Análise da autoavaliação institucional pela perspectiva da teoria da resposta ao item: uma aplicação na UFSM

• Ano de defesa: 2018
• Autor(a) principal: Kronbauer, Elisete
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Santa Maria; Brasil; Administração Pública; UFSM; Programa de Pós-Graduação em Gestão de Organizações Públicas; Centro de Ciências Sociais e Humanas
• Palavras-chave em Português: Avaliação institucional; Autoavaliação; Ensino superior; Teoria da resposta ao item; Institutional evaluation; Self-evaluation; Higher education; Item response theory; CNPQ::CIENCIAS SOCIAIS APLICADAS::ADMINISTRACAO::ADMINISTRACAO PUBLICA
• Link de acesso: http://repositorio.ufsm.br/handle/1/16201

Institutional self-ealuation is an important component from Brazilian National Higher Education Assessment System (SINAES), constituting in an instrument to subsidize the continuous efforts to improve the quality, efficiency and effectiveness of university management. It is an opportunity for all segments from university community to self-reflect about the institution's performance. In this context, this study aimed to analyze the instruments of institutional self-evaluation from Universidade Federal de Santa Maria (UFSM), applied during the year of 2016, which were nominated as General Questions, Professor Segment and Technical-Administrative in Education Segment. For this, the study was based on the Classical Theory of Tests (TCT) in order to analyze the reliability, the internal consistency and the unidimensionality of the construct; and in the Item Response Theory (IRT), from the application of 2 Parameters Logistic Model (ML2), to verify the bisserial correlation and to estimate the parameters of difficulty and discrimination by the items. Results showed that the set of questionnaire items presents internal consistency; however, according to the factorial analysis, they are not unidimensional, that is, they do not measure the latent institutional performance, thus the item parameters of the questionnaires could not be estimated through ML2. In this way, it was necessary to perform partial analyzes by segment and by axis. Some axes presented unidimensionality and it could be measured the bisserial correlation and the parameters from the items. Partial analyzes revealed that all measured items had a good bisserial correlation (>0.3) and good discriminatory power (a). In general, the work contributed to demonstrate the validity and usefulness of ITR in processes of higher education institutional evaluation.