Skip to main content
NCBI home page
Sage Choice logoLink to Sage Choice
. 2023 May 2;50(5):662–670. doi: 10.1177/10901981231165339

Measuring Shifts in Mental Models in the Prevention of Childhood Obesity in Rural Australia

Tiana Felmingham 1,✉, Kristy A Bolton 1, Penny Fraser 1, Steven Allender 1, Andrew D Brown 1
PMCID: PMC10492428  PMID: 37128853

Abstract

Group model building is a participatory workshop technique used in system dynamics for developing community consensus to address complex problems by consensus building on individual assumptions. This study examines changes in individual mental models of the complex problem of childhood obesity following participation in group model building (GMB), as part of a larger community-based system dynamics project. Data are drawn from GMB participants across six community sites in the Whole of Systems Trial of Prevention Strategies for Childhood Obesity (WHO STOPS) in rural and regional Victoria, Australia. Each community participated in two GMB sessions resulting in a causal loop diagram (CLD) of drivers of childhood obesity for each community. Presurvey and postsurvey captured participants’ perspectives before and after (n = 25) participation in both GMB sessions and a blend of inductive and deductive qualitative content analysis was used to code individual responses. Three calculations were used to determine the number of responses, whether responses were a result of persuasion from others, and comparison of responses to those found in the CLD. Our study found participant mental models shifted during the course of the GMB sessions, with some responses persuaded by others and 75% of new insights identified in CLDs created by communities. The GMB process created a platform for participants to share ideas and learn from each other. In addition, participants listed new insights about childhood obesity in their community through developing CLDs.

Keywords: childhood obesity, systems thinking, community-based system dynamics, mental models, prevention

Introduction

Community health problems are complex and require community participation to effectively address them (Nickel & von dem Knesebeck, 2020; J. Scott et al., 2020). Systems thinking has shown promise in tackling complex social and community problems (Baugh Littlejohns et al., 2018; Calancie et al., 2018; Carey et al., 2015; Chughtai & Blanchet, 2017; Rutter et al., 2017). Community-based system dynamics (CBSD), a method within systems thinking, uses a series of facilitated participatory workshops, primarily focused on a technique called group model building (GMB). Along with several other outcomes, GMB maps participants’ understanding of causal factors within a complex system and shows how they interact with each other in feedback loops, creating a causal loop diagram (CLD). In doing so, the GMB process builds a shared mental model of participants’ understanding, defined by Doyle and Ford (1998) as “a relatively enduring and accessible, but limited, internal conceptual representation of an external system.” The CLDs from GMB make a community’s mental models explicit (Dwyer & Stave, 2008), providing a shared, grounded model of the complexity driving a problem. This shared contextually relevant understanding represents a basis for more collaborative action than traditional, externally derived, generic, linear program logic models.

As a graphical representation of community consensus about cause and effect in a complex problem, CLDs can then be used to identify and agree on potential intervention points across a system and support cohesive planning of local action. These nonlinear models identify feedback within a system theoretically known to be compounding existing problems and limiting the strength of corrective actions (Hovmand, 2014). GMB sets out to make mental models explicit among communities, strengthen relationships and collaboration and generate consensus and commitment to prioritize and implement agreed actions (Andersen et al., 1997; Black & Andersen, 2012; McCardle-Keurentjes et al., 2018; Rouwette et al., 2002; R. J. Scott et al., 2016a). The resultant shifts and alignment in mental models among stakeholders are cited as a major positive of the GMB process (Black & Andersen, 2012; R. J. Scott et al., 2016b).

Capturing the impacts of GMB on mental models among participants presents a challenge (Andersen et al., 1997; Rouwette et al., 2011; R. J. Scott et al., 2016b). Fokkinga et al. (2009) developed a method to measure changes in mental models within individuals. The tool captures participants’ mental models immediately before (pre) and after (post) GMB workshops. Their study developed and tested a measurement instrument that evaluated shifts in mental models showing it was possible to measure changes in individual mental models using this tool. While the approach has been used within organizations (R. J. Scott et al., 2013), this instrument has not been used in the community context.

The Whole of Systems Trial of Prevention Strategies for Childhood Obesity (WHO STOPS) was a 5-year study set in regional and rural Victoria, Australia with the aim of building community capacity in systems thinking to improve child health. WHO STOPS conducted a separate series of GMB workshops creating CLDs for childhood obesity for six communities. As outlined by Allender et al. (2016), this series of GMB workshops were used by community members to design and implement actions targeting obesity. In this article, we examine how GMB changed participants’ mental models of childhood obesity (Fokkinga et al., 2009; R. J. Scott et al., 2013).

