papers UPCommons Portal del coneixement obert de la UPC http://upcommons.upc.edu/e-prints Aquesta és una còpia de la versió author’s final draft d'un article publicat a la revista Waste Management. URL d'aquest document a UPCommons E-prints: http://upcommons.upc.edu/handle/2117/108970 Article publicat / P ublished paper: Derqui, B.; Fernandez, V. The opportunity of tracking food waste in school canteens: guidelines for self-assessment. "Waste Management", Novembre 2017, vol. 69, p. 431-444, DOI: 10.1016/j.wasman.2017.07.030 © 2017. Aquesta versió està disponible sota la llicència CC-BY-NCND 3.0 http://creativecommons.org/licenses/by-nc-nd/3.0/es/ http://upcommonsdev.upc.edu/ http://upcommonsdev.upc.edu/ http://upcommons.upc.edu/e-prints http://upcommons.upc.edu/handle/2117/108970 https://doi.org/10.1016/j.wasman.2017.07.030 http://creativecommons.org/licenses/by-nc-nd/3.0/es/ 1 The opportunity of tracking food waste in school canteens: 1 guidelines for Self-Assessment 2 3 Derqui, B.; Fernandez, V. 4 5 Abstract 6 Reducing food waste is one of the key challenges of the food system and addressing it in the institutional 7 catering industry can be a quick win. In particular, school canteens are a significant source of food waste 8 and therefore embody a great opportunity to address food waste. The goal of our research is the 9 development of guidelines for audit and self-assessment in measuring and managing food waste produced 10 at school canteens. The purpose of the tool is to standardise food waste audits to be executed either by 11 scholars, school staff or by catering companies with the objective of measuring and reducing food waste at 12 schools. We performed a research among public and private schools and catering companies from which 13 we obtained the key performance indicators to be measured and then pilot-tested the resulting tool in four 14 schools with over 2,900 pupil participants, measuring plate waste from over 10,000 trays. This tool will help 15 managers in their efforts towards more sustainable organisations at the same time as the standardisation 16 of food waste audits will provide researchers with comparable data. The study suggests that although there 17 is low awareness on the amount of food wasted at school canteens, managers and staff are highly interested 18 in the topic and would be willing to implement audits and reduction measures. The case study also showed 19 that our tool is easy to implement and not disruptive. 20 Keywords 21 Food waste; School Catering; Self-assessment Tool; Sustainability Metrics; Food Waste Audit, Awareness 22 building, Food waste prevention. 23 24 1. Introduction 25 26 The global food system still has to solve deep problems in order to be truly sustainable. One of the key 27 sustainability challenges brought up by researchers (e.g. Clarke et al., 2015; Finn, 2014; Garrone et al., 28 2014) in the last few years is waste. In particular, reducing food waste (FW) would aid in the path towards 29 a more sustainable global food system as it would imply a more efficient (and ethical) use of scarce natural 30 resources at the same time as helping reduce its significant environmental footprint (Buzby and Guthrie, 31 2002). This is particularly challenging in developed countries, as food waste is very closely related to 32 individual behaviour and cultural attitudes towards food (Godfray et al., 2010). 33 Business managers are at present considered the major actors trying to implement sustainable 34 development, opposed to some years ago, when focus was put on local authorities (Dyllick et al., 2002). In 35 fact, many companies and institutions, particularly schools (Rickinson et al., 2016), have initiated a full set 36 of sustainable development initiatives to address the demands of public and private stakeholders. With 37 regard to food waste, progress has been slow, mostly due to lack of awareness (Finn, 2014). Hence, 38 increasing visibility and awareness on food waste through audits is an obvious place to start. Once food 39 waste has come to light, people will probably be willing to act against it, managers will probably become 40 more concerned about its financial impact and kitchen staff about its social implications (Goonan et al., 41 2014). In any case, food waste auditing should be the starting point of a food waste awareness campaign. 42 2 As schools are a natural place for education, and making the most of the near universal attendance of 43 school by children and the fact that they are on the premises for many hours a day (Dehghan et al., 2005), 44 addressing food waste at school canteens becomes noteworthy. However, regulators, school managers, 45 and catering companies very rarely concentrate on reducing food waste. Instead, they usually focus on 46 analysing how effective nutritional programmes are (Wilkie, 2015). For this reason, most researchers have 47 limited their studies on food waste at schools to the analysis of plate waste (PW), concerned with the 48 nutritional value of effective dietary intake. Our research has a broader purpose, offering a more holistic 49 approach on school food waste. Indeed, standard criteria for measuring school catering food waste is novel 50 in the literature, particularly as we propose to include both pre-consumer and post-consumer waste in our 51 assessment tool, while most researchers in this area have focused their work on analysing plate waste (e.g. 52 Adams et al., 2005; Byker et al., 2014; Cohen et al., 2013; Marlette et al., 2005; Rodriguez Tadeo et al., 53 2014). Moreover, through a standardised tool, researchers will be able to compare results and data from 54 different studies. The goal of this research is to provide schools and educators as well as catering companies 55 with a set of principles and tools that unveil and quantify food waste at school canteens and therefore 56 facilitate the implementation of reduction measures and result tracking. With this purpose, we first analyse 57 the nature and types of food wasted at schools as well as cafeteria managers´ attitudes toward food waste 58 and end with the development of a self-assessment waste tool. This research has a very precise managerial 59 implication. As a final outcome, a simple and easy to implement auditing tool has been developed. Through 60 it, we aim to help managers and pupils in their efforts to increase the sustainability of the food system. The 61 study is particularly relevant for schools with in-house kitchens, no matter if the service is outsourced - 62 managed by a catering company - or not. Nevertheless, the tool could be applied to other business models 63 too, with little modification. The scope of this research includes school canteens in both public and private 64 schools. To achieve the goals of this research, we collected primary data from public and private schools in 65 Spain. 66 67 2. Literature Review 68 69 2.1. The opportunity of addressing Food Waste in Institutional Feeding Systems. 70 Food waste can be defined as all the products that are discarded from the food chain while still preserving 71 their nutritional value and complying with safety standards (Falasconi et al., 2015). Estimates on the amount 72 of food wasted globally are striking: FAO estimates that up to one third of global food produce is wasted, a 73 fact that places food waste as one of the top challenges for global sustainability (FAO, 2011). In Europe, 74 despite acknowledging that food waste is a data-poor area across the main sectors in which it arises, the 75 European Commission has quantified current average annual food waste at 200 kilos per capita, stating 76 that this figure will increase significantly in the next years if no action is taken. They therefore recommend 77 member estates to act, setting the objective of halving EU disposal of edible food by 2020 (European Union 78 Committee, 2014). 79 On the other hand, researchers mention that a big impact may be achieved when addressing food waste at 80 places where there are many individuals dining at the same place (Mirosa et al., 2016).This is especially 81 true in the institutional catering industry (schools, hospitals and prisons) where, as underlined by Mirosa et 82 al. (2016), many individuals dine similarly, and therefore both efficiency along the supply chain and plate 83 waste can be addressed. Moreover, Goonan et al. (2014) state that food service institutions are big 84 3 producers of food waste, mostly during service, but also as a result of overproduction. In particular, 85 researchers state that school canteens embody a significant source of food waste (Adams et al., 2005; 86 Smith and Cunningham-Sabo, 2014) and represent an ideal opportunity for minimising food waste foot print 87 (Wilkie, 2015). Food waste was found by Wilkie (2015) to be the predominant component in a school 88 canteen waste audit in three schools in Florida (US): between 58% and 69% of total waste weight was food, 89 far more than paper, plastic & glass wastage. The mean daily food waste per pupil was averaged between 90 60.1 and 95.33 g. in schools with an in-house kitchen in this research. Therefore we can state that the 91 institutional catering industry represents an ideal opportunity to divert food waste from landfills thanks to 92 their concentrated food waste stream due to the fact that they serve a high number of meals at a single 93 location, resulting in food waste collected at only one location too (Wilkie, 2015). As a consequence, the 94 institutional catering industry becomes crucial in the fight against food waste (Mirosa et al., 2016). 95 Food waste at school canteens could be reduced through educating pupils and staff in order to change 96 behaviours that cause food waste (Wilkie, 2015). Youths concerned about food waste were found, by 97 Principato et al.(2015), to be more likely to reduce leftovers. Furthermore, we can assume that these 98 improved behaviours and habits will prevail into their adulthood (Guthrie and Buzby, 2002). Mirosa et al. 99 (2016, p.12) found one of the key reasons for consumers not to waste food was a cultural tradition: “those 100 who had grown up with the belief that they need to clean their plates” produced less plate waste. These 101 more sustainable habits could be passed on further and have an effect on the amount of waste produced 102 by future generations (Mirosa et al., 2016). There is evidence in the literature on the effectiveness of waste 103 reduction initiatives. For instance, Ensgtröm (2004) carried out research aiming to measure the impact of a 104 food waste reduction campaign in a school in Sweden resulting in a 35% reduction in plate waste compared 105 to a baseline score. It is also acknowledged by researchers that people with a high knowledge of issues 106 related to food waste are more likely to avoid waste (Principato et al., 2015). 107 Reducing Food Waste has obvious environmental and ethical benefits at the same time that it also has 108 relevant economic implications as its associated costs are not only related to procurement of food 109 ingredients, but also to disposal costs (Papargyropoulou et al., 2014). Moreover, both schools and families 110 could save some money by reducing food waste: pupils who eat more at school are less likely to spend 111 money on substitutive products outside the canteen (Cohen et al., 2013). 112 2.2. Food Waste Auditing and Reporting 113 Good sustainability performance is linked to a full and honest commitment of management to sustainability 114 and to the adoption of incentives, something that should be done by setting appropriate goals, monitoring 115 and evaluating progress (Székely and Knirsch, 2005). As stated by Gerbens (2003), measuring tools 116 offering light on the sustainability performance of a firm turns out to be the very first move towards 117 sustainability. More precisely, food waste inventories are claimed to be critical for the development of 118 effective reduction initiatives and monitoring progress overtime (Hanson et al., 2016) . Conducting a waste 119 audit in both the preparation and the display areas (kitchen and service line) as well as in the pupils’ canteen 120 is the first step towards reducing food waste produced at schools (Bradley, 2011). 121 2.2.1. Framework 122 The World Resources Institute (Hanson et al., 2016) together with partners such as WRAP, UNEP and 123 FUSIONS have developed a Global Food Loss and Waste Accounting and Reporting Standard aiming to 124 provide guidance for governments and organisations to carry out inventories on food loss and waste. We 125 have used this standard as a framework for waste auditing analysis. 