by L.A.J.W. van de Sande

THE ROLE OF SUPERMARKET PROMOTIONS

IN CREATING FOOD WASTE

by

L.A.J.W. van de Sande

University of Groningen

Faculty of Economics and Business

Msc Marketing Intelligence

Master Thesis

June 21, 2015

ABSTRACT

Do consumers’ waste more food when a product is bought on promotion? Whereas prior research on food waste has mainly focused on a survey approach in order to determine the factors that contributes to households’ food waste, this study focuses on a food-diary approach in order to find out whether supermarket promotions play a role in creating food waste, as well as to what extent this effect is moderated by the type of food. Data was collected via a three-weeks food diary from a variety of households in the Netherlands (N = 97). This diary was used to record daily volume of food waste and to indicate the amount of food bought on promotion. The findings of this study revealed that only price discounts have a positive effect on consumers’ food waste and this effect was greater for virtues bought with a price discount than for vices since this effect was not significant.

Keywords: food waste, promotions, price discount, quantity-based promotion, vices and

MANAGEMENT SUMMARY

Food waste is one of the world’s most pressing issues since more than one-third of all food produced worldwide is wasted or lost. So, over the years the food waste problem gained more attention by society and this resulted in global initiatives such as the campaign “I Value Food” in order to tackle this problem by making people more aware of this pressing food waste problem. Especially since consumers are not aware of the amount of food they waste because their perceptions of waste and reality does not match. Extensive research has explored consumers’ role in this food waste problem and these studies suggest that a lot of this problem lies with consumers since they waste the food. However, consumers are not the only one to blame because there are more factors that contributes to food waste since consumers’ waste can be traced back to their consumption behavior and this behavior is influenced by the tempting supermarket promotions.

These promotions encourage excessive consumption and this results in buying more food than consumers’ need and it is likely that this increase in consumption will result in some food disposal. So, therefore, this study seeks to explore whether supermarket promotions play a role in consumers’ share of food waste. The study will focus on two commonly used supermarket promotions (e.g. price discounts and quantity-based promotions) and the product category, which is classified as vice- (e.g. unhealthy food) and virtue food (e.g. healthy food). This distinction is made because it is assumed that the type of promotion and the product category will have a different effect on consumers’ food waste.

PREFACE

This is my final piece to end my study in Groningen.

I worked with several feelings on my master thesis because this work symbolizes the end of my student period but also a start of something new.

I would like to thank some people during my process of writing my master thesis. First, I would like to thank my supervisor Jenny van Doorn for her guidance, help and suggestions in order to improve and finalize my thesis.

TABLE OF CONTENTS

1. INTRODUCTION ... 7

2. THEORETICAL FRAMEWORK ... 10

2.1. Food Waste ... 10

2.2. Impact of Promotions ... 11

2.3. Consumers’ Responses to Vices and Virtues ... 12

2.4. Conceptual Model ... 13

3. DEVELOPMENT OF HYPOTHESES ... 14

3.1. Type of Promotion ... 14

3.2. Vices versus Virtues ... 16

4. RESEARCH DESIGN ... 18 4.1. Data Collection ... 18 4.2. Measurements ... 19 5. METHODOLOGY ... 24 5.1. Data Exploration ... 25 5.2. Multilevel Models ... 25 6. RESULTS ... 28

7. DISCUSSION AND RECOMMENDATIONS ... 32

REFERENCES ... 37

APPENDIX A – FOOD DIARY AND QUESTIONNAIRE ... 50

1. INTRODUCTION

Food waste is a serious and enormous global problem and over the years this problem received more and more attention by society. In the Netherlands, SIRE has launched a new campaign “Kliekipedia” with a corresponding website kliekipedia.nl, this website provides tips about how to get the most out of your food instead of throwing it away (SIRE, n.d.). The aim of this initiative is to change consumers’ food behavior in order to decrease food waste. This food waste problem is also recognized in other countries since more than one-third of all food produced worldwide is wasted or lost (FAO, 2013). In America, the organization Sustainable America has launched the campaign “I Value Food” in order to make the people more conscious about the pressing food waste problem considering that 40 percent of all edible food is wasted (Gunders, 2012). These campaigns make people aware of how much food is wasted and what kind of impact this has on the environment. This is important since the average Dutch citizen wasted approximate 47 kilos of food per year in 2013 (van Westerhoven, 2013). However, consumers perceptions of how much food they throw away differs substantially from this fact because in reality they wasted three times more food than they think they do (MilieuCentraal, 2014). Therefore, it is important to make consumers and retailers aware of this problem because even the smallest changes and simple actions can help in reducing food waste. Nevertheless, all these initiatives are focused on changing people’s behavior instead of understanding it.

waste but this influence is determined based on questions in a survey asking which factors consumers’ think causes food waste in their households, instead of measuring actual behavior. So, these measurements are not precise in measuring the influence of promotions on consumers’ food behavior because what consumers say is not univocally consistent with their behavioral patterns (Vermeir & Verbeke, 2006).

So, up to now researchers have not yet measured whether promotions increases consumers’ share of food waste, while the role of promotions is already acknowledged by the European Union Committee because their report concluded that retailers can pass food waste on to the consumer by seducing them with promotions such as “Buy one, get one free” (European Union Committee, 2014). This issue is also addressed in the literature because it revealed that promotions are a popular technique to encourage consumers to stockpile products which will help the retailer to sell their inventory (Blattberg, Eppen & Lieberman, 1981) because promotions increases demand and this will reduce retailers’ share in food waste (Theotokis, Pramatari & Tsiros, 2012). The study of Tsiros and Heilman (2005) confirms this because this research emphasized that retailers sell their excess food products at discounted prices in order to reduce the amount of food wasted. Thus, up to now, the role of retailers in consumers’ amount of food waste is only suggested since these studies are only viewed from a retailer perspective instead of both a consumer- and retailer point of view. So, yet it is not clear whether these tempting supermarket promotions can actually pas on retailers’ share of food waste to consumers’ households.

So, to that end, this study seeks to explore whether consumers’ waste more food when a product is bought on promotion. This resulted in the following problem statement: “What is the role of

supermarket promotions in consumers’ share of food waste?”. This role will be determined

with the help of a food diary approach because previous studies successfully used this method in order to get more insights into households’ food waste (Silvennoinen et al., 2014; Koivupuro, Hartikainen, Silvennoinen, Katajajuuri, Heikintalo, Reinikainen & Jalkanen, 2012; WRAP, 2009) . The greatest advantage of this approach is, that this diary not only shows how much food a consumer waste on a weekly basis but also which products are bought on promotion.

also the type of food (e.g. virtue- or vice food) bought on promotion will be examined. The type of food is divided into virtue- and vice food since this distinction addresses the healthy- (e.g. virtues) and unhealthy (e.g. vices) food sold in the supermarkets. These are included because prior studies revealed that consumers’ preference for the type of promotion depends on the type of food (Mishra & Mishra, 2011; Wertenbroch, 1998) and therefore it is expected that the promotions in vice- and virtue categories might have a different effect on food waste. Especially since the promotional effect is greater for perishable product categories (Sun, 2005).

So, in this study the following research questions will be addressed:

1) Do consumers waste more food when a product is bought on promotion? 2) Does the type of promotion have an influence on consumers’ food spoilage?

3) Is consumers’ food waste behavior influenced by a specific product category on promotion?

