Designing for Behavior Change: In Depth Analysis

We asked Amy Bucher, Ph.D., is Vice President of Behavior Change Design at Mad*Pow, to discuss some of the issues from her new book. Amy is psychologist who specializes in health behavior change with a special emphasis on technology.

In your new book, Engaged: Designing for Behavior Change, you discuss the psychology of behavior change. What is the psychology behind controlled versus autonomous motivation?

The concept of controlled versus autonomous motivation comes from self-determination theory, which posits that the source of a person’s motivation is a key determinant of its lasting impact on behavior. SDT considers motivational quality. You can think of the different types of motivational sources arranged along a continuum with extrinsic motivation at one end and intrinsic motivation at the other.

The controlled motivational qualities are extrinsic and introjected. They’re called “controlled” because the source is external to the person. You can think of them as being imposed from without. Extrinsic motivation is reward- or punishment-driven, while introjected motivation is the internalization of others’ expectations. An important thing about controlled sources of motivation is that they’re relatively vulnerable to challenge. I use the example sometimes of someone who’s made a dietary change in order to satisfy a family member. When that family member goes out of town, how likely is the person to stick to their diet?

Autonomous forms of motivation are more internal to the person, and therefore tend to withstand obstacles more easily. On the autonomous side of the spectrum, you have identified, integrated, and intrinsic motivational qualities. People have identified forms of motivation when a behavior supports other goals they care about, while integrated forms of motivation align with people’s sense of self and values. Having that north star of something personally meaningful behind a behavior means people tend to be more persistent with autonomous motivational sources.

The most autonomous form of motivation, intrinsic, is when something is inherently pleasurable. I don’t think we encounter it often when people are newly embarking on a behavior change—doing anything new can be scary and difficult—but many people eventually take pleasure in health behaviors like movement or preparing a nutritionally balanced meal. One of my goals in behavior change design is to craft a process that gets beginners to that place where they can start to actually enjoy their new behaviors.

What are ability blockers and how do you overcome them?

I use the term “ability blockers” to refer to the barriers that make a behavior more difficult or less likely for someone. One way to organize categories of blockers is using COM-B and the Behaviour Change Wheel; capability, opportunity, and motivation offer nice top-level headings that describe most of the common behavioral barriers people encounter. Something like low health literacy could be an ability blocker to medication adherence, as could difficulty getting to the pharmacy for a refill, inability to afford a medication, or disinterest in taking the treatment.

One of the reasons I like using COM-B to investigate and classify blockers is that it links potential solution sets, intervention functions, to each category. If you’ve identified a psychological capability issue like not understanding how to minimize side effects, the COM-B model suggests you might think about training or education as top-level solution categories. From there we to design to increasing levels of detail based on the tools available to us, the specific patient needs we’ve observed in our research, and any business considerations.

It’s important to consider that solving for one set of ability blockers has the potential to introduce another. One of the positive effects of many face-to-face services being made unavailable due to coronavirus is that, at least in the US, we’ve seen governments and payers quickly change the rules governing access to and reimbursement for telemedicine. That very quickly changed the ability blockers in play for behaviors around seeking and receiving care. For many health issues, it no longer matters if someone is able to get transportation to the doctor’s office. But we also have a new set of ability blockers to consider. What if someone doesn’t have an internet connection and a computer? What if they don’t understand the software? These sorts of considerations are why we emphasize doing primary research in context so that whatever you design to overcome blockers doesn’t create insurmountable obstacles for your specific user group.

Virtual coaching has proven effective in Diabetes Prevention Programs. How does this work and is a coach necessary for health behavior change?

I’ve seen two virtual coaching models used for Diabetes Prevention Programs. One brings the coaching entirely virtual, where a core program curriculum would be defined in advance and then delivered to participants in discrete modules based on their needs and progress. The other uses technology to connect a live coach to a patient who is likely receiving education and providing progress data to the system to facilitate the coaching conversations. Both models can be quite effective. Of course, the less an actual person is involved in delivering the coaching, the more scalable it is, so there’s often a business consideration going into the decisions about delivery method.

I don’t think a live coach is necessary for health behavior change, with heavy caveats. The first caveat is that any health behavior change program should be designed by subject matter experts so that it’s aligned with the advice a coaching professional would provide. Th Diabetes Prevention Program is an excellent example in that there’s a clinically established framework on which it’s based, and any DPP program is going to follow its basic outline. The second caveat is that an important ingredient in successful behavior change is personalization. A live coach personalizes the experience for people just by being human; her responses and advice are going to reflect the patient’s inputs. Technology can also deliver more or less personalized experiences. Research shows that when a technology experience is personalized to someone, it’s more effective, and it seems to be that more personalization is better.

Is social support an essential ingredient in health behavior change? Why?

One of people’s basic psychological needs is relatedness, or feeling connected to others, and it’s important that any health behavior change experience supports that need as much as possible. But the resulting experience does not necessarily have to incorporate social support in the form that we tend to think of it.

