Horizontal Equating of Science Test Forms Using Generalized Partial Credit Model (GPCM) in Secondary Education

Abstract

Ensuring fair comparisons between different test forms is a central concern in educational assessments. This study explores the horizontal equating of two versions of a science academic test using the Generalized Partial Credit Model, which is suitable for items scored across multiple categories. Student responses were analyzed using the Mean–Sigma method, involving shared anchor items to align the scales of the two test forms. The analysis revealed consistent item parameters and student ability estimates after transformation. A good model fit was observed based on residual and approximation measures, although further refinement may be needed due to limited index values in structural comparisons. The threshold distributions became more stable after equating, and graphical analyses confirmed that item characteristics and measurement information were preserved. This approach proved successful because it aligned the measurement scales without distorting original ability estimates. The findings support the development of fairer assessment systems that uphold validity, reliability, and comparability in science education.