This project was created as a part of my HCI Foundations class at Georgia Tech. The theme for the class was "wellbeing" and we given the liberty to choose our own problem statement. Adhering to the theme of the class, our team decided to work on the problem of stress-eating among college students.
College is a big transition for many students as it presents them with a volatile environment and ever-changing academic, social, and financial pressures. The college environment is often highly stressful and students either do not have helpful resources at their disposal or are not aware of them. In such situations, one of the maladaptive strategies that college students use to cope with their environment is stress-eating.
First, to understand the problem space, we conducted literature reviews and comparative analysis. Then, to understand the scope of stress as a trigger among our target users, we conducted surveys. Finally, to develop a contextual understanding of how stress impacts everyday eating habits and test out potential solutions, we conducted a week-long diary study.
To understand the behavior of stress-eating at large and among college students, we looked at existing literature. We read published studies, journal articles, and excerpts from books on stress.
In our exploration of the problem space, we spoke to college students and two registered dieticians at Georgia Tech to better understand how stress affects the eating habits of students. Through our conversations, we were able to define the target population and their goals.
Before coming up with a solution, we decided to look at tools that exist in our defined problem space. While we did not find a tool that addressed stress eating among college students, we found applications that address emotional eating through mindfulness.
Noom uses research and statistics to motivate changes in their user's behavior. They also use periodical journal entries that prompt users to consciously think about their eating habits.
Am I Hungry intervenes before the stress eating episode is about to happen. The app takes the user through a series of questions every time the user wants to engage in emotional eating.
In the Moment app uses gamification and a series of questions to guide users through spontaneous eating urges. Using positive affirmations, it helps steer users away from boredom eating, emotional eating, and stress eating.
We decided to do surveys because, first, surveys would provide us with clear, quantitative data about students’ snacking habits and coping strategies, allowing us to identify trends and patterns clearly. Second, due to the relative ease of distributing and analyzing surveys, we would be able to collect data from a larger number of target users, giving us a set of data that is reliable.
Since our problem space specifically involved college students, our primary focus was to send out the survey to as many students as possible. The survey was distributed through various communication channels within educational institutions like Georgia Tech, Georgetown University and University of Minnesota. In total, we received 74 survey responses.
Once my teammates created the survey, I collected feedback about it from my peers and iterated on the survey design as required. I actively participated in the distribution of the survey. I also reviewed the raw data to extract the findings presented above.
Creating and improving on this survey underscored the importance of the iterative design process for me. If our team had not sought feedback on the survey, we would have missed the logical inconsistencies or flawed conditional logic that was present in the survey. Further, while I have a background in research, trying to keep the bigger picture in mind during data analysis is always a challenge. I found that the more time that I spent looking at the data, the better I could disambiguate it and form a coherent story through our findings.
To better inform our design implications, we wanted to identify which interventions would help our target users overcome the urge to stress eat. Since we wanted highly contextual data that varies from day-to-day, we decided to conduct a diary study.
The diary study was designed in a survey format on Qualtrics. Each participant received an anonymized link to their personal survey, which they could access using the same link everyday. Overall, we were able to collect seven days worth of data from six participants, five days worth of data from two participants, and one participant did not end up filling out the diary study at all.
I created the diary study design from the ground up. Since my peers had not done diary studies before, I relied on existing literature about diary studies in the realm of stress eating and emotional eating. Once I had the study design in place, I and Suyash created the screening survey and the diary study survey on Qualtrics. We also created an instructional document that explained the goals and procedures of the diary study to our participants. After the diary study, we created qualitative and quantitative metrics on the basis of which we analyzed the data and reached the findings presented above.
Creating a diary study, as I had never done so before, was a new and enriching experience for me. Diary studies are highly contextual and thus, the researcher has minimal control on the study conditions. Given this fact, I had to keep the constraints of my participants in mind. Since my participants were college students, I had to make sure that they did not have to spend too much time filling out their diary each day. Further, I paid special attention to the phrasing and tone of the survey questions as I did not want my participants to view their eating habits and dietary lifestyle in a critical light.
The data analysis portion of the diary study was definitely one of the most challenging and rewarding parts of this project. I spent a lot of time thinking about our research goals, our survey questions and how to create metrics that answer our research questions with the most clarity. While reaching and formulating the findings was challenging, seeing them form the foundation of the design of our application was a proud and satisfying moment for me.
We started this phase of the project by establishing functional and non-functional design requirements. Then we used brainwriting to develop divergent solutions. Finally, we established clear criterion to determine which design to move forward with.
Before the brainwriting session we looked over our requirement gathering and everyone from the team thought of different ideas. With some basic preparation, the team got together and went through a brainwriting exercise. We used a brainwriting template to guide us through the process. The process followed four rounds for our team. Each round was five minutes long and each team member either continued the idea that was previously mentioned or came up with a new idea of their own. Once the round was completed, everyone passed their sheet on to the team member. All rounds were coordinated by a moderator. We were able to generate 25 unique ideas using the brainwriting exercise. Following the exercise, we engaged in a group discussion and clubbed the ideas that were complementary to each other; we removed ideas that did not meet our design requirements.
