Language Test Performance and Test-Takers’ Degree of Confidence
Anna Mouti, Aristotle University of Thessaloniki & University of Thessaly (Greece)
George Ypsilandis, Aristotle University of Thessaloniki (Greece)
Abstract
This study examines whether test-takers’ language test performance and degree of confidence are related, proving that they may influence each other. Participants were asked to complete the tests also by marking their confidence in every item on a 100mm bar. Our statistical analysis concentrated on the number of correct responses and degree of confidence. Findings reveal that accuracy and confidence scores positively correlate significantly. A possible application of confidence marking may be found in the design of computer language tests.It is our view that confidence marking should be integrated into the computer language tests because it may increase test-takers’ engagement and activate their metacognitive strategies leading to higher language performance. Furthermore it may provide a more fair and precise language scores and better discrimination, when it comes to a computer adaptive language test. Confidence should be considered as a variable which could provide more information about the test-taking process and the test-takers themselves.