CS247B Final Reflection

Before taking CS247B, I took a intro class to psychology and immediately became interested in human behavior. Because the psychology facts we learned were mostly intuitive and rational, I thought it would relatively easy for people to change their behavior (given that they wanted to change). Going into the class, I thought the main challenge of designing for behavior change was going to be the “design” part: prototyping a UI and determining the user interaction.

Turns out – it is much more difficult to change a person’s behavior than I expected. CS247B showed me how irrational and unpredictable people could be. My team focused on the problem of reducing online shopping – our goal was to redirect users from online browsing by recommending substitute activities to do instead. We had a brilliant insight behind this solution – people who shopped compulsively typically do so to fulfill some emotional need, such as relieving anxiety or avoiding negative emotions. Our hypothesis was that if users were self-aware of their emotional need and how it could be fulfilled alternatively, then they would likely attempt a substitute activity. For example, if they were aware that they were stressed and needed to distract themselves, then they would be more willing to watch TV to distract themselves instead of shopping. Hence, our solution was an app that recommended activities based on the user’s identified emotion and emotional need. All of this made logical sense. Yet, from our studies, we found that most people continued to shop anyways – which was not rational. Granted, our study’s sample size was too small, so we couldn’t conclude any statistical significance, however I was still surprised at these initial results. This is wildly different from debugging computer programs, where the internal logic was always robust. If I had more time, I would love to dive deeper into this (unresolved) issue and conduct more studies.

This leads to the second thing I found surprising – studies are more difficult to conduct that I expected. Psychology studies often appear simple and elegant, but in practice it can be messy – there are many factors that could affect people’s behavior, and it was difficult to separate them completely from the study. For example, a simple data collection method (e.g., filling out a spreadsheet) could easily alter users’ behavior and confound the phenomenon we wanted to study.

Despite the surprises, there were also many things that went the way I expected from an HCI class. Notably, rapid prototyping and synthesis were important skills that I learned from this class. Before taking this class, I did research in graphics/HCI but was not familiar with UI prototyping tools, so I’m glad I got to learn Figma. 

Now, after completing CS247B, I have a different perspective on human behavior. First of all, people are very unpredictable – debugging humans is very different from debugging a machine. It requires a lot of insight, empathy, planning, and testing. Second, the main challenge of design for behavior change is the “human” part, not the “design” part. In that sense, design for behavior change is more pertinent to psychology than CS. Third, prioritizing and scoping is very important in carving out a good and do-able project, not unlike research. Fourth, while taking this class, I also tried applying some of the behavior change tactics to myself as well – starting small (tiny habits) turned out to be very effective for me. I learned that the fear of change is often scarier than the actual change. Finally, this class experience made me understand why people have trouble changing – it helped me develop a more compassionate view of the world.

Next time, when I design for behavior change, I will first read lots of existing literature pertaining to the issue. Specifically, I will try to gain a solid background of psychology in the area before attempting to design a solution. Then, I will test any assumptions and run conclusive studies before prototyping the UI or the specific user interaction.

 

Ethical considerations:

 

  • Nudging and manipulation: it’s important that users know how they will be manipulated. Informed consent is what makes a nudge acceptable, as opposed to manipulative. A nudge is manipulative if it is against user’s will or used in a way that does not align with the user’s original intention.
  • Privacy: the best arrangement for the user is to not give data to us (remote servers) in the first place. Ideally, they will store all their app usage and info locally. To avoid future misuse, the app terms of use should contain a clause that forbids the company from selling user data at any point in the future.
  • Interface Design & Design Justice: our team designed an app layout which includes common elements shared with other apps (e.g., bottom menu tabs) so it’s easy to pick up for the average mobile user. Due to time constraints and our focus on the Spencer persona, we did not design many accessible elements or universal, non first-world activities. Given more time, we would love to make improvements to enhance the universality of our app.
  • Design for Well-Being: I learned that there are many different theories for measuring well-being, such as desire satisfaction and hedonism. In future design efforts, I will consider these theories at earlier stages of my design so I can better incorporate its value throughout the system. I will also inform the user what well-being theory the system relies on so the user can better understand the system’s core values and know what to expect.
Avatar

About the author