126 4 As stated by the WRI, a Food Loss and Waste inventory must be based on the five principles of relevance, 127 completeness, consistency, transparency, and accuracy (C. Hanson, B. Lipinski, K. Robertson, D. Dias, I 128 Gavilan and J. Fonseca, 2016, p. 29). Relevance because it should contain the necessary information for 129 the intended user to make decisions and because the quantification method should be selected based on 130 the specific goals to achieve. Completeness because no relevant data or component should be excluded 131 from the inventory, unless justified. WRI researchers go further adding that auditing methods should be 132 consistent, allowing comparable measurements along-time in order to permit the identification of trends and 133 the assessment on the performance of the audited institution. Transparency is gained by clearly reporting 134 the quantification method. Finally, they acknowledge a trade-off between accuracy and completeness and 135 cost and suggest choosing the optimal method based on the needs and resources of the institutions. 136 Regardless of the objective and scope of the audit, entities should report on the following four elements (C. 137 Hanson, B. Lipinski, K. Robertson, D. Dias, I Gavilan and J. Fonseca, 2016) (World Resources Institute, 138 2016): 139 1. Time frame. Exact start and end date of the audit should be recorded. It is recommended to take 140 seasonal variations into account when planning waste audits. 141 2. Boundary (organisation, geography, etc.) and particularities of the sample. 142 3. Scope (types of waste included). Records must include the type of food waste, the reason that 143 caused it (e.g. overproduction, spoilage, trim waste…) as well as the estimate of loss (by weight or 144 portions). 145 4. Waste destination (where waste goes after being discarded) must be accounted and reported 146 because there are a wide range of possible destinations for food waste with very different 147 associated environmental impacts. 148 The WRI Food Loss and Waste standard (World Resources Institute, 2016) establishes that methods, 149 estimates and possible bias must be clearly documented and disclosed in a neutral manner. The auditing 150 system should also register who recorded the data. Moreover, Bradley (2011) strongly recommends that 151 the results of the audit are shared and discussed with the kitchen team and suggests that it could also be a 152 great learning opportunity for pupils. 153 Due to their interest and particularities, in this section we shall further develop both the scope of the audit 154 and waste destination. 155 2.2.2 Audit Scope and Categorisation 156 The scope of the audit must be clarified before beginning to measure food waste. Papargyropoulou et al. 157 (2014) mention the relevance of distinguishing between avoidable and unavoidable food waste as a key 158 factor in a food waste prevention strategy. Wrap’s definition of avoidable food waste includes food discarded 159 because it is unwanted or has been allowed to pass its best (Ventour, 2008), therefore avoidable food waste 160 had previously been edible, although it might or might not be edible at the time of disposal. Papargyropoulou 161 et al. (2014) explains that avoidable food waste includes foods or parts of food, usually considered edible, 162 while unavoidable food waste is food that has never been edible, such as bones, fruit skins, etc. As 163 described by Wrap, this includes waste from food that one would not expect people to eat (Wrap, 2011). 164 Despite this classification being subjective, unveiling avoidable food waste reveals the substantial potential 165 for food waste prevention (Papargyropoulou et al., 2014). 166 This leads us to the very first key characterisation when analysing food waste: whether it could possibly be 167 avoided or not. Potentially avoidable waste might not have ended up as waste with better management 168 5 while inedible food conforms to unavoidable waste. Whether to quantify both food and associated inedible 169 parts removed from the food supply chain when performing a waste audit, the choice of studying only food, 170 or only associated inedible parts, is to be decided depending on the purpose of the waste audit (Hanson et 171 al., 2016). 172 The vast majority of studies use some kind of further classification for the discarded food, usually related to 173 the place or moment where waste is generated. Table 1 shows a few examples of classifications for 174 avoidable and possibly avoidable waste used by researchers when analysing food waste. 175 Table 1. Characterisation of food waste by researchers, some examples 176 177 Additionally, as noted by Papargyropoulou (2014), distinguishing between food waste and food surplus is a 178 must when addressing food waste: food surplus is food produced beyond our nutritional needs while food 179 waste is a consequence of food surplus. Proper meal planning will help caterers minimise food surplus and 180 therefore the planning process should in some way be included in a waste audit. 181 With regard to plate waste, there is consensus in the literature on its definition (Mirosa et al., 2016). The 182 term plate waste is used by researchers to refer to the amount of food served to pupils that is finally 183 discarded. Its measures have been used with two main purposes: in order to decide how much food to 184 prepare or order and more importantly to judge how well pupils accept the meals offered (Buzby and Guthrie, 185 2002) and assess their dietary intakes. 186 On top of the above mentioned classifications, most researchers measure food types in each of the previous 187 categories separately. Depending on the purpose of the study, food type classifications can be broad, like 188 the one used by Byker et al. (2014) or Cohen et al. (2013) who classify food types into only four groups 189 (main entree, fruit, vegetables and milk) or more detailed, like Marlette et al. (2005, p. 1), who mentions 190 plate waste by the specific food item, such as applesauce, green peas, etc. using a more comprehensive 191 Author (Derqui et al., 2016) (Engström and Carlsson-Kanyama, 2004, p. 206) (Ferreira, Martins, & Rocha, 2013, p. 1630) (Falasconi et al., 2015) (Clarke et al., 2015, p. 2) Sector Food Service food service institutions University Catering School Catering Consumer (Households) Boundary Spain Sweden Portugal Italy USA Characterisation of food waste PRE- CONSUMER WEIGHT Storage losses Weight of raw and cooked food not distributed (“leftovers”) “Avoidable” unserved food Losses during cooking and preparation Preparation losses (mostly seeds, peel, etc. from fruits and vegetables) Serving loss (left on serving dishes and in canteen kitchens and food wells) “Physiological” unserved food (cooked in excess to ensure some extra portions) Food discarded due to preparation of too much food, expired use-by/open dates, or spoilage Leftovers (prepared food never served) POST- CONSUMER WEIGHT Plate waste (what the diner leaves on the plate) Plate waste (items returned at tray collection, after scraping of non- edible discards such as bones, peels, etc.) Food served but not consumed (“serving dish leftovers”) Plate waste or loss 6 classification with 10 food type groups:(a) mixed dishes (b) meats(c) grains (d) milk (e) cheese (f ) 192 vegetables (g) fruits (h) sweet snacks (i) savoury snacks and (j) beverages. Moreover, as mentioned before, 193 other researchers use the nutrient content of food for their analysis instead of food types (e.g. Bergman et 194 al., 2004). 195 196 2.2.3 Waste Destination 197 Whenever the goal of the audit might include an analysis of environmental impacts or at least an increase 198 of the awareness on food waste environmental footprint, waste destination should be recorded. The 199 environmental impact of food waste varies greatly depending on how it is discarded (Creedon, M., 200 Cunningham, D., & Hogan, 2010)(Creedon, M., Cunningham, D., & Hogan, 2010; Papargyropoulou et al., 201 2014). Typical destinations of food waste can be landfills, animal feed, anaerobic digestion, biomaterial and 202 compost, among others (C. Hanson, B. Lipinski, K. Robertson, D. Dias, I Gavilan and J. Fonseca, 2016). In 203 fact, destinations differ significantly, from the most favourable to the least favourable environmental option 204 in the waste management hierarchy (Papargyropoulou et al., 2014). Using the waste hierarchy as a 205 framework, Papargyropoulou et al. suggest different options for dealing with food surplus and food waste 206 where food surplus prevention is at the highest level of the pyramid. At the following step they suggest 207 redistribution for human consumption, animal feed and compost. Finally, at the lower levels, they list the 208 worst environmental options, such as energy recovery (e.g. anaerobic digestion) and disposing of food 209 waste in landfills - which they state should be used as the last option (Papargyropoulou et al., 2014). 210 Following the above-mentioned hierarchy, Creedon et al. (2010) state that from an environmental 211 perspective, the best way would be of course not to produce food waste or to prevent food waste from over 212 preparation, over trimming, etc. Secondly, he mentions reusing food for feeding people by reusing it in other 213 meals, donating to the needy, or even diverting it to feed animals. Thirdly, he states that food waste should 214 be recycled by composting or other processes. Finally, landfill disposal arises as the worst option for the 215 environment and is at present regulated in many countries (Creedon, M., Cunningham, D., & Hogan, 2010). 216 217 2.3. Methods for measuring Food Waste 218 Most of the academic work on food waste in the catering industry has been conducted in schools or hospitals 219 (e.g. Cohen et al., 2013; Williams and Walton, 2011) and is often focused on plate waste (Adams et al., 220 2005; Buzby and Guthrie, 2002), being researchers concerned with the nutritive intake of children as well 221 as with the efficiency of school nutrition programmes (e.g. Adams et al., 2005; Marlette et al., 2005; Smith 222 and Cunningham-Sabo, 2014). Quantification methods in the literature are diverse. Comstock (1979) 223 analysed and compared seven methods of measuring plate waste in the institutional food service, classifying 224 them into two groups: direct and indirect measures of waste, depending on whether waste was actually 225 weighed or estimated. 226 Direct (physical) measurement of plate waste is the most commonly used method by researchers, aiming 227 to measure food intake at schools by the actual weighing of food discarded by children (e.g. Bergman et al., 228 2004; Cohen et al., 2013). Aggregate measures involve collecting all food waste and weighing the total bulk 229 amount for a population (e.g. all meals from one sitting), while individual measures record either the total 230 food remaining on each individual tray or the weights of each food component on each plate (Williams and 231 Walton, 2011). Individual weighing is reported by researchers to be more accurate, despite its high logistical 232 burden being a relevant disadvantage and it may make it difficult to implement without disrupting or delaying 233 7 normal foodservice operations (Comstock, 1979; Jacko; C. C.; Dellava; J.; Ensle; K.; & Hoffman; D. J., 234 2007). Furthermore, when measuring waste individually there is a high risk of children changing their 235 consumption patterns if being observed, thus biasing results (Guthrie and Buzby, 2002; Jacko; C. C.; 236 Dellava; J.; Ensle; K.; & Hoffman; D. J., 2007). 237 Moreover, individual or aggregate measurements can be done selectively, that is, differentiating the weight 238 of each food component, or non-selectively. Comstock (1979) criticised aggregate non-selective plate waste 239 for not providing enough information and actually recommended aggregate selective plate waste defending 240 that it was fast, accurate and easy to learn while at the same time providing adequate information. Going 241 further on aggregate measures of plate waste, Jacko (2007) recommends the plate-waste method, which 242 he describes as follows: first the mass of food being served is measured by weighing each item in the menu; 243 then, after finishing eating, pupils are asked to discard individual food items into different labelled plastic 244 tubes for waste (e.g. #1 beans, #2 bread, #3 meat,…) Then, total weight per item is recorded (net of the tub 245 weight) obtaining the total amount of food waste. The difference between mass of each item served and 246 wasted is the estimated food intake. Jacko (2007) concluded from his research that there were no 247 statistically relevant differences between the estimations on energy and nutrient intake in children at school 248 obtained using aggregate selective or individual physical measurements of plate waste. 