Students’ Intention to Reduce Food Waste

Students’ Intention to Reduce Food Waste

An approach with an extended version of the Theory of

Planned Behavior

Authors: Eline Wajon and Johanna Richter

Course: Master Thesis, 15 credits, Uppsala University

Supervisor: Matilda Dahl (Ph.D.)

Abstract

The aim of this research is to develop the understanding of food waste behavior by analyzing student’s intention to reduce household food waste. The determinants Attitude, Subjective Norm and Perceived Behavioral Control (scope of Theory of Planned Behavior), as well as the Anticipated Emotions were therefore investigated. Data from a sample of 209 students at Uppsala University, Campus Gotland (Sweden) were collected with a web-based survey and used to identify the relevant factors. A multiple linear regression analysis showed that Attitude and Perceived Behavioral Control has a significant positive relation to the students’ Intention to reduce food waste. Subjective Norm and Anticipated Emotions did not reach statistical significance and could therefore not be used to draw conclusions. As a limitation, it must be considered that the focus is purely on the intention and the actual behavior was not part of the research. In addition, a missing universal definition of food waste leaves space for interpretation. What food (parts) is seen as edible depends on individual perception. People have different perceptions of what is edible. The findings of the research are helpful to recommend strategies on how to increase the intention to waste less food. Therefore it contributes to address the global issue of food waste. It outlines the factors that appear to drive the largest change in altering the intention to reduce food wastage.

Acknowledgments

First of all, we would like to thank our supervisor Matilda Dahl (Ph.D.). We are grateful for all the support and input you gave us. You made us believe in our topic and evoke the potential of this thesis with all your strength and advice. Thank you for pushing us when needed and also for always encouraging us to believe in ourselves. Moreover, we appreciate all the assistance we got from other lecturers (faculty members). You always had the time to give us guidance, when we struggled with specific issues.

A big thank you goes to our classmates. Due to your valuable feedback and suggestions, our study could develop to what it is today. At the same time, we appreciate your effort to make this study as pleasant as it was.

Table of content

2.1 Food Waste and Food Supply Chain 9

2.2 Household Food Waste 10

2.2.1 Classification of food waste 10

2.2.2 Origin of food waste 11

2.2.2.1 Behavior 11

2.2.2.2 Food Industry 12

2.2.2 Impact of food waste 13

2.2.3 Sustainable consumption 14

2.2.3 Situation in the EU 14

2.2.4 Characteristics of Students 15

3 Theoretical Framework: Theory of Planned Behavior 17

3.1 Original TPB 17

3.1.1 Overview 17

3.1.2 Criticism 18

3.2 Conceptual Research Model 19

3.2.1 Extending the TPB 19

3.2.2 Hypothesis 21

3.2.2.1 Attitude 21

3.2.2.2 Subjective norms 22

3.2.2.3 Perceived Behavioral Control 22

4.3.1 Overall 27

4.3.2 Intention 28

4.3.3 Attitude 28

4.3.4 Subjective norm 28

4.3.5 Perceived Behavioral Control 29

4.3.6 Anticipated Emotions 29 4.3.7 Demographics 30 4.3.8 Pilot 30 4.4 Data analysis 31 4.4.1 Procedure 31 4.4.2 Reliability 32 4.4.3 Validity 34 4.5 Ethical aspects 34 5. Empirical findings 35 5.1 Descriptive statistics 35 5.2 Correlations 37 5.3 Hypotheses tests 38 6. Discussion 41 6.1 Implications 41 6.2 Limitations 43 6.3 Further research 43 7. Conclusion 44 References 45

Appendix I - Item overview 51

Appendix II - Survey 53

Appendix III - Reliability Tests 57

Appendix IV - Validity Tests 59

Appendix V - Descriptive statistics outcomes 61

Appendix VI - Pearson Correlation Analysis 67

List of abbreviations

ATT Attitude

AE Anticipated emotions

CATA Check all that apply

EU European Union

FAO Food and Agriculture Organization

FSC Food supply chain

GHG Greenhouse gases

INT Intention

PBC Perceived behavioral control

REV Reversed item

SD Standard deviation

SDG Sustainable Development Goals

SMS Short Message Service

SN Subjective norm

SPSS Statistical Package for the Social Sciences

TPB Theory of Planned Behavior

UK United Kingdom

UN United Nation

USD United Stated Dollar

VIF Variance inflation factor

List of tables

Table 1 Cronbach’s Alpha per construct Table 2 Descriptive statistics per construct Table 3 Regression Analysis - Model 1 & 2

List of figures

Figure 1 Food supply chain and the key sources of waste creation per tier Figure 2 Sources and disposal routes of household food and drink Figure 3 Theory of Planned Behavior by Ajzen

Figure 4 Conceptual model of determinants on food waste behavior Figure 5 Living situation

Figure 6 Environmental concern Figure 7 Nationalities

1 Introduction

1.1 Motive

What is the first thing that comes to your mind when thinking about food? Is it a necessity, something that is always there, a daily treat or a struggle to get? In developed countries, enough food is available for everyone. It is even estimated that one-third of all food produced is not being consumed (Parfitt et al., 2010). Today it is common knowledge, that wasting food is bad. But it happens anyway. Society seems to have lost the meaning and importance of food because it is always available in any shape and form and at any time of the day. It is taken for granted, that stores are always fully stocked with a large variety of food.

Especially compared to developing countries, food in general is not highly valued anymore. In some parts of the world, people struggle to survive from one day to the other. If there are regrets about throwing food away, it is often because of the money we spend on it (Stancu et al., 2015). There is no way to transport it to the third world anyway. With the growing population, the problem of food maldistribution becomes more pressing. While people are starving, most homes in developed countries are full of food. Food that sometimes just sits and goes bad. This abundance of food developed over time with the economic welfare and the increasing expectations by consumers (Parfitt et al., 2010).

Food waste behavior is determined by various factors. Different consumer/household practices and routines must be considered since their decisions regarding planning, shopping, food preparation, and eating have consequences for the amount of food waste generated. The decisions made are interrelated. In an ideal situation, a household is able to balance the amount of food that is purchased and that is consumed (Stefan et al., 2013).

1.2 Problematization

What is actually referred to when one talks about food waste or loss? In general, food waste or food loss is defined as ‘food that was originally meant to human consumption but which fortuity gets out the human food chain is considered as food loss or waste even if it is then directed to a non-food use (feed, bioenergy…)’ (Gustavsson et al., 2011, p. 2). Food loss happens at the production, postharvest and processing stage, which leads to a decrease in edible food mass throughout these three stages of the food supply chain (FSC). When one discusses food waste, it generally refers to thrown away food at the end of the chain, e.g. in supermarkets or in households (Parfitt et al., 2010). This study focuses on this last stage of the FSC and in particular on the so-called ‘household food waste’.

