Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
Consumer purchase behaviour and date labelling of food: Selection behaviour as a contributor to food waste
An investigation of consumers’ tendency to select food items with a longer expiration date in retail stores
Bo van Geel 290878 (ANR)
Master Thesis
MSc Marketing Analytics July 2018
CentERdata
Submission date: July 31st, 2018 Master of Marketing Analytics (MSc MA) Tilburg School of Economics and Management Tilburg University Master’s Thesis supervisors: dr. M. Elsen (CentERdata & Tilburg University) dr. A.M.M. Bosmans (Co-reader Tilburg University)
Management Summary
A substantial amount of food produced for human consumption is wasted nowadays, which entails
negative economic, environmental, and ethical consequences (Gustavsson, Cederberg, & Sonesson,
2011). It is presumed in the literature that consumers’ tendency to select food items with a longer or the
longest expiration date in retail stores contributes to food waste. This particular consumer purchase
behaviour has been referred to as date labelling selection behaviour. In two experimental studies, it is
investigated (I) how big the proportion of consumers is that engage in this selection behaviour, (II)
which consumer motives underlie this behaviour, (III) when consumers are more/less likely to engage
in this behaviour, and (IV) what the retailer could do about it. Based on the empirical results it can be
concluded, with respect to limited shelf life food, that almost two-thirds engaged in date labelling
selection behaviour. This finding is in line with the existing literature suggesting that consumer’s date
labelling selection behaviour is omnipresent and therefore problematic. Consumer’s primary reason for
engaging in selection behaviour was that this would give them more time to consume the food and
therefore more flexibility. Consumers turned out to be less inclined to engage in date labelling selection
behaviour when they had the intention to consume the food on the same day as the day on which the
food was purchased, when they are aged fifteen to twenty-four, are higher educated, have a higher net
monthly household income, and have bigger households. Unfortunately, it can be concluded that the
first attempt at finding an effective solution against date labelling selection behaviour has failed. The
proposed retailer interventions – messages that challenge the underlying consumer motives for selection
behaviour, printed on sticky labels attached to food items in front of the shelf – turned out not to have
the desired negative effect on date labelling selection behaviour. Further research is necessary in order
to find and test successful solutions against consumer’s selection behaviour.
Keywords: Date labels, expiration date, retailer-generated food waste, consumer motives, behavioural
interventions, date labelling selection behaviour
Preface
In February 2018 I started as a graduate intern at CentERdata to write my Master thesis. This Master
thesis represents the final part of my Master Marketing Analytics at Tilburg University. The Master
Marketing Analytics suited my interests perfectly and I studied all its facets with great pleasure. The
same goes for my Master thesis: I enjoyed studying and investigating consumer’s purchase behaviour
in food retail stores. I have always been interested in the way consumers act, and it felt satisfactory and
rewarding to connect this to the social issue of food waste. If you do something you like, it often will
not feel like work.
Nonetheless, while writing my thesis it sometimes felt that there were not enough hours in the day. This
thesis would not have been possible in the way as it is now without Millie Elsen. Millie’s expertise,
feedback and guidance were of great value, and I hereby would like to thank her for that. Further thanks
go out to everyone else who helped make this thesis possible. I would like to thank CentERdata for the
internship opportunity, the pleasant working atmosphere, and the facilities to work with. It was a
pleasure to be part of CentERdata. Furthermore, I would like to thank my family and friends for their
support and encouragement. I hope this Master thesis might be helpful to some extent in the battle
against food waste.
I am excited to see what the future has in store for me,
Bo van Geel
Tilburg, July 2018
Table of Contents
Management Summary ....................................................................................................................................
Preface ..............................................................................................................................................................
Table of Contents .............................................................................................................................................
Introduction and Problem Indication ............................................................................................................ 1
1.1 – Introduction and Problem Background ................................................................................................. 1
1.2 – Problem Indication and Statement ........................................................................................................ 3
1.3 – Conceptual Model ............................................................................................................................... 6
1.4 – Thesis Structure ................................................................................................................................... 7
Study 1 ............................................................................................................................................................ 9
2.1 – Purpose ............................................................................................................................................... 9
2.2 – Theory and Predictions ........................................................................................................................ 9
2.3 – Methodology ..................................................................................................................................... 13
2.3.1 – Experimental Design and Participants ........................................................................................ 13
2.3.2 – Materials.................................................................................................................................... 14
2.3.3 – Procedure and Measures............................................................................................................. 15
2.4 – Results .............................................................................................................................................. 17
2.4.1 – Main Analysis ............................................................................................................................ 17
2.4.2 – Follow-up Analysis: Consumer Profiling .................................................................................... 21
2.5 – Preliminary Conclusion ..................................................................................................................... 23
Study 2 .......................................................................................................................................................... 25
3.1 – Purpose ............................................................................................................................................. 25
3.2 – Theory and Predictions ...................................................................................................................... 25
3.3 – Methodology ..................................................................................................................................... 27
3.3.1 – Experimental Design and Participants ........................................................................................ 27
3.3.2 – Manipulation of the Type of Message ......................................................................................... 28
3.3.3 – Materials.................................................................................................................................... 28
3.3.4 – Procedure and Measures............................................................................................................. 30
3.4 – Results .............................................................................................................................................. 31
3.4.1 – Main Analysis ............................................................................................................................ 31
3.4.2 – Follow-up Analysis: Moderators of the type of message effect .................................................... 32
3.5 – Preliminary Conclusion ..................................................................................................................... 37
Conclusions and Recommendations ............................................................................................................. 39
4.1 – General conclusion and discussion ..................................................................................................... 39
4.2 – Recommendations ............................................................................................................................. 41
4.3 – Limitations and suggestions for further research ................................................................................. 41
References .................................................................................................................................................... 43
Appendix 1 – Questionnaire Study 1 ........................................................................................................... 47
Appendix 2 – Results Study 1 ....................................................................................................................... 49
Appendix 3 – Instruction Study 2 ................................................................................................................ 51
Appendix 4 – Post-Experiment Questionnaire Study 2 ............................................................................... 53
Appendix 5 – Results Study 2 ....................................................................................................................... 55
Appendix 6A – Follow-up Analysis Study 2 with ‘time’ .............................................................................. 59
Appendix 6B – Follow-up Analysis Study 2 with ‘intention to avoid food waste’ ....................................... 61
Appendix 6C – Follow-up Analysis Study 2 with ‘noticing the message labels’ .......................................... 65
1
Introduction and Problem Indication
1.1 – Introduction and Problem Background
In this era of prosperity, consumerism and abundance of choice in Western societies, food waste has
become a noticeable part of everyday life. In both Europe and the United States, the amount of available
food per capita in retail stores, supermarkets and restaurants has increased rapidly during the last
decades (Gustavsson, Cederberg, & Sonesson, 2011). In addition, contemporary consumers can simply
afford to waste food. The abundance of food, together with consumer attitudes, leads to substantial food
waste and disposal nowadays (Gustavsson et al., 2011).
Although almost impossible to obtain precise figures, United Nations’ Food and Agriculture
Organization (2013) states that approximately one-third of produced food for human consumption is
lost or wasted globally nowadays, which results in 1.3 billion ton of disposed food per year. The amount
of food losses and waste per capita are the highest in North America and Oceania (almost 300 kg/year),
followed by Europe (280 kg/year) according to Figure 1 (Gustavsson et al., 2011, p. 5).
Food is lost and wasted throughout the entire food supply chain; from agricultural production to
postharvest treatment, to processing and distribution to catering and retailers, and eventually at
household consumption (Gustavsson et al., 2011). The food supply chain can best be described as the
connected activities used to produce, process, distribute and consume food (Stenmarck, Jensen,
Quested, & Moates, 2016). Whereas food losses occur primarily at the production, postharvesting and
processing stage of the food supply chain, food waste occurs near the end of the food supply chain at
the caterer, retailer, and consumption stage (Parfitt, Barthel, & Macnaughton, 2010). Food waste can
be described as food fit for human consumption that is nonetheless unconsumed (Parfitt et al., 2010).