249 Indirect measures include both visual estimation and dietary recall (named self-estimation of plate waste by 250 Comstock (1979). Although Comstock (1979) considered visual estimation by trained observers as being a 251 non-obtrusive method, not too time consuming, they did not recommend it as its accuracy had not been 252 adequately tested at that moment.. More recent researches (e.g. Rodriguez Tadeo et al., 2014) have 253 concluded that it can be a valuable method. Visual estimation is done based on different grading scales for 254 plate waste, Comstock’s is the most commonly used, with 6 grades: full plate, almost full plate, ¾ plate, ½ 255 plate, ¼ plate and empty plate (Rodriguez Tadeo et al., 2014). Despite Buzby (2002) mentioning that ratings 256 can differ among observers as being a disadvantage of this method, Rodriguez Tadeo et al.’s (2014) 257 research concluded that the visual scale was a reliable tool for measurement, although acknowledging the 258 need for training catering staff as being inconvenient. Williamson (2004) performed research aiming to 259 validate digital photography for measuring food portions (food served, food intake and plate waste) 260 comparing it with direct visual estimations and weighed foods, concluding that both the direct visual 261 estimation method and digital photography results were highly correlated with actual weighed food, and 262 therefore, are valuable methods, although they acknowledge that both methods tended to slightly 263 overestimate portion sizes compared to weighed food methods. Williamson (2004) supports the validity of 264 both digital photography and direct visual estimation methods, based on the results of his research 265 comparing results of both methods with actual weighing. He recommends digital photography for being less 266 obtrusive and less disruptive in the eating environment. 267 On the other hand, when using the dietary recall method, children are asked about the type and amount of 268 food eaten. Despite this method being easy to implement and low cost, results are highly biased by 269 children’s ability to recall (Jacko; C. C.; Dellava; J.; Ensle; K.; & Hoffman; D. J., 2007), as well as by the fact 270 that children may want to please educators (Buzby and Guthrie, 2002). Comstock (1979) criticised both food 271 preference questionnaires and self-estimation for not being reliable. 272 Table 2 summarises the pros and cons mentioned by researchers of the different measurement methods, 273 based on Comstock’s (1979) classification of methods in direct or indirect measures of waste. 274 275 8 Table 2. Methods for measuring food waste 276 Method Advantages Disadvantages DIRECT MEASURES OF WASTE Individual Plate Waste Accuracy Specific information provided (e.g. by sex, age, etc.) High cost Time consuming Biased results Aggregate Selective Little disruption Easy to learn No specific information provided by pupil Non Selective Fast and easy Little information provided Rubbish Analysis Non-obtrusive Highly inaccurate Time consuming INDIRECT MEASURES OF WASTE Visual Estimation Direct Visual Non-obtrusive Non-disruptive Time Consuming Subjective ratings Need for training Digital Photography Food preference Easy to implement Low cost Low accuracy Biased results Dietary Recall 277 Actually, the most accurate method for measuring food intake has been reported to be weighing foods 278 before and after eating although it is reported to be time consuming, costly and disruptive (Williamson et al., 279 2004). This said, it is interesting to recall Smith’s (2014) research in which, in order to confirm observer 280 reliability he weighed 20% of pupil trays after consumption and compared the result with visually estimated 281 plate waste using digital photography, resulting in a 92% agreement. This is consistent with the 282 Environmental Protection Agency - EPA (2014), which suggests that when there are space and time 283 limitations, visual assessment may be more appropriate. 284 Jacko et al. (2007) in their research suggest that an accurate measure of plate waste at schools should be 285 done without direct contact with the children because this could influence their behaviour and bias results. 286 They therefore recommend the use of aggregated methods. Moreover, they compare aggregate vs 287 individual methods to measure plate waste, finding no relevant statistical differences. They conclude that 288 aggregated selective plate waste measurements provide accurate results for groups of children without the 289 complexity of implementing actual weighed food measurements (Jacko; C. C.; Dellava; J.; Ensle; K.; & 290 Hoffman; D. J., 2007). However, individual plate waste data would provide more specific information such 291 as correlations between sex and age (Jacko; C. C.; Dellava; J.; Ensle; K.; & Hoffman; D. J., 2007). 292 Therefore, even when using an aggregate method it might be useful to individually measure a small part of 293 the sample. Furthermore, in order to generate useful comparators when using aggregate methods, total 294 recorded kilos of waste are usually presented per pupil (Buzby and Guthrie, 2002). 295 296 2.4.Food Waste Research Objectives and Indicators in the Literature 297 Before going deep into the particularities of our research scope, school canteens, as a baseline we used 298 general recommendations from researchers on measuring food waste. Nevertheless, food waste studies in 299 the catering industry have been performed mainly in the education and health sectors. 300 Generally speaking, before performing a food waste audit, an entity should clearly define why it wants to 301 quantify food waste. The results may be used for internal decision making, reporting to the institution 302 stakeholders or to develop a Food Waste reduction policy or initiatives (C. Hanson, B. Lipinski, K. Robertson, 303 D. Dias, I Gavilan and J. Fonseca, 2016). The way in which results are presented is closely related to the 304 purpose of the audit, where the most recurrent research objectives observed in our review of the literature 305 are assessing novel policies nutritional and analysing the efficiency of the food system. 306 9 Food waste audit results are typically expressed by researchers through one or a combination of the 307 following indicators: 308 • Plate waste weight in grams per pupil (e.g. Ferreira et al., 2013; Wilkie, 2015), which can be calculated 309 as the mean of individual measures or as a result of dividing total waste obtained in the audit when 310 using an aggregate method by the number of diners. This output is useful when a comparison between 311 different institutions is considered useful. 312 • Plate waste index, calculated as a percentage by weight on served food that is discarded or eaten (e.g. 313 Byker et al., 2014; Rodriguez Tadeo et al., 2014). This more explicit indicator is very often used for its 314 conclusiveness and clarity. Ferreira (2013) highlights the fact that the plate waste index shows the 315 interaction between the diner and the food, regardless of kitchen or system efficiency. We find in the 316 literature researchers that present their results in either of two ways: as percentage wasted (e.g. 317 Marlette et al., 2005) or as percentage consumed out of total amount served (e.g. Cohen et al., 2013). 318 • Energy value of the waste, expressed as percentage of nutrients consumed against nutrients offered 319 (e.g. Bergman et al., 2004). This indicator is used when the purpose of the study is assessing the 320 dietary intake of pupils, without considering sustainability impacts of wasting food. 321 • Total kilos wasted (e.g. Wrap, 2011). This indicator is normally used together with average grams per 322 pupil with the purpose of increasing awareness on waste as big figures (kilos, tons) are more 323 impressive than grams. 324 • Monetary value of waste (e.g. Cohen et al., 2013) is very seldom used by researchers due to the fact 325 that the research objectives are rarely related to cost. In order to determine the cost of plate waste, 326 Buzby et al (2002) suggest multiplying the percentage waste estimate by the total budget allocation for 327 food in the institution, although acknowledging this method does not adjust for differences in costs of 328 food items wasted (e.g. bread vs meat or processed food). 329 • Efficiency of the food service system (e.g. Falasconi et al., 2015), a ratio of the relation between 330 processed food (kg) and unserved food (kg and %). As stated by Ferreira (2013, p. 3), the “Leftovers 331 index” relates all food discarded in the food service process to the quantity of food consumed. 332 333 3. Research Objective and Methods 334 335 3.1. Objectives and Scope of the Study 336 We conclude from the literature that there is relevance in measuring food waste and the need to provide a 337 standardised method that can ease its measurement as well as track its evolution along time. The 338 development of a food waste measurement reduction protocol has been highly recommended by 339 researchers like Lipinski et al. (2013) who go further by suggesting the need to link it to setting reduction 340 targets and supporting collaborative initiatives to reduce food waste. Moreover, Wilkie (2015) states that 341 before any food waste reduction or recycling initiative can be implemented, it is necessary to know the 342 amount of food waste that is generated. With regard to plate waste, Jacko (2007) observes that more and 343 more schools are acting to prevent child obesity, initiating changes in dietary education programmes and 344 lunch menus; consequently, he states that it is vital to have an accurate and cost effective validated method 345 to measure and track plate waste through which changes can be assessed. 346 Provided that food waste seems to be a challenge for schools in their path towards sustainability, and since 347 as stated by Szekely (2005), there is a need to establish clear, user-friendly methods and tools to measure 348 10 progress that companies are making toward sustainability (Székely and Knirsch, 2005, p. 1) the availability 349 of a food waste self-auditing tool becomes valuable. A standard criteria for measuring school catering food 350 waste is novel in the literature, particularly as we propose to include in our assessment tool both pre-351 consumer and post-consumer waste, while numerous studies on school food waste focus on analysing 352 plate waste (e.g. Adams et al., 2005; Byker et al., 2014; Cohen et al., 2013; Marlette et al., 2005; Rodriguez 353 Tadeo et al., 2014). 354 In order to contribute towards filling this gap, we conducted research in the catering industry at school 355 canteens. The central objective of this study is to shed light on how initiatives and practices aiming to reduce 356 food waste at schools can be measured and tracked. In order to attain this research goal, the following 357 specific objectives were set for the research: 358 O1: To analyse how research measures, assesses and reports food waste at schools. 359 O2: Comprehend the level of awareness on food waste and its relevance for school and catering managers. 360 To identify the elements that influence the generation of food waste at schools, understand its nature and 361 the types of food being wasted as well as at which point waste is generated. 362 O3: To develop a self-assessment auditing tool to be used by educational centres and researchers to 363 measure and track food waste at school canteens. 364 Our practical perspective is also novel, a fact that gives our research a very useful and precise managerial 365 implication. Our aim is to develop an easy to implement self-assessment tool to be applied by school 366 catering managers without the need of external assistance. Our auditing tool targets not only plate waste 367 but also any losses before food is served with the purpose of assessing on the sustainability of the food 368 service system. 