This study contributes to the existing literature in food waste because this is the first study that examines the impact of promotions on consumers’ actual food waste behavior, with a food diary approach instead of a questionnaire. Second, this study takes a marketing perspective on the food waste problem, while previous studies mainly looked to the food waste issue from a consumer- (Principato et al., 2015; Silvennoinen et al., 2014) and supply chain (Kummu, Moel de, Porkka, Siebert, Varis & Ward, 2012; Mena, Adenso-Diaz & Yurt, 2011) point of view. Third, to the best of the author her knowledge, this is the first study that examines the impact of the type of food on promotion on consumers’ food waste. In addition, this study explores whether promotions have an effect in consumers’ home environment since prior studies only examined the effects of promotions in store on consumers’ buying behavior (Farr-Wharton et al., 2014; Schultz & Block, 2011).

2. THEORETICAL FRAMEWORK

This section reviews prior literature on food waste, promotions and consumers’ responses to vices and virtues in order to get more insights in whether promotions play a role in consumers’ share of food waste.

2.1.Food Waste

In consumers lives food plays a central role and a large amount of this food is wasted. The report of CREM revealed that on average 47 kilos of food per person per year is thrown away in the Netherlands (van Westerhoven, 2013). This waste occurs when food is unintendedly lost which could have been prevented since 64 percent of consumers’ waste is avoidable (Porpino, Parente & Wansink, 2015). The high amount of food wasted conflicts with consumers’ aversion to waste (Bolton & Alba, 2012) and the increased concern for the environment and community and its positive impact on consumers’ waste minimization (Tonglet, Phillips & Bates, 2004). This conflict might be explained by consumers’ perceptions of waste and reality since consumers’ waste three times more food than they think they do (MilieuCentraal, 2014).

The interest in food waste has led to an increased amount of studies with food waste as subject (Principato et al., 2015; Graham-Rowe, Jessop & Sparks, 2014; Silvennoinen et al., 2014; Stefan et al., 2013). Although these studies offer evidence for the contributors to food waste, overall these studies suggest that households are responsible for the volume of waste. For example, the study of Silvennoinen, Katajajuuri and Hartikainen (2014) discussed avoidable food waste and revealed that preparing more food than needed, foods’ expiring dates and plate leftovers are the main reasons for disposing food, whereas the research of Principato, Secondi and Pratesi (2015) revealed two major issues of wasting food and that is consumers’ misunderstanding of expiration dates and the concern that food is no longer fresh. However, consumers are not the only one to blame for this problem (Evans, 2011) since the route of households’ waste starts in the supermarket (WRAP, 2009).

of Porpino, Parente and Wansink (2015) revealed that excessive consumption (e.g. purchasing) is an important antecedent of households’ waste. Excessive consumption is influenced by promotions because the persuasiveness of promotions results in an increase in consumption (Wansink, Kent & Noch, 1998). This increase in consumers’ demand might result in grocery items that will never be used and therefore be wasted (Wansink, Brasel & Amjad, 2000).

2.2.Impact of Promotions

Researchers have spent a lot of time in examining the effect of sales promotions. These studies provided insights into how promotions affect consumers’ behavior such as their value perceptions and purchase intentions (Darke & Chang, 2005; Hardesty & Bearden, 2003), but also how promotions affect supermarkets. These sales promotions have the effect to increase supermarkets short-term sales (van Heerde, Leeflang & Wittink, 2004). This increase in sales is expected because promotions have a positive effect on consumers’ stockpiling- and purchase quantity behavior (Gupta, 1988; Bell, Iyer & Padmanabhan, 2002; Pauwels, Hanssens & Siddarth, 2002). Nevertheless, it stimulates consumers towards buying because promotions offer consumers a reason for immediate purchase (Kotler & Armstrong, 2004). This immediate urge to buy is extensively explained in the literature of impulsive buying behavior because different studies revealed that (in-store) sales promotions have a positive effect on consumers’ impulse buying behavior (Karbasivar & Yarahmadi, 2011; Zhou & Wong, 2008). In other words, consumers’ in-store behavior is affected by sales promotions and this results in consumers buying more products (Wansink et al., 1998) than planned. Moreover, Mela, Jedidi and Bowman (1998) found that consumers also have the tendency to wait for good promotions in order to buy larger quantities of a certain item. So, promotions not only stimulates buying but it also results in purchasing larger quantities of a product.

2.3. Consumers’ Responses to Vices and Virtues

Vice food can be described as “wants” because these products give consumers instant pleasure such as chocolate, while ignoring the long-term negative consequences (e.g. weight gain and the related health risks) (van Doorn & Verhoef, 2011), whereas virtue food is known as “shoulds” because as compared to vices it is less attractive and satisfying but the choice is more prudent (van Doorn & Verhoef, 2011). The vice food/virtue food distinction can be compared with the distinction between hedonic- and utilitarian goods (Dhar & Wertenbroch, 2000). Vices and hedonic goods both share the same main characteristic and that is providing (immediate) pleasure (Chitturi, Raghunathan & Mahajan, 2008), while virtues and utilitarian goods both offer consumers concrete benefits (Sela, Berger & Liu, 2009). This comparison helps to explain why consumers purchase larger amounts of virtues than vices since it is easier to rationing purchase quantities for virtues than vices (Wertenbroch, 1998).

This rationing might be explained by consumers’ tendencies to justify their decisions to others and themselves (Simonson, 1989). These justifications are especially important when consumers decide to indulge because when there is no reason for indulging vice food, consumers will experience more negative feelings (Xu & Schwarz, 2009). So, consumers will only succumb to the indulgence if they can justify their decision. For example, the research of Hui, Bradlow and Fader (2009) revealed that consumers are more likely to shop at the aisle with vice categories when they first purchased virtue categories. Nevertheless, this might differ per person because the need to justify decisions and their outcomes depends on the individual self-imposed system of rules (Cheema & Soman, 2006). For instance, one consumer could justify the vice consumption with eating more healthy meals that day, while another consumer will eat the whole week healthy in order to have one cheat day.

2.4.Conceptual Model

Figure 1 represents the conceptual model (see page 14). This model tests the effect of promotions on consumers’ food waste, as well as the specific food category on promotion. This relationship will be tested with two key types of promotions: price discounts and quantity-based promotions, since prior research has shown that monetary- (e.g. price discounts) and non-monetary (e.g. volume discounts) sales promotions are the most common techniques used in stores (Chandon, Wansink & Laurent, 2000). In order to determine this effect, it is important to have a common understanding of the term food waste. Therefore, consistent with most recent scientific articles about food waste, the term food waste is described as “All (edible) food that is wasted that could have been prevented and therefore eaten if it had been prepared or stored differently” (Schneider, 2008; Silvennoinen et al., 2014; Principato et al., 2015).

This study expects that the influence of promotions on consumers’ food waste will vary on the type of promotion. This is expected since consumers prefer quantity-based promotions over price discounts (Chen, Marmorstein, Tsiros & Rao, 2012) and such preferences will lead to greater purchases because this type of promotion gives consumers the feeling that they got a good deal by getting more for their money (Haws & Winterich, 2013). Especially, since shoppers are encouraged by this specific demand for food such as “Buy one, get one free”, that they will buy more food than needed (Caswell, 2008). However, the persuasiveness of promotions could differ per food category (e.g. vice- and virtue food) (Mishra & Mishra, 2011). In such cases, consumers’ choice for an item on promotion will likely to be determined not only by consumers’ preference for the promotional type but also by the specific food on promotion.

not need a situation that facilitates the justification of purchasing virtue food since the consumption is free from guilt (Mishra & Mishra, 2011). This means that the (perceived) healthfulness of food should have a different effect on consumers’ response to promotion types. Therefore, it is expected that the effect of promotions on food waste is moderated by the product category since consumers’ hold specific preferences for the promotional- and food type.

