Data and code for: Investigating the Washback Effect on Generative AI

This repository holds the files used in the paper Investigating the Washback Effect on Generative AI Strategies: Insights from Engineering Students.The dataset comes from a mixed-methods study of how first- and second-year engineering students use generative AI (e.g. ChatGPT) as a study aid in relation to course assessment and perceived relevance. It contains anonymised survey responses from 336 students in mandatory foundation courses (mathematics, physics, programming, chemistry), including Likert-scale measures of procedural vs. conceptual GenAI use, perceived assessment criteria (procedural vs. conceptual), perceived course relevance, and basic background variables. These data were used to estimate a psychometric measurement model (EFA/CFA) and a structural equation model testing how assessment orientation and relevance predict different types of GenAI use, supplemented by 11 qualitative interviews (not included in this repository).The Rmd file used for data analysis, figure and table production is available in the folder "Data and Code" along with the raw survey data in the file "complete.csv"The file Survey.pdf shows the exact questionnaire used for data collectionAdditionally, the folder "Summary report" contains an HTML file generated by Quarto as well as supplementary files to run it. The HTML shows all relevant graphs and statistics for the paper.Various licenses apply to the files in this item:complete.csv: CC-BY 4.0Washback_data_analysis.Rmd: MIT License Copyright © 2025 Technical University of DenmarkAll files in the folder "Summary report": MIT License Copyright © 2025 Technical University of DenmarkSurvey.pdf: CC-BY 4.0readme.txt: CC-BY 4.0