369 3.2.Research on food waste at schools 370 With the purpose of doing an in-depth analysis of how research measures, assesses, and reports food 371 waste at schools, our first research objective, we gathered over 20 studies by means of a Scopus search 372 using as key search words - food waste and schools. Later, we found a few additional ones from 373 bibliographies and citations. We analysed their objectives, methods, procedures and outputs in order to 374 understand their strengths and weaknesses and then used the knowledge to create the foundations for the 375 development of a standardised auditing tool. 376 Studies performed in order to quantify the amount of food that is wasted daily at school dining facilities (e.g. 377 Byker et al., 2014; Falasconi et al., 2015; Smith and Cunningham-Sabo, 2014; Wrap, 2011) show the effect 378 of pupils’ preferences and behaviour, and the effect of the food service regime on food waste from school 379 meals (Wilkie, 2015). Although research objectives are diverse (see Table 3), the vast majority (80%) of 380 studies focus on analysing plate waste. However, most of these studies are not complete food waste audits 381 and do not account for food waste from kitchen preparation, or waste from serving lines nor food pupils 382 bring from home. Despite being plate waste the most frequently reported measure in school food waste 383 studies, it is not the only source of food waste at schools. Interestingly, Falasconi et al.(2015) undertook 384 research in 6 schools in Italy and found a significant level of inefficiency in school catering services: over 385 15% of the overall processed food was not served to the pupils, according to their measurement. 386 Nevertheless, only a few of the studies found in the literature aim to measure the efficiency or sustainability 387 of the school food system, as most of them are focused on pupils nutritional intake, and therefore limiting 388 the analysis to plate waste. 389 11 Plate waste measures show a considerable variation between the different schools (Wilkie, 2015). Typical 390 results range from 20% to 50% of the food served being wasted, with vegetables and fruit in the higher 391 range (Wilkie, 2015). For instance, Rodriguez Tadeo et al. (2014) did a research in Spanish schools 392 estimating leftovers by visual estimation, being up to 26% of total served food and Byker (2014) obtained a 393 45.3% of waste on total food served. Other studies mentioned by Wilkie (2015) give results that range 394 between 52 g and 227 g per pupil per day. He explains such differences were likely due to the different ages 395 of pupils and methods of food service (Wilkie, 2015). It is interesting to point out that there was significant 396 variability in the amount of food wasted during the week, vegetables ranged from 26.1% to 80%, depending 397 on the day. Although researchers acknowledge some plate waste is unavoidable (Cohen et al., 2013), they 398 agree that in excess is a sign of inefficiency or even irresponsibility (Buzby and Guthrie, 2002).The wide 399 range of waste generation rates shown in these studies also suggest the need for more standardised waste 400 audit methods to measure waste produced at school cafeterias. 401 From our review of the literature (n=20), we present a summary in table 3 of the most relevant features of 402 the studies performed by researchers quantifying food waste in school canteens as well as their weight on 403 the analysed studies. 404 Table 3. Empirical research quantifying food waste in schools (% on total analysed studies) 405 Boundary Research Scope Research Objective Methods Indicators used USA 75% UK10% SPAIN 5% ITALY 5% AUSTRALIA 5% Plate waste 80% Kitchen and PW 10% Kitchen waste 5% Total Waste 5% Dietary Assessment 40% Drivers of Plate waste 30% Method comparison 10% Economic cost of food waste 10% FS efficiency 5% Waste assessment 5% Individual 69% Aggregate 31% Selective 94% Non selective 6% Weigh 69% Visual 31% % waste on served 29% % Consumed on served 17% Nutrients consumed or wasted 21% Grams of waste per pupil 13% Waste economic value 13% Total kilos of waste 4% Food surplus 4% 406 3.3. Methods 407 The development of a standardised self-assessment tool should take into consideration the diverse 408 frameworks in which school canteens operate which involve a set of complex social phenomena. In order 409 to analyse this complexity, we designed research with an explorative/inductive approach through primarily 410 qualitative data as proposed by Pratt (Pratt, 2009). 411 With the purpose of developing a useful and practical assessment tool, we designed exploratory research 412 in two phases. First, we collected data through qualitative research with a range of stakeholders in order to 413 understand the factors that generate food waste at school canteens. Semi-structured, individual interviews 414 with 12 managers and staff of 9 different institutions and collectives that play a role in school meals were 415 conducted (see appendix A for details).In this first phase of the research we obtained insights from 416 managers, both at schools and catering organisations from which a first draft of the tool was designed. In 417 the second phase of our research, once the assessment tool was pre-designed, we tested it in four of the 418 participating schools in the former phase, in order to validate and improve it. At the same time as the tool 419 was being tested, we gathered the opinion of canteen and school staff through 9 further individual interviews 420 as well as the opinion of 8 pupils too. Data collection was performed during November and December, 2014. 421 12 The sample selection of the first part of the study followed a strategy of quotas according to the type of 422 school (semi-public, public and private institutions) and catering organisation. Due to the nature of the 423 research, all schools should satisfy the following criteria: offer in-house cooked meals in a canteen and a 424 minimum of 300 pupils having lunch daily at school. Catering companies had to have a revenue in Spain of 425 at least 10 M € in the last year and a significant market share in the institutional food service channel. To 426 identify our sample, we explored their web sites and existing reports and visited their locations. The final 427 sample was made up of 4 catering companies and 5 schools in Barcelona city. Semi-structured interviews 428 with school principals, canteen managers and food service organisation management were carried out (see 429 Appendix A for interview and organisation characteristics).Due to the complexity of an analysis of this kind 430 of process, we have developed a protocol as a conceptual and practical guide on data collection during 431 interviews. The protocol proposes a semi-structured interview design with open questions and unlimited time 432 in order to capture possible unexpected results and redirect the interview according to the responses of the 433 interviewee. The questions were grouped in three sections; the first one about the management system, 434 followed by specific questions related to each production stage (procurement, kitchen, service and waste 435 disposal) and finishing with questions on their interest in applying reduction measures and best practices. The 436 interviews lasted an average of 60 min and all of them were conducted in places suggested by the interviewees 437 to maintain their comfort and privacy. In addition, the interviews were recorded using an audio recorder. The 438 protocol also suggests the annotation of interviewees’ reactions (e.g. behaviour or non-verbal communication) 439 when responding to questions. The transcript of the interviews was conducted following a process of double 440 review by the authors. In the second phase of the research, more informal interviews with school and 441 catering staff as well as professors and pupils were conducted. 442 The next step was the codification of the interviews through the methodological proposals of Bogdan and 443 Biklen (1997) implementing a qualitative data analysis software (MaxQDA). The first step of interview coding 444 was to identify the blocks or paragraphs where the interviewees spoke about one of the elements suggested 445 by Bogdan and Biklen, such as Setting, Definition, Process, and Method. This first coding allowed us to 446 define the starting point from which we analysed the structure of each interview. The second step of coding 447 consisted in assigning to paragraphs (or a part of them) a list of preconceived codes from the theoretical 448 framework of the research. The initial list of codes contained 7 codes (Players, Places, Food Type, Waste 449 Drivers, Initiatives, Waste Hierarchy, Key Performance Indicators (KPIs)). The third and final step consisted 450 in coding the paragraphs with a more inductive approach (encoding in vivo), recoding some of the interviews 451 as new codes emerged. The final code book contains a total of 63 codes that classify data into 10 codes 452 (the former 7 plus three new ones: Management, Resources and Culture). 453 After the encoding process, we analysed each interview and later we analysed them all in block following 454 the suggestions of Miles and Huberman (1994) and Jurgenson (2005) with the goal of obtaining a specific 455 vision of each case and a final conclusion for all cases. The first step of this part of the analysis was to build 456 a checklist matrix to coherently organise several components for every case. These matrices showed the 457 different sources of data (interviews) in rows and the topics or codes (both the codes from the second and 458 the third step of the coding process) in columns. The matrices allowed us to display the interviews of the 459 codified elements and their reliability and importance according to the number of sources that corroborated 460 them. 461 From each case, we generated a Time-Ordered Matrix that showed the several processes throughout the 462 study period. Based on the matrices, we re-analysed the assessment tool that we had previously developed. 463 13 After the analysis of each case, we carried out a Cross-Case Analysis in order to enhance generalisability 464 and potential self-execution of the outcome. Following a code-oriented strategy, we developed a Case-465 Ordered Effects Matrix (based on Miles and Huberman,(1994), which allows us to see how the effects play 466 out across the seven interviewees. In other words, we could sort the seven cases and show the diverse 467 effects for each case in the same picture. The matrix has the cases in rows and the main features of the 468 school, their strategies and point of view on sustainability, the point of view of the catering company, and 469 some short-run effects. From this matrix, we were able to start analysing the relationship between schools 470 and food waste. 471 Once a first draft of the tool was developed based on the insights obtained from the qualitative phase of the 472 research, we addressed 4 schools in Barcelona in order to test its performance and improve its deficiencies. 473 The test lasted three to five consecutive weekdays at each school with the objective of comprising different 474 menus and therefore avoiding potential bias due to meal preferences. The schools were selected so as to 475 ensure different catering arrangements, medium to large size schools, public and private institutions and a 476 mix of socio-economic statuses. The four selected schools for the trial each had an in-house kitchen in 477 which daily meals were prepared managed by a specialised firm because this is the most common 478 procedure at Spanish schools, as mentioned by C4 (see Appendix A) in our research. We weighed and 479 measured waste from their canteens during 11 school days, in the four schools (Table 4). School staff 480 cooperated in the audits through setting aside the waste collected from the different areas and providing 481 access to the areas where collection stations were placed. The schools in our sample had different cafeteria 482 layouts but their lunch schedules were similar. Meals were composed of a starter (legumes, rice, pasta or 483 vegetables), main dish (meat or fish), white bread and a dessert (fruit or yoghurt) and tap water. Children 484 did not have the option of choosing their menu, except for secondary graders in school C.6 where they 485 chose from two different options for each course. Special regime meals were usually also offered on 486 demand. None of the schools offered a la carte items such as potato chips, as this very rarely happens in 487 Spanish schools. Pupils in the study ate in one common lunchroom in three of the schools, while one of the 488 schools had seven different lunchrooms. This latter school had 4 serving lines, two of the schools had one 489 single serving line, and in one - school children were served by the staff at their tables. With regard to 490 serving lines, food was presented in stainless steel containers (called Gastronorm) in the serving lines and 491 kitchen staff served students on their trays when they passed by. 492 According to Engström (2004), food waste at the canteens was collected and aggregately weighed 493 separately depending on the point where it had been produced (pantry, kitchen, service station or plate 494 waste), distinguishing whether it was avoidable (e.g. out of date ingredients, plate waste) or unavoidable 495 (e.g. bones, peels) waste. Research assistants weighed the aggregated discarded food at each step in the 496 process every day, recording total kilos as well as the approximate % of the different types of food. For this 497 purpose, we used industrial transparent plastic bags (100 litres) so that research assistants could visually 498 estimate the percentage of the different types of food once the bags were full. This was possible because, 499 as mentioned before, the variety of dishes usually offered at school canteens in Spain in one day is limited, 500 typically one entrée plus one main dish and one dessert or at the most two options of each, resulting in no 501 more than three to five different food types per meal. 502 Research assistants arrived at schools three hours before lunchtime, in order to prepare collection bins and 503 track kitchen preparation tasks. Bins were placed in different spots, labelled in order to collect food at each 504 stage. First of all they measured food wasted during meal preparation, making a note of its alleged cause. 