Figure 1 - Per capita food losses and waste, at consumption and pre-consumption stages, in different regions. Reprinted from “Global Food Losses and Food waste” by J. Gustavsson, C. Cederberg, U. Sonesson, 2011, Food and Agriculture Organization of the United Nations, 37, p. 5.
2
In medium- and high-income countries1, food is to a great extent wasted as opposed to lost (Gustavsson
et al., 2011). There are multiple serious negative consequences of food waste (Graham-Rowe, Jessop,
& Sparks, 2014). Food waste is, to start, obviously a waste of money and good food (Gustavsson et al.,
2011). Since more value in terms of money and resources is added for every step in the food supply
chain, food waste entails a substantial loss of value (Eriksson, 2015). From an ethical perspective, food
waste has a substantial moral impact since a large part of the world’s population is starving and/or
malnourished (Williams, Wikstrom, Otterbring, Lofgren, & Gustafsson, 2012). According to the Food
and Agriculture Organization (2013), the food that is currently wasted in Europe could feed 200 million
people. In addition, food waste increases global food prices and therefore makes food less accessible
for people in underdeveloped countries (Graham-Rowe et al., 2014). Furthermore, food waste is also
harmful to the environment. Eventually, a huge amount of natural resources in food production – such
as land, water and energy - are used for nothing. Greenhouse gas emissions caused by the production
of food that in the end gets wasted, are eventually emissions in vain (Gustavsson et al., 2011; Graham-
Rowe et al., 2014). The volume of food waste should be downsized in order to sustain the world’s
limited natural resources and to ensure enough food for all humans (Williams et al., 2012).
Considering the negative economic, environmental, and ethical consequences of food waste, the
question should be raised and addressed by whom, and on what grounds, so much food intended for
human consumption is wasted nowadays. Food wastage clearly is a complex issue with multiple factors
impacting disposal (Aschemann-Witzel, de Hooge, Amani, Bech-Larsen, & Oostindjer, 2015). It
appears that, in Western societies, both the retail and consumption stage of the food supply chain play
a significant role in the problem of food waste (Aschemann-Witzel et al., 2015). Particularly, the retail
stage appears to be an important pawn as regards the problem of, and in the fight against, food waste.
Food retailers play a considerable role in causing food waste, as well as in offering potential solutions
to lower food waste at the same time given their crucial position at the food supply chain - consumer
interface (Aschemann-Witzel, de Hooge, & Normann, 2016). Food retailers have the tools and power
to exert substantial influence in shaping both the features of food production upstream, as well as the
purchase preferences and behaviour of consumers downstream via e.g. certain marketing tactics
(Cicatiello, Franco, Pancino, Blasi, & Falasconi, 2017).
The share of European food losses and waste attributable to the retail stage is around 5 percent
(European Commission, 2015). In the Netherlands, the retail sector is accountable for approximately 7
percent of total food disposal (Soethoudt, Vollebregt, & van der Burgh, 2016). The Netherlands
currently has around 4300 supermarkets located across the country (Centraal Bureau
Levensmiddelenhandel, 2018). Annually, Dutch supermarkets discard food worth around 400 million
euro (De Vocht, 2018). Food waste recording procedures differ considerably between retailers, and
1 Europe (including Russia), the United States, Australia, New Zealand, China, Japan, and South – Korea (Gustavsson et al., 2011)
3
food waste related information is most of the time kept private (Parfitt et al., 2010). Country- and
supermarket-specific case studies best represent the current literature on the quantification of food waste
in the retail sector, combined with an attempt to extrapolate the findings to the bigger picture
(Lebersorger, & Schneider, 2015; Cicatiello, Franco, Pancino, & Blasi, 2016). Based on multiple
studies, (semi-)perishable food with a short to limited shelf life accounts for the highest proportion of
food waste in the retail sector (Parfitt et al., 2010). Among the food product categories analysed, bread,
fresh fruit and vegetables, meat, and dairy products stand out in terms of retailer-generated food waste
(Cicatiello et al., 2017). At the bottom of retailer’s disposal list there is canned, dried and frozen food
all with a long shelf life (Stenmarck, Hanssen, Silvennoinen, Katajajuuri, & Werge, 2011). But what
makes this food wastage? There are multiple routes that cause food waste in retail stores, either relating
to stock management practices as well as to purchasing behaviour of consumers (Cicatiello et al., 2017;
Gustavsson et al., 2011). Part of retailer-generated food waste is caused by a mismatch between supply
and demand due to misjudgment of the retailer, leaving abundant products unsold (Schripsema, van der
Burgh, van der Sluis, & Bos-Brouwers, 2015). Notably, there is always some degree of surplus stock
in supermarkets (Storup, Mattfolk, Voinea, Jakobsen, Bain, Reverte, & Oliveira, 2016). However, the
mismatch between supply and demand has grown relatively smaller over the years due to strongly
improved, data-driven computer-generated mechanisms to predict demand and to keep track of stock
records. In addition, consumer behaviour and preferences in-store are also strong drivers of food waste
at the retail level (Parfitt et al., 2010). For example, perceived suboptimal food by consumers – e.g.
damaged food, food not perfectly shaped, or food with the shortest remaining shelf life – is often left
unsold and wasted in the end in retail stores (Storup et al., 2016). The consumer role with respect to
food waste in retail stores is very decisive, while at the same time susceptible to influences from food
retailers (Aschemann-Witzel et al., 2015). Therefore, the retailer-consumer interface is a valuable field
to focus upon in the battle against food waste (Aschemann-Witzel et al., 2015).
1.2 – Problem Indication and Statement
Date labelling of food products has been practiced for decades, and it is used as a communication tool
to achieve stock rotation at retail while at the same time providing information to consumers (Parfitt et
al., 2010). Date labels on food are either based on product quality (best before date, also known as “ten
minste houdbaar tot datum” in Dutch), or on product safety (use-by date, also known as “te gebruiken
tot datum” in Dutch). Quality date marks represent a date by which the product should be consumed for
ideal or optimal quality; it is a date of minimum durability and after that date the food quality starts to
deteriorate from its peak. Such food products can be consumed after the printed date, it is then
encouraged to use one’s own senses to determine product quality and freshness before simply throwing
it away (Newsome et al., 2014). On the contrary, safety date marks represent an expiry date after which
4
the product must be discarded due to health safety reasons. The printed date stands for the endpoint of
food in terms of ‘microbiological load’ rather than consumer taste (Milne, 2012). Despite the good
communicative intentions, the food wastage discussion has pointed to date labelling as one of the
biggest contributors to the unnecessary waste of edible food (Milne, 2012; Newsome et al., 2014).
Approximately 10 percent of total food waste at the caterer, retailer and consumption stage is solely
attributable to date labelling of food (European Commission, 2018). Considering retail stores, studies
have found that the majority of its food waste is ascribable to date labelling (Garrone, Melacini, &
Perego, 2014). Problematic is that food retailers are not allowed to sell food products that have passed
their date mark, so expired food products will inevitably be disposed and removed from the shelves
regardless of their current, actual quality and edibility (Storup et al., 2016). This prohibition to sell
expired food is regulated by law and therefore difficult to counter. Although disposed food is to some
extent donated to local foodbanks, used for animal feed, used for biogas, or used as fertilizer in the form
of compost, the majority of disposed food in supermarkets is actually wasted (Schripsema et al., 2015).
