Designing Behaviour Change: A Behavioural Problem/Solution (BPS) Matrix

"[T]he proposed solution principles operationalise insights at the design level and complement the higher-level segmentation of behavioural design solutions."

There is an increasing recognition that behaviour change is essential to resolving many of the major challenges facing society. For example, although people are generally concerned about the environment, it is often difficult for them to act to protect it, due to ingrained habits. To overcome the attitude-behaviour gap, designers have an increasing interest in understanding how to design for behaviour change. Behavioural design focuses on creating positive, ethical behaviour change that balances individual and collective concerns through the design of artefacts, systems, and other interventions. It can be difficult to connect problem features, i.e., characteristics that determine a problem's nature and challenge, to potential solution principles, i.e., general approaches to problem resolution. The behavioural problem/solution (BPS) matrix proposed in this article is an effort to provide concrete guidance for defining behavioural problems and subsequently identifying relevant solution principles.

The researchers explain that, when discussing problem features in the behavioural design context, designers face a number of challenges, including: the multiplicity of theories of behaviour change (e.g., one scholar identified 83 such theories, dealing with 1,659 often ambiguously defined constructs); the complex interactions between behaviour, user self-image, interaction and experience, and perceived intent; and the fact that many of these theories are highly specific in domain (e.g., Health Belief Model) or in scope, such as focusing on specific cognitive or social mechanisms (e.g., Social Norms Theory). There is a need to operationalise these general strategies in concrete principles relevant to designers.

As outlined here, problem features can be conceptualised in terms of an interaction between: (i) the required change demand (hereafter, CD) - that is, the degree of difficulty associated with a planned behaviour change - and the degree of behavioural constraint (hereafter, BC), which is the degree of difficulty associated with a planned intervention. With regard to the latter, in the behavioural design context, designers must consider: environmental and social constraints on a desired behaviour, degree of freedom within the project to influence these systems, and constraints on the degree to which the objects and tools used in a specific behaviour change intervention (e.g., a mobile phone) can be altered.

Per the researchers, few works explicitly link problem features and solution principles. Four existing models are examined here that provide a strucutre for understanding behavioural design solutions: Fogg's (2009) Behaviour Model (FBM), Michie et al.'s (2011) COM-B model, Bhamra et al.'s (2011) Design for Sustainable Behaviour (DfSB) Strategies, and Tromp et al.s (2011) Influence framework.

With this groundwork in place, the paper explains how the researchers analysed 218 behavioural design interventions drawn from 139 cases across design domains and foci. Based on this analysis, they: provide an overview of the intervention data, evaluate the proposed problem features with respect to current models, and develop their proposed BPS matrix. Highlights:

• Table 4 on page 71 of the paper features a list of 23 solution principles relevant to designers that represents a synthesis of recommendations from across domains and foci. Sample solution principle: Tell or show what others do - that is, highlight the supportive behaviours of peers to incite the desire for social conformity (social norms). Among the most widely used principles in the interventions considered were: providing types of feedback (e.g., via physical visits, emails, ambient lighting on a watch, or digital counters in the home), using triggers (e.g., pop-up signage explaining required action or digital prompts via email or app), and making it salient (e.g., adding coloured floor markings, highlighting areas of a digital document, or translating behaviour into monetary terms).
• Linking back to their conceptualisation of problem features (what they deem the "behavioural problem feature framework" - see above), the anslysis showed that, among the cases they considered: those with a high CD/high BC (49 interventions) were 82% successful; those with a high CD/low BC (33 interventions) were 64% successful; those with a low CD/high BC (79 interventions) were 76% successful; and those with a low CD/low BC (57 interventions) were 75% successful. However, the paper identifies a number of potential systemic challenges in the reporting and evidencing of behavioural design interventions. For instance, approximately 51% of the interventions they considered used cohort controlled trials, which "falls significantly short of the standards associated with typical health interventions..., and aligns with prior findings of relatively weak evidence for behaviour change interventions..."
• After examining the interaction between their behavioural problem feature framework and solutions included in the four existing models named above (FBM, COM-B, DfSB, and Influence model), the researchers propose the BPS matrix. (See Figure 7 on page 77.) In the matrix, the use of the various solution principles logically corresponds to the different requirements imposed in each quadrant. For example, in the high CD-high BC quadrant, the five top principles can all be achieved with means independent of the behaviour. All of these deal with the delivery of person-specific information needed to support the high CD, which can be achieved via multiple means both within the environment and via personal technologies due to the high BC. The matrix also helps us understand that, when the desired CD is high, it is beneficial to implement combinations of solution principles in order to successfully tackle the various challenges (e.g., lack of motivation, lack of skills) that people may face.

This work has two main implications for design practice, in the researchers' estimation:

1. The behavioural problem feature framework provides a means for unpacking behavioural challenges by enabling designers to identify high/low CD and BC at various levels of detail. For example, observations, interviews, and other user studies can be used to identify the degree of novelty (have users done something similar in the past?), the scope (how elaborate is the behaviour in question, and to what degree is it connected with other behaviours?), and the frequency in daily life. These questions allow a design team to segment their population with respect to their different features and also to identify principles common across areas, such as presenting feedback in various forms.
2. The proposed solution principles and combinations provide designers with ranked suggestions for how to resolve behavioural challenges across domains and foci. For example, a design team can start the intervention development stage by identifying the top-ranked principles and combinations in order to support their own ideation or to move directly into structured testing.

In conclusion, the researchers argue that the insights they share here "extend both theory and practice surrounding problem-solution mapping in behavioural design, and form a foundation for further theory development and synthesis in this area."