Understanding how designing physical data visualisations can influence behaviour change: a case study on consumer food waste reduction in Australia

Data visualisations from physical self-trackers have been studied as persuasive tools for behaviour change by encouraging reflection and action. This paper presents the preliminary findings of a longitudinal nine-week study among 30 households in Australia (104 participants with a control group) using data visualisations from physical self-trackers to reduce food waste. The study focused on households with working parents and their children, identified as the primary contributors to food waste. Combining theories of consumer behaviour psychology, interaction design and data humanism to form the basis of the research methodology, participants were given visual nudges and information to encourage food-saving behaviours, and physical self-trackers to visualise and share their actions. Results were triangulated through content and thematic analysis, and paired samples frequency comparison to provide comprehensive insights. The group with access to the probe was able to sustain newly formed habits for a longer duration compared to the control group. Creating their own methods of encoding data enabled agency, holding households accountable for their own actions and facilitating a deeper understanding of their own unique dynamics and collective behaviours that lead to food waste. Furthermore, the social and physical aspects of the study emerged as the most influential factors in driving long-term behaviour change.


Introduction
'Food waste' refers to edible food of good quality that is unconsumed and discarded due to spoilage or expiration.It primarily occurs at the retail and consumer stages within the food supply chain (Forbes et al., 2021).In Australia, consumer food waste poses significant economic, environmental, and social challenges.It is responsible for 3% of greenhouse gas emissions and incurs an annual cost of $18 billion (Van Biene et al., 2021).Consumers are the primary contributors, discarding approximately 20% of the food they purchase, which accounts for up to 40% of household waste (Mahmudul et al., 2022, p.2). Karunasena et al. (2022) profiled three consumer groups related to food wasting behaviours in Australian households, which informed participant recruitment for this study.The first profile, Over Providers, consisted of young families with working parents and children, generating the highest

