A Systematic Review of Pre-Post Studies Testing Behaviour Change Interventions to Reduce Consumer Food Waste in the Household

Abstract

Since the United Nations announced their Sustainable Development Goal 12.3 to halve per capita food waste by 2030, prevention has become an international focus. Consumers are responsible for a significant portion of food waste, and much of this waste is avoidable by improving food management routines and planning in the household. There is a growing body of research focused on developing and evaluating domestic behaviour change interventions which can improve these behaviours. However, evidence of intervention efficacy on a household level is inconsistent, and best-practice approaches for researchers, policymakers, and practitioners have not been identified. Furthermore, the magnitude of this problem across environmental, social, and economical aspects of life necessitates meaningful long-term change. Many reviews have synthesised household food waste intervention studies, yet there is a gap exploring whether new habits can or will stick. We identify 16 peer-reviewed articles applying behaviour change interventions in the household, with a pre–post design to measure food waste both before and after implementation. The review reveals a paucity of studies that evaluate intervention efficacy relative to their baseline, as well as a significant longitudinal evidence gap. Our recommendation for further research is for the robust replication of effective short-term interventions to be tested longitudinally. Overall, this review outlines potential areas for prioritisation to enable large-scale sustained household behaviour changes in the fight against food waste.

1. Introduction

Globally, a third of all food produced is wasted, and a disproportionate amount of this comes from high-income countries [1,2]. Food intended for human consumption, referred to as edible food, is often disposed of, rather than eaten, across the supply chain, with consumers in the household responsible for contributing the greatest portion [3]. This waste can be avoided by better food planning, shopping, preparation, and storage [4]. Food waste has significant implications on climate change, food insecurity, and wasted resources [5]. Given the magnitude of this phenomenon, policymakers, practitioners, and researchers are increasingly collaborating to improve consumer behaviour in order to remedy it [4,6,7].

A plethora of studies have implemented interventions thought to be effective in changing food waste behaviours, such as information-sharing and education, role modelling, influencing food habits for the better (nudging), and providing feedback [8,9]. These techniques can improve awareness of one’s own food waste [10,11], food management skills [12,13], and provide motivation and opportunity to change behaviour [14,15,16].

There is a growing body of literature recognising the threat of household food waste and proposing appropriate solutions; however, there remains a paucity of evidence validating the efficacy of interventions aimed at addressing it [17,18]. Though the existing research is promising, in comparison to the threat of climate change, the outcomes to this point have been unable to meet the demands of the challenge [19,20]. It is important to determine what works, and what does not, in order to focus resources and funding on areas that are more likely to generate significant and timely positive outcomes [7].

Furthermore, the literature focusing on household waste behaviour changes has been critiqued for inconsistent measurement methods [21] and findings [22,23]; while some intervention studies have been found to be effective, research reveals such improvements in behaviour can wane over time [24,25]. Few behaviour change studies account for the potential of this diminishing improvement in their methodologies [7,26,27]. This is despite researchers having long recognised the difficulty in developing short-term interventions that lead to sustained behavioural changes [28,29] and the breaking of habits [30,31].

A structured and methodical approach is required to facilitate intervention testing, simultaneously by measuring ‘before and after’ repeated measures to determine if behaviour differs at each point [32,33], and measuring again to determine if there is any long-term impact. A growing concern here is the potential of a measurement effect that research itself may have on behaviour, which can be mitigated through the use of a control group [13]. Thus, this review turns to studies using this approach to determine the ways in which interventions to reduce consumer food waste have been implemented in the household.

To the best of the authors’ knowledge, the current article is the first systematic review encompassing empirical food waste studies with both pre- and post-test measures, with the purpose of illuminating promising interventions that have been rigorously applied and validated. This systematic review contributes to the literature in the two following ways: first, it provides an overview of the empirical research measuring the impact of behaviour change interventions on household food waste, and second, it identifies the longitudinal gap in intervention studies where further research is required [34].

2. Materials and Methods

2.1. Search Strategy

A systematic review of peer-reviewed literature was undertaken following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines and recommended PICOS (Problem/Intervention/Comparator/Outcomes/Study design) framework [35,36,37], facilitating a standardised, high-quality approach to systematic reviews [35].

Five large and commonly used online academic databases/search engines (Sage, ScienceDirect, Scopus, ProQuest, and Wiley) were searched, using key words (see search strings in Appendix A). The titles and abstracts of the search results were initially screened to meet criteria, including having been written in English, a peer-reviewed article, a quantitative study, and research conducted in a developed country. The titles and abstracts were verified for suitability (e.g., set in the household, aimed at avoiding or reducing food waste). Finally, the remaining articles were individually screened in full, and reference lists were checked for additional records. Articles were excluded at the first sign of not meeting a PICOS criterion (see Section 2.1.1) [37], and recorded in the PRISMA chart (see Figure 1). Screening included an informal quality appraisal of the articles, determining whether the authors adequately reported the “6Qs”: “what, which, where, how, whereby, whom” [38].

2.1.1. Inclusion and Exclusion

The inclusions and exclusions for the PICOS article screening are outlined in

Table 1. PICOS Inclusion and exclusion criteria for relevant paper appraisal.

Subject	Inclusion and Exclusion Criteria	Keywords	Justification
Population	Include studies of consumers, set in domestic household. Exclude studies of students, universities, canteens, restaurants, retail or work environments.	consumer, household, domestic	Household waste makes up the largest portion of post-purchase food waste in developed countries [39,40]
Population	Adult consumers		Responsibility for major food provisioning primarily consumers aged 18+ [4,12,41]
Population	Include studies set in developed countries. Exclude developing countries.		Food management behaviours found to be similar among developed countries [39,42,43]
Intervention	Include articles aimed at changing consumer food in the household through implementation of an intervention or nudge.	behaviour change interventions, behavior change interventions, behavioural insights, behavioral insights, behaviourally-informed policies, behaviorally informed policies, behaviour change theory, behavioural public policy, behavioral public policy, theory of planned behaviour, theory of planned behavior, TPB, theory of reasoned action, TRA, behavioral intention, subjective norms, perceived behavioral control, attitude, nudges, nudging, nudge	Interventions have successfully been implemented to change consumer behaviour [19]
Comparison	Include articles with or without a control.		To minimise possibility of outcome being as a result of external impact [44]
Outcome	Measured improvement in behaviours associated with reducing or preventing food waste, or food wasted.	reduce food waste, food waste cessation, reduce food disposal, waste management, food waste reduction, FWR, food/waste, food waste/loss, food waste/loss prevention, F-WL, habit, pro-environment, pro-social, sustainable, ethical, socially responsible	Food management linked to food waste [45]. Longevity of change in these rarely evaluated [7]
Study	Include quantitative empirical research. Exclude review papers, conference papers, grey literature, or qualitative research.	quantitative, empirical, mixed methods, mixed research	Journal articles that have been through the peer-review process [4,46]

below.

3. Results

3.1. Search Results

The initial search undertaken in October 2020 yielded 975 articles. From this search, 760 unique articles remained after duplicate records were removed. Due to the time that has lapsed, additional searches were undertaken for publications up until December 2023, resulting in an additional 1424 articles. These supplementary searches revealed overlapping articles from the 2020 search, equating to 792 new articles, and taking the total review to 1764 articles. The titles and abstracts for each of the 1764 records were screened against the inclusion and exclusion criteria (see Table 1), resulting in 950 articles that were irrelevant, leaving 601 articles for full-text screening. Finally, 586 articles were removed from the dataset, based on the criteria shown in Figure 1, resulting in 16 articles for final analysis.

The reviewed articles are shown thematically across Table 2 (Study design and methods) and Table 3 (Study outcomes), followed by results and discussion. This systematic review compares these remaining studies that implemented and evaluated food waste behaviour change interventions in the household. The review identifies the strategies and tools that have been applied, revealing commonalities and limitations in the existing literature, as well as any implications for future research.

3.2. Literature Appraised

3.2.1. Overview

All 16 studies incorporated interventions or treatments targeting food management behaviour at the household level of the supply chain. The strict screening criterion requiring a rigorous pre- and post-intervention measure resulted in fewer-than-expected articles in the final review. Notably, a minority of studies using this design elected to conduct a third measurement, enabling the longitudinal evaluation of interventions.

Figure 1. PRISMA Flow diagram of included studies.

Figure 1. PRISMA Flow diagram of included studies.

Table 2. Study design and methods.

