Using group model building to frame the commercial determinants of dietary behaviour in adolescence – proposed methods for online system mapping workshops

Complex systems thinking has gained popularity among public health researchers [1, 2]. This is partly because traditional evidence models have been criticised for being ill-suited to account for the “real-life” contexts in which health decisions are made and the broader systems in which policies and interventions take place [3, 4]. Therefore, a “complex systems approach” has been advocated as a way of understanding the many influential factors and different sectors, and their interactions that affect complex health issues like diet and obesity [5]. Incorporating systems thinking methods into policymaking has the potential to increase policy efficacy and better anticipate unintended consequences of well-intentioned practices from other contexts [6]. However, the applications of systems science methodologies in public health research are relatively underdeveloped [7]. 

The commercial determinants of obesity are strategies used by the food and beverage industry to produce, promote and increase the sales of their products, sometimes at the expense of public health, especially where their promotion strategies particularly target children and adolescents with foods high in salt, fats, sugar and energy density through integrated marketing techniques [5]. Incorporating systems thinking can actively foster dialogue between diverse stakeholders and enlighten the mechanisms by which commercial drivers influence core dietary behaviour and can affect health a long time into the future.

Group Model Building (GMB) is a facilitated modelling method based on system dynamics (SD) [8] whereby system actors (i.e. stakeholders from civil society, academia, policymaking and business) share their views and ideas about a problem to create a collective understanding of a complex system [9]. The aim of GMB is to uncover the causal structure of complex systems driving its outcomes [8]. The causal structure of the system is presented in the form of a Causal Loop Diagram (CLD), or system map, which provides a visual representation of the most influential factors, their interconnections, and feedback structures in the system that are thought to be responsible for creating a problem, or making an existing problem worse. It also specifically identifies what factors are included and excluded in the system of interest (i.e. its boundaries), and enables the identification of leverage points at which system change can potentially be initiated.

Existing literature on conducting GMB workshops mainly focus in in-person facilitation [10, 11], which has been successful in fostering discussion among stakeholders from multiple sectors and with different beliefs and values on a topic [12]. However, with technological advancements in recent years, and the impact of COVID-19 pandemic, other methods for developing CLDs of systems are gaining attention, including digital collaborative tools, online discussion environments and meeting platforms [13]. Conducting GMB workshops online may have several advantages, such as reduced costs and time required to run the workshops, inclusion of a broader range of stakeholders regardless of their physical location and breaking down GMB activities into several short online sessions rather than whole day workshops, which would allow for reflective processes between sessions. Despite the aforementioned advantages, we only found two examples of fully online GMB workshops in the literature [14, 15], which were held online due to COVID-19 restrictions.

We ran a series of Group Model Building (GMB) workshops to create a system map of the commercial determinants of dietary behaviour associated with obesity and, due to the COVID-19 global pandemic, we assessed adapting and running the GMB workshops fully online. Specifically, we ran GMB workshops with adolescents (16–18 years) living in the Southwest of England and aimed to create a system map of the of the commercial determinants of dietary behaviour associated with obesity, and to encourage them to identify and formulate relevant policy ideas to tackle these commercial influences and to improve their food choice environment. In a second workshop, we gathered policymakers and public health practitioners to discuss the system map created by the adolescents and explore the potential for the implementation of some of the policy ideas proposed at a local authority level.

The detailed account of the results from the workshops will be presented elsewhere, but for illustrative purposes, we will present the final map here. The aim of this paper is to report on our experiences of adapting face-to-face GMB processes to an online format, with a particular focus on our methodological approach. We use our experiences to produce recommendations on how online GMB can be delivered.

