速度与准确率权衡：被试反应状态评价与建模  

The Trade-off between Speed and Accuracy:Evaluation and Modeling for Participants in the Response States

作　　者：郭小军[1] 罗照盛[1] GUO Xiaojun;LUO Zhaosheng(Psychology College,Jiangxi Normal University,Nanchang 330022)

机构地区：[1]江西师范大学心理学院

出　　处：《心理与行为研究》2019年第5期589-595,共7页Studies of Psychology and Behavior

基　　金：国家自然科学基金（31660279）

摘　　要：在测验中,被试总是被要求在规定的时间内完成测验作答。在测验时间限制内,被试在不同项目上可能存在不同的权衡倾向。在有的项目上,随着反应时增加,准确率提高到一定程度后反而会降低。针对这种情形,本研究从能力、速度以及能力与速度补偿角度对其原因进行探索,并以此构建了三个反应时与准确率非单调关系的分层模型。结果表明,MCMC(Markov chain Monte Carlo)算法能准确地估计模型各参数;而且在实际数据分析中发现反应时与准确率的非单调关系更可能与被试的能力有关。Most tests are administered within an allocated time.Within the limited time,participants might have different trade-offs on different items.With the increase of response time,the accuracy reduces when it is improved to a certain point in some items.In view of this situation,this study explored the causes from the point of view of ability,speed,ability and speed compensation.Therefore,three hierarchical models of the non-monotonic relationship between response time and accuracy were constructed.The results obtained in the simulation studies indicated that the MCMC algorithm performed well and enabled simultaneous estimation.More specifically,increasing the sample size of the participants led to better estimation for item parameters;and increasing the test length,the person parameters became better.Finally,three proposed hierarchical models have been empirically studied by a real-data example in the Raven's Standard Progressive Matrices.The results showed that the non-monotonic relationship between response time and accuracy is more likely to be affected by the ability level.

关 键 词：反应时 准确率 非单调关系 分层模型 

分 类 号：B841[哲学宗教—基础心理学]