ABSTRACT When item response models are applied in equating, the
assumption of local independence is required. Polytomous item response
theory (IRT) models can be considered as alternatives to dichotomous
models if the assumption is violated. This study compares the performance
of the dichotomous IRT model and a combination of dichotomous and
polytomous IRT models in equating two forms of the Examination for the
Certificate of Proficiency in English (ECPE). Traditional equating
methods are used as a baseline for comparison. The results reveal that a
combination of the three-parameter logistic model and the generalized
partial credit model yield results similar to the traditional equating
functions for the listening section, and the three-parameter logistic model
performs better in the GCVR section.