For any given behavior or life domain, people have different preferences about how they want to involve others. One segmentation model I worked on to understand workplace wellness found that a large number of people are “booster clubbers,” who enjoy group activity and cheering each other on, while an almost equally large number are “free agents” who strongly believe that their fitness should be separate from their work. Setting a free agent up with an accountability buddy would backfire because it goes against how that person wants to experience relatedness in that context. But an algorithm-driven activity tracker that offers feedback and tips on workouts could offer that same sense of accountability and support in a way that better resonates with a free agent’s preferences.

Offering people options about how to use social support or not can help meet different preferences. An example I love that shows both the benefits of social support in general and the importance of letting people choose how to use it comes from the Truth Initiative’s BecomeAnEx platform. In one study, they found that members of their online social community were significantly more likely than non-members to successfully quit smoking. This was true whether they actively posted in the community, or just lurked and read the conversations. Participation doesn’t have to look the same for everyone.

Many health app developers talk about behavioral economics. What are the pros and cons of this approach?

I think of behavior change designers as having a toolkit of approaches they can use depending on the problem at hand. Behavioral economics is an excellent tool to have in the behavior change toolkit. The issues with it arise when people want to use it as the tool to solve every challenge. Just like a hammer is the right choice for pounding nails but not for loosening lugnuts, behavioral economics is great for some behavior change projects and not for others.

In terms of pros: Behavioral economics is incredibly useful in designing for one time or once in a while behaviors, like making a doctor’s appointment, getting a vaccination, or choosing an appropriate health insurance plan. There are often elements of choice architecture to consider within a design project, and understanding the cognitive biases that are likely to influence people’s decisions helps us to anticipate their needs and design for better outcomes. I find myself considering behavioral economics principles whenever I’m designing a selection process. Behavioral economics is also very helpful in considering incentives; although I’m not generally a fan, they’re a fact of the health industry, and behavioral economics helps us design incentives schemes that are more likely to produce results.

The biggest con to the behavioral economics approach is that since it focuses on choice points, it’s not sufficient on its own for creating solutions that drive sustained behavior change. For something like medication adherence, you can nudge people to choose a particular medication or sign up for mail order pharmacy, but it’s much more difficult to create nudges in daily life that get the pill from bottle to mouth. Those sorts of ongoing actions are better addressed with tools like habit formation and by linking them to autonomous forms of motivation.

How does behavior change design adapt to those with low health and computer literacy?

Low health and computer literacy are both examples of ability blockers that might prevent someone from effectively engaging with a digital health intervention. The fact that both are pervasive speaks to the importance of usability and accessibility expertise as part of the product development team. Beyond making digital health as easy to use as possible for everyone, there are lots of other steps to address low health and computer literacy.

For low health literacy, using plain language is critical. We try to target a 4^th to 6^th grade level in our content as a best practice. When there’s medical terminology we need to include, we define it right away and give relatable examples. It’s also important to chunk out any content in a logical way so that people encounter foundational information first, and then can layer on additional knowledge. Pairing content with real world activity and feedback can also help people understand concepts experientially that are difficult to understand in writing. This is also an area where a health coach can be extremely helpful by helping someone with low health literacy during onboarding to a digital health intervention.

As for low computer literacy, one thing I try to keep in mind is that not all health interventions need to be digital. I worked on one project for people with diabetes in India; when we visited the areas where the intervention was to be deployed, we saw that internet access in the home was rare, and some of the older people, especially the women, did not know how to read. So we rethought how our intervention would be delivered. Ultimately we designed a program where a trained educator could interview the patient in the clinic, inputting their responses to a digital system. This would generate a printout that included a caregiver version for a literate family member. Rather than having patients input their self-monitoring of blood glucose (SMBG) and other biometric data into an app, they were able to text it using basic SMS, which was ubiquitously available. Even when we are working on an intervention that’s primarily digital, we try to think about the context in which people might use it and how we can design services and supports that improve the experience for people who have low computer literacy.

Amy Bucher, Ph.D., is Vice President of Behavior Change Design at Mad*Pow. Amy designs engaging and motivating solutions that help people achieve personal goals, especially related to health, wellness, learning, and financial well-being. Her research interests include motivational design, patient and user engagement, happiness, how social relationships influence health and well-being, and cross-cultural behavior change strategies. Prior to joining Mad*Pow, Amy worked with CVS Health as a Senior Strategist for their Digital Specialty Pharmacy, and with Johnson & Johnson Health and Wellness Solutions Group as Associate Director of Behavior Science. Amy received her A.B. magna cum laude in psychology from Harvard University, and her M.A. and Ph.D. in organizational psychology from the University of Michigan, Ann Arbor. Amy is the author of the Rosenfeld Media book Engaged: Designing for Behavior Change.