3.1 MIRT vs CDMs

3.1.1 Multidimensional item response theory (IRT) models

Attributes are assumed to be continuous

Examples include multidimensional 2- and 3PL models, the multicomponent latent trait model, and the multidimensional random coefficients multinomial logit model

Challenges in using multidimensional IRT models

  • handling a large number of attributes

  • structuring the latent space

  • accommodating diverse item-trait interactions

3.1.2 Cognitive Diagnosis models

Attributes are assumed to be categorical

Also called Diagnostic classification models, Cognitive psychometric models, Multiple classification (latent class) models, Latent response models, Restricted latent class models, Structured located latent class models, Structured IRT models

Defining characteristics of CDMs (Rupp et al., 2010)

  • Multidimensional nature

  • Confirmatory nature

  • Complexity of their loading structure

  • Types of observed response variables

  • Types of latent predictor variables

  • Interactions of the latent predictor variables

  • Criterion-referenced interpretations

  • Diagnostic nature of the interpretations

  • Types of heterogeneity they can model

References

Rupp, A. A., Templin, J., & Henson, R. A. (2010). Diagnostic measurement: theory, methods, and applications. Guilford Press.