Before taking this class, I thought that behavior change was achieved through changing people’s values and beliefs. In this class, I learned that in reality, attitudes tend to follow behaviors, not lead them. Therefore, the best thing isn’t to change someone’s mind and then hope that their habits shift, but to instead create the positive conditions for a new behavior, and let their mindset catch up later. This was a really pivotal learning while working on a behavior change project in this class this quarter.
I loved my team and genuinely enjoyed the project we built together, SleepPea. Our team meshed together very well as we rarely had conflicts, and any disagreements we had were very easily and quickly resolved.
While I loved my group and working on our project SleepPea, I would have enjoyed less deadlines and busy work. Every week of this class felt like a sprint. I would love some way to teach the same objectives but to make the workload more intentional and meaningful. I believe the same learning objectives could be achieved with a more consolidated workload. What worked well for me in this class was the hands-on, project-based structure of the class; I enjoyed working on a team. Having a real problem to solve and getting to choose that problem as well as our solution was also particularly meaningful.
After this class, Figma is a tool I will absolutely use again. I enjoy doing group work in the Figma tool. Building our prototype in React Native was also a highlight.
The biggest surprise was how involved the baseline and intervention studies turned out to be. While I learned a lot, I expected a lighter research component on the order of conducting isolated interviews. Running week-long studies with real participants was meaningful, but recruiting dedicated participants during the busy winter quarter was a big challenge. I expected the prototyping work to be more difficult than it was; in reality, once we had a clear direction, and due to the vibe coding lecture, building out our prototype was a lot better than I expected.
One specific problem we ran into was figuring out how to demo the social Sleep Pod aspect of our app without having a full community of real users and real sleep data to pull from. We resolved this by hardcoding sample data and pre-made pods, and adding a skip-to-morning feature so that anyone demoing the app could experience the full loop without waiting 24 hours.
What remains unresolved for me is the question of the effectiveness of our final prototype. We validated that people responded positively to the assumptions behind our app, but we did not have the opportunity to conduct a study with our actual final prototype over a sustained period of time. I think that kind of study could be really useful in showing us if our assumption tests results translated to our final prototype.
One social tension this project surfaced for me was whether social pressure is always positive. Designing a system like SleepPea that uses your friends to nudge your behavior can go wrong if the nudges end up causing anxiety or feelings of shame.
This project connected directly to my interests in product design and human-computer interaction. As a computer science major our work on the final prototype definitely cultivated skills I will bring into future classes and future roles.
In regards to the ethics of our group’s project, the mechanisms Sleep Pea uses to change behavior are social accountability, streak visibility, and a shared garden that reflects the pod’s collective consistency. What makes these acceptable nudges rather than manipulation is that they are fully transparent and opt-in. Users choose their pod, set their own goals, and can leave at any time.
However, our app could be ethically questionable for users prone to anxiety or who have friends that are toxic/push unhealthy sleep habits. A pod dynamic that feels motivating for one person could feel shaming for another, and that asymmetry is something a “production” version of our app would need to address and account for.
SleepPea respects user privacy in its current form because the only data shared is bedtime goals and lock-in status, and only within a small group the user has chosen themselves. However, this could break down if the app were to expand into using real Health data for sleep duration tracking, health integrations, or any form of data aggregation beyond the Pod. It would be important that the user’s data was stored in a way that was ethical.
Overall, my thinking has evolved significantly on what behavior change actually requires. I came into this class believing that if people understood the consequences of their choices, then they would make better choice, and I am leaving with a much more grounded understanding of what actually drives behavior change.
Next time when I am faced with a similar situation and I am on a team working on a behavior change initiative, I’ll think back to what I learned in this class, and make sure that the solution that I develop is grounded in user research, user testing, and facts about behavior change.