Literature Review
This section discusses how behaviour can be explained by the Motivation, Opportunity and Ability (MOA) framework and why this is an effective model to use as a basis for the design of the selftracker.It then explores three areas borrowed from human computer interaction (HCI) -play, data physicalisation, and vernacular design -and how these can influence behaviour using the MOA as a lens.The discussion ends with data humanism, a concept that encapsulates these different areas, and references the "Dear Data" project as a relevant example.
The MOA framework proposes that behaviour change occurs when individuals perceive its relevance and consequences (motivation), have accessible options to support the behaviour (opportunity), and possess the necessary skills (ability) (Soma, Li & Maclaren, 2021, p2).Figure1 below explains the basic relationship between motivations, opportunities and attitudes and how these an influence behaviour.The framework has been used in a number of studies to understand factors that shape behaviours leading to food waste (Figure 2).Van Geffen et al. (2020) looked at food waste as a consequence of competing motivations, lack of opportunities, and insufficient abilities.The study revealed that social goals, such as financial status, successful dieting, and being organised, competed against moral, financial, and environmental motivations for reducing food waste.In terms of opportunities, the results showed that systemic issues hindered them from reducing their levels of food waste: unforeseen events (leaving food at home unused), lack of proper storage containers to prolong shelflife, inaccessibility of grocery stores and overly large package sizes.Lack of abilities was also cited by the study as a hindrance to food waste reduction: the majority of participants felt that if they could improve the accuracy of their planning and cooking skills and have knowledge on estimating food safety and prolonging shelf-life of edible products.Another study conducted by Soma Li & Maclaren (2021) used the MOA framework (Figure 3) to evaluate three food waste interventions in Canada over 12 weeks: 1) information-only (booklet, newsletters delivered via email, fridge magnet as a nudge), 2) community engagement (learning workshops) + information, and 3) gamification (online trivia game) + information (Figure 3).While they reported a general increase in motivation after engaging in all three interventions, some participants did not want to change established habits and felt unmotivated due to information fatigue and were overwhelmed by the amount of information in the materials.They did however feel that nudges, rather than information, were important in shaping their behaviour.The fridge magnet in particular was found to play the same nudging role as the online trivia game.Several participants noted that the changes they made had nothing to do with improved awareness or information.
Based on the previous examples, the MOA framework suggests that behaviour change requires individuals to possess physical and psychological abilities, have external opportunities in their environments, and have the motivation to act.We propose that to address all three aspects of the MOA framework more effectively, three areas of HCI can be used to influence behaviour: play, data physicalisation and vernacular design.The next section describes these three areas briefly to give rationale of the design of a kit consisting of visual nudges and informational materials encouraging all household members to engage in food waste reduction behaviours and track their progress using an analogue self-tracker.
Play and ambiguity: Despite a surge in popularity of gamification and games that support behaviour change, Van der Kooij, et al (2015, p.53) found contradictory evidence of their efficacy: while anecdotal findings confirm their motivational value, most quantitative findings from randomized controlled trials have yielded negative results or results that are difficult to interpret.Heimann & Roepstorff (2018, p.12) also support this stance from a behavioural psychology point of view: the main kind of motivation established in gamification is extrinsic (i.e., external rewards), rather than intrinsic (i.e., internal motivation), and that the benefits may not transfer outside the specific context of the game.Play on the other hand, rather than rigid rules imposed by a game designer, is proposed as a more effective means to enhance intrinsic motivation.According to Breen et al. (2021, p.163), play is characterized by selfmotivation and voluntary engagement, promoting exploration, experimentation, and novel ways of interacting with objects and space, akin to how young children approach play.Heimann & Roepstorff (2018, p.1) emphasise the importance of autonomy and agency in adopting a playful mindset, enabling self-awareness and enhancing the user experience for behaviour change interventions.Ambiguity, despite being often seen negatively in HCI, can be valuable for playful design.Gaver, Beaver & Benford (2003, p.223) propose that ambiguity promotes personal engagement with systems and artefacts.By allowing users to interpret imprecise interactions, inaccurate sensors, inexact mappings, and lowresolution displays, ambiguity encourages users to understand them with their own interpretations and beliefs.Ambiguity therefore aligns with the study's aim of designing analog habit trackers for a deeper understanding of behaviors related to household food waste.
Data physicalisation: Data physicalisation is the study of representing data through tangible objects that encode self-gathered data using material properties.It often involves the body and senses, enabling reflection on behaviour in new ways (Khot, Hjorth & Mueller, 2020, p.3).Without a screen as a buffer, data physicalisations can be more perceptible and easily interpretable to owners and others interacting with them (Lupton 2017, p.1600).Lupton (2017, p.1611) describes creating engaging objects to challenge assumptions and engage with data-driven issues.Different approaches, such as using bespoke, playful forms or everyday materials, can provide opportunities for people to connect with their data in meaningful ways (D'Ignazio & Bhargava, 2015; Thudt et al., 2018).Studies have shown that manual crafting of data representations promotes self-awareness and behaviour change, because it facilitates an understanding of the link between data collection, visualisation, and selfreflection.Personable and affective physical visualisations in a personal context hold promise for motivating people in collecting and reflecting on their everyday life data (Botros et al., 2016, p.113).Creating data physicalisations together can empower individuals and foster social interaction.Nissen & Bowers (2015, p.2467) suggest that collaborative digital fabrication of personal data within a group can invest meaning into artefacts, facilitate social interaction, reflect on activities, and explore new forms of data representation.Lupton (2017, p.1609) highlights the potential of communal multisensory data physicalisation to enhance individuals' sense of control over personal data and reduce feelings of individualism.Gourlet & Dassé (2017, p.258) note that these physical data collection methods not only incentivize survey participation but also improve the survey experience, change behaviour, and increase participation rates.They caution that social influence may impact data reliability as previous answers are visible to new participants, potentially influencing response behaviours.
Vernacular visualisation design: We use the term vernacular visualisations as every day, adaptive designs where users or non-data visualisation experts rely on their intuitive visual fluency to create their own encoding systems for data representation.In the context of self-tracking and interaction design, limited literature exists on vernacular visualisations but a study by Snyder et. al (2019) posits that vernacular visualisation should not replace professional practices in data analysis and visualisation, but rather complement existing frameworks to enhance the potential of data visualisations.While these types of visualisations may have technical limitations and lack sufficient information for expert evaluation, they hold value in terms of personal cognitive understanding, insight generation, and sensemaking.
"Dear Data" (Lupi & Posavec, 2016), widely discussed in academic literature and media, is a postcard exchange project that encapsulates theories in play, data physicalisation and vernacular visualisation and how designed artefacts using a combination of these theories can be used as a point of reflection on behaviour (Sorapure & Fauni, 2020; Byrd 2021; Krekhov, Michalski & Krüger 2019).It involved a weekly postcard exchange over a year between designers Giorgia Lupi (New York) and Stefanie Posavec (London), visually representing data on various aspects of their daily lives (objects, phrases, behaviours) (Figure 4).The front of the postcards indicates different instances of experiences, while the back provides a decoding key.This data-driven communication uses data as creative material and a way to engage with the world (Lupi & Posavec, 2016, p.xi).The Dear Data project is relevant to this study because it shows how social interaction can be another factor in influencing behaviour change by increasing motivation to perform a task.It also demonstrates how visualisations designed with familiar materials enable agency, understanding, insight and hold people accountable for their actions.The playful approach to the visualisations emphasises ambiguity's role in critical thinking and personal interpretations.
The following sections bring insight on how designing physical data visualisations using can influence behaviour change, using the case study of reducing consumer food waste in Australia.