Study	Target Behaviour(s)	Theoretical Underpinning	Intervention(s) and Group(s)	Intervention Methodologies Detailed	Data Collection	State of Food	Intervention Duration	Study Duration
Cooper et al., 2023 [47]	Planning, preparation, cooking, storage	Theory of Planned Behaviour, Motivation Opportunity Ability	Study 1: Intervention a: Information, online goal setting intervention and a salience tool (Group A + B + C) No intervention: (Group D— control group). Study 2: Online and physical intervention (Group A + C) Online only (Group B) No intervention: (Group D—control group) Random group allocation.	Two studies, both included in review. Study 1: Intervention a: Information and goal to make a ‘bonus meal’ with ingredients in the household (Group A + B + C). Intervention b: Increasing salience; track (Group A), collect (Group B), tag (Group C).	Baseline questionnaire with intervention or no intervention, 4 follow-ups, and a longitudinal questionnaire. Measured waste in questionnaires.	Study 1: Edible food limited to fruit, vegetables, bread/grains (drinks excluded). Study 2: As above, plus cheese, eggs, meat, fish	Study 1: Five weeks. Study 2: Five weeks (Group A + B) Three weeks (Group C).	13 weeks. One questionnaire a week. Longitudinal questionnaire eight weeks after final follow-up.
Everitt et al., 2023 [24]	Planning, purchasing, preparation, storage, consuming leftovers (by increasing intention to prevent food waste, perceived behavioural control)	Theory of Planned Behaviour	Intervention a: Awareness and Information (Group A). No intervention: (Group B—control group). All agreed to be recontacted for this study after participating in van der Werf et al.’s study [48]. Random group allocation.	Intervention a: Materials provided in 4 L container with information on local average value and quantity of FW, printed tools and links to website for education on improving behaviours, postcard linking wasted food and wasted money, five emails throughout intervention period.	Waste audit, intervention or no intervention, follow-up waste audit. Measured waste in garbage collection.	Edible and inedible food (dairy included, other drinks excluded)	Intervention a: Two weeks	Longitudinal questionnaire 31 months after intervention.
Gimenez et al., 2023 [33]	All (by increasing awareness of own FW and reducing cognitive dissonance)	Theory of Planned Behaviour, Social Practice Theory	Intervention a: Information (Group A). No intervention (Group B— control group). Random group allocation.	Intervention a: information provided online after questionnaire about FW, with the task to highlight text they agreed and disagreed with (Group A).	Baseline questionnaire with intervention or no intervention, follow-up questionnaire. Measured waste in questionnaires.	Edible food (drinks included)	Intervention a: Provided at single point in time	Two weeks from baseline questionnaire to final questionnaire. No longitudinal follow-up.
Graham-Rowe et al., 2019 [49]	All (by reducing denial of own waste)	Self-affirmation theory	Intervention a: Self-affirmation (Group A). Intervention b: Integrated self-affirmation (Group B). No intervention: (Group C— control group). Random group allocation.	Interventions a + b: shown different lists, asked to identify their most important values. These values were then tied to their perceived behavioural control over their FW, and the impacts on the environment (Groups A + B). Intervention b: shown list and also asked to evaluate why this value was important to them and how it tied in with past behaviour (Group B). Intervention c: shown same list as Intervention a and asked to instead identify the least important value. The value was then tied in with their perceived behavioural control over their FW, and with impacts of FW on the environment (Group C).	Baseline questionnaire with intervention or no intervention, follow-up questionnaire. Measured waste in questionnaires.	Edible fruit and vegetables (drinks unspecified)	Intervention a + b + c: Provided at single point in time	One week from baseline questionnaire to final questionnaire. No longitudinal follow-up.
Pelt et al., 2020 [50]	All (by increasing awareness of own FW and reducing cognitive dissonance)	Theory of Planned Behaviour, Social Practice Theory	Intervention a: Awareness (Group A). Intervention b: Cognitive dissonance (Group B). Intervention c: Information (Group C—control group). Random group allocation.	Interventions a + b + c: door-to-door pamphlet distribution. Pamphlet information on FW consequences and means to reduce it (Groups A + B + C). Intervention a: participants weighed and recorded FW in food diary, to be collected. Weighing instructions provided (Group A). Intervention b: participants asked to preach support for reducing FW, intention implementation plan, and listed recent past behaviours in opposition to this (Group B).	Waste audit, door-to-door visits with Intervention a, Intervention b, or no intervention, follow-up waste audit, longitudinal waste audit. Measured waste in garbage collection.	Edible and potentially edible food (only food in solid form, drinks excluded)	Intervention a: One week. Intervention b: on-the-spot activity	Six weeks from baseline waste audit to final waste audit. One week between audits and intervention. Longitudinal waste audit five weeks after intervention.
Roe et al., 2022 [51]	Planning, shopping, preparation, cooking, storage, consuming leftovers	Theory of Planned Behaviour, Theory of Reasoned Action, Tailored pro-environmental behaviour interventions	Intervention a: App goal setting and FW coaching (Group A). Intervention b: App and unrelated coaching (Group B— control group). Randomly allocated into groups.	Intervention a: One-on-one coaching about FW, with personalised tips and ways to reduce it. Asked to set goals, and received regular reminders by email/text/call, as preferred. Provided app to record FW and instructed on its use (Group A). Intervention b: One-on-one coaching and information provided about unrelated topic matched in intensity of communication. Provided app to record FW with and instructed on its use (Group B).	Training (in-person), baseline app-based FW diary, intervention or no intervention, follow-up app-based FW diary. Measured waste in app.	Edible and inedible food (drinks included)	Intervention a + b: One week	Two or more weeks. One baseline week followed by training, intervention and follow-up measurement week.
Romani et al., 2018 [43]	Food planning and preparation	Theory of Planned Behaviour	Intervention a: Information (Group A + C). Control group: (Group B + D). Solomon four-group design. Group allocation unspecified.	Intervention a: provided a printed/digital article on creating a weekly food menu, with advantages, recipes, and suggestions for family involvement. Two groups involved in pre-test (Group A, Group B), two groups involved an intervention (Group A, Group C), four groups involved in post-test (Group A, Group B, Group C, Group D).	Study 3 in larger body of research. Baseline food diary or no diary, intervention or no intervention, follow-up food diary. Measured waste in diary.	Edible food (milk and other potable dairy products included, other drinks excluded)	Intervention a: Provided at single point in time	Four weeks from pre-test to post-test diary. One week between diaries and intervention. No longitudinal follow-up.
Schmidt 2016 [52]	All	Integrative influence model of pro-environmental behaviour	Intervention a: Commitment and goal setting (Group A). No intervention: (Group B— control group). Randomly allocated into groups (restricted block formation)	Intervention a: provided five behavioural suggestions to reduce FW on an online platform, personalised with prior self-reported behaviour in baseline questionnaire. Then, presented with a public commitment and goal-setting measure and asked to rate willingness to follow recommendation (Group A).	Baseline assessment, intervention or no intervention, follow-up assessment, longitudinal post-test assessment. Measured performance of waste-causing/waste-reducing behaviours.	Edible food (drinks unspecified)	Intervention a: Four weeks (roughly), commencing at varying points in time	Seven months from baseline assessment to second assessment. Two months of recruiting and pre-test, another two between end of recruitment and intervention, approximately four before post-test assessment. Eight weeks after the intervention, a third assessment was undertaken, with low uptake.
Shu et al., 2021 [53]	Planning, storage, preparation, consuming leftovers	Unspecified	Intervention 1a: FW awareness (Group A). Intervention 1b: Food awareness (Group B—control group). Intervention 2a: Information on financial impact of food waste (Group C). Intervention 2b: Information on unrelated topic (Group D—control group). Randomly allocated into groups (data missing, presumed Intervention 2 allocated independently to Intervention 1 allocation)	Intervention 1a: Asked to pay attention to FW for the week preceding first FW measurement (Group A). Intervention 1b: Asked to pay attention to food in household for week preceding first FW measurement (Group B). Intervention 1 (a + b): In first questionnaire measuring FW, participants were provided instructions on how to estimate it based on size of their own hand (i.e., size of thumb, palm, fist). Intervention 2a: Information provided about environmental impact of FW (Group C). Intervention 2b: Information provided about unrelated topic (Group D—quasi-control).	Pre-study communication, baseline questionnaire with intervention or control message with training, follow-up questionnaire with intervention or control message, longitudinal questionnaire. Measured waste in questionnaires after prior notice to monitor FW.	Edible food (drinks included)	Intervention a: One week. Intervention b: On-the-spot message	Baseline questionnaire, one week before follow-up questionnaire, longitudinal questionnaire.
Shu et al., 2023 [54]	Storage	Unspecified	Intervention a + c: FW awareness and storage information (Group A). Intervention b + c: FW awareness and storage and composting information (Group B) Intervention c: FW awareness (Group C—control group) No intervention: (Group D—control group). Allocated into groups A + B + C based on household location in city district and waste collection day. Allocated into group D based on household location outside of district and part of national control.	Intervention a: Information on storage and offered storage containers (Group A + B). Intervention b: Information on storage and compost and offered composting tools (Group B). Intervention c: Information on FW prevention and diversion via community-based communication (Group A + B + C).	Baseline questionnaire, group-specific questions, intervention, follow-up questionnaire. Measured waste in garbage collection, on a community level (all) and household level (additional consent).	Edible food (drinks included)	Intervention a + b: One month. Intervention c: Three months.	Baseline questionnaire, one week before follow-up questionnaire.
Soma et al., 2020 [55]	Planning, shopping, cooking, storage, consuming leftovers (by increasing awareness of own waste, education on date labels)	Expanded Theory of Planned Behaviour	Intervention a: Information (Group A). Intervention a + b: Information and gamification (Group B). Intervention a + c: Information and community workshops (Group C). No Intervention (Group D—control group). Group allocation unspecified.	Intervention a: Information booklet on FW consequences and reduction strategies. Provided fridge magnet as prompt, and newsletters via mail or email with tips (Group A + B + C). Intervention b: Information plus an online quiz about FW using gamification techniques (Group B). Intervention c: Information plus community workshops with group activities, discussions, resources and prizes (Group C).	Waste audit, baseline questionnaire (in person), intervention or no intervention, follow-up questionnaire (online, over phone, or by mail), follow-up waste audit, (n.b. follow-up focus groups). Measured waste in garbage collection.	Edible and inedible food (dairy and non-dairy drinks included)	Intervention a: Provided at single point in time. Intervention b + c: 12 weeks.	12 or more weeks. One week for waste audit, 12 weeks between baseline questionnaire and follow-up questionnaire, one week for follow-up waste audit. Recruitment window unspecified. (N.b. three between follow-up audit and focus group discussions).
van der Werf et al., 2021 [48]	Planning, purchasing, preparation, storage, consuming leftovers (by increasing intention to prevent food waste, perceived behavioural control)	Theory of Planned Behaviour	Intervention a: Awareness and Information (Group A). No intervention: (Group B—control group). Randomly allocated into groups (with minor modification for practicality).	Intervention a: Materials provided in 4 L container with information on local average value and quantity of FW, printed tools and links to website for education on improving behaviours, postcard linking wasted food and wasted money, 5 emails throughout intervention period.	Waste audit, intervention or no intervention, follow-up waste audit. Measured waste in garbage collection.	Edible and inedible food (dairy included, other drinks excluded)	Intervention a: Two weeks	Five weeks from baseline audit to final waste audit. One week for audit, one week before intervention, one week for follow-up waste audit.
van Herpen et al., 2023 [56]	All	Motivation Opportunity Ability	Intervention a: Tools (Group A) Intervention b: Norms (Group B). Randomly allocated into groups.	Intervention a: Tools provided including shopping list notepad, stickers, measuring cup, app, leaflets, recipe cards, fridge thermometer (Group A + B) Intervention b: Motivational messaging with social norms (Group B).	Questionnaire, intervention, follow-up questionnaire.		Two weeks.	One month from baseline to follow-up measure.
Wharton et al., 2021 [57]	Purchasing, cooking, storage (by information, education on date labels)	Theory of Planned Behaviour	Intervention a: Information (Group A). No group allocation.	Intervention a: Provided with information and tools to prevent FW. Weekly themed modules on household food management framed by impact on health, finances, and the environment. Participants trained in-person to weigh and log FW in a standardised way on a provided logbook (hard copy or digital), and provided printed instructions, 3.8 L buckets and a digital scale.	Training and assessment (in person), baseline questionnaire, baseline FW log, intervention with weekly FW log, follow-up FW log, follow-up questionnaire, assessment (in person), (n.b. semi-structured interviews). Waste collected and weighed on scales.	Edible food (drinks excluded)	Five weeks	Seven or more months. Four months of recruitment, seven weeks from baseline to final week. One week for baseline measure, five weeks for intervention, one week for follow-up measure.
Young et al., 2017 [58]	Planning, purchasing, cooking, consuming	Social influence theory	Intervention a: Social influence (Group A). Intervention b: Information 1 (Group B). Intervention c: Information 2 (Group C). No intervention: (Group D—quasi-control group). Group allocation dependent on self-reported prior awareness of intervention materials. Not discrete groups.	Intervention a: Leftovers campaign on supermarket’s Facebook asking customers to submit their favourite recipes, provided link to tips from WRAP, promote discussion. Intervention b: Supermarket magazine (available in store and online, and distributed widely) featuring tips on reducing FW and recipes for commonly wasted foods. Intervention c: Supermarket e-newsletter (distributed widely) discussed using leftovers and provided link to Intervention a social media campaign, encouraging customers to take part. Provided storage tips and a link for purchase.	Baseline questionnaire, intervention, follow-up questionnaire, longitudinal questionnaire. Measured performance of waste-causing/waste-reducing behaviours.	Edible food (drinks included)	Intervention a: published once, available digitally on social media page. Intervention b: published once, circulated once digitally and available in store. Intervention c: published once, circulated once digitally and available in store.	Six or more months. One month from baseline questionnaire to intervention. Follow-up questionnaire two weeks after intervention. Longitudinal questionnaire five months after intervention.