This study is one approach to evaluation of WHOSTOPS and compliments other published literature about the intervention. WHOSTOPS intended to build stakeholders’ capacity in systems thinking to enable systemic action to prevent childhood obesity (Allender et al., 2016). Epidemiological results from the outcome evaluation of the study examined body mass index and health-related quality of life over 4 years and are published elsewhere (Allender et al., 2021), but indicate WHOSTOPS played a role in keeping takeaway food intake low and sustaining health-related quality of life. Several process evaluations that examined how systems thinking was applied in practice, barriers, and enablers to engaging community and implementing systems change, and equity of health outcomes from the trial have been published to build understanding of how the intervention worked in communities (Allender et al., 2019; Bolton et al., 2022; Jacobs et al., 2021; Jenkins et al., 2020). This article contributes to the process evaluation by examining the degree to which stakeholders’ thinking about childhood obesity changed after participation in a GMB process.

Impact Statement.

Group model building (GMB), a method used in system dynamics, can be used to empower and mobilize communities in response to some of the most complex community problems. While GMB attempts to make individual mental models explicit and may align collective effort to address a problem, documenting shifts in mental models is a challenge. Previous work used presurvey and postsurvey to document shifts in mental models in organizational settings. This study extends that work by investigating the effects of GMB on individual mental models in the community setting and across multiple sites, documenting shifts in individual thinking following participation. These shifts may provide insights into how GMB can support community led systems change efforts.

Methods

GMB Process

This article focuses on the first two workshops in the WHOSTOPS community process, which focus on the development of a CLD. Within each community, a community leader with strong networks and a remit for improving health was identified. This person helped to recruit other community leaders to attend the two GMB sessions (GMB1 and GMB2). Community leaders were defined as those who had authority to change where children eat and play. GMB1 was a 90-minute session where community-specific, current childhood obesity and associated risks were presented resulting from a high quality monitoring study (Crooks et al., 2017). Content for each session was guided by GMB scripts (WikiBooks, 2022) (Table 1). Graphs over time (GOT) is a GMB process that helps participants visualize how each factor has changed over time. This helped to identify drivers of the observed patterns in obesity and associated risks. Using bespoke software, called STICKE, these drivers were translated into connection circles to build an initial CLD (Hayward et al., 2020; Savona et al., 2021). In GMB2, another 90-minute session, the CLD was refined and additional variables and connections included. The time gap between first and second GMB sessions varied by community, but was approximately 1 to 3 months. More information about the process used in these communities has been published elsewhere (Bolton et al., 2022). The presurvey was completed by participants immediately before GMB1, followed by the postsurvey completion immediately after GMB2.

Table 1.

Content Used in the Delivery of GMB1 and GMB2 Sessions.

GMB GMB script Description
GMB1 Graphs over time Helps participants to frame a problem by eliciting variables that inform building of the CLD
Connection circles Helps participants identify connections and feedback between variables
GMB2 Model review Helps to summarize insights and stories, clarify ideas and identify additional information from the participants perspective
Open in a new tab

Note. GMB = Group Model Building; CLD = causal loop diagram.

Participants

GMB sessions were conducted within six communities between 2016 and 2019. All communities were based in rural and regional Victoria, with a population range of 1,202 to 20,972. Key industries across communities were similar and included agriculture (dairy cattle/beef farming), health (hospitals and aged care), and food manufacturing (meat, ice cream) (Australian Bureau of Statistics, 2022). A total of 94 people attended the six GMB1 sessions and 67 people attended the six GMB2 sessions. Attendees were from the following sectors: local government, health and medical services, education, legal, real estate, sport, community-based organizations and clubs, nongovernment organizations, disability services, social and community services, faith-based services, and environment.

Collection of Mental Model Evaluation Data

Participants were briefed at the beginning of each GMB and provided a signed consent form before being assigned a de-identified identification number to link individual data pre-GMB1 and post-GMB2. Presurvey and postsurvey using Fokkinga et al.’s (2009) questionnaire collected mental models among workshop participants. These surveys were administered immediately prior to GMB1 and again at the conclusion of GMB2. As a new measurement technique, this is one of the first examples of this tool and corresponding analysis being used. The questionnaire (Fokkinga et al., 2009) was adapted to collect detailed information in free text format about individuals’ belief about:

  1. the causes of childhood obesity,

  2. the effects or consequences of childhood obesity,

  3. three options to prevent childhood obesity.