Methodology
This study used a longitudinal, mixed method approach that consisted of three research instruments (triangulation)(Figure 5): 1)the design probe, the analogue self-tracking device, deployed in the field over a period of 9 weeks, 2) surveys distributed at three points of measurements (PoMs) to set baseline behaviours and measure the effects of the probe on food saving behaviours after the probe and one month after the probe, and 3) semi-structured interviews to gather qualitative data on the probe's ability to nudge behaviour and support the results of the probe and surveys.The design probe: The analogue tracker, deployed as a cultural probe (Černevičiūtė & Liebutė 2022, p.174), involved open-ended tasks to document participants' daily experiences.The playful and simple design of the self-tracker encouraged participants to creatively document their food-saving behaviours without feeling intimidated.To ease any non-expert audience's apprehension, a staggered approach was used.
Timeline of study: During the first three weeks, participants used coloured markers to indicate their performance of food-saving behaviours using the provided items (Figure 6).Behaviours included checking hunger levels, marking food to be used, using leftovers, involving children in preparation, creating a meal plan, making a shopping list, and following tips or recipes.Participants were given specific symbols to draw when all or none of the behaviours were performed, accounting for laziness or external events.In weeks four to six, participants used googly eye stickers, a black pen, and the provided key to record their food-saving behaviours.This step also allowed respondents to create their own symbol for when they performed all the behaviours, slowly introducing them to the idea of creating their own system of symbols (Figure 7).In last three weeks, participants were encouraged to create their own system of symbols to track their food saving behaviour using materials of their choice (Figure 8).The probe also included various nudges and interventions to support participants in performing actions suggested.The complete the kit that was provided to participants is illustrated in Figure 9.
The complete kit included: 1. Use it Up tape: to mark a section of their fridge or pantry where food needed to be used up.2. Googly eye stickers: to mark individual items of food in the fridge, freezer, or pantry that needed to be used up, making them more visible.3. Use it All book: it provided suggestions on ingredient combinations and maximising their use.4. Meal planner and shopping list: Participants could take the pages with them when shopping, take a photo, or create a digital version.They were encouraged to share the plan and list within the household.The booklet also included tips, information on food waste, and advice from recruited role models.It promoted inclusive food preparation involving children.

Participants selection:
The Fight Food Waste CRC research (Karunasena et 2010) studies, none included control groups.However, this research employed a control group (without a habit tracker) to assess the impact of tracking habits through playful, physical means on food waste behaviour.Six Considerate Planner households served as role models, providing insights for food wasting tips given to Over Providers.Self-reported behaviours before and after the study, and at one-month follow-up, were measured using a survey with Likhert scales.
Surveys: Four surveys were used in the study.The first gathered demographic information to determine eligibility as Over Providers or Considerate Planners.The remaining surveys were conducted at three Points of Measurement (PoMs): two months before the probe as a baseline, immediately after probe completion, and one month later.Practical considerations, such as research milestones and holiday periods, influenced the PoMs chosen.Longitudinal research often adapts PoMs based on various factors, including location, relationships, funding, and resources (Table 2).Semi-structured interviews: Two semi-structured interviews were conducted: one prior to probe deployment and one month later.The first interview aimed to assess participants' motivation levels, determine household members' roles in food preparation and planning, and inform tracker personalisation.Personalised trackers were intended to enhance participant engagement and promote accountability.Additionally, the first interview gathered advice from potential role model households (Considerate Planners) on reducing food waste, which was incorporated into the meal planner document.The post-probe interview aimed to clarify survey and probe results, explore the reasoning behind the symbol system design, and gather additional insights from the tracker.It also provided qualitative feedback on participants' perception of the longitudinal study.Conducting the interview was essential, as seemingly negative survey feedback proved to be valuable.
Participant groups: Participants were divided in 3 groups: Group A, Group B, and Group C. Group A: Over Providers with trackers (N=12 households, 46 participants); Group B: Over providers without trackers (N=12 households, 46 participants); and Group C: Considerate Planners (N=6 households, 12 participants).The participants in groups A and C posted the trackers back to the researchers via a post office box every three weeks.The trackers were then scanned, de-identified and uploaded to a shared Miro board for each household to view.The experiment group (Group A, 46 participants) were given parts 1 and 2, while the control group (Group B, 46 participants) were given only part 1. Changes in adopting food-saving behaviours between Groups A and B were compared through paired samples frequency analysis.A third group consisting of couples 55 years and over without children (Group C) were also given parts 1 and 2, to serve as social influencers to the experiment group.An overview of these interactions is found in Figure 11.