Notes. ‘or more months’ used to indicate minimum longitudinal time between measures if the specific timing or data were missing in the original reviewed articles. Terminology such as ‘Group A’, ‘intervention’, and ‘follow-up’ have been consistently used where possible to reduce variations in language between the original reviewed articles and may not reflect the same grouping order. ‘Follow-up’ refers to the post-test measure succeeding the intervention treatment period. Groups were labelled with capital letters (i.e., A, B, C) and interventions were labelled with lowercase letters (i.e., a, b, c) as they may not correspond in all articles. ‘Food waste’ abbreviated to ‘FW’.

Table 2. Study design and methods.

Study	Target Behaviour(s)	Theoretical Underpinning	Intervention(s) and Group(s)	Intervention Methodologies Detailed	Data Collection	State of Food	Intervention Duration	Study Duration
Cooper et al., 2023 [47]	Planning, preparation, cooking, storage	Theory of Planned Behaviour, Motivation Opportunity Ability	Study 1: Intervention a: Information, online goal setting intervention and a salience tool (Group A + B + C) No intervention: (Group D— control group). Study 2: Online and physical intervention (Group A + C) Online only (Group B) No intervention: (Group D—control group) Random group allocation.	Two studies, both included in review. Study 1: Intervention a: Information and goal to make a ‘bonus meal’ with ingredients in the household (Group A + B + C). Intervention b: Increasing salience; track (Group A), collect (Group B), tag (Group C).	Baseline questionnaire with intervention or no intervention, 4 follow-ups, and a longitudinal questionnaire. Measured waste in questionnaires.	Study 1: Edible food limited to fruit, vegetables, bread/grains (drinks excluded). Study 2: As above, plus cheese, eggs, meat, fish	Study 1: Five weeks. Study 2: Five weeks (Group A + B) Three weeks (Group C).	13 weeks. One questionnaire a week. Longitudinal questionnaire eight weeks after final follow-up.
Everitt et al., 2023 [24]	Planning, purchasing, preparation, storage, consuming leftovers (by increasing intention to prevent food waste, perceived behavioural control)	Theory of Planned Behaviour	Intervention a: Awareness and Information (Group A). No intervention: (Group B—control group). All agreed to be recontacted for this study after participating in van der Werf et al.’s study [48]. Random group allocation.	Intervention a: Materials provided in 4 L container with information on local average value and quantity of FW, printed tools and links to website for education on improving behaviours, postcard linking wasted food and wasted money, five emails throughout intervention period.	Waste audit, intervention or no intervention, follow-up waste audit. Measured waste in garbage collection.	Edible and inedible food (dairy included, other drinks excluded)	Intervention a: Two weeks	Longitudinal questionnaire 31 months after intervention.
Gimenez et al., 2023 [33]	All (by increasing awareness of own FW and reducing cognitive dissonance)	Theory of Planned Behaviour, Social Practice Theory	Intervention a: Information (Group A). No intervention (Group B— control group). Random group allocation.	Intervention a: information provided online after questionnaire about FW, with the task to highlight text they agreed and disagreed with (Group A).	Baseline questionnaire with intervention or no intervention, follow-up questionnaire. Measured waste in questionnaires.	Edible food (drinks included)	Intervention a: Provided at single point in time	Two weeks from baseline questionnaire to final questionnaire. No longitudinal follow-up.
Graham-Rowe et al., 2019 [49]	All (by reducing denial of own waste)	Self-affirmation theory	Intervention a: Self-affirmation (Group A). Intervention b: Integrated self-affirmation (Group B). No intervention: (Group C— control group). Random group allocation.	Interventions a + b: shown different lists, asked to identify their most important values. These values were then tied to their perceived behavioural control over their FW, and the impacts on the environment (Groups A + B). Intervention b: shown list and also asked to evaluate why this value was important to them and how it tied in with past behaviour (Group B). Intervention c: shown same list as Intervention a and asked to instead identify the least important value. The value was then tied in with their perceived behavioural control over their FW, and with impacts of FW on the environment (Group C).	Baseline questionnaire with intervention or no intervention, follow-up questionnaire. Measured waste in questionnaires.	Edible fruit and vegetables (drinks unspecified)	Intervention a + b + c: Provided at single point in time	One week from baseline questionnaire to final questionnaire. No longitudinal follow-up.
Pelt et al., 2020 [50]	All (by increasing awareness of own FW and reducing cognitive dissonance)	Theory of Planned Behaviour, Social Practice Theory	Intervention a: Awareness (Group A). Intervention b: Cognitive dissonance (Group B). Intervention c: Information (Group C—control group). Random group allocation.	Interventions a + b + c: door-to-door pamphlet distribution. Pamphlet information on FW consequences and means to reduce it (Groups A + B + C). Intervention a: participants weighed and recorded FW in food diary, to be collected. Weighing instructions provided (Group A). Intervention b: participants asked to preach support for reducing FW, intention implementation plan, and listed recent past behaviours in opposition to this (Group B).	Waste audit, door-to-door visits with Intervention a, Intervention b, or no intervention, follow-up waste audit, longitudinal waste audit. Measured waste in garbage collection.	Edible and potentially edible food (only food in solid form, drinks excluded)	Intervention a: One week. Intervention b: on-the-spot activity	Six weeks from baseline waste audit to final waste audit. One week between audits and intervention. Longitudinal waste audit five weeks after intervention.
Roe et al., 2022 [51]	Planning, shopping, preparation, cooking, storage, consuming leftovers	Theory of Planned Behaviour, Theory of Reasoned Action, Tailored pro-environmental behaviour interventions	Intervention a: App goal setting and FW coaching (Group A). Intervention b: App and unrelated coaching (Group B— control group). Randomly allocated into groups.	Intervention a: One-on-one coaching about FW, with personalised tips and ways to reduce it. Asked to set goals, and received regular reminders by email/text/call, as preferred. Provided app to record FW and instructed on its use (Group A). Intervention b: One-on-one coaching and information provided about unrelated topic matched in intensity of communication. Provided app to record FW with and instructed on its use (Group B).	Training (in-person), baseline app-based FW diary, intervention or no intervention, follow-up app-based FW diary. Measured waste in app.	Edible and inedible food (drinks included)	Intervention a + b: One week	Two or more weeks. One baseline week followed by training, intervention and follow-up measurement week.
Romani et al., 2018 [43]	Food planning and preparation	Theory of Planned Behaviour	Intervention a: Information (Group A + C). Control group: (Group B + D). Solomon four-group design. Group allocation unspecified.	Intervention a: provided a printed/digital article on creating a weekly food menu, with advantages, recipes, and suggestions for family involvement. Two groups involved in pre-test (Group A, Group B), two groups involved an intervention (Group A, Group C), four groups involved in post-test (Group A, Group B, Group C, Group D).	Study 3 in larger body of research. Baseline food diary or no diary, intervention or no intervention, follow-up food diary. Measured waste in diary.	Edible food (milk and other potable dairy products included, other drinks excluded)	Intervention a: Provided at single point in time	Four weeks from pre-test to post-test diary. One week between diaries and intervention. No longitudinal follow-up.
Schmidt 2016 [52]	All	Integrative influence model of pro-environmental behaviour	Intervention a: Commitment and goal setting (Group A). No intervention: (Group B— control group). Randomly allocated into groups (restricted block formation)	Intervention a: provided five behavioural suggestions to reduce FW on an online platform, personalised with prior self-reported behaviour in baseline questionnaire. Then, presented with a public commitment and goal-setting measure and asked to rate willingness to follow recommendation (Group A).	Baseline assessment, intervention or no intervention, follow-up assessment, longitudinal post-test assessment. Measured performance of waste-causing/waste-reducing behaviours.	Edible food (drinks unspecified)	Intervention a: Four weeks (roughly), commencing at varying points in time	Seven months from baseline assessment to second assessment. Two months of recruiting and pre-test, another two between end of recruitment and intervention, approximately four before post-test assessment. Eight weeks after the intervention, a third assessment was undertaken, with low uptake.
Shu et al., 2021 [53]	Planning, storage, preparation, consuming leftovers	Unspecified	Intervention 1a: FW awareness (Group A). Intervention 1b: Food awareness (Group B—control group). Intervention 2a: Information on financial impact of food waste (Group C). Intervention 2b: Information on unrelated topic (Group D—control group). Randomly allocated into groups (data missing, presumed Intervention 2 allocated independently to Intervention 1 allocation)	Intervention 1a: Asked to pay attention to FW for the week preceding first FW measurement (Group A). Intervention 1b: Asked to pay attention to food in household for week preceding first FW measurement (Group B). Intervention 1 (a + b): In first questionnaire measuring FW, participants were provided instructions on how to estimate it based on size of their own hand (i.e., size of thumb, palm, fist). Intervention 2a: Information provided about environmental impact of FW (Group C). Intervention 2b: Information provided about unrelated topic (Group D—quasi-control).	Pre-study communication, baseline questionnaire with intervention or control message with training, follow-up questionnaire with intervention or control message, longitudinal questionnaire. Measured waste in questionnaires after prior notice to monitor FW.	Edible food (drinks included)	Intervention a: One week. Intervention b: On-the-spot message	Baseline questionnaire, one week before follow-up questionnaire, longitudinal questionnaire.
Shu et al., 2023 [54]	Storage	Unspecified	Intervention a + c: FW awareness and storage information (Group A). Intervention b + c: FW awareness and storage and composting information (Group B) Intervention c: FW awareness (Group C—control group) No intervention: (Group D—control group). Allocated into groups A + B + C based on household location in city district and waste collection day. Allocated into group D based on household location outside of district and part of national control.	Intervention a: Information on storage and offered storage containers (Group A + B). Intervention b: Information on storage and compost and offered composting tools (Group B). Intervention c: Information on FW prevention and diversion via community-based communication (Group A + B + C).	Baseline questionnaire, group-specific questions, intervention, follow-up questionnaire. Measured waste in garbage collection, on a community level (all) and household level (additional consent).	Edible food (drinks included)	Intervention a + b: One month. Intervention c: Three months.	Baseline questionnaire, one week before follow-up questionnaire.
Soma et al., 2020 [55]	Planning, shopping, cooking, storage, consuming leftovers (by increasing awareness of own waste, education on date labels)	Expanded Theory of Planned Behaviour	Intervention a: Information (Group A). Intervention a + b: Information and gamification (Group B). Intervention a + c: Information and community workshops (Group C). No Intervention (Group D—control group). Group allocation unspecified.	Intervention a: Information booklet on FW consequences and reduction strategies. Provided fridge magnet as prompt, and newsletters via mail or email with tips (Group A + B + C). Intervention b: Information plus an online quiz about FW using gamification techniques (Group B). Intervention c: Information plus community workshops with group activities, discussions, resources and prizes (Group C).	Waste audit, baseline questionnaire (in person), intervention or no intervention, follow-up questionnaire (online, over phone, or by mail), follow-up waste audit, (n.b. follow-up focus groups). Measured waste in garbage collection.	Edible and inedible food (dairy and non-dairy drinks included)	Intervention a: Provided at single point in time. Intervention b + c: 12 weeks.	12 or more weeks. One week for waste audit, 12 weeks between baseline questionnaire and follow-up questionnaire, one week for follow-up waste audit. Recruitment window unspecified. (N.b. three between follow-up audit and focus group discussions).
van der Werf et al., 2021 [48]	Planning, purchasing, preparation, storage, consuming leftovers (by increasing intention to prevent food waste, perceived behavioural control)	Theory of Planned Behaviour	Intervention a: Awareness and Information (Group A). No intervention: (Group B—control group). Randomly allocated into groups (with minor modification for practicality).	Intervention a: Materials provided in 4 L container with information on local average value and quantity of FW, printed tools and links to website for education on improving behaviours, postcard linking wasted food and wasted money, 5 emails throughout intervention period.	Waste audit, intervention or no intervention, follow-up waste audit. Measured waste in garbage collection.	Edible and inedible food (dairy included, other drinks excluded)	Intervention a: Two weeks	Five weeks from baseline audit to final waste audit. One week for audit, one week before intervention, one week for follow-up waste audit.
van Herpen et al., 2023 [56]	All	Motivation Opportunity Ability	Intervention a: Tools (Group A) Intervention b: Norms (Group B). Randomly allocated into groups.	Intervention a: Tools provided including shopping list notepad, stickers, measuring cup, app, leaflets, recipe cards, fridge thermometer (Group A + B) Intervention b: Motivational messaging with social norms (Group B).	Questionnaire, intervention, follow-up questionnaire.		Two weeks.	One month from baseline to follow-up measure.
Wharton et al., 2021 [57]	Purchasing, cooking, storage (by information, education on date labels)	Theory of Planned Behaviour	Intervention a: Information (Group A). No group allocation.	Intervention a: Provided with information and tools to prevent FW. Weekly themed modules on household food management framed by impact on health, finances, and the environment. Participants trained in-person to weigh and log FW in a standardised way on a provided logbook (hard copy or digital), and provided printed instructions, 3.8 L buckets and a digital scale.	Training and assessment (in person), baseline questionnaire, baseline FW log, intervention with weekly FW log, follow-up FW log, follow-up questionnaire, assessment (in person), (n.b. semi-structured interviews). Waste collected and weighed on scales.	Edible food (drinks excluded)	Five weeks	Seven or more months. Four months of recruitment, seven weeks from baseline to final week. One week for baseline measure, five weeks for intervention, one week for follow-up measure.
Young et al., 2017 [58]	Planning, purchasing, cooking, consuming	Social influence theory	Intervention a: Social influence (Group A). Intervention b: Information 1 (Group B). Intervention c: Information 2 (Group C). No intervention: (Group D—quasi-control group). Group allocation dependent on self-reported prior awareness of intervention materials. Not discrete groups.	Intervention a: Leftovers campaign on supermarket’s Facebook asking customers to submit their favourite recipes, provided link to tips from WRAP, promote discussion. Intervention b: Supermarket magazine (available in store and online, and distributed widely) featuring tips on reducing FW and recipes for commonly wasted foods. Intervention c: Supermarket e-newsletter (distributed widely) discussed using leftovers and provided link to Intervention a social media campaign, encouraging customers to take part. Provided storage tips and a link for purchase.	Baseline questionnaire, intervention, follow-up questionnaire, longitudinal questionnaire. Measured performance of waste-causing/waste-reducing behaviours.	Edible food (drinks included)	Intervention a: published once, available digitally on social media page. Intervention b: published once, circulated once digitally and available in store. Intervention c: published once, circulated once digitally and available in store.	Six or more months. One month from baseline questionnaire to intervention. Follow-up questionnaire two weeks after intervention. Longitudinal questionnaire five months after intervention.