3. DEVELOPMENT OF HYPOTHESES 3.1.Type of Promotion

Quantity-based promotions such as “Buy one, get one free” are seen as segregated gains, while price discounts are perceived as reduced losses (Diamond & Campbell, 1989). These losses are weighted more heavily than gains, which means that losses are more intensely experienced than gains (Kahneman & Tversky, 1979). Consumers may experience these losses because (frequent) price discounts may reduce the perceptions of value (Grewal, Krishnan, Baker & Borin, 1998). This reduction can be explained by consumers’ internal reference price. The internal reference price differs per person because each individual has a specific reference point on their internal judgment scale which is used to make judgments about the offered price, for example when the price exceeds this reference point it will be perceived as high (Thomas & Menon, 2007). So, it is expected that consumers’ internal reference price is lower due to price discounts because this type of promotion increases consumers’ price sensitivity (Mela, Gupta & Lehmann, 1997). The study of Lowe and Barnes (2012) confirms this. This study found that price discounts have a larger impact on consumers’ reference-price perceptions, whereas quantity-based promotions have the smallest effect on these perceptions (Lowe & Barnes, 2012).

Thus, whether consumers perceive a promotions as a gain or loss depends on the nature of the promotion. According to the study of Diamond and Sanyal (1990), consumers’ preference

Type of Promotion: - Price Discounts - Quantity-Based Promotions Food Waste Product Category: - Virtue Food - Vice Food

“Figure 1 – Conceptual Model”

H1

goes out to promotions that are framed as gains given that these promotions provide better value. This is in line with the research of Sinha and Smith (2000) because the transaction value is perceived as higher when the promotion is framed as a gain. Transaction value is “Consumers’ perceptions of the psychological satisfaction or pleasure from taking advantage of getting a deal” (Xia & Monroe, 2010: 884) and these transaction values are an important value during consumers’ shopping trips (Davis & Hodges, 2012). So, these findings suggest that consumers hold specific preferences for quantity-based promotions because it is seen as segregated gains (Diamond & Campbell, 1989). Moreover, consumers view quantity-based promotions more positively while price discounts are viewed more negatively (Chandran & Morwitz, 2006). The positive view of “Buy one, get one free” promotions can be explained through consumers reactions towards getting one item for free. Consumers perceive this “free” product if it has additional value and it creates a much higher (positive) affective reaction compared to price discounts (Shampanier, Mazar & Ariely, 2007). That is why consumers are more prone to quantity-based promotions during food shopping than to price discounts (Farr-Wharton et al., 2014).

Quantity-based promotions also have an impact on consumers’ stockpiling behavior. Sinha and Smith (2000) explains that consumers prefer quantity-based promotions for stock-up (stockpiling) items because this strongly infers gain through the additional item. Besides that, the “Buy one, get one free” promotions resulted in consumers buying larger quantities of the promoted product (Diamond, 1992). This effect might be greater on product items that consumers tend to plan (e.g. yoghurt) (Stilley, Inman & Wakefield, 2010). So, this may result in a growth in consumers’ inventories. However, these inventories may have a negative side effect because consumers’ inventory estimations are seldom accurate and this can lead to overstocking and spoilage (Chandon & Wansink, 2006).

behavior since consumers’ experience a more intense force of attraction towards this type of promotion (Higgins & Scholar, 2009). This attraction may cause consumers buying more of an item than needed and this may contribute to food waste because for example not all items are used for cooking (Farr-Wharton et al., 2014). So, this has led to the following hypothesis:

H1: Quantity-based promotions have a greater effect on consumers’ food waste than price discounts.

3.2.Vices versus Virtues

Consumers’ response to promotions differs per product category (Blattberg, Briesch & Fox, 1995). This response may lead to an increase in the demand of a specific category (Ailawadi & Neslin, 1998) or it even may have a cross-over effect on the demand of other categories (Erdem & Sun, 2002). The difference in consumers’ response to a specific product category promotion might be explained by the type of food. According to the research of Mishra and Mishra (2011), consumers favor a bonus pack (quantity-based) for virtue food and price discounts for vice food. This finding is in line with the study of Parreño-Selva, Mas-Ruiz and Ruiz-Conde (2014) because they revealed that price discounts for vice products have a greater effect on sales than virtue products. The reason that consumers lean towards a specific type of promotion for a specific food category might be explained through consumers’ inner conflict. For example, vice food has the effect of inducing consumers’ guilt (Ramanathan & Williams, 2007) because this is mostly purchased for personal pleasure and indulgence (Alba & Williams, 2013; Khan & Dhar, 2010) and that is why it is hard to justify vices (Kivetz & Simonson, 2002). This inner conflict can be resolved through justification because justification is an important aspect to clarify your choices (Levav, Kivetz & Cho, 2010). Price discounts provides consumers’ a reason to indulge because this discount justify their choice for buying vice food since they cannot overconsume it, as compared to quantity-based promotions (Wertenbroch, 1998), and at the same time they will save money (Mishra & Mishra, 2011).

2014). ). So, consumer’s motivation to exert mental effort is lowered due to the discount, which means that these choices are relatively effortless because it relies on quick associations and heuristics (Masicampo & Baumeister, 2008). Therefore, it is more likely that consumers will purchase more vice food due to the use of heuristics (Cools, Schotte & McNally, 1992) and this might result in buying more food than planned and buying too much food causes food waste (Cox & Downing, 2007). So, this has led to the following hypothesis:

H2: Price discounts for vices will lead to more overall food waste than quantity-based promotions for vices.

In contrast, virtue food is seen as healthy but less enjoyable compared to the vice food (Chen & Sengupta, 2014). For this type of food (e.g. virtues) consumers prefer a bonus pack because of the absence of consumption guilt, so that give consumers a free pass to consume more of this product (Mishra & Mishra, 2011). This preference does not depend on the amount of discount because even when the quantity discounts are relatively shallow, consumers will still buy larger amounts of virtue food (Wertenbroch, 1998). Also the study of Haws and Winterich (2013) revealed the preference for larger packages, which increases the size of the consumption of healthy food. This effect could be explained through the rule of justification because consumers can easier justify their spending on virtue food than on vices (Okada, 2005). So, this may explain why the promotion effect is greater on consumers consumption behavior for perishable product categories (Sun, 2005).

However, the downside is that this effect may lead to overconsumption which could result in food waste because 48 percent of virtue food (such as fresh vegetables and salads, bread, diary, eggs, and fresh fruit) is wasted which could be avoided (WRAP, 2013). This waste could be due to buying more food than needed (Caswell, 2008) and it is likely that this is induced by the quantity-based promotions because getting one item for free is perceived as additional value and this results in a positive affective reaction (Shampoanier et al., 2007). Therefore, it is assumed that consumers will purchase more virtue food on promotion and this might result in throwing some food away because of the amount of food that is bought and given the limited shelf life of this product (WRAP, 2013). So, this has led to the following hypothesis:

4. RESEARCH DESIGN 4.1.Data Collection

To test whether promotions play a role in creating food waste, this study measured the impact of promotions in consumers’ home environment. Therefore, two types of methods are used 1) a food diary and 2) a questionnaire. The data was collected from a variety of households in the Netherlands in which the participants are part of a single- or multi-person households. In total a sample of 97 households participated in this study. The participants used the food diary to register the food that have been wasted (but could be avoided) and the questionnaire is used to acquire additional information about the head of the household such as their socio-demographic characteristics (e.g. gender, age, level of education, household size, employment status, income level) and participants’ opinions towards promotions. The study period was three weeks and the data was collected in the months April and May because the respondents started on different days with the food diary.

