Main content
Separating content trait and varying thresholds in IRT approaches accounting for response styles
Date created: | Last Updated:
: DOI | ARK
Creating DOI. Please wait...
Category: Project
Description: IRT models with varying thresholds account for response styles by treating each threshold as a random effect. However, it usually constrains trait and thresholds to be unrelated: an assumption that is likely to be violated in empirical data. I propose a sum-to-zero constraint for varying threshold models that separates trait and response style effects and allows for the full estimation of the variance-covariance matrix. The new modeling approach is relevant for psychometric and applied researchers: it increases the informative value of varying threshold models while applying minimal a priori assumptions. R Code to estimate the new model with the package TAM is provided in this repository.
Start managing your projects on the OSF today.
Free and easy to use, the Open Science Framework supports the entire research lifecycle: planning, execution, reporting, archiving, and discovery.