Analysis of C-Tests with the Equidistance and the Dispersion Models

Alabdallah, Ziad Ahmad; Ismail, Ismail Abdulwahhab; Mutar, Hamid Khalaf; Mohammed, Aisha; Alghazali, Tawfeeq; Mansoor, Mohammad Salman; Anber, Akram Ali; Mohammed Ali, Yusra; Ghaleb, Maabreh Hatem; Georgievna, Goncharova Velichka

doi:10.22034/ijlt.2023.403640.1264

Analysis of C-Tests with the Equidistance and the Dispersion Models

Document Type : Special Issue

Authors

Ziad Ahmad Alabdallah ¹

Ismail Abdulwahhab Ismail ²

Hamid khalaf Mutar ³

Aisha Mohammed ⁴

Tawfeeq Alghazali ⁵

Mohammad Salman Mansoor ⁶

Akram Ali Anber ⁷

Yusra Mohammed Ali ⁸

Maabreh Hatem Ghaleb ⁹

Goncharova Velichka Georgievna ⁹

¹ College of Veterinary Medicine, Al-Furat University, Deirez-Zor, Syria

² English Department, Al-Noor University College, Nineveh, Iraq

³ English Language and Literature Department, Al-Mustaqbal University College, Babylon, Iraq

⁴ College of Pharmacy / Al-Farahidi University, Baghdad , Iraq

⁵ College of Media, Department of Journalism/ The Islamic University in Najaf, Iraq

⁶ Department of Media / Mazaya University College/ Iraq

⁷ Al-Esraa University College, Baghdad, Iraq

⁸ Department of Medical Laboratory Techniques, Al-Zahrawi University College, Karbala, Iraq

⁹ Peoples Friendship University of Russia,Moscow, Russia

10.22034/ijlt.2023.403640.1264

Abstract

C-tests are commonly used as measures of second language reading comprehension and general language proficiency. Analysis of C-tests with item response theory models is problematic due to the interdependent structure of C-test items or gaps. An approach to facilitate item response theory (IRT) analysis of C-tests involves treating each passage as a polytomous super-item. This approach facilitates the application of polyomous IRT models for ordered response data to the C-test passages. Usually, Andrich’s rating scale model or Masters (1982) partial credit model are used to analyse the data. In this study, we aim to employ two alternative modelling techniques, namely, the equidistant model (Andrich, 1982) and the dispersion model (Rost, 1988) to C-test data. Our findings showed that the C-test data have a good fit to the both models and can be used for psychometric analysis of C-test passages. Information criteria showed that the dispersion model has a better fit compared to the equidistance model.

Keywords

C-Test

dispersion model

equidistance model

local item dependence

polytomous Rasch models