Results and Discussion
As previously mentioned, our study explored the use of vernacular data visualisations, that is, representations of data created by non-experts using their own visual fluency, to track habits that lead to eventual behaviour change in adopting environmentally sustainable practices.Prompted by the kit materials, the aim was to use data visualisation and physical self-trackers as persuasive tools for behaviour change as they have been found to encourage reflection and action.List of instructions for using the self-tracker can be found in Figure 12.

Visual and textual content analysed through inductive content analysis
The visual and text-based data from the design probe (9-week habit tracker) and photos that participants had voluntarily sent through were analysed using content.Highly varied, rich qualitative visual data was received, with participants visually representing their food saving behaviours in aside range of ways, from sheep to rainbows to numerals (Figure 13).The trackers were analysed in terms of their visual content and how participants chose to represent their data.Twelve households used indexicality as a method of signifying their behaviours.An index is a sign that shows evidence of the concept or object being represented and while it does not bear visual resemblance to it, it points towards something that implies that object or concept (Peirce, 1991).As behaviours can be quite complicated to portray accurately in a small space, indexical signs like check marks, crosses, stars and emojis were the preferred choice.Only two households attempted to depict behaviours directly using icons (drawings of a list).Nine households opted for abstracted lines and shapes to indicate specific behaviours, but for combined behaviours like "Yes, we did all of these behaviours today" or "No, we were too tired or lazy to do any of these today", they opted for indexical representation using emojis or drawings alluding to human facial expressions.Examples are shown in Figure 16.
The predominance of the use of smiley or frowny faces resulted in a high number of indexical, anthropomorphic signs.This could be indicative of the influence of communication using mobile phone technology.Furthermore, despite the study focusing on habits around food, only four households chose to represent their data using food-related images.The use of anthropomorphism indicates that behaviours are centred around emotion and feelings rather than what the behaviour is about.The use of metaphors and abstractions also points towards taking action being given precedence over what the behaviour is about.Ten households opted to use a combination of joined symbols (i.e., when different elements formed a whole image) to encode how specific behaviours related to each other and discrete symbols to encode combined behaviours.This is a significant finding as the precedents in weeks 1 to 6 clearly indicate joined symbols for both specific and combined behaviours.Also relevant is the fact that all ten households used emoji or emoji-related signs to encode combined behaviours, again pointing towards affect as a reward or punishment for completing all or none of the behaviours.Only one household used numbers to represent specific behaviours, but still used emojis to represent combined behaviours.The signs chosen by the four households who elected purely discrete method of encoding were selected by the child/children in the household.This indicates that children may not be able to grasp how different behaviours relate to each other.Where children and adults answered together, the method of encoding was joined.In terms of medium and hue to depict behaviours, only one household used purely physical materials (stamps and stickers), indicating how convenience is an important factor in engaging with the trackers.Only two households chose only black to represent their data, indicating how colour is an important factor in delineating different types of data.Two-thirds of the households chose to completely stray from the method of encoding presented to them in Weeks 1 to 6.This is points towards using vernacularity and selecting their own method of representing data instils a sense of agency and accountability for one's own action.
The highest number of voluntary comments focused on explaining or justifying "external events," suggesting feelings of guilt or responsibility.Participants also voluntarily reported using leftovers or ingredients they would typically discard, indicating these behaviours as strong internal measures of food-saving habits.

