Beck Stephen M, Perry J Christopher
Depression Research Clinic, Institute of Community and Family Psychiatry, Sir Mortimer B Davis Jewis General Hospital, Montreal, Quebec, Canada.
Psychiatry. 2008 Fall;71(3):219-33. doi: 10.1521/psyc.2008.71.3.219.
In a companion report (Beck & Perry, 2008), we reviewed the literature with regard to interview structure from which we derived seven operationalized quantitative measures. This report examines these measures as applied to five commonly used interview types--psychodynamic therapy sessions, dynamic interviews, Relationship Anecdote Paradigm (RAP) interviews, the Guided Clinical Interview and the Structured Clinical Interview for the DSM-IV axis I--each administered to the same six patients (n = 30). Two clinicians independently rated each interview using the Global Level of Interview Structure Scale (GLISS). Both the GLISS and six of the seven operationalized measures differed across interview types but not between subjects. Factor analysis yielded a single factor solution composed of five measures, not including a sixth measure (percentage of interviewer interventions that were questions) which was used as a solitary variable. Together the single factor and the percentage of questions predicted 75.2% of the variance in GLISS ratings, although no association was found between the factor and the percentage of questions. The GLISS and the operationalized measures captured distinct but complementary dimensions of interview structure. Discriminant analysis indicated that, on average, 80% of all interviews were correctly classified as to their type. Our main findings confirm that we can now accurately measure the degree of interview structure. Further research is needed to examine how these measures apply to other interview settings, such as psychoanalytic or cognitive-behavioral treatments, in the social sciences.
在一篇相关报告中（贝克和佩里，2008年），我们回顾了关于访谈结构的文献，从中得出了七种可操作的定量测量方法。本报告考察了这些测量方法应用于五种常用访谈类型的情况，即心理动力治疗会话、动态访谈、关系轶事范式（RAP）访谈、指导性临床访谈以及《精神疾病诊断与统计手册》第四版轴I的结构化临床访谈，每种访谈类型均对相同的六名患者进行（n = 30）。两名临床医生使用访谈结构总体水平量表（GLISS）对每次访谈进行独立评分。GLISS以及七种可操作测量方法中的六种在不同访谈类型之间存在差异，但在不同受试者之间不存在差异。因子分析得出了一个由五种测量方法组成的单因素解决方案，不包括用作单独变量的第六种测量方法（访谈者干预中为问题的百分比）。尽管在该因素与问题百分比之间未发现关联，但单因素和问题百分比共同预测了GLISS评分中75.2%的方差。GLISS和可操作测量方法捕捉到了访谈结构中不同但互补的维度。判别分析表明，平均而言，所有访谈中有80%能被正确分类到其所属类型。我们的主要发现证实，我们现在能够准确测量访谈结构的程度。需要进一步研究来考察这些测量方法如何应用于其他访谈情境，如社会科学中的精神分析或认知行为治疗。
