The Study of the Extent of Bias by Persian Native Raters towards Non-Iranian Persian Language Learner's Writings Based on Many-Facet Rasch Measurement Model

The present study examined the errors of Persian native speakers in rating the essays of Persian language learners in the framework of a many-facet Rasch measurement model. To do so, Persian native speakers were requested to rate 56 writings of 28 male and female Persian language learners in advanced level. The applicants, studying in Humanities and Engineering fields, were learning Persian as a second language in Persian Language Center at Imam Khomeini International University in winter 2014. The raters rated the learners’ writings using a researcher-made based on 5-point analytic rating scale. The collected data were analyzed using Facets software. The results of Facets analyses indicated that raters showed more bias toward both topics in writings. The results also, revealed that raters were more biased toward coherence. Findings represented significant differences in raters’ bias towards learners’ writings. The findings would shed light on raters training to reduce the errors they may commit.