Hybrid factor reduction as a data mining technique for large scale educational assessment

• Autor(a) principal: Lopes, Simone Mágna Menezes Carneiro
• Data de Publicação: 2022
• Outros Autores: Freire, José Carlos da Silveira
• Tipo de documento: Artigo
• Idioma: por
• Título da fonte: Revista de Gestão e Avaliação Educacional
• Texto Completo: http://periodicos.ufsm.br/regae/article/view/70944
• topic: Avaliação educacional; Avaliação em larga escala. IDEB. Qualidade do ensino.; Análise fatorial

Large-scale assessment has aroused the interest of researchers, governments and civil society. Such interest lies in the fact that its results have guided public policies on management and financing of basic education. However, the data produced by the Brazilian Basic Education Evaluation System are underused by school management as a diagnostic and learning promotion tool. In this perspective, we sought to identify factors that influenced the Ideb index in the 2017 assessment for the 9th grade of elementary education among schools in the state network of Tocantins. To this end, an exploratory multivariate factor analysis was performed to identify the factors that are related to a higher or lower performance in the Ideb assessment. Given the variation in the scales of the Saeb questionnaires and the Ideb scores that vary from 0 to 10, there was the need to dichotomize the scales. Therefore, the multivariate factor analysis was based both on the factor extraction by means of the principal components method from Pearson's correlation and on that obtained by means of the tetrachoric correlation. It was concluded that the application of the method will help the manager to understand the indexes raised and draw a perspective with specific points where improvement is needed, besides making it possible to extract important information so that the management can intervene in a focused way in the application of resources and guide public policies.