【作者】 彭旺； 赵守盈；

【Author】 PENG Wang;Zhao Shouying;School of Psychology,Guizhou Normal University;

【机构】 贵州师范大学心理学院；

【摘要】 自我批评反刍,是个体反复进行消极自我评价的现象。有研究发现,自我批评反刍与正念水平呈负相关,与压力呈正相关;同时还发现自我批评反刍能够调节依恋焦虑和感知压力之间的关系。由此可见,自我批评反刍对个体心理健康具有重要影响,可作为一个较好的预测变量。本研究的目的是运用Rasch模型和Mokken模型两种不同取向项目理论模型对中文版自我批评反刍量表的单维结构进行验证。使用Winsteps软件进行Rasch分析,检验数据和模型的拟合程度、项目功能差异(DIF);使用R程序中的Mokken包进行Mokken分析,进行单维性、单调性和题目难度排序不变性检验。结果:(1)自我批评反刍量表第一特征根值为6.346,第二特征根值为0.821,比值为7.730,符合单维性假设,可进行Rasch分析;被试和题目的分离系数分别为2.84和2.15,被试和题目的信度分别为0.82和0.82,均达到测量学标准;10个题目的 INFIT MNSQ的取值范围在0.72-1.34之间,OUTFIT MNSQ的取值范围在0.69-1.45之间,模型拟合较好;进行性别的DIF检验时,有三个题目存在DIF,可能与中国的传统文化、社会环境等有关;(2)总的量表系数H为0.608,10个题目的 Hi均大于0.5,90组试题对的Hij未出现小于0的结果,表明满足单维性假设;使用Monotonicity方法进行单调性检验,10个题目的项目反应曲线图均呈现为单调递增,说明没有违反单调性假设;使用Pmatrix和Restscore两种方法检验题目难度排序不变性发现不存在严重违反假定的情况,但是MIIO方法结果显示HT仅为0.04,说明不满足题目难度排序不变性;Mokken模型共报告四种信度:SM量表信度0.929,Cronbachα0.929,葛特曼信度系数0.930,潜在类别信度系数0.930,表明量表信度良好;Mokken分析结果说明量表满足单调同质模型。结论:Rasch模型和Mokken模型分析结果表明中文版自我批评反刍量表的单维结构良好。更多还原

【Abstract】 Self-criticism ruminating is a phenomenon in which individuals repeatedly conduct negative self-evaluation. Some studies have found that self-criticism rumination is negatively correlated with mindfulness level and positively correlated with stress. At the same time, it is also found that self-criticism rumination can regulate the relationship between attachment anxiety and perceived stress. Therefore, self-criticism rumination has an important impact on individual mental health and can be used as a better predictor. The purpose of this study is to verify the one-dimensional structure of the Chinese version of the Self-critical Ruminant Scale by using Rasch Model and Mokken Model, two theoretical models with different orientations. Rasch analysis was carried out by Winsteps software to check the fitting degree of data and model and the difference in project function(DIF); Mokken analysis was carried out by using Mokken package in R program, and the invariance tests of unidimensionality, monotonicity and problem difficulty ranking were carried out. Results:(1) The first characteristic root value of the self-criticism rumination scale was 6.346, the second characteristic root value was 0.821, and the ratio was 7.730, which was consistent with the unidimensional hypothesis and could be analyzed by Rasch. The separation coefficients of subjects and subjects were 2.84 and 2.15 respectively, and the reliability of subjects and subjects were 0.82 and 0.82 respectively, which reached the measurement standard. The value range of INFIT MNSQ for 10 items is 0.72-1.34, and the value range of OUTFIT MNSQ is 0.69-1.45. The model fits well. In the DIF test of gender, there are three questions that have DIF, which may be related to China’s traditional culture and social environment.(2) The total scale coefficient H is 0.608, the Hi of 10 questions is greater than 0.5, and the Hij of 90 pairs of questions does not show a result less than 0, which indicates that the unidimensional hypothesis is satisfied. Monotonicity method was used to test the monotonicity, and the item response curves of 10 topics showed monotonic increase, which showed that the monotonicity assumption was not violated. The two methods of Pmatrix and Restscore are used to check the invariance of topic difficulty ranking. It is found that there is no serious violation of the assumption, but the MIIO method results show HT is only 0.04, which indicates that the invariance of topic difficulty ranking is not satisfied. Mokken model reported four kinds of reliability: SM scale reliability 0.929, Cronbach α 0.929, Gutman reliability coefficient 0.930, potential category reliability coefficient 0.930, indicating that the scale reliability is good. The results of Mokken analysis show that the scale satisfies monotone homogeneous model. Conclusion: The analysis results of Rasch model and Mokken model show that the Chinese version of Self-critical Ruminant Scale has a good one-dimensional structure.更多还原