Item Latent Structure Analysis by Marginalizing Latent Variable(<Special Issue>New Generation Learning Assessments)

DOI Web Site 13 References
  • HASHIMOTO Takamitsu
    Graduate School of Information Systems, The University of Electro-Communications:Research Division, National Center for University Entrance Examinations
  • UENO Maomi
    Graduate School of Information Systems, The University of Electro-Communications

Bibliographic Information

Other Title
  • 潜在変数周辺化による項目潜在構造分析(<特集>新時代の学習評価)
  • 潜在変数周辺化による項目潜在構造分析
  • センザイ ヘンスウ シュウヘンカ ニ ヨル コウモク センザイ コウゾウ ブンセキ

Search this article

Abstract

Item Relational Structure (IRS) analysis and fuzzy graphing are known as popular methods to express relations among test items. These methods can estimate the item structure from test data through easy calculation. However, when two items are affected strongly by the latent ability variable, relations can be detected incorrectly because of the relation through the latent variable. This paper introduces Item Latent Structure (ILS) analysis, which uses the Latent Conditional Independence (LCI) test, to assess the conditional independence between two items given a latent variable. After simulation and application to actual data, results demonstrate that ILS analysis can detect conditional independence correctly given a latent ability variable.

Journal

References(13)*help

See more

Keywords

Details 詳細情報について

Report a problem

Back to top