Figure 1 shows the process we followed to plan, adapt, pilot, and deliver the GMB workshops with adolescents, policymakers and public health practitioners. In a first phase (Fig. 1. I. Planning and Adapting), because of COVID-19 restrictions on running face-to-face activities, we had to develop an adapted version of the GMB project for an online format. In a second phase (II. Adapting and Piloting), we conducted a process mapping exercise to develop a shared understanding of the overall GMB process, discuss the number of workshops, and select which and how many people would be involved in each one. The activities were based on standard scripts that outline the elements to include in a GMB workshop [16, 17]. We identified the scripts to include in the GMB workshop and then developed the online agenda and the timings for each activity. To avoid “screen fatigue” we divided the GMB workshop with adolescents into 3 shorter online workshops, i) the introductory workshop, ii) the GMB workshop with all the adolescents and iii) a validation workshop. We adapted the scripts to a single one-hour online workshop with the policymakers and public health practitioners since the time available from this group was limited due to COVID-19 response requirements. To test the online platform and the timings for the activities, we did two online piloting workshops with volunteers from the University of Bristol. In a third phase (III. Facilitation and Delivery) we delivered the three GMB workshops with adolescents (1. Introduction – individual; 2. GMB – group; and 3. Validation – individual), and the one workshop with policymakers and public health practitioners.

Shared objectives:

Specific objectives for adolescents:

Specific objectives for policymakers and public health practitioners:

The GMB workshops were based on Scriptapedia [16, 18], an online freely distributed book which uses scripts for structured group exercises as guides to conduct GMB practice. The process mapping enabled drafting the different activities required to be able to build the system map with the adolescents. We used BlueJeans, a virtual online platform, and STICK-E [19], a licenced application developed by system mapping experts in Deakin University, Australia. Table 1 gives a brief explanation of the Scriptapedia scripts we used.

We found many advantages of doing GMB workshops online in terms of cost and commuting time. Initially we costed ≈£650 for room hire, refreshments, stationery and travel costs for the facilitator and note-taker. In addition, the time between each of the workshops with the adolescents allowed for a reflective process to occur making the following workshops more engaging. Adolescents were very keen to share their experiences and interaction with their food environments and how the first workshop allowed them to see their food choice processes more critically, for example, when they went to the shops or used social media, they had a critical eye on the food marketing that “popped-up” on their screens.

We enhanced the development of system’s mapping experience into an online environment [14] by implementing Causal Mapping with Seed Structure and by exploring the use of Graphs Over Time and Action Ideas scripts, as suggested by Wilkerson et al. (2020). We started the GMB workshop’s discussion with a seed structure (i.e. the synthesised map from the 11 individual workshops with the adolescents), which allowed us to quickly illustrate how the problem being discussed involved a system of interacting feedback loops. Using the Graphs Over Time script prompted participants to think in terms of variables (i.e. factors which are amenable to change) and it facilitated the initiation of the mapping process having decided on the variables they thought to be the most important ones. Finally, using the Action Ideas scripts allowed adolescents to think about policy ideas to intervene in the system map they had created. These ideas were then shared with the policymakers and public health practitioners.

Validation of the system map was achieved through consensus among workshop participants and by comparing the map to existing literature on the topic, and the method was evaluated through an anonymous survey at the end of the workshops. These processes could be carried out equally well in an in-person or in an online workshop. However, we do not recommend having a mixed setup (some participants online, and some in person) due to power dynamics and facilitation challenges [14] and because participants online might miss the in-person dynamic, causing an unbalanced communication.

Online GMB should not be thought of a substitution for face-to-face delivery, but rather as a complementary option for system mapping. As more research emerges on this topic it would be valuable to compare and evaluate the effectiveness of online versus in-person GMB workshops in terms of the quality of the results and long-term mental model changes on participants.

We were interested in introducing participants to systems thinking concepts (i.e. interconnectedness, feedback loops), but within the timeframe of the study were not able to also explore concepts of accumulation or feedback loops with time delays, neither could we compare system maps at different time points, between groups or delivery modes (i.e. online vs face-to-face). A deeper understanding of complex systems methodologies, including the aforementioned interacting feedback loops, time-delays, and accumulations, might have further improved the final maps [25, 26].