Table 1 – Sample Characteristics

Gender Age Person in Households

Education Occupation Net Income Per Month

Male Female 15,5% 84,5% Min. Max. Mean SD 20 82 46,54 16,903 1 = 25,8% 2 = 38,1% 3 = 20,6% 4 = 14,4% 5 = 1,0% MBO HBO University Other 25,8% 37,1% 17,5% 19,6% Job Student Retired Other 67,0% 9,3% 11,3% 12,4% €0 – €999 €1000 – €1999 €2000 – €2999 €3000 – €3999 €4000 or more 7,2% 23,7% 16,5% 21,6% 14,4%

4.2.Measurements

The participants all received a document that included a three weeks food diary and a questionnaire, with detailed instructions on how to weigh and report food waste on a daily basis. The document is carefully constructed with three other marketing students, who also investigate a variety of factors that may contribute to food waste. This means that the diary and questionnaire contains elements that are not applicable for this study, and to that end these elements will be excluded from this research.

Food Diary

The participants were asked to weigh the amount of waste they produce, and record this using the food diaries. This self-measurement method is chosen because it helped to collect the data over time, provides a more extensive view on the composition of food waste (Koivupuro et al., 2012) and this is the most complete accurate real measurement of food waste (Langley, Yoxall, Heppell, Rodriguez, Bradbury, Lewis, Luxmoore, Hodzic & Rowson, 2010). The food diary approach has been used in past researches of food waste (Silvennoinen et al., 2014; Koivupuro et al., 2012; WRAP, 2009). Measuring by the consumer can be perceived as a self-monitoring method because participants supervision their own actions (Kilbourn, 1991). This self-weighing method (a.k.a. food diary) is an important motivator for behavior change because in most cases people become aware of their own waste, and this is a visible reminder to change their behavior (Sharp, Giorgi & Wilson, 2010). So, this may have an influence on the amount of food wasted and therefore it is important to control for this effect.

Table 2 – Food Classificationa

Food Categories Products

Milliliter (ml) Gram (g) Virtue Category Dairy Products

Soup

Fresh (Fruit) Juice

Bread Cereals Eggs Fruit Ready-made Meals Vegetables Vice Category Alcohol

Desserts Soft Drinks

Cheese Chocolate

Cookies and Pastries Crisps, Nuts and Candy Neither Category

(neither virtue or vice)

Coffee and Tea Sauces

Chicken and Meat Products (+Fish) Rice, Pasta and Potatoes

Leftovers Baby Food

a. Food classification is based on the categorization of van Doorn & Verhoef (2015) and Hui et al. (2009)

Besides reporting food waste, participants also have to indicate on a daily basis whether and what type of food is bought on promotion in the supermarket(s). In addition, they also have to specify which products are bought with a price discount and/or quantity-based promotion. All these elements in the food diary are tested multiple times in order to make sure that the diary was understandable and user friendly. These pre-tests resulted in a final food diary that has been used in this study (see appendix A page 51 for the diary).

An overview of consumers’ food waste and the different category of waste are provided in table 3 (see page 21). The amount of waste was calculated in gram and therefore are the items measured in milliliter converted to gram. In order to convert volume (milliliter) into weight (gram), the report of Charrondiere, Haytowitz and Stadlmayr (2012) was consulted because they provide composition tables of food and beverages in gram and milliliter. This report suggest that for the beverages in this study the following guideline can be used; 1 milliliter is 1 gram (Charrondiere et al., 2012).

Table 3 – Overview of Consumers’ Food Wastea

Min. Max. Mean SE Percentageb

Virtue Category Waste 0 4810 464,79 592,156 50,09% Vice Category Waste 0 1500 90,76 205,469 9,77% Neither Category Waste 0 2450 373,39 528,160 40,14% Total Waste 0 6825 927,97 988,490 100%

a. The amount of waste is specified on a weekly basis. b. This is the percentage of the total waste.

Questionnaire

In every participating household, the questionnaires was given to the person that does the groceries. The questionnaire was designed to obtain additional information about participants’ background, such as their demographics, as well as consumers’ attitudes towards sales promotions. These questionnaires were completed in the study period and the questions helped them to retrieve from memory the shopping details of their grocery trips. Consumers’ attitudes towards various promotions such as price discounts and quantity-based promotions are measured via three items, adopted from the study of Chandon et al. (2000), which are based on measuring the overall evaluations of promotions. The three measurements all started with “With this type of promotion…” and are completed with the following items: 1) “I really save money”, 2) “I feel that I am getting a good deal” and 3) “I really spend less” (Chandon et al., 2000: 68). The participants rated these items for the two types of promotions: price discounts and quantity-based promotions, on a seven-point Likert Scale ranging from (1) completely disagreeing, to (7) completely agreeing.

The questions that indicates consumers’ attitudes- and behavioral intentions towards promotions are tested for internal consistency reliability in order to determine how closely the set of items are related as a group. The method that have been used is the Cronbach’s Alpha. The reliability of consumers’ attitudes towards price promotions is .832, while attitudes towards quantity-based promotions have a reliability of .873, and participants behavioral intentions have a α of .679. The Cronbach’s alpha for all measures are above 0.60, which means that these items are closely related to be count as one variable, so this results in three variables: attitudes towards price discounts, attitudes towards quantity-based promotions, and behavioral intentions towards promotions.

Control Variables

This study accounts for several variables because of the potential influence on consumers’ food waste. These are the five socio-demographic characteristics: gender, household size, level of education. age, and income, and participant’s perceptions of promotions: consumers’ attitudes and behavioral intentions towards supermarket promotions. Prior research shows that the amount of food wasted can be explained by the gender of a person that is mainly responsible for the groceries but also the size of the household since more people in a household waste more food (Silvennoinen et al., 2014; Koivupuro et al., 2012). But also the level of education could be an indicator of food spoilage (Koivupuro et al., 2012). In addition, there can also be differences across age groups since prior research shows differences across ages in the amount of food thrown away (Jörissen, Priefer & Bräutigam, 2015). For example the research of Cox and Downing (2007) shows that younger people tend to thrown away more food than older people. Similarly, the household income also helps to explain the amount of food wasted since the consumers’ in high income nations have the largest contribution to food waste (Graham-Rowe et al., 2014). This is confirmed by the study of Stefan, van Herpen, Tudoran and Lähteenmäki (2013) which revealed that a higher household income will lead to more food waste.

consumers’ behavioral intentions towards promotions are included (Laroche et al., 2003). These behavioral intentions focuses on consumers’ intentions to stockpile promotional items that they normally use since the amount of items purchased is influenced by consumers’ product usage characteristics (Chan, Narasimhan & Zhang, 2008).

Consumers’ attitudes and behavioral intentions towards promotions are included as control variables because how consumers’ perceive promotions might help explain the link between supermarket promotions and food waste. The study of Seipel (1971) found that sales promotions have a positive effect on consumers’ attitudes towards the specific food on promotion and these (positive) attitudes might also have an influence on the amount of food wasted. But also consumers’ intentions to stockpile promotional items might have a positive effect on consumers’ food spoilage since promotions increase consumers’ demand (Wansink et al., 1998) and encourage consumers to stockpile products (Blattberg et al., 1981). Stockpiling might result in overstocking and spoilage since consumers’ inventory estimations are seldom accurate (Chandon & Wansink, 2006).

Tables 4 and 5 (page 24) provide an overview of all the included variables and the descriptive statistics.

