应用Rasch模型分析Rosenberg自尊量表

采用Rosenberg自尊量表(RSES)对425名在校大学生进行施测,应用项目反应理论的Rasch模型对项目指标进行分析及DIF检验。结果表明,Rosenberg自尊量表具有单维性,量表的信度为0.84; 除项目8以外,其他项目拟合指标良好,较适用来区分中等及偏低自尊水平的个体,项目功能差异检验发现在项目1和项目5上存在DIF,表现为男生自尊水平要高于女生。相对于经典测量理论,应用Rasch模型分析Rosenberg自尊量表具有优势,为进一步的完善和使用该自尊量表提供依据。

Use Rasch Model to Analyze the Rosenberg Self-Esteem Scale

The study aimed to use the Rasch Model in analyzing the Rosenberg Self-Esteem Scale(RSES)to explain its model fitness of each item,explored the subjects individual differences of DIF,and it provided a basis for effectiveness and usage in RSES.425 undergraduates(mean age=21.71 years old,191 males)were sampled from Northeast China,they were measured by the Rosenberg Self-Esteem Scale(RSES)which containing 10 items.The data was collected by SPSS and analyzed through the software Conquest2.0 and R package.The results showed the Rosenberg Self-Esteem Scale's reliability was.84,its item difficulty distribution was from-1.006 to 1.497,more items difficulty levels indicated that the test is relatively easy for participants,the RSES can discriminated the examines with moderate and lower self-esteem rather than the upper.All except item 8 have acceptable model fit indices,the mean square of both Infit and Outfit MNSQ of item 8 in RSES were 1.51 and 1.55.Differential item functioning in gender was item 1 and 5,it showed that the male had higher self-esteem than female.All of these findings may provide implications of using Rasch Model for the Rosenberg Self-esteem Scale development and evaluation,it also support discussions and suggestions for future research.