Rasch analysis and differential item functioning of English language anxiety scale (ELAS) across sex in Egyptian context

Abstract

Background: English language anxiety (ELA) is a prevalent phenomenon in language education. It is one of the most commonly investigated non-linguistic variables in studies of language learning. Accordingly, numerous studies have paid great attention to the factors leading to FLA. Methods: As instruments of ELA have not been subjected to a rigorous test of item response theory (IRT), this study conducted such analysis using the Rasch rating scale model. ELAS scale developed using classical testing theory (CTT), which consists of 32 items measuring four sub-scales (listening, speaking, reading and writing anxiety), was analysed using IRT. WINSTEPS software and SPSS version 26 were used to examine the psychometric properties, sex differential item functioning (DIF) and Rasch analysis of the ELAS in the Egyptian context. A total of 604 participants were recruited for data collection. Results: The main findings indicated that the scale yielded a good approximation of Rasch assumptions and appears to be a valid and reliable tool. DIF was computed using the Mantel–Haenszel (MH) method and Welch’s t-test, which indicated that no sex bias was detected for any item of the ELAS. Conclusions: This paper presents a promising ELA instrument characterized by adequate validity, reliability and objectivity in addition to potential for precise use in comparison between males and females because it is invariant across sex.