Nudging Toward Sustainable Food Consumption at University Canteens: A Systematic Review and Meta-analysis

Introduction: This systematic literature review and meta-analysis investigated the effectiveness of the nudging approach toward sustainable food consumption in the university canteen context. Methods: The systematic literature search was carried out in 5 databases, Web of Science, PubMed, Sco-pus, ProQuest, and the Royal Library, identifying 14 eligible studies and selecting 9 articles containing adequate information for meta-analysis. The nudging strategies were classiﬁed using the typology of interventions in the proximal physical microenvironments framework that resulted in 5 different intervention types: availability, position, size, presentation, and information that belonged to either intervention class-altering properties or placement. Results: The study identiﬁed presentation, availability, and information as the most promising nudge intervention for achieving sustainable food consumption at the university canteen or similar settings. Nudging by altering the properties had a small effect size ( d = 0.16), and nudging by altering placement showed a medium effect size ( d = 0.21). Discussion: Nudging interventions implemented after understanding consumers’ current behavior showed positive effectiveness toward sustainable food consumption rather than implementing random nudges. Conclusions and Implications: It is important that future studies aim to achieve sustainable food consumption by understanding canteen user food preferences and food choice motives before designing a nudging strategy.


INTRODUCTION
The global food system is complex, and currently, it is a significant contributor to global warming, with an estimated 37% of greenhouse gas emissions. 1,2According to the United Nations (UN), the environmental impact of the food systems can be significantly reduced through a shift toward responsible production and consumption. 3As a result, the concept of a sustainable diet and sustainable food system is gaining attention and is increasingly explored in research and international politics. 4,5ccording to the Food and Agricultural Organization (FAO), sustainable diets have low environmental impacts that contribute to food and nutrition security and healthy life for present and future generations. 6−9 However, policymakers and civil society leaders are approaching different strategies to promote sustainable food consumption and to achieve the UN Sustainable Development Goals (SDGs) by 2050.−16 It is an effective way for policymakers to design and evaluate food-related policy instruments. 16The nudging approach does not replace laws, regulations, or economic instruments but allows behavioral observation integration into the policy implementation. 17Furthermore, nudging may be applied to reach policy goals such as the new dietary guidelines published by the Danish government. 18,19nterventions using nudging strategies are significantly nonobtrusive and do not require much cognitive effort to trigger consumer response in various settings. 20,21It steers consumer behavior in a preferred direction that can play a crucial role, for instance, in motivating consumers toward sustainable food consumption. 22,23Importantly, nudges are highlighted in the Intergovernmental Panel on Climate Change's Sixth Assessment Report as having slightly positive transformational potential and environmental effects and highly enabling effects on other food policies. 24Moreover, the latest Intergovernmental Panel on Climate Change's report included nudging within the feasible strategies with positive mitigation potential for combating climate change and achieving the SDGs. 24ustainable consumption (optimal use of resources to minimize the environmental impact) and proenvironmental behavior (behavior to reduce the adverse effects on the environment) could be promoted widely by applying a nudging approach to motivating consumers toward consuming environmentally friendly products. 23,25Previous systematic reviews have addressed implementing nudging strategies to promote sustainable behavior through the food supply chain, 19 healthy food choices, 26 food waste behaviors, 27 and sustainable protein consumption (meat substitutes and alternative protein). 28The article by Ferrari et al 19 identified the effectiveness of green nudges in promoting environmentally sustainable behaviors among food supply chain actors (producers and consumers) in a broader context.However, the authors highlighted a need for more indepth investigations to understand the effectiveness of different nudging strategies in a specific setting, as nudges may influence specific food chain actors differently.
At the consumer level, nudges effectively achieve the desired healthy eating behavior. 29For instance, previous research has shown that altering placement had a modest effect on choosing fruits and vegetables (Cohen's d = 0.3). 30Furthermore, changing portion size showed increased consumption of healthy items (Cohen's d = 0.45), 31 and modification of the plate size seems to have a large effect on calorie intake (Cohen's d = 0.7). 32However, nudges are difficult to upscale to food service and may have little effect, as observed when applying a default nudging intervention-setting a dish by labeling it as the dishof-the-day to reduce meat consumption by promoting novel plant-based alternatives as a replacement.The default nudging intervention did not affect food service when choices were among familiar and similar dishes. 33till, it was effective when choices were made between unfamiliar dishes in a restaurant setting. 34anteen meals are an essential part of the diet for young adults.As the young population enters university, they become more explorative and get influenced by different social environments, which may lead to poor dietary practices. 35,36Furthermore, it is a setting in which a large population of young adults spends their ample time, and the food environment at the university canteen has the potential to play a crucial role in shaping their eating behavior. 37,38Most health beliefs and practices are developed during young adulthood, shaping their and their family's lifestyle. 39,40verall, the main objective of this study is to systemize and synthesize the findings of recent research on nudging approaches to promote sustainable food consumption in university canteens.Furthermore, it aims to quantify the effectiveness of different nudging approaches for promoting sustainable food consumption in university canteens and their heterogeneity.

