Um método baseado na Teoria da Resposta ao Item para avaliação e feedback automático no contexto de educação digital

Abstract

The assessment of students' knowledge is a common task in the field of education, either to assess learning or to mediate the selection of candidates in entrance exams. Exams that employ an assessment based on multiple choice questions, in which the student marks items during the exam time and then receives feedback on his grade, do not offer the student significant contributions to understanding his performance. The general objective of this study is to provide, for both students and teachers, formative feedback from the estimation of the ability of students and difficulty of the items, through a statistical technique called Item Response Theory (IRT). This technique produces a model of data on student performance based on the analysis of the responses collected in the exams. Two experiments were conducted using data from an exam administered in a public high school in Amazonas. The results obtained indicate a valuable amount of information that allows analyzing the relationship between the various items and the estimated abilities. It is possible to classify students on a hability scale so that the student himself can locate his position in a discipline and simulate which topics can be studied to obtain greater hability. From the teacher's point of view, it is possible to analyze which topics were more difficult for the class and analyze whether an elaborated item is taught. Otherwise, the teacher can redo the item or change the difficulty level. The results provided significant information that allows the elaboration of formative feedback to provide the examinees and evaluators with the necessary guidelines to investigate difficulties and contribute to a better performance in the teaching-learning process.