Item Ranking Comparison between GRA and IRT Rasch Model

Abstract

In this paper, the samples are randomly selected from a CSL (Chinese as second language) computerized test. Follow by performing utilization of Grey Relational Analysis (GRA) to calibrate and analysis the rank of each item difficulty. The major objective of this paper is to compare the rank difference between method of GRA under limited samples and Rasch model with sufficient data available in Item Response Theory. All data was collected from a CSL computerized test conducted overseas in Philippine during 19(superscript th) to 24(superscript th) of October 2009. There were 269 examinees participated in this test. Our study aimed to use GRA on decision making under uncertainty and with insufficient or limited data available for analysis and to prove its effectiveness. This analyzing procedure will contribute and re-productively applied into other areas, such as ”minimum sample requested for pre-testing” during the test item assembling in the futures.