ABSTRACT
Food waste has globally become an increasingly important topic because of the environmental, social and economic consequences. More than one third of all produced food for consumer consumption gets disposed and consumers are for a large part responsible for the amount of food waste. Therefore this research investigates how consumers can be encouraged to consume their purchased food instead of wasting it. Despite the importance of reducing food waste, research regarding influencing consumer food waste patterns is scarce. Moreover, the decision-making process at home has received little attention in academic literature as well. Using a field experiment, this study is the first to investigate the effectiveness of a storage management intervention in reducing food waste, while taking into account the moderating roles of hedonic values and frugality on this effect. We propose that the storage management intervention will reduce the amount of food waste within households. However, we expect that strong hedonic values decrease the effectiveness of the intervention on food waste reduction. We expect frugality to increase the effectiveness of the intervention on food waste. Data was collected by means of a food waste diary and a questionnaire. Although our study could not find evidence to support these expectations, our study shows that hedonic values have a u-shaped linear effect on food waste. Moreover our study confirms findings of former food waste studies. Older people tend to waste more food than younger individuals, and food waste is higher for larger household sizes.
1. INTRODUCTION
In recent years food waste has globally become an increasing important topic because of the environmental, social and economic impact. Substantial amounts of food are wasted throughout the entire food supply chain. More than one third of all produced food is never consumed (Gustavsson, Cederberg, Sonesson, Otterdijk & Meybeck, 2011), while around 795 million people in the world do not even get enough food (FOA Stats, 2016). This means that about one in nine people on earth is starving whilst edible food is thrown away. In the Netherlands consumers and organizations yearly waste approximately 5 billion euros worth of food (Ministry of Economic Affairs, 2014). Half of the waste is at the expense of the consumers. Each year, Dutch households dispose a total of 800 million kilos of food. This is equivalent to over 50 kilos or about 155 euro per person. 14% of all the purchased food is discarded uneaten. Consumers for instance are buying or cooking more than needed and are often insufficiently informed about the possible consequences of food waste (Milieu Centraal & NIBUD, 2004). Virtue foods, such as bread, dairy, vegetables and fruit are the most disposed food products by consumers (Gustavsson et al. 2011). Consumers are thus heavily responsible for a huge part of the amount of food waste. To a large extent this household food waste is preventable (O’Donnel, 2014). Therefore it is necessary to motivate and learn people to better manage their food waste behaviour. The raising question is, how can consumers be encouraged to consume their purchased food instead of wasting it?
Food waste is a complex matter. Over the years food waste and consumer waste behavior is examined in various studies and numerous reasons and factors are discovered (Gustavsson et al., 2011; Koivupuro, 2012; Porpino, Parente, & Wansink, 2015; van Doorn, 2016). Factors that seem to be of most importance at the individual consumer level can be categorized in: 1) lack of planning, 2) management of food purchase, storage and preparation, and 3) not reusing food or leftovers (Aschemann-Witzel, et. al, 2015, Parfitt et al., 2010).
These factors are related to two important consumer decision-making processes that influence food waste behavior, namely consumer decision-making at the point-of-purchase (e.g. supermarket) and decision-making of consumption and disposing food at home (Porpino, Parente, and Wansink, 2015). These two processes are interrelated, because the amount of purchased food affects the options consumers have at home and the amount of disposed food products influences how much people are willing to buy in their next grocery-shopping trip (Van Doorn, 2016). In other words, one positive change in one process will impact the other decision-making process as well. Yet, the process of buying (food) products is a well-reviewed topic in literature studies (Rook and Fisher, 1995). Controversially, the decision-making process such as consumers’ food waste behavior at home has not received that much attention in academic literature (Cappellini 2009; Cappellini, and Parsons 2012; Koivupuro et al. 2012). The fact that food waste occurs at the last phase of the consumption process might explain why this has not been investigated and described intensively in marketing literature (Cappellini, 2009).
Looking at the immense global economic and environmental effects of food waste, marketing research should focus more on consumers’ food waste behaviour at home. It should focus more on including non-purchase elements of consumer behavior (e.g. product use and disposal). Despite the importance of reducing food waste due to several negative consequences, research about influencing consumer food waste behaviour is very scarce. Investigating drivers of change for consumer’s food waste behaviour at the storage phase might enable researchers to derive at opportunities to reduce the amount of food waste. Unfortunately, most studies examined the amount of food waste or the reasons for food disposal; only little research has been done in finding an effective solution to decrease food waste. This study therefore attempts to gain more insight on this matter.
are found to be successful in influencing consumer performance or behaviour (Abrahamse, Steg, Vlek & Rothengatter, 2005). Hence, interventions or tools might also positively influence food waste behaviour. This research consequently examines and evaluates the effectiveness of a storage management intervention in reducing consumer food waste. Accordingly, the problem statement of this research is as follows:
‘To what extent is a storage management intervention effective in changing consumer behaviour and reducing food waste?’
Some people are more open for persuasion or engage more in information processing and interventions than others (Hoyer et al. 2013). This implies that the effectiveness of an intervention aimed to change consumer behaviour might be affected by consumer characteristics. Previous studies have found that consumer characteristics (e.g. values, personality traits and lifestyles) influence consumer behaviour. Actual consumer behaviour is influenced by factors such as social-demographic factors (e.g. age, household size, and occupation) and personality factors (e.g. personality traits and values). Influences of social-demographics on food waste behaviour have already been investigated in previous studies. However, personality factors have had little attention in food waste research, while for example consumer values are highly related to (pro-environmental) consumer behaviour (Hoyer, Pieters & MacInnis, 2013; Vinson et al., 1977; Zeithaml, 1988). Given the growing importance of hedonic values and frugality in consumption behaviour, these two characteristics seem to be most interesting for this research. Frugality is conceptualized as a characteristic combining values and lifestyle dimensions.
effect on environmental behaviors (see also Batra & Ahtola, 1991) and suggest that these values are essential in environmental researches. Given the environmental impact of food waste (Koivupuro, 2012), we assume that hedonic values are also essential to food waste behaviour. Moreover, researchers propose that interventions aimed to stimulate pro-environmental actions should take into account hedonic values, because they may obstruct behavioural change (Steg et al. 2014). Hedonism is assumed to negatively affect the effectiveness of the storage management intervention because the aim of the intervention is not in line with the values of hedonism (Bolderdijk, Gorsira, Keizer and Steg, 2013). Therefore, the second research question of this study is:
“To what degree are hedonic values affecting the effectiveness of the storage management intervention?”
“To what degree is the level of frugality affecting the effectiveness of the storage management intervention?”