Every year, a third of the food that is produced for human consumption is lost or wasted. Economically speaking this amounts to an estimated cost of USD750 billion (FAO, 2018) and result in about 1.3 billion tons of generated waste (Gustavsson et al., 2011). The industrialized world produces per-capita much more food than the developing countries. Gustavsson et al. (2011) estimate a 95-115 kg/year food waste by consumers in North-America and Europe, whereas South-Southeast and sub-Saharan Africa only waste 6-11kg/year per capita. In general, the amount of food waste per capita is higher in developed countries than in developing countries (FAO, 2018). It is calculated that in high-income countries households are generating more than 50% of all food waste (Stancu et al., 2015; Kowalewska & Kollajtis-Dolowy, 2018).

Food waste has an impact on all three pillars of sustainability; economic, social and environmental. Firstly, food waste has an economic impact, since it is costly to produce food, that becomes waste without being consumed. It is purchased without fulfilling its purpose. Energy consumption contributes not only to the production and transportation cost of food, but it can also be looked at from an environmental perspective based on the emissions (Girotto et al., 2015). It is estimated that in the EU 22% of all greenhouse gases (GHG) are emitted by the food industry (Winker & Aschemann, 2017). Another environmental consequence of food crops, especially monocultures1_{, is the severe impact on}

biodiversity. Deforestation for agricultural purposes is also facilitating climate change. These are only selected effects on the environment, but the spectrum is much wider (Thyberg & Tonjes, 2015). Lastly, food waste is an important social issue when considering food security and poverty. While some have an abundance of food, others struggle with starvation. Also, the consequences of climate change can bring different conflicts to society (Stancu et al., 2015).

Scholars see the prevention of food waste as the best way to deal with the problem. This way the resources put into its production, processing, distributing and preparation cannot only be saved but also

greenhouse gases that emerge from the decomposition of food can be avoided (Quested et al., 2011; Stancu et al., 2015; Thyberg & Tonjes, 2015).

1.3 Research question

Studies reveal that there are gaps in research and data available with regards to consumer behavior towards global food waste (Stancu et al., 2015). The goal is to contribute to this field and provide a better understanding of how the intention to reduce food waste is formed. What factors are playing a role when it comes to this intention? This study is conducted from the perspective of developed countries with an emphasis on Sweden and puts a focus on a new generation of food providers in the form of students. Based on four hypotheses the connection between the intention to reduce food waste and the attitude, the social pressure, possible anticipated emotions and the confidence of ability will be analyzed. This research adds to the existing literature on human behavior towards food waste and its implications for students when it comes to different factors that influence their intention to throw away food. Through a quantitative study, based on a conceptual framework derived from literature, the study aims to answer the following question:

What factors influence the intention to reduce food waste for students? An approach with an extended version of the Theory of Planned Behavior

1.4. Outline

2 Food Waste

2.1 Food Waste and Food Supply Chain

Food waste is a much-debated definition in literature and researchers agree that a universal definition of food waste is missing (Parfitt et al., 2010; Stancu et al., 2015; Ponis et al., 2017; Girotto et al., 2015; Delsignore et al., 2017). The most commonly used definition is provided by the Food and Agriculture Organization (FOA). It refers to food waste as “the wholesome edible material intended for human consumption, arising at any point in the Food Supply Chain (FSC) that is instead discarded, lost, degraded or consumed by pests” (FOA, 1981). Stuart (2009) extends this definition by “including edible material that is intentionally fed to animals or is a by-product of food processing diverted away from the human food”. In 1945 the FAO was established with the aim to reduce food losses. The main focus was to decrease food losses in the early stages of the FSC. The first World Food Conference was held in 1974 and aimed to address world hunger, where an emphasis was put on post-harvest losses as the main food waste indicator. By 1990 it was recognized that the focus was solely on the technical part and they began to develop a more holistic approach to tackle the issue (Parfitt et al., 2010).

The distinction between food waste and food loss is explained by the particular points of occurrence in the FSC. One refers to ‘food loss’ when talking about post-harvest waste, it decreases the quantity or quality and therefore gets unfit for human consumption. It usually relates to systems in infrastructure that require investment. The term food waste generally applies when it occurs at consumer or retail stages. This type of waste generally relates to human behavioral issues. (Parfitt et al., 2010). In this paper, the term food waste will be used to discuss ‘Post-consumer losses’, which refers to food waste caused by operations and activities at the stage of consumption.

Food differs from other waste products by being biodegradable and having diverse nutritional value. This helps to integrate the waste again into the ecological cycle and have the possibility to create new value out of the left value (Parfitt et al., 2010). Food waste can be composed and used as fertilizer to grow new plants. However, much of the food waste enters landfills together with other waste instead. This applies especially to industrialized countries. There the waste decomposes over a long time releasing GHG like methane (Montoneri, 2018). Furthermore, it needs to be considered that nowadays the FSC can spread over the whole globe. Due to globalization, food parts can be produced, modified, processed and consumed in different parts of the world with a complex transportation system between these steps (Govindan, 2017).

and 6 are referred to as food waste stages. This research will focus on tier 6 “Food Preparation, Consumption & Disposal” and therefore concentrates on human behavior with regards to food waste.

Figure 1. Food supply chain and the key sources of waste creation per tier (based on Ponis et al., 2017).

2.2 Household Food Waste

2.2.1 Classification of food waste

This research will analyze household food waste in particular. Household food waste is the amount of waste that is generated from the domestic consumption of food and drinks, which also comprises home-grown products and takeaways. However, the consumption of food and drinks outside of the home is not included (Parfitt et al., 2010).

Household food waste (or kitchen waste) can be divided into three different categories: avoidable, possibly avoidable and unavoidable waste. The avoidable and possibly avoidable parts are also referred to as ‘edible waste’. Avoidable waste is “food and drinks thrown that was, at some point prior to disposal, edible in the vast majority of situations” (Parfitt et al., 2010, p. 3073). Possibly avoidable waste is “food and drinks that some people eat and others do not (e.g. bread crusts), or that can be eaten when food is prepared in one way but not in another (e.g. potato skins)” (Ibid., p. 3073). Unavoidable waste is “arising from food preparation that is not, and has not been, edible under normal circumstances” (Ibid., p. 3073). Examples of this type include eggshells and bones. But even with this common definition, the personal conception of ‘edible’ and different cultures play an important role when categorizing the type of food waste (Ibid., 2010).

disposed of in different ways. For example, the waste goes into the sewer, is part of the household waste or food waste collections or ends up in home composting or as animal food (Ibid., 2010).

Figure 2. Sources and disposal routes of household food and drink (based on Parfitt et al., 2010).

