Benchmarking de avaliação dos cursos de licenciatura em educação física dos institutos federais: uma aplicação da data envelopment analysis (DEA)

Abstract

This thesis investigates the efficiency of on-campus Bachelor’s degree programs in Physical Education offered by Federal Institutes (IF) in Brazil, with the central research question being: how can we assess this efficiency using Data Envelopment Analysis (DEA) and develop an interactive benchmarking dashboard that identifies best practices and contributes to the improvement of academic management? The study is based on two hypotheses: (H1) identifying the most relevant SINAES indicators will allow for measuring the efficiency of the programs; and (H2) applying DEA will enable the construction of an interactive benchmarking dashboard to inform academic management. The overall objective is to evaluate the efficiency of these courses using DEA modeling, proposing a dashboard that highlights the most efficient indicators. This is applied, exploratory, and quantitative-qualitative research, guided by the quadripolar methodology. Data were collected through a structured questionnaire based on the INEP guidelines, administered via Google Forms, semi-structured interviews with coordinators, and a document analysis of the current Course Program Plans (CPPs) of eight DMUs. The pre test yielded a Cronbach’s alpha greater than 0.90, with high internal consistency in the dimensions of Didactic-Pedagogical Organization, Faculty, and Infrastructure. Hierarchical cluster analysis identified the following as priority outputs for DEA modeling: Course Objectives (OUT1), Supervised Curricular Internship (OUT2), Supervised Internship – Basic Education Network (OUT3) and Supervised Internship – Theory and Practice (OUT4), Core Faculty (OUT5), Course Committee Activities (OUT6), and Coordinator’s Workspace (OUT7). The output-oriented BCC model (BCC-O), implemented in OSDEA-GUI-v0.2.jar, revealed that DMU 7 (IFMG – Bambuí Campus) was the only fully efficient program according to the Pareto-Koopmans criterion (θ = 1.0, zero slacks), serving as a reference peer group for all other DMUs in four of the seven indicators, notably OUT2, OUT3, OUT4, and OUT6. The other DMUs exhibited inefficiencies due to an excess of inputs not converted into maximum evaluative quality, with scores ranging from 0.600 to 0.964. The qualitative analysis of the interviews and current PPCs, using content analysis, identified the following as key best practices of DMU 7: data-driven management with documented internal evaluation cycles (2023–2025), which increased the alignment of teaching plans with the PPC from 58% to 92%; the Articulated Curricular Component (CCA), with student teachers’ immersion in Basic Education schools starting from the first semester; mandatory participation in class councils; a requirement for an original Intervention Project; and systematic integration between the NDE, CPA, and PPC, updated in accordance with the 2024 DCNs. Qualitative-quantitative triangulation confirmed that the full efficiency of DMU 7 stems from verifiable and replicable management decisions that are independent of size, budget, or location. As a technical product of the thesis, an interactive benchmarking dashboard was developed in Power BI that measures the relative efficiency of each course and makes the practices of the benchmark DMUs—which are suitable for institutional transfer readily available, thereby contributing to the improvement of teacher training in Brazilian public education.