using Kelley’s correction for coarse grouping on the corre-lation coefficient, and concluded that the optimal number of cate-gories to maximize scale reliability is seven, and that the increase in reliability when more categories are used is negligible. However,