1 Pontifícia Universidade Católica de Minas Gerais (BRAZIL)
2 University of Castilla-La Mancha (SPAIN)
3 University of Alicante (SPAIN)
About this paper:
Appears in: INTED2022 Proceedings
Publication year: 2022
Pages: 5796-5804
ISBN: 978-84-09-37758-9
ISSN: 2340-1079
doi: 10.21125/inted.2022.1484
Conference name: 16th International Technology, Education and Development Conference
Dates: 7-8 March, 2022
Location: Online Conference
Large-Scale Exams (LSE) presuppose teaching based on competences, are based on reference matrices, are structured in item banks, and are based on statistical data treatments. Several works investigate the LSE impacts on educational systems around the world. Although it is essential to keep a critical view on the meanings and consequences of an Educational System embodied in LSE, the ENEM (Brazilian High School Examination) is considered as a mechanism that expands the possibilities of access to Brazilian Higher Education.

In 2016, ENEM surpassed 9 million candidates, losing numerically only to the Chinese Exam. Since 2009, ENEM has been composed of an essay and 4 (four) tests prepared by area of knowledge (Natural Sciences, Human Sciences, Languages and Mathematics), each one of them comprising 45 (forty five) items. Each knowledge area test responds to a reference matrix, materialized in 5 general competences and 30 specific skills. ENEM uses the statistical treatment of IRT (Item Response Theory) in the three-parameter model (difficulty, discrimination, and random hit), which essentially depends on the quality and calibration of items.

The IRT allows equalization between different editions of the test, as it is based on a statistic that allows the construction of scales in which subjects and items are located. By placing someone on a scale, it is intended to detect his pedagogical coherence in that area of knowledge, that is, he is expected to get right questions that are below his proficiency level, and that, on the contrary, he is expected to miss questions above it.

Thus, subjects who get the same number of correct questions may have different grades, with the one who demonstrates greater pedagogical coherence having better performance. The items are randomly distributed throughout the test, applied in four versions (yellow, blue, pink and gray), however, it is useful to understand the complexity test sequence, in order to better face it from the easiest to the most difficult one. The academic quality analysis of the tests and their items is not disclosed, but all the ENEM editions and their microdata are available. Based on these data, it is presented the equivalence in the complexity randomness tests sequence in the four versions generated from the same test. It is also presented, in the clipping of the Physics content, studies on the ENEM complexity, considering Bloom and Webb's Taxonomies as references.
Innovation, technology, research projects, Brazilian Higher Education.