2. LITERATURE REVIEW
2.1 Household food waste
Figures show that half of the food waste is at the expense of the consumer and therefore an important source for food waste. Consumers are heavily responsible for a huge share of food waste that to a large part is preventable (O’Donnel, 2014). In this research food waste is referred to as avoidable food waste. Avoidable waste means that the food was still eatable before it was thrown away (Quested and Johnson, 2009). WRAP defines it as food that is thrown away because it has not been approved to be at his best anymore or is no longer desired. Avoidable food waste is causing the largest amount of food waste (Aschemann-Witzel, de Hooge, Amani, Bech-Larsen, & Oostindjer. 2015). To illustrate, research shows that in the UK half the amount of food was disposed because it wasn’t used in time (WRAP, 2015).
way food is sold in stores. Waarts and colleagues (2009) also underline that food waste can be prevented during the steps of buying, preparation or storing of food items. Purchasing too much food results in higher food waste, however cooking, preparing and serving too much food than necessary accounts for even more (40%) of the disposed food items within households (Quested et al., 2013, Waarts et al., 2009). Van Doorn (2016) describes the phenomenon of food waste as a discrepancy between consumer purchases and the actual consumption. Thus there are two important decision-making processes that influence food waste; decisions consumers make at the point-of-purchase (e.g. supermarket) and decisions about consumption and discarding food at home (Porpino, Parente, and Wansink, 2015). Both decisions are related because the amount of purchased food affects the options consumers have at home, and the quantity of disposed food products impacts how much consumers will buy in their next grocery-shopping trip (Van Doorn, 2016). This implies that influencing one of these processes will result in a change in both decision-making processes.
influence the behavior (Abrahamse, et al., 2005). Interventions may target individuals’ perceptions, preferences or abilities aiming to change behavior. Or alternatively they may be used for changing the context in which individuals make decisions (e.g. laws or rewards) (Abrahamse et al., 2005). Experiments and interventions can contribute to a shift in analysis to solutions (Aschemann-Witzel, et. al, 2015) and Evans (2011) suggests that interventions in material context are crucial.
Interventions generally are more effective if they are aimed at important antecedents of the appropriate behaviour. Consequently, understanding of the factors that stimulate or impede the behaviour is important. Research indicates that Dutch consumers are willing to decrease their individual food disposal, yet they run into some obstacles (Bogerd, 2001). Food wastage is largely caused by habits that concern the organization of the daily household (Bogerd, 2001). Educating and informing can be effective, only consumers will not likely change deeply imbedded consumption behavior (Abrahamse et al., 2005). Consumer values and traits are important predictors of consumer behavior and therefore the following section will elaborate on the role of consumer values and traits on consumers’ food waste behaviour.
2.3 The role of consumer values and traits on food waste
dependent on values and other psychological foundations. The following paragraphs elaborate on how hedonic values and the consumer trait frugality handle different trade-offs regarding environmental behavior and food waste.
2.3.1 Hedonic values and food waste
Consumers are increasingly operating on the values of hedonism (Hoyer et al., 2013). Consumers with strong hedonic values tend to seek a more satisfying need in consumption than only basic needs. Hedonic consumption behaviour can be described as; pursuing enjoyment, seeking sensual pleasure or fun, and searching for experiences that make them feel good (Dhar and Wertenbroch, 2000; Hoyer et al, 2013). According to Fischer & Arnold (1990) hedonists fulfill their satisfaction from instant hedonic pleasure experience of consumption. Individuals with strong hedonic values focus on ways to improve their feelings in certain situations like avoiding effort or seeking enjoyment. Hedonic eating goals usually end in indulgent (luxury) consumption (Tice and Bratslavsky, 2000). Eating for pleasure rather than to maintain energy is consequently referred to as hedonic hunger for a reason. Food consumption seems to be increasingly driven by pleasure and other reasons rather than the daily need for calories (Lowe and Butryn, 2007). Palatability appears to be essential to hedonic hunger and has a great effect on whether food will be desired and consumed (Painter, Wansink & Hieggelke, 2002). Hoyer et al. (2013 p. 411) state that the rise of, for example, chains as gourmet cupcake bakeries “illustrate the hedonistic search for food that tastes good and makes people happy”. Although they might be concerned about their health, hedonic consumers will not eat healthy unless it taste good (Hoyer et al., 2013). These individuals are the ones that might not want to eat the same meal or leftovers, dispose food that does not taste good, and throw away food if new food variations that do not match their expectations (Cappellini, 2009).
Pro-environmental behaviour is mostly considered as the right thing to do, though in most cases it is less profitable, less pleasurable, more time-consuming or more effortful compared to less environmental actions (e.g. wasting food). These consequences are not appealing for individuals with hedonic values as these values generally focus on improving feelings and reducing effort. Hence, people may withhold from pro-environmental activities because they value the environment lower than they value their assets, comfort, or enjoyment (Steg, et
al., 2014). Nevertheless people just might also want to engage in pro-environmental behaviour for hedonic reasons because it may be perceived as enjoyable (Bolderdijk et al., 2013) or because people feel good about themselves if they are doing the right thing.
2.3.2 Frugality and food waste
The economic downturn and interest in the environment has resulted in a growth in frugality (Birkner, 2013 and Bove et. al., 2009) by which people have noticeably changed their saving and spending behaviors (Hoyer et al., 2013). The word “frugality” is derived from the Latin word ‘frugalitas’ and means virtue or thriftiness. Frugality is defined as “a uni-dimensional consumer lifestyle trait characterized by the degree to which consumers are both restrained in acquiring and in resourcefully using economic goods and services to achieve longer-term goals” (Lastovicka, Bettencourt, Hughner and Kuntze, 1999, p. 88). There is some discussion if frugality must be seen as a value or a lifestyle trait. Todd and Lawson (2003) propose that frugality can be view as either both. “Frugality as a value refers to the importance that an individual attributes to frugality as a guide to action and judgments across specific situations” (Bove, Nagpal and Dorsett, 2009). In contrast, frugality as a lifestyle denotes the occurrence and intensity of frugal actions that consumers undertake, related to the abovementioned definition by Lastovicka et al. (1999). The latter is of most importance for this research and is measured accordingly. As already mentioned before, lifestyle is dependent on consumer values; hence frugal values will affect the frugal activities undertaken by consumers.
emphasis lies more on pursuing collective rather than personal goals (Schultz, et. al, 2005). A characteristic of frugality is that people tend be less materialistic, less impressionable by others and more price- and value conscious. Frugal consumers tend to live on a tight budget and have great self-discipline in spending and efficient use of existing resources (Lastovicka et al., 1999). For example, they are more motivated to eat leftovers for lunch instead of buying a take out menu (Hoyer et. al, 2013). Frugality is not the same as poverty, as one voluntary chooses to sacrifice for a more worthy goal (Jeurissen & van de Ven, 2011). Frugality is associated with a simple lifestyle, however this is not the key motive. Frugality is about reducing waste instead of reaching some higher-order goals like personal growth (Bove, Nagpal and Dorsett, 2009). Frugality might be seen as the antithesis of hedonism, because consumption is not about fun, pleasure and materialism, but rather about consuming and using resourceful products to realize long-term goals (Lastocivka, 1999). Hence, people that tend to pursue a frugal lifestyle are expected to throw away less food.
