一种新的处理迫选量表数据的累积模型研究.pdfVIP

Abstract Personality tests’ widely application in the talent assessment arouse the scholars’ concern on it. While using the single stimulus format scale to select employees is not so effective. That’s because that kind of scale do not do well in controlling the faking motives of applicants. Researches show that forced choice format scale is highly effective on controlling these motives. But now, most of the researches on forced choice format scale are based on CTT, from which the datum we can get are ipsative scores. Then we cannot compare with participants by using those ipsative scores and it is also hard to explain the scores. On the basis of Pairwise-Preference Model put forward and developed by Andrrich Strak, this study propose a new mode—2PLM-MU—to deal with the data fits with the Cumulative Model by using 2PLM to handle unidimensional data and combining 2PLM and Pairwise-Preference Model. Simulation study reveals that 2PLM-MU has high accuracy in estimating the participants ’ latent traits, among which test length affects the parameter estimation significantly and the proportion of unidimensional items also have a certain impact on it. But how the proportion of unidimensional items impacts needs further studies. Key words ：forced-choice questionnaire ； dominance model ； simulation II 目 录 摘 要 I Abstract II 目 录 III 引 言 1 1 文献综述 2 1.1 人格因素对员工绩效的预测效度2 1.2 应聘情境下人格测验的特点3 1．2．1 应聘情境下被试作伪的界定3 1．2．2 被试存在作伪的能力4 1．2．3 应聘情境下被试是否作伪的争论4 1．2．4 作伪的模式5 1．2．5 作伪的影响5 1.3 自陈量表控制作伪的能力比较6 1．3．1 单刺激形式量表控制作伪能力研究6 1．3．2 迫选量表控制作伪能力