Department of Psychology, McGill University, Montreal, Quebec, Canada.
J Child Psychol Psychiatry. 2010 Sep;51(9):1040-9. doi: 10.1111/j.1469-7610.2010.02244.x. Epub 2010 Apr 7.
Although structured interviews are assumed to be scientifically superior to checklists for measuring youth psychopathology, few studies have tested this hypothesis. Interviews place a much greater burden on respondents, making it critical to determine their added value when quantifying psychiatric symptoms.
Confirmatory factor analysis was used to compare interviews and checklists in community (N = 251) and clinically referred (N = 406) samples of youth aged 5 to 17 years. We examined the associations between mother-reported externalizing symptoms assessed by interview versus checklist against (a) teacher-reported externalizing symptoms, and (b) child's gender, academic performance, single- versus two-parent family, and family income. Models in which associations were estimated freely were contrasted to models in which the interview and the checklist were constrained to have equal associations with the variables. Finding these models fit comparably would suggest no difference between interviews and checklists.
In the community sample, both the constrained and unconstrained models provided comparable fit to the data, suggesting no marked differences between interviews and checklists. In the clinical sample, associations with the interview were generally stronger. Reducing the number of items on the interview to match those on the 6-item checklist eliminated these differences, suggesting that the increased reliability of the interview scales, afforded by additional items, enhanced their quantification of psychopathology.
Consistent with previous studies, interviews were not notably superior to checklists for the measurement of externalizing symptoms. When only a few items are used, small performance differences between checklists and interviews may be due to scale length.
虽然结构式访谈被认为在测量青少年精神病理学方面比清单更具科学性，但很少有研究对此假设进行检验。访谈给受访者带来了更大的负担，因此在量化精神症状时，确定其附加价值至关重要。
采用验证性因子分析比较了社区（N=251）和临床转诊（N=406）样本中 5 至 17 岁青少年的访谈和清单。我们考察了访谈与清单评估的母亲报告的外显症状与（a）教师报告的外显症状，以及（b）儿童性别、学业表现、单亲家庭还是双亲家庭以及家庭收入之间的关联。对比了自由估计关联的模型和访谈与清单具有相等关联的模型。如果这些模型拟合相当，则表明访谈和清单之间没有差异。
在社区样本中，约束和非约束模型都为数据提供了相当的拟合度，表明访谈和清单之间没有明显差异。在临床样本中，与访谈的关联通常更强。将访谈中的项目数量减少到与 6 项清单相匹配消除了这些差异，表明访谈量表通过增加额外项目而提高的可靠性增强了它们对精神病理学的量化。
与之前的研究一致，访谈在测量外显症状方面并不明显优于清单。当只使用几个项目时，清单和访谈之间的小性能差异可能归因于量表长度。