505 14 “potentially avoidable” waste was differentiated from “unavoidable” waste such as egg shells, bones, etc. 506 and only potentially avoidable waste was weighed. For this purpose, rubbish bags were placed at different 507 points of the kitchen with specific labels. We therefore used 6 differently labelled bins and placed them at 508 the different collection stations: 1) “Out of date or damaged raw ingredients”; 2) Unavoidable “kitchen 509 scraps”; 3) Potentially avoidable “kitchen scraps”; 4) “Service line leftovers”; 5)Unavoidable “Plate waste”, 510 and 6) Potentially avoidable plate waste. Once the audit was finished, only four of them were weighed (using 511 a Pelouze scale in all but one school where we used a Campesa K3 balance), as we did not measure 512 unavoidable waste, in accordance with Papargyropoulou et al.’s (2014) suggestion. 513 We decided to combine a direct measure of waste method, aggregately weighing waste at the different 514 collection stations with a less accurate method to measure food typology shares. Once total weight was 515 measured, research assistants visually estimated the approximate percentage of total weight per food 516 category. We opted for the aggregate selective method for its easy execution and simplicity, as schools 517 should be able to implement it without external help later on. 518 Table 4 shows the total number of trays included in the trial as well as the number of days the audits lasted 519 in each school. Overall, we measured the aggregated avoidable waste weight of over 10,000 trays, and 520 2,991 children took part in the audit. 521 522 Table 4. Trays and pupils audited 523 524 Participating pupils Trial Duration (# Days) Elementary Pupils’ trays Secondary Pupils’ trays Total Audited Trays School C5 986 5* 2,815 2,113 4,928 School C7 465 2 534 396 930 School C6 1,316 3 1,881 2,067 3,948 School C8 225 1 225 0 225 TOTAL 2,991 11 5,455 4,576 10,031 *(secondary pupils were present 4 four days only) 525 526 During the audit days, we interviewed 9 canteen and school staff in order to get insights from those who 527 work closely with the day to day operations of the canteen. We also performed 9 quick interviews with 528 children eating in the canteen. The interviews in this case lasted 20 minutes on average with staff and 10 529 minutes with pupils and we encoded the transcripts following the same method and codes as in the former 530 phase of the study. 531 The number of pupils actually eating lunch in the canteen each day was registered in order to be able to 532 estimate the average weight per pupil and day, as this was the measure found by Wrap (2011) to be the 533 most meaningful way to compare data from different schools. This figure was compared with the planned 534 number of diners, a figure that we asked the cooks each audit day in order to assess potential food surpluses 535 as suggested by Papargyropoulou et al.(2014). 536 It is important to recall that the primary objective of the auditing tool is to analyse and track food waste 537 produced at schools, not the amount of food going in, nor the nutritional intake of pupils. Therefore, the 538 output is given in grams of waste per pupil and not as % of waste on food prepared or served nor percentage 539 of energy or nutrients consumed vs offered. Nevertheless, the tool can be easily adapted for these purposes. 540 541 15 4. Results and Discussion 542 4.1. Perspectives on food waste by school caterers and canteen managers 543 We found a very low real awareness of managers on the amount of waste produced in the canteens. Only 544 one of the schools in the sample had ever performed a waste audit at the canteen and only one of the 545 participating catering companies does waste audits in the kitchens they operate in on a regular basis. This 546 said we nevertheless found a high interest on the topic, especially among public funded school managers 547 and personnel: we appreciated that many school managers would be willing to implement initiatives to 548 measure and minimise the amount of food wasted at their canteens, especially after observing our pilot-test 549 results. It was acknowledged by the interviewed managers that food waste is a data-poor area and therefore 550 when suggested, a waste inventory was reflected as the starting point for the application of reductive 551 initiatives. They largely agreed on the fact that it would be useful to increase awareness on waste through 552 the measurement and tracking potential of reduction initiative results. 553 Consistent with the literature (Wrap, 2011), avoidable food waste accounted for the greatest amount of 554 waste generated at schools in our pilot test. Plate waste accounts for the biggest source of food waste, 555 followed by food from serving lines. Average weight of food wasted per elementary school pupil in Barcelona 556 ranged between 40 and 100 grams per meal and pupil. Secondary pupils’ average waste was higher in two 557 of the three secondary schools analysed, exceeding 80 grams daily waste per pupil in two of the four studied 558 schools. 559 In our trial of the auditing tool, school’s institutional and pedagogical principles showed a very direct 560 influence on the amount of food wasted at the canteen. Some schools consider the canteen as part of their 561 learning project and therefore try to educate children in finishing their food through different activities, 562 training, and workshops. These schools resulted in lower levels of waste, and especially of plate waste. 563 Conversely, whenever top management of the school did not consider food waste a priority, plate waste 564 ratios were higher, at the same time as the level of awareness on the amount of food wasted was very low. 565 Just one school mentioned they regularly performed initiatives with the purpose of reducing food waste. In 566 fact, in this school we found the lowest rate of plate waste in our pilot-test. We concluded this was due to 567 the fact that its management had a strong focus on reducing food waste and this strong focus was translated 568 into multiple ongoing initiatives.C6.1:“We settle specific objectives every year. At present we are focusing on three 569 food types: lentils, fish and oranges. Last year we achieved an important reduction on discarded bread. We are also 570 currently focused on reducing dairy packaging, as its disposal costs are high”. 571 Moreover, schools with a stronger management focus on sustainability, or with wider pedagogical objectives 572 showed high interest in the results of our pilot audit at the same time as they declared their purpose of 573 repeating the audit in the near future. 574 On the other hand, we also found food service providers with very different perspectives and visions on food 575 waste. One of the food service managers interviewed, who worked for a catering company with a strong 576 sustainability culture mentioned that school managers’ scepticism and lack of awareness was a barrier for 577 improving results:C1.2 “Implementing sustainable initiatives is difficult sometimes, as schools are often not 578 very sustainability conscious; We have had customer complaints when trying to reduce food waste arguing 579 that our only purpose was to reduce our costs!”. 580 She nevertheless recalled that when they had formerly performed waste audit assessments in schools, the 581 results had been touching for both organisations and stated that it had been easier to introduce reduction 582 initiatives in those institutions since then. We concluded from this that increasing visibility and awareness 583 on food waste is crucial: C1.2“We recently measured aggregated plate waste in one of our customers, one 584 16 big sized school in Madrid resulting on a daily average of 350 kilos of food discarded. Then they launched 585 an awareness campaign by putting together 350 kilos of packaged food ingredients at the entrance of the 586 lunchroom with the purpose of increasing awareness on food waste among children” 587 Moreover, we observed very different attitudes toward plate waste among canteen and school staff. Such 588 attitudes range from strict control on pupils so that they completely finish their meal, to passiveness, 589 acceptance or even denial of the real situation regarding plate waste. These diverse attitudes are also 590 related to dissimilar school management ideologies regarding school meals: from those considering the 591 canteen as a fringe service offered to the parents (with no educational responsibility by the school), to those 592 who consider it as part of the school´s pedagogical mission. This is very closely related to the means and 593 resources dedicated to minimise plate waste, such as the number of caretakers and their role regarding 594 leftover control and pupils eating habits as well as food waste reduction awareness campaigns. 595 596 We concluded from these observations that the role performed by school top management is the most 597 relevant factor influencing sustainability issues such as the level of canteen food waste. Those institutions 598 with a strong focus on sustainability or which were at an advanced stage on “greening” their organisations 599 usually allocated more resources to reducing food waste and were thus more likely to be looking for 600 performance indicators and initiatives to reduce waste. This was confirmed in our pilot-test, as the one 601 school with a clear focus on sustainability recorded the lowest plate waste rate. The higher management 602 focus on sustainability was translated into diverse procedures impacting the different waste driver areas, 603 resulting generally in lower waste rates. Moreover, green conscious managers tend to be concerned not 604 only with food waste but also with related packaging waste. An informative campaign addressing public 605 funded schools with the purpose of increasing awareness on food waste could therefore be highly efficient. 606 Actually, as mentioned by Papargyropoulou et al. (2014) we verified that food waste arises at all the different 607 stages as a result of very diverse causes and thus the ways to tackle them must be different too. We 608 concluded from our research that food waste drivers can be categorised in three groups. First, those related 609 to management practices such as the meal planning process or procurement practices. Secondly, 610 infrastructures and equipment also impact food waste levels, especially at the storage and serving stages. 611 Finally, human resources issues, such as staff awareness (or lack of awareness) on food waste is also 612 reflected at the different levels of food waste in canteen operations. In the next paragraphs we shall develop 613 these drivers, relating them to adequate indicators that will allow managers and researchers to measure 614 and track performance in their related areas. 615 Regarding management practices, cooks and caterers mention communication between school and kitchen 616 as key in order to accurately plan the number of menus to elaborate. As mentioned by C1.2, this is absolutely 617 relevant for special regime diets such as allergenic: C1.2 “Special menus such as diet or allergenic produce 618 higher amounts of waste per pupil than regular ones as they are more difficult to plan”. From this insight we 619 can infer the relevance of tracking deviations between planned and real numbers of diners. 