期刊
出版社
COLEGIO OFICIAL PSICOLOGOS MADRID
DOI: 10.5093/jwop2019a11
关键词
Forced-choice questionnaires; Invariance; IRT; Likelihood Ratio test; MUPP-2PL
资金
- Spanish government's Ministerio de Economia y Competitividad [PSI 2015-65557-P, PSI 2017-85022-P]
Multidimensional forced-choice questionnaires are widely regarded in the personnel selection literature for their ability to control response biases. Recently developed IRT models usually rely on the assumption that item parameters remain invariant when they are paired in forced-choice blocks, without giving it much consideration. This study aims to test this assumption empirically on the MUPP-2PL model, comparing the parameter estimates of the forced-choice format to their graded-scale equivalent on a Big Five personality instrument. The assumption was found to hold reasonably well, especially for the discrimination parameters. In the cases in which it was violated, we briefly discuss the likely factors that may lead to non-invariance. We conclude discussing the practical implications of the results and providing a few guidelines for the design of forced-choice questionnaires based on the invariance assumption.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据