However, food waste as a result of date labelling is not solely attributable to the retailer because they
are not allowed to sell expired food; food waste as a result of date labelling is also ascribable to the
consumer. Date labelling of food influences consumer purchase behaviour in retail stores; it influences
consumer preferences with respect to which food product they want and – more importantly with respect
to food turning into waste – not want to buy. Literature suggests that many consumers regularly look at
and make use of date labels to guide purchase decisions, and it turns out that these decisions often
contribute to (even more) food waste (Lyndhurst, 2011). To elaborate, a ‘first in, first out’ (FIFO)
issuing policy is most often in place as regards retailer’s stock management and product display in-
store. The FIFO principle is regularly used as a shelf-filling system for food products that are subject to
decay (Ferguson, & Ketzenberg, 2006). When stocking and replenishing store shelves based on the
FIFO principle, new food items are added at the rear end of the shelf or at the bottom of a pile. FIFO
issuing has the objective of minimizing the amount of outdated food (Ferguson, & Ketzenberg, 2006).
However, it turns out that food items closest to expire, and thus in front of the shelf or at the top of a
pile, are often ignored by consumers (Gustavsson et al., 2011). Retailers stock food products with
multiple expiration dates which has the implication that food items closest to expire are often ignored
by consumers, with some of these ignored food items turning into waste eventually (Hug, Asnani, Jones,
& Cutright, 2005). It can be argued that date labels influence consumer’s acceptance and desirability of
food items in-store (Newsome et al., 2014). A study conducted by Lyndhurst (2011) shows that 87
percent of participants indicated to refer to date labels while shopping, and 39 percent of participants
indicated to look for the longest date available while shopping. To be more precise, the relationship
between date labelling of food and consumer behaviour during the retailer-consumer interface is
particularly reflected in ‘selection behaviour’, implying that consumers will often opt for the product
on the shelf with the latest expiry date (Soethoudt, Sluis, Waarts, & Tromp, 2013). This ‘selection
5
behaviour’ relates to sorting out food items based on the length of their remaining shelf life, and
subsequently buying food items with a longer, or the longest, remaining shelf life. Consumer preference
for younger products often involves reaching for products at the back of a shelf or at the bottom of a
pile (Newsome et al., 2014). This somewhat peculiar behaviour has been referred to as ‘date labelling
selection behaviour’ or ‘sorting dillema’ (Newsome et al., 2014), or ‘datumduiken’ as we named it in
Dutch. The problem with this date labelling selection behaviour of consumers – under the assumption
that a substantial part of consumers engage in this behaviour – is that it causes older food items to
remain on the shelf for a longer period of time, approaching their date mark as each day goed by thereby
becoming (even) less and less attractive (Soethoudt et al., 2013). Eventually, if these older food items
do not get picked, they will most likely turn into waste (Aschemann-Witzel et al., 2015). In other words,
date labelling selection behaviour increases the likelihood of (older) food items becoming outdated and
turning into food waste. It is expected that a lot could be gained from changing this selection behaviour
of consumers since this would mean reducing the amount of retailer-generated food waste together with
its negative environmental, ethical, and economic consequences.
Surprisingly little is known about date labelling selection behaviour. There are several studies which
suggest that consumers’ tendency to select food items with a longer expiration date contributes to food
waste and is therefore problematic (Gustavsson et al., 2011; Newsome et al., 2014; Southoudt et al.,
2013). Furthermore, a small cluster analysis has been conducted by Lyndhurst (2011) on behalf of the
‘Waste and Resources Action Programme’ (WRAP), where participants were clustered based on the
degree of ‘attention paid to date labels’ in-store and at home. This study showed that 39 percent was
‘date driven’ (always looking for the longest date mark in-store and keeping an eye on date labels at
home), 48 percent was ‘date savvy’ (sometimes looking for the longest date mark in-store and keeping
an eye on date labels at home), and 13 percent was ‘date disengaged’ (Lyndhurst, 2011). However, so
far it has not been empirically studied how big the proportion of consumers exactly is that engage in
date labelling selection behaviour, why consumers do so, when consumers are more likely to do so
etcetera. Because of all this, it should first be determined whether date labelling selection behaviour is
really a problem in the first place by examining the share of consumers that actually engage in this
selection behaviour. If only a relatively small part of consumers engage in date labelling selection
behaviour it does not seem to be a real, far-reaching problem. But if it turns out that a substantial part
of consumers engage in this selection behaviour, it might be valuable to understand and to successively
examine why consumers engage in date labelling selection behaviour, when consumers are ‘more’
versus ‘less’ inclined to engage in date labelling selection behaviour, and - most importantly – what can
be subsequently done by retailers to intervene with this selection behaviour. It is often argued that it
lies within the power of retailers to influence the behaviour of their customers (Young, Russel,
Robinson, & Chintakayala, 2018). Based on the aforementioned, the following research questions are
focused on:
6
1. To what extent do consumers engage in date labelling selection behaviour? (I.e. How big is the share of consumers that engage in date labelling selection behaviour? Is it really a problem?)
2. Why do consumers engage in date labelling selection behaviour? (I.e. Which consumer motives underlie date labelling selection behaviour?)
3. When are consumers more/less inclined to engage in date labelling selection behaviour? (I.e. In which situations is date labelling selection behaviour ‘more’ versus ‘less’ strong?)
4. What can the retailer do about date labelling selection behaviour? (I.e. What can the retailer do to improve consumer acceptance of food items with the shortest remaining shelf life?)
As a final remark, consumers are more likely to opt for younger food items for some particular product
types than for other product types. The literature suggests that the use of date labels by consumers varies
according to product type (Lyndhurst, 2011). Food can broadly be subdivided into 3 categories (van der
Vossen, & van Dooren, 2016):
- Very fresh, short shelf life food products (e.g. meat and prepacked ready-to-eat meals)
- Limited shelf life food products (e.g. packed meat products and dairy products)
- Long shelf life food products (e.g. soda and confectionery)
According to a study conducted by Lyndhurst (2011), the deliberate use of date labels by consumers
when making purchase decisions primarily occurs with (semi-)perishable food products with a short or
limited shelf life. Participants in Lyndhurst’ study reported paying most attention to date labels on meat
and dairy products (especially yoghurts), whereas paying the least attention to date labels on bread, and
fresh fruit and vegetables during the moment-of-purchase in-store (Lyndhurst, 2011). This research
focuses on food products with a limited shelf life with respect to the problem of date labelling selection
behaviour. This focus decision is guided by the general consensus that date labelling selection behaviour
is not really in operation at food products with a long shelf life, together with the accompanied danger
of focusing on fresh food products with a short shelf life. In almost all instances, fresh, highly perishable
food products with a short shelf life carry a ‘use-by’ date mark denoting product safety. After the expiry
date, such products should be discarded due to health safety reasons and therefore this forms a rather
dangerous product category to focus upon. One wants consumers to handle this product category with
care, and behavioural changes with respect to consumer’s date labelling selection behaviour could
jeopardize this.
1.3 – Conceptual Model
Based on the four research questions (denoted by I, II, III and IV) formulated in the previous section,
the following conceptual model is proposed and used as a guide throughout this research (Figure 2).
7
This conceptual model revolves around the assumption that date labelling selection behaviour is
problematic because a substantial part of consumers engage in this selection behaviour. This conceptual
model shows that consumer motives lead to, and are the underlying reason of, date labelling selection
behaviour by consumers. In turn, retailers can impose interventions aimed to weaken or disempower
these consumer motives in an attempt to reduce date labelling selection behaviour. Whether the retailer
interventions are effective, depends on the presence or absence of various situational/contextual factors
which in turn influence consumer motives. The exact content of this conceptual model will be explained
in more detail in next sections in Study 1 and Study 2.