Notes. ‘or more months’ used to indicate minimum longitudinal time between measures if the specific timing or data were missing in the original reviewed articles. Terminology such as ‘Group A’, ‘intervention’, and ‘follow-up’ have been consistently used where possible to reduce variations in language between the original reviewed articles and may not reflect the same grouping order. ‘Follow-up’ refers to the post-test measure succeeding the intervention treatment period. Groups were labelled with capital letters (i.e., A, B, C) and interventions were labelled with lowercase letters (i.e., a, b, c) as they may not correspond in all articles. ‘Food waste’ abbreviated to ‘FW’.

Table 3. Study outcomes.

Study	Sample Eligibility	Country/Region	Sample Size n=	Data Analysis	Intervention Results	Effect Size of Intervention	Longitudinal Intervention Results and Effect Size
Cooper et al., 2023 [47]	Families with children aged 3–18, equally or mainly responsible for food preparation and shopping. Recruited through research panel.	Canada and United States	Study one: Canada 1205 (baseline) 909 (follow-up) 647 (longitudinal) Study two: US 756 (baseline) 482 (follow-up) 447 (longitudinal)	Linear mixed effects model, ANOVA, type-III sums of squares, Satterthwaite’s denominator degrees of freedom, post hoc tests.	Study one + two: Significant FW reduction in intervention groups, and relative to Group D (control group). Three-week and five-week interventions, online only and online and physical interventions all found effective.	Study one: Group A + B + C: at follow-up, a reduction of 33.4% compared to baseline F(1,1141) = 11.10, p = 0.001. Group D (control group) a 14.4% reduction compared to baseline. Group A + B + C 26.5% more than Group D. Study two: Group A + B + C: at follow-up, an average reduction of 46% compared to baseline, and 33% compared to control. Group D (control group): difference from baseline not significant.	Study one: Reduction in FW found at follow-up relative to Group D (control) was no longer present. Study two: Reduction in FW found in multiple follow-up surveys in treatment groups A + B + C lessened by final longitudinal measure.
Everitt et al., 2023 [24]	Single-family households, drawn from the participants of a prior study (van der Werf et al., 2021 [48]), randomly selected from the sample that agreed to participate in further research.	Ontario, Canada	99	Split-plot ANOVA	Follow-up: Data missing, reported in van der Werf et al., 2021 [48].	Data missing.	Group A + B: No significant changes in total edible FW compared to baseline (p = 0.27), though insignificant change from follow-up (p = 0.97) indicates 30% sustained reduction. Group A: 1.6% increase, Group B (Control group): 0.6% decrease in edible FW.
Gimenez et al., 2023 [33]	25–69, regular participation in grocery shopping and food preparation (more than once a week), high proficiency in English. Recruited through research panel.	Australia	1117	Logarithmic transformation of data, inference tests, t-test, ANCOVA, Cronbach’s alpha coefficient.	Group A: Significant reduction of FW, from baseline and control (Group B). Group B: Also, significant reduction of food waste from baseline.	Group A: 120.7 g at follow-up, a reduction of 22.1% compared to baseline. Group B: 143.3 g at follow-up, a reduction of 17.1% compared to baseline (partial η2 0.004).	N/A
Graham-Rowe et al., 2019 [49]	Participants must have reported wasting food in the baseline questionnaire, and completed follow-up questionnaire. Recruited through advertising from fruit delivery companies, local council waste management departments.	United Kingdom	283	Hierarchical multiple regression analyses, one-way ANCOVA	Group A + B: Non-significant main effect in averaged treatment groups. Group A: Significant main effect on follow-up FW moderated by baseline.	F(2,279) = 2.73, p = 0.067, ηp2 = 0.02	N/A
Pelt et al., 2020 [50]	Single-family households drawn from randomly selected household list from rural regions, recruited based on timing of household garbage collection.	France	64	Mixed-design ANOVAs comparing follow-up FW and longitudinal FW.	Follow-up: Data missing.	N/A	Group B: Reduction in FW between follow-up and 5 weeks later (F(1,19) = 4.675 (p < 0.05), partial η2 = 0.197).
Roe at al., 2022 [51]	18–65, responsible for part of household shopping. Restrictions based on BMI and immune system health, and exclusions based on pregnancy. Has access to Apple iPhone. Recruited by ads on the research centre’s website, electronic mailing list, and social media posts.	Louisiana, United States	40	Wilcoxon Signed Rank test (two-sided) regression analysis	Group A: Effective in reducing FW at consumption stage, not significantly different at preparation and storage. Sample size should be interpreted with caution.	Group A: Reduction of 390.8 g, or 78.8% compared to baseline plate waste (p ≤ 0.05).	N/A
Romani et al., 2018 [43]	Mainly responsible for shopping and cooking, 18+, have a child in their household. Recruited by convenience sampling (via Master’s degree students).	Italy	210	2 × 2 ANCOVA, mediation analysis, bootstrapping	Group A: Significant effect on FW reduction. Non-significant effect found on planning for meals skills.	FW (t(208) = 3.36, p < 0.001). Meal planning (−97.69, p < 0.10).	N/A
Schmidt 2016 [52]	Unrestricted household make up (included flat share, single-family). Recruited through convenience sampling (social network, university newsletters), in a local event and local newspapers. Included people aged 17–71.	Magdeburg, Germany	217 (baseline) 130 (follow-up)	Factorial repeated measures ANOVAs, Spearman–Brown correlations	Follow-up: Group A: Significant effect on food waste-preventing behaviours at follow-up, relative to baseline and control (Group B).	(F(1,128) = 7.00, p < 0.01 and partial η2 = 0.05) and in intervention a: (Mt1 = 3.45, SD = 0.55, Mt2 = 3.98, SD = 0.51) compared to control (Mt1 = 3.47, SD = 0.52, Mt2 = 3.78, SD = 0.58).	Longitudinal: Indicates continued performance of behaviours but sample size too small to reliably confirm effect (Subsample of Group A, n = 20). Data not shown.
Shu et al., 2021 [53]	Aged between 18–35 or over 55, live in small households (1–2 people) or large households (5+ people) and with access to internet. Recruited via Qualtrics survey platform.	United States	114	Censored regression	Group A: Non-significant reduction of FW compared to baseline.	Group A: Reduction of 25.4% or 151 g (p = 0.109).	N/A
Shu et al., 2023 [54]	Equally or mainly responsible for household food preparation, 18+. Reside in local city district receiving community-based campaign, with waste collection on specific days. Recruited by letters, social media and local media. Alternately, national control resided outside of district and not subjected to campaign, recruited by research panels.	Ohio, United States	897 (baseline) 818 (follow-up) Residents that were not part of the baseline were invited to participate in the follow-up survey.	Censored regression, linear regression, controlled for demographic differences within group	Group A + B + C: Reduced amount of FW at follow-up in averaged treatment groups including the regional control group, relative to the national control group (Group D). Group B: Significant reduction in avoidable FW at follow-up compared to all other groups averaged. (Group A + C + D) Group A + B: Less FW reduction found in averaged treatment groups compared to averaged regional control group (Group C).	Group A + B reduction of 9%. Group B reduction of 53% avoidable FW. Group A + B + C reduction of 23% (p < 0.01). Group C reduction of 44%. Group D increase of 29% (p < 0.01).	N/A
Soma et al., 2020 [55]	Unrestricted household make up (single-family and multiple-family). Responsible for at least half of food preparation. Drawn from randomly selected household list from local administrative district. Completed one survey and two waste audits at minimum, or both surveys and 0–2 waste audits.	Toronto, Canada	501	Wilcoxon Signed Rank test (one-tailed for intervention groups, two-tailed for control group)	Group B: Non-significant reduction in FW at follow-up waste audit, compared to baseline. Subsample of Group B: Marginally significant difference within group at follow-up, with a greater reduction among regular online quiz players than the rest of Group B. No significant differences with Group B compared to Group A, Group C, or Group D on follow-up. Groups A + C + D: No statistically significant differences found from respective baseline measures. Intervention groups self-reported less FW than control group at follow-up, data not shown.	Group B: (z = −1.47, p = 0.07) compared to Group B baseline. Group B frequent gamers: 0.67 kg at follow-up waste audit, compared to infrequent gamers 1.00 kg (p = 0.030).	N/A
van der Werf et al., 2021 [48]	Single-family households, drawn from volunteer households and selected in clusters for convenience, but a spread across urban and suburban waste collection zones to maximise range of socioeconomic households.	Ontario, Canada	112	Multiple linear regression models, statistical interaction terms	Group A: Significant reduction of total FW and edible FW, compared to baseline and control (Group B).	Group A: reduction of 1.044 g or 31% total FW (12%), and reduction of 634 g or 30% edible FW. (F = 3.881, p = 0.05). Significant relative to control. Group B (control): increase of 18 g or 1% edible FW.	N/A
van Herpen et al., 2023 [56]	Recruited through social media and the mailing list of previous research participants who agreed to future contact.	The Netherlands	150	Repeated measures ANOVAs, Tobit regression models, bootstrapping.	Group A + B: Reported performing more waste-reducing behaviours, increase in perceived ability, and reduced FW than baseline.	Group A + B: Behaviours (F(1,148) = 49.86, p < 0.001, n2p = 0.25) Perceived ability (M = 5.43, F(1,148) = 25.67, p < 0.001, n2p = 0.15) FW reduction of 145.6 g or 39.2% total FW (F(1,148) = 24.63, p < 0.01, η2p = 0.14).	N/A
Wharton et al., 2021 [57]	Only single-family households. Eat dinner at home on at least 3 days per week, and primary food purchaser in the home. Aged 18 or over. Recruited through social media, flyers, community events and farmers’ markets. Excluded for composting.	Arizona, United States	53	Non-parametric Friedman’s test, Z-score effect sizes compared (r = Z/√n), non-parametric Wilcoxon signed rank test.	Group A: Significant FW reduction from baseline to follow-up. Fewer extreme values as intervention progressed.	Group A: (MedianBaseline = 28.22, MedianFollow-up = 20.36; p = 0.008)	N/A
Young et al., 2017 [58]	Customers of large-scale supermarket chain Asda. Participants completed all three questionnaires. Aged 16–70.	United Kingdom	2038	One-way repeated measures ANOVAs, Wilks’ Lambda, post hoc tests	Follow-up: Frequency and quantity of FW: No significant difference seen overall by follow-up. Mixed results by intervention, with some means increasing at Time 2.	Data missing.	Significant effect of time: FW frequency (F(2,2008) = 4.78, p < 0.01) and FW quantity (F(2,2009) = 13.65, p < 0.001 Frequency: Group A + B + C: No significant effect found. Group D (control): Baseline (M = 1.27, SD = 1.42), Longitudinal (M = 1.14, SD = 1.31); t(2.32, p ≤ 0.05). Quantity: Group A + C: Significant reduction in quantity from baseline. Group A: Baseline (M = 1.28, SD = 1.36), Longitudinal (M = 1.17, SD = 1.33); t(1.99, p ≤ 0.05). Group C: Baseline (M = 1.43, SD = 1.34), Longitudinal (M = 1.16, SD = 1.26); t(2.29, p ≤ 0.05) Group D: Significant reduction in frequency in control group from baseline. Group D (control): Baseline (M = 1.27, SD = 1.42 [sic]), Longitudinal (M = 1.14, SD = 1.31); t(2.32, p ≤ 0.05). Group B: No significant effect found. Non-statistically significant differences. Data not shown. Engagement with more than one intervention revealed mixed results. Data not shown.

