Learning analytics (LA) has become central to personalized and data-driven education, but ethical concerns remain. While technical and policy-based approaches have aimed to safeguard privacy, less attention has been given to participatory consent practices, particularly about how student data is collected and shared. Longin, Briceno, and Poquet (2025) explored this issue by investigating how contextual factors and group discussions shape students’ willingness to share learning data. 

Drawing on contextual integrity theory and the wisdom of crowds literature, the authors hypothesized that:  

• Group discussion would alter initial individual decisions. 
• Willingness to share data would vary based on the type of data, its intended use, and the recipient.
• The content of group discussions would reflect these contextual shifts.

Methodology

Using an in-person experimental design, 60 university students discussed and rated eight different data-sharing vignettes. Each scenario varied by data type (process vs. outcome), recipient (government vs. company), and purpose (individual vs. collective benefit). Process data refers to how learners interact with educational technologies, such as click patterns or navigator behavior, while outcome data refers to results like grades or test scores. Regarding purpose, individual benefit means that data is used to support the learner personally, such as personalized feedback, whereas collective benefit involves using data to improve the learning environment or informing educational research. 

Participants rated acceptability individually, then as a group, and again individually post-discussion. Transcripts of the group discussions were analyzed using epistemic network analysis grounded in contextual integrity theory. 

Results

After group discussion, students became more cautious. Their acceptability ratings dropped significantly, with the lowest willingness to share when data were shared with governments for collective benefit. Sharing with private companies was seen as more acceptable than with governments. This is contrary to findings in health data studies. Furthermore, the data type (process vs. outcome) had minimal influence, and trust was a key factor. The higher trust in the recipient predicted higher acceptability ratings. 

Conclusion 

This study highlights that data-sharing decisions in education should move beyond static consent forms and embrace context-sensitive, participatory models. Involving learners in consent decisions and addressing trust and perceived utility are essential for future learning analytics frameworks. Educational technologies should thus move beyond technical solutions by integrating dynamic, learner-centered consent practices that transparently communicate data use and adapt to who collects the data and for what purpose. 

References

• Longin, L., Briceno, D., & Poquet, O. (2025). Data sharing in learning analytics: how context and group discussion influence the individual willingness to share. Humanities and Social Sciences Communications, 12(849). https://doi.org/10.1057/s41599-025-05175-2