1.4 – Thesis Structure
The structure of this thesis is built around a general framework established by Steg and Vlek (2009)
about encouraging pro-environmental behaviour. One wants consumers to behave pro-environmentally
and to reduce consumer-related food waste in retail stores. Steg and Vlek (2009) propose the following
steps in effectively encouraging and establishing pro-environmental behaviour:
1. Identification of the behaviour to be changed
2. Examination of the main factors underlying this behaviour
3. Design and application of interventions to change behaviour to reduce environmental impact
4. Evaluation of the effects of the behavioural interventions
In addition to this introductory chapter, this thesis will consist of two studies each addressed in separate
chapters, followed by a final, concluding chapter. Study 1, a large-scale online study, will examine the
extent to which consumers engage in date labelling selection behaviour (research question 1) and why
they do so (research question 2). As a follow-up analysis, various consumer types will be profiled. This
first study aims to (inter alia) demonstrate the relevance of this research by investigating the extent to
Retailer Interventions Consumer Motives Date Labelling Selection Behaviour
Figure 2 – Conceptual model
I II IV
III
Situational / Contextual Factors
8
which consumers engage in date labelling selection behaviour. Study 2, a small-scale follow-up study,
will test potential retailer interventions in a first step to reduce date labelling selection behaviour
(research question 4). In addition, both studies will examine situations in which consumers could be
more/less inclined to engage in date labelling selection behaviour (research question 3). The final,
concluding chapter will provide a coherent and comprehensive discussion regarding the four research
questions, and it will also address possible limitations and suggestions for further research.
9
Study 1
2.1 – Purpose
Primarily, this large-scale online study aims to seek answers to the following research questions:
1. To what extent do consumers engage in date labelling selection behaviour?
2. Why do consumers engage in date labelling selection behaviour?
By examining how big the share of consumers is that engage in date labelling selection behaviour, it
will be determined whether this selection behaviour of consumers is really detrimental or not. Because
of the assumption that consumer’s date labelling selection behaviour ís problematic, it will be
simultaneously examined which consumer motives underlie this selection behaviour together with
which motive is most important for consumers. These insights can in turn be used as valuable input for
the design and application of retailer interventions aimed to reduce consumer’s date labelling selection
behaviour (which will be the topic of the second study). In addition, in this first study it will be attempted
to disempower one of the which we believe to be important consumer motives underlying date labelling
selection behaviour, by trying to turn it into a non-sense argument, to see whether that influences the
extent to which consumers engage in date labelling selection behaviour. As a follow-up analysis,
consumer types related to this study will be profiled in terms of socio-demographic characteristics.
2.2 – Theory and Predictions
It is often argued that a substantial amount of consumer behaviour is unconscious and guided by
automated response processes (Dijksterhuis, Smith, van Baaren, & Wigboldus, 2005; Steg, & Vlek,
2009). In fact, much of consumer purchase behaviour in retail stores is said to occur ‘mindlessly’ and
‘automatically’, driven by norms (Cialdini, Kallgren, & Reno, 1991; Dijksterhuis et al., 2005). The
theory of normative social behaviour argues that norms have a strong impact on behaviour and that
norms often offer a decisional shortcut when one is choosing how to behave in a specific situation
(Cialdini et al., 1991). Norms describe what is typical or normal (Cialdini et al., 1991). In the situation
where consumers want to buy a specific food product in retail stores, they can opt for various items on
the shelf. For all that, ‘the norm’ is to pick the food item with the shortest remaining shelf life in front
of the shelf because of the ‘first in, first out’ (FIFO) shelf-filling principle employed by retailers
(Ferguson, & Ketzenberg, 2006). For that reason, many consumers will automatically pick the food
item with the shortest remaining shelf life in front of the shelf because that is the norm-consistent way,
without even thinking about it. However, it is not necessarily rational for consumers to pick the food
item in front of the shelf. As a matter of fact, when consumers would consciously think about which
food item to pick from the shelf, they might find it more desirable to opt for a food item further on the
10
back of the shelf with a longer remaining shelf life (i.e. engage in date labelling selection behaviour).
The rationale behind this will be explained below.
Although automatic, almost thoughtless behaviour is said to be most common, it is expected that a
substantial part of consumers do engage in conscious decision-making when opting for a food item on
the shelf in retail stores, especially with regard to short and limited shelf life food products (Dijksterhuis
et al., 2005). Retailers often stock multiple food items of the same product on the shelf, with varying
date marks (Hug et al., 2005). When consumers want to buy a specific food product in retail stores some
may make a conscious decision as to which food item to choose from the shelf, thereby trading off pros
and cons between food items of the same product with differing date marks. In that case, consumers’
food item choice is likely to be determined by cost-benefit trade-offs (Ratchford, 1982; Ariely, &
Shampan’er, 2004). Each decision alternative has certain benefits associated with it as well as certain
costs (Payne, Bettman, & Johnson, 1991), and individuals will in turn choose alternatives with highest
benefits against lowest costs (Steg, & Vlek, 2009). This cost-benefit analysis may result in behaviour
that deviates from the norm (i.e. deliberately picking a food item from the back of the shelf, with a
longer shelf life) or behaviour that is consistent with the norm (i.e. deliberately picking the first food
item, with a shorter shelf life). But which costs and benefits do consumers perceive with respect to
picking food items with longer versus shorter shelf lifes? It is suggested in the literature that consumers
make use of date labels on food during the product decision-making process in retail stores to ensure
that they (1) will buy the freshest, finest quality food products and/or to ensure that they (2) will buy
food products that will give them the maximum amount of time to use up (i.e. consume) the food
products (WRAP, 2010). Furthermore, it is also suggested that some consumers deliberately pick the
food item with the shortest shelf life because of the altruistic reason of avoiding food waste (WRAP,
2010). Therefore, the following possible costs and benefit could be - fully or only partially - in place
during the trade-off and decision-making process of consumers with respect to whether or not to pick
the food item with the shortest remaining shelf life:
FOOD ITEM IN FRONT OF THE SHELF
COSTS BENEFITS
- A slightly less fresh food product
- Less flexibility with respect to when to consume the food product
- Less food waste
11
In a sense, the cost side tends toward consumers’ self-interest whereas the benefit side is more in interest
of the overall collective good. This proposed cost-benefit trade-off therefore constitutes a clash between
consumers’ self-interest and collective interest; these interests are directly opposing one another. With
many other sustainability-related consumer decisions the self-interest and collective interest go hand in
hand. For example, self-interests and collective interests are aligned with respect to the sustainable
decision to turn off the lights when leaving the house (collective interest) and thereby saving money on
the energy bill (self-interest). Or, to give another example, buying an energy-saving refrigerator
(collective interest) later cuts costs with respect to the energy bill (self-interest). However, in the case
of selecting a specific food item from the shelf, pro-environmental behaviour has no self-interest. In
other words, consumers really have to decide between individual self-interest and the interest of the
community as a whole. The rational choice theory prescribes that people pursue their self-interest
(Miller, 2015). Furthermore, it is assumed by the standard economic theory that people often behave
rationally and therefore self-interestedly (Arrow, 1986). Therefore, when the proposed cost-benefit
trade-off would be the trade-off a consumer makes in real-life, it is assumed that a rational consumer
will be unlikely to opt for the food item with the shortest remaining shelf life. In other words, a rational
consumer is likely to engage in date labelling selection behaviour because self-interest outweighs
collective interest (i.e. costs outweigh the benefits).
Motives are the underlying reasons behind behaviour (Lyndhurst, 2011). So, if one wants to understand
why consumers engage in date labelling selection behaviour, one should examine which consumer
motives underlie this selection behaviour. Consumer motives in favour of date labelling selection
behaviour relate to the cost side of the proposed cost-benefit trade-off (a slightly less fresh food product
and less flexibility with respect to when to consume the food product), whereas consumer motives
against date labelling selection behaviour relate to the benefit side (less food waste). The three possible
consumer motives will now be discussed individually, and it will be tested later in this first study
whether these three motives are indeed important consumer motives in real-life.
I. The argument that food items in front of the shelf are slightly less fresh than food items
further on the back of the shelf, relates to quality perceptions of consumers. With respect
to these quality perceptions, it is assumed that consumers (often) infer greater quality of
food items that have a longer remaining shelf life, even if the difference in the length of the
remaining shelf life is very small (e.g. one or two days) (Aschemann-Witzel, J. Jensen, M.