The amount of food waste is affected by the size of a household. It has been detected, that smaller households generate more waste per person than larger households. Also, the presence of children and their age are influencers. The same applies to the demographics, the culture and possibly the income of the household. To calculate the exact amount of food waste in households is difficult. Many studies have been conducted, but with different approaches. According to Parfitt et al. (2010), some calculated the percentage based on consumed calories, others in weight of consumed food. Some studies did not consider sink disposals. This makes it very difficult to relate them to each other. To determine the individual amount of food waste is especially difficult since a lot of waste is created ‘on-the-go’ by eating or drinking in restaurants, at work or other places (Parfitt et al., 2010).

2.2.2 Origin of food waste 2.2.2.1 Behavior

Food waste is generated in many different ways and is not based on a single behavior, but results from multiple behaviors. This is plausible since numerous factors have an impact on food waste behavior: planning, shopping, storage, preparation and consumption of food. These steps represent different possibilities to avoid food waste. Once food is thrown away that opportunity has passed. The activity causing the wastage might have occurred already several days earlier (Quested et al., 2013). The disposal of food is only the last step in the process of “a series of food-related behaviors” (Stefan et al., 2013, p.375).

al., 2013). A study conducted by Ponis et al. (2017) in Greece demonstrated that consumers with thoughtless shopping behavior produce more food waste. In addition, people cooking their own food tend to throw away less than others. They also apply better food management strategies. On the one hand, low-income households make better use of their resources and therefore the household food waste generated can be less. But on the other hand, they also tend to cook more from scratch ending up with more waste (Ponis et al., 2017). This makes it difficult to say whether the income of a household has a clear relation to wasteful behavior.

Moreover, Girotto et al. (2015) point out that there is also an emotional and psychological aspect when it comes to reducing food waste. There is a desire to have (healthy) food constantly available at home to meet the needs. This results in overstocking and consequently in a food surplus.

In developed countries the gap between the consumer and the production of food increases. Usually, supermarkets connect producer/processor and consumer. The food price is an essential factor for customers when it comes to decision making (Parfitt et al., 2010). The food industry is dynamic and adapts to changing customer demands. The customers shape the way they are buying, transporting, conserving, preparing and consuming food (Govindan, 2017). Consequently, they have power over the FSC and can pressure the other stakeholders towards more sustainable practices. More consumers support green labels for food products. Therefore transparency along the entire FSC is necessary to gain the consumer’s trust towards the end product (Ibid.). During the last years, society started to recognize the impact of their food waste (Ponis et al., 2017). An important aspect, contributing to this change, is spreading information regarding the effect of food waste among consumers. Only if they are aware of the impacts of their consumption, they are able to make more sustainable choices (Staniškis, 2012). Governmental organizations and media can impact the preferences of customers and thereby contribute to their decision for a more sustainable lifestyle (Govindan, 2017).

2.2.2.2 Food Industry

In industrialized FSCs, food is centrally processed. The factories generate waste for multiple households and can make better use of the resources. Consumers tend to cook less from scratch at home and therefore processed food has arguably a positive effect of household food waste (Parfitt et al., 2010). Nevertheless, households still throw away a lot of food. As a result, the benefit of processing food centralized is being offset (Ibid.).

over time. Food is leftover because the amount prepared was too large or it has not been used in time (Ponis et al., 2017). Package sizes play also an important role to reduce food waste. They should be suitable for household needs. Preparing too much food causes not only obesity but can also result in food waste. Also, the consumption of more delicate products with shorter shelf-life is associated with an increasing amount of food waste. Perishable goods make up for a large part of food waste. In contrast, better technologies, such as cookers, smart fridges, and meal planning tools can have a positive effect on the reduction of food waste (Parfitt et al., 2010). They support food management in a sustainable manner and can reduce food waste.

2.2.2 Impact of food waste

Globally there has been an increased focus on this issue. In 2013, the Sustainable Development Goals (SDG) were established by the United Nations (UN). Food waste is intertwined in multiple goals, however, goal 12 ‘responsible production and consumption’ is mainly focused on this issue. It connects the environmental problems (e.g. land degradation, gas emissions), social problems (e.g. hunger) and economic problems (e.g. high costs) to targets that should be reached globally by 2030 (UN, 2013). With the rapid depletion of earth’s resources, pressure on global economies and political and social instability in many countries, one can see an increase in war zones and economic inequality on the one hand. On the other hand, consumption volumes and world population are constantly increasing (Ponis et al., 2017). Nixon (2015) estimates this will lead to USD600 billion of costs by 2030, which amounts to 50% in the global cost for food waste.

Food waste impacts all three pillars of sustainability. From an economic perspective, it is costly to waste food with an impact on the consumer's income (Girotto et al., 2015; Stancu et al., 2015). With the growing population on earth, the demand for food increases simultaneously. Wasting food affects food security negatively and therefore creates a shortage of food (Stancu et al., 2015; Thyberg & Tonjes, 2015). Along the whole FSC, which covers stages like production, processing, transportation, storage, distribution, and marketing, food makes an environmental impact (Girotto et al., 2015). The production of food requires water, cropland and fossil fuels (Stancu et al., 2015). In addition, fertilizers can create eutrophication, diversity suffers, and other deforestation and agricultural practices; the environment gets polluted in different ways (Thyberg & Tonjes, 2015). According to Delsignore et al. (2017), the single largest type of waste ending up in landfills is food waste. From there GHG like methane are released during the process of composing, which accelerates climate change (Delsignore et al., 2017; Girotto et al., 2015; Stancu et al., 2015; Clark & Manning, 2017).

food waste is seen as the best solution, especially at the stage of consumption a great potential for savings is present. That way all the resources such as energy and water, that are needed for processes like growing, harvesting, transporting, processing and selling the food can be saved together with the emissions arising of storing and cooking (Stancu et al., 2015; Thyberg & Tonjes, 2015; Quested et al., 2011). According to Quested et al. (2013), the prevention of food waste can decrease GHG by about eight times more compared to letting it digest anaerobically in a landfill. Even if not all food waste can be prevented, they add that it is, in any case, better to let the food waste compost at home or use council food waste collection services compared to using landfills.

2.2.3 Sustainable consumption

The food sector has an influence on sustainable development. A common understanding of sustainability arose from the Brundtland Commission by being defined as a “development that meets the needs of the present without compromising the ability of future generations to meet their own needs” (Geissdoerfer et al., 2017, p. 758). Consequently, sustainable consumption of food aims to minimize the economic, social and environmental impact and ensures a long-lasting and prospering existence of natural resources. So the need of future generations is not endangered. Sustainable consumption is based on a decision-making process. It combines the consumers’ social responsibility with their individual needs (Vermeir & Verbeke, 2007).

However, the present consumption behavior is leading to the degradation of resources and unsustainable production (Govindan, 2017; Longo et al., 2017). Therefore a change in behavior is crucial. This can be achieved by either reducing the current level of consumption or by modifying the consumption procedures. Reducing means scaling-down the number of products that are consumed. In contrast, alternative ways of consumption take also the environmental and social influence of the purchase into consideration. E.g. sourcing from the local market is an option for the latter. These two changes in behavior can be applied independently or interdependently (Longo et al., 2017).

