Individual Presentation Summary
Share...

Direct link:

On Posterior Probability of Latent Classifications in Diagnostic Models

Sun, April 14, 1:15 to 2:45pm, Convention Center, Floor: First, 121C

Abstract

Parametric estimation of person parameters in cognitive diagnosis models hinges on posterior probabilities that are sample-dependent. This study investigates the impact of sample dependency on model calibration and scoring, proposing a multiple imputation procedure to correct it. Simulation results highlight the superiority of the new method with small samples.

Authors