Analysis of Mental Model Data

Qualitative content analysis was used to analyze the data (Fokkinga et al., 2009; Prasad, 2019; Schreier, 2014). A blend of inductive and deductive coding was used across coding cycles (Fereday & Muir-Cochrane, 2006). First cycle coding (initial coding) was completed by three coders (TF, AB, and PF) who iteratively applied codes independently to responses from each workshop participant (Saldaña, 2009). TF collated codes from all coders and grouped codes into three areas: agreement (three coders have the same code), some discussion (two coders with the same or similar codes, with one coder substantially different), or deep discussion (all three coders have different codes). Coders came together to discuss and agree on remaining codes for each response. Codes were framed as dynamic variables consistent with system dynamics conventions and several responses received multiple codes to reflect the complexity of their response. For example, where a participant reported that “access to physical activity and fresh food” was a factor that contributed to childhood overweight and obesity, two codes were applied: “access to physical activity” and “access to healthy food.” A codebook was developed to capture all agreed codes (Saldaña, 2009).

Data were then re-grouped, connecting each individual response back to its original predataset and postdataset along with its connection to the individual who contributed the response. These codes formed the basis of subsequent analysis and interpretation of data providing comparisons. Analysis followed previously defined methods to categorize number of responses, GMB as a persuasive technique and qualitative changes (Fokkinga et al., 2009; R. J. Scott et al., 2013).

Number of Responses

This measure was used to determine whether participants refined their mental models, indicated by fewer responses in the postsurvey than the presurvey (R. J. Scott et al., 2013). The number of responses was calculated by counting the number of codes applied to each question completed by each participant. All participants entered at least one response for every question. In some instances, separate responses may have been given the same code for a participant, and were counted as two responses. For example, where one participant suggested two separate causes of overweight and obesity in children locally, “busy lives” and “time poor,” these were both coded as “busyness” and counted as two responses for the purpose of this analysis.

GMB as a Persuasive Technique

This measure was used to determine whether participants were either persuaded by others in the room or if new insights were listed after participating in GMB1 and GMB2 (Rouwette et al., 2011; R. J. Scott et al., 2013).

Three calculations were used to determine persuasion using responses from participants:

  1. How many responses per participant were retained from presurvey to postsurvey? That is, the total number of post-survey responses that were also present in the same participant’s presurvey divided by the total number of participants in a community to get the average number of retained responses per participant.

  2. How many responses per participant were a result of persuasion from others in the community? That is, the total number of postsurvey responses that did not appear in the same participant’s presurvey, but did appear in another participant’s presurvey divided by the total number of participants in a community to get the average number of responses that were taken from one participant by another.

  3. How many responses per participant were new insights listed after participating in the GMB process? i.e., The total number of responses in the postsurveys that did not appear in any participants’ presurvey within a community, divided by the number of participants in that community to give the average number of new insights per participant.

Qualitative Changes

Several responses in the postsurveys did not appear on any presurvey, and these responses were compared with the CLD generated by the two GMB sessions to test whether the new ideas being prioritized were reflected in the group’s conversation. All three coders categorized new insights in the postsurvey responses as present, implied, not present, or referred to a systems approach. Present responses were present in the CLD either with the exact same wording or a clearly similar concept. Implied responses could not be linked to one variable in the CLD explicitly but appeared to be implied when looking holistically at a set of variables and their connections within the CLD. Not present responses were not in the CLD at all. Systems approach responses were from participants indicating that a systems approach is an important driver, result, or solution to the problem.

Results

Of all participants who attended both GMB1 and GMB2 (n = 46), 54.4% (n = 25) completed both presurvey and postsurvey (Table 2). This study considers responses from the 25 participants who completed both presurvey and postsurvey.

Table 2.

Number and Percentage of Participants Who Completed Both Presurvey and Postsurvey, and Total and Average Number of Responses Per Participant in Each Community in Presurvey and Postsurvey, and all Communities Combined.

Community Attendance at GMB1 Attendance at GMB2 Attendance at both GMB1 and GMB2 (retention rate) Both presurvey and postsurvey completion (percentage) Responses
Total number in presurvey Average number of responses/person a Total number in postsurvey Average number of responses/person b
Community 1 13 9 6 (46%) 5 (83%) 66 13.2 56 11.2
Community 2 16 16 11 (69%) 7 (64%) 119 17.0 87 12.4
Community 3 17 10 8 (47%) 6 (75%) 72 12.0 81 13.5
Community 4 21 18 10 (48%) 3 (30%) 40 13.3 39 13.0
Community 5 10 7 6 (60%) 2 (33%) 31 15.5 30 15.0
Community 6 17 7 5 (29%) 2 (40%) 50 25.0 38 19.0
All communities 94 67 46 (49%) 25 (54%) 378 15.1 331 13.2
Open in a new tab

Note. GMB = Group Model Building.

a

Total number of codes in presurvey/number of participants from the corresponding community. bTotal number of codes in postsurvey/number of participants from the corresponding community.