Notes. ‘data missing’ refers to data not reported in the original reviewed article. Data ‘not shown’ refers to information not reported in this review for brevity and simplicity. Terminology such as ‘Group A’, ‘intervention’, and ‘follow-up’ have been consistently used where possible to reduce variations in language between the original reviewed articles and may not reflect the same grouping order. ‘Follow-up’ refers to the post-test measure succeeding the intervention treatment period. ‘Food waste’ abbreviated to ‘FW’.

Table 3. Study outcomes.

Study	Sample Eligibility	Country/Region	Sample Size n=	Data Analysis	Intervention Results	Effect Size of Intervention	Longitudinal Intervention Results and Effect Size
Cooper et al., 2023 [47]	Families with children aged 3–18, equally or mainly responsible for food preparation and shopping. Recruited through research panel.	Canada and United States	Study one: Canada 1205 (baseline) 909 (follow-up) 647 (longitudinal) Study two: US 756 (baseline) 482 (follow-up) 447 (longitudinal)	Linear mixed effects model, ANOVA, type-III sums of squares, Satterthwaite’s denominator degrees of freedom, post hoc tests.	Study one + two: Significant FW reduction in intervention groups, and relative to Group D (control group). Three-week and five-week interventions, online only and online and physical interventions all found effective.	Study one: Group A + B + C: at follow-up, a reduction of 33.4% compared to baseline F(1,1141) = 11.10, p = 0.001. Group D (control group) a 14.4% reduction compared to baseline. Group A + B + C 26.5% more than Group D. Study two: Group A + B + C: at follow-up, an average reduction of 46% compared to baseline, and 33% compared to control. Group D (control group): difference from baseline not significant.	Study one: Reduction in FW found at follow-up relative to Group D (control) was no longer present. Study two: Reduction in FW found in multiple follow-up surveys in treatment groups A + B + C lessened by final longitudinal measure.
Everitt et al., 2023 [24]	Single-family households, drawn from the participants of a prior study (van der Werf et al., 2021 [48]), randomly selected from the sample that agreed to participate in further research.	Ontario, Canada	99	Split-plot ANOVA	Follow-up: Data missing, reported in van der Werf et al., 2021 [48].	Data missing.	Group A + B: No significant changes in total edible FW compared to baseline (p = 0.27), though insignificant change from follow-up (p = 0.97) indicates 30% sustained reduction. Group A: 1.6% increase, Group B (Control group): 0.6% decrease in edible FW.
Gimenez et al., 2023 [33]	25–69, regular participation in grocery shopping and food preparation (more than once a week), high proficiency in English. Recruited through research panel.	Australia	1117	Logarithmic transformation of data, inference tests, t-test, ANCOVA, Cronbach’s alpha coefficient.	Group A: Significant reduction of FW, from baseline and control (Group B). Group B: Also, significant reduction of food waste from baseline.	Group A: 120.7 g at follow-up, a reduction of 22.1% compared to baseline. Group B: 143.3 g at follow-up, a reduction of 17.1% compared to baseline (partial η2 0.004).	N/A
Graham-Rowe et al., 2019 [49]	Participants must have reported wasting food in the baseline questionnaire, and completed follow-up questionnaire. Recruited through advertising from fruit delivery companies, local council waste management departments.	United Kingdom	283	Hierarchical multiple regression analyses, one-way ANCOVA	Group A + B: Non-significant main effect in averaged treatment groups. Group A: Significant main effect on follow-up FW moderated by baseline.	F(2,279) = 2.73, p = 0.067, ηp2 = 0.02	N/A
Pelt et al., 2020 [50]	Single-family households drawn from randomly selected household list from rural regions, recruited based on timing of household garbage collection.	France	64	Mixed-design ANOVAs comparing follow-up FW and longitudinal FW.	Follow-up: Data missing.	N/A	Group B: Reduction in FW between follow-up and 5 weeks later (F(1,19) = 4.675 (p < 0.05), partial η2 = 0.197).
Roe at al., 2022 [51]	18–65, responsible for part of household shopping. Restrictions based on BMI and immune system health, and exclusions based on pregnancy. Has access to Apple iPhone. Recruited by ads on the research centre’s website, electronic mailing list, and social media posts.	Louisiana, United States	40	Wilcoxon Signed Rank test (two-sided) regression analysis	Group A: Effective in reducing FW at consumption stage, not significantly different at preparation and storage. Sample size should be interpreted with caution.	Group A: Reduction of 390.8 g, or 78.8% compared to baseline plate waste (p ≤ 0.05).	N/A
Romani et al., 2018 [43]	Mainly responsible for shopping and cooking, 18+, have a child in their household. Recruited by convenience sampling (via Master’s degree students).	Italy	210	2 × 2 ANCOVA, mediation analysis, bootstrapping	Group A: Significant effect on FW reduction. Non-significant effect found on planning for meals skills.	FW (t(208) = 3.36, p < 0.001). Meal planning (−97.69, p < 0.10).	N/A
Schmidt 2016 [52]	Unrestricted household make up (included flat share, single-family). Recruited through convenience sampling (social network, university newsletters), in a local event and local newspapers. Included people aged 17–71.	Magdeburg, Germany	217 (baseline) 130 (follow-up)	Factorial repeated measures ANOVAs, Spearman–Brown correlations	Follow-up: Group A: Significant effect on food waste-preventing behaviours at follow-up, relative to baseline and control (Group B).	(F(1,128) = 7.00, p < 0.01 and partial η2 = 0.05) and in intervention a: (Mt1 = 3.45, SD = 0.55, Mt2 = 3.98, SD = 0.51) compared to control (Mt1 = 3.47, SD = 0.52, Mt2 = 3.78, SD = 0.58).	Longitudinal: Indicates continued performance of behaviours but sample size too small to reliably confirm effect (Subsample of Group A, n = 20). Data not shown.
Shu et al., 2021 [53]	Aged between 18–35 or over 55, live in small households (1–2 people) or large households (5+ people) and with access to internet. Recruited via Qualtrics survey platform.	United States	114	Censored regression	Group A: Non-significant reduction of FW compared to baseline.	Group A: Reduction of 25.4% or 151 g (p = 0.109).	N/A
Shu et al., 2023 [54]	Equally or mainly responsible for household food preparation, 18+. Reside in local city district receiving community-based campaign, with waste collection on specific days. Recruited by letters, social media and local media. Alternately, national control resided outside of district and not subjected to campaign, recruited by research panels.	Ohio, United States	897 (baseline) 818 (follow-up) Residents that were not part of the baseline were invited to participate in the follow-up survey.	Censored regression, linear regression, controlled for demographic differences within group	Group A + B + C: Reduced amount of FW at follow-up in averaged treatment groups including the regional control group, relative to the national control group (Group D). Group B: Significant reduction in avoidable FW at follow-up compared to all other groups averaged. (Group A + C + D) Group A + B: Less FW reduction found in averaged treatment groups compared to averaged regional control group (Group C).	Group A + B reduction of 9%. Group B reduction of 53% avoidable FW. Group A + B + C reduction of 23% (p < 0.01). Group C reduction of 44%. Group D increase of 29% (p < 0.01).	N/A
Soma et al., 2020 [55]	Unrestricted household make up (single-family and multiple-family). Responsible for at least half of food preparation. Drawn from randomly selected household list from local administrative district. Completed one survey and two waste audits at minimum, or both surveys and 0–2 waste audits.	Toronto, Canada	501	Wilcoxon Signed Rank test (one-tailed for intervention groups, two-tailed for control group)	Group B: Non-significant reduction in FW at follow-up waste audit, compared to baseline. Subsample of Group B: Marginally significant difference within group at follow-up, with a greater reduction among regular online quiz players than the rest of Group B. No significant differences with Group B compared to Group A, Group C, or Group D on follow-up. Groups A + C + D: No statistically significant differences found from respective baseline measures. Intervention groups self-reported less FW than control group at follow-up, data not shown.	Group B: (z = −1.47, p = 0.07) compared to Group B baseline. Group B frequent gamers: 0.67 kg at follow-up waste audit, compared to infrequent gamers 1.00 kg (p = 0.030).	N/A
van der Werf et al., 2021 [48]	Single-family households, drawn from volunteer households and selected in clusters for convenience, but a spread across urban and suburban waste collection zones to maximise range of socioeconomic households.	Ontario, Canada	112	Multiple linear regression models, statistical interaction terms	Group A: Significant reduction of total FW and edible FW, compared to baseline and control (Group B).	Group A: reduction of 1.044 g or 31% total FW (12%), and reduction of 634 g or 30% edible FW. (F = 3.881, p = 0.05). Significant relative to control. Group B (control): increase of 18 g or 1% edible FW.	N/A
van Herpen et al., 2023 [56]	Recruited through social media and the mailing list of previous research participants who agreed to future contact.	The Netherlands	150	Repeated measures ANOVAs, Tobit regression models, bootstrapping.	Group A + B: Reported performing more waste-reducing behaviours, increase in perceived ability, and reduced FW than baseline.	Group A + B: Behaviours (F(1,148) = 49.86, p < 0.001, n2p = 0.25) Perceived ability (M = 5.43, F(1,148) = 25.67, p < 0.001, n2p = 0.15) FW reduction of 145.6 g or 39.2% total FW (F(1,148) = 24.63, p < 0.01, η2p = 0.14).	N/A
Wharton et al., 2021 [57]	Only single-family households. Eat dinner at home on at least 3 days per week, and primary food purchaser in the home. Aged 18 or over. Recruited through social media, flyers, community events and farmers’ markets. Excluded for composting.	Arizona, United States	53	Non-parametric Friedman’s test, Z-score effect sizes compared (r = Z/√n), non-parametric Wilcoxon signed rank test.	Group A: Significant FW reduction from baseline to follow-up. Fewer extreme values as intervention progressed.	Group A: (MedianBaseline = 28.22, MedianFollow-up = 20.36; p = 0.008)	N/A
Young et al., 2017 [58]	Customers of large-scale supermarket chain Asda. Participants completed all three questionnaires. Aged 16–70.	United Kingdom	2038	One-way repeated measures ANOVAs, Wilks’ Lambda, post hoc tests	Follow-up: Frequency and quantity of FW: No significant difference seen overall by follow-up. Mixed results by intervention, with some means increasing at Time 2.	Data missing.	Significant effect of time: FW frequency (F(2,2008) = 4.78, p < 0.01) and FW quantity (F(2,2009) = 13.65, p < 0.001 Frequency: Group A + B + C: No significant effect found. Group D (control): Baseline (M = 1.27, SD = 1.42), Longitudinal (M = 1.14, SD = 1.31); t(2.32, p ≤ 0.05). Quantity: Group A + C: Significant reduction in quantity from baseline. Group A: Baseline (M = 1.28, SD = 1.36), Longitudinal (M = 1.17, SD = 1.33); t(1.99, p ≤ 0.05). Group C: Baseline (M = 1.43, SD = 1.34), Longitudinal (M = 1.16, SD = 1.26); t(2.29, p ≤ 0.05) Group D: Significant reduction in frequency in control group from baseline. Group D (control): Baseline (M = 1.27, SD = 1.42 [sic]), Longitudinal (M = 1.14, SD = 1.31); t(2.32, p ≤ 0.05). Group B: No significant effect found. Non-statistically significant differences. Data not shown. Engagement with more than one intervention revealed mixed results. Data not shown.

