The Revision and Development of the HELP Placement Test

Abstract

This project intends to reform HELP’s placement test by observing its strengths and weaknesses. This project, contrary to the classical Item Response Theory (IRT), employed the Rasch model (Bond & Fox, 2007) for measurement. The Rasch model incorporates an algorithm that expresses the probabilistic expectations of item and person performances. The following tasks were performed subsequently: (1) Analyze the Michigan EPT test results from HELP students, to inform whether this can be used as an effective placement exam at HELP, (2) Evaluate and reform the reading and writing tests, (3) Develop a new speaking test, and (4) Revise written test administration and scoring procedures. The procedure and results of test reform is to be discussed, and recommendations for new placement tests will also follow.