Jensen, & Kulikovskaja, 2017; Newsome et al., 2014). Because of the perception that the
food item with a longer remaining shelf life is of (much) better quality, it is often preferred
over ‘older’ food items (Tsiros, & Heilman, 2005). However, this consumer perception is
incorrect; in reality quality does not fluctuate much during the shelf life of a food product
(Aschemann-Witzel et al., 2017). Only when the date mark of limited shelf life food
products has passed, the quality starts to deteriorate from its peak (Newsome et al., 2014).
12
The date mark is often set very conservatively because it is a manufacturer’s best guess as
to how long the food item will be at its best quality (Tsiros, & Heilman, 2005). However,
consumers often seem to assume that the difference in quality between a food item with the
shortest shelf life and a food item with a longer shelf life is of substantial difference, even
if the shortest shelf life food item is (often) only slightly ‘older’ and still long before expiry
(WRAP, 2010). This proposed consumer motive will be further referred to as the ‘higher
quality’ motive.
II. The argument that food items in front of the shelf offer less flexibility with respect to when
to consume the food than food items further on the back of the shelf, turns out to be an
often-reported reason by consumers of using date labels on food (Lyndhurst, 2011). This
flexibility argument has to do with the preference of consumers to have the maximum
amount of time possible to consume food (Lyndhurst, 2011). It is about having the option
to consume a food product somewhere ‘in the undetermined future’. Date labels aid
selection of the ‘best’ product in terms of the length of the remaining shelf life so that
consumers would have as long as possible to use food up (Lyndhurst, 2011). With respect
to flexibility, many consumers prefer food items with a longer/the longest remaining shelf
life. At first glance, it does not seem that this flexibility motive suffers from incorrect
consumer perceptions such as the ‘higher quality’ motive does. More flexibility seems to
be a fairly understandable consumer motive underlying date labelling selection behaviour.
However, although the rationale behind this flexibility motive seems well-grounded, there
are situations in which this motive seems less valid. For example, with respect to limited
shelf life food, consumers typically still have at least a week to consume the food item with
the shortest remaining shelf life. Therefore, it is questionable whether more flexibility is
really needed with respect to limited shelf life (semi-perishable) food products. To give
another example, in the situation in which consumers already have the intention to consume
the food product at short notice, the flexibility arguments seems to be a complete non-sense
argument. This proposed consumer motive will be further referred to as the ‘more
flexibility’ motive.
III. The argument that opting for food items in front of the shelf would lead to less food waste
than opting for food items further on the back of the shelf would, relates to consumers’
food waste awareness and importance attached to avoiding food waste. It is assumed that
some consumers are aware of the food waste streams caused by selecting food items with
a longer remaining shelf life, and do therefore select food items with the shortest remaining
shelf life (Theotokis, Pramatari, & Tsiros, 2012). However, it is questionable whether most
consumers actually associate date labelling selection behaviour with food waste. In other
words, are consumers really aware of the benefit side of the proposed cost-benefit trade-off
13
related to the food item in front of the shelf? The extent to which consumers are aware of
the fact that ‘date labelling selection behaviour’ leads to food waste is assumed to be
insufficient (Aschemann-Witzel et al., 2015). In addition, according to WRAP (2010), there
is a general lack of concern about (avoiding) food waste. Because of this, it is expected that
currently many consumers engaging in conscious decision-making do not face a trade-off
at all. It seems likely that consumers are solely aware of the cost side related to the food
item with the shortest shelf life which would make the decision rather easy. To elaborate,
when consumers are unaware of the benefits of picking the food item with the shortest
remaining shelf life, costs will automatically outweigh and have the upper hand which will
make the rational consumer pick the food item with a longer remaining shelf life (i.e. will
make them engage in date labelling selection behaviour). Therefore, it is questionable
whether the proposed cost-benefit trade-off always constitutes a personal, internal trade-
off and conflict. However, it sure represents a wider, social conflict surrounding date
labelling selection behaviour which would ideally be considered by all consumers at all
times. This proposed consumer motive will be further referred to as the ‘less food waste’
motive.
As a final remark, it will be attempted in this first study to weaken the ‘more flexibility’ motive by
trying to turn it into a non-sense argument, to see whether that influences the extent to which consumers
engage in date labelling selection behaviour. It is predicted that by eliminating the need for flexibility,
less consumers will engage in date labelling selection behaviour. Although not of primary focus, this
will be examined in order to see if there are in a sense ‘natural situations’ in which consumers are
more/less inclined to engage in date labelling selection behaviour. It was chosen to manipulate the ‘more
flexibility’ motive (and for example not the ‘higher quality’ motive) because it was intended to find a
natural situation in which a consumer motive underlying date labelling selection behaviour would not
be applicable. In other words, it was intended to find a situation in which consumers would be more/less
inclined to engage in date labelling selection behaviour, without needing (retailer) interventions or
consumer education.
2.3 – Methodology
2.3.1 – Experimental Design and Participants
The study consists of an online questionnaire where the real-life situation is simulated in which a
consumer (i.e. the participant) has to make a choice between which food item to buy in the supermarket,
accompanied by various follow-up questions. The study employs a two-group experimental design with
two conditions (moment of food consumption: ‘unspecified’ versus ‘on the same day of purchase’).
14
Each participant is randomly assigned to one condition. In condition 1, the moment of consumption is
unspecified. In condition 2, the moment of consumption is specified as being on the same day as the
day on which the food item is purchased. In condition 2, consumers’ need for flexibility with respect to
when to consume the food is in a sense eliminated. Because of the random assignment of participants,
any variation in the extent to which participants engage in date labelling selection behaviour and why
they do so can be attributed to the treatment of eliminating the need for flexibility through specifying
the moment of consumption to be on the same day of purchase. The manipulation is reflected in the
instruction each group receives.
For the data collection, primary data was collected through an online questionnaire which was sent to
the CentERpanel2, which represents a good reflection of the Dutch-speaking population. In total, 2023
members of the CentERpanel filled in the questionnaire (Mage = 56.47, SDage = 16.28; male = 53.2%,
female 46.8%). An overview of the sample characteristics can be found in Table 1. In total, 1036
participants were assigned to condition 1 and 987 participants were assigned to condition 2.
2.3.2 – Materials
In order to examine the extent to which consumers engage in date labelling selection behaviour, it is
necessary to simulate in the questionnaire the situation in the supermarket where consumers are about
to buy a specific limited shelf life food product and have to decide which product item on the shelf
2 The CentERpanel is part of the data collecting and research institute ‘CentERdata’. The CentERpanel consists of around 1900 households, who complete questionnaires via the Internet at home on a weekly basis against payment.
Percentage Absolute
Age
15 until 24 2.2 % 45 25 until 34 8.9 % 181 35 until 44 15.5 % 314 45 until 54 16.1 % 326 55 until 65 19.2 % 389 65 and older 38.0 % 768
Gender
Male 53.2 % 1077 Female 46.8 % 946
Education
Primary education 3.5 % 71 VMBO 23.8 % 481 HAVO/VWO 10.3 % 209 MBO 22.9 % 464 HBO 26.0 % 526 WO 13.4 % 272
Table 1 – Sample descriptives Study 1
15
(which one in the row) to pick. Therefore, the participants were exposed to a visual representation of a
supermarket shelf and were subsequently asked to indicate which item they would opt for. The product
selected for this study is yoghurt of 500 ml, based
on the assumption that yoghurt is prone to date
labelling selection behaviour as being a chilled,
limited shelf life food product. The visual
representation of the yoghurt shelf in the
supermarket encompasses seven yoghurts, with the
first two yoghurts having a date mark of the 15th of
March (oldest yoghurts) and the last five yoghurts
having a date mark of the 17th of March (youngest
yoghurts). Each yoghurt was assigned a number
from 1 to 7 (1 = yoghurt in front of the shelf, 7 =
yoghurt on the far back of the shelf). Figure 3 shows
the picture of the yoghurt shelf used in this study.
