Interviews.

Key informant interviews were conducted between December 2015 and April 2017. Participants were identified through purposive sampling, supplemented by snowball sampling [45]. Purposive sampling was facilitated by the positioning of one of the researchers (BC) in the policy context, which helped with identification of relevant and appropriate interview participants through the researcher’s direct observations and interactions with policy makers. Participants were selected based on their anticipated ability to provide detailed insight and first-hand experience into the policy processes related to each of the included HTV policy interventions (assessed separately for each policy). The total number of interviewees conducted in relation to each of the HTV policy interventions ranged from six (Jamie’s Ministry of Food) to 24 (Land Use Planning Policies) as shown in S1 Appendix

The varying number of interviewees for each policy reflects the differing scale, prominence and number of stakeholders involved in the policy processes regarding each of the HTV initiatives. Participants included: current and former politicians (n = 4); senior ministerial staff (n = 2); government officials at state (n = 26) and local levels (n = 12); academics (n = 4); and senior representatives from key public health organisations (n = 7) and private sector organisations (n = 2), such as food retailers. All but one of the potential interviewees who were invited to participate accepted in relation to each of The Achievement Program, Jamie’s Ministry of Food, LiveLighter® and Healthy Catering Policies investigations, whilst two identified interviewees did not participate in relation to Menu Kilojoule Labelling Legislation and Land Use Planning Policy (n = 6 in total). There were also a small number of potential interviewees (n = 4), identified through the snowball sampling process, who were not contactable in relation to The Achievement Program, Jamie’s Ministry of Food and Menu Kilojoule Labelling Legislation, as they had either left their respective organisations or were on extended leave at the time of data collection.

Interviews were guided by a semi-structured interview schedule (S2 Appendix), which was informed by a review of the literature and the political science frameworks underpinning the study [46]. Participants selected the location of the interviews, which occurred either within a workplace or neutral public setting (e.g., café), except for four participants who elected to participate by phone interview. The duration of interviews ranged from 16 minutes to 75 minutes. Whilst most participants were asked about one specific HTV policy intervention, some interviewees provided insights in relation to multiple policies. Eight participants from the local government level were interviewed in relation to both the Healthy Catering Policies and Land Use Planning Policies.

Documents.

Documents relevant to the decision-making processes of the policies under investigation were obtained through DHHS contacts. Document collection was facilitated by the embedded nature of the research and occurred in an iterative manner, with interviewees asked whether they could suggest any documentation that may be relevant to the study.

Documents included internal policy briefings, reports, consultation papers, evaluation reports and other documents. Additional searches were undertaken to identify relevant public documents of the Victorian Hansard database, which contains public records of parliamentary debates, and of the Factiva database for public media reports and documents. Further details of the documents analysed are provided in S3 Appendix

As with the interviews, the varying number of documents analysed for each policy reflects the differing characteristics and scale of the policy processes regarding each of the HTV initiatives.

Observations.

Open ended narrative field notes were captured by one researcher (BC) through observing and taking notes on daily work-related activities within the HTV policy context from April 2015 until December 2017 [47]. This ‘overt’ observation [45] allowed examination of how public health practitioners and policy actors worked to influence policy related to obesity prevention. All members of the Obesity Prevention Unit consented to this observation. Organisational consent was also obtained.

Analysis of individual policy processes.

The analysis of the policy process for each policy intervention involved a deductive thematic analytical approach underpinned by multiple political science theories. The theories employed were the Advocacy Coalition Framework (ACF) supported by the Institutional and Analysis Framework (IADF), and the Multiple Streams Theory (MST). These theories were chosen based on a systematic review of the application of political science theory to the study of obesity prevention policy processes, which identified the value of each theory and recommended employing multiple theoretical perspectives in obesity prevention policy analysis [46]. Data (interviews, documents and field notes) were systematically examined and coded against key constructs of the selected theories using a defined codebook (available on request), with coding facilitated by the qualitative software NVivo10. Then, as described by Kim and Andersen [32], coded data were ‘micro-analysed’ to identify cause and effect variables, and the relationships between variables. This process involved iteratively documenting variable names, direction of linkages between elements and the association direction (i.e., polarity). These causal structures were used to develop the final graphical representation of the CLDs using Vensim software (Ventana Systems, Harvard, MA). Detailed findings of the individual analyses are reported elsewhere [48–51].

Analysis to produce a meta-CLD.