2.2.3 Situation in the EU

After the second world war, the FOA of the United Nations directed the reduction of food losses. Over time post-harvest losses could be reduced due to improving technology and infrastructure among other factors. Compared to developing countries, the food loss rate in industrialized countries is low (Parfitt et al., 2010).

Generally, industrialized countries are characterized by an aging population and an increase in single person households (Parfitt et al., 2010). Therefore, different preconditions have to be considered when dealing with food waste compared to developing countries.

The European Commission is implementing actions to lower the amount of food waste generated. By doing so, not only money can be saved, but it also is an opportunity to reduce the environmental consequences of food production and consumption, and work on the social impacts (Graham-Rowe et al., 2014; Mondéjar-Jiménez et al., 2016). Hereby one aspect is the planned transformation towards a circular economy making the economy in Europe more resource efficient. Practical steps are central to cut the amount of food waste by 2025 in half (Mondéjar-Jiménez et al., 2016). Preventing food waste is central in the EU’s Circular Economy Package and public authorities aim to increase the awareness for food management (Ponis et al., 2017).

In the EU a good infrastructure and advanced technologies are available to minimize the amount of food loss. Still, at the end of the FSC, consumers waste a lot of food. Most of it is considered to be avoidable waste. By throwing away food at this stage, the use of resources in the previous stages becomes useless together with the value of the food wasted (Stancu et al., 2015; Ponis et al., 2017). There are many initiatives in Europe that try to minimize waste and/or create new value out of food loss and food waste. Grassroots innovations2 _{can raise awareness and launch change (Tartiu & Morone, 2017). Despite the}

efforts, consumers still engage too little in this problem. There is a discrepancy between the actual practices and the belief in sustainability if present at all (Longo et al., 2017). Despite all efforts put into lowering food waste, it is estimated that household food waste in the EU amounts to about 42% of all food produced, whereby two-thirds of this amount is assessed to be avoidable or possibly avoidable (Mondéjar-Jiménez et al., 2016). Winkler & Aschemann (2017) add that the production of food accounts for about 9.4-14.5% of total GHG emissions and assess that “between 89 and 178.3 million tons of food waste accumulate each year in the EU, which will generate roughly 70-170 Mt CO2eq/year of emissions” (p. 47).

2.2.4 Characteristics of Students

In the EU the 15-24 year-olds form a group that tends to generate a large amount of food waste. It is estimated that more than 6% of their food purchased per week is wasted (Mondéjar-Jiménez et al., 2016). Consequently, they have the greatest potential to reduce their personal level of food waste and thereby contribute to the overall amount. Educating them at pre-consumption stage (e.g shopping behavior) and at post-consumption stage (e.g. handling of leftovers) can contribute to improving their food management skills. Next to this socio-demographic factor also other factors like attitudes,

intentions, household habits and the context of purchase have an influence on food waste (Quested et al., 2013). In addition, the younger generation is also considered to be less educated when it comes to food management (Clark & Manning, 2017).

In their study regarding food waste among young people in Italy, Principato et al. (2015) found a general awareness about the problem of food waste among their sample. However, their concern was mainly about the economic consequences than about environmental impact. This can be linked to the assumption that students cannot afford to waste food and are more aware of its monetary value (Graham-Rowe et al., 2014). Particularly young consumers demand a rising amount of convenience products to manage their time and work more efficiently. At the same time, more consumers buy ethical or sustainable products like organic, animal-friendly, environmentally friendly, or locally produced food (Vermeir & Verbeke, 2007).

When looking particularly at students, one has to consider that they often live in short-term rented accommodations and have to adapt to the present conditions. Many live in shared accommodation with other students. Therefore a special living situation applies to them. Their food waste behavior is influenced by a number of factors, such as being poor, deficient planning, lacking storing and cooking skills, impulse purchases and not eating what needs to be eaten first (Clark & Manning, 2017).

It can be expected from higher educated people to have a better knowledge of the concept of sustainability and being aware of the basic environmental issues. Students form the next generation of consumers and are able to make a change within the next decades. They are likely to take their beliefs, values, and habits into older age (Kim et al., 2013; Vermeir & Verbeke, 2007).

3 Theoretical Framework: Theory of Planned Behavior

3.1 Original TPB

3.1.1 Overview

Since decreasing the amount of generated waste is viewed as the best solution, the intention to reduce waste is at the center of attention in this study. One model that is concerned with behavioral intention is the Theory of Planned Behavior (TPB), which was developed by Icek Ajzen in 1991. This theory is a conceptual extension of the Theory of Reasoned Action by Fishbein & Ajzen (1975) which did not incorporate the so-called ‘Perceived Behavioral Control’ in the model. With this addition, it is believed to be valuable for predicting human behavior (Tommasetti et al., 2018).

Intention to perform a certain behavior is the central factor in the theory, which entails the motivational factors that influence behavior. It assumes that the stronger the intention towards a behavior, the more likely it is performed. This can only be expected when the behavior investigated is under volitional control, meaning that a certain behavior is carried out purposefully and under a person’s will. Prior to the behavior a cognitive decision to act was made. Or in other words, the basic rule is that the stronger the intention is, the more likely is it that certain behavior is carried out (Vermeir & Verbeke, 2007). The TPB suggests that intention is determined by three conceptually independent predictors: Attitude, Subjective Norm, and Perceived Behavioral Control (PBC). The Attitude describes whether a person is in favor of carrying out a certain behavior. The Subjective Norm defines to what extent a person feels social pressure to perform the behavior. Both address the motivation for a certain behavior, while Perceived Behavioral Control is concerned with the ability and opportunity of an individual. In other words, whether the person feels in control over the activity and is confident in his or her ability to perform it (Vabø & Hansen, 2016). Overall, the performance of a behavior is a combination of perceived behavior control and intentions (Ajzen, 1991). Figure 3 shows how all above-mentioned aspects are interrelated.

The TPB is able to combine individual, social and behavioral aspects in one model (Vabø & Hansen, 2016). Another strength of this model is its ability to detect the impact of social pressure on the intention with the variable Subjective Norm. It has been used in a wide range of research, that are both businesses related, and non-business related. This includes job performance, political behavior. motivation to exercise, illegal activities, participating in recycling programs, consumption of halal food among others (Londono et al., 2017). In addition, the theory provides the opportunity to investigate the impact of personal determinants, the social environment and non-volitional factors on intention (Kim et al., 2013). The wide range of researches conducted with the TPB demonstrates that is it easily adaptable and flexible for studying additional determinants that are not included in the TPB (Mondéjar-Jiménez et al., 2016).