2.3.3 – Procedure and Measures
Procedure
Participants filled out the questionnaire online. First, participants received the following general
instruction when starting the questionnaire: ‘This questionnaire is about grocery shopping. We are
curious about how you make certain choices in the supermarket. There are no right or wrong answers.’
A more detailed briefing followed this instruction on the next screen, which was different for
participants assigned to condition 1 (moment of consumption ‘unspecified’) versus participants
assigned to condition 2 (moment of consumption ‘on the same day of purchase’):
- Condition 1: ‘You will soon see an image of a supermarket shelf with packages of yoghurt.
Imagine that you are shopping for groceries in the supermarket. You have a shopping list, but
you also just look around for other things. Because you sometimes have appetite for yoghurt in
the morning, you decide to get a pack of yoghurt.’
- Condition 2: ‘You will soon see an image of a supermarket shelf with packages of yoghurt.
Imagine that you are going to make a dessert for tonight for which you will need 500 ml of
yoghurt. You are in the supermarket to buy, among other things, a pack of yoghurt.’
Figure 3 – Picture used in experiment
16
As a final remark, it was mentioned to participants that for the success of the research is was important
to imagine that it was the 3rd of March on the day they filled out the questionnaire. The picture of the
yoghurts on the supermarket shelf was taken on the 3rd of March. Because the whole experiment
revolves around consumer decisions based on date marks, participants were requested to imagine it was
the 3rd of March to make the decision as realistic as possible. In that case, the yoghurt pack with the
shortest remaining shelf life would still have two weeks before reaching its date mark, allowing
sufficient time for consumers to consume the yoghurt (as was the case in the simulated real-life
situation).
Measures
The extent to which consumers engage in date labelling selection behaviour was measured via one
question referring to the picture in Figure 3 (shown on the same screen): ‘If you were planning to buy
a pack of yoghurt [if condition 2: for tonight’s dessert], which one would you take?’ (1 = yoghurt in
front of the shelf, 7 = yoghurt on the far back of the shelf). Participants did not engage in date labelling
selection behaviour if they chose a pack of yoghurt with assigned number 1 or 2 (date mark of March
the 15th), whereas participants did engage in date labelling selection behaviour if they chose a pack of
yoghurt with assigned number 3, 4, 5, 6 or 7 (date mark of March the 17th).
The consumer motives underlying date labelling selection behaviour were measured by various
successive questions as shown in Figure 4 on the next page. First of all it was asked whether participants
made a deliberate decision as to which yoghurt pack to select. If the answer was yes, participants were
subsequently asked why they opted for a particular yoghurt pack. With respect to consumer motives in
favour of date labelling selection behaviour, the extent to which the ‘more flexibility’ motive was an
important motive for consumers is assessed by the answer option ‘It gives me more time to consume
the yoghurt before it reaches its date mark’ whereas the ‘higher quality’ motive is assessed by the answer
option ‘I expect the yoghurt to taste better than the yoghurt with the shorter shelf life because it is
slightly fresher’. On the contrary, with respect to consumer motives against date labelling selection
behaviour, the extent to which the ‘less food waste’ motive was an important motive for consumers is
assessed by the answer option ‘I think it is a shame if the yoghurt packs with a shorter shelf life should
eventually be thrown away’. In addition, with respect to consumer motives against date labelling
selection behaviour, the extent to which consumers understand that the ‘higher quality’ and ‘more
flexibility’ motives are no – at least in this case – true costs of the food item in front of the shelf, is also
assessed. The importance attached by consumers to these motives is assessed by the answer option ‘The
yoghurt has twelve days until it reaches its date mark, which would give me enough time to consume
the yoghurt’, and the answer option ‘I expect the yoghurt to taste as good as the yoghurt with a longer
shelf life’. Noteworthy, the answer option ‘other, namely ..’ allows consumers to add other, personal
motives; in this way it is ensured that all possible consumer motives are incorporated in this study. The
complete questionnaire and accompanying set-up can be found in Appendix 1.
17
2.4 – Results
This results section is subdivided into two separate paragraphs. The first paragraph shows the results of
the main analysis addressing the research questions previously formulated, whereas the second
paragraph show the results of a follow-up analysis profiling several consumer types.
2.4.1 – Main Analysis
To start, some basic descriptive statistics are derived in order to get a first overall impression of the
results. For the sake of clarity, these descriptive results are presented in a flowchart (Figure 5).
Noteworthy, the answers that participants provided when selecting the open-ended ‘other, namely..’
Q1: 'If you were planning to buy a pack of yoghurt [if condition 2: for tonight's dessert], which one
would you take?'
Q2a: 'There were yoghurt packs on
the shelf with a date mark of March the
17th. Did you deliberately choose a yoghurt pack with
a date mark of March the 15th?'
'No I just picked a yoghurt pack in front of the shelf'
'Yes I deliberately picked a yoghurt pack with a
shorter shelf life'
Q3a: 'Why did you deliberately pick a yoghurt pack with a shorter shelf
life?'**choose the answer that applies to
you the most
'The yoghurt still has twelve days until it reaches its date mark, which would
give me enough time to consume the yoghurt'
'I expect the yoghurt to taste as good as the yoghurt with
a longer shelf life'
'I think it is a shame if the yoghurt packs with a
shorter shelf life should eventually be thrown away'
('less food waste' motive)
'Other, namely..'
Q2b: 'There were yoghurt packs on the shelf with a
date mark of March the 15th. Did you
deliberately choose a yoghurt pack
with a date mark of March the 17th?'
'No I just picked a yoghurt pack'
'Yes I deliberately picked a yoghurt pack with a longer
shelf life'
Q3b: 'Why did you deliberately pick a yoghurt
pack with a longer shelf life?'*
*choose the answer that applies to you the most
'It gives me more time to consume the yoghurt before
it reaches its date mark' ('more flexibility' motive)
'I expect the yoghurt to taste better than the yoghurt with
the shorter shelf life because it is slightly
fresher' ('higher quality' motive)
'Other, namely..'
Yoghurt Item 3, 4, 5, 6 or 7
Figure 4 – Flowchart of questionnaire items
Yoghurt Item 1 or 2
18
option to express their underlying motives, were recoded3; it appeared that, while recoding, none of the
provided answers were essentially different from the already existing answer options. Based on the
output of this study, the first, second, and (partially the) third research can now be answered.
‘To what extent do consumers engage in date labelling selection behaviour?’
The descriptive results displayed in Figure 5 provide insight into how big the share is of consumers that
engage in date labelling selection behaviour. The first research question can easily be answered by
comparing, within group 1 (moment of consumption ‘unspecified’), the percentage of participants that
chose a yoghurt pack with assigned number 1 or 2 with the percentage of participants that chose a
yoghurt pack with assigned number 3, 4, 5, 6, or 7. It appeared that, out of 1036 participants, 64.3
percent (666 participants) engaged in date labelling selection behaviour whereas 35.7 percent (370
3 Originally, in condition 1 (moment of consumption ‘unspecified’), out of the participants (not) engaging in date labelling selection behaviour, 2.3 (2.4) percent selection the ‘other’ option. In condition 2 (moment of consumption ‘on the same day of purchase’) out of the participants (not) engaging in date labelling selection behaviour, 6.2 (4.4) percent selected the ‘other’ option.
Moment of Consumption:
Group 1. Unspecified (n = 1036)Group 2. On the same day of
purchase (n = 987)
Picked a yoghurt with the shortest remaining shelf life (March the 15th) with assigned
number 1 or 2
35.7% 54.4%
Non-deliberate choice43.8%40.1%
Deliberate choice*56.2%59.1%
'Still enough time for consumption'
70.2%61.4%
'Equally good taste'2.4%3.1%
'Shame if these yoghurts would be thrown away'
27.4%35.5%
'Other'0.0%0.0%
Picked a yoghurt with a longer remaining shelf life (March the 17th)
with assigned number 3, 4, 5, 6, or 7
64.3%45.6%
Non-deliberate choice2.4%3.1%
Deliberate choice*97.6%96.9%
'More time for consumption'
90.0%86.2%
'Better taste because its fresher' 10.0%13.8%
'Other'0.0%0.0%
Figure 5 – Descriptive results Study 1 * Only one answer possible: ‘choose the answer that applies to you the most’.