Notes. ‘data missing’ refers to data not reported in the original reviewed article. Data ‘not shown’ refers to information not reported in this review for brevity and simplicity. Terminology such as ‘Group A’, ‘intervention’, and ‘follow-up’ have been consistently used where possible to reduce variations in language between the original reviewed articles and may not reflect the same grouping order. ‘Follow-up’ refers to the post-test measure succeeding the intervention treatment period. ‘Food waste’ abbreviated to ‘FW’.

3.2.2. Context

As a condition of the review, the studies were conducted in a developed country, sampling the target populations that need to reduce their food waste with greater urgency [2]. Of the final 16 papers, most studies were undertaken in North America (n = 8) (Canada and the United States), followed by Europe (n = 4) (Italy, France, Germany, and The Netherlands), the United Kingdom (n = 2), and Australia (n = 1), suggesting a pre-and-post research design may be more commonly employed in particular regions over others. Sample sizes varied across the studies, ranging from pilot studies with samples too small to enable insight extrapolation to the wider population, n = 40 participants across two groups [51], through to robust samples, namely up to n = 2038 participants [58]. Though larger samples are recommended for the reliability of statistical analysis, the small-scale approaches offer a personalised approach which can complement quantitative findings, to allow for deeper insights to emerge, thus strengthening the breadth of the literature [59].

The reviewed study by Everitt, van der Werf, Seabrook, and Gilliland in 2023 [24] was a continuation of an earlier study by van der Werf, Seabrook, and Gilliland in 2021 [48], adding a longitudinal element. The remaining 14 reviewed articles were unique studies, though it was common for authors to be involved in writing multiple papers on the topic in subsequent years (e.g., [53,60]). This underscores their prolific contribution to the literature, and enriches the discourse on food waste behaviour change through consistent terminology, collaboration, and deepening insights [61].

The studies administered interventions to these samples with an intention of shaping future behaviour, facilitating less wasteful food management responses. This was conducted through tangible interventions (pamphlets with information, magnets, measuring cups), technology (such as a mobile phone app, online quiz, emails, social media), personalised touchpoints (face-to-face interviewing, community workshops), and on-the-spot or one-off messages. The variety of the intervention approaches tested suggests that the most effective approach for changing behaviour short-term remains unknown.

3.2.3. Defining Food Waste

Disposing of some food is unavoidable, as not all parts are edible [62]. However, notions of what is or is not edible vary based on many factors, including upbringing, culture, and diet [63,64]. The reviewed articles focused on definition for the purpose of measuring avoidable and unavoidable food waste; however, missed the opportunity to educate and increase the appetite for more parts of foods fit for human consumption.

Studies in the review differed in aim, with some focusing on only changing behaviour of edible or avoidable food waste, and others also including inedible and potentially unavoidable food and drinks. Simply, this can be determined by what is fit-for-purpose, balancing practicality with rigor.

An argument for the inclusion of all types of food waste is that it reduces the subjectivity of what participants include in self-reported measurements, though studies of this nature may not adequately be able to determine the efficacy of their interventions if the analysis is on an aggregate level, compared with those that only include avoidable waste. Furthermore, the inclusion of both inedible and edible food cannot easily be compared to other studies focusing on only edible food.

Another challenge with choosing to focus on edible foods only is that some methodologies do not lend themselves to waste separation; inedible food cannot be differentiated as clearly in a waste audit as it can in a self-reported diary or questionnaire. Thus, the 16 studies vary in their inclusion of parts of food, and this makes the direct comparison of their intervention efficacy imprudent.

3.2.4. Target Behaviours

Research indicates that consumer food routines and management play a crucial role in predicting food waste [3,65]. In particular, people with better food planning habits tend to waste less than people with worse or no planning habits [45,66]. The inclusion of planning was evident in all bar two studies [53,57], which concentrated on food shopping, cooking, and storage. In the studies that incorporated planning behaviours, the interventions targeted what food to buy, what to cook, how to be flexible for changes in the number of diners, how food is stored, and so on. Furthermore, five of these studies had a holistic approach to food management, which encompass food planning in the wider food management system [33,49,50,52,56], rather than focusing solely on food planning [24,43,47,48,51,53,55,58].

Intervention personalisation was found to have merit in the review [51,52,67], if based on an individual’s poorest performing behaviours, including planning or other stages in food management [65,68]. Broad behaviour change approaches like those used in the reviewed studies were also validated, offering a cost-effective solution that may be easier to implement on a large scale [23]. This trade-off between individualisation and cost-effectiveness is a key consideration for policymakers and change advocates.

3.2.5. Theoretical Underpinning

Through the use of theoretical frameworks, researchers have tried to understand and explain why food waste occurs [60,69], thus identifying behaviours that need to change, and providing guidance for the most appropriate interventions to address them [70]. The incorporated frameworks can broadly be categorised under cognitive (theory of planned behaviour [71], motivation opportunity ability framework [72], self-affirmation theory [73], procedural (nudging [30]), and social (practice theory [74], social influence theory [75]) aspects of behavioural science.

Basing interventions on well-supported frameworks can facilitate persistent change [8]. The theory of planned behaviour is a prominent foundation for intervention design [45,76], cited in nine of the sixteen articles. Other theories used were wider pro-environmental behaviour research [51,52], the motivation opportunity ability framework [47,54], and self-affirmation theory [49]. Similarly, but offering a different perspective, social influence theory [58] and social practice theory [33,50] were drawn on to incorporate social norms and influences on participant behaviour. These theories purport that behaviours are influenced by the people and environment around us, not undertaken in isolation [23,77]. Two studies provided no theoretical basis to inform its design [53,54].

3.2.6. Treatment Group Allocation

In all but one study [57], participants were allocated groups, receiving different amounts of information, and being tasked with varying intensities of participant engagement with an intervention. Ten studies randomly allocated participants into their groups, to prevent bias and heterogeneity that may occur from purposeful grouping [78]. However, in the case of one study [48], participants were regrouped based on geographical proximity in order to minimise the risk of the treatment and control groups interacting. Furthermore, another determined group allocation by the location of their household [54], negating reliability and validity benefits that come with the robust process of randomisation, and risking heterogenous group characteristics [44].

