Differentiation among Equated Scores for General Abilities Test Forms Using Classical Equating Methods with Common-Item Non-Equivalent Groups Design

Abstract

The study aims to differentiate among equated scores for the General Abilities Test forms within common item non-equivalent group design using classical equating methods. Two new forms are equated to a reference form for the Verbal ability test and the Quantitative ability test. The study uses five classical equating methods: Linear equating methods (Tucker, Levine, and Braun-Holland) and Nonlinear equating methods (Equipercentile and Smoother Equipercentile). The results show differences among the descriptive statistics and frequency distributions for test forms. After equating, results show that the equating score from equipercentile methods at the majority of test scores, and different equating scores are obtained at the two extreme test scores (low and high scores). In addition, equipercentile method show smaller standard error of equating.