620 Also related to management practices we found menu planning closely related to food waste. In fact, many 621 of the pupils interviewed complained about the quality of the food offered. Pupils’ acceptance of food can 622 be increased by menu planning policies. As suggested by C12: “The different acceptance rates of dishes 623 by pupils makes a difference. We try to balance our menus: if the first course is “difficult” (like for example 624 chickpeas), the main course should be “easier” (for instance not offering fish)”. Pupils’ acceptance of meals 625 can also be enhanced by giving them the option of choosing between more than one alternative for each 626 course. Only one of the schools studied offered the pupils different dish alternatives to choose from. 627 17 On the other hand, procurement policies were admitted as closely related to waste. Suppliers’ delivery 628 frequency and product formats are managed to prevent pantry losses. Public policies were highlighted as a 629 key potential tool to entice good purchasing practices at schools, although this was not clearly related to the 630 generation of waste and should be tracked by selective measures of plate waste. C2.1: “Public procurement 631 policies are aimed to guarantee that children have a diverse and complete diet, but effective food intake by 632 children varies a lot between schools, closely related to school management priorities and consequent child 633 education on food habits and supervision during meals”. 634 Research also shows that kitchen food waste is strongly influenced by school infrastructure and equipment. 635 Caterers need to adapt their processes to school facilities and often complain that some of them are very 636 old. They recognise this fact as a limitation: C2.1:“It is really hard sometimes”. Furthermore, the availability 637 of recycling facilities strongly determines the destination of waste. C8.”Since we own a vegetable garden , 638 we compost most of the kitchen scraps and peels we generate”. Recording regularly the destination of food 639 waste as well as its disposal costs might increase awareness on potential improvements. Waste bins at 640 schools in our sample were normally emptied into dumpsters. Although three schools in our sample had a 641 vegetable garden, only one of them composted food waste from the canteen. 642 Better storage facilities was mentioned by cooks as a way in which they could reduce the amount of raw 643 materials that had to be discarded, at the same time as it could also be a way of permitting excess cooked 644 food to be stored for later consumption. We also found a relevant source of waste related to the number of 645 serving lines in which children were served or where they could help themselves to food. Whenever there 646 is one unique serving station, waste at this stage was significantly lower than when there were several. 647 Schools with more than one service line tend to generate more food waste per pupil at this stage. This was 648 due to the fact that all types of food needed to be displayed until the end of the service time at all service 649 stations, inevitably causing a certain amount of waste at each station. One of the schools where we pilot-650 tested the auditing tool had four serving lines. Waste at this stage in this school varied significantly among 651 the dates studied and we weighed over 70 kilos of cooked food not served that was discarded in one day. 652 Bread has a relevant role here. In our case study plate waste accounted for the greatest part of food waste 653 in three of the schools studied and serving waste in the fourth one. Moreover, due to the fact that bread is 654 low priced, no attention was paid in general to the amount discarded. In most serving lines, bread was 655 placed at the beginning, together with the trays and cutlery, and diners used to take it before knowing 656 whether they were going to like the menu. Bread was in our test one of the food categories with highest 657 waste. 658 Finally, the role of canteen supervisors was emphasised as crucial, the lack of control on pupils leftovers 659 being a relevant driver of plate waste. It was acknowledged that plate waste is closely related to effective 660 supervision. Actually, schools with the lowest rates of food waste in our pilot were those where there was 661 stricter control by canteen supervisors on top of a wider educational perspective. Measuring and tracking 662 plate waste can be used by managers to encourage caretaker supervision. Managers therefore will find it 663 useful to unveil the amount of plate waste as this will allow them to set reduction objectives and measure 664 their effect or even compare results with other schools. 665 Tracking and disseminating these key performance indicators will facilitate school managers when choosing 666 the most adequate correction measures and evaluating results. Necessary correction measures are different 667 depending on the cause and the place where waste is generated. Table 5 summarises the most relevant 668 school canteen food waste drivers and the indicators or variables that might be useful for running a diagnosis 669 and describing the main improvement areas and help in the management of each of them. 670 18 671 Table 5. School food waste drivers and key performance indicators (KPI) 672 Related Area Food Waste Driver Institution Culture and Values Top Management (low) Focus on Sustainability Pedagogical Vision Management Practices Communication between Kitchen and School staff Meal planning process Menu planning (and acceptance of food by pupils) Procurement practices Infrastructure Kitchen equipment and facilities Recycling & Reuse facilities Canteen Layout Human Resources Supervision by caretakers 673 Age is highlighted as a relevant factor too. Canteen staff and caretakers agree on the fact that children of 674 different ages usually have different eating patterns. There was a consensus on the fact that younger 675 children produce less plate waste, as stated by C.5.4a “The younger they are, the more they eat. Three to 676 five year olds leave no plate waste at all!” This insight shades light on the relevance of measuring waste 677 from different collectives separately. Interestingly, even though the amount of waste generated per pupil 678 varied a lot among the different schools, food wasted by elementary pupils was much lower than by 679 secondary graders in our research. This result is consistent with the outcome of the first stage of our 680 research although we found opposite results in several of the studies (e.g. Guthrie and Buzby, 2002; Niaki 681 et al., 2016). 682 It is interesting to note that catering and school staff did not consider the proposed auditing method 683 disruptive. On the contrary, cafeteria staff, teachers and caretakers who collaborated in the trial were proud 684 to share their experience with other colleagues. They were often impressed by the results and willing to 685 collaborate when ideas for FW reduction were brought up. Research findings strongly support the relevance 686 of sharing results with canteen staff, as suggested by the World Resources Institute (C. Hanson, B. Lipinski, 687 K. Robertson, D. Dias, I Gavilan and J. Fonseca, 2016). 688 689 4.2. Self-assessment food waste auditing tool 690 Based on our research, we can group the information to be measured and tracked when auditing food waste 691 at school canteens, into four categories: accuracy of the planning system, physical measure of waste, waste 692 destination and economic cost of food waste. In the following paragraphs we develop the four categories 693 and describe related key performance indicators that should be included in a waste audit. 694 4.2.1.Accuracy of the planning system. Conformity between real versus planned number of diners should 695 be measured, with the objective of analysing and tracking actual deviations between the information used 696 by cooks when preparing food and the final amount of food needed at lunchtime. Differences between these 697 two figures are often the cause of generation of food surplus (excess food cooked). In order to assess the 698 accuracy of the planning system, we suggest using the following indicator: deviation rate between planned 699 and served meals. 700 19 A daily estimation of the difference between planned meals and the real final number of diners should be 701 tracked. For this purpose, a deviation rate should be recorded daily, noting both the number of planned 702 diners before the cooking process begins and the actual number of effective pupils that eat at the canteen 703 each auditing day. Deviations should be recorded in % of actual vs planned diners per sitting (whenever 704 there is more than one). Special menus such as allergenic or diet lunches should be recorded separately 705 too. If there is a known cause for the deviation it should also be briefly explained in the record. Needless to 706 say, elementary versus secondary grades should be recorded separately. 707 4.2.2.Physical measure of waste. Different food categories (e.g. fruit, bread, etc.) should be recorded 708 separately in order to be able to assess the efficiency of the food service system as well as dietary and 709 nutritional intake and food acceptance and preferences. This measure will shed light on the potential 710 improvement that can be achieved by performing reduction initiatives and will be helpful for their design. 711 Due to the nature of the physical measure of waste, we suggest two indicators, weight of food waste and 712 number of zero waste trays, discussed below. 713 (a)Aggregate and Selective weight of food waste at each different stage of the process. This should 714 be measured at each collection station, in order to differentiate the four typologies of waste:pantry loss, 715 cooking loss, prepared food surplus and plate waste, as explained in section 2.2. At each stage, potentially 716 avoidable food waste should be measured separately from unavoidable waste, which does not need to be 717 included in this record. Collection stations must differentiate the place and stage in the process where waste 718 has been generated and categorised food should be recorded at each collection station. We suggest 719 estimating the share (percentage on total food waste) of each food type by visual estimation. For this 720 purpose we recommend the use of transparent rubbish bags or bins for the aggregate measurement, 721 recording the approximate % of each food category after weighing. To do this, we suggest using the 722 classification used by Betz et al. (2015): meat/fish, starch, vegetables, fruit, desserts (e.g. yoghurt), and 723 others, adding bread and legumes as separate additional categories. As mentioned before, unavoidable 724 waste such as peels, bones, etc. must be separated at collection stations and withdrawn before weighing. 725 Recording total weight of unavoidable waste is optional. 726 We shall therefore measure four different waste indicators in this section, one per each stage of the 727 process: 728 A. Pantry loss: food waste generated in raw ingredient storage (mostly out of date produce). We 729 shall record the total kilos wasted at this stage, the approximate % of total weight per food type and 730 the place where it occurred (e.g. pantry, fridge, etc.) as well as its alleged cause (e.g. out of date, 731 spoilt, etc.). 732 B. Cooking loss: waste produced during the cooking process. Unavoidable waste should be 733 discarded separately at this stage because only potentially avoidable waste needs to be weighed. 734 Total kilos of avoidable waste should be recorded, as well as the approximate % per food type, the 735 place of generation and the reason that probably caused it (e.g. burnt, aesthetics, etc.). 736 C. Prepared food surplus: food cooked but not served. This comprises waste produced at serving 737 lines or other means of distribution or display. Here, total weight of cooked food not served to the 738 pupils should be recorded as well as the approximate percentage per type of food, noting the most 739 probable reason that caused it as well as its most likely end: reuse (e.g. staff meals, soups, 740 donations,…), recycling (e.g. compost), or disposal. 741 D. Plate Waste: Food Served but not eaten. We recommend measuring plate waste using the 742 aggregated and selective method, once having withdrawn inedible food or parts of food. Again, total 743 20 kilos of waste should be recorded before noting the approximate percentage per food type which will 744 be measured by visual estimation. 