Categorisation of participants' visual approaches
Content analysis is an observational research method commonly used in communications and media studies for quantifying recorded visual, verbal or audio texts or a combination of these using explicitly defined variables.Each variable consists of values, categories of the same logical kind within each value (Bell, 2004).As a technique for making replicable inferences from the contexts of their use, these variables can be determined by examining content that is either: 1) contained in the text, 2) a property of the source of text or 3) that has emerged in the process of a researcher's analysis (Krippendorff, 2013).In the preliminary analysis of data in this research, the variables were based on the first and third types of content, using theories from semiotic and communication design studies as a lens.Signifier for example, follows Peirce's (1991) categories of signs -icon (has a physical resemblance to the object being represented), index (shows evidence of the concept or object being represented and while it does not bear visual resemblance to it, it points towards something that implies that object or concept) and symbol (a type of sign that does not have any resemblance to the concept or object and its meaning must be culturally learnt).Representation on the other hand has emerged as a variable from analysing visual content of the drawings in the context of food waste and elements in this study -food, objects other than food, anthropomorphism and total abstraction, like lines, dots and other ambiguous symbols.Using a deductive method of determining variables, the visual approaches were categorised according to signifier, representation, encoding, medium, hue and originality as illustrated in Figure 17.The themes were metaphor, the type of signifier, the representation and encoding method used, as well as how each behaviour was portrayed through medium and hue.Originality of the symbols -whether participants copied methods of encoding set by us, the researchers, or whether it was completely original, was another factor taken into consideration.the visual approaches were categorised according to signifier, representation, encoding, medium, hue and originality.A summary of these categories can be found in Table 3.
Table 3: Variables used to categorise participants' visual approaches

Thematic analysis of semi-structures interviews.
The transcripts of the semi-structured interview conducted at the one-month follow-up were analysed using thematic analysis.It is important to note that time for engaging with the trackers was necessary in forming new habits emerged also as a theme as well as the factors, such as complementary goals, situational factors and social conditions growing up, that influenced the formation of new food-saving habits.A more detailed overview of emerging themes, codes, subcodes and quotes from participants can be found in Table 4. First, subcodes, codes, and themes were made based on participants comments.For example, some subcodes included: new behaviours and habits as a result of the tracker; the selftracker as catalyst; new behaviours and habits as a result of other items in the probe.These then were synthesised to codes: Impact on behaviour, there was a perception that food waste was reduced despite no quantitative measures, continuing lessons learnt into the future."The whole process was quite enjoyable and seeing our habits shift and being ingrained now and using the tips and tricks that we've picked up through the process is fantastic."OP06 "I've noticed habits of the way my kids eat, that I hadn't thought about or notice really before."OP08 Having a tracker allowed agency and made households accountable for their actions.
The tracker was necessary in developing habits.
"If I didn't have them, I would not probably be as diligent."OP02 "I felt like the tracker was kind of holding me to task making sure grip because my behavior was reflected, kind of on the tracker, that was happening.But also, just the planning out every weekend, what was going to happen across the next week?Yeah, mapping it out."OP05 "I think the accountability of the tracker helps you realize that you're following something and that there's a plan.So, I think the tracker is important.Otherwise, it's really, it's like buying a cookbook and using it for one recipe and then putting it on a shelf.Right.Whereas if you were told, you know, to use this three times a week and tracked it attracted, you'd use it three times a week."OP08 Those without the tracker: "(A tracker) will force me to do it more.There is a little pressure to do something more."OP13 "I think we like, okay, engage really well with them at the start.And then we dipped a little bit.And then we were like, oh, we should be doing this better.And then we engaged again.So a tracker might have helped through the middle part."OP18 "I was feeling like, I'm not doing enough.And maybe some sort of tracker, or more of even when you send me you know, those reminder emails could have helped?"OP20 Vernacularity and playfulness were crucial factors in leading to insight.
It led to being able to understand a concept through visual metaphors "Because like it 'cos you know how like ninjas they have to, like, be like, so precise and like so like, you have to like, try and like make like the right choices so that you don't like create danger on a mission or whatever."OP01, 9 years old "Helen's favourite animal is a sheep.And she came up with the concept of a sheep with like, each of the elements representing a different aspect.So, the feet, the blue, and the wool part.And then we realized that some didn't have legs because whatever the legs represented, it meant that we didn't do one behaviour.So, in fact the sheep was 'asleep' when we didn't complete all the behaviours for that day."OP08 "When I got to doing the little triangles, I realized I could use them to like indicate time, like going backwards or looking forwards going backwards for leftovers or looking forwards for planning."OP12 A mix of both inductive and deductive approaches were used in identifying the themes.The main themes that emerged are as follows and are illustrated in Figure 18: 1.
The self-tracker had a significant impact on food-wasting behaviours.