To enable understanding of common influences on policy processes and opportunities for change within obesity prevention policy processes, a process of comparing the six individual HTV policy system CLDs was undertaken to identify recurrent associations and feedback mechanisms between system components. A meta-CLD of the HTV policy system was produced that represented system dynamics that were consistent across each of the six individual CLDs. The causal relationships and system feedback structures were documented in a simplified manner to enable readability and in an effort to communicate insights effectively [52]. The synthesis process was undertaken using an iterative approach to ensure the CLD was consistent with the findings from the qualitative data [53].

Reinforcing loops.

R1 and R2: Increasing policy actor capacity and capability. Synthesis of the thematic analyses, underpinned by the MST and ACF theories, helped to identify the importance of policy actors (either inside or outside of government) who played entrepreneurial [22,55] or brokerage [56,57] roles to secure policy adoption. Government policy makers and external policy actors that were able to frame policy narratives to align policy proposals to decision-maker beliefs (R1) played an important role in reinforcing policy change. According to interview participants, these actors commonly framed policies as part of a broader suite of complementary initiatives under HTV, particularly where there was criticism of the likely impact of a single policy. Policy documentation and interviews demonstrated how, in some instances, such as the Menu Kilojoule Labelling Legislation, policy makers also emphasised the ways in which the policy supported informed consumer choice, so as to align with the predominant political ideologies of neoliberalism and individual responsibility that were current at the time [48]. This strategic alignment of policies to decision-maker beliefs acted to increase support for policy intervention from decision-makers, in turn, leading to increased adoption of obesity prevention policies. Critically, the level of capability and use of these strategic skills by policy actors increased with each successful policy adoption over time, thus creating a virtuous reinforcing loop. For example, in interviews, policy makers reflected on how, during the development of Menu Kilojoule Labelling Legislation, they learnt to adapt policy framing to align with decision-maker beliefs and increase decision-maker acceptability.

“… The framing around the policy did get a bit more focused in this third [iteration of the policy] around providing Victorians with informed choice. So that kind of wording was quite strong in this government’s, in this Minister’s speaking. So that was a kind of transformation of the framing”

                                                  (DHHS Senior Policy Advisor 1)

A related reinforcing loop was the alignment of policy proposals to other government objectives (R2) by policy actors. This alignment of policy proposals to other government objectives was an important driver of decision-maker support for several HTV policy interventions. For example, Jamie’s Ministry of Food was framed in policy documentation as aligning with broader preventive health policies, such as the Victorian Public Health and Wellbeing Plan, and government objectives related to increased productivity and reduced healthcare costs.

“There is no ‘silver bullet’ to reducing the prevalence of obesity. JMoF [Jamie’s Ministry of Food] can work alongside other government and non-government health initiatives and add value.”

                                                  (DHHS policy documentation, 2011)

These positive associations became virtuous reinforcing loops when the effectiveness of these framing tactics was shared with other public health policy staff or when continuity of staff allowed relevant policy actors to implement learnt successful tactics in future policy processes.

R3: Stakeholder consultation as part of policy development. The case study revealed that government staff involved in crafting policy proposals sought involvement from a diverse range of stakeholders as part of policy development processes. Interview participants noted that decision-makers recognised the advantages of stakeholder consultation in policy development, particularly for the LiveLighter®, Menu Kilojoule Labelling Legislation, and the Achievement Program policy processes, whereby numerous stakeholder groups who would be impacted by the policy were reportedly made to feel that they had an opportunity to contribute to the policy development process [48,49,51]. In addition, stakeholder groups were given the opportunity to learn, at an early stage of the process, about how the proposed policy was likely to impact them [58,59]. Across the HTV policies investigated, consultations resulted in greater commitment to the success of the policy (or at least acquiescence), and, often, shared ownership of success [22]. Consequently, according to interview participants, stakeholder consultations reduced the perceived political risks associated with policy adoption, leading to increased support from decision-makers, and thus creating a reinforcing loop (R3). Indeed, a former Minister identified stakeholder consultation as a critical facilitator in the adoption of the LiveLighter® campaign.

“I think part of [the reason that the policy was adopted] was the process of consultation itself, actually. Often people then feel involved, and they have a deeper understanding of what’s going on, what’s trying to be achieved, any potential negatives and how you might manage those. I think [the facilitating factor] was doing [the consultations]”

                                                  (Former Minister 1)

R4: Policy viability. The MST emphasises the importance of developing viable policy solutions in order for approval to be granted by decision-makers [22]. According to interviewees, policy adoption was enhanced when policy actors developed viable policy proposals that met the requirements of decision-makers. In the case of legislated policy (e.g., Menu Kilojoule Labelling Legislation), prescriptive documentation requirements were met [48]. For several other policy interventions, evidence from relevant contexts (e.g., tobacco policy or other Australian jurisdictions) was utilised to demonstrate implementation feasibility [48,51]. The presentation of viable and feasible policy solutions was described by participants as important in increasing support for policy interventions from decision-makers, leading to increased obesity prevention policy adoption. Over time, successful implementation of HTV policies increased the capacity and capability of policy actors to develop viable solutions, thus creating a virtuous loop (R4).

