Differential Weighting of Items to Improve University Admission Test Validity

Abstract

This paper gives an evaluation of different ways to increase university admission test criterion-related validity, by differentially weighting test items. We compared four methods of weighting multiple-choice items of the Basic Skills and Knowledge Examination (EXHCOBA): (1) punishing incorrect responses by a constant factor, (2) weighting incorrect responses, considering the levels of error, (3) weighting correct responses, considering the item’s difficulty, based on the Classic Measurement Theory, and (4) weighting correct responses, considering the item’s difficulty, based on the Item Response Theory. Results show that none of these methods increased the instrument’s predictive validity, although they did improve its concurrent validity. It was concluded that it is appropriate to score the test by simply adding up correct responses.

Keywords

• Ponderación de ítems
• calificación ponderada
• métodos de evaluación
• validez predictiva
• exámenes de admisión.