Czobor P, Bitter I, Volavka J
Department of Psychiatry, New York University School of Medicine, NY.
Psychiatry Res. 1991 Feb;36(2):129-39. doi: 10.1016/0165-1781(91)90125-9.
Because overlapping psychometric scales are used frequently in psychiatric research, examination of the relationship between scales has become increasingly important. The concept of relationship is the focus of this article. By way of illustration, the Brief Psychiatric Rating Scale (BPRS) and the Scale for the Assessment of Negative Symptoms (SANS) were compared for correlation and redundancy. Since these scales are frequently represented by derived summary variables (e.g., factors, total scores), it is also important to assess the effect of such representation on measures of relationship. The SANS and the BPRS were found to be highly intercorrelated. Nevertheless, the individual items and the subscale scores of the SANS contain information independent from the BPRS: the best BPRS predictor variates can explain only approximately half of the total variance of the SANS. When the SANS, however, is represented by a single variable (composite score), it becomes highly redundant with the anergia factor of the BPRS.
由于重叠的心理测量量表在精神病学研究中经常被使用,因此对量表之间关系的考察变得越来越重要。关系的概念是本文的重点。作为例证,对简明精神病评定量表(BPRS)和阴性症状评定量表(SANS)进行了相关性和冗余性比较。由于这些量表经常由派生的汇总变量(如因子、总分)来表示,因此评估这种表示方式对关系测量的影响也很重要。结果发现,SANS和BPRS高度相互关联。然而,SANS的单个项目和分量表得分包含独立于BPRS的信息:最佳的BPRS预测变量只能解释SANS总方差的大约一半。然而,当SANS由单个变量(综合得分)表示时,它与BPRS的无活力因子变得高度冗余。