It is argued that frugality is linked to age, as older people more likely pursue a frugal lifestyle. Frugal values have an emphasis on social goals and can be seen as the counterpart of goals prioritized by material values. Above that, frugality leads to consumers buying larger packages (bulk buying) because it is perceived as relatively cheaper (Dacyczyn 1992). However buying larger packages increases the likelihood of throwing away the unused surplus (Porpino, et al., 2015) and may lead to disposal. Research states that lifestyles have a significant effect on consumption. Frugality has been linked to sustainable consumption with specific importance for avoidance of food waste (Evans, 2011). Frugal consumers are focused on the long-term rather than the short-term orientation. Considering the values attached to frugality by Lastovicka (1999), frugal consumers are expected to eat leftovers and first use food products that they already purchased before buying anything new. A reason for frugal consumers to act environmental friendly is that better use of their resources makes them feel good (Lastovicka, 1999).
2.4 Conceptual model
Based on the literature review a conceptual model was established and presented in figure 1. The storage management intervention will positively affect the amount of household food waste, as other motivational techniques were successful in changing behaviour before e.g. love food hate waste campaign (Graham-Rowe∗, Jessop, Sparks 2013; WRAP, 2014). Informational strategies are generally effective for pro-environmental behaviour if not taking to much effort from consumers. However, according to literature consumer values and lifestyles might influence and moderate consumer behaviour. Certain consumer characteristics or values can inhibit the particular behaviour. Hedonic values have found to have a negative influence on the effectiveness of interventions, as it may impede behavioral change (Steg et al., 2014). Frugality is assumed to have a positive effect on the effectiveness of the intervention oi reducing food waste, as frugal tend to seek for ways to reduce or recycle waste (Evans, 2011; Schreurs, Martens & Kok, 2012) and have the self-discipline (Lastovicka, 1999) and motivation to use the intervention because their strong values are linked to sustainable consumption (Bolderdijk, 2013). FIGURE 1 – conceptual model 2.5 The effect of a storage management intervention on food waste
Research suggests that people should be trained in food management because this will empower them to reduce food waste to a minimum (Graham-Rowe∗,
Storage management intervention
Amount of household food waste Consumer values and traits:
Jessop, Sparks 2013). Incorrect food management is linked to lack of knowledge about store management and Porpino, Parente and Wansink (2015) therefore suggest that campaigns regarding food storage will encourage better practice of food management. Urgent action is necessary in the form of education, for example to store food correctly (Porpino, et al., 2015). Steg and Vlek (2009) indicate that informational strategies are generally also effective when the expected pro-environmental behaviour is fairly convenient and it won’t take too much effort in terms of costs, and time. Yet, only educating and informing consumers is not likely to change deeply imbedded consumption behavior. Though there are already several motivational techniques in use for food waste reduction of which some are noticed successful (e.g. the Love Food Haste Waste campaign and Kliekipedia). Other sustainable behaviour research has also found a positive effect of interventions on consumer waste behaviour: Bernstad, La Cour Jansen and Aspegren (2013) proved that an intervention campaign was effective for food recycling behaviour, and an intervention with tailored information, including ways to reduce waste, directly positively influenced household energy waste (Abrahamse, Steg, Vlek & Rottengather, 2007)
A combination of actions seems the most promising in an intervention; 1) provide practical tips or tools and 2) raise motivation and involvement by stressing ethical reasons together with highlighting a win-win situation (Aschemann-Witzel, et. al, 2015). This is exactly what the storage management intervention in this research incorporated. Above that, the storage management intervention might especially be effective because the tool functions as a visual reminder (Bernstad, 2014).
visual reminder and informing the consumer about choice options (Achemann-Witzel, 2015). Above that, we assume that this storage management intervention will be effective because there is a lot to gain within the storage phase (Porpino et al, 2015). If households better manage their storage management with the use of this tool, it will lead to a reduction in food waste (Aschemann-Witzel, et. al, 2015; Graham-Rowe∗, Jessop, Sparks 2013). Therefore we propose that the intervention will have a positive effect on the amount of household food waste. Hence the hypothesis is the following: H1. A storage management intervention will reduce consumers’ food waste 2.6 The role of hedonic values on intervention effectiveness Informational interventions, like awareness campaigns are developed to inform people about environmental consequences of their actions hoping that it will result in pro-environmental improved intentions or behavior (Bolderdijk, Gorsira, Keizer and Steg, 2013). Nevertheless Bolderdijk, Keizer and Steg (2013) argue that interventions will not directly affect people’s motivation if the aim is not in line with what an individual values. Their research findings suggest that values determine the effectiveness of interventions. Hedonic values are therefore assumed to weaken the effectiveness of the storage management intervention, given the various values involved that are not in line with the pro-environmental intention to reduce food waste.
that utilitarian orientation is negatively affected by hedonic values (Wang et al., 2000). If the intervention is perceived as too practical or functional, these strong hedonic values will impede the responsiveness to the stimuli. Leftovers are seen as boring because they lost their novelty, since they have been served and or eaten before (Cappellini, 2009). Moreover, although leftovers can be seen as an easy or quick way to have meal, it also requires time and effort in re-using them. It involves a process in which the food needs to be discovered, evaluated and re-used as illustrated by Parsons (2008). As hedonists are novelty seekers and have the tendency to avoid effort, it may well be that pro-environmental behavior as re-using food or increasing food management skills are less of interest for them than people with weak hedonic values. Values are influencing the objectives of individuals in certain situations and their perspective on the diverse consequences, which in turn affect the options that people prefer and the choices they make (De Groot, Steg & Poortinga, 2013). Participants with strong hedonic values care less about food waste and more about seeking pleasure as Steg et al. (2014) suggest that hedonic values affect environmental behaviour in a negative way. Therefore we assume that they will be less inclined to participate or be motivated to intentionally use the intervention with the aim to reduce food waste (Wang et al., 2000). Consequently, taking into account that hedonic values are found to be important barriers for behavioral change within interventions (Steg et. al 2014) concerning pro-environmental behaviour, we assume that hedonic values will negatively affect the effectiveness of the intervention of in reducing food waste. Thus, the following hypothesis is presented:
H2. Strong hedonic values will reduce the effectiveness of a storage management intervention to reduce consumers’ food waste.