METHODS AND DATA
A systematic review was performed to select articles that applied nudges to motivate dietary transitions in university canteens, followed by a metaanalysis to estimate its effect size.The systematic review follows the approach of Littell et al 41 to effectively locate and synthesize the research by carefully assessing the quality of the research.The meta-analysis summarises the effect sizes reported in different studies and corrects errors and biases in the research.The research follows the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. 42he literature search was conducted in December 2021 and included all records with titles and abstracts using the following string: "nudg*" AND "sustain*" AND "food consumption" AND ("university*" OR "campus") AND ("canteen" OR "caf e").The search was conducted using the following databases: Web of Science, Scopus, PubMed, the Royal Library, and ProQuest.The search string was adapted to the syntax of each database.Sustainable food consumption is an umbrella concept for sustainable food consumption behavior, covering various aspects such as proenvironmental behavior related to food, climate-friendly food choices, green food consumption, and ethical food consumption. 43herefore, articles that promote sustainable food consumption were selected by including ecolabels, climate labels, meat reduction, and portion-size sustainable dietary choices.
The systematic literature review included peer-reviewed articles published in English between 2011 and 2021.In addition, articles implementing a nudging approach to promote sustainable food consumption were included.The review included randomized control trials and interventional studies conducted in university contexts such as cafeterias, restaurants, dining halls, and canteens.The meta-analysis included articles with sufficient statistical data for effect size calculation (ie, the total number of participants, P, and effect direction). 44urthermore, the review excluded review articles, book chapters, dissertations, work-in-progress or papers-inprogress, articles without access to full texts, interventions not conducted at the university canteen, not using the nudging approach, not focusing on sustainable food consumption, and not being peer-reviewed.Studies not specifying the total number of participants, P, and effect direction were excluded from the meta-analysis.
The articles were selected on the basis of the defined inclusion and exclusion criteria.Titles and abstracts were screened, and full texts were reviewed for relevant articles.Altogether, 250 articles were excluded after refining the inclusion and exclusion criteria.This resulted in 32 articles that were further screened and excluded, if not peer-reviewed, works on progress and dissertation/ thesis.The title and abstract were reviewed by 2 reviewers, identifying 14 articles for the systematic literature review.The selected articles were screened thoroughly to determine the eligibility for the article on the basis of the inclusion criteria.Out of these 14 articles, only 9 were included in the meta-analysis as they provided sufficient statistical data for effect size calculation.The PRISMA flow diagram illustrates the screening criteria for selecting the articles (Figure).
Descriptive data of the final set of 14 selected articles comprising study participants, country of study, year of data collection, intervention strategy, outcome measured, the applied method for data analysis, and findings on the effectiveness of nudging applied.
The typology of interventions in the proximal physical microenvironments (TIPPME) framework was used to identify the fundamental concept of the nudging strategies applied.The purpose of using this framework is to classify and describe the effects of nudging intervention. 45The framework identifies placement (change of environment) and properties (characteristics of the product) as 2 different classes further described according to 6 different intervention types (availability, position, functionality, size, presentation, and information).The classification and type of intervention implemented for selected articles are provided in Supplementary Table .The type of nudging approach used in the selected articles was outlined on the basis of the intervention focus of selected studies that resulted in 5 different intervention types: availability, position, size, presentation, and information that belonged to either intervention class properties or placement.Furthermore, some of the selected articles used more than 1 nudge type; thus, in this case, the effect of nudges was analyzed separately and was out-lined in more than 1 intervention type and class (Supplementary Table ).The meta-analysis used SPSS soft-ware (version 22, IBM, 2013) for com-puting sample size, P, and effect direction for each selected article.The main outcome variables (eg, means, percentage, and total sales of items) of the intervention reported in the 9 articles were included in the meta-analysis.Moreover, some of the selected articles implemented more than 1 nudge and reported more than 1 result; thus, the outcomes were analyzed separately, which re-sulted in a total of 13 studies for the meta-analysis.The analysis was con-ducted with continuous outcomes using precalculated effect size using SPSS software.The standard difference in mean, confidence intervals, and P were computed for each study.The significance of effect size is based on Cohen's d; thus, effect size ≤ 0.2 is small, 0.2−0.5 is medium, and > 0.5 is large.44 The result from homogeneity tests shows heterogeneity between the studies; thus, a random-effect model was performed for the meta-analysis as the model presumes that the effects in the studies follow some distribution and are not alike.46

