Beyond the Noise: Uncovering the Effects of Careless Respondents in Cognitive Diagnostic Modeling

• In Event: Virtual Poster Session
  In Virtual iPresentation Only: Division D - Section 1: Measurement, Psychometrics, and Assessment Virtual Poster Session

Abstract

This study investigates the impact of careless respondents on the estimation of category-level guess and slip parameters in the sequential-GDINA (s-GDINA) model using Monte Carlo simulations. Four types of careless response patterns were examined: random carelessness, acquiescence, long-string responses, and mixed types. Data were simulated under 24 conditions with varying proportions of careless respondents (5% to 50%). Results showed that increasing careless response rates led to higher RMSE values, especially for slip parameters. Acquiescence responses caused the greatest bias, while long-string responses had the least effect. These findings highlight the vulnerability of s-GDINA parameter estimates to careless responding and underscore the need for further research on detection methods within cognitive diagnostic modeling frameworks.

Authors

• Tugay Kaçak, Trakya University

• Mehtap Aktaş, Trakya University