2.7 The role of frugality on intervention effectiveness
more motivated to use the intervention since their values are in line with the communal goal of the intervention, reducing food waste (Bolderdijk et al., 2013). Moreover, individuals pursuing a frugal lifestyle tend to have a great desire to learn and do things independently, which suggest a need for information and or cognitive resources. Frugal consumers are expected to increase the effectiveness of the intervention because making better use of their resources is enjoyable (Lastovicka et al., 1999) and the intervention can be viewed as an helpful tool in accomplishing their goals regarding avoiding waste.
Moreover studies show that consumers pursuing a frugal lifestyle have great desire to learn and do things independently, which suggest a need for information and or cognitive resources. However Goldsmith, Reinecke & Clark (2014) findings suggest that frugal consumers are more independent in making decisions and do not rely on others. Moreover, they argue that independence results in making consumption decisions in which people do not rely on society or external sources (Goldsmith, Reinecke & Clark, 2014). Hence, this might inhibit the responsiveness to the intervention. Another, reason why frugal consumer might not be encouraged to use the intervention is that consumers with a frugal lifestyle have self-discipline and already know how to resourcefully use their supplies (Lastovicka et al., 1999). In that case it is expected they already are likely to be committed to minimize their food waste (Schreurs et al, 2012) and the storage management intervention would be needless. Though, there is always room for improvement. Besides, the intervention can lead to more convenient and fun use of resources.
consumers are found to be less likely to waste food (Koivupuro et al., 2012). However, we assume that frugality most likely will positively affect the effectiveness. Hence, the following hypothesis is formed:
H3. A high level of frugality will increase the effectiveness of the storage management intervention to reduce food waste.
3. METHODOLOGY
3.1 Research design A quantitative research is performed in the form of a field experiment in order to be able to answer the research question of this study. The aim of the research is to determine the effectiveness of the storage management tool (IV) in reducing food waste (DV) and if this relationship is affected by consumer characteristics (hedonic values and frugality). An experimental field research design was used because experimentation allows for greater certainty in interfering the causal relationship between variables (Aronson, Wilson & Brewer, 1998). Household’s food waste behaviour takes places in people’s homes, which is in their natural environment. Therefore this field experiment that takes place in the homes of the participants is an appropriate method to gain insight in household’s food waste behaviour. FIGURE 2 - Experimental design Between-subject design This experiment measures the effectiveness of the storage management tool by comparing two conditions (intervention vs. no intervention), see figure 2. Participants were randomly assigned the one of the two experimental groups: 1) the study group (this group received the intervention tool) or 2) the control group (this group received no intervention tool). Random assignment of participants ensures the prior equality of the two groups, given that the sample size is sufficiently large. Randomization guarantees that extraneous factors are equally distributed between the groups and thus unlikely will affect the outcome (Malhotra, 2010). Any differences between the groups will be a result from the experimental procedures rather than differences between the participants (Aronson et al., 1998). The minimum amount of respondents has to be 25 per condition (n≥25). The sample size must be at least 50 for this research.Condition Groups
No intervention Control group
3.2 Method of Data Collection
Two combined methods were developed to collect data: 1) a food waste diary and 2) a questionnaire. All materials were distributed in Dutch because the participants (supermarket customers) are Dutch. Data collection was held in November 2016 at Albert Heijn Paterswolde, The Netherlands.
3.2.1 Food waste diary
A food waste diary approach is used to measure the food waste of the participants for a period of seven days. The diary consists of two parts (see Appendix A). In the first section participants can mark the frequency of their food purchases per day: if they 1) ordered take away food, 2) went out for dinner, 3) or did grocery shopping at the supermarket (large, small or extra groceries). In the second part participants are asked to keep a daily record of their wasted quantities of 24 suggested categories (food and drinks) and additionally a category named ‘other’ is added for certain food products that cannot be ascribed to one of the abovementioned categories. These 24 categories can be divided into virtue category (e.g. bread, diaries, and fruits), vice category (chocolate and candy), and neither both categories (coffee, rice, and meat) according to (Van Doorn & Verhoef, 2015). In this research the food waste will be divided into liquid food waste (=ml) and solid food (=gr). The solid food waste is of most importance for this research, however the proportion of wasted liquid food will also be taken into consideration in the analysis. 3.2.2 Questionnaire
this study. Furthermore participants could evaluate the intervention tool. The questions are discussed in the measurement section.
3.3 Procedure of Experiment
The experiment took place at Albert Heijn, a supermarket in Paterswolde, the Netherlands. Participants were recruited on two days, Friday 11th and Friday 18th of November 2016. These days have been chosen because figures show that Friday is one of the most popular shopping days of the week among customers of this supermarket and therefore the most suitable for recruiting participants. In total 140 sets of materials have been distributed. Each set consisted of a letter explaining the study process, food diary, and a questionnaire. Among 70 of the participants the intervention tool was included and the other half did not received the intervention tool.
Respondents were asked to keep a food waste diary for a period of seven days and had to fill in the questionnaire at the end. The questionnaire could be filled out on paper or online via Qualtrics (an online survey software tool). A link to the online survey was given in the information letter and also a reminder was send via e-mail. E-mail addresses were obtained via the informed consent, as participants were kindly asked to fill in their e-mail on the form, to enable the researchers to contact them in case the participant would not return the materials. Participants were informed that the experiment was anonymously and personal data would not be used for other purposes, contact data was solely used for sending a reminder about the study materials and the lottery.
behavioral changes. After signing the informed consent form, participants were given one of the two sets of materials (intervention vs. no intervention).
After one week (Saturday 19th or Saturday 26th of November 2016) participants were asked to hand in their food waste diary and questionnaire in the delivery box or at the service counter at Albert Heijn Paterswolde. Participants that brought back the materials received a luxurious box with nuts as a token of appreciation for their participation. Above the present and the lottery, as a little incentive to encourage people to bring back the study materials, we promised that if 75% of the participants would hand in the materials, the researchers would make a financial donation to the food bank of Paterswolde.