2.
The self-tracking visualisations enabled reflection on behaviour.

3.
Having a tracker allowed agency and made households accountable for their actions.4.
Vernacularity and playfulness were crucial factors in leading to insight. 5.
Designing analogue self-tracking data visualisations enabled emotional, intellectual, social and physical engagement.
Figure 18 A summary of the themes from semi-structured participant interviews Having children as respondents supported the argument that vernacular visualisations allow for more democratic access to data-driven narratives, regardless of age, visual and numerical literacy.Karunasena & Pearson's (2022) research recommended that families with children need 'personalised food waste audits to make the problem real.'The design probe responded to this, with members of the household collectively constructing bespoke data representations which served as an audit on food waste behaviours and encourage reflection.It is important to note, however, that the findings revealed that only two families would have preferred an audit of actual numerical food wasted or dollar values on food waste but nearly all reported that food waste in their households have significantly reduced using different benchmarks (i.e., being able to see the back of the fridge, emptying the bin less often).This points towards the importance of habit formation and performing actions in the behaviour change process rather than quantifiable results.

Frequency analysis
As mentioned in the methods section, measurements of food-saving behaviour were taken before probe deployment, immediately after probe deployment, then a follow-up a month after.The focus of the study was on measuring the following food-saving (rather than food-wasting) behaviours: B1 Plan the meals to be cooked for a set number of days B2 Involve other adult members of the household in meal planning B3 Make a complete shopping list before food shopping B4 Have a back-up plan that considers last minute changes B5 Stick to a shopping list and only buy what is needed B6 Check hunger levels before cooking B7 Involve other adult members of the household in meal preparation B8 Check hunger levels before serving B9 Use oldest food first B10 Use leftovers when cooking B11 Arrange/label/mark food in fridge/freezer/pantry that needs to be used up B12 Involve children in the household in meal planning The number of daily actions to track on the outset might seem overwhelming, but all these were aimed at achieving the three behaviours of preparing the right amount of food, eating leftovers and storing food properly.Table 5 shows a summary of how each action leads to these food-saving behaviours.The large list of actions also pointed towards the fact that food waste can be reduced through a series of actions performed collectively by all members of the household everyday over a period of time, as opposed to doing larger actions less often.This was an insight that most participants came to realise after having engaged with the trackers.

Table 5. Summary on how daily actions can contribute to behaviour change.
A paired samples frequency comparison was conducted between Groups A and B, from pre-probe to post-probe, then post-probe to follow-up.Only the results for "Almost all the time (over 90%)" were considered as this points to behaviours performed consistently every day.While the results show that both groups show an increase in frequency in most of the 12 food-saving behaviours, the households in the experiment group with the tracker showed that they were able to sustain or increase the behaviours a month after.With the exception of Behaviour 4: "Have a back-up plan that considers changes in plans", the control group without the tracker showed less households were able to maintain their behaviours a month after the probe was deployed.
The self-tracker had a significant impact on food-wasting behaviours, with households who were given the tracker being able to sustain new habits longer term.New behaviours and habits were formed as a result of using the tracker, with using a meal planner, checking hunger levels before serving and arranging/labelling food that needs to be used up, being the top 3 behaviours adopted.These habits needed to be easy to implement and attached to already existing behaviours.By being able to use their own visual fluency in representing their food-saving data, the participants were intellectually, physically, socially and emotionally engaged with the tracker, holding them more accountable for their actions.The discrepancy in Behaviour 4: "Have a back-up plan that considers last-minute changes in plans" could be due to the fact that Group A didn't consider having a back-up plan necessary as they were already forming other new habits like eating leftovers or labelling food that needs to be used up, which can be considered an alternative to back-up plans.There was also a sharp increase in the number of households in Group B despite not having the tracker, indicating that encouraging people to have backup plans is an intervention that doesn't need constant reminding.Behaviours 5: "Make a shopping list and buy only what is needed", 7: "Check hunger levels before serving" and 11: "Arrange/label/mark food in fridge/freezer/pantry that needs to be used up" were the behaviours in Group A that saw the biggest increase in adoption and habit formation.These behaviours point to the fact that these were easily adoptable and required the least amount of time and effort in implementing around busy schedules.These are also behaviours that are also linked to already existing activities, like shopping, cooking and storage.They require very little or no additional resources to implement.
Other findings to note were that in Behaviours 2 and 7 which required households to involve other members of the family in meal planning and preparation, there were low counts in both Groups A and B. this suggests that meal prep is mostly done by just one person rather than a shared task.There were also no responses for the rarely/never category in Behaviours 10 and 11: "Use oldest food first" and "Use leftovers when cooking" indicating that these behaviours were already being done by most households and implying that they understood that this behaviour can lead to food waste.The results of each of the three methods used in this research reveals a confirmatory pattern in the observed data.They validate and confirm findings, leading to understanding of the primary aim of this study to understand how the design of a physical self-tracker can influence food wasting behaviour.