19
participants) did not. So almost two-thirds of the participants (i.e. the vast majority) engaged in date
labelling selection behaviour.
‘Why do consumers engage in date labelling selection behaviour?’
The descriptive results displayed in Figure 5 also provide insights with respect to the second research
question as to which consumer motives underlie date labelling selection behaviour. As predicted, with
regard to group 1 (moment of consumption ‘unspecified’) it turned out that almost all participants (97.6
percent) who engaged in date labelling selection behaviour deliberately and consciously did so. In other
words, these 97.6 percent out of all participants engaging in date labelling selection behaviour made a
deliberate choice to pick a yoghurt pack from the shelf with a longer remaining shelf life. With respect
to the proposed underlying motives, it appeared that the ‘more flexibility’ motive was by far the most
important reason (90 percent) for participants to engage in date labelling selection behaviour. In
contrast, only 10 percent of participants argued that they engaged in date labelling selection behaviour
out of ‘higher quality’ reasons. It turns out that, in the general situation of uncertainty with respect to
when to consume the food (i.e. the yoghurt), participants who engage in date labelling selection
behaviour predominantly do so because that will give them more time to consume the yoghurt before it
reaches its date mark. Not surprisingly, unconsciously, non-deliberately picking a yoghurt pack from
the shelf with a longer remaining shelf life represents an almost non-existing subgroup (2.4 percent).
With respect to group 1 (moment of consumption ‘unspecified’), it remarkably turned out that out of
the participants who did not engage in date labelling selection behaviour, merely 43.8 percent indicated
that they ‘just picked a yoghurt pack from the shelf’ when asked whether they deliberately picked a
yoghurt pack in front of the shelf or not. In fact, most participants (56.2 percent) who did not engage in
date labelling selection behaviour, indicated that they deliberately opted for a yoghurt pack in front of
the shelf with the shortest remaining shelf life. In other words, the behavioural act of picking a food
item in front of the shelf appears to be not as automatic and thoughtless as one would expect it to be.
However, we may have turned this consumer behaviour into more conscious and deliberate behaviour
than it actually is in reality. By asking participants directly whether their choice was deliberate or not,
it could be that these participants started to think about their behaviour more than they would normally
do. Nonetheless, picking a food item in front of the shelf turned out to be much less of a conscious and
deliberate act compared to picking a food item further on the back of the shelf (43.8 percent versus 2.4
percent). Interestingly, with respect to the participants who indicated to have deliberately picked a
yoghurt pack in front of the shelf, the ‘less food waste’ motive turned out to be not so salient. When
participants deliberately picked a yoghurt pack in front of the shelf, the ‘less food waste’ motive was
not the most important reason for participants. In fact, the primary, most important reason for
participants to deliberately pick the yoghurt pack in front of the shelf was because that they believed
that yoghurt pack would still give them enough time for consumption before it would reach its date
20
mark (70.2 percent). Furthermore, 2.4 percent deliberately picked a yoghurt pack in front of the shelf
because they expected that that yoghurt pack tasted as good as yoghurt packs with a longer remaining
shelf life on the back of the shelf. It could thus be argued that the link between date labelling selection
behaviour and its contribution to food waste is not yet strong enough, or it could be that participants do
not care enough about food waste.
‘When are consumers more/less inclined to engage in date labelling selection behaviour?’
It was additionally attempted to disempower the ‘more flexibility’ motive by eliminating the need for
flexibility with respect to when to consume the yoghurt, thereby trying to turn the flexibility argument
into a non-sense argument. Are consumers less inclined to engage in date labelling selection behaviour
when the intended moment of consumption is on the same as the day on which the food (i.e. the yoghurt)
is purchased? By looking at the results provided in Figure 5, it seems that the extent to which consumers
engage in date labelling selection behaviour is more profound in case one is uncertain about when to
consume the yoghurt (64.3 percent) than when one has the intention of consuming the yoghurt on the
same day of purchase (45.6 percent). Whether or not consumers engage in date labelling selection
behaviour is significantly determined by the manipulation of the moment of consumption (Wald χ2 (1)
= 70.520, p < .001; Appendix 2). Noteworthy, the elimination of the need for flexibility as a motive has
weakened the extent to which consumers engage in date labelling selection behaviour, but did not lead
to a complete disappearance of this selection behaviour; in group 2 (moment of consumption ‘on the
same day of purchase’) still nearly half of the participants (45.6 percent) engaged in date labelling
selection behaviour. Furthermore, because the need for flexibility was in a sense eliminated, one would
expect that consumers who still engaged in date labelling selection behaviour did so out of the ‘higher
quality’ motive. Remarkably, however, 86.2 percent out of the participants still deliberately engaging
in date labelling selection behaviour (with the intention of consuming the yoghurt on that same day) did
so out of the ‘more flexibility’ motive. In other words, although the ‘more flexibility’ motive is slightly
toned down and a less important reason for participants that had the intention of consuming the yoghurt
on the same day (86.2 percent) compared to participants where the moment of consumption was
uncertain (90 percent) (Wald χ2 (1) = 17.916, p < .001; Appendix 2), 86.2 percent still represents a
substantially high percentage. It turns out that the ‘more flexibility’ motive is a very persistent motive
which is difficult to disempower. Although it was attempted to eliminate the need for flexibility by
trying to turn it into a non-sense argument, it seems that for many participants (86.2 percent) it was
never a non-sense argument. Even though the participants in group 2 were instructed that the intended
moment of consuming the yoghurt was on the same day of purchase, many participants might have kept
21
thinking (for whatever reason) ‘but what if tonight’s dessert does not take place?’4 In that case,
flexibility with respect to when to consume the yoghurt might be desirable.
2.4.2 – Follow-up Analysis: Consumer Profiling
The aim of this follow-up analysis is to profile consumer types related to this study. For instance, what
is the typical profile of a consumer that engages in date labelling selection behaviour in retail stores?
More specifically, in this study three types of consumers appeared that can in turn be profiled:
1. Consumers that unconsciously comply with the norm by automatically selecting the food
item in front of the shelf with the shortest shelf life (Group 1)
2. Consumers that consciously deviate from the norm by deliberately selecting the food item
further on the back of the shelf with a longer shelf life (Group 2)
3. Consumers that consciously comply with the norm by deliberately selecting the food item
in front of the shelf with the shortest shelf life (Group 3)
Consumers in a particular subset often share similar characteristics. There are a number of consumer
characteristics (i.e. segmentation bases) that can be used for profiling and segmenting consumers
(Wedel, & Kamakura, 2012). In fact, there are three main types of segmentation bases; decriptive,
behavioural and benefit bases (Wedel, & Kamakura, 2012). In this follow-up analysis, consumers will
be profiled on a descriptive bases (i.e. on socio-demographic variables) because the CentERpanel
contains a lot of background information from panel members with respect to socio-demographic
characteristics. The three identified consumer groups were profiled on several of these socio-
demographic variables (see Table 2 on the next page). To be more precise, the following variables are
examined (with ANOVA’s5 and crosstabs with Pearson χ2); ‘age’, ‘gender’, ‘education’, ‘net monthly
income in the household’, and ‘number of members in the household’. Age, education, net monthly
income in the household, and number of members in the household significantly differ between the
three consumer groups (see Table 2). Surprisingly, there are no differences in gender between the
groups.
