DATA of Validating a Reduced Technology Acceptance Model 3 in Engineering Education: A Five-Year Study of a Constructionist 3D Learning Microworld
Description
This study assessed whether a reduced Technology Acceptance Model 3 provides a valid and stable framework for evaluating students’ acceptance of ARPAID across five academic cohorts (2019–2024) and three engineering disciplines: Aerospace, Electrical, and Electronic Engineering. ARPAID is a constructionist 3D microworld, inspired by Seymour Papert’s theory, that supports exploratory learning through the functional, and graphical representation of mechanical assemblies. A five-year dataset was analysed using a four-dimensions instrument comprising perceived usefulness, perceived ease of use, enjoyment, and behavioural intention. Factorial validity, reliability, measurement invariance, and coherent structural relationships supported the instrument’s consistency across disciplines and cohorts. Acceptance was consistently favourable. Aerospace and Electrical Engineering students reported greater perceived usefulness and behavioural intention than Electronic Engineering students. The 2019/2020 cohort showed the strongest acceptance, possibly reflecting pandemic-related digital learning conditions. The findings support the reduced model as a robust framework for evaluating constructionist 3D learning technologies in engineering education.
Files
Steps to reproduce
The analyses reported in this study were conducted in R using RStudio 2025.05.1 Build 513 and documented through a Quarto workflow. The anonymised dataset, 5anos.xlsx, contains the participant-level responses used in all descriptive, psychometric, measurement-invariance, group-comparison, and structural equation analyses. The rendered document, TAM3Reduced.pdf, provides a complete record of the analytical workflow, including data preparation, model specifications, statistical procedures, R code, tables, figures, and numerical outputs. Questionnaire responses were treated as ordinal Likert-type variables, and the primary factor and structural models were estimated using WLSMV, with MLR employed as a sensitivity analysis where indicated. A fixed random seed (set.seed(1234)) was used to support the reproducibility of procedures involving random-number generation. The software environment, required R packages, analytical decisions, and output-generation procedures are described within the rendered report. However, the executable Quarto source document (TAM3Reduced.qmd) is not deposited in a public repository. Researchers wishing to reproduce, verify, or extend the analyses may request the source document and any necessary supporting materials from the corresponding author. Materials will be made available upon reasonable request, subject to applicable ethical, institutional, authorship, and data-protection requirements. Because the PDF is a rendered analytical record rather than an executable source file, exact automated reproduction requires the original Quarto document, an appropriate R and Quarto installation, the required package dependencies, and a compatible TeX distribution for PDF generation. Requests should specify the intended academic or research use of the materials and may be directed to the corresponding author at fjfraf@unileon.es.
Institutions
- Universidad de LeónCastille and León, León