The role of playful, physical, vernacular self-trackers in influencing consumer food wasting behaviour.
The self-tracker had a significant impact on food-wasting behaviours, with households who were given the tracker being able to sustain new habits longer term.New behaviours and habits were formed as a result of using the tracker, with using a meal planner, checking hunger levels before serving and arranging/labelling food that needs to be used up, being the top three behaviours adopted.These habits needed to be easy to implement and attached to already existing behaviours.By being able to use their own visual fluency in representing their food-saving data, the participants were intellectually, physically, socially and emotionally engaged with the tracker, holding them more accountable for their actions.Using their own metaphors also allowed deeper insight into how specific behaviours related to each other and an understanding of how specific actions need to work together to form a whole.Collectively designing the tracker with family members was found to have a significant impact on consumer food wasting behaviour.Getting everyone on board as well as realising that foodsaving goals need to be addressed by the whole household was an insight revealed in both the semistructured interviews and the probe data.The trackers also revealed that the action of doing something frequently and taking pride that action leads to long-term behaviour change; whether that action is related to food waste or another topic is almost irrelevant.

4.4.2.
Factors that influence motivation, reflection, and behaviour change in food.Having complementary goals, like going on a health kick or new year's resolutions made food-saving habits easier to implement.Another factor that made behaviour change easier to achieve was the presence of related situational circumstances, such as a newly implemented government scheme to introduce green bins or a recent increase in food prices.A further confounding factor were socio-economic backgrounds of the participants growing up: those not born in Australia seemed to have formed more food-saving habits prior to the study.Length of time in engaging with the tracker was also seen to have a significant effect, with participants stating that if the study had been shorter, new habits would not have been adopted.

Determining the value of vernacular self-trackers in promoting behaviours aimed at sustainable practices.
The results that have come out of these research methods informs a new framework that proposes the value of vernacular self-tracking data representations in influencing behaviour.This model was based on an existing framework put forward by Wun, Hoffmann & Rodgers (2019, p.15) that suggests that the value of physical data representations take into account the 'emotional qualities of experiences are part of sense-making'.Their approach considers hedonic qualities and values in a manner that goes beyond data representations' traditional focus on efficiency, comprehension or insight: V = C + E (A, P, I, S), where the value of a data representation derives from: C: its creativity.In terms of introducing new and original ideas.E: its ability to engage beyond the raw information content: A: affective (emotional) engagement.P: physical interaction being invited through touch and movement, real or imagined; I: intellectual engagement.S: social engagement.
From the analysis of the results of this study, a new framework is proposed: V(sb) = U + E + L + CV where, V(sb): the value of a physical data visualisation in encouraging behaviour aimed at sustainable practices U: the uniqueness in its design, brought about by playfulness and vernacularity E: its capacity for engagement in an affective, physical, intellectual and social sense L: length of time it takes for users to create and engage with the physical visualisation CV: complementary variables such as interrelated goals, situational circumstances and social background.
We propose to validate it by an in-depth discussion of each of the elements' connection to the literature on the role of play, vernacular design, data physicalisation and data humanism in reflection and behaviour change.As the intersection of these areas have not been established before at this time of writing, each of these areas will be linked to the findings and analysis.Relationships to the MOA framework will also be established in terms of how the design of the kit and the trackers have influenced motivation, opportunity and ability of the participants.It is also recommended that this study design be replicated in other areas where consumer behaviour can have a positive impact on the environment such as recycling or taking public transport.