RESULTS
Table 1 briefly describes the aim, country, nudging class/type, data analysis method, main findings, out-come measured, study design, length, sample size, and effect sizes of the selected articles.In total, 10 of the study designs were randomized controlled trials, and the length of the intervention varied from 4 days to 2 years.The sample size of the studies was based on the number of participants or meals served.The sample size in some studies was rela-tively small compared with others; however, all selected articles had > 100 participants, which is the mini-mum sample size requirement for a meaningful result.47 Most studies were conducted in Europe (n = 11), whereas the rest (n = 3) were con-ducted in the US.Most articles (n = 11) applied altered properties, 2 altered placements, and only 1 altered placement and properties.The effect size estimation for the meta-analysis based on Cohen's d shows that most included studies showed medium to small effect sizes.Table 2 shows the forest plot of the meta-analysis.The overall effect size of Cohen's d = 0.06 (95% confidence interval [CI], 0.009−0.063)(P = 0.27) was obtained from the random-effect model for the studies using the nudging approach toward sustainable food consumption.The result shows that 5 out of 13 studies show a medium effect size.The studies that showed modest effect size were Piester et al 49 ) was used to interpret the publication bias in the study that shows the asymmetrical distribution, indicating that the studies are likely to have publication bias and study heterogeneity. 62ltering placement of the food product: 1 out of 13 articles implemented nudging by altering the placement that showed a highly significant effect (P = 0.004) with a medium effect size of d = 0.21 (95% CI = 0.067−0.356). 53ltering properties of the food product: 12 out of 13 studies implemented nudging by altering the properties in which the studies that showed significant effect were Jalil et al (P = 0.04), 63 Lorenz-Walther et al (P = 0.01), 58 Piester et al (P = 0.04), 49 and Visschers and Siegrist (P < 0.01) 51 with mostly medium effect sizes d = 0.27 (95% CI, 0.011−0.539), 49d = 0.22 (95% CI, 0.081−0.364), 51and d= 0.28 (95% CI, 0.011−0.555) 61but low effect size for d = 0.17 (95% CI, 0.041−0.307). 58A medium effect size d = 0.37 (95% CI, 0.149−0.597)with a highly significant effect P < 0.01 in study 1 and a small effect size d À0.07 (95% CI, À0.291 to 0.154) with insignificant effect P = 0.53 in study 2 were observed. 50Other studies did not show any significant effect with very small effect sizes Brunner et al (P = 0.05), 50 Griesoph et al (P = 0.38), 52 Lorenz-Walther et al (P = 0.07), 58 Visschers and Siegrist (P = 0.24) 51 and in both phases of study by Slapø and  P = 0.38). 56The overall effect of nudging by altering the properties was not significant (P = 0.08), with a small effect size d = 0.16.

DISCUSSION
This systematic literature review and meta-analysis aimed to investigate the effectiveness of nudging approaches implemented at the university canteens.Most articles implemented the nudging intervention to achieve behavioral change toward increased environmental sustainability and lower greenhouse gas emissions.Other articles applied behavioral nudges promoting nutritionally balanced food while promoting environmentally friendly behavior.The nudging interventions were classified and type identified according to the TIPPME framework.The framework identified 2 articles that applied altering placement, 11 applied altering properties, and 1 applied altering of properties and placement of the food product.
The key outcomes are discussed following the different classes of nudging identified through the TIPPME framework.