Compared to face-to-face workshops, the online workshop process could have impacted on the validation by limiting the consensus on relevant variables due to online fatigue, or by making it more difficult for the facilitator to track the non-verbal behaviour of participants. However, previous research suggests that discussions in online workshops, compared to in-person, does not impact on quality while also moderators report less overtalking than in in-person ones [23]. Additionally, we placed particular emphasis on encouraging participants to voice their ideas, both in the group, and subsequently individually in case they did not feel confident in expressing their ideas in the group. Therefore, we believe that our methodological evaluation supports our study’s findings. However, we recognise that validation methods for GMB models, such as meditation analysis, structural-equation-modelling, exploratory factor analysis could help quantitatively test the relationships among the elements and underlying factors, and account for their non-linear relationships in the model and produce more robust models to support policymaking efforts [15].

Designing and delivering online GMB is achievable, however it comes with its challenges due to the many aspects and roles needed to deliver the workshop activities. We have provided some recommendations to overcome these in Table 3 Additionally, guidance on best practices for online power dynamic facilitation would be beneficial to have more balanced relational processes when performing participatory activities, like GMB workshops, where one of the aims is to ensure everyone has an equal opportunity to share their views about the causes of a problem and its potential solutions to enable a systemic change [27]. Having relatively small groups with minimal levels of power differences seemed to be a beneficial factor when facilitation online GMB workshops. Additionally, having separate workshops with participants to introduce them to system dynamic concepts and to the system mapping platform seemed to have favoured active and confident participation during the group workshop.

Adapting GMB workshops to an online setting has its limitations. Aiming for small groups, for ease of facilitation and communication, can limit the diversity of perspectives about the problem [20]. GMB activities can be cognitively demanding and further exacerbated by screen fatigue [13, 20]. In our case, breaking down the activities into short workshops (i.e. 30 to 90 min) on different days enhanced participants’ engagement. Having GMB workshops fully online risks limiting participation of people with limited technological knowledge who might be discouraged to take part, it excludes people with no or poor internet access, and people living in space-limited accommodation might not feel comfortable participating due to restricted privacy whilst taking part in the workshop. It was difficult to reach minority groups and participants from different ethnicities or socio-economic backgrounds. This might have been due to having limited time for recruitment. Additionally, COVID-19 restrictions had just started, therefore the transition to online environments was still developing and potential participants might not have felt encouraged enough to participate in an online workshop. It would be especially valuable to explicitly design online GMB to include these groups which tend to be left out of online discussions and could further exacerbate social and health inequalities [28]. GMB workshops are designed to be delivered in person, and one could potentially miss the rapport built in a face-to-face setting. However, we believe that breaking down the activities into different workshops and having an initial individual introductory meeting allowed participants to become familiar with the concepts, the platforms and the problem before having the group activity. Due to COVID-19 we had very limited time with the practitioners and policymakers and they felt unsatisfied with how much they could contribute during the workshop. Having an initial introductory meeting (one-on-one) with them could save time during the group workshop and allow more time for discussing the map and policy ideas. Extending the group workshop by 30 min (1.5 h total) might also help.

Through a systems thinking approach and using an adapted GMB methodology, individuals were able to have a better understanding of the complex system that influences dietary behaviour in the light of food industry influence. By using systems thinking tools, adolescents confirmed their understanding of this complex system at the level of recognising that all the factors they mentioned were interconnected, they recognised their interdependencies and highlighted important reinforcing feedback loops when analysing system outcomes (i.e. a food system which gives preference to unhealthy vs healthy food choices), and suggested interventions to balance the system in favour of healthy food choices [25].

In our experience, designing and delivering GMB online is feasible, engaging, cost-saving and an enjoyable experience. Participants gave positive feedback in terms of engagement, enjoyment and it allowed them to recognise and accept different points of view about the same problem.

Online GMB workshops are achievable and an enjoyable experience. Participants became familiarised with system thinking and system dynamics concepts, they developed a shared understanding of a complex issue and they were able to portray it in a system map that depicted the most important factors in a causal structure and how they interact at various levels. In Table 3 we share the main recommendations of the most important things to take into consideration when running an online GMB workshop.

Standardising online GMB approaches would be valuable to compare effectiveness and quality between online and in-person workshops [14, 29, 30]. Evaluations of online versus in-person GMB workshops will enrich our knowledge of the effect online system mapping workshops has on stakeholders, the trade-off between effectiveness, quality and efficiency, and how this might be enhanced to identify leverage points and achieve systemic changes in complex issues.