Conclusion
The design strategy involved a kit with nudging techniques and concise information to enhance participants' ability and provided them with an opportunity to reduce food waste.An analogue selftracker was also included in the kit to track their food-saving behaviours as a result of the nudges.The visualisations arising from the tracker served as a motivating factor, offering opportunities to change behaviours in the context of food waste.Social influence within households as well as across households promoted positive change among participants, increasing motivation, engaging broader audiences, fostering conversation, and cultivating long-term habit formation.To sustain participant interest and promote sustainable habits, the study incorporated play as a strategy to keep all participants engaged throughout the nine-week longitudinal study.By leveraging participants' existing skills, the research emphasised the value of vernacular visualisations in meaning-making, reflection, and personal accountability.These findings have been captured in a new framework for the value of physical data visualisations in encouraging behaviours aimed at sustainable practices, V(sb).It can be seen as the sum of the uniqueness in its design (U) brought about by playfulness and vernacularity, capacity for engagement (E) in an affective, physical, intellectual and social sense, length of time (L) it takes for users to create and engage with the physical visualisation and complementary variables (CV) such as interrelated goals, situational circumstances and social background.Analysing the data in more depth is necessary as is testing the framework in a different context in sustainability to bring more insight on how to design vernacular self-tracking visualisations that influence behaviour change.

Acknowledgement
We would like to thank all the participants for their time in this study.The work has been supported by the Fight Food Waste Cooperative Research Centre whose activities are funded by the Australian Government's Cooperative Research Centre Program.

Figure 2
Figure 2 How factors in MOA affect food waste (Van Geffen et al. 2020, p.4)

Figure 5 :
Figure 5: Triangulation of mixed methods

Figure 6 :Figure 7 :
Figure 6: Self-tracking cards to document food-saving behaviours using symbols provided (Photograph by Regine Abos)

Figure 8 :
Figure 8: Self-tracking cards to document food-saving behaviours using participants' own system of symbols (Photograph by Regine Abos)

Figure 9 :Table 1
Figure 9: Complete kit provided to participants (Photograph by Regine Abos)

Table 2 :
Rationale for selection of PoMs Each survey included a set of questions using a 5-point Likert scale to assess participants' frequency of performing food-saving behaviours.Considerate Planners were not asked about involving children in meal planning, as they were identified as empty nesters.The survey questions are illustrated in Figure 10.ISDRS-2023 IOP Conf.Series: Earth and Environmental Science 1304 (2024) 012001 IOP Publishing doi:10.1088/1755-1315/1304/1/01200113

Figure 11 :
Figure 11: Overview of participant groups and interactions

Figure 12 :
Figure 12: Instructions provided as part of the participant kit (Photograph by Regine Abos)

Figure 13 :
Figure 13: Participants' visual representations of food saving tracking behaviours (Photographs by Regine Abos)

Figure 14 :Figure 15
Figure 14: Participants volunteering text-based data from the probe (Photographs provided by Regine Abos)

Figure 15 :
Figure 15: Participants engaging in the study beyond the probe and kit (Photographs provided by research participants)

Figure 16 :
Figure 16: Examples of participants' visual representation of their data (Photographs by Regine Abos)

Figure 17 :
Figure 17: Examples of participants' visual representation to determine variables (Photographs provided by research participants) Results of behaviour 1 to 9 are shown in Figures 19 to 30.

Figure 20 :
Figure 20: B2.Involve other adult members of the household in meal planning

Figure 21 :Figure 22 :
Figure 21: B3.Make a complete shopping list before food shopping

Figure 25 :Figure 27 :Figure 28 :
Figure 25: B7.Involve other adult members of the household in meal preparation

Figure 29 :
Figure 29: B11.Arrange/label/mark food in fridge/freezer/pantry that needs to be used up

Figure 30 :
Figure 30: B12.Involve children in the household in meal planning ISDRS-2023 IOP Conf.Series: Earth and Environmental Science 1304 (2024) 012001 IOP Publishing doi:10.1088/1755-1315/1304/1/01200138 al. 2021, p.23) categorised three consumer profiles in Australian households to guide intervention design for reducing food waste: Over providers, Under Planners, and Considerate Planners (see Introduction section).They used a TwoStep Cluster Analysis to generate cluster groups based on key variables, employing a data-driven behaviour-based segmentation approach.Table 1 provides an overview of the Over Providers segment, respondent percentages, food waste levels, demographic and psychographic profiles, and recommended interventions and behaviours to encourage.Over Providers, while representing only 23% of respondents, had the highest weekly food waste (6.33kg), almost twice that of Considerate Planners.ISDRS-2023 IOP Conf.Series: Earth and Environmental Science 1304 (2024) 012001 IOP Publishing doi:10.1088/1755-1315/1304/1/012001

Table 4 :
Snapshot of thematic analysis including themes, codes, subcodes and quotes from participants