R5 and R6: Utilisation of policy evidence. The ACF’s notion of ‘policy-oriented learning’, whereby new evidence can shape decision-maker beliefs and result in policy change, was helpful in highlighting the role that evidence played in policy adoption [57]. Interviews and documents revealed that evidence of various forms were utilised in the HTV policy processes. Evidence of policy impact (R5), particularly evidence generated through evaluations of ‘real-world’ obesity prevention policy adoption and implementation, was identified as an important part of securing decision-maker support [60]. According to participants, evidence from evaluations of implementation elsewhere in Australia was used to demonstrate the likely positive impact and increase acceptability of several HTV policies. For example, evidence of the impact of the LiveLighter® campaign in Western Australia was reported to have helped convince decision-makers that the policy was likely to be effective in Victoria.

“We’ve seen the [Western Australia] material [for LiveLighter®], and we’re able to adapt that… There was good research behind [the LiveLighter® policy] to suggest that it may well be effective”

                                                  (Former Minister 1)

In addition, evidence of implementation feasibility (R6), often independent of evidence of policy effectiveness, was noted by participants as critical in regards to several policy interventions, including the Achievement Program, Jamie’s Ministry of Food, Live Lighter®, and Menu Kilojoule Labelling Legislation [48,49,51]. An example of the use of evidence of implementation feasibility was in relation to Menu Kilojoule Labelling Legislation, where government policy makers introducing the policy emphasised the established nature of similar policies in other Australian jurisdictions.

“Kilojoule labelling laws have been in operating for nearly five years now in New South Wales. The scheme is well understood and accepted by industry. Similarly, we expect a smooth transition to the laws in Victoria. It is estimated that 80 per cent of the businesses required by the legislation to display kilojoule content area already doing so in other states. Therefore, the measurements have already been calculated and will be easily available”

                                                  (Victorian Parliament Hansard, Legislative Council, 13 October 2016)

This use of evidence of policy impact was in addition to the provision of evidence of the policy problem as part of almost all documentation with respect to each policy intervention. For example, evidence of the high prevalence of obesity, and its health and economic consequences, was frequently cited in policy documentation. The various forms of evidence were important in increasing the support for obesity prevention policy options among external stakeholders and decision-makers. These virtuous reinforcing loops (R5 and R6) contributed to policy adoption in all the HTV policies, except for the Land Use Planning Policies, for which there was insufficient evidence regarding the causal association between fast food outlet density and obesity to convince decision-makers of the need for policy change in respect of fast food outlets.

R7: ‘Softening-up’ of key stakeholders. The MST theory suggests that ‘softening up’ processes, whereby policy proposals are discussed with stakeholders who have to ‘soften’ up to new policy ideas, is an important method to increase the acceptability of policy solutions [22, p. 123]. There was strong evidence from interviews for this effect across the HTV policy process investigations, with examples of internal government staff members facilitating network coordination and negotiation processes with stakeholders to reduce resistance to policy options. In the adoption of the Achievement Program, policy makers commented on how they strategically engaged stakeholders from the Department of Education early and throughout the policy process in order to increase their support.

“[The] original steering group only had two representatives from the Department of Health and eight from Department of Education. We deliberately sought to shift some of that power to Education as a strategy to engage them, to give them a bit more power and control over the direction to inform it right from the beginning”

                                                  (DHHS Senior Policy Advisor 2)

Interviewees identified that the support for a policy from key stakeholders helped to reduce the political risks of that policy, which, in turn, increased the support for the policy from decision-makers. The processes regarding the adoption of LiveLighter® also demonstrated how some decision-makers (e.g., a Minister) who supported the policy also worked to soften-up other stakeholders and decision-makers (e.g., other members of parliament) in order to increase support and reduce the potential for backlash [51]. This process created a virtuous reinforcing loop over time (R7) to increase support for obesity prevention policy adoption.

Balancing loops.

B1: Organisational culture of risk aversion. The ACF analysis, supported by the IADF, was particularly useful in highlighting the role of the organisational norm of risk aversion in resisting policy change. Across HTV policy process investigations, organisational risk aversion resulted in an increased number of policy proposal reviews, undertaken to reduce the perceived political risk of policy change. This risk aversion was particularly evident for the more controversial LiveLighter® social marketing campaign and the Menu Kilojoule Labelling Legislation whereby additional consultation or modified policy proposals were required to overcome perceived risks [48,51]. In relation to the Land Use Planning Policies, interview participants reflected on how organisational risk aversion increased reluctance to attempt policy changes unless they were aligned with decision-maker attitudes.

