Back to work
Research · Education Design · HCI

Teaching Interaction Designers
to Code

A 3.5-year research-through-design project exploring how to make creative coding accessible, empowering, and genuinely student-centered for non-computing design students.

University of Sydney · 2018–2022 My role: Researcher & Thematic Analyst Published: Frontiers in Computer Science Collaborators: K. Grace, L. Bray, A. Elton-Pym
3.5
Years of research
1,250+
Students across 3 cohorts
60+
In-depth interviews
3
Design principles developed

The problem with teaching designers to code

Interaction designers increasingly need programming skills, but traditional CS pedagogy wasn’t designed for them. Design students learn differently: they’re motivated by visual output, creative autonomy, and making things that help people. A lecture-and-exam model left many feeling like they’didn’t belong’ in technical spaces.

At the University of Sydney, the Interaction Design unit DECO1012 served around 400 students per year - mostly first-year undergraduates and postgraduate students from design backgrounds. The challenge: how do you design a programming curriculum that meets design students where they are, while building real technical competence?

Project at a glance
Platform
Creative Coding Challenges (CCCs) - a custom learning platform using p5.js
Approach
Research-through-design with iterative cycles and summative meta-analysis
My contribution
Led thematic meta-analysis across 60+ student interviews; co-authored published findings
Context
Ran across face-to-face, blended, and fully remote (COVID-19) delivery modes

Three iterative cycles of design & research

We used a research-through-design methodology: building, deploying, studying, and rebuilding over three distinct cycles. Each cycle generated student interview data that fed directly into the next design iteration - a genuine feedback loop between research and practice.

1
2018–2019 · Face-to-face
Discovery & Prototyping
Built the initial CCC platform as a set of visual p5.js challenges. Required challenges formed the core curriculum; optional extension challenges gave confident students room to stretch. This cycle established the platform’s architecture and surfaced the first round of student attitudes toward open-ended creative coding.
2
2019–2020 · Blended then remote
Blended Learning & COVID Pivot
Refined the platform based on cycle 1 findings. A pilot pair-programming exercise was introduced mid-year. COVID-19 arrived mid-cycle, forcing a rapid pivot to fully online delivery - inadvertently giving us a rare comparison between face-to-face and remote open-ended learning in the same cohort.
3
2020–2021 · Remote-first
Pivot to Protégé
Introduced the RSDL (Recursive Student-Directed Learning) task - asking students to design a challenge that would have helped them earlier in the course. Expanded pair programming as a formal component. This cycle generated the most polarizing but also most insightful data, particularly around student autonomy and metacognition.

Making sense of 60+ voices

My core contribution was the thematic meta-analysis: coding and interpreting over 60 student interviews conducted across all three research cycles. Using an ecological triangulation approach, I worked to surface the patterns that weren’t visible in any single cycle alone.

This meant sitting with a lot of complexity and contradiction. Some students found the open-ended format deeply empowering. Others found it anxiety-inducing. The challenge wasn’t to reconcile these, but to understand why both were true - and what it revealed about the conditions under which open-ended learning works.

“I found it was a turning point in my learning where I actually could freely explore.”

Student participant, Cycle 1

The thematic analysis produced seven broad themes that together told a nuanced story about student attitudes toward open-ended programming. Rather than a simple “did it work?”, the answer was: it worked for some students, undermined others, and the difference had less to do with prior coding skill than with metacognitive confidence.

Seven themes, one central tension

The meta-analysis revealed a consistent tension at the heart of open-ended blended learning: the same pedagogical choices that felt empowering to some students felt anxiety-inducing to others. The difference wasn’t about ability - it was about metacognitive confidence and knowing where you stand.

🧠
Learning as a Skill
Students who thrived treated learning itself as something to get better at. The platform’s reflective prompts supported metacognitive growth - but only for those already comfortable with ambiguity.
💻
Learner Technology
p5.js’s immediate visual feedback was genuinely motivating. But in remote contexts, the physical/digital disconnect left many students feeling isolated from the platform they were supposed to learn through.
🔑
Learner Autonomy
Choice created anxiety for students who didn’t trust their own judgment. Without clear external markers of progress, freedom felt less like opportunity and more like uncertainty.
🤝
Social Learning
Peer learning was among the most positively received interventions - particularly pair programming. Social connection helped address isolation, and having someone struggle alongside you normalised difficulty.
🧭
Learning Support
Students craved reliable, authoritative guidance - especially one-on-one tutor access. Open-ended tasks without clear scaffolding amplified anxiety rather than curiosity.
💡
Content Complexity
Difficulty itself wasn’t the problem - unexpected difficulty was. Students needed to know when they were struggling appropriately, not just that they were struggling.

What we learned for future HCI educators

From the research, we distilled three actionable design principles for educators building open-ended programming curricula for design students:

1
Design open-endedness within activities, not between them
Embedding open-endedness inside a required task (e.g., “create a composition using nested loops”) is far more effective than offering a choice of which tasks to do. The latter creates assessment anxiety; the former creates creative space. Students with lower coding confidence found choice anxiety-inducing, while open-ended tasks embedded in required work gave them permission to explore without fear of missing something testable.
2
Even a single other student makes remote learning better
Pair programming was among the most positively received interventions in the entire project. Even in challenging remote conditions, working alongside one other person reduced fear of failure, provided social validation, and transformed individual struggle into shared problem-solving. For design students, who are trained to collaborate, this is especially apt.
3
Student-led teaching is human-centered design
Asking students to design a challenge for their earlier selves was one of the most provocative and insightful interventions we tried. It reframed programming as something to design for others - a natural motivation for design students. The most effective learning happened when students could apply their human-centered instincts to the act of teaching itself.

What changed, and what it means

This research contributed to an ongoing conversation in HCI education about how to design programming pedagogy that genuinely serves non-computing students. The CCCs platform reached over 1,250 students across three years, and the findings have since informed how the Interaction Design unit approaches open-ended assessment design.

For me personally, this project sharpened something I care about deeply: the intersection of research and teaching practice. Good research should change how you teach. Good teaching should raise questions worth researching. This project was both.

Publication details
Title
An Open-Ended Blended Approach to Teaching Interaction Designers to Code
Authors
Grace K, Klaassens B, Bray L, Elton-Pym A
Journal
Frontiers in Computer Science, Volume 4, 2022
DOI
10.3389/fcomp.2022.813889
Want to read the full paper?

The published article is open access and available on Frontiers in Computer Science.

🥚 You found me

5 things you didn’t ask,
but here we are anyway.