Table 4 – Overview Time-Varying Predictors and Descriptive Statistics

Variable Label Mean SE

Time-Varying Predictors Price Discounts

The number of products bought with a price discount for customer i in week t

PDti 1.03 1.662

Quantity-Based Promotions

The number of products bought with a quantity-based promotion for customer i in week t

QPti .36 .905

Price Discounts for Vice Food

The amount of vice food that is bought with a price discount for customer i in week t

PDisVcFti .23 .578

Price Discounts for Virtue Food

The amount of virtue food that is bought with a price discount for customer i in week t

PDisVrFti .47 .951

Price Discounts for Neither Food

The amount of neither food that is bought with a price discount for customer i in week t

PDisNtFti .33 .765

Quantity-Based Promotions for Vice Food

The amount of vice food that is bought with a quantity-based promotion for customer i in week t

QProVcFti .11 .392

Quantity-Based Promotions for Virtue Food

The amount of virtue food that is bought with a quantity-based promotion for customer i in week t

QProVrFti .16 .465

Quantity-Based Promotions for Neither Food

The amount of neither food that is bought with a quantity-based promotion for customer i in week t

Table 5 – Overview Control Variables and Descriptive Statistics

Variable Label Mean SE

Control Variables Attitudes towards Price Discounts

Consumers’ attitudes towards the monetary savings of price discounts

APDi 4.5358 1.38684

Attitudes towards Quantity-Based Promotions

Consumers’ attitudes towards the monetary savings of quantity-based promotions

AQPi 4.9444 1.57761

Behavioral Intentions towards Stockpiling

Consumers’ intentions towards stockpiling of promotional items

BISi 4.7111 1.64109

Age

Customers’ age in years

Agei 46.90 16.470

Gender

(male = 0, female = 1)

Genderi .84 .363

Household Size

The number of persons in the participants’ household

Householdsizei 2.28 1.067

Education

The level of education is ranked from Lower Vocational Education to University

Educationi 4.21 1.527

Household Income

Different classes of Households’ monthly net income

Incomei 5.28 1.924

5. METHODOLOGY

In this study a sample of 97 households in the Netherlands are observed over a period of 3 weeks of which 90 households are analyzed, this means that seven participants are excluded from the sample because they failed to specify the promotions according to the guidelines of this study. The three weeks have led to a total of 270 observations of which 47 percent of the observations (N = 127) revealed cases of promotions. In total, 375 products are bought with a promotion and 67 participants are responsible for this amount because they have at least bought one food item on promotion in this period. In addition, 73,87 percent (N = 277) of these products are bought with a price discount, while 26,13 percent (N = 98) of the items are bought on a quantity-based promotion. The most popular food item bought on promotion is virtue food (N = 171 (45,6%)), followed by the neither category (N = 111 (29,6%)), and 24,8 percent (N = 93) for vice food.

(Twisk, 2006). The exact statistical method that was conducted is the Linear Mixed Models (LMM) in SPSS. This method is applicable because not only will it examine the within-person variations over multiple time points, but it also provide information about between-person variations. The model for this study is specified in section 5.2, but first the data will be explored in order to get some insights into the data.

5.1.Data Exploration

This study detected four cases of outliers but for several reasons it is decided to leave them in the dataset. First, this study examines human behavior and the aim is to find the underlying causes of consumers’ food waste behavior. Second, the detection of cases of outliers in this dataset might be part of human nature since people differs in the amount of food they waste because some people waste a lot of food, while others do not waste any food. Finally, given the size of the sample it is better to leave these participants in the dataset in order to prevent loss of data. Moreover, as expected the food diary approach resulted on average in a decrease of food waste over time (see figure 2) since this method makes people aware of their food waste behavior (Sharp et al., 2010). Therefore, it is decided to include “Week” as the repeated effects variable since this will account for this effect.

The specific food bought on promotion is classified according to the vice-, virtue-, and neither category distinction (van Doorn & Verhoef, 2015; Hui et al., 2009) in order to examine whether vice- and/or virtue food items bought with a

price discount and/or quantity-based promotion have an influence on consumers’ food waste. However, since 29,6% of the promotional items consists out of neither vice or virtue food it is therefore decided to also include the neither food category items bought on promotion in the analysis in order to avoid omitted variable bias.

5.2.Multilevel Models

Several models are used to test the relationships in the conceptual model. Multiple tests are conducted for the final specification of these models since several issues must be taking into account for the development of the multilevel models such as the variability and the inclusion of random- and/or fixed effects (Heck, Thomas & Tabata, 2014; Twisk, 2006). The independent

variables are included as fixed effects because the intrinsic interest of this study is whether promotions play a role in creating food waste, so these effects represent the relationship between food waste and predictor variables for an entire population and not per respondent (West, Welch & Gatecki, 2007). Another reason is that this study is conducted at micro-level, in others words the predictors are measured at participant level, which means that random regression coefficients are not appropriate because it can only be considered to be random at a level above the participant-level such as group-level (Twisk, 2006). So, only a random intercept is included in this model because variability in food waste between participants is expected since previous studies revealed that the magnitude of food waste varied with households (Parizeau, von Massow & Martin, 2015; Suthar & Singh, 2015; Williams, Wikström, Otterbring, Löfgren & Gustafsson, 2012). This is confirmed by the Wald Z static and the total variability, the Wald Z statistic (Wald Z = 5.657, p = .000 < .01) indicates that the intercepts vary significantly across the sample, and the Intraclass Correlation Coefficient (ICC) suggests that about 37,59% (529419,476229 / (529419,476229 + 878988,378663) = .3759) of the total variability in food waste lies between participants. The volume of waste not only varies between participants but also within participants because the estimates suggests that about 62,41% (878988,378663 / (529419,476229 + 878988,378663) = .6241) of the waste varied within participants over time, this is confirmed by the Residual parameters because there is a significant variance to be explained within groups over time (Week 1: Wald Z = 5.480, p = .000 < .01; Week 2: Wald Z = 3.916, p = .000 < .01; Week 3: Wald Z = 3.230, p = .001 < .01). So, the inclusion of a random intercept and a person- and time specific residual is recommended.

Two models are used to test the hypotheses. Model 1, the type of promotion model, tests whether quantity-based promotions and price discounts affect consumers’ food waste, whereas model 2 examines the relationship between the specific food bought on promotion and households’ food waste, in which t is used to indicate a time point and participants are specified with index i: (1) 𝐹𝑊_𝑡𝑖 = 𝛼_0𝑖+ 𝛽₁(𝑄𝑃𝑡𝑖) + 𝛽2(𝑃𝐷𝑡𝑖) + 𝛽3(𝐺𝑒𝑛𝑑𝑒𝑟𝑖) + 𝛽4(𝐴𝑔𝑒𝑖) + 𝛽5(𝐻𝑜𝑢𝑠𝑒ℎ𝑜𝑙𝑑𝑠𝑖𝑧𝑒𝑖) + 𝛽6(𝐸𝑑𝑢𝑐𝑎𝑡𝑖𝑜𝑛𝑖) + 𝛽7(𝐼𝑛𝑐𝑜𝑚𝑒𝑖) + 𝛽8(𝐴𝑃𝐷𝑖) + 𝛽₉(𝐴𝑄𝑃_𝑖) + 𝛽₁₀(𝐵𝐼𝑆_𝑖) + 𝜀_𝑡𝑖 𝛼_0𝑖 = 𝛾₀₀+ 𝑢_0𝑖 (2) 𝐹𝑊𝑡𝑖 = 𝛼0𝑖+ 𝛽1(𝑄𝑃𝑟𝑜𝑉𝑟𝑡𝑖) + 𝛽2(𝑄𝑃𝑟𝑜𝑉𝑐𝐹𝑡𝑖)+ 𝛽3(𝑄𝑃𝑟𝑜𝑁𝑡𝐹𝑡𝑖) + 𝛽₄(𝑃𝐷𝑖𝑠𝑉𝑟𝐹_𝑡𝑖) + 𝛽₅(𝑃𝐷𝑖𝑠𝑉𝑐𝐹_𝑡𝑖) + 𝛽₆(𝑃𝐷𝑖𝑠𝑁𝑡𝐹_𝑡𝑖) + 𝛽₇(𝐺𝑒𝑛𝑑𝑒𝑟_𝑖) + 𝛽₈(𝐴𝑔𝑒_𝑖) + 𝛽₉(𝐻𝑜𝑢𝑠𝑒ℎ𝑜𝑙𝑑𝑠𝑖𝑧𝑒_𝑖) + 𝛽₁₀(𝐸𝑑𝑢𝑐𝑎𝑡𝑖𝑜𝑛_𝑖) + 𝛽₁₁(𝐼𝑛𝑐𝑜𝑚𝑒_𝑖) + 𝛽₁₂(𝐴𝑃𝐷_𝑖) + 𝛽₁₃(𝐴𝑄𝑃𝑖) + 𝛽14(𝐵𝐼𝑆𝑖) + 𝜀𝑡𝑖 𝛼0𝑖 = 𝛾00+ 𝑢0𝑖