Number of Responses

The average number of responses per participant was calculated for each community (Table 2). The average number of responses per participant decreased slightly between presurvey and postsurvey in Communities 1, 2, 4, 5, and 6. Given that the number of preresponse and postresponse were relatively similar to one another, there was insufficient evidence to indicate that mental model refinement occurred.

Modeling as Persuasion

Survey results indicate that participants were persuaded by others over the course of GMB1 and GMB2. Some responses were retained from presurvey to postsurvey for participants. A vast majority of individuals listed at least one response that was either reflective of being persuaded or identified as a new insight, and many individuals had multiple of these responses. In five communities, the average sum of the number of new insights answers per person was higher than the average number of retained responses per person. Individuals from all communities also had new insights emerge for every question after participating in the GMB process. While each community was different, examples of retained responses in some communities included access to healthy food, busyness, mental health and access to screens, while persuaded responses included community leadership, self-esteem, and physical activity infrastructure, and examples of new insights included culture change, food supply chain, and a systems approach.

Responses for individuals across communities are summarized in Table 3. For all questions, Communities 1, 2, and 3 had the highest number of postsurvey responses that appeared in other participants’ presurveys, indicating they had been persuaded by others in the group. While Communities 4, 5, and 6 experienced the lowest amount of persuasion, they had the highest number of new insights develop after participating in the GMB process. This contrasted with Community 2, who had the lowest new insights emerge following their participation.

Table 3.

Responses Retained, Persuaded by Others and New Insights Gained Per Participant Within Each Community for All Questions Combined.

Community Responses
Retained a Persuaded by others b New insights c
Community 1 3.80 3.20 4.40
Community 2 6.86 4.00 1.57
Community 3 3.17 4.50 6.00
Community 4 4.33 2.00 6.67
Community 5 5.50 1.50 7.50
Community 6 7.50 2.00 9.50
All questions, all communities (mean) 4.84 3.32 4.76
Median 4.92 2.60 6.33
Range 3.17 to 7.50 1.50 to 4.50 1.57 to 9.50
Open in a new tab
a

Total number of postsurvey responses that were also in an individuals’ presurvey responses/total number of participants from the community who completed both presurvey and postsurvey. bTotal number of postindividual responses that were also in another participants presurvey (but not their own)/total number of participants from the community who completed both presurvey and postsurvey. c Total number of postsurvey responses that were did not appear in any presurvey responses/total number of participants from the community who completed both presurvey and postsurvey.

Qualitative Changes

In all communities, a majority of new insight responses were also present in the CLD developed by the community. New insights were categorized as either present (i.e., new insights identified in individual postsurveys also appeared in that community’s CLD in the same or very similar wording), implied (i.e., where new insights in individual postsurveys did not appear as a single variable in the CLD but the concept was identified when looking at a collection of variables in the CLD), a systems approach (i.e., where new insights identified in individual postsurveys described a systems approach as important when contributing to or addressing the problem) or not present (not appearing in the CLD at all) (Table 4). Examples of new insights that were present in both individual postsurveys and the respective community’s CLD included mental health, consumption of healthy food and access to transport. Examples of implied insights within communities included social disadvantage and access to unhealthy food. Examples of new insights that described a systems approach included community leadership, a systems approach and policy.

Table 4.

Number and Percentage of All of New Insights From Postsurveys to CLDs From Individuals Across Communities.

Community Present Implied Systems approach Not present
a b (%) c d (%) e f (%) g h (%)
Community 1 4 18 5 23 4 18 9 41
Community 2 4 36 4 36 0 0 3 27
Community 3 25 69 8 22 1 3 2 6
Community 4 12 60 1 5 3 15 4 20
Community 5 3 19 10 63 0 0 3 19
Community 6 5 26 4 21 0 0 10 53
All communities 53 43 32 26 8 6 31 25
Open in a new tab
a

Number of responses present in the CLD either with the exact same wording or a clearly similar concept in each community. bTotal number of “present” responses in each community/total number of responses in the community × 100. cNumber of responses that could not be linked to one variable in the CLD explicitly but appeared to be implied when looking holistically at a set of variables and their connections within the CLD. dTotal number of “implied” responses in each community/total number of responses in the community × 100. eNumber of responses that indicated that a systems approach is an important driver, result, or solution to the problem. fTotal number of “systems approach” responses in each community/total number of responses in the community × 100. gNumber of responses were not in the CLD at all. hTotal number of “not present” responses in each community/total number of responses in the community × 100.