3.4 Measurements
3.4.1 Food waste
The dependent variable food waste was measured by means of data collected with the food waste diary (see Appendix A), in which participants measured their own waste. A food waste diary allows for detailed information on reasons for food waste and gives more detailed information for each disposal (Koivupuro et al., 2012). The food waste diary was merely focused on avoidable food waste categories. As mentioned before, avoidable food waste are food products that were edible before it was thrown away and this food waste could have been prevented; such as leftovers or unprepared food products that are forgotten and have passed their expiration date. Weights were attached to the categories to make it easier for the participants to fill in their food waste.
According to Koivupuro (2010) a diary method is demanding for participants and it is possible that they neglect or decide to not keep track of their food wastage. Also, the diary might cause some awareness of the true purpose of the research and this can cause a decrease in the quantity of wasted food compared to a normal unobserved situation. Consequently these some limitations may lead to a small underestimation of the actual share of food waste.
3.4.2 Storage management intervention
The ‘Voedingscentrum’ created a storage management intervention with the aim to reduce food waste (see appendix B). The storage management intervention tool consisted of three different ‘fridge stickers for organizing one’s fridge and freezer inventory’: 1) An ‘eat-me-fast zone’ sticker; people can put food products in this ‘zone’ that need to be consumed quickly. 2) ‘Forget-me-not, eat me’ stickers divided into three categories: for leftovers, for freezer items (with expiration date) or for in-home use. 3) A ‘leftover inspiration ideas’ sticker, which provides 4 basic recipes for leftovers (soup, salad, stir-fry, and omelet) which can be used in combination with the ‘forget-me-not’ leftover stickers.
3.4.3 Hedonic values
Hedonic values scales are based on validated scale items proposed by Schwartz (1992). Participants rate the importance of three values (pleasure, enjoying life and gratification for oneself) “as guiding principles in their lives”. The items were measured on a nine-point Likert scale ranging from -1 (opposed to my principles), 0 (not important), to 7 (extremely important). See table 1.
3.4.4 Frugality
The set of eight items of frugality has also been validated in previous research and is developed by Lastovicka et al.’s (1999). Respondents are asked to rate eight statements on a five-point Likert scale, ranging from 1 “describes me not at
all” to 5 “describes me extremely well”. All responses were summed up, which
TABLE 1 – Overview of scale items Schwartz (1992) Hedonic values Moderator 1. Pleasure 2. Enjoying life 3. Gratification for oneself 9-point scale Opposed to my principles -1, Not important 0 – 1 – 2 – 3 – 4 – 5 – 6 – 7 Extremely important Lastovicka et al. (1999) Frugality Moderator 1. If you take good care of your possessions, you will definitely save money in the long run
2. There are many things that are normally thrown away that are still quite useful
3. Making better use of my resources makes me feel good
4. If you can re-use an item you already have, there's no sense in buying something new 5. I believe in being careful in how I spend my money 6. I discipline myself to get the most from my money 7. I am willing to wait on a purchase I want so that I can save money 8. There are things I resist buying today so I can save for tomorrow 5-point scale Describes me not at all 1 – 2 – 3 – 4 – 5 describes me extremely well 3.4.5 Control variables
stability of the model. Due to the small data set and the risk for over specification of the model, only household size was incorporated as control variable in this research (a comparable approach was used by Bijmolt, van Heerde, and Pieters 2005). Household size was controlled for in case it may affect the main effects in this research since household size is found to be a strong explanatory variable for household food waste (Cox, and Downing, 2007). The control variables will be further elaborated on in the results section, paragraph 4.8. 3.5 Correlation and reliability
The two constructs in this research are measured with multiple scale items. Before proceeding it was checked whether these items correlate with each other to examine whether these items could be averaged.
3.5.1 Hedonic values
The correlation analysis showed that the items of hedonic values (HV1, HV2 and HV3) are significantly correlating with each other. Reliability analysis on the three items measuring hedonic values presented a Cronbach’s Alpha of 0,841. When ‘gratification for oneself’ would be excluded the Cronbach’s Alpha would increase to 0,960. A Cronbach’s Alpha of 0,960 is extremely satisfactory, however a value of 0,841 is satisfactory enough. Therefore the three items were averaged for measuring hedonic values. 3.5.2 Frugality
Reliability analysis on the eight variables for measuring frugality showed a Cronbach’s Alpha of 0.797. However correlation analysis showed that not all variables correlate significantly with each other. Therefore a factor analysis (PCA analysis) was conducted to see if the items have common factors. First it was examined if a factor analysis was appropriate. The Kaiser-Meyer-Olkin (KMO) analysis showed a KMO value above the cut-off value of 0.5 (KMO = 0.684) and the Barlett’s Test of Sphericity was highly significant thus a factor analysis was applicable. Moreover, the sums of the squared loadings are all above 0.400 which is desirable (except for F2),
The factor analysis yielded two factors with an eigenvalue larger than 1, explaining a total of 58,84%, see table 2. The first factor is considered as frugality concerning the present and is more short-term focused (F3, F4, F5 and F6). This factor explained 42,48% of the variance. The second factor reflected frugality regarding the future and is more long-term oriented (F1, F7 and F8), explaining 16,36%. Item F2 did not load on any of the factors and was excluded for further analysis.
TABLE 2 –Rotated factor loadings and Cronbach’s Alpha of Frugality
Items Factor 1 Factor 2
analysis was continued with factor 1, the items of factor 2 were therefore excluded for further analysis. The correlation analysis showed that the items within factor 1 (F3, F4, F5, F6) significantly correlate with each other and reliability analysis shows a pleasing Cronbach’s Alpha of 0,783. Hence, the four items measuring frugality were averaged into one variable.
3.6 Analytical Method
Hypothesis 1 is tested with a one-way ANOVA analysis, to determine if the amount of food waste was different for the group that used the storage management intervention from the control group. Multiple regression analyses were conducted to test the remaining hypotheses. An OLS multiple regression analysis was performed with the independent variable and the moderator(s) to estimate the contribution by each factor to the total variation of the amount of food waste. Hypotheses 2 (model 1) and 3 (model 2) were separately tested with a moderating regression analysis.
A three-step procedure was used in the analysis. We first regressed the independent variable (intervention) on the dependent variable (food waste in gr), while controlling for household size. Then the independent variable (model 1: hedonic values, model 2: frugality) was added to the regression equation in step two. In the last step, the interaction term (model 1: hedonic values and intervention, model 2: frugality and intervention) was entered in the model. The moderating effect was examined by the change in the R-squared after adding the interaction term. Additionally, as a robustness check, the analysis was run considering the additional food waste variable in milliliters, referred to as: food waste (ml).
test procedure. All continuous variables were standardized for better interpretability of the results and were used as a means for reducing multicollinearity. Given the circumstances of this research and its experimental design, an alpha level of 0.10 is used within this research (Pagano, 2013).