“We were going to do a policy around limiting fast food outlets and put in a Land Use Planning Policy review but we felt that [the local government area] wasn’t really ready to listen to that… because what’s happened in the past was that things have tried to get into the review of Land Use Planning Policy, even around food, and it’s been ‘chucked out’. It has got up to the council and a particular Councillor said “you can’t restrict that sort of stuff” and took it out”

                                                  (Local Government Manager 1)

A consequence of the increased number of policy reviews was that these processes increased time for policy change. Interviewees indicated that this created a barrier to policy adoption due to the relatively short-term (4 year) political cycles in Victoria, especially considering that there were changes in the governing political party over the course of the policy study period. The reduced likelihood of policy change, over time, was related to both the shifts in political inclinations of leadership and the flow-on effects that the political party turnover had on the institutional structures.

B2: Time needed for effective consultation. As HTV progressed, experience in successful obesity prevention policy adoption helped increase the capability of policy makers in regard to effective stakeholder consultations. As noted in regards to R3, these capabilities worked to reduce political risks and garner policy support among decision-makers. However, according to interview participants, consultation processes elicited opposition to policy proposals from several different stakeholder groups, including the food industry (in relation to Menu Kilojoule Labelling Legislation), the construction industry (in relation to Land Use Planning Policies) and mental health groups (in relation to LiveLighter®) [48,51]. Interviewees indicated that stakeholder consultation processes resulted in substantially increased time required for policy change, which, as identified earlier (B1), presented a risk to policy adoption. For example, in the case of the Land Use Planning Policies, it was reported that the requirement for public exhibition and consultation increased the complexity of policy change and the likelihood of delayed policy decisions.

“The [strategic planning team] are still dealing with the [public] submissions they’ve had from the [land use planning policy] exhibition, and I think they’re trying to avoid going to Panel [a Planning Panel hearing–a formal process whereby a planning authority reviews proposed amendments]. The trouble with planning scheme amendments is that if one person makes a submission and they won’t withdraw it–even if it’s irrelevant, you can end up at a panel hearing”

                                                  (Local Government Policy Officer 1)

The time needed for effective consultations thus served as a balancing loop that can help explain the policy resistance that needed to be overcome in order for policy change to occur.

B3, B4 and B5: Consequences of a whole-of-government approach. The theoretical analysis, and in particular the IADF, helped illuminate how the provision of increased resources for policy processes, acquired through the HTV funding, enhanced the capacity of obesity prevention policy advocates to engage in whole-of-government policy processes. This was particularly evident at the local government level, where there has historically been limited funding for health promotion workforce capacity. The theoretical analysis also demonstrated that, for policies for which approval was required at a Cabinet level (the LiveLighter® campaign, Menu Kilojoule Labelling Legislation and the Land Use Planning Policies), a collaborative, whole-of-government approach was required as part of policy decision-making [48,51]. However, in undertaking whole-of-government policy development processes, several competing forces emerged across the HTV policy process investigations. Firstly, whole-of-government processes resulted in increased pressure from competing policy issues (B3). For example, with respect to Land Use Planning Policies, whole-of-government consultation processes brought to light multiple other problem representations, such as the need for improved public transport infrastructure, actions on climate change, housing affordability, community and social infrastructure, and protection of agricultural land, among others. Interviewees indicated that this large ‘problem load’ had a substantial negative effect on the potential inclusion of mechanisms to support obesity prevention, with policy makers only able to attend to a limited number of priority issues. Secondly, whole-of-government consultation processes necessitated involvement of a diverse range of stakeholders (B4), including at both state and local government levels, which increased the complexity of policy processes. For example, according to participants, in order for the Healthy Catering Polices to be adopted, decision-maker support and formal approval was required at numerous hierarchical levels and across several departments of multiple organisations. This required policy brokers to convince executive managers of the benefits of the policy for each organisational unit.

Moreover, organisational rules meant that substantial consultation processes within each organisational unit were required before formally proposing the policy to decision-makers. The complexities associated with whole-of-government consultation often served to reduce support for proposed policy interventions from decision-makers. The nature of the required processes to adopt a whole-of-government approach to decision-making also substantially increased the time required for the policy process (B5). As explained above (B1 and B2), the long duration of policy processes reduced the likelihood of policy adoption. Moreover, the time taken for whole-of-government and external consultation reduced the opportunity for policy actors to learn from policy adoption success, thus creating several balancing loops that contributed to policy resistance.