项目反应理论在居民健康素养标准参照测验中的应用研究

Item Response Theory in the Application of ‘Residents Health Literacy Criterion-Referenced Test’

目的运用项目反应理论（IRT）对居民健康素养调查问卷的条目进行测评,探讨基于信息函数建立居民健康素养标准参照测验并评价项目性能。方法以1000例健康素养调查样本为例,利用单维三参数logistic模型（3PLM）拟合样本,估计项目参数与信息函数,以掌握水平比例的能力水平作为决策点（测验划界分数）。结果问卷73个项目的三参数logistic模型拟合优度98.6%,难度在-4.27~4.23,区分度在0.32~1.58,猜测度在0.009~0.5,测验在能力水平为-0.13时获得最大信息量与最小测量误差,对应的掌握比例为0.62（划界分数）。结论 IRT在问卷的测评编制中能深刻而合逻辑地分析每个测验项目与整个测验间的定量关系,2013版《江苏省居民健康素养调查问卷》在中低能力水平的被试中具有较高测量精度。

Objective To analyze items of the ＇ Residents＇ health literacy questionnaire＇ using item response theory （IRT）. Using the information function of IRT to establish ＇ Residents Health Literacy Criterion-Referenced Test＇ and evaluate its item performance. Methods Unidimensional three parameters logistic model （3PLM） was used to fit the sample including 1000 cases and to calculate item parameters and information function. The grasp proportion of ability parameter was as a decision point（ The cut-off score）. Results The goodness of fit of the questionnaire of 73 items is 98.6%. Among the 73 items, the dif- ficulty parameters of all items were between -4. 27 - 4. 23, the discrimination parameters of all items were between 0. 32 ~ 1.58 and the guessing parameters of all items were between 0. 009 - 0. 5. The test maximized the information and minimize the meas- urement error at ability parameter of -0. 13 with the corresponding grasp proportion of 0. 62. The test should decrease the diffi- culty when using grasp proportion 0. 6 as the cut-off score while increase the difficulty and discrimination when using grasp pro- portion 0. 8 as the cut-off score. Conclusion IRT analysis questionnaire＇ could deeply and logically comprehend the quantitative relation between each item and the whole test. Residents＇ health literacy questionnaire of Jiangsu province applied to middle or low level ability subjects.