4. RESULTS
4.1 Descriptive Statistics
In total 55 participants (39,3% response rate) participated in this research. However three respondents did not include their respondent number in the online questionnaire and cannot be linked to a food waste diary. Moreover two respondents only handed in their food waste diary and not included their questionnaire. Furthermore 13 respondents have filled in their consumption during seven days instead of their food waste. Which leads to a total of 37 valid respondents that have handed in both materials correctly. Due the small sample no data was excluded from the dataset and all data was used when the type of analysis allowed for it. The amount of missing values for all descriptive variables was under 10%, therefore these missing values were neglected as Bennet (2001) claimed that statistical analysis is not biased when less than 10% are missing.
The descriptive statistics of respondents’ socio-demographics are reported in table 3 and figure 3. The majority of the sample is female (80%) with an average age of 57 (Mean = 57,68, SD = 13,98) ranging from an age of 25 to 86 years. Of all (valid) respondents 70,5% is older than 50 years. An explanation for this could be that older people were more willing to participate in our research or that the average age within Paterswolde is relatively high. The sample is on average highly educated with a mean score of 3,98 on a 6-point Likert-scale, with 59,6% of the people having a HBO degree or higher. Results indicate that the respondents have a relatively high income. In total 47,3% of the participants have indicated that their collective household income is greater than €3.000, and even 25,5% of them earn more than €4.000. A possible explanation for this could be that inhabitants of Paterswolde in general are higher educated and they have higher incomes. Half of the sample lives in a 2-person household (50%).
8 male participants. The overall low amount of male participants can explain this difference. TABLE 3 – Socio-demographics FIGURE 3 – Socio-demographics Gender Education level
Table 4 displays the distribution of participants between the two conditions (intervention and control group); the two groups are not equally large. The descriptive statistics for the independent variables and the dependent variable are presented in table 5. The minimum amount of waste was zero, as some
N Mean Std. Deviation
Age 44 57,68 13,981
Household size 48 2,44 1,109
Working hours 46 16,59 15,356
Income Frequency Percent
€ 1000 - € 1499 7 12,7% € 1500 - € 1999 4 7,3% € 2000 - € 2999 10 18,2% € 3000 - € 3999 12 21,8% € 4000 or more 14 25,5% I rather not say 6 10,9%
participants did not waste any food during the week. The average amount of food waste was 627,92 grams and 234,87 milliliter per household. TABLE 4 – Distribution of conditions TABLE 5 – Descriptive statistics variables
4.2 Manipulation check The results indicate that the instruction for the food waste diary was not clear enough or participants did not read carefully enough to understand that they should have filled in their waste instead of their consumption. Above that, most participants have stated that they did not use the intervention tool (only 4 stated they used the tool). However, according to Adams & Lawrence (2015) manipulations can still have an effect unconsciously. The storage intervention tool may have not been used, however it might have triggered the respondents unconsciously to adjust their storage habits or keep track of their food waste. Therefore, as a robustness check, the data is analyzed with and without those who have ‘failed’ the experiment to test if excluding them might lead to other results. 4.3 One-way ANOVA: intervention on food waste In order to analyze whether or not households’ food waste differs between the control group (n = 19) and intervention group (n = 20), a one-way ANOVA was performed, with independent variable intervention on dependent variable food waste in grams. There was no statistically significant difference between the
Intervention Frequency Percent
Intervention group 31 56,4%
Control group 24 43,6%
N Mean Std. Deviation Cronbach’s Alpha
Hedonic values 52 7,9167 1,141 0,841
Frugality 53 3,406 0,717 0,783
Food waste (ml) 39 234,87 455,069 -
group means (F (1,37) = 2,492, p = 0.123). An overview of the results can be found in the table 6. This implies that there is no significant difference in food waste (gr) between the control group (M= 883,11 SD= 1236,32) and the intervention group (M= 385,50, SD= 467,79), see figure 4. This means that people from the intervention group had no lower amounts of food waste compared to the control group; hence the intervention did not reduce the amount of food waste.
As robustness check, an additional test was performed with food waste in milliliters as dependent variable. Likewise no statistically significant difference was found for food waste in milliliters (F (1,37) = 0,050, p = 0.824). Which indicates that there is no significant difference between in food waste (ml) between the control group (M=251,84, SD= 367,81) and the intervention group (M=218,75, SD = 534,28). Also here the intervention did not lead to reduction in food waste. These average means are visualized in figure 4.
FIGURE 4 – Mean plot intervention on food waste 4.4 Validation This section reviews the pre-tests and validation assumptions that must be met before data of the regression analyses can be interpreted. As mentioned before, two multiple regression analyses were conducted, referred to as model 1 (hedonic-model) and model 2 (frugality-model) based on the dependent variable food waste in grams. Hence assumptions were examined these two both models.
4.4.1 Independence of observations
The assumption of independence of observations (e.g. independence of residuals) can be checked with the Durbin-Watson statistic. Field (2009, p. 139) recommends that only values below 1 or values above 3 will definitely be a cause for concern. Considering the Durbin-Watson statistics of both models it could be concluded that there was independence of observations, 2.510 (model 1) and 2.278 (model 2).
4.4.2 Normality check
increase the possibility of a false positive with the use of a parametric test (e.g. multiple regression).
TABLE 7 - Shapiro Wilk Test for non-normality
However, literature review showed that the Shapiro-Wilk test is extreme sensitive to small deviations from the normal distribution (Ahad, Yin, Othma and Yaacob, 2011). Therefore visual inspections of the histogram, Q-Q plot and Boxplot serve as additional checks on the normal distribution. This can be judged individually because there are no strict rules that indicate boundaries of non-normality for a parametric test (McDonald, 2014). A visual inspection of the histograms, Q-Q plots, and Boxplots showed that the residuals of food waste were not normally distributed. It showed a right skewed distribution of the variables, which meant that outliers were mainly positioned at the high values of the scales. The histograms for hedonic values and frugality showed a reasonable normal distribution. Researchers argue that an OLS regression is fairly robust to reasonable deviations and one should not worry about non-normality particularly when sample size is small (Osborne and Waters, 2002). Hence, only food waste was still a cause for concern. To reduce the impact of the high values of the variables, a log transformation was used to overcome this skewness of the distribution, which is widely used to deal with skewed data (Feng et al, 2014). Consequently, food waste was thereafter normally distributed, see column 2 of table 7. Hence considering the above, it was decided that the normality distribution is no problem in this dataset.