Where, FWti is the overall food waste for participant i in week t, 𝛼_0𝑖 is the random intercept for participant i, 𝛾00 is the fixed sample grand mean intercept, 𝑢0𝑖 is the between-person variation

in intercepts, and 𝜀_𝑖𝑡 is the variation in food waste within participants over time. See tables 4 and 5 page 23 and 24 for the labels of the independent variables.

Models 1 and 2, do not account for the delay effect between the consumption of promotions and consumers’ wasting food since the promotional items that are bought now, can be discarded later. Therefore, it is decided to test whether the promotional lagged effects have an influence on consumers’ food waste. However, the lagged effects might cause some statistical problems because the inclusion of these effects results in a decrease in the number of usable observations (Leeflang et al., 2014). Therefore, it is decided to run four models in which the first two models (3 and 5) examines the effect of promotions from a period back in week 2 and 3, while the other two models (4 and 6) inspects the promotional effect from two periods ago in week 3:

(3) 𝐹𝑊_𝑡𝑖 = 𝛼_0𝑖+ 𝛽₁(𝑄𝑃_𝑡𝑖) + 𝛽₂(𝑄𝑃_𝑖𝑡−1) + 𝛽₃(𝑃𝐷_𝑡𝑖) + 𝛽₄(𝑃𝐷_𝑖𝑡−1) + 𝜀_𝑡𝑖

(4) 𝐹𝑊_𝑡𝑖 = 𝛽₀+ 𝛽₁(𝑄𝑃_𝑡𝑖) + 𝛽₂(𝑄𝑃_𝑖𝑡−2) + 𝛽₃(𝑃𝐷_𝑡𝑖) + 𝛽₄(𝑃𝐷_𝑖𝑡−2) + 𝜀_𝑡 And, (5) 𝐹𝑊_𝑡𝑖 = 𝛼_0𝑖+ 𝛽₁(𝑄𝑃𝑟𝑜𝑉𝑟_𝑡𝑖) + 𝛽₂(𝑄𝑃𝑟𝑜𝑉𝑟_𝑖𝑡−1) + 𝛽3(𝑄𝑃𝑟𝑜𝑉𝑐𝐹𝑡𝑖) + 𝛽4(𝑄𝑃𝑟𝑜𝑉𝑐𝐹𝑖𝑡−1) + 𝛽5(𝑄𝑃𝑟𝑜𝑁𝑡𝐹𝑡𝑖) + 𝛽6(𝑄𝑃𝑟𝑜𝑁𝑡𝐹𝑖𝑡−1) + 𝛽₇(𝑃𝐷𝑖𝑠𝑉𝑟𝐹𝑡𝑖) + 𝛽7(𝑃𝐷𝑖𝑠𝑉𝑟𝐹𝑖𝑡−1) + 𝛽8(𝑃𝐷𝑖𝑠𝑉𝑐𝐹𝑡𝑖) + 𝛽9(𝑃𝐷𝑖𝑠𝑉𝑐𝐹𝑖𝑡−1) + 𝛽10(𝑃𝐷𝑖𝑠𝑁𝑡𝐹𝑡𝑖) + 𝛽11(𝑃𝐷𝑖𝑠𝑁𝑡𝐹𝑖𝑡−1+ 𝜀𝑡𝑖 𝛼0𝑖 = 𝛾00+ 𝑢0𝑖 (6) 𝐹𝑊_𝑡𝑖 = 𝛽₀+ 𝛽₁(𝑄𝑃𝑟𝑜𝑉𝑟𝑡𝑖) + 𝛽2(𝑄𝑃𝑟𝑜𝑉𝑟𝑖𝑡−2) + 𝛽₃(𝑄𝑃𝑟𝑜𝑉𝑐𝐹_𝑡𝑖) + 𝛽₄(𝑄𝑃𝑟𝑜𝑉𝑐𝐹_𝑖𝑡−2) + 𝛽₅(𝑄𝑃𝑟𝑜𝑁𝑡𝐹_𝑡𝑖) + 𝛽₆(𝑄𝑃𝑟𝑜𝑁𝑡𝐹_𝑖𝑡−2) + 𝛽7(𝑃𝐷𝑖𝑠𝑉𝑟𝐹𝑡𝑖) + 𝛽7(𝑃𝐷𝑖𝑠𝑉𝑟𝐹𝑖𝑡−2) + 𝛽8(𝑃𝐷𝑖𝑠𝑉𝑐𝐹𝑡𝑖) + 𝛽9(𝑃𝐷𝑖𝑠𝑉𝑐𝐹𝑖𝑡−2) + 𝛽₁₀(𝑃𝐷𝑖𝑠𝑁𝑡𝐹_𝑡𝑖) + 𝛽₁₁(𝑃𝐷𝑖𝑠𝑁𝑡𝐹_𝑖𝑡−2+ 𝜀_𝑡 6. RESULTS

In this study, the outcome is measured several times for a participant and these observations are usually correlated because they come from the same person. So, therefore, the correlation of repeated measurements for each participant is taking into account by including “Week” as the repeated effects variable because this identifies the repeated observations. The corresponding covariance structure could be the first-order autoregressive (AR1) covariance structure since this structure is mostly applied to longitudinal data (Hedeker & Gibbons, 2006). However, whether this is the best structure to choose from will be tested by examining different covariance structures in order to decide which structure better fits the data for models 1 and 2 (see table 6 for the comparisons). Based on this comparison, the model with the diagonal covariance structure is selected for both models (1 and 2), these choices are based on the Akaike information criterion (AIC) index in which the smallest AIC is preferred, regardless of the amount of parameters (Heck et al., 2014).

Table 6 – Comparison Models (AIC)

Model 1a _{Model 2}b

Covariance Structure AIC Parameters AIC Parameters First-order autoregressive 4333,286 20 4337,867 24

Diagonal 4328,707 21 4334,030 25

Scaled Identity 4333,957 19 4337,507 23 a. Time-varying predictors: type of promotion.