The conceptual framework is seen as useful for dealing with the complexity in social human behavior and can be used as a starting point to change it. With a complex topic like sustainability, which is highly volatile for human behavior and social issues, the theory seems to be a good choice to understand the decision making processes in this field. Ajzen (1991) explains that attitude towards behavior, perception of behavior control, the subjective norm and intention, all expose a different part of behavior.

3.1.2 Criticism

As any widely used and recognized theory, the TPB has received some criticism over the past years. It arguably has insufficient variables and poor predictive efficacy to undertake a given behavior. Sheeran et al. (2013) found that unconscious influences on behavior are not taken into consideration and the focus lies purely on rational reasoning. Russell et al. (2017) agree by stating “the TPB rests largely on the assumption that individuals make rational and reasoned choices” (p. 108). De Pelsmaeker et al. (2017) elaborate on this in more detail. The model is only considering rational aspects when it comes to decision making. It excludes unconscious and non-rational (affective processes) factors like emotions, that also play a role in the decision-making process. Differences in the situational context are overlooked. Particularly in situations where the behavior is connected to a negative outcome, anticipated effective processes can have a significant influence on intention and the behavior itself (De Pelsmaeker et al., 2017).

3.2 Conceptual Research Model

3.2.1

Extending the TPB

When investigating consumer decision making in the fields of sustainability and specifically minimization and recycling of waste, as well as food-related behavior, the TPB was chosen multiple times as underlying framework (Mondéjar-Jiménez et al., 2016; Tommasetti et al., 2018). Over the last years, the TPB has been used widely to predict and explain environmental behavior. Klöckner (2013) performed a meta-analysis which showed that the TPB was used as the theoretical framework in 40% of all papers that were published in the environmental psychology field. It has been successfully applied to various behaviors related to the environment, including travel mode choices, water consumption, public transport, recycling and more recently also to predict food waste behavior. The usefulness of the TPB in predicting intention has typically been supported by the findings of the studies. Moreover, as specified by the model, there is an indication that intention can underwrite the prediction of environment-related behavior (Graham-Rowe et al., 2015).

Since different behaviors have an influence on food waste, making it complex, it is challenging to identify a suitable framework. The food waste level is not only influenced by the post-consumption stage, but the process starts already at pre-consumption-stage with planning the shopping activities (Mondéjar-Jiménez et al., 2016). It is proposed by Quested et al. (2013) that a model working with food waste behavior should be contextual and include factors like the Awareness, Attitudes and Values, Habits, Subjective Norms and Perceived Behavioral Control. Since Ajzen’s TPB contains already three of the factors, it is a good model to start with (Mondéjar-Jiménez et al., 2016). However, an extended version is helpful to set the model into a specific context and incorporate the critics the original model received.

As a complex topic like sustainability, which is highly volatile for human behavior and social issues, the theory seems like an appropriate choice of exploration when it comes to food waste. Although the TPB has received strong empirical support in explaining environmentally related behaviors, particular criticism addresses the lack of emotions and habits in the model. Based on this critique and other possibly missing factors of the theory in mind, different researches such as Russell et al. (2017) and De Pelsmaeker et al. (2017) developed an extended version of the TPB.

factors (Londono et al., 2017). Also, Richard et al. (1996) agree that anticipated effects need to be investigated separately from the broad attitude. Consequently, it is logical to display them as a separate factor. In addition, incorporating emotions can raise the explanatory power of the model (Kim et al., 2013).

When discussing the TPB regarding food waste behavior, it is assumed that an individual is more likely to have a positive intention towards not wasting food, when he or she:

1. thinks favorable about not throwing food away (Attitude);

2. believes that important people in their life would approve of him/her carrying out the behavior of not wasting food (Subjective Norm);

3. is confident about the ability of not wasting food (Perceived Behavioral Control).

4. has experienced or expects negative feelings when throwing food away (Anticipated Emotions).

This positive intention is connected to a greater probability that the behavior will be performed; meaning that the person will actually waste less food. It is also believed that in some situations the Perceived Behavioral Control has a direct effect on the behavior, not influenced by any intention (Ajzen, 1991). In this context that would mean we throw food away because it is easy and we do not know what else to do with it. Even though our intention is not to waste any food.

Figure 4. Conceptual model of determinants on food waste behavior (based on De Pelsmaeker et al. (2017).

3.2.2 Hypothesis 3.2.2.1 Attitude

Ajzen (1991) explains the ‘attitude toward the behavior’ by referring to whether the behavior in question has a favorable or unfavorable evaluation by a person. Attitudes are formed based on information about the attitude object and direct experiences with it. They cannot be observed directly but are measured based on responses e.g. in the form of beliefs. Beliefs are considered cognitive responses. The stronger the attitude is the likelier it is a predictor of behavior (Vabø & Hansen, 2016). Hence, attitude comprises the behavioral beliefs of a person towards a specific activity and the outcome evaluation of these actions. Previous studies see this often as the most important factor influencing intention (De Pelsmaeker et al., 2017). However, attitudes can change over time with a change in information at the given time (Glasman & Albarracín, 2006). In addition, social factors can have an impact on the development or change of attitudes (De Pelsmaeker et al., 2017). Different levels of situational and personal factors can have an influence on the extent to which attitudes are forecast intention (Vermeir & Verbeke, 2007).

Glasman & Albarracín (2006) found that if an attitude is upheld with confidence, the behavior can be better predicted. Also, easily rememberable attitudes and such attitudes that are based on direct experience are a better predictor for behavior. People that have low confidence in their attitude are more likely to adapt/replace it. Attitudes are supposed to be firmest when the relevant information is steady over time for the person carrying out a behavior. The relationship between attitude and behavior is stronger when the individual has considered the issue carefully. All in all, the more attitude object is thought through, the larger the amount of reports or expressions of the attitude, and the more direct behavioral experience has been made, the stronger is the attitude-behavior relationship (Ibid., 2006). Their review suggests that attitudes can better direct future behavior when the consequences of an action are considered beforehand, the gathered information is properly stored, and the attitude is held with confidence. Accessible attitudes are not only more stable but are also better to forecast behavior (Glasman & Albarracín, 2006). Morris et al. (2002) add that attitude measures are mostly depended on cognitive processes and miss to incorporate feelings into the measurement. Therefore attitude display in this study the general mindset of an individual and will be discussed separately from emotions. Concerning students’ intention to reduce food waste attitude is displayed in the preference to not throw food away, especially when it is avoidable. This conviction can be formed based on the knowledge about the consequences of food waste, sustainable consumption and the influence of media. Many organizations report that too much food is wasted and aim for a change in society (Thyberg & Tonjes, 2015). Hence the following hypothesis is formed:

3.2.2.2 Subjective norms

Ajzen (1991) outlines a social factor named ‘Subjective Norm’. This refers to the perception of social pressure to accomplish a specific behavior. This determinant consists of normative beliefs and motivation to comply. The social influence can also occur unconscious (De Pelsmaeker et al., 2017). The significant others can vary in different situations. The effect of subjective norms is considered to be depending on whether the individual is characterized to be individualistic or collectivistic. Individualism entails that people view themselves different from others and pursue their personal value. In contrast, collectivism means that people see themselves connected to others and follow common goals and norms (Vabø & Hansen, 2016). Hassan et al. (2015) suggest that in collectivistic societies social norms would have a greater influence on the behavior. Consequently, culture plays a role when applying the TPB. Riemer et al. (2014) agree that Hofstede’s individualism/collectivism dimension influence the role of norms when it comes to forming an intention.