745 We suggest weighing discarded food without separating the different types of food at each collection station, 746 as categorisation can be visually estimated after collection by the use of transparent rubbish bags. This 747 suggested method will ease audit implementation despite possibly being less accurate. This is consistent 748 with the literature, as Smith (2014), in a study measuring individual plate waste, concluded that visual 749 estimation was close enough to selective weighing when measuring plate waste. Due to the nature of the 750 audit we prioritise easy execution over accuracy. 751 Nevertheless, plate waste usually being the main source of waste at school cafeterias, it can be helpful to 752 deepen the analysis in a small sample of pupils, in order to get insights on the reasons that caused leftovers. 753 This sample should be taken at random and it is recommended to take digital photos of these pupils’ trays, 754 both before they start dining and when they return their trays. The amount of plate waste found in this study 755 is consistent with plate waste reported in previous research in schools although high differences were found 756 among them. Moreover, most food waste types in our pilot study were legumes, vegetables and bread. This 757 is consistent with the literature, as most studies highlight the high waste of vegetables. 758 Although the aggregated method is recommended for its convenience, results should also be given in grams 759 per pupil, calculating the ratio between total waste amount and the number of real diners, using the figure 760 of real diners previously recorded. We must consider that using this ratio will only be comparable among 761 schools with the same catering system. Using this method, only plate waste ratios will be comparable among 762 schools with different catering systems. Whenever it is possible, a measure of efficiency would also be 763 recommended, recording the percentage of wasted food related to prepared food. This ratio is particularly 764 relevant for transported meals catering systems. 765 766 (b) Number of zero waste trays, as a percentage of total trays. Tracking how many pupils empty their 767 trays completely will shed light on meal acceptance and caretakers’ control. Moreover, the study suggests 768 the dissemination of this information may encourage other pupils to reduce plate waste. C.6.1: “Since we 769 started the zero tray project (a contest among classes in which the class with a higher percentage of fully 770 empty returned trays were rewarded), plate waste has been reduced significantly”. 771 4.2.3.Waste destination or use. Improvement opportunities can also arise by noting and tracking the 772 destination of waste from the canteen. Good sustainability initiatives could include setting objectives of 773 reducing waste sent to landfills and reducing food waste footprint by reducing waste that is discarded at the 774 lower levels of the waste hierarchy pyramid. The indicator proposed to manage waste destinations is simple. 775 We recommend recording the way food waste is discarded (e.g. rubbish bin, compost) or reused. The waste 776 destination indicator implies noting the approximate % of waste which will probably end in landfills, compost 777 or that will be reused, recording its intended purpose in this case. Whenever more than one disposal method 778 is used, the approximate % on total waste weight of each one should be recorded. 779 4.2.4.Economic cost of food waste. An economic estimation of food waste is recommended as it will 780 increase the relevance that school and catering managers give to tracking and measuring waste as a means 781 of reducing food waste could be seen as a potential profit increase. As mentioned by one of the caterers in 782 our sample C1.1: “Canteens are a source of business for schools, they make profit out of them”. School 783 managers with a low focus on sustainability, and therefore not motivated to reducing food waste for 784 sustainability related reasons, may find an attractive incentive in this indicator. The approximate cost of 785 21 waste can be calculated in different ways. We suggest using an average cost per meal estimated on a year 786 basis (including procurement and service) and multiply it by the equivalent of meals thrown away. This can 787 be calculated by dividing the total kilos of waste by the average weight of meal (g) and multiplying the result 788 by the average cost per meal. This should be done with the support of the financial manager. Although this 789 method may not be accurate as it does not distinguish the diverse cost of different food ingredients, we 790 prioritise ease of execution over accuracy due to the purpose of the measurement. 791 By tracking appropriate KPIs related to the above mentioned four areas and their probable causes, school 792 caterers and managers will be able to diagnose and describe main improvement areas. Materials needed 793 in order to perform the audit include a scale, six labelled waste bins or waste containers and transparent 794 rubbish bags. 795 Table 6 summarises the four main data categories, relating them to the goal of the analysis and their related 796 KPI. You will find the auditing tool in Appendix B. 797 798 799 Table 6: Summary of selected KPIs and their purpose. 800 Data Category Purpose Food Waste Indicators short list Accuracy of the planning system Better adjustment of quantities cooked 1. Planned vs real number of meals Physical measure of waste Assess system efficiency & dietary intake 2. Selective aggregate food waste by type of food 3. Zero waste trays Waste destination Reduce environmental impact 4. Food waste destination Economic cost of food waste Increase awareness of Food Waste Relevance to management 5. Total Euros/Dollars/Pounds in cost of food waste 801 Kitchen and service staff highlight that there are some dishes which typically generate low or no plate waste, 802 such as rice or pizza, while others such as fish or vegetables generate high plate waste rates. Despite menu 803 planning often taking this into consideration, we found a wide range of plate waste ratios on different dates, 804 a fact that we attributed to the different acceptance of the menus. Plate waste one day in a specific school 805 could double or even triple a previous day’s ratio. For this reason auditing a full school week is urged in 806 order to include diverse meals and avoid bias due to different meal acceptance from pupils. Strong 807 differences were also found among the sample schools in our pilot-test. 808 Once the audit is finished, it is recommended to share the results with professors, supervisors and pupils 809 as this would contribute to increase awareness on the issue. Lack of visibility and therefore lack of 810 awareness is one of the key reasons for the low level of measures taken to reduce food waste in the food 811 service channel (Derqui et al., 2016). The first measurement will be used as a baseline and the reference 812 for improvement goals. Successive measurements will shed light on the efficiency of initiatives as well as 813 on the room for improvement. We suggest that the audit project be led by a “project leader”, a person in 814 charge who will be responsible for coordinating the different players needed for the success of each 815 improvement initiative. 816 817 5. Conclusion 818 As suggested by Gerbens-Leenes et al. (2003), it is important to bridge the existing gap between theoretical 819 scientific knowledge and practical company knowledge in measuring sustainability. Nevertheless, as they 820 22 state, this is in general difficult, as research as a rule emphasises accuracy and completeness while 821 business needs easy to handle, practical and cheap tools to assess their sustainability performance 822 (Gerbens-Leenes et al., 2003).Through our research, we designed a self-assessment tool that can be easily 823 used by schools and caterers to measure and track food waste at school canteens yet comprehensive and 824 accurate. In addition, through the implementation of the tool, academics will have further relevant 825 quantitative and comparable data as well as visibility to food waste, a field of information which is not widely 826 available. Moreover, managers and researchers can adapt and use the tool in different countries and 827 environments in order to obtain metrics and insights on food waste and benefit from benchmarking and 828 shared experiences under homogenous criteria and standardised concepts. 829 Our paper provides new contributions to the literature on food waste. Firstly, a standardised and easy to 830 implement self-assessment tool is developed to be implemented at school canteens. Secondly, it sheds 831 light on the potential good acceptance that sustainable initiatives may get from school managers and staff. 832 Finally, it relates food waste drivers to key performance indicators that would help managing potential 833 initiatives to address them. On the one hand, our main contribution for researchers is the availability of a 834 standardised tool that will permit the comparison of food waste assessments in schools among different 835 cities and environments. On the other hand, we provide school and food service managers with an easy to 836 implement tool that will help them along their path towards more sustainable organisations. 837 838 6. Acknowledgements 839 This research was partly funded by the Spanish Ministry of Food and Agriculture. The authors want to thank 840 the Ministry for their initiative “More Food, Less Waste”, under which framework this research was done. 841 The authors thank the principals, teachers, pupils and cafeteria staff of the participating schools. The 842 contributions of Antonio Agustin are highly appreciated. 843 844 845 846 847 23 APPENDIX A. Sample characteristics 848 849 INSTITUTION Type of organisation Number of employees/pupils Profile & Number of people interviewed C1 SODEXO Food service 18,000 Million € Global Revenue 420,000 employees Operates in 80 countries Headquarters in FR C.1.1 Marketing Manager C.1.2 Opex Manager C.1.3 Social Responsibility Manager C2 CATSCHOOLS Food service Headquarters in Spain, operates regionally (Barcelona only) C.2.1 Sales Managers C.2.2 Purchasing Manager C3 EUREST (Compass Group) Food service Headquarters in the UK. 17,000 million pounds in 50 countries (group) C.3 Regional Sales Manager C4 ARAMARK Food service 14,329 billion USD revenue 270,000 employees in 21 countries. Headquarters in the US C.4 Regional Sales Manager C5 SAGRAT COR SCHOOL Elementary & Secondary School 1,500 pupils eat daily 2 dining rooms and two service lines C.5.1 Canteen manager C.5.2 Cook C.5.3 a & b: 2 kitchen assistants C.5.4 a, b & c: 3 caretakers C.5.5 a to e:5 pupils C6 ESCOLA PIA SCHOOL Private Elementary & Secondary School 1,500 pupils eat daily Seven dining rooms and 4 service lines Compost facilities C.6.1 Canteen manager C.6.2 a&2b supervisors C.6.3 a to d: 4 pupils C7 ISABEL DE VILLENA SCHOOL Private Elementary & Secondary School 670 daily diners C.7.1 Canteen coordinator C.7.2 Cook C8 ESCUELA JUNGFRAU SCHOOL Public Elementary School 250 daily diners Pupils are served at their table C.8 Canteen coordinator C9 COSTA LLOBERA SCHOOL Public Elementary & Secondary School C.9 Canteen coordinator 850 851 24 Appendix B SCHOOL CANTEEN FOOD WASTE AUTO – ASSESSMENT TOOL 852 853 A. Record the number of planned meals and the real number of diners Planned number of diners Actual Diners % Deviation Deviation causes 1st shift E.g. excursions, sick kids,… 2nd shift Allergenic menus Diet menus B. Selective weight by stage of the process FOOD TYPE WEIGH % on total (% Approx.) PLACE WHERE IT OCCURRED E.G.: fruit, bread… e.g. pantry, fridge… FOOD TYPE WEIGH % on total (% Approx.) PLACE WHERE IT WAS PRODUCED E.g.while cooking, already cooked …. 3. TOTAL KILOS (Approx.): ………..……………………………………………………….....…...………………….……………………………………………………………….…….………………………………… Kg Record cooked food that is not served FOOD TYPE Quantity (kilos) Cause E.g. Roasted chicken 4. TOTAL KILOS (Approx.): ………..……………………….………………………………………………………………………………………………………….…………………………………… Kg FOOD TYPE WEIGH % on total (% Approx.) KG E.g. Vegetables, legumes, etc. % waste on food served (C.2 / C.1) KG Approximate % on total weigh TOTAL AVOIDABLE KILOS (1b+2b+3b+4b) C.1 Average per pupil 1. TOTAL KILOS (Approx.): …………………………………………………………………………………..……………………………………………………….……..……………………………………………… Kg 3.b TOTAL POTENTI ALLY AVOI DABLE KI LOS (Estimate) ………….……………………………………………………………………………………………….……..