Before conducting the Linear Mixed Models to test the hypotheses, first the pattern of correlations among the various predictors and the outcome will be examined for the two models (The correlation matrixes are reported in appendix B in table 7 and 8, see page 59). Price discounts correlated positively with food waste (r = .238, p < .01) and quantity-based promotions (r = .357, p < .01), but this relationship is rather weak than strong. This is the same for the relationship between quantity-based promotions for vices and food waste, since the correlation is below .20. Furthermore, also the positive correlations between price discounts for virtues and food waste can be considered as weak (r = .275, p < .01). This means that the value of food waste only slightly increases in response to an increase in the amount of virtues bought with a price discount. In sum, both matrixes shows that the correlations in the two models are rather weak than strong since the correlations are below .60. Therefore, it is expected that there are no problems with multicollinearity. But to make sure that collinearity is not a problem the Variance Inflation Factors (VIF) will be computed. The VIFs of both models had values below 2 which suggest that there is no problem with collinearity since the values are smaller than 5 (Leeflang, Bijmolt, Pauwel & Wieringa, 2014).

Results - Models 1 and 2

The interpretation of the results for models 1 and 2 is central, but first the model fit will be examined with the help of the -2* log likelihood. Both models will be compared to the null model and this test reveals that the model with predictors (Model 1: 4317,846 – 4286,707 = 31,139 > χ2 _{26,296 with p = .05 for 16 degrees of freedom; Model 2: 4317,846 – 4284,030 =}

33,816 > χ2 _{31,410 with p = .05 for 20 degrees of freedom) is favored over the null model, so}

the results can be interpreted. In contrast with H1, the results suggest that price discounts have a significant positive effect (B = 72,95 , p = .009 < .01) on consumers’ food waste, while the effect of quantity-based promotions is insignificant (p = .433 > .10). This implies that products bought with a price discount will lead to more overall food waste. Interestingly to note, is that consumers are not aware of this effect since the majority of the sample is not convinced that their household food waste is due to price discounts (79,6%).

effect of quantity-based promotion for virtues (p = .970 > .10) on consumers’ food waste, so H3 is not confirmed. This result implies that consumers’ waste more food when virtue food is bought with a price discount. Furthermore, in both models, household size has a positive effect on food waste (Model 1: B = 255,30, p = .004 < .01; Model 2: B = 258,92, p = .003 < .01), which means that the number of people in a household contributes to the amount of food wasted. See table 9 for the results of both models.

Table 9 – Estimates of Fixed Effects (Models 1 and 2)

Model 1a _{Model 2}a

Parameter B SE p-value Parameter B SE p-value

Intercept 211,081018 576,052852 .715 Intercept 264,203045 576,824321 .648 PDc _{72,949510 27,667938 .009*** QP-Virtue}d _{3,770605 100,821792 .970} QPd _{39,966743 52,056448 .443 QP-Vice}d _{77,516912 125,618229 .538} Income (1) -542,950777 463,915824 .245 QP-Neitherd _{105,439717 138,249235 .447} Income (2) 611,589284 445,849581 .174 PD-Vicec _{69,247377 80,699222 .392} Income (3) 52,953360 303,021371 .862 PD-Virtuec _{122,083762 45,722449 .008***} Income (4) -188,528934 298,684383 .530 PD-Neitherc _{6,481382 58,062469 .911} Income (5) -245,181199 262,309110 .352 Income (1) -542,490109 465,373809 .247 Income (6) -115,076643 249,116664 .645 Income (2) 588,780218 446,320978 .191 Income (7) 250,629207 272,128438 .360 Income (3) 27,162879 303,343355 .929 Income (8) 0b ₀b _{. Income (4) -208,646551 298,851352 .487} Age -2,488090 5,076233 .625 Income (5) -250,272978 262,655493 .343 Gender 3,162593 210,043884 .988 Income (6) -141,792988 250,441979 .573 Education 48,952290 53,897387 .366 Income (7) 226,375964 273,277059 .410 Household 255,362168 85,805602 .004*** Income (8) 0 0 . Attitudes PDc _{16,135964 62,688345 .797 Age -2,503753 5,076055 .623} Attitudes QPd _{-47,807397 53,343559 .373 Gender -6,109737 210,619287 .977}

Intention Stockpiling 15,732743 57,087968 .784 Education 48,579926 54,119749 .372 Household 258,921076 85,831931 .003*** Attitudes PDc _{10,506954 62,774771 .867}

Attitudes QPd _{-45,812607 53,553573 .395}

Intention Stockpiling 11,079530 57,247479 .847

a. Dependent Variable: Overall Food Waste.

b. This parameter is set to zero because it is redundant. c. PD = Price Discount.

d. QP = Quantity-based Promotion.

***. significant at the 1% level

Results Lagged Effect Models

(AIC) indicates that the lagged effect models (Model 3: AIC = 2875,164; Model 4: 1452,628) better fit the data than the null model (AIC = 4327,846).

The lagged predictors in model 3 are significant for price discounts (B = 67,09, p = .031 < .05) but insignificant for quantity-based promotions (p = .492 > .05), which means that the price discounts purchased one period ago has an positive effect on consumers’ food waste in the following period. In addition, it has been found that the effect of promotions from two periods ago is insignificant for quantity-based promotions (p = .465 > .10) but significant for price discounts (B = 104,75, p = .014) on consumers’ food waste in the next period. This means that the items bought with a price discount in week 1 have a positive effect on consumers’ food waste in week 3 (see table 10 for the results).

Table 10 – Estimates of Fixed Effects (Lagged Effects Models T-1 and T-2)

Model 3a _{Model 4}a

Parameter B SE p-value Parameter B SE p-value Intercept 698,706111 88,136630 .000*** Intercept 620,573786 100,836032 .000*** QPb _{7,580871 62,409653 .903 QP}b _{1,537689 105,742383 .988}

PDc _{10,800161 38,051452 .777 PD}c _{-39,214712 72,969979 .592}

QP T-1b _{38,493472 55,841305 .492 QP T-2}b _{77,016622 104,944212 .465}

PD T-1c _{67,090935 30,740564 ,031** PD T-2}c _{104,750369 41,740367 .014**} a. Dependent Variable: Overall Food Waste

b. QP = Quantity-based Promotion. c. PD = Price Discount.

**. significant at the 5% level ***. significant at the 1% level

Table 11 – Estimates of Fixed Effects (Lagged Effects Models T-1 and T-2)

Model 5a _{Model 6}a

Parameter B SE p-value Parameter B SE p-value

Intercept 701,112399 88,675431 .000*** Intercept 593,055456 97,537183 .000*** QP-Virtueb _{10,501386 143,142406 .942 QP-Virtue}b _{-674,449477 657,071519 .307} QP-Viceb _{-65,618604 184,630047 .723 QP-Vice}b _{417,269488 977,555840 .671} QP-Neitherb _{94,508389 180,824015 .602 QP-Neither}b _{109,185561 264,443974 .681} PD-Vicec _{-16,489146 123,191481 .894 PD-Vice}c _{-107,938877 283,959099 .705} PD-Virtuec _{47,036130 60,219307 .436 PD-Virtue}c _{-70,838373 134,407711 .599} PD-Neitherc _{-7,034690 80,425915 .930 PD-Neither}c _{296,358598 200,388053 .143} QP-Virtueb _{T-1 -10,142071 116,644640 .931 QP-Virtue}b _{T-2 5,076738 199,227991 .980} QP-Viceb _{T-1 12,125110 134,912746 .928 QP-Vice}b _{T-2 525,360144 230,484643 .025**} QP-Neitherb _{T-1 107,730137 191,487629 .575 QP-Neither}b _{T-2 -419,378499 291,565934 .154} PD-Vicec _{T-1 154,474155 87,809584 .081* PD-Vice}c _{T-2 -52,991464 117,710553 .654} PD-Virtuec _{T-1 82,137616 53,187564 .125 PD-Virtue}c _{T-2 135,528523 69,653953 .055*} PD-Neitherc _{T-1 -28,631062 71,561795 .690 PD-Neither}c _{T-2 140,816847 95,291268 .143} a. Dependent Variable: Overall Food Waste

b. QP = Quantity-based Promotion. c. PD = Price Discount.

