top of page

Invited Speakers -

Prof. John Quan ZHANG

Prof. John Quan ZHANG

Professor @

Faculty of Foreign Studies,
University of Jiaxing

Jizxing, China

Topic of Presentation:

Winsteps and GITEST: Different but Alike
An Application of Rasch-based Software to Test Equating in China Mainland

Speaker Bio: 

Got his B.A of English major in 1982, M.A. of applied linguistics in 1989 and Ph.D of language testing in 1993. He was invited by ETS as senior visiting scholar to pursue further studies (2002-2002) and invited by Professor Lyle F. Bachman as senior research scholar (2006-2009).
With over 30 years’ of experience in language testing, Prof. Zhang has been working as Dean and chair professor in several universities in China and officially got involved in test equating in China. In 2013, he was invited as an adjunct professor to supervise Ph.D students conducting research in language testing by City University of Macau. From 2016 on,he was invited to work as Ph.D supervisor at DEEP University, Wisconsin, USA. Prof. Zhang is the editors in chief of two international journals. He has published books and many articles and good at BILOG, PARSCALE, GITEST and Winsteps. 

Abstract: 

This presentation falls into two parts: an introduction to the status quo and prospect regarding Rasch Model in China Mainland and test equating via Rasch model. The concept of “equating” discussed herein refers to linking of separate test forms through common items so that scores derived from the tests which were administered separately to different test takers on different occasions, after conversion (using Rasch-based software), will be comparable on the same scale (Hambleton & Swaminathan, 1985). The relevant procedures are illustrated in the parallel test forms based on a set of minimum yet representative data collected from 40 Chinese students of non-English major of a university in Guangdong Province, China. Two types of Rasch-based software: GITEST and Winsteps are used to process the data. The results are compared and observed to be highly correlated (cor=0.99). Two important findings are to be shared with testing counterparts: though Winsteps is widely used for objective measurement of various purposes, it is NOT applied for test equating purpose. The present author is the pioneer in applying Winsteps to test equating in China Mainland. And the results obtained from Winsteps are compared with those by GITEST are highly correlated. The author is thus convinced that GITEST, albeit developed in Basic in 1986, is still an effective tool to perform the same task.
bottom of page