Altering Placement of the Food Product
The article by Garnett et al 53 implemented nudging by altering the placement that, showed a medium effect size (d = 0.21) and reported an increase in sales of vegetarian meals by 41% in the observational and 79% in the experimental field study.The finding aligns with a previously conducted meta-analysis that reported a medium effect size (d = 0.39) when altering the placement of fruits and vegetables. 30Another article by Garnett et al 60 altered the position by placing the vegetarian option first.The finding showed an increase in the sale of vegetarian options by 2.3% at dinner time but did not show any effect at lunchtime that may have been influenced by other aspects of the choice environment.Previous studies have shown that items displayed at the start and end can significantly influence consumers' cognitive decision-making, 64,65 and repositioning food items impacted promoting healthy food choices. 63hus, future interventions aiming to reduce meat consumption could increase the availability of vegetarian options by repositioning food items that may influence consumers' choices.

Altering Properties of the Food Product
A negative effect size (d = À0.07) was reported in the article by Campbell-Arvai et al 54 that provided information on appealing or unappealing meat-free options on the menu.The article evaluated the perceived level of appeal on the basis of Positive and Negative Affect Schedule scales by Watson et al 66 using a series of bipolar scales, for instance, desirable vs undesirable.Moreover, a medium effect size (d = 0.37) was obtained in the same article when presenting appealing or unappealing meat-free meal options in the default menu that increased the consumption of meat-free options by 89.7%.The findings are similar to what has been observed in previous research that modified menus to promote plantbased meals with dish-of-the-day and other default interventions. 67,68Furthermore, the articles by Andersson and Nelander 59 and Kurz 55 also implemented menu modifications by increasing the visibility of vegetarian options, which resulted in a significant increase in sales of vegetarian meals.These findings align with previous research showing a willingness to purchase vegetarian foods instead of meat because of menu modifications. 69,70iester et al 49 modified the menu using sustainability labels based on the degree of the environmental impact of food items.The application of sustainability labels in the menus reported a medium effect size with a significant increase in the purchase of veggie burgers among women.However, a similar approach has been implemented in supermarkets and restaurants that did not significantly affect consumers' decisionmaking, 71,72 indicating that these sustainability labels may create confusion and overload consumers with complex information.−75 The application of traffic-light labeling as an indicator of the degree of the environmental impact of the food item reported a reduction in meat sales and an increase in sales of the green-labeled meat items; however, no effect was seen when using green and red colors separately. 50,56urthermore, the article by Spaargaren et al 57 also implemented a blackand-white carbon label that showed no effect in Phase I and only a modest impact in Phase II, in which a colored carbon label (influenced by a traffic light) with comprehensive information was implemented.Previous interventions showed the effectiveness of using traffic-light labeling with additional information 76 and for a longer time, 77 leading to positive shifts toward sustainable and healthy choices.Traffic-light labeling triggers consumer responses unconsciously, thus easing the cognitive burden of making informed choices 78 ; thus, these efforts could be integrated with redesigning the meal offers for effective nudging toward plant-based meal choices and nutrient intake. 35Therefore, implementing traditional trafficlight labeling, together with some additional information for a more extended period, may boost consumers' awareness to achieve better outcomes in the future.
The article by Visschers and Siegrist 51 implemented a combination of climate-friendly labels and information posters, which increased the purchase of climate-friendly meal choices by 9.7%, showing a medium effect (d = 0.22).However, in the same study, a negative effect (d = À0.16) was seen when consumers rated the taste of the meal, which was then related to the meal's global warming potential and perceived environmental impacts.
Moreover, the effect of unsustainable food choices on climate change and the benefits of reducing meat consumption was promoted by conducting lectures and providing information in the article by Jalil et al. 61 The article reported a medium effect size (d = 0.28) that showed a decrease in the purchase of meat-based meals and a simultaneous increase in the purchase of plant-based meals.
However, the effectiveness of information posters on consumer behaviors showed a negative effect size (d = À0.18) in the article by Lorenz-Walther et al. 58 In contrast, implementing information, education, and posters has previously been used in school canteens as an effective tool. 79These differences indicate that the successful use of information as a nudging tool is highly determined by the quality of meals offered at the canteen. 35urthermore, the article by Griesoph et al 52 implemented descriptive and guessed norms as a nudging strategy that showed a negative effect size (d = À0.08).The descriptive norms were based on how people behave, whereas the guessed norms were interindividually changing guesses of a particular descriptive norm based on an individual mindset about behavioral patterns.The findings from the article show that using guessed norms as a nudging tool did not show any significance in promoting climate-friendly meal choices.
The article by Lorenz-Walther et al 58 reported a small effect size (d = 0.17) with a significant indirect effect of reducing the plate size on plate waste.However, previous research showed that portion size reduction was an effective nudging tool in promoting sustainable consumer behavior at a store. 80

