Abstract: To explore the psychometric properties of the Chinese internet addiction scale(revision)for college students using the multivariate generalizability theory, we make the generalizability design of random measurement model, and 1200 students were surveyed with the Chinese internet addiction scale(revision). The results showed that the correlation coefficients among each factor were bigger than 0.92 in bi-factor structure, and their coefficients in five-factor structure were between 0.76 and 0.97. The generalizability coefficient and reliability index of the whole scale were bigger than 0.94. For each factor, they were about 0.90 in bi-factor structure and were between 0.74 and 0.85 in five-factor structure. In conclusion, the reliability was excellent for the whole scale and the proportion of each factor in the scale. It could be used for the norm-referenced test and the criterion-referenced test for college students.

Key words: internet addiction, multivariate generalizability theory, generalizability coefficient, reliability index.

摘要： 用多元概化理论考察大学生网络成瘾量表在大学生群体中应用的测量学性能。以随机测量模式的概化设计,针对1200名在校大学生进行问卷调查。结果显示双因子结构的相关程度在0.92以上,五因子结构的相关程度均在0.76~0.97间;整体量表的概化系数和可靠性指数均达到了0.94以上,而双因子结构各因子在0.90左右,五因子结构各因子均在0.74~0.85间。所以,整体量表及各因子在大学生群体中应用的信效度较高,可用作常模和标准参照测验;无论双因子还是五因子结构,CIAS-R各因子在分值比和项目数上,设计非常合理和完善。

关键词: 网络成瘾, 多元概化分析, 概化系数, 可靠性指数。