Developing Grade Calibration Methods Based on Linear Double-Linking

Abstract

The objectives of this study were to develop and compare three methods for grade
calibration: one-section linking; two–section additive linking; and two–section interactive linking. Method quality was assessed by correlating modified grades with the ordinary National Educational Test (O-NET); a discrepancy index was computed between modified grades and criterion grades. The secondary data used was derived from the Central University Admission System in the academic year 2007.

Results indicated that:
1. the correlations between O-NET and modified grades based on one-section linking,
two–section additive linking, and two–section interactive linking were all higher than the
correlation between O-NET and raw grades at the .05 level.

2. with regard to the discrepancy index between modified grades and criterion grades: the
root mean square difference, and the mean absolute difference, from one-section linking,
two–section additive linking, and two–section interactive linking were significantly lower than the corresponding discrepancy index computed between raw grades and criterion grades at the .05 level.