STATISTICAL ANALYSIS OF ROMANIAN AS A FOREIGN LANGUAGE PRETEST AND EXAM DATA
STATISTICAL ANALYSIS OF ROMANIAN AS A FOREIGN LANGUAGE PRETEST AND EXAM DATA

Summary/Abstract: Statistical Analysis of Romanian as a Foreign Language Pretest and Exam Data. In the European Union, linguistic policies have gained an essential role within the European policy in its entirety. In this context, in 2014 we started an experiment with the aim of creating an instrument for analysing the quality of tests of Romanian as a foreign language. The instrument thus created contains a succession of statistics tests including factor structure, internal consistency and convergent validity evidence, IRT and ROC analysis for validation. Our creation reached the point where it successfully fulfills the task of Reading the Mind in the process of learning Romanian, also applicable to other languages. The present article shows the manner in which the authors apply the IRT models. It will demonstrate that the IRT models can help improve test scoring and facilitate the development of superior quality tests in Reading Mind applied to the process of studying foreign languages. In the authors' perspective, the present work has a significant contribution to the applicability and usability of IRT models in educational testing, as the presented results show. Furthermore, the paper brings arguments in favour of utilizing IRT and ROC curve models in common, which leads to the conclusion that the achieved results are important in the process of improving the precision of marking and the manner in which the tests are administered through the adaptive use of discriminating elements pertaining to the testing procedure, which saves time, space and impacts on the number of assigned tests.

• Details
• Contents