Somma Antonella, Borroni Serena, Maffei Cesare, Giarolli Laura E, Markon Kristian E, Krueger Robert F, Fossati Andrea
LUMSA University, Rome, Italy.
San Raffaele Hospital, Milan, Italy.
J Pers Disord. 2017 Oct;31(5):624-646. doi: 10.1521/pedi_2017_31_272. Epub 2017 Jan 10.
In order to assess the reliability, factorial validity, and criterion validity of the Personality Inventory for DSM-5 (PID-5) among adolescents, 1,264 Italian high school students were administered the PID-5. Participants were also administered the Questionnaire on Relationships and Substance Use as a criterion measure. In the full sample, McDonald's ω values were adequate for the PID-5 scales (median ω = .85, SD = .06), except for Suspiciousness. However, all PID-5 scales showed average inter-item correlation values in the .20-.55 range. Exploratory structural equation modeling analyses provided moderate support for the a priori model of PID-5 trait scales. Ordinal logistic regression analyses showed that selected PID-5 trait scales predicted a significant, albeit moderate (Cox & Snell R values ranged from .08 to .15, all ps < .001) amount of variance in Questionnaire on Relationships and Substance Use variables.
为评估《精神疾病诊断与统计手册》第五版人格量表（PID - 5）在青少年中的信度、因子效度和效标效度，对1264名意大利高中生施测了PID - 5。参与者还接受了人际关系与物质使用问卷作为效标测量。在全样本中，除怀疑量表外，PID - 5各量表的麦克唐纳ω值均足够（中位数ω = .85，标准差 = .06）。然而，所有PID - 5量表的平均项目间相关值在.20 - .55范围内。探索性结构方程模型分析为先验的PID - 5特质量表模型提供了适度支持。有序逻辑回归分析表明，选定的PID - 5特质量表预测了人际关系与物质使用问卷变量中显著但适度（考克斯和斯内尔R值范围为.08至.15，所有p < .001）的方差量。