4.4.3 Linear relationships
The linearity assumption argues that the relationship between the independent and dependent variable should be linear. A scatterplot between the individual
1 Shapiro-Wilk 2 Shapiro-Wilk
Statistic Sig. Statistic Sig.
Food waste gr .632 .000 Food waste gr (ln) .938 .294
Hedonic values .859 .000 Hedonic values (ln) .933 .246
independent variables and the dependent variable, log of food waste (gr), was created by means of curve estimation. This measure was used because it controls which regression line has the best match with the data, based on its R-squared. Linear, logarithmic and quadratic functions were tested, as these are the most common functions. First, no linear relationship was found for hedonic values on food waste (gr) with a R-squared of .019 (β = -.21, p > .10) and no logarithmic relation (R2 = .030, β = -2.02, p = > .10). However a quadratic function was discovered with an explained variation of 21,6% (R2 = 0.216, β1 = -9.36, β2 = 0.60, p = 0.042). Therefore the variable of hedonic values was transformed in a quadratic variable and this higher-order variable (hedonic values 2) was added to the model to coax a linear relationship. The relationship between hedonic values and food waste is further examined by means of a regression analysis in section hypothesis testing, paragraph 4.5.1. Secondly, no linear relationship was found for food waste in grams (R2 = .030, β = -2.02, p = > .10) and no quadratic function was found (R2 = .118, p = > .10). Yet, the results indicated a significant logarithmic relation (R2 = .100, β = -1.55, p = < .10). Hence, a log transformation of frugality was performed to overcome this non-linearity.
4.4.4 Homoscedasticity
FIGURE 5 – Scatterplot standardized residuals vs. predicted values Model 1 Model 2 4.4.5 Correlation and multicollinearity
A pre-analysis was conducted to check for any correlations among the four variables in the conceptual model. Table 8 specifies an overview of the correlation analysis. This analysis shows low correlations (< .4). Hence, the independent variables are not highly correlated with each other. TABLE 8 – Correlation matrix *** Significant at .01 level ** Significant at .05 level * Significant at .10 level Moreover the multicollinearity was examined. In model 1 multicollinearity was found since VIF >4 and tolerance <0.25, as the table below specifies. The VIF’s are high in this model because of the interaction term and higher-order term
(the squared predictor; hedonic values 2). These terms are correlated with the main effects as they include main effect terms. The second regression model also showed high multicollinearity due to the interaction term. These circumstances of multicollinearity will have no unfavorable consequences according (Allision, 1999, p. 141; Wooldridge, 2013, p. 97). Moreover, another measure to test for multicollinearity is the condition index. A condition number above 50 indicates significant multicollinearity (Belsley, Kuh an Wlesch, 1980). The condition number in the two models, 32 (model 1), 23 (model 2), was high but showed no severe multicollinearity. Yet, although the multicollinearity can be explained or negated, we should agree for some degree of collinearity and therefore we should urge some caution when interpreting the results.
TABLE 9 – Multicollinearity test
Model 1 (Hedonic-model) Model 2 (Frugality-model)
Variables Tolerance VIF Variables Tolerance VIF
of statistical power (Sheskin, 2007). Taken into account our small dataset and the fact that the data is legitimate, it was decided that removing outliers was arbitrary. Above that, after log-transforming food waste (gr) no outliers could be detected anymore in the dependent variable. FIGURE 6 – Outliers
Frugality Hedonic values Food waste (gr) 4.5 Model 1: Interaction effect of hedonic values
Based on the literature we wanted to examine whether the level of hedonic values changes the effectiveness of the storage management intervention on amount of food waste. This moderating relationship was examined by running a multiple regression analysis with the independent variables intervention and hedonic values and the dependent variable food waste (gr), controlling for household size. The higher-order variable of hedonic values was added to the model, as explained in paragraph 4.4.3. The output of this three-step procedure regression analysis can be found in table 10 (model 1: A – C). The final overall model (model 1C) was significant (F(5,22)= 4,545, p = .005) with a total
explained variance of 39,6% (adjusted R2 = .396.) In experiments trying to predict human behaviour, the R-square is often relatively low. Hence we can conclude that the model has a good fit and the data is reliable.
the amount of food waste (β-.35, p > .10), after controlling for household size. This confirms, as already stated before in 4.3, that the results are not in line with hypothesis H1; we did not find supporting evidence that the storage management intervention reduces consumer’s food waste. This means that the storage management intervention appears to have no influence on people’s food disposal patterns and therefore cannot be assumed to be an effective tool in reducing food waste.
TABLE 10 – The moderating role of hedonic values Dependent variable: Food waste (gr)
Model 1 Variables Beta t-stat. Sig. VIF
expected to dispose less food, until a certain point, where consumers with even higher hedonic values are expected to dispose more food (when controlling for household size).
Lastly, in model 1C all predictors were entered, including the interaction terms. Results showed that the interaction effect of hedonic values is not statistically significant (β-.07, p > .10). Moreover, the explained variance decreased while adding the interaction term, indicating that hedonic values do not change the effectiveness of the intervention on food waste. These results are therefore not in line with hypothesis H2, we did not find supporting evidence that a higher level of hedonic values negatively affects the effectiveness of the storage management intervention. Hence, hypothesis H2 is rejected. This means that ones’ high level of hedonic values does not change the effect of the storage management intervention with the aim to reduce food waste.
4.5.1 Regression analysis hedonic values on food waste
As was also established in paragraph 4.4.3 (linearity), a u-shape effect was found for hedonic values and food waste. A regression analysis was conducted with hedonic values regressed on food waste (gr) while controlling for household size, to examine the effect of hedonic values on food waste. The overall model is significant, see table 11. The results confirmed that hedonic values negatively (β1 = -9.11, p<.05) and the higher-order variable hedonic values positively (β2= .573, p <.05) were significantly affecting food waste (gr). Hence, as mentioned above, this means that the higher the hedonic values the lower the food waste, until a certain point where higher hedonic values lead to higher amounts of food waste (figure 7). This u-shaped effect is an interesting finding as literature merely assumes a negative effect of hedonic values on food waste.
TABLE 11 – Regression analysis hedonic values Dependent variable: Food waste (gr)
Unstandardized Coefficients t-statistic Sig.