StressAR is a mobile application(iOS/Android) which uses Augmented Reality to help users curb stress eating. As soon as the application is opened, the user is asked to scan the room. While scanning, the application gives callout type buttons for objects of interest. These objects will be items and activities that the user can interact with to distract them from their stressor. This application has a shallow information architecture. Due to this, the user can access the to-do task in just two steps. There is no data collected about users and there is an integrated UI button that they can see while scanning to call a loved one if they so choose.
CompanionAI is a mobile application that interfaces with a wearable device. The wearable device can track physiological data such as heart-rate and skin conductance. Once the device detects anomalous spikes in the user's physiological data, it sends a signal to the CompanionAI app on the mobile device which triggers the voice assistant on the phone to speak up and ask if the user is stressed. Based on how the user responds, the voice assistant replies and suggests activities or calls a loved one.
Once we narrowed down on our top two designs, we considered the following three criterion while choosing between them:
Based on the above 3 questions, we created comparison tables between our top two designs. The comparison table for functional requirements is presented below.
Users see exactly what they would expect to see when they first open a new application; a login screen. Our login screens utilizes subtle contrast and include softer edges for user input fields to make the content appealing. Above, on the right, is the part of the onboarding that allows users to connect their wearable to our application. This allows the application to receive real-time physiological data about the users stress and prompt them to engage in stress-alleviation activities,
Once the user logs into the application, they will be taken to the Activity screen, which is perhaps the most important feature of CompanionAI. Here, they can "browse" through stress alleviation activities, choose to perform the activities, and add them to their favorites. Once the user selects an activity, they can choose to do it with a friend or a group of friends, which is in turn enabled by the Community feature of the application. Finally, users can create and add their own stress-alleviation activities.
The community feature fulfills the functional requirement that the system must facilitate interaction with loved ones. The Community feature is also accessible from the bottom navigation bar and allows users to search and add new friends, send positive affirmations to friends, and chat with friends.
Another important feature of CompanionAI is the Search feature. The Search feature is accessible from the bottom navigation bar and allows users to search for friends and activities.
The user profile feature is another salient feature within CompanionAI that users can access from the bottom navigation bar. Here, users can view their favorite activities and access application settings.
Another important feature of our application is the notification feature. CompanionAI offers strategic system-initiated interventions to suggest activities to users and does by using iOS notifications like the ones displayed below. Timely notifications by the application fulfills the functional requirement that the system must initiate interaction with the user.
We conducted usability tests with 4 college students. The goal of the usability tests was to determine the nature and scope of usability issues within the application.
For the evaluation, we conducted four one-on-one discount evaluations with college students. In each session, the facilitator began by providing a brief overview of the problem statement and the primary aim of our application. The first 15 minutes of the session were spent observing the user interact with the prototype using a think-aloud protocol. The facilitator would then ask the designated user to perform a series of benchmark tasks to test the viability of the application. A few of the benchmark tasks are listed below:
At the end of the evaluation sessions, participants were asked to rate our application on the system usability scale.
I prepared the usability testing benchmark tests and questions. I also created a usability testing script to ensure that the tests were conducted the same way across all my team members.
Performing usability tests with our target users was equal parts exciting and nerve-racking. I was excited to present our prototype to our target users but also apprehensive about whether we had created something that would fulfill the needs of our users. In that light, getting a high system usability score was immensely satisfying. However, I did not lose sight of the usability problems in the application and the improvements suggested by our users. Due to time constraints, we were unable to iterate on the prototype but I cannot underestimate the value of the design recommendations presented above and am eager to implement them into this project in the future.
Since we are designing an application that aims to solve the problem of stress eating, it may be concerning that our solution does not specifically focus on food or it’s nutritional value. This was an intentional choice because through our conversations with dietitians at Georgia Tech, we found that redirecting people to healthier food choices does not change the fact that they use eating as a way to cope with stress. Being aware of this information gave us adequate reason to not focus on food, but instead on the triggers and behaviors behind the need to reach for food when people feel stressed.
Tying into the point above, it can also be argued that our application does not directly address stress eating. This was also an informed decision made by the team because we realized that to break the habit of stress eating, we would have to intervene between the exact point at which the stressor is encountered and the maladaptive coping strategy. While our application can intervene with stress eating; it could also intervene with other, similarly functioning, maladaptive coping mechanisms like drinking, smoking, and other types of substance abuse. While we recognize that our application can be used to reduce a variety of maladaptive coping behaviors, we are confident in our decision because our solution still addresses the problem of stress eating.
© Aditi Bhatnagar 2020. Created with Webflow. Made with love and powered by coffee.