In the context of student food waste behavior, subjective norms are influenced by the living situation. If a student lives with others, his/her food waste behavior is more revealed to others than when living in solitary. If others are aware of a person’s food waste, they can exert social pressure. It can also imply that a person living with others is more exposed and perceives more pressure. Therefore, subjective norms can play a more important role. The same situation can occur when a group of students cook together and are directly involved in each other’s waste behavior. Therefore the following hypothesis is tested:

H2: Subjective norms are positively related to students’ intention to reduce food waste.

3.2.2.3 Perceived Behavioral Control

barriers are external difficulty factors. PBC is impacted by perceived consumer effectiveness (the extent to which an individual believes his efforts contribute to solving a problem) and perceived product availability (belief about how easily a product can be obtained/consumed) (Vermeir & Verbeke, 2007). Connecting this variable to the intention to reduce food waste means to investigate the confidence of an individual in his/her ability to influence the amount of food waste generated. This can involve e.g. store location and opening hours, storage opportunity in rented accommodation as external factors. In contrast, internal factors would be planning activities, purchasing behavior, and personal cooking skills. Based on this information following hypothesis is constructed:

H3: Perceived Behavioral Control of food waste is positively related to students’ intention to reduce food waste.

3.2.2.4 Anticipated Emotions

This study only addresses anticipated emotions since research suggest that they are of interest for intention to carry out a specific behavior. However, the model of De Pelsmaeker et al. (2017) also includes that emotions can have a direct impact on behavior (in the moment), but these are not considered in this study. De Pelsmaeker et al. (2017) define anticipated emotions as “emotional beliefs about an emotion” (p. 1984), that have an influence on behavior / behavioral intention. On the one hand, they depend on experienced emotions in the past when carrying out a specific behavior. On the other hand, anticipated emotions can influence as well. The term ‘anticipated’ indicates that the feelings are related to potential consequences of the behavior (Londono et al., 2017).

The TPB understands anticipated emotions as part of Attitudes (Ajzen & Sheikh, 2013). De Pelsmaeker et al. (2017) conducted a study regarding the consumption of filled chocolates based on the TPB without and also with including the concept of anticipated emotions. Their findings suggest that anticipated emotions are a meaningful contribution to the TPB because it helped to explain the structure of intention and increased its explained variance. Therefore, emotions are essential when it comes to decision making and should not be disregarded (De Pelsmaeker et al., 2017). This is in line with the extended model by Russell et al. (2017). Also, Kim et al. (2013) expanded the TPB with the anticipated emotion of regret in their study of consumers’ intention to choose eco-friendly restaurants. Similarly, they found anticipated emotions as a supportive determinate to explain behavioral intentions. This makes a stronger argument to include anticipated emotions in our research.

sorrow, or distress. It has been shown, that this emotion is perceived to have a strong connection to intention, even when categorized as an attitudinal factor (Kim et al., 2013). Also, Rivis et al. (2009) found that specific anticipated emotions such as regret are stronger connected to intentions than general anticipated emotions. Therefore a final hypothesis is built:

4 Methodology

4.1 Research philosophy

The philosophy behind the research determines how the study is conducted and on which assumptions and principles it is based. These define the perspective of the researcher and guide interpretation of data. This research focuses on the behavior in student households regarding food waste based on the philosophy of critical realism. Critical Realism suggests that the researcher can only observe sensations of reality, only a representation of the real. Our senses influence the intake of information and need to be mentally processed to make sense of it. The understanding of a sensation is dependent on beliefs and thoughts. The comprehension of the social world is based on the social structures and it is recognized that reality is under constant change. Critical realists see the necessity to study phenomena from a multi-layered perspective (individuals, groups, and organizations) (Saunders et al., 2012). This study not only looked at the internal intention/motive of students to waste less food, but it also considered other members of society as influential factors. This can be roommates, friends, and family (Subjective Norms) or stores like supermarkets with special offers (Perceived Behavioral Control).

According to the epistemology of realism the collected data is viewed as trustworthy. Critical realism does not aim to make generalizable findings. Instead, it aims to uncover a social phenomenon. However, in critical realism phenomena create sensations which can be misinterpreted (Saunders et al., 2012). Therefore, the focus is laid on explaining the findings within a specific context. This study addresses the phenomena of food waste in the specific setting of student households. The findings are analyzed within this context.

Realism considers research to be loaded with values. The researchers are guided by the world-views and therefore biased, which affects the research (Saunders et al., 2012). In this case, the researchers are part of the population and therefore have a different understanding and insights than an external researcher would have. Consequently, this study cannot be conducted totally free of personal values.

4.2 Research strategy

4.2.1 Approach

factors; Subjective Norm, Perceived Behavioral Control, Attitude, and Anticipated Emotions. The hypotheses are tested again primary data which is collected with questionnaires. Consequently, the research is of quantitative nature applying a mono-method (Saunders et al., 2012).

The study uses a cross-sectional design, also called a social survey design. It is supported by four factors due to the nature of the research, namely; the data will consist of a lot more than one case; it is collected at a single period in time; the data is quantifiable and connected with two or more variables to detect patterns (Bryman & Bell, 2011). The project entails survey research where the data will be collected through a questionnaire.

4.2.3 Sampling

The population of this study consists of all full-time students registered at Uppsala University, Campus Gotland in the spring semester 2019 who are living on the island. The University has 1100 full-time on-campus students (Uppsala University, 2019) and therefore N=1100. It aimed to include as many students as possible in this study.

A mixture of convenience sampling and snowball sampling was used as the survey was distributed among the acquaintances of the researchers as well as unknown students present at Campus and member of digital groups on Facebook and Whatsapp. They were asked to forward it among their peers. Since the researchers are part of the population, it was easy to access the target group. Because of the close connection to the target group, acquainted members of the population were more likely to become part of the sample than others due to a higher level of cooperation. Although this approach reduces the external validity (Bryman & Bell, 2011), it is accepted in order to increase the response rate in the research. To pursue this goal an additional step was reminding people about the survey (Bryman & Bell, 2011).

At the end of the collection period, a total of 224 people responded to the survey. 15 responses were unusable for the research; 4 respondents indicated not to be a student and were therefore removed from the study. After this first selection, 6 other respondents were eliminated as they indicated not to provide for their own food. Lastly, 5 additional paper-filled surveys were not usable, as they were not completely filled out when returned to the authors. Therefore, the final sample size of 209 respondents reflected a response rate of 19,3% (209/(1100 - 15)) to the questionnaire.

