Application of Pseudo-Equivalent Groups (PEGs) Linking for Testing Mode Adjustment

• In Event: 29.082 - Roundtable Session 9
  In Roundtable Session: 29.082-14 - Linking, Equating, and Multidimensional Item Response Theory

Abstract

In this study we investigated how different ways of adjusting testing mode differences affect scoring conversions and proficiency level cut scores. We compared three methods (linear and curvilinear pseudo-equivalent groups linking and multiple linear regression) to adjust testing mode differences on a large scale K-12 assessment. The overall results suggested that the scoring conversions for the three adjustment methods were very similar in the middle of the score range where the cut score exists, but the conversions differed somewhat at the top and bottom of the score ranges. The pseudo-equivalent groups methods had slightly larger deviations, standard errors, and root mean squared errors than the regression approach.

Authors

• Hyeonjoo J. Oh, ETS

• Junhui Liu, Educational Testing Service

• Hanwook (Henry) Yoo, Educational Testing Service