Comparing Linear Equating and Equipercentile Equating Methods Using Random Groups Design

In this study, raw scores which taken from 9th grade 2009 BBS Form C of Social Sciences were equated to 2009 BBS Form A of Social Sciences with linear equating and three different (unsmoothed, presmoothed and postsmoothed) equipercentile equating methods. The random equating error of different equating methods was compared by Root Mean Squared Difference (RMSD) index and mean of bootstrap standard errors (MBSE). The results indicated that there was a linear relationship between equivalent scores of Form A and raw score of Form C, and Form C was easier than Form A along the score scale. It was seen that results based different equipercentile equating method were consistent with each other, there was a nonlinear relationship between them and test difficulty varied along the score scale. Finally, although the least MBSE and RMSD coefficients were got from linear equating because of nonlinear relationship between raw scores and equivalent scores, concluded that the most appropriate equating method was postsmoothed equipercentile equating which had relatively the lowest random equating error among equipercentile equating methods. 2013 IOJES. All rights reserved