Of the grouped studies, all but one [56] incorporated a control group in the study design, reducing confounding external factors, and enabling a change in behaviour to be attributable to the success of an intervention [44,79]. Recent literature has indicated that measurement may impact on behaviour, and thus differences between the control group and treatment group are necessary to indicate if an intervention works, if all groups show a reduction in food waste post-test [13]. In Young, Russell, Robinson, and Barkemeyer’s study [58], participants were placed in groups dependent on the prior self-reported awareness of campaign materials, thus it cannot be assumed that the participants were like-for-like across groups, as purposeful allocation renders the control group a quasi-control [58,80].

Control groups were used in the following two fundamental ways: (a) to control against interventions provided in other groups, i.e., participants not given an intervention or treatment [24,33,43,47,48,49,52,54,55,58], or, alternately, (b) to enable different groups to receive different treatments, i.e., each group completes a survey common to all participants, with the control group missing the intervention or secondary treatment [50], or instead receiving an unrelated ‘placebo’ intervention or information [51,53].

3.2.7. Behaviour Change Interventions

The studies were comprised of methodologies whereby volumes or quantities of food waste or food waste-causing behaviours were quantified. To quantify the food waste or behaviour, data were collected using self-reporting, most commonly through online questionnaires where participants were asked to estimate or measure their behaviour or waste, keeping a log, or a physical collection and audit of their waste bins.

In addition, one or more interventions were applied to improve the food waste management behaviours found to be associated with lower food waste, and a second measurement was taken to reveal any differences. Interventions themselves centred around the education of food waste and its impacts, providing resources, tools, and tips to impart better behaviours that improve food management, thus building awareness of participants’ wasteful behaviours, and influencing them towards behaviours associated with reduced waste.

The reviewed studies skewed towards generalised, cost-effective solutions able to be implemented on a larger scale [23], rather than personalising treatments for the individual like Roe, Qi, Beyl, Neubig, Apolzan, and Martin [51]. Designing interventions aligned with targeted behaviours has been found more successful, and this trade-off reflects a key consideration for policymakers [67,70].

The papers present varied and creative approaches to improving food waste behaviour, targeting a myriad of complex antecedents to the issue, as detailed in

Table 4. Household food waste intervention approaches employed by pre- and post-test studies.

Approach/Intervention	Result	Contribution to Literature
Activities for the recovery of food that would otherwise be wasted (use-up day, flexibility in planning)	Effective short-term Showing promise long-term effect lessened in the treatment group/no difference between treatment group and control group Three-week intervention found as efficacious as five-week intervention [47]	A new pro-environmental behaviour change approach designed with FW in mind An alternative approach to educating on the impacts of FW Good respondent engagement with interventions (easy to participate, enjoyable), which provides opportunity to minimise attrition and facilitate longitudinal studies Evidence for shorter intervention enables reduced costs and increases feasibility
Activities for meal planning (weekly menu) with information (digital material)	Effective on food waste quantity short-term in a convenience sample Meal planning skills not significantly changed [43]	Promising results to be replicated in a sample of the general population The intervention may be built upon with the addition of more educational information or another food management behaviour to have an effect on skills
Website with information (themed modules and downloadable digital material)	Effective on FW quantity and FW behaviours short-term [57] Requires long-term measurement	Provides an objective method for waste measurement Impact on FW warrants further research
Provision of food management tools (e.g., storage container, measuring cup, shopping list, freezer stickers, fridge magnets) with information (digital and printed materials)	Mixed effects; information and tool package found effective [48,56] and ineffective [55], short-term Indication that change was sustained long-term, requiring further validation and replication Efficacious for certain avoidable food products, such as bread and baked goods, fruit and vegetables [24]	Impact on FW warrants further research; intervention design may impact efficacy and intensive package may be more efficacious than simple intervention Can be utilised for frequent wasters of particular products
Social norms and the provision of food management tools with information	No effect found short-term due to little interaction with email containing social norms No difference found on behaviours, perceived skill, or food waste quantity when received with tools and information [56] Requires the further implementation in short-term and long-term studies to determine efficacy	Poor engagement with intervention
Cognitive dissonance with information	No effect found one week after intervention [50], but significant effect found two weeks on average after intervention [33] and five weeks after intervention [50] Requires long-term measurement	Impact on FW warrants further research Targets psychological barrier to reducing FW Cost-effective and digital, enabling widespread and large-scale studies
Self-affirmation with information	Mixed effects; a standard self-affirmation treatment having a significant effect, moderated by baseline FW behaviour, but an integrated self-affirmation treatment not showing a significant change of food waste behaviour in the short-term Efficacious for certain avoidable food products: fruit and vegetables [49] Requires long-term measurement	Findings that participants who wasted more benefitted the most from the intervention, suggesting this approach could work for those who need behaviour change the most Can be utilised for high-volume wasters of particular products
Goal-setting (tailored) with online component (app/website)	Effective in a short-term study [52] and a smaller pilot study [51], when incorporating interaction with digital component Indication that after committing to goals, change in behaviour was sustained longer term (eight weeks after intervention), albeit among small sample size, requiring further research [52]	One-on-one app coaching can be scaled by providing online training videos of objective food waste measurement A personalised approach to goal setting can be achieved by relating goals to an individual participant’s baseline food waste behaviours from the questionnaire This is a promising area for further research
Pre-announcement survey message	No effect found short-term [53] Requires further implementation in short-term and long-term studies to determine efficacy	Using a simple survey pre-announcement warms respondents up to quantifying their FW to enable better self-reported measurement This light-touch intervention style could be incorporated with other interventions
Community-based campaign with information (storage and compost)	Effective for short-term treatment groups and quasi-control group Combined intensive storage and compost materials had the greatest impact [54]	Significant results found warranting further study, such as applying the campaign in other countries Yet to be tested with a control group recruited by same method and timing as treatment groups Campaign materials designed and available for replication Yet to be tested long-term on household FW behaviour
Gamification (online quiz) with information	Mixed effects short-term; the gamification intervention did not have a significant effect overall, but was effective for participants that engaged with the online quiz frequency [55] Requires longer-term measurement	Findings that participants who engaged more frequently benefitted the most from the intervention Good respondent engagement with intervention (easy to participate, enjoyable) which provides opportunity to minimise attrition and facilitate longitudinal studies Evidence of engagement for lengthy intervention (12 weeks)
Community workshops (presentations, videos, interactive group discussions, and activities) with information	No effect found short-term due to little participation turn-out at workshops [55] Requires further implementation in short-term and long-term studies to determine efficacy	Poor engagement with intervention
Social influence campaign on supermarket social media (leftovers) with information provided (digital and downloadable material) and information-sharing facilitated (discussion, contributing recipes)	Results indicate the Facebook intervention took time to change behaviour, and no reduction was shown short-term (two weeks after intervention) Effective at reducing the quantity of FW long-term (five months after intervention) Effective at reducing quantity of FW combined with digital supermarket newsletter and in-store supermarket magazine interventions Changes in FW quantity also seen in control group and not randomly allocated, requiring more research [58]	Provides commercial utility for large-scale implementation Promising results to be replicated in a randomised controlled trial

. Shy of providing a preferred method or solution, the data provide utility in designing future interventions. Promisingly, the review consistently found that supplementing interventions with an educational component was efficacious, supporting the existing pro-environmental literature [81]. Providing information on the social, environmental, or economical impacts of food waste, combined with tools and processes by which to change behaviour, illustrates why the change is beneficial and, in turn, facilitates it [82].

3.2.8. Study and Intervention Duration

For studies with a pre- and post-test measurement but with no follow-up, study durations ranged substantially, from one week [49] to over seven months [57]. Five were longitudinal studies that measured waste or waste behaviours over six weeks [24,47,50,52,58]. Detailed durations are found in Table 1.

Interventions were provided at a single point, or had a longer duration, ranging from one week [50,51,53] to the study’s close, allowing participants to access materials until follow-up measures [58]. Given habits do not form instantly [30], longer access to intervention tools and resources has been purported to be more effective in getting new behaviours to be maintained, with an ideal treatment time still yet to be established [83].

More thought should be given to developing enjoyable and simple interventions to minimise participant attrition. By maximising participant retention over the course of a study, sample representativeness will be greater, and bias from drop-outs can be reduced, thus enabling greater generalisability of results [84].

3.2.9. Study Design

Tailoring study design to ensure the effective delivery and evaluation of interventions is pivotal [85]. Yet, in food waste research, measurement itself varies, targeting food waste quantities, volumes, or behaviours that contribute to or prevent waste. In the reviewed articles, quantification entailed self-reporting, typically through online questionnaire estimation or measurement, log keeping, or the physical collection and audit of waste bins.

Challenges with each method of measurement prevent a consistent approach from being recommended [84]. While bin audits offer the greatest reliability through actual measurement, they may discourage “mainstream consumers” from participation, due to the perceived invasiveness of their bin contents being observed, and this may raise issues of consent if participants are unaware of the audit. Additionally, general waste bins do not take into account foods wasted by other disposal methods such as the kitchen sink or compost [1,58], and make the delineation of inedible and edible food difficult (see Section 3.2.3.).

Conversely, research shows that self-reported diaries and online questionnaires underestimate food waste [86], and individuals find it difficult to recall waste [50]. However, they are scalable and cost-effective [53], offer less onerous data collection for participants, and can lead to lower levels of attrition and bias, and a greater sample size [87,88]. For the purpose of longitudinal studies, questionnaires complement repeated measures as they are consistent, allowing for variability within individuals to be identified, and distinguishing changes in behaviour or waste over time [53].

Of the reviewed articles, the following two novel approaches were utilised to address this challenge: one study developed a protocol to consistently measure food waste with scales and buckets to individually train participants [57]. This additional touchpoint improves the accuracy and objectivity which is lacking in self-reporting, while offering a scalable solution via the digital sharing of instructional materials. Another study gave participants the advance notice of the intention to self-report food waste in a questionnaire, enabling them to pay attention to their food waste prior to the first survey. Though this may change their behaviour, it reduces the discrepancy between the first measurement and follow-up measurements, for which participants are aware of the upcoming task [53]. These methods of measurement warrant comparison against the commonly employed surveys, diaries, and curb-side audits to validate their use for future studies.

In tandem with the issue of fallible measurements across household food waste literature, the review of 601 papers also revealed a paucity of articles taking before and after measurements, with any of these methods. Though the majority of the articles in the eligibility phase of PRISMA were already excluded on a fundamental issue, such as the study not being set in the household or not implementing any interventions, little of the literature utilised a pre–post methodology. Resultingly, the culmination of 16 articles is illuminating, and underlines the importance of a reliable study design to evaluate intervention efficacy.