Discussion

This study found evidence that participants’ mental models may have shifted following the GMB process either through the persuasion of others or through creating new insights. Two previous studies have demonstrated similar shifts using these methods (Fokkinga et al., 2009; R. J. Scott et al., 2013) though this is the first time this instrument has been used in the community setting. The Fokkinga et al. (2009) study tested the questionnaire they developed to measure shifts in mental models as a result of GMB intervention in applied settings using a GMB group and a control group. They concluded it was possible to measure a shift in individual and group mental models, pre-GMB and post-GMB interventions, using their questionnaire. R. J. Scott et al. (2013) also aimed to measure changes in mental models, using a combination of evaluation methods, including prequestion and postquestion, and interviews with four groups within an organization as a case study, to determine whether changes in mental models could endure over time. They found that shifts in mental models occurred as result of GMB, and that these changes endured over time. As with both of these studies, we also found that mentals’ models appeared to change following GMB and we were able to capture these changes using the prequestionnaire and postquestionnaire designed by Fokkinga et al. (2009). We were also able to meet Fokkinga et al.’s (2009) suggestion of applying the method in a real life setting for the first time, with teams of local decision-makers using this method across six different communities. We also extended this to include an analysis of individual mental model changes in comparison with community-built CLDs. There were a range of results, though most individuals decreased the number of responses between presurvey and postsurvey, others remained equal or increased. One explanation for this may be that there were some participants who broadened their perspectives because of the GMB process, resulting in more responses. While others may have consolidated their thinking, reducing the number of responses (Hinsz et al., 1997; Hovmand, 2014; McCardle-Keurentjes et al., 2018).

Some responses in postsurveys did not appear in any presurveys, indicating that at the conclusion of the workshop, the participants were possibly considering or prioritizing ideas that they had not been considering prior to the workshop. It is impossible to know with certainty what drives those ideas, but to test whether they possibly could have been a result from the workshops and drawing the CLDs, the new insights that emerged were compared with the CLDs in each community.

Our study differed from previous studies in that we did not find mental model alignment across participants, defined as fewer unique codes per person on average between the pre- and postsurveys (Huz et al., 1997; R. J. Scott et al., 2013). The context and overall purpose of GMB is critical in measuring its effectiveness particularly in terms of the existing activities, relationships, previous activities and collaboration and competing priorities (Hinsz et al., 1997; Hovmand, 2014). The GMB process in this study aimed to collect mental models, engage communities, increase participant understanding of the underlying system driving obesity and identify connections and points of intervention (Jenkins et al., 2020; Maitland et al., 2019; Owen et al., 2018). Therefore, there may not have been the same focus on narrowing down to a shared understanding and specific shared actions as in other applications of GMB.

The comparison of individual responses (or codes) with the content of the CLD is a new method trialed in this study. While there were variations across individuals and communities, three quarters of new insights for all questions for individuals across communities were present in the CLD in some form, suggesting that participation in the GMB process to build a collective CLD may result in the creation of new insights.

The retention rate for the presurvey and postsurvey was highest for individuals from Community 1 and only 47% of new insights from individuals in that community were in the CLD or were confirmed to be a “systems approach.” During the GMB process, individuals from Community 1 displayed a low level of engagement and trust in the process. There was obvious division among community members and organizations, including education and health services.

Individuals from Community 2 also had a high number of retained responses, but in a difference to those from Community 1, 73% of the new insights were reflected in the CLD or were a call for “systems approaches”. A larger than normal number of participants in GMB1 had previous experience with and exposure to both GMB processes and systems thinking.

The GMB process alternates between divergent and convergent exercises to develop a wide range of insights into an issue followed by a consensus of the most relevant of these insights. This may explain why these results differ from other studies. While one of the main outcomes of GMB is to establish consensus or agreement across a group (Rouwette, 2011) this is not always the case. This may be a result of differences in the overall purpose of the GMB process, particularly where the context between communities differs, or where there may have less emphasis on alignment and greater emphasis on identifying relationships and creating new insights as part of the workshop delivery. These results may also be influenced by the stage of the program, the number and experience of facilitators, in addition to the structure of each session.

