Exploring Student Feedback Needs and Design Opportunities in Data Storytelling Education

Paper Title

Exploring Student Feedback Needs and Design Opportunities in Data Storytelling Education

Publication Info

• Topic area: AI-assisted feedback in data storytelling education
• Keywords: Data storytelling, AI feedback, educational design, participatory design, formative feedback, narrative visualization, scaffolding, student learning, creative workflows, intelligent tutoring systems

Background and Problem

• Problem / challenge: Data storytelling education requires integrating analytical, design, and narrative skills, but instructors face challenges in providing timely, personalized feedback due to high enrollment and resource constraints. Existing tools focus on automating storytelling rather than teaching the process.
• Significance: Effective feedback mechanisms are critical for developing data storytelling skills, which are increasingly recognized as essential for data science professionals. Addressing this gap can improve learning outcomes and support creative workflows.
• Motivation and related work: Prior research highlights the importance of structured feedback in education, but existing frameworks focus on linear tasks and offer limited guidance for creative, multi-stage processes like data storytelling. This paper builds on literature in educational psychology, creativity, and HCI to explore feedback design in complex workflows.

Solution

• Proposed approach: Story Studio, an AI-assisted narrative storytelling platform designed to provide adaptive feedback tailored to the stages of the data storytelling process.
• Novelty:
  1. Empirical insights into learners’ and educators’ feedback needs during data storytelling workflows.
  2. Evaluation of user-preferred feedback modes (on-demand, automatic, process, outcome) for AI-assisted tools.
  3. Design implications for balancing accountability, creativity, and seamlessness in feedback systems.
• Procedure and key techniques:
  • Conducted a two-phase participatory design study:
    1. Phase 1: Observations (N=8) and interviews (N=6) to identify feedback needs and challenges.
    2. Phase 2: Two design workshops (N=8/10) to co-design and evaluate feedback strategies for Story Studio.
  • Developed and iteratively refined prototypes (V0, V1, V2) based on user feedback.
  • Evaluated six feedback modes (on-demand, automatic, skip, address, process, outcome) through surveys (N=7).

Results

• Concrete findings:
  • On-demand feedback rated as more effective (M=4.57) and useful (M=4.57), while automatic feedback was perceived as more persuasive (M=4.00).
  • Process feedback rated highest for effectiveness (M=4.86), while outcome feedback was more persuasive (M=4.00).
  • Addressing feedback was preferred over skipping, with participants valuing enforced revision for reflection but expressing concerns about limiting creative flexibility.
• Advantage over baselines:
  • Story Studio provides adaptive, stage-sensitive feedback, addressing gaps in existing tools that focus on automating storytelling rather than teaching the process.
• Experiments / evaluation:
  • Observations, interviews, and workshops informed prototype development.
  • Surveys assessed perceptions of feedback modes across effectiveness, persuasiveness, and usefulness.
  • Feedback modes evaluated included timing (on-demand vs. automatic), requirement (address vs. skip), and locus (process vs. outcome).
• Limitations and future work:
  • Limited generalizability due to single-institution study and small sample size (N=7 for surveys).
  • Future work will include multi-institution studies, A/B testing, and evaluation of learning outcomes.
  • Plans to explore data storytelling with immersive media and resource-constrained systems.

Summary

This study investigates the design of AI-mediated feedback for data storytelling education, addressing challenges in providing personalized, stage-sensitive feedback. Through participatory design and iterative prototype development, the authors identify user preferences for feedback timing, requirement, and locus, emphasizing the need to balance accountability with creative flexibility. Results show that on-demand feedback is perceived as more effective, while automatic feedback is more persuasive. Process feedback supports reflection-in-action, while outcome feedback aids retrospective reflection. Design implications include adaptive feedback requirements, avoiding over-scheduling, and clarifying roles of AI and instructors. Future work will focus on broader evaluations and learning outcomes.