3.2.10. Sample Eligibility

Similar to food waste measurement, there is no singular, standardised approach for recruiting household participants. In the reviewed articles, participants were recruited via a mix of random or convenience sampling methods, or both. Methods utilised to recruit participants include council district lists, waste management regions waste collection [48,50,55], customer lists, companies [49,51,58], and local newsletters and community events [52,57]. Participants were found through researchers or student networks [43,52]. In the reviewed articles, research-specific panels were used the least frequently to recruit participants (e.g., [53]).

Different outcomes require different approaches, and a variety of sampling increases validity in the literature. Thus, a single preferred method of recruitment has not been identified. Random sampling via research panels has a strength of achieving a representative sample, whereas council or waste management district lists offer the benefit of reaching parts of the population who have not signed up to research-specific panels, but can have methodological limitations around data collection. Targeting specific networks can lead to a focused, relevant sample of high food wasters with a greater-than-average need for behavioural change. Opposingly, convenience sampling undertaken by Romani et al. offers lower external validity and is not recommended to be replicated in future research, as it provided neither a general approach fitting the whole population, nor a targeted approach specific to high food wasters [43]. On this basis, studies that recruited people responsible for food management in their household are adequate, as they have the ability to answer questions and complete exercises relating to food waste, and, likewise, have room for improvement in food waste behaviour.

The study samples limit the applicability of some findings from the studies; for example, through networks of students, rather than a random cross-section of the population [43], using supermarket customers of a big chain [58], or based off a list of participants from previous research, mostly for postgraduate studies [56]. By group design, some participants do not undertake each questionnaire or waste audit, leading to incomplete data [55]. Other methodological limitations included showing different intervention messages in addition to the interventions, thus introducing too many variables where any change in behaviour could be put down to methodology [53]).

3.2.11. Data Analysis

In the reviewed articles, the data were analysed using inferential statistics to determine if any statistically significant difference emerged between the starting food waste measures prior to the intervention, and the measures afterwards [89]. Internal validity quality checks and between-group analyses to determine the reliability of samples were reported in the studies, but not shown in Table 3 for brevity.

The statistical approaches varied between authors, dependent on whether the data enabled them to use parametric tests, or if the data were non-parametric [80]. Aligned with other food waste research, commonly used methods included regression analyses, ANOVAs, and ANCOVAs. The analyses undertaken are appropriate for determining if groups are appropriate to compare and contrast, and to check if any differences occur between the baseline, or pre-intervention, and post-intervention follow-up surveys.

3.2.12. Immediate Intervention Results

Promisingly, the reviewed studies indicate a pattern of reduced food waste and the improvement of behaviour between the pre- and post-intervention measures, suggesting that behaviour change interventions can temporarily reduce food waste. However, the studies were often hampered by attrition and insufficient sample sizes post-measurement, resulting in outcomes that could not be validated as a statistically significant improvements [49,53,55]. In one study, greater changes were found in the combined control groups when compared to the treatment groups, and insufficient sample sizes prevented any sufficient further analysis of independent groups [54]. Missing data for two published articles at the post-intervention point results in the efficacy of their short-term interventions remaining unknown [50,58]. The results of their follow-up measures based on the available information are reported below in Section 3.3.

Notably, an insight that emerged from the review suggests that a more involved intervention does not necessarily result in greater behavioural change, as a standard self-affirmation intervention led to a more significant impact than an integrated self-affirmation intervention did [49].

Table 3 presents the analyses and key findings at the post-test and follow-up measures by article. The full results, such as the analysis of variances between different points in time for subsamples, or between subsamples to check for statistically significant demographical differences, were withheld for brevity, and can be found in the original articles.

3.2.13. Longitudinal Intervention Results

Just five of the sixteen studies measured food waste quantity or behaviours at a subsequent point following the intervention, enabling the evaluation of potential enduring changes [24,47,50,52,58]. The remaining studies focused on determining the immediate effects of interventions on food waste, or, in one case [48], undertaking further evaluation at a much later point and subsequently publishing the new results [24].

Of the studies with a longitudinal element, the first [52] indicated continued improvement on baseline behaviours for participants who completed a third assessment eight weeks on from the intervention. However, given the sample size had reduced to just 20 participants by this point, the author does not purport that the intervention caused any sustained changes in behaviour. The intervention used provides a starting point to replicate before the extrapolation of its long-term efficacy can be made.

Secondly, a study found significant reductions in food waste behaviour at the five-month follow-up compared to the baseline, though this was present in both treatment groups and the control group [58]. Of the multiple interventions reported, participants shown either an electronic newsletter or a Facebook intervention displayed a statistically lower quantity of food waste five months on. Following this, these reductions in the quantity of food waste on the post-test and follow-up measures tapered off from time two to time three, indicating that the improvement in behaviour was slowing over time. The statistically significant reduction seen in the control group could have multiple possible explanations, such as external factors equally affecting each group [27], or that each participant could simply be more cognisant of food waste after repeated exposure in the survey design, leading to a retest bias [90]. In light of these confounding factors, this study shows a promising starting point for the longitudinal measurement of interventions, but the temporary reductions cannot be causally linked to the interventions, and their efficacy cannot be sufficiently argued [58].

In 2020, the third study successfully found continued reduction in food waste five weeks after the intervention [50]. However, no measures were taken after this point, and, regrettably, the longevity of the improved behaviours after two months cannot be determined.

More recently, a 2023 study in the United States showed promising results at four post-intervention points in time among the treatment groups, relative to the control group; however, the difference the treatment made was no longer seen at the final longitudinal survey, indicating that any potential changes in behaviour were temporary [47]. Though these results demand further longitudinal research, the authors provided valuable insight into the nature of hard-copy and online intervention materials; they found the difference in efficacy between the treatment groups over time was not significant, and thus a trade-off could be made by reducing costs and going digital, thus enabling a scalable, accessible, and cost-effective solution for a marginally less efficacious intervention [47].

Finally, also in 2023, the ‘Reduce Food, Save Money’ study was revisited [48], recontacting the same sample 31 months later [24]. Despite the findings in 2021 indicating a better outcome for treatment groups than the control group, by 2023, the difference to the baseline was not statistically significant, raising doubt that the changes in behaviour were maintained.

Due to the divergent results of these longitudinal evaluative studies, there are not enough pre- and post-test longitudinal studies on this topic to provide robust evidence of intervention efficacy. The limitations and limited evidence render the review inconclusive on whether behaviour change interventions can lead to a sustained and meaningful long-term improvement on food waste behaviours. After extensive attempts to prevent avoidable food waste, the problem persists, supporting this notion that food waste researchers have not converged on a common framework of approaches.

3.3. Future Research Directions

Throughout the review, important areas for future research have been identified. Evaluations of interventions aimed at reducing household food waste have been conducted, but no standardised behaviour change approaches have been settled on.

For sampling, in order to account for attrition, a robust sample size in each group is a key requirement of longitudinal studies going forward. For the intervention design, the authors advocate for researchers to replicate elements from the plethora of the existing, validated studies to advance the development of methodologies; promising short-term improvements in food waste and planning behaviours were demonstrated through the provision of educational materials at the post-intervention point [43]; and personalised public commitment and goal-setting on a longer-term basis could further this, with a focus on food planning and management [52]. Additionally, using an online approach enables scalability to reach wider and larger audiences [47]. Alternately, the novel method of providing buckets and instructing participants on how to collect and weigh food waste provides a meaningful advancement in scalable, objective food waste measurements that warrants further evaluation, and lends itself to repeated measures over time, suited for a longitudinal study [57].

Studies are recommended to adopt these methodologies within a theory-based framework, drawing on wider contextual social, procedural, and cognitive factors surrounding food behaviour in the household.

Furthermore, studies should be designed to establish causal links between the efficacy of the intervention and waste reduction should there be a change in behaviour. This is supported by the majority of reviewed articles employing randomised controlled trial features, including control groups and random allocation [79].

The unique contribution that this review exposes is the paucity of studies on household food waste reduction through behavioural changes that incorporate a pre- and post-test measurements; the authors argue that one cannot effectively gauge whether an intervention is successful in reducing waste without both a baseline reading to compare changes against, and a repeated measure to evaluate behaviours or waste quantities immediately after the intervention and after some time has passed. Additionally, to enable the attribution of changes to a treatment rather than confounding factors, including a control group in the research design is encouraged.

This research concentrates on the priority areas of food waste among adult consumers in the households of developed countries. Future reviews may look to expand on this in order to address waste in developing countries, or target the behaviour of children to prevent poor habits forming.

4. Conclusions

This review examined the behaviour change interventions applied to consumers in the household, which were aimed at improving the food management behaviours linked with avoidable food waste. To the best of the authors’ knowledge, the current study is the first systematic review which appraises empirical studies taking a before and after measure of food waste in the household, against which to compare outcomes and determine intervention efficacy.

The review finds that just five studies have looked at the impact of interventions over time to inconclusive ends, due to unreliable sample sizes [52], quasi-control groups with changes in behaviour also reflected in control groups [58], the waning impacts of interventions [24,47], and short-term conclusions [50], made too early to determine if the effects diminished or if new habits had remained. Furthermore, the review reveals that few studies are using a robust pre–post methodology when evaluating the efficacy of their interventions, thus requiring further research, and reducing the reliability of the findings in the literature.

Despite the importance of addressing food waste in relation to reducing carbon emissions and mitigating climate change, there is not enough evidence that interventions aimed at changing these behaviours can create the sustained change required to meet this need. The review synthesises 16 implemented and evaluated interventions to provide a foundation for further intervention designs that build upon existing studies with inconclusive results with the aim of streamlining efforts to change future behaviours.

On the whole, the literature provides robust studies across developed countries where the need for change is greatest, and a range of diverse approaches to reducing household food waste. However, this strength also illuminates that little has been accomplished to replicate the successful findings of interventions, and point to a single intervention approach that can address the scale and urgency of the issue. These findings purport that further longitudinal research is required in the behaviour change space to find a sustainable solution that can be implemented on a mass scale to address the complex issue of consumer food waste.

• Limitations and study risk of bias assessment:

A challenge to conducting a systematic review is determining an appropriate research scope that balances both breadth and depth [91]. Due to the exponential increase in articles published on the topic of food waste, screening criteria were chosen to present a focused and nuanced look at the food waste research available. The interventions reviewed can be evaluated for their utility in designing future approaches for reducing the food waste of adult consumers in the household.

The full-text review process revealed there is scope for relaxing the pre- and post-intervention measure exclusion criterion if this process is replicated in future reviews for a meta-analysis. The review results should be interpreted with the focused scope in mind of adult consumers within the household in countries that tend to generate more food waste. The following notable studies may be relevant for similar reviews with widened PICOS parameters: Davenport et al., 2019 [92]; Giordano et al., 2019 [25]; Neff et al., 2019 [93]; Revilla and Salet, 2018 [94]; Russell et al., 2017 [60]; Septianto et al., 2020 [95]; van Dooren et al., 2020 [96]; and Ramos et al. 2023 [13]. To the authors’ knowledge, there are no missing results from this review that fit the inclusion and exclusion criteria or bias due to missing results [36].