Supporting Self-Experimentation for Behavior Change

Although people’s health and wellbeing are tightly linked to their behaviors, failure to sustain desired behaviors is common. Driven by the importance of behavior change and the struggle to achieve it, many behavior-change technologies are designed by experts. As an alternative and complementary approach for supporting more personalized and precise behavior change, my work aimed to help individuals create their own behavior change plans. To this end, two levels of supports were explored with a series of user studies: Support for creating personalized behavioral plans and Support for creating context-aware just-in-time interventions.

Support for behavioral plans

With intention to provide theoretically grounded support for users’ self-creation of behavioral plans, I focused on a wide range of behavior change techniques proposed by behavioral scientists. For professional practitioners designing behavior change interventions, behavior change techniques are considered as the "active ingredients". As such, I sought to simplify these techniques so that individuals could build their own interventions that are based on “what works” in general.

In exploring how to connect users’ behavioral plan creation and practices from behavioral science, I defined three primary components of a behavioral plan: 1) a goal; 2) a consciously chosen behavior-change technique(s) that is personalized by and for the individual; and 3) self-monitoring. To guide users’ creation of behavioral plans including these components, a structured protocol was established as below:

The protocol had five steps: 1) Choosing a behavior to attempt to change (the target behavior), 2) Setting a goal, 3) Generating ideas for attainment of the goal by applying behavior-change techniques, 4) Formulating a final plan, consisting of one or more complementary behavior-change techniques, and 5) Devising self-tracking measures to determine if the goal was accomplished.

One of the design challenges was to present users a large number of behavior change techniques so that it inspired creativity but was not overwhelming to them. Based on previous work, a “meta-model” consisting of four components was developed, and the behavior change techniques were grouped by these four components. Within the first session of creation, users receive one technique from each group. In following sessions, users are provided with the meta-model and then provided with some structure to help them self-diagnose, which area they most likely need to work on. Based on that self-diagnosis, the users are then provided additional techniques specifically for that group.

The meta-model states, “Behavior occurs when ability, motivation, opportunity, and a trigger align.” Behavior change techniques can be classified according to which of the four they mainly target.

Elaborating the protocol further, I made a low-fidelity prototype, powerpoint presentation. It delivers concepts with text, suggestive pictures, and recorded narratives, allowing users' easy understanding.

1 / 13
🔊 At first, you bring out all issues you have before choosing one of them to work on. For example, you may have made a New Years Resolution in the last January...Please choose one that you really want to fix.
2 / 13
🔊 Now, you think of what you aim to do or not to do, in order to reduce or eliminate the issue. We emphasize you set behavioral goals. Let’s first learn about what it is...Can you choose one that you’d like to work toward most?
3 / 13
🔊 Once goals are generated, they should be checked if they are SMART goals...Now, you check your goal is SMART, and transform it into a SMART goal. Write down every idea that come to the mind. Then make a final formulation.
4 / 13
🔊 Though there are many others, four techniques will be given today. They will be introduced one by one. After you learn about one, you will work on how to apply it for you, and then move to the next one.
5 / 13
🔊 This technique suggests that there is often a specific time and place when you engage in a behavior. It is a bit different if you are trying to do more vs. do less of a behavior. When you are trying to do more of a behavior,...
6 / 13
🔊 The general goal here is to create a nudge or “trigger” to do the desired behavior. This technique differs slightly if you are trying to increase or decrease a behavior. If you are trying to increase a behavior...
7 / 13
🔊 The goal of this technique is to make sure you know exactly what you are going to do. Often, when people try to change their behaviors, they get “stuck” because they do not think about the small “mini-steps”...
8 / 13
🔊 One of the strategies you can use to help increase your motivation is to reward yourself with things you like, that aren’t counter to your personal goals. Perhaps it could be going out with friends, spending time with family,...
9 / 13
🔊 OK, you generated ideas with each technique. Please walk through the ideas, and integrate ideas from each technique to formulate a final plan...Please picture a day or part of it when you carry out the plan successfully...
10 / 13
🔊 Keeping track of how well you are doing at keeping your goals can be a very valuable tool. There are a variety of ways to keep track of things that can be helpful for you when you are trying to achieve a goal.
11 / 13
🔊 We would like you to do some self-tracking using one of two possible tracking styles, one that we are calling the structured questions approach and the second we call unstructured journaling. We will describe both approaches...
12 / 13
🔊 First, ‘Structured Questions Approach’... It involves creating very specific questions that you think will help you to understand not only if you met your goal or not but factors that might impact your goal and plan...
13 / 13
🔊 This approach is used if you aren’t quite sure what you should be tracking or if you just don’t like all of the structure of the other technique. For this one, the goal is to set aside times throughout the day...

This prototype could eventually be used to design interactive digital materials allowing users to work alone. In the future, I envision several aids for users' easy creation of effective plans, including function to relieve users' burden in completing detailed descriptions for making actionable plans, merging plans into existing daily routines, and self-tracking.

Support for just-in-time interventions

The second approach was to support self-creation and testing of just-in-time (JIT) interventions using context-aware computing. I have been working on a toolkit, GaLLaG, that provides integrated hardware and software for rapid prototyping of sensor-based responsive home environments. The toolkit is considered easy to learn and versatile to realize various application ideas for behavior change. For further information on the toolkit, please visit this page.

The protocol for creating behavioral plans (described above) is extended for users' generation of JIT intervention ideas. It presents basic concepts on the technology and intervention examples related to the behavior change techniques. For instance, the below example is given related to the technique, 'Script critical actions':

Imagine there is a person who wants to have more energetic morning by waking on time and doing some jogging. He made this idea, “At 7 in the morning, I wake up with peaceful music. I get out of bed, and hear a song coming from the bathroom. I go to the bathroom. After a while, when I enter into the closet to change clothes, weather news get played, which help me choose what to wear.” He thinks it may help him go through the steps and begin the target activity.

I conducted a 7-week evaluation, which tested these two approaches for improving sleep relative to a sleep education control that mimics the sort of unstructured approach that is common of those engaging in Quantified Self activities. Fundamentally, all three conditions can be considered as users’ self-experimentation, although the control group only received sleep education and a minimum structure for planning. Results indicate that all three conditions appeared to improve sleep quality over 7 weeks, with the high likelihood that our two interventions resulted in a small to moderate improvement in sleep quality relative to the control.


Self-Experimentation for Behavior Change: Design and Formative Evaluation of Two Approaches

Jisoo Lee, Erin Walker, Winslow Burleson, Matthew Kay, Matthew Buman, and Eric B. Hekler
CHI 2017 [PDF]

Video preview

Understanding Users’ Creation of Behavior Change Plans with Theory- Based Support

Jisoo Lee, Erin Walker, Winslow Burleson, and Eric B. Hekler
CHI 2015 [PDF]

Exploring Users’ Creation of Personalized Behavioral Plans

Jisoo Lee, Erin Walker, Winslow Burleson, and Eric B. Hekler
UbiComp 2014 [PDF]