Strengths and Limitations of the Study

Relative to other similar studies, we report a large sample size. This study was completed with participants from six different rural communities, providing an opportunity to observe patterns in shifting individual mental models over time and location. It allowed questions to be asked of context, consistency of GMB structure and facilitation, the importance of people in the room, and provided the opportunity to test a new measurement tool in an applied setting. Consistency of facilitators’ ensured content was consistent across sites, with the role of facilitators potentially positively impacting the variables elicited. While it may be possible that facilitators and workshop content influenced participant postworkshop responses, we are not able to determine the extent of this influence. The approach to coding strengthened validity and rigor, through the use of three coders first coding independently, followed by discussion and agreement on each code, and later trends across new insights and CLDs.

The data collected from participants at the beginning of GMB1 and end of GMB2 may have also been impacted by the limited time available for participants to complete the surveys. This potentially impacted the number of responses in each question. Participant responses also varied in presentation (e.g., dot points, full text sentences with multiple ideas/concepts) which sometimes made it difficult to determine the exact number of responses. This was strengthened using multiple coders and a decision to analyze codes opposed to responses. While some responses were retained by participants, some did not appear and some were new, it is beyond the scope of this study to determine the reasons behind these changes. Our study considered individual shifts across all questions included in the questionnaire. A limitation of this approach was the possibility of missing individual shifts per question, which may have surfaced new insights and modified results. Future studies could examine this.

Implications for Practice

This study reinforces that the GMB process can be used as an effective way of surfacing mental models of a problem and thus providing a basis to support community collaboration to address complex social and health problems. This study suggests timing, structure and context of GMB workshops are important and these results should direct subsequent studies to consider these aspects when designing implementation of new GMB efforts in the prevention of childhood obesity. These lessons likely also apply for practitioners seeking to use GMB approaches for other complex social and community issues.

It is critical that participants from GMB1 also attend GMB2 if comparison of mental models is an objective of the process. GMB1 and GMB2 may be considered as a single event, split over two sessions, leading into later work focused on action opposed to model development or refinement. Future research examining the impact on mental model shifts in community members on community level decision-making would also be of value.

Conclusion

This study provides one example of how to measure shifts in mental models in the community setting following participation in the GMB process, where there are few others. It highlights that when the GMB process is used to tackle complex community issues, it can change the perspectives of participants by being persuaded by others, and creating new insights.

Acknowledgments

The authors thank the participants (and their organizations) for their involvement in the GMB workshops.

Footnotes

Authors’ Note: The opinion and analysis in this article are those of the author(s) and are not those of the Department of Health, the Victorian Government, the Secretary of the Department of Health, or the Victorian Minister for Health.

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding: The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by a NHMRC Partnership Project titled “Whole of Systems Trial of Prevention Strategies for childhood obesity: WHOSTOPS childhood obesity” (APP1114118). The work described herein has also received funding support from the Western Alliance. Allender, Bolton, Fraser and Brown were researchers within the NHMRC Center for Research Excellence in Obesity Policy and Food Systems (APP1041020) at the time the study was conducted. Community partners also providing support to the research include Portland District Health, Western Alliance, Southern Grampians and Glenelg Primary Care Partnership, Colac Area Health, Portland Hamilton Principal Network of Schools, Colac Corangamite Network of Schools, The Glenelg Shire Council, Southern Grampians Shire Council, Warrnambool and District Network of Schools, Western District Health Service, and Victorian Department of Health and Human Services.