Altering Placement and Properties of the Food Product
The article by Colleoni et al 48 implemented 3 different types of nudges, in which altering the position of wholemeal bread and white bread to make the former more visible resulted in a 200% increase in consumption.The study also changed the layout, raised the fruits and vegetable basket, and achieved 50% more consumption of raw vegetables.The study further verified a piece of simple information labeling (ie, so good on healthier food items) that positively influenced canteen users' choices.In addition, varied food choices resulted in less food waste as the amount of unused food was reduced.Thus, future studies could implement a similar approach to prevent food waste in canteens.These nudges were designed on the basis of findings from a survey exploring eating behaviors instead of implementing random nudges; thus, future studies need to consider implementing nudges after understanding the current behavior of the users. 81he review has a clear scope and predefined inclusion and exclusion criteria for the type of intervention and study setting.The search strategy for the systematic literature review has followed the requirements of the PRISMA statement.Furthermore, the metaanalysis identified publication biases using a funnel plot, and asymmetry was assessed using the Egger method.
The study has several limitations.First, the study's findings are limited to the few articles studied.Second, the results are based on the primary studies and, thus, carry their limitations.For instance, some of the primary studies reported a loss of individual purchase data, missing data, and data misclassification that could have resulted in measurement errors in the findings of their studies. 55,59Furthermore, in the study by Andersson and Nelander, 59 the meat option was displayed on top of the website menu for 6 weeks, which could have significantly affected the intervention outcome.Finally, Piester et al 49 fail to explain the long-term effect of sustainable menus and labels on buying sustainable foods, as the purchase is only measured once.
Moreover, some studies did not measure the effect of each nudging tool, limiting an explanation of the impact of nudges separately.Furthermore, because of the lack of previous studies among Danish young adults, the background and comparisons are limited to research conducted among other age groups in Denmark.Finally, a risk of bias assessment could not be conducted for this study as only limited studies have specified the risk of biases in the primary studies.

IMPLICATIONS FOR RESEARCH AND PRACTICE
This study identified 14 articles for the systematic literature review, out of which 9 articles were selected for the meta-analysis.The study identified increasing availability, modifying the menu, and providing information through labels and lectures as the most effective nudges for promoting sustainable food consumption behavior.The most used nudging types were traffic-light labels/carbon labels and information/posters; however, their effects on behavior are ambiguous.For instance, implementing traditional traffic lights (red, green, and yellow) showed a modest effect, whereas modified traffic-light labels (black-andwhite) did not show a significant result. 57Furthermore, combining 2 or more nudging strategies effectively yields sustainable food consumption behavior change. 48,58Thus, the findings from this review may support designing a nudging strategy that could further verify the integration of multiple nudging strategies for behavioral change and future food policy and guidelines.
In addition, it is important that future studies investigate the effect of integrated nudging strategies over time and their cost-effectiveness.As the primary studies support, it is recommended to prioritize baseline studies for understanding consumer behavior to select appropriate nudging strategies rather than implementing random nudges. 48,49,51−84 This is essential as there might be an attitude-behavior gap in sustainable food consumption as consumers constantly make tradeoffs, such as price, health, and nutrition information. 71These nudging efforts might also be integrated with pricing strategies to influence consumers' purchasing behavior, particularly in university canteens. 85,86Finally, future studies might also consider comparing and contrasting the effectiveness of nudging strategies between other forms of canteens, as well as restaurants and grocery stores.

Figure .
Figure.Preferred Reporting Items for Systematic Reviews and Meta-analyses flow diagram indicating the selection process of publications.

Table 1 .
Summary of Interventional Studies That Applied the Nudging Approach to Promote Sustainable Food Consumption With a Description of the Country, Aim, Nudging Class/Type, Analysis, Findings, Outcome, Design, Length, Sample Size, and Effect Sizes of the Study

Table 2 .
Result of Meta-analysis for Nudging Sustainable Food Consumption