4.2.4 Data collection

and low cost, which is helpful to increase the response rate (Bryman & Bell, 2011). Mondéjar-Jiménez et al. (2016) found that self-reported behavior seemed to be suitable for the conceptualization of the relationship between behavior and attitudes, according to the TPB principle.

The data was collected once from May 3rd to May 12th, 2019. Since the survey is web-based, the link to the online version was shared on different groups addressing students on Campus Gotland on Facebook. In addition, it was shared on Whatsapp and Messenger to the researchers known and unknown contacts with the inquiry to participate and share it with peers. The authors were also on Campus to collect data and asked present students face-2-face to fill out the survey. On request, the link was sent by email and SMS. The authors also printed out copies of the questionnaire when they collected data since they found that some participants were more likely to fill it out on paper directly than with the shared link to the online version. The print outs were left with the respondent and collected later, so the respondent was free in his/her response. The answers on paper were entered manually into the online survey by the authors to unite all data.

4.3 Survey Design

4.3.1 Overall

English was chosen as the language for this survey. Other languages were not provided as it is presumed that the whole population speaks this language. The survey was designed with closed questions as they are faster to fill out and the responses can be compared more easily (Saunders et al., 2012). Most items can be classified as a Likert-style rating question with a 5-point-scale and are both regular and reverse. The respondents were asked to indicate their (dis)agreement with statements to keep it simple for them. Response categories were chosen with the guidance of Saunders et al., (2012). However, the formulation of the items was aimed to be varied to keep the respondent attentive (Bryman & Bell, 2011; Saunders et al., 2012). In fact, about half of the items that do not address the demographics were structured in a reverse from. In addition, they aimed to be not suggestive (‘yes of course I try to reduce food waste’).

Furthermore, the respondent was informed about an estimated length of the survey and the anonymity on this page. The complete survey can be found in Appendix II.

4.3.2 Intention

Initially, a definition of household food waste is provided. The respondent needed to mark ‘I understand’ to ensure that the respondent was aware and understands the defined framework. The dependent construct Intention (The Menu) was started with, as this reduced the chances that respondents were influenced by later topics when filling out the survey. Item I.1 “How often do you write a shopping list” was adapted from Stefan et al. (2013) as well as Mondéjar-Jiménez et al. (2016). By writing a shopping list, which is part of a planning routine, a person is estimating meals and amounts of food in advance. Also Clark & Manning (2017) state preplanning and sticking to a shopping list is important when it comes to reducing food waste. The follow-up question I.2 was adopted from Stefan et al. (2013) to get a better sense of how people perceive their purchasing amounts even if people do not write a shopping list. This can give conclusions on whether their consumption volume is sustainable or not. Item I.3 (Mondéjar-Jiménez et al., 2016) directly measured the intention to reduce food waste by connecting it to portion sizes. Item I.4 was derived from the same source but was adapted into a reversed question that also includes the intention to waste food. Item I.5 was self-developed to include the topic of expiry dates and how people handle it.

4.3.3 Attitude

The items regarding Attitude (Soup) were all adapted with a personal connection such as “I think…” and were based on the impacts food waste has economically, socially and environmentally. In addition, the general emergence of food waste was included (A.1). Latter was formed on the findings made by Graham-Rowe et al. (2014). Some respondents in their qualitative study viewed food waste as inevitable and had therefore no incentive to reduce their personal level. Item A.2, adapted from Stefan et al. (2013) and Principato et al. (2015), can show a lack of concern but is more related to the personal circumstances than the general situation. Item A.3 – A.5 were adopted statements from Principato et al. (2015) covering a field of impact each.

4.3.4 Subjective norm

was developed by the authors to get information on whether people living together actually interfere with the food waste of their partners/mates. As a final step SN.4 was also developed by the authors to display the willingness to adapt, only based on other suggestion of valued people.

4.3.5 Perceived Behavioral Control

In the fifth section, the Perceived Behavioral Control (Salad) was addressed. PBC.1 concerned a potential barrier for planning activities, adapted from Stefan et al. (2013) changing it from a household scale to a more personal level. PBC.2 focused on the shopping activities and was adopted from Mondéjar-Jiménez et al. (2016). From the same source, PBC.3 was adapted focusing on cooking skills but leaving out the meal preparation for a family. The last items were directly concerned with the waste. PBC.4 and PBC.5 (adapted from Russell et al., 2017) focused on individual food waste. The reversed item PBC.6 was adapted from Russell et al. (2017) and Mondéjar-Jiménez et al. (2016), scaling it down again on the personal amount of food wasted.

4.3.6 Anticipated Emotions

The last part of the theoretical framework is the Anticipated Emotions (Main Course). King et al. (2012) recommend measuring emotions after other items (e.g. general attitude) to make them less influential on them. It was aimed to include both positive and negative feelings in this section. The time of the day when the survey is conducted does not play a role in the results when it comes to emotions related to food (King et al., 2012). To measure emotions with single items has been proven effective in previous studies (Harmon-Jones et al., 2016), but it needs to be taken into account that they are more likely to contain error variance. This disadvantage is considered and accepted in this study. De Pelsmaeker et al. (2017) suggest to rather test specific affective reactions such as regret or guilt than simple positive/negative reactions. Common positive emotions are happy or joyful, while common negative emotions are sad, down, grieving or angry feelings (Harmon-Jones et al., 2016). Collecting data about emotions with two to four items can be more valid since the respondents tend to get less bored or tired by answering the questions (Ibid., 2016). When measuring emotions with rating scales, the order of appearance of the emotions is less relevant compared to the check all that apply (CATA) format. In general, rating scales are more sensitive compared to CATA questions (King et al., 2012). Therefore each item in this survey addressed a single emotion and was asked to be rated.

of regret in the act of wasting. Item AE.6 discussed another positive emotion: relief. Based on the author's assumption that throwing away food can also create positive feelings when it is connected to cleaning. As guilt seems to have the strongest connection to food waste this emotion was placed last (item AE.7). It is adapted from De Pelsmaeker et al. (2017) by changing the context from eating pralines into wasting food. Also, Stefan et al., (2013) contain the item “When I throw away food I feel guilty” but as part of moral attitudes. They did not separate between attitudes and emotions. All of the items were incorporated into a matrix to reduce the length of the section (Saunders et al., 2012).

