A partial mastery, higher-order latent structural model for polytomous attributes in cognitive diagnostic assessments

Peida ZHAN, Wen Chung WANG, Xiaomin Nicky LI

Research output: Contribution to journalArticle

Abstract

The latent attribute space in cognitive diagnosis models (CDMs) is often assumed to be unstructured or saturated. In recent years, the number of latent attributes in real tests has often been found to be large, and polytomous latent attributes have been advocated. Therefore, it is preferable to adopt substantive theories to connect seemingly unrelated latent attributes, to replace the unstructured or saturated latent structural models (LSMs) with structured or parsimonious ones, with simplified parameter estimation. In the present study, we developed a partial mastery, higher-order LSM for polytomous attributes, which was built upon the framework of adjacent-category logit models to account for a higher-order latent structure of multiple polytomous attributes. The proposed model can be incorporated into many existing CDMs. We conducted simulations to evaluate the psychometric properties of the proposed model and obtained good parameter recovery. We then provided an empirical example to demonstrate the applications and the advantages of the proposed model. Copyright © 2019 The Classification Society.
Original languageEnglish
Pages (from-to)328-351
JournalJournal of Classification
Volume37
Issue number2
Early online dateApr 2019
DOIs
Publication statusPublished - Jul 2020

Citation

Zhan, P., Wang, W.-C., & Li, X. (2020). A partial mastery, higher-order latent structural model for polytomous attributes in cognitive diagnostic assessments. Journal of Classification, 37(2), 328-351. doi: 10.1007/s00357-019-09323-7

Keywords

  • Cognitive diagnosis
  • Latent class models
  • Polytomous attributes
  • Higher-order structure
  • Latent structural model
  • DINA model

Fingerprint Dive into the research topics of 'A partial mastery, higher-order latent structural model for polytomous attributes in cognitive diagnostic assessments'. Together they form a unique fingerprint.