ORCID iD: Tiana Felmingham Inline graphic https://orcid.org/0000-0003-2278-7249

References

  1. Allender S., Brown A. D., Bolton K. A., Fraser P., Lowe J., Hovmand P. (2019). Translating systems thinking into practice for community action on childhood obesity. Obesity Reviews: An Official Journal of the International Association for the Study of Obesity, 20(Suppl.2), 179–184. 10.1111/obr.12865 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Allender S., Millar L., Hovmand P., Bell C., Moodie M., Carter R., . . . Morgan S. (2016). Whole of systems trial of prevention strategies for childhood obesity: WHO STOPS childhood obesity. International Journal of Environmental Research and Public Health, 13(11), Article 1143. 10.3390/ijerph13111143 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Allender S., Orellana L., Crooks N., Bolton K. A., Fraser P., Brown A. D., . . . Millar L. (2021). Four-year behavioral, health-related quality of life, and BMI outcomes from a cluster randomized whole of systems trial of prevention strategies for childhood obesity. Obesity, 29(6), 1022–1035. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Andersen D. F., Richardson G. P., Vennix J. A. (1997). Group model building: Adding more science to the craft. System Dynamics Review, 13(2), 187–201. [Google Scholar]
  5. Australian Bureau of Statistics. (2022). 2016 census all persons QuickStats. https://www.abs.gov.au/census/find-census-data/quickstats/2016/0
  6. Baugh Littlejohns L., Baum F., Freeman T., Lawless A. (2018). The value of a causal loop diagram in exploring the complex interplay of factors that influence health promotion in a multisectoral health system in Australia. Health Research Policy and Systems, 16(1), Article 126. 10.1186/s12961-018-0394-x [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Black L. J., Andersen D. F. (2012). Using visual representations as boundary objects to resolve conflict in collaborative model-building approaches. Systems Research and Behavioral Science, 29(2), 194–208. 10.1002/sres.2106 [DOI] [Google Scholar]
  8. Bolton K. A., Fraser P., Lowe J., Moodie M., Bell C., Strugnell C., . . . Allender S. (2022). Generating change through collective impact and systems science for childhood obesity prevention: The GenR8 Change case study. PLOS ONE, 17(5), Article e0266654. 10.1371/journal.pone.0266654 [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Calancie L., Anderson S., Branscomb J., Apostolico A. A., Hassmiller Lich K. (2018). Using behavior over time graphs to spur systems thinking among public health practitioners. Preventing Chronic Disease, 15, Article 170254. 10.5888/pcd15.170254 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Carey G., Malbon E., Carey N., Joyce A., Crammond B., Carey A. (2015). Systems science and systems thinking for public health: A systematic review of the field. BMJ Open, 5(12), Article e009002. 10.1136/bmjopen-2015-009002 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Chughtai S., Blanchet K. (2017). Systems thinking in public health: A bibliographic contribution to a meta-narrative review. Health Policy and Planning, 32(4), 585–594. https://doi.org10.1093/heapol/czw159 [DOI] [PubMed] [Google Scholar]
  12. Crooks N., Strugnell C., Bell C., Allender S. (2017). Establishing a sustainable childhood obesity monitoring system in regional Victoria. Health Promotion Journal of Australia, 28(2), 96–102. 10.1071/HE16020 [DOI] [PubMed] [Google Scholar]
  13. Doyle J. K., Ford D. N. (1998). Mental models concepts for system dynamics research. System Dynamics Review, 14(1), 3–29. 10.1002/(SICI)1099-1727(199821)14:1<3::AID-SDR140>3.0.CO;2-K [DOI] [Google Scholar]
  14. Dwyer M., Stave K. (2008, July 20–24). Group model building wins: The results of a comparative analysis [Paper presentation]. Proceedings of the 2008 International Conference of the System Dynamics Society, Athens, Greece. [Google Scholar]
  15. Fereday J., Muir-Cochrane E. (2006). Demonstrating rigor using thematic analysis: A hybrid approach of inductive and deductive coding and theme development. International Journal of Qualitative Methods, 5(1), 80–92. 10.1177/160940690600500107 [DOI] [Google Scholar]
  16. Fokkinga B., Bleijenbergh I., Vennix J. A. M. (2009, July 26–30). Group model building evaluation in single cases: A method to assess changes in mental models [Paper presentation]. Proceedings of the 27th International Conference of the System Dynamics Society, Albuquerque, NM, United States. [Google Scholar]
  17. Hayward J., Morton S., Johnstone M., Creighton D., Allender S. (2020). Tools and analytic techniques to synthesise community knowledge in CBPR using computer-mediated participatory system modelling. NPJ Digital Medicine, 3(1), Article 22. 10.1038/s41746-020-0230-x [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Hinsz V. B., Tindale R. S., Vollrath D. A. (1997). The emerging conceptualization of groups as information processors. Psychological Bulletin, 121(1), 43–64. 10.1037/0033-2909.121.1.43 [DOI] [PubMed] [Google Scholar]