*. significant at the 10% level **. significant at the 5% level ***. significant at the 1% level

An overview of the supported and non-supported hypothesis are provided in table 12. In sum, in this study some opposite effects are found. For example, price discounts have a greater effect on consumers’ food waste, while the effect of quantity-based promotions was insignificant. But also, consumers’ waste more food when virtues are bought with a price discount instead of with a quantity-based promotion.

Table 12 – Overview (Non)-Supported Hypotheses

Hypotheses Supported

H1: Quantity-based promotions have a greater effect on consumers’ food waste than price discounts.

X

H2: Price discounts for vices will lead to more overall food waste than quantity-based promotions for vices.

X

H3: Quantity-based promotions for virtues have a larger positive effect on consumers’ waste than price discounts for virtue food.

X

7. DISCUSSION AND RECOMMENDATIONS

consumers’ food waste behavior. This study revealed that consumers waste more food when a product is on promotion, however this effect differs per type of promotion (e.g. price discounts and quantity-based promotions) and per food category on promotion (e.g. vice- and virtue food).

Consumers’ overall food waste is affected by price discounts, which implied that products bought with a price discount will lead to more overall food waste. However, it was expected that quantity-based promotions had a greater effect on consumers’ food waste than price-based promotions, but this study did not found a significant effect for quantity-based promotions. This was somewhat surprising but it is assumed that this effect might be influenced by consumers’ consumption patterns since 73,87% of the items in this study have been bought with a price discount. This amount can be explained by Darke and Chung (2005) because price discounts are the most commonly used form of sales promotions and these promotions (monetary) are easier noticed than nonmonetary promotions (such as “Buy one, get one free”), even when the monetary savings are small (Campbell & Diamond, 1990). These monetary savings allow consumers to spend the extra money, that have been saved with the price promotion, on buying other products (Palazon & Delgado-Ballester, 2009) and this might result in buying more products than planned. Especially since consumers have the tendency to wait for good deals in order to buy larger quantities of a certain item (Mela et al., 1998). This tendency will likely result in an acceleration of product purchases (Neslin et al., 1985) and this might affect consumers’ food waste because stockpiling results in a growth in consumers’ inventories and according to Chandon and Wansink (2006), consumers’ inventory estimations are seldom accurate and these miscalculations can lead to overstocking and food spoilage.

Although the type of promotion had an influence on the spoilage of food, this relationship is somewhat different for a specific product category on promotion. This study only found a positive effect for virtue food purchased with a price discount on the amount of food wasted, while the effect of quantity-based promotions for virtues on food waste was not significant. But also the effect of price discounts for vices on the overall food waste was not significant. These results were unexpected given consumers’ specific preferences for the promotional- and food type (Mishra & Mishra, 2011). So, in theory the specific product category on promotion (e.g. price discounts for vices and quantity-based promotions for virtues) would have had a positive effect on food waste because of consumers’ preferences but in practice it did not had the intended outcome and this could be influenced by the supermarkets because they offer both type of promotions for vice- and virtue food.

For example in this study, virtue food is the most popular item bought on promotion (45,6%, N = 171), whereas the promotions for vices only has a percentage of 24,8 (N = 93). In addition, 74,85% (N = 128) of the virtues are bought with a price discount and 25,15% (N = 43) is purchased with a quantity-based promotion, while for vices 66,67% (N = 62) is bought with a price discount and 33,33% (N = 31) is purchased with a quantity-based promotion. These numbers suggest that most of the food is purchased with a price discount and this high number can be explained by the fact that price discounts is the most popular type of promotion used by stores (Darke and Chung, 2005).

Implications for Policy Makers

This study revealed that supermarket promotions have a share in consumers’ food waste and this finding is interesting for policy makers who are striving for a world in which less food is wasted. These policy makers such as SIRE could educate consumers about the influence of promotions on their share of food waste. Especially, since people believe that quantity-based promotions is the main driver of their household waste, while this study revealed that price discounts is the key contributor to food wastage. This is important to emphasize because as opposed to previous studies (WRAP, 2011; Caswell, 2008; Cox & Downing, 2007), consumer preference for quantity-based promotions and the assumed effect on food waste does not hold in practice since the measurement of consumers’ actual food waste behavior indicates that only price discounts affect households’ waste.

Moreover, it is also important to raise awareness about the delay between the consumption of price discounts and consumers’ food waste since consumers might not be aware of this effect. Raising awareness might help in reducing the amount of products bought on promotion and this is important since today approximately one-third of all food and beverages are on promotion in supermarkets (Williams, Gill & Whitehead, 2014). This increased amount not only attracts consumers’ attention more (Ellickson & Misra, 2008) but it is also harder for consumers to resist since they are surrounded by promotions. Therefore, it is important that consumers become aware of the role of supermarkets price discounts on their share of food waste.

Limitations and Future Research

willing to participate in a three-week study. This challenge was known beforehand because the food diary is a time-consuming method for participants.

So, for future research it is recommended to partner up with a supermarket (or supermarkets) to acquire shopping trip data in order to alleviate some of the participants’ burden such as writing down what type of food has been bought on promotion on which day. But in return participants are requested to shop with a scanner in order to acquire this data. The shopping trip data will provide accurate information on when and what type of food have been bought but also whether or not the purchase was made on promotion. Not only will this eliminate the mistakes of specifying the promotions according to the study guidelines but participants will now have only one task, and that is recording the amount of food wasted. So, this might increase the number of participants who are willing to participate in this study but it may also result in a decrease in the dropout rate.

REFERENCES

Ailawadi, K.L., & Neslin, S.A. 1998. The effect of promotion on consumption: buying more and consuming it faster. Journal of Marketing Research, 35(3): 390-398.

Alavi, S., Bornemann, T., & Wieseke, J. 2015. Gambled price discounts: a remedy to the negative side effects of regular price discounts. Journal of Marketing, 79(2): 62-78.

Alba, J.W., & Williams, E.F. 2013. Pleasure principles: a review of research on hedonic consumption. Journal of Consumer Psychology, 23(1): 2-18.

Aydinili, A., Bertini, M., & Lambrecht, A. 2014. Price promotion for emotional impact.

Journal of Marketing, 78(4): 80-96.

Bell, D.R., Iyer, G., & Padmanabhan, V. 2002. Price competition under stockpiling and flexible consumption. Journal of Marketing Research, 39(3): 292-303.

Blattberg, R.C., Briesch, R., & Fox, E.J. 1995. How promotions work. Marketing Science, 14(3): 122-132.

Blattberg, R.C., Eppen, G.D., & Lieberman, J. 1981. A theoretical and empirical evaluation of price deals for consumer nondurables. Journal of Marketing, 45(1): 116-129.

Blattberg R.C., & Neslin S.A. 1989. Sales promotion: the long and the short of it. Marketing

Letters, 1(1): 81-97.

Bolton, L.E., & Alba, J.W. 2012. When less is more: consumer aversion to unused utility.

Journal of Consumer Psychology, 22(3): 369-383.

Burnett, M.S., & Lunsford, D.A. 1994. Conceptualizing guilt in the consumer decision-making process. Journal of Consumer Marketing, 11(3): 33-43.