4.3.7 Demographics

The final section included items about the Demographics (Dessert) of the respondents. They formed the control variables and started with the age (D.1) in years, which had to be entered manually. Item D.2 asking about the gender the respondents were able to select between female, male and non-binary. The nationality (D.3) must be entered in a separate input field. Item D.4 was used to make sure that the respondent is a student at Uppsala University, Campus Gotland and therefore a member of the population for this study (control item). Just like nationality, item D.5 about the study-program includes short open answers to give the respondent the opportunity to answer freely and not be restricted to a selection. Item D.6 concerned the current living situation and provided four possible scenarios. To make sure that the respondents are their own food providers and therefore can relate to all the different activities questioned item D.7 was created. To avoid confusion a definition of food provider was included in the questioning. The last item D.8 covered the environmental interest of the respondent. According to Vermeir & Verbeke (2007), the concern about the environment is an indicator that an individual purchases more sustainable foods. In this survey, the effect of environmental interest will be tested in relation to food waste.

4.3.8 Pilot

Prior to the data collection, a pilot for the survey was carried out to test its design. The members of the thesis group (class) were asked to provide feedback on the survey and complementary a handful of other students. While the former was familiar with the topic due to reading a draft of this thesis, the latter was not as deep into the topic of food waste and come from other backgrounds (diverse nationalities and programs). The size of the pilot groups meets the by Saunders et al. (2012) suggested minimal number of 10 individuals for student research.

were moved to the end to finish with simple questions when the motivation might be low. The definition of household food waste was not always noticed, thus an answer to confirm was added.

Some items needed to be reformulated after the feedback. The items regarding Intention were all phrased ‘in the next few weeks’ based on literature (Stefan et al., 2013; Mondéjar-Jiménez et al., 2016; Russell et al., 2017). This was changed into a more general formulation. Moreover, the items regarding Attitude were strengthened into a statement with a personal connection (e.g. in my opinion…). In addition, the three pillars of sustainability were put into a more relatable context (e.g. “waste of money” and “biodegradability”).

Initially, the survey contained only anticipated emotions regarding the waste of food. Due to recommendations, other activities related to food were included as well. Originally, age was clustered into groups. Since an appropriate division was hard to find, this item was changed into an open answer. Another topic that created confusion in the pilot group was the term household size as it was not clear what range of people this includes (apartment vs. house). Therefore this item was changed into addressing the living situation. This provided the researchers with the sought information. The term ‘home’ included multiple living situations and was therefore better understood by the respondent.

4.4 Data analysis

4.4.1 Procedure

The collected data was analyzed by a computer, which is a standard procedure for this research design. This study used the program SPSS, which is recommended by Bryman & Bell (2011) and provided by Uppsala University. The received digital data was transferred into the program using an extracted Excel sheet from Google Forms.

Before the data could be analyzed a cleaning was necessary. The items regarding study program, age, and nationality were open questions. In the answers typos were removed and other spellings corrected (e.g. ‘Swedish’ and ‘Sweden’ were all changed into ‘Swedish’). Multiple nationalities were brought into a uniform order by putting the same nationality first. Provided minors of different programs such as Game Design were removed to have only one program.

As a first step of the analysis, the descriptive statistics were derived. The distribution of age, gender, nationality, program, living situation, and environmental concern were analyzed to get a general understanding of the respondent group. Frequency tables and pie charts were used as an appropriate tool to visualize these univariate analyses (Bryman & Bell, 2011).

as a next step. As all items were ordinal, the mean and the standard deviation were reviewed. This enabled the comparison between the different items per construct, e.g. mean comparison of all intention questions. Furthermore, the individual items (per construct) were compared to the descriptive statistics.

The third step consisted of a Pearson Correlation. This tool was used to measure how strong two variables were related to each other (Saunders et al., 2012). The outcome gave an overview of which variables were more relevant when forming the intention to reduce food waste.

Lastly, a number of diagnostic tests were done to ensure the reliability and validity of the constructs, and therefore also the prediction value of the conceptual research model. Thereafter, two models were created using a multiple regression analysis to assess the relationship between the dependent variable, the six control variables and the four independent variables. This form of analysis is the appropriate measure to assess the strength of relationship if the model consists of one dependent and multiple independent variables (Saunders et al., 2012). Afterward, these findings were tested against the hypotheses, which were supported or rejected.

When analyzing the correlations and regression, a significance level of p<0,05 indicated a statistically significant relationship (Saunders et al., 2012). One star signified a significance level of p<0,05 and two stars a significance level of p<0,01. The Beta Coefficient indicated the contribution of the independent variable to the dependent variable, with a constant of all other independent variables. The value lies between -1 and +1, indicating that -1 is the strongest negative effect on the dependent variable, +1 the strongest positive effect, and 0 no effect at all (Hair et al., 2006).

4.4.2 Reliability

Construct Number of Items Cronbach’s Alpha

Intention 4 0,859

Attitude 5 0,701

Perceived Behavioral Control 6 0,740

Anticipated Emotions 4 0,724

Subjective Norm 4 0,769

Table 1. Overview of Cronbach’s Alpha per construct.

For the construct Intention, Cronbach’s Alpha was with 0,753 already at an acceptable level without eliminating items. It could be increased to a higher level of 0,859 by removing item I.1 (shopping list). It was decided to take out that item as students do not relate it directly to reducing food waste.

The variable Attitude scored a Cronbach’s Alpha of 0,701. This value was therefore acceptable. The removal of an item would only lower this score and therefore all items are incorporated. The same accounted for the Subjective Norm, with a Cronbach’s Alpha of 0,769 all items were included to form the construct.

Including all items in the calculation of Perceived Behavioral Control yielded a value of 0,606. This score was improved by eliminating item PBC.3 (cooking skills) to 0,740. By deleting this item, the construct PBC had an item scale of 6 left. Apparently, cooking skills are not a factor related to food waste for students.

4.4.3 Validity

Validity is concerned with the correspondence of the content between the empirical data and the concept behind the research (Saunders et al., 2012). After all, the data should be useful to answer the research question. The previously described structure of the survey design aimed to explain the validity of this research. The pilot increased the face validity of the survey e.g. by clarifying the terms ‘household’ and ‘home’. Thereby it increased the sense-making for the respondents.

When analyzing the data a multiple regression analysis was carried out. According to Saunders et al. (2012), a multiple regression analysis needs to ensure that a set of assumptions are met in order for the test to be valid. Therefore the normality, linearity, homoscedasticity, and multicollinearity were tested before the evaluation of the conceptual research models’ dependent and independent variables. These tests showed the internal validity from a statistical perspective (Appendix IV). Firstly, the normality test showed a bell-shaped form and therefore normal distribution existed. A Normal P-Plot proved linearity as the visual examination shows that the points lie close to the straight line and is therefore said to be normal. Moreover, homoscedasticity examined the variances of the dependent value and was confirmed through an observed scattering of the data. Lastly, Hair et al. (2006) specify that multicollinearity is present if the tolerance value is 0,10 or below and/or the large variance inflation factor (VIF) value is 10 or above. All values were not meeting the described criteria and therefore no issues with multicollinearity were found (Saunders et al., 2012).

4.5 Ethical aspects