FIGURE 8 – Estimated means plot hedonic values on food waste in grams (ln) 4.6 Model 2: Interaction effect of level of frugality
A second regression analysis was performed to test whether the degree of frugality changes the effectiveness of the storage management on the amount of food waste. This regression analysis was run with the independent variables intervention and frugality and the dependent variable food waste (gr), controlling for household size. The results can be found in table 12 (model 2: A-C). The final overall model (model 2C) was significant with a 10% alpha level (F = 2.631, p < .10) with a total explained variance of 19,5% (adjusted R2 = .195). Hence we can conclude that the data is reliable. Only household size remained significant in all models.
Finally, in model 2C all predictors were added, incorporating the interaction term frugality on intervention. The results showed that the interaction effect did not statistically significant (β-.87, p > .10) affect the amount of food waste. Moreover, by adding the interaction term to the model, the significance of the overall model decreased as well as the explained variance. Hence, it can be concluded that the degree of frugality has no influence on the effectiveness of the intervention. Consequently, this is not in line with hypothesis H3, we did not find supporting evidence that a higher level of frugality increased the effectiveness of the storage management intervention. Hence hypothesis H3 rejected. This means that the effectiveness of the storage management intervention with the aim to reduce food waste does not depend on ones frugal lifestyle.
TABLE 12 – The moderating role of frugality Dependent variable: Food waste (gr)
Model 2 Variables Beta t-stat. Sig. VIF
A Intervention -.352 -0.0871 .392 1.006 Household size .593 3.114 .005 1.006 F = 5.046, p = .014, Adjusted R2 = .231, N = 28 B Intervention -.319 -.782 .442 1.016 Household size .593 2.615 .015 1.145 Frugality -.171 -0.819 .421 1.143 F = 3.544, p = .030, Adjusted R2 = .220, N = 28 C Intervention -2.068 -0.566 .577 78.573 Household size .627 2.600 .016 1.254 Frugality -.196 -0.899 .378 1.213 Intervention * Frugality .873 0.482 .634 78.641 F = 2.631, p = .061, Adjusted R2 = .195, N = 28 4.7 Robustness checks
tests did not appear to be significant, thus no reliable conclusions can be drawn from these tests. An overview of the results can be found in appendix C. Yet, regarding the hedonic-model it appears that hedonic values had a direct influence on food waste (ml) as well. Considering the frugality-model the results turned out to be insignificant as well.
Secondly, additional tests were performed considering the additional food waste variable. Robustness of the two regression models was assessed using the other dependent food waste (ml) variable. The data and models cannot meet all the assumptions necessary for regression analysis. Furthermore the overall model of hedonic values was insignificant (adjusted R2 = 0, F(5,12)= .352, p > .10) and the frugality-model appeared to insignificant as well (adjusted R2 = 0, F(4,13)= .673, p > .10). Therefore the results are unreliable and no conclusions can be drawn from these tests. Hence, neither the level of hedonic values, nor the degree of frugality changes the effectiveness of the intervention on liquid food waste.
4.8 Control variables
were not supported in this study, since these were found to have no statistically insignificant influence on food waste. Hence these control variables were not included in the model.
4.9 Evaluation of the intervention
5. DISCUSSION
In this chapter the research questions and hypotheses will be answered based on the findings of this research and literature. After that, recommendations are given for managerial implications. Lastly, the limitations of this study will be elaborated on and suggestions for future research are made.
5.1 Overview of Results
5.2 The effect of a storage management intervention on food waste
The reviewed literature suggested that interventions can lead to less (food) wastage (Abrahamse, 2007; Bernstad, La Cour Jansen and Aspegren, 2013); Graham-Rowe∗, Jessop, Sparks 2013). Results of this research are not in line with these findings. The storage management intervention was expected to reduce food waste by helping people to better manage their food storage, as Porpino, et al., 2015 recommend that educating people to store food properly is needed for reducing food waste. However, as revealed by the results the intervention was not effective in reducing the amount of food waste in households within this research. Yet, the average amount of food waste between the groups did seem to be higher for the control group compared to the intervention group. One explanation for the insignificant result could be the small sample. Small sample sizes frequently lead to non-significant results while with a larger sample effects could have been discovered (Hoyle, 1999). Subsequent research with a larger sample might lead to better results.
(Abrahamse et al., 2005). Or they could not be motivated to reduce their food waste behaviour and were not encouraged to use the tool (Bolderdijk, et al, 2013). Moreover, the tool might have been not needed for certain individuals, as they already dispose no food. Some participants mentioned the latter as reason for not using the intervention. 5.3 The role of hedonic values on intervention effectiveness Considering the literature review we expected that hedonic values would reduce the effectiveness of the storage management intervention. The results of this study did not support this expectation. Unlike the expected, the interaction effect of hedonic values on the effectiveness of the intervention was found to be not significant. This is contrary to findings of Steg et al. (2014) who state that hedonic values may impede behavioral change in interventions. This also denies the applied theory of Bolderdijk et al. (2013) that declared that values determine the effectiveness of an intervention if the aim is not in line with the individual’s values, as was the case for hedonic values. Taken into account the relatively high level of hedonic values within this sample it could be that the intervention was perceived as too practical or functional, which might be an explanation for the low responsiveness to the stimuli (Wang et al., 2000). Moreover, people might not be motivated to use the intervention because it was perceived as too functional, too time consuming or not convenient (Steg and Vlek, 2009).
counterargument for this result could be that other factors that were not controlled for have affected this pattern. As described in the literature review, people might also engage in pro-environmental behaviour (e.g. reducing food waste) for hedonic reasons because it brings them enjoyment (Bolderdijk et al. 2013). Or the amount of food waste may have been situation dependent (De Groot et al., 2013).
5.4 The role of frugality on intervention effectiveness
People pursuing a frugal lifestyle were expected to be motivated to use the tool, as their values (e.g. avoid waste) are in line with the aim of the intervention (reduce waste) (Bolderdijk, 2013). Results of this study cannot support these findings. Frugality did not increase the effectiveness of the storage management intervention and did not lead to less food waste. However to some degree it was expected that frugality might not have any effect, as indicated before. One explanation for this could be that frugal consumers already have minimized their food waste (Evans, 2011). In that case the intervention would logically be unnecessary, which clarifies why some participants did not used the tool. In total six respondents have indicated to have not waste any food during the week. Furthermore, frugal consumers might not have been responsive to the intervention because they did not want to rely on an external source, such as the intervention (Goldsmith, Reinecke & Clark, 2014).
5.5 Control variables
The socio-demographic control variables gender, work hours, income and education this research had no effect on food waste. Age and household size did have a significant effect of food waste. Aschemann-Witzel et al. (2015) propose that social-demographics are less explanatory for explaining household food waste in comparison to psychographic variables.
5.6 Managerial implications