……………………………………………… Kg Cost of food waste (€) (Equivalent meals thrown away * average cost per meal) 2.a UNAVOIDABLE WASTE: BIN#2 E.G. potato peels, egg shells, etc.: ………………………………………………………………………………….……………… Kg 2.b POTENTI ALLY AVOI DABLE W ASTE: Cooked but not served, burnt, damaged, etc. (indicate type) BIN#3 CAUSE E.g.burnt food; less dinners than expected,.. 4.b POTENTI ALLY AVOI DABLE W ASTE BIN#6 ………….Kg Grams / STUDENT …………… % Approx. 4.a UNAVOIDABLE WASTE BIN#5 e.g. Banana peels, bones, ….: Grams / STUDENT 2. TOTAL KILOS (Approx.): ………..………………………………………………………………………………………………………………………………………………………………..…………………..…… Kg Compost COOKING LOSS (Kitchen Waste): PANTRY LOSS (Out of date and damaged food): BIN #1 PREPARED FOOD SURPLUS (Display): BIN#4 PLATE WASTE: Equivalent of meals thrown away (Total food waste kilos / weight of meal) How it was discarded TOTAL KILOS WASTED (1+2+3+4) C.2 Average weight of meal served per tray (g) Average cost/meal (including preparation cost) (€) C.- Waste Economic Cost D. Waste Destination Garbage Bin Reuse (Mention for what purpose) Most probable end (disposal or use) E.g. Staff meals, soup, donations, etc. CAUSE 1.b TOTAL POTENTI ALLY AVOI DABLE KI LOS (Estimate) …………………………………………………………………………………………………….……..……………………………………………… Kg 25 References 854 Adams, M.A., Pelletier, R.L., Zive, M.M., Sallis, J.F., 2005. Salad bars and fruit and vegetable consumption in 855 elementary schools: a plate waste study. J. Am. Diet. Assoc. 105, 1789–92. doi:10.1016/j.jada.2005.08.013 856 Bergman, E., Buergel, N., Englund, T., Femrite, A., 2004. Relationships of Meal and Recess Schedules to Plate 857 Waste in Elementary Schools. The Journal of child Nutrition and Management. 858 Betz, A., Buchli, J., Göbel, C., Müller, C., 2015. Food waste in the Swiss food service industry - Magnitude and 859 potential for reduction. Waste Manag. 35, 218–26. doi:10.1016/j.wasman.2014.09.015 860 Bogdan, R., Biklen, S., 1997. Qualitative research for education. 861 Bradley, L., 2011. Food Service/Cafeteria Waste Reduction [WWW Document]. Northeast Recycl. Counc. URL 862 https://nerc.org/documents/schools/FoodServiceWasteReductionInSchools.pdf (accessed 6.25.16). 863 Buzby, J.C., Guthrie, J.F., 2002. Plate Waste in School Nutrition Programs Final Report to Congress. 864 Byker, C.J., Farris, A.R., Marcenelle, M., Davis, G.C., Serrano, E.L., 2014. Food Waste in a School Nutrition 865 Program After Implementation of New Lunch Program Guidelines. J. Nutr. Educ. Behav. 866 doi:10.1016/j.jneb.2014.03.009 867 C. Hanson, B. Lipinski, K. Robertson, D. Dias, I Gavilan, J. Fonseca, 2016. Food Loss and Waste Accounting and 868 Reporting Standard. 869 Clarke, C., Schweitzer, Z., Roto, A., 2015. Reducing Food Waste : Recommendations to the 2015 Dietary Guidelines 870 Advisory Committee 1–9. 871 Cohen, J.F.W., Richardson, S., Austin, S.B., Economos, C.D., Rimm, E.B., 2013. School lunch waste among middle 872 school students: Nutrients consumed and costs. Am. J. Prev. Med. doi:10.1016/j.amepre.2012.09.060 873 Comstock, E., 1979. Plate Waste in School Feeding Progams. USDA. 874 Creedon, M., Cunningham, D., & Hogan, J., 2010. Less Food Waste More Profit. 875 Dehghan, M., Akhtar-Danesh, N., Merchant, A.T., Nicklas, T., T., B., K.W., C., G., B., Parsons, T., Power, C., Logan, 876 S., Summerbell, C., Whitaker, R., Wright, J., Pepe, M., Seidel, K., Dietz, W., Livingstone, M., James, P., 877 Kelishadi, R., Pour, M., Sarraf-Zadegan, N., Sadry, G., Ansari, R., Alikhassy, H., Bashardoust, N., AlNuaim, A., 878 Bamgboye, E., AlHerbish, A., McCarthy, H., Ellis, S., Cole, T., Daniels, S., Arnett, D., Eckel, R., Gidding, S., 879 Hayman, L., Kumanyika, S., Robinson, T., Scott, B., Jeor, S.S., Williams, C., Mossberg, H., Must, A., Jacques, 880 P., Dallal, G., Bajema, C., Dietz, W., Flegal, K., Carroll, M., Ogden, C., Johnson, C., Kuczmarski, R., Flegal, K., 881 Williams, D., Going, S., Lohman, T., Harsha, D., Srinivasan, S., Webber, L., Berenson, G., Flegal, K., Wei, R., 882 Ogden, C., Himes, J., Dietz, W., Lazarus, R., Baur, L., Webb, K., Blyth, F., Danford, L., Schoeller, D., Kushner, 883 R., Deurenberg, P., der, K. van, Paling, A., Withagen, P., Deurenberg, P., Kusters, C., Smit, H., Deurenberg, 884 P., Pieters, J., Hautvast, J., Phillips, S., Bandini, L., Compton, D., Naumova, E., Must, A., Stevens, J., Hill, J., 885 Peters, J., Goodrick, G., Poston, W., Foreyt, J., Eckel, R., Krauss, R., Grundy, S., Link, K., Moell, C., Garwicz, 886 S., Cavallin-Stahl, E., Bjork, J., Thilen, U., Ahren, B., Erfurth, E., Willett, W., Nicklas, T., Prentice, A., Jebb, S., 887 Bellisle, F., Rolland-Cachera, M., Deheeger, M., Guilloud-Bataille, M., Griffiths, M., Payne, P., Maffeis, C., 888 Zaffanello, M., Pinelli, L., Schutz, Y., Troiano, R., Briefel, R., Carroll, M., Bialostosky, K., Wright, J., Kennedy-889 Stephenson, J., Wang, C., McDowell, M., Johnson, C., Gregory, J., Lowe, S., Maffeis, C., Pinelli, L., Schutz, Y., 890 Tucker, L., Seljaas, G., Hager, R., Heaney, R., Davies, K., Barger-Lux, M., Pereira, M., Jacobs, D., Horn, L. 891 Van, Slattery, M., Kartashov, A., Ludwig, D., Carruth, B., Skinner, J., Putnam, J., Allshouse, J., Ludwig, D., 892 Peterson, K., Gortmaker, S., Gittelsohn, J., Wolever, T., Harris, S., Harris-Giraldo, R., Hanley, A., Zinman, B., 893 Heitmann, B., Kaprio, J., Harris, J., Rissanen, A., Korkeila, M., Koskenvuo, M., Swinburn, B., Egger, G., 894 Tremblay, M., Willms, J., Gordon-Larsen, P., Griffiths, P., Bentley, M., Ward, D., Kelsey, K., Shields, K., 895 Ammerman, A., Carriere, G., Muller, M., Mast, M., Asbeck, I., Langnase, K., Grund, A., Campbell, K., Crawford, 896 D., Jackson, M., Cashel, K., Worsley, A., Gibbons, K., Birch, L., Gortmaker, S., Peterson, K., Wiecha, J., Sobol, 897 A., Dixit, S., Fox, M., Laird, N., Dietz, W., Gortmaker, S., Stone, E., McKenzie, T., Welk, G., Booth, M., Dwyer, 898 T., Coonan, W., Leitch, D., Hetzel, B., Baghurst, R., Guo, X., Popkin, B., Mroz, T., Zhai, F., Jacobson, M., 899 Brownell, K., Young, L., Swinburn, B., Caterson, I., Gill, T., Chomitz, V., Collins, J., Kim, J., Kramer, E., 900 McGowan, R., Freedman, D., Srinivasan, S., Valdez, R., Williamson, D., Berenson, G., Zametkin, A., Zoon, C., 901 Klein, H., Munson, S., Kotani, K., Nishida, M., Yamashita, S., Funahashi, T., Fujioka, S., Tokunaga, K., 902 Ishikawa, K., Tarui, S., Matsuzawa, Y., Lobstein, T., James, W., Cole, T., Moreno, L., Sarria, A., Popkin, B., 903 Rolland-Cachera, M., Deheeger, M., Thibault, H., GE, K., T, T., C, T., T, K., 2005. Childhood obesity, 904 prevalence and prevention. Nutr. J. 4, 24. doi:10.1186/1475-2891-4-24 905 26 Derqui, B., Fayos, T., Fernandez, V., 2016. Towards a More Sustainable Food Supply Chain: Opening up Invisible 906 Waste in Food Service. Sustainability 8, 693. doi:10.3390/su8070693 907 Dyllick, T., Hockerts, K., Thomas Dyllick, K.H., 2002. Beyond the business case for corporate sustainability. Bus. 908 Strateg. Environ. 11, 130–141. doi:10.1002/bse.323 909 Engström, R., Carlsson-Kanyama, A., 2004. Food losses in food service institutions Examples from Sweden. Food 910 Policy 29, 203–213. doi:10.1016/j.foodpol.2004.03.004 911 European Union Committee, 2014. Counting the Cost of Food Waste: EU Food Waste Prevention. Auth. House 912 Lords 78. 913 Falasconi, L., Vittuari, M., Politano, A., Segrè, A., 2015. Food Waste in School Catering: An Italian Case Study. 914 Sustainability 7, 14745–14760. doi:10.3390/su71114745 915 FAO, 2011. Global food losses and food waste: extent, causes and prevention, Save Food! 916 doi:10.1098/rstb.2010.0126 917 Ferreira, M., Martins, M.L., Rocha, A., 2013. Food waste as an index of foodservice quality. Br. Food J. 115, 1628–918 1637. doi:10.1108/BFJ-03-2012-0051 919 Finn, S.M., 2014. VALUING OUR FOOD: MINIMIZING WASTE AND OPTIMIZING RESOURCES. Zygon® 49, 992–920 1008. doi:10.1111/zygo.12131 921 Garrone, P., Melacini, M., Perego, A., 2014. Opening the black box of food waste reduction. Food Policy 46, 129–922 139. doi:10.1016/j.foodpol.2014.03.014 923 Gerbens-Leenes, P.W., Moll, H.C., Schoot Uiterkamp, A.J.M., 2003. Design and development of a measuring 924 method for environmental sustainability in food production systems. Ecol. Econ. 46, 231–248. 925 doi:10.1016/S0921-8009(03)00140-X 926 Godfray, H.C.J., Crute, I.R., Haddad, L., Lawrence, D., Muir, J.F., Nisbett, N., Pretty, J., Robinson, S., Toulmin, C., 927 Whiteley, R., 2010. The future of the global food system. Philos. Trans. R. Soc. B Biol. Sci. 365, 2769–2777. 928 doi:10.1098/rstb.2010.0180 929 Goonan, S., Mirosa, M., Spence, H., 2014. Getting a taste for food waste: A mixed methods ethnographic study into 930 hospital food waste before patient consumption conducted at three new zealand foodservice facilities. J. Acad. 931 Nutr. Diet. 114, 63–71. doi:10.1016/j.jand.2013.09.022 932 Guthrie, J.F., Buzby, J.C., 2002. Several Strategies May Lower Plate Waste in School Feeding Programs. Food Rev. 933 25, 36. 934 Hanson, C., Lipinski, B., Robertson, Kai; Dias, D., Gavilan, I., 2016. FLW Protocol Steering Committee and Authors 935 Other Contributing Authors, Food and Agriculture Organization of the United Nations. 936 Jacko; C. C.; Dellava; J.; Ensle; K.; & Hoffman; D. J., 2007. Use of the Plate waste. J. Ext. 937 Jurgenson, Á., 2005. Cómo hacer investigación cualitativa. Ecuador: reimp. Paid. pag. 938 Lipinski, B., Hanson, C., Lomax, J., Kitinoja, L., Waite, R., Searchinger, T., 2013. Reducing Food Loss and Waste. 939 World Resour. Inst. 1–40. 940 Marlette, M.A., Templeton, S.B., Panemangalore, M., 2005. Food Type, Food Preparation and competitive food 941 purchases impact school lunch plate waste by sixth-grade students. J. Am. Diet. Assoc. 105, 1779–1782. 942 doi:10.1016/j.jada.2005.08.033 943 Miles, M., Huberman, A., 1994. Qualitative data analysis: An expanded sourcebook. 944 Mirosa, M., Munro, H., Mangan-Walker, E., Pearson, D., 2016. Reducing waste of food left on plates: interventions 945 based on means-end chain analysis of customers in foodservice sector. Br. Food J. 118. 946 Niaki, S.F., Moore, C.E., Chen, T.A., Weber Cullen, K., 2016. Younger Elementary School Students Waste More 947 School Lunch Foods than Older Elementary School Students. J. Acad. Nutr. Diet. 117, 95–101. 948 doi:10.1016/j.jand.2016.08.005 949 Papargyropoulou, E., Lozano, R., K. Steinberger, J., Wright, N., Ujang, Z. Bin, 2014. The food waste hierarchy as a 950 framework for the management of food surplus and food waste. J. Clean. Prod. 76, 106–115. 951 doi:10.1016/j.jclepro.2014.04.020 952 Pratt, M., 2009. From the editors: For the lack of a boilerplate: Tips on writing up (and reviewing) qualitative 953 research. Acad. Manag. J. 954 Principato, L., Secondi, L., Pratesi, C.A., 2015. Reducing food waste: an investigation on the behaviour of Italian 955 27 youths. Br. Food J. 117, 731–748. doi:10.1108/BFJ-10-2013-0314 956 Rickinson, M., Hall, M., Reid, A., 2016. Sustainable schools programmes: what influence on schools and how do we 957 know? Environ. Educ. Res. 22, 360–389. doi:10.1080/13504622.2015.1077505 958 Rodriguez Tadeo, A., Patiño Villena, B., Periago Caston, M.J., Ros Berruezo, G., Gonzalez Martinez Lacuesta, E., 959 Alejandra, R.T., Bego??a, P.V., Jesus, P.C.M., Gaspar, R.B., Eduardo, G.M.L., 2014. Evaluando la aceptacion 960 de alimentos en escolares; registro visual cualitativo frente a analisis de residuos de alimentos. Nutr. Hosp. 29, 961 1054–1061. doi:10.3305/nh.2014.29.5.7340 962 Smith, S.L., Cunningham-Sabo, L., 2014. Food choice, plate waste and nutrient intake of elementary- and middle-963 school students participating in the US National School Lunch Program. Public Health Nutr. 17, 1255–63. 964 doi:10.1017/S1368980013001894 965 Székely, F., Knirsch, M., 2005. Responsible leadership and corporate social responsibility: Metrics for sustainable 966 performance. Eur. Manag. J. 23, 628–647. doi:10.1016/j.emj.2005.10.009 967 U.S. Environmental Protection Agency, 2014. A Guide to Conducting and Analyzing a Food Waste Assessment. 968 doi:10.1017/CBO9781107415324.004 969 Ventour, L., 2008. The food we waste, Wrap. 970 Wilkie, A., 2015. Food Waste Auditing at Three Florida Schools. Sustain. 971 Williams, P., Walton, K., 2011. Plate waste in hospitals and strategies for change. E. Spen. Eur. E. J. Clin. Nutr. 972 Metab. 6, e235–e241. doi:10.1016/j.eclnm.2011.09.006 973 Williamson, D.A., Allen, H.R., Martin, P.D., Alfonso, A., Gerald, B., Hunt, A., 2004. Digital photography: A new 974 method for estimating food intake in cafeteria settings. Eat. Weight Disord. 9, 24–28. doi:10.1007/BF03325041 975 World Resources Institute, 2016. Food Loss and Waste Accounting and Reporting Standard, World Resources 976 Institute. 977 Wrap, 2011. Food waste in schools. Banbury, Oxon. 978 979