融合反应时的多级评分IRT模型开发及其应用研究

当前大多数融合反应时的IRT模型仅适用于0-1评分数据资料,极大的限制了IRT反应时模型在实际中的应用。本文在传统的二级计分反应时IRT模型基础上,拟开发一种多级评分反应时模型。在层次建模框架下,分别采用拓广分部评分模型(GPCM)和对数正态模型构建融合反应时的多级评分IRT模型(本文记为JRT-GPCM),并采用全息贝叶斯MCMC算法实现新模型的参数估计。为验证新开发的JRT-GPCM模型的可行性及其在实践中的应用,本文开展了两项研究:研究1为模拟实验研究,研究2为新模型在大五人格-神经质分量表中的应用。研究1结果表明,JRT-GPCM模型的估计精度较高,且具有较好的稳健性。研究2表明,被试的潜在特质与作答速度具有一定的正相关,且本研究结果支持Ferrando和Lorenzo-Seva(2007)提出的“距离-困难度假设”,即当被试的潜在特质与项目的难度阈限距离越远,那么被试会花费更多的时间对项目进行作答。总之,本研究为拓展反应时信息在心理测量及教育中的应用提供新的方法支持。

Research on Development and Application of Polytomous IRT Model Incorporating Response Times

With the development of computer testing technology,collecting reaction time has become a routine work of many large-scale tests.However,most current IRT models for fusion reaction time are only applicable to 0-1 score data,which greatly limits the application of IRT model in practice.Based on the traditional two-level scoring response time IRT model,this paper intends to develop a multilevel scoring response time model.Under the framework of hierarchical modeling,the extended partial scoring model(GPCM)and the log-normal model(jrt-gpcm)were used to construct the multi-stage scoring IRT model(jrt-gpcm)for fusion reaction,and the parameter estimation of the new model was realized by the holographic bayesian MCMC algorithm.In order to verify the feasibility of the newly developed jrt-gpcm model and its application in practice,this paper carried out two studies:Study 1 for simulation experiment research,the use of 2 x 2 double factor experiment design,one factor for the number of participants(1000 and 2000 respectively,the two level),another factor for the test number(20 and 30 respectively two levels),all items of 0,1,2,3 multistage grading,using holographic bay leaf,MCMC algorithm for parameter estimation,and validates the feasibility of MCMC algorithm and JRT-GPCM model to estimate accuracy; Study 2 for JRT-GPCM model in the application of the big five personality-neurotic subscales,testing group for college students,this paper USES the computer answer way,collected a total of 1030 data(including the answer in each available data that reaction time),by eliminating the invalid data(such as too many missing data/answer exception)on lie detection problem,the final valid data is 845.Study 1 results show that under the JRT-GPCM model,the estimated method of MCMC algorithm by fairly robustness,and the precision of the item and the person the parameters was preferably great,model has good robustness,and the topic,the more the higher estimation precision,It indicated that the number of subjects indicated that the rrt-gpcm model was reasonable and feasible.Study 2 shows that the parameter estimation indexes of all items are basically less than 1.1,indicating the convergence of parameter estimation of MCMC algorithm.The variance of each parameter and the standard deviation of the covariance are small,which indicates that the model has good robustness in empirical research.The 12 questions on the neurotic subscale ranged from 0.895 to 1.209,all of which were greater than 0.7(Fliege,2015),indicating that the 12 questions were of good quality.There was a positive correlation between the potential traits and the response speed of the subjects.The higher the neurotic tendency of the subjects,the higher the potential traits and the faster the response speed.Project step parameters(the location parameter)and its parameters is related to the intensity of time,the greater the absolute value of that project step parameters(off center value,then represents to the characteristics of extreme levels),so the participants answers in the less time needed for the project,namely the time intensity is small,the results support Ferrando and Lorenzo-Seva(2007)proposed “distance-difficult holiday”.In conclusion,this study provides a new method to expand the application of response time information in psychological measurement and education.