Cairns V, von Zerssen D, Stutte K H, Mombour W
J Psychiatr Res. 1982;17(1):19-28. doi: 10.1016/0022-3956(82)90031-0.
A full set of data has been gathered on admission of 1654 inpatients and included information on all 90 items from the Inpatient Multidimensional Psychiatric Scale (IMPS). The sample was divided into two by means of an odd-even split and principal component analyses with Varimax rotation as well as Ward hierarchical cluster analyses of symptoms were run on both subsamples. Comparisons of results were than made between the two subsamples, between the two methods of analysis, and with the studies of Lorr and Klett and Behrends et al. In general, earlier results concerning the structure of the symptom groupings have been confirmed. However, by comparing all the different sets of results, those symptom groupings that are particularly stable or unstable have been identified. In addition, comparisons have been made between the results from analyses using the full range of scores on the items and those using dichotomized scores in order to determine whether the use of different rating scales for different items has an effect on the resulting item groupings.
已收集了1654名住院患者入院时的全套数据，其中包括来自住院多维精神病量表（IMPS）所有90个项目的信息。通过奇偶分割将样本分为两组，并对两个子样本进行了主成分分析（采用方差最大化旋转）以及症状的沃德分层聚类分析。然后对两个子样本之间、两种分析方法之间的结果进行了比较，并与洛尔和克莱特以及贝伦兹等人的研究进行了比较。总体而言，早期有关症状分组结构的结果得到了证实。然而，通过比较所有不同组的结果，确定了那些特别稳定或不稳定的症状分组。此外，还对使用项目全部分数进行分析的结果与使用二分法分数进行分析的结果进行了比较，以确定对不同项目使用不同评分量表是否会对最终的项目分组产生影响。