  19. Hovmand P. S. (2014). Community based system dynamics. Springer. [Google Scholar]
  20. Huz S., Andersen D. F., Richardson G. P., Boothroyd R. (1997). A framework for evaluating systems thinking interventions: An experimental approach to mental health system change. System Dynamics Review, 13(2), 149–169. [DOI] [Google Scholar]
  21. Jacobs J., Strugnell C., Allender S., Orellana L., Backholer K., Bolton K. A., . . . Nichols M. (2021). The impact of a community-based intervention on weight, weight-related behaviours and health-related quality of life in primary school children in Victoria, Australia, according to socio-economic position. BMC Public Health, 21(1), 1–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Jenkins E., Lowe J., Allender S., Bolton K. A. (2020). Process evaluation of a whole-of-community systems approach to address childhood obesity in western Victoria, Australia. BMC Public Health, 20, Article 450. 10.1186/s12889-020-08576-x [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Maitland N., Williams M., Jalaludin B., Allender S., Strugnell C., Brown A., . . . Wardle K. (2019). Campbelltown—changing our future: Study protocol for a whole of system approach to childhood obesity in South Western Sydney. BMC Public Health, 19, Article 1699. 10.1186/s12889-019-7936-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. McCardle-Keurentjes M. H. F., Rouwette E. A. J. A., Vennix J. A. M., Jacobs E. (2018). Potential benefits of model use in group model building: Insights from an experimental investigation. System Dynamics Review, 34(1–2), 354–384. 10.1002/sdr.1603 [DOI] [Google Scholar]
  25. Nickel S., von dem Knesebeck O. (2020). Effectiveness of community-based health promotion interventions in urban areas: A systematic review. Journal of Community Health, 45(2), 419–434. 10.1007/s10900-019-00733-7 [DOI] [PubMed] [Google Scholar]
  26. Owen B., Brown A. D., Kuhlberg J., Millar L., Nichols M., Economos C., Allender S. (2018). Understanding a successful obesity prevention initiative in children under 5 from a systems perspective. PLOS ONE, 13(3), 1–10. 10.1371/journal.pone.0195141 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Prasad B. D. (2019). Qualitative content analysis: Why is it still a path less taken? Forum: Qualitative Social Research, 20(3), 1–21. 10.17169/fqs-20.3.3392 [DOI] [Google Scholar]
  28. Rouwette E. A. J. A. (2011). Facilitated modelling in strategy development: Measuring the impact on communication, consensus and commitment. Journal of the Operational Research Society, 62(5), 879–887. 10.1057/jors.2010.78 [DOI] [Google Scholar]
  29. Rouwette E. A. J. A., Korzilius H., Vennix J. A. M., Jacobs E. (2011). Modeling as persuasion: The impact of group model building on attitudes and behavior. System Dynamics Review, 27(1), 1–21. 10.1002/sdr.441 [DOI] [Google Scholar]
  30. Rouwette E. A. J. A., Vennix J. A. M., Van Mullekom T. (2002). Group model building effectiveness: A review of assessment studies. System Dynamics Review, 18(1), 5–45. 10.1002/sdr.229 [DOI] [Google Scholar]
  31. Rutter H., Savona N., Glonti K., Bibby J., Cummins S., Finegood D. T., . . . White M. (2017). The need for a complex systems model of evidence for public health. The Lancet, 390(10112), 2602–2604. 10.1016/S0140-6736(17)31267-9 [DOI] [PubMed] [Google Scholar]
  32. Saldaña J. (2009). The coding manual for qualitative researchers. SAGE. [Google Scholar]
  33. Savona N., Macauley T., Aguiar A., Banik A., Boberska M., Brock J., . . . Knai C. (2021). Identifying the views of adolescents in five European countries on the drivers of obesity using group model building. European Journal of Public Health, 31, 391–396. 10.1093/eurpub/ckaa251 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Schreier M. (2014). Qualitative content analysis. In Flick U. (Ed.), The SAGE handbook of qualitative data analysis (pp. 170–183). SAGE. [Google Scholar]
  35. Scott J., Dardas L., Sloane R., Wigington T., Noonan D., Simmons L. A. (2020). Understanding social determinants of cardiometabolic disease risk in rural women. Journal of Community Health, 45(1), 1–9. 10.1007/s10900-019-00710-0 [DOI] [PubMed] [Google Scholar]
  36. Scott R. J., Cavana R. Y., Cameron D. (2013). Evaluating immediate and long-term impacts of qualitative group model building workshops on participants' mental models. System Dynamics Review, 29(4), 216–236. [Google Scholar]
  37. Scott R. J., Cavana R. Y., Cameron D. (2016. a). Mechanisms for understanding mental model change in group model building. Systems Research & Behavioral Science, 33(1), 100–118. 10.1002/sres.2303 [DOI] [Google Scholar]
  38. Scott R. J., Cavana R. Y., Cameron D. (2016. b). Recent evidence on the effectiveness of group model building. European Journal of Operational Research, 249(3), 908–918. 10.1016/j.ejor.2015.06.078 [DOI] [Google Scholar]
  39. WikiBooks. (2022). Scriptapedia. https://en.wikibooks.org/wiki/Scriptapedia

Articles from Health Education & Behavior are provided here courtesy of SAGE Publications

RESOURCES