To elaborate, it appeared that consumers over the age of 65 are relatively more likely to deliberately
engage in date labelling selection behaviour than other age categories, whereas consumers between the
age of 15 and 24 are relatively less likely to engage in selection behaviour compared to other age
4 In the open-ended answer option ‘Other, namely ..’, two participants mentioned explicitly that ‘maybe tonight’s dessert will not go on, and then I need more time to consume the yoghurt’ 5 Levene’s test for homogeneity of variances for ‘number of members household’ was significant (p = . 033) implying unequal variances and a Games-Howell post hoc test with respect to Table 2. Levene’s test for homogeneity of variances for the natural logarithm of ‘net monthly income household’ was nonsignificant (p = .410) implying equal variances and a Tukey’s post hoc test with respect to Table 2
22
categories. Consumers between the age of 15 and 24 are relatively more likely to automatically and
unconsciously select the food item in front of the shelf, but still 48.9 percent engaged in date labelling
selection behaviour. Furthermore, it appeared that the higher educated consumers are, the less likely
they are to engage in date labelling selection behaviour and the more likely they are to deliberately and
consciously select a food item in front of the shelf. In addition, consumers that deliberately decide to
select a food item in front of the shelf, often have a higher net monthly household income than
consumers that deliberately decide to select a food item further on the back of the shelf (i.e. engage in
date labelling selection behaviour) and than consumers that automatically select a food item in front of
the shelf. Finally, it appeared that consumers that deliberately decide to select a food item in front of
the shelf, often have slightly bigger households (i.e. a larger number of members in the household) than
consumers that deliberately decide to select a food item further on the back of the shelf. With respect
to bigger households, consumers could argue that the food will be consumed anyways. The consumer
that automatically and unconsciously selects a food item in front of the shelf hardly differs, on this
socio-demographic characteristic of number of members in the household, from the consumer that
deliberately and consciously selects the food item in front of the shelf and further on the back of the
shelf.
GROUP 1
In front of the shelf,
non-deliberate
(n = 378)
GROUP 2
On the back of the
shelf, deliberate
(n = 1116)
GROUP 3
In front of the shelf,
deliberate
(n = 529)
Socio-demographic variables Statistic p
Age 15 – 24 years (%) 31.1 48.9 20.0
Age 25 – 34 years (%) 17.7 54.7 27.6
Age 35 – 44 years (%) 19.7 51.9 28.3
Age 45 – 54 years (%) 19.0 52.8 28.2
Age 55 – 64 years (%) 14.7 54.8 30.6
Age 65 + (%) 19.6 58.3 22.2
Age in categories χ2 (10) = 19.871 < .05
Male (%) 19.9 53.6 26.6
Female (%) 17.2 57.0 25.7
Gender χ2 (2) = 3.084 > .05*
Basisonderwijs (%) 28.2 63.4 8.5
VMBO (%) 17.1 61.7 21.3
HAVO/VWO (%) 21.1 55.5 23.4
MBO (%) 15.5 56.3 28.2
HBO (%) 19.8 50.2 30.0
WO (%) 20.3 49.1 30.6
Education in CBS-categories χ2 (10) = 36.112 < .05
Net monthly income household (mean)** 7.87b 7.87b 7.96a F (2) = 5.969 < .05
Number of members household(mean) 2.39ab 2.39b 2.55a F (2) = 3.349 < .05
Note: the different superscripts (a-b) indicate significant statistical differences (p < .05) * p = .214 ** The natural logarithm of ‘net monthly income household’ was used in this analysis
Table 2 – Socio-demographic characteristics per consumer group
23
2.5 – Preliminary Conclusion
To conclude, it turned out that in the case of yoghurt, almost two-thirds of the participants engaged in
date labelling selection behaviour. Furthermore, it turned out that the ‘more flexibility’ motive was by
far the most important reason for participants to engage in date labelling selection behaviour. At the
same time, the ‘more flexibility’ argument proved to be a very persistent and hard to disempower
consumer motive underlying selection behaviour. Nonetheless, consumers turned out to be less likely
to engage in date labelling selection behaviour when they had the intention to consume the food on the
same day of purchase. To end, it appeared from the consumer profiling that consumers are relatively
less likely to engage in date labelling selection behaviour when they are higher educated, have a higher
net monthly household income, have a bigger household, and when they are aged fifteen to twenty-four
(as opposed to older).
24
25
Study 2
3.1 – Purpose
This study represents a first attempt at finding an effective solution to successfully intervene with
consumer’s date labelling selection behaviour in retail stores. Primarily, this small-scale follow-up
study aims to seek answer to the fourth research question: ‘What can the retailer do about date labelling
selection behaviour?’ In other words, it will be assessed what the retailer could do to improve consumer
acceptance of food items with the shortest remaining shelf life. In addition, various situations will be
examined in which consumers are suggested to be more/less inclined to engage in date labelling
selection behaviour, since this could potentially impact the effectiveness of retailer interventions.
3.2 – Theory and Predictions
According to the ‘ABC’ (Antecedents, Behaviour, and Consequences) model of behavioural change,
behavioural interventions can either be classified as antecedent-focused strategies or consequence-
focused strategies (Lehman, & Geller, 2004). In this second study it was decided to make use of
antecedent-focused strategies for behavioural change; it is attempted to intervene with the causes (i.e.
antecedents) of consumer behaviour. In other words, an attempt was made to influence consumer
behaviour by influencing the underlying consumer motives. To specify, one possible way to counteract
consumer’s date labelling selection behaviour is to take away the consumer motives underlying this
selection behaviour. It can be argued that in the battle against date labelling selection behaviour, the
retailer could either try to eliminate or at least disempower consumer motives in favour of this selection
behaviour or the retailer could try to reinforce or enhance consumer motives that work against this
selection behaviour. In other words, in order to increase acceptance of food items in front of the shelf
with the shortest remaining shelf life, one could try to take away or weaken the cons (costs) or
emphasize the pros (benefits) of those food items. Therefore, as became apparent from the first study,
there are essentially three possible ways to approach the problem of date labelling selection behaviour:
One could try to eliminate or disempower the ‘higher quality’ motive (I) or the ‘more flexibility’ motive
(II), or one could try to enhance the ‘less food waste’ motive (III).
Based on the results of the first study as well as the general consensus that consumers are insufficiently
aware of the food waste streams caused by date labelling selection behaviour (Aschemann-Witzel et al.,
2015), raising the concern about (avoiding) food waste could be a way to go for retailers in order to
intervene with consumer’s selection behaviour. In fact, in the current, general fight against food waste,
retail stores often employ strategies that emphasize waste issues in an attempt to create awareness and
to reinforce consumers’ collective interest. An example of such a strategy can be given with reference
26
to ‘Albert Heijn’ and its often-recurring message “It’s a shame to throw it away”6 addressed at (often
perishable) food items that are about to expire. However, in the current context of limited shelf life food
products, it may not always be logical to attach an “It’s a shame to throw it away” message to food
items closest to expire because it could perfectly be that those food items will not expire in one or two
weeks. In that case, such a message might be odd and incomprehensible for consumers: “Why would a
retailer throw away this food item, while it will not reach its date mark in (e.g.) two weeks?” Therefore,
in this context, much more would be needed than merely attaching this message to the food item in
front of the shelf. Besides, it is generally argued that self-interest is often the primary basis for economic
decisions (Miller, 2015). According to the literature, it is easier to motivate individuals to pursue their
self-benefits than collective benefits (Frantzich, 2008). In other words, it is often suggested in the
literature that individuals are more inclined to act out of self-interest than to act purely out of interest
of the collective good (Miller, 2015). Because of all this, as a retailer intervention it is decided to not
employ a food waste campaign that tries to enhance collective interest.
As a matter of fact, it became apparent from the first study that consumers engage in date labelling
selection behaviour primarily out of flexibility reasons and some out of higher quality reasons. The
perception of slightly lower quality and less flexibility (often) form the self-centered ‘costs’ for
consumers to select a food item in front of the shelf with the shortest remaining shelf life. The retailer
interventions in this second study will focus on these ‘higher quality’ and ‘more flexibility’ motives. In
order to disempower these consumer motives in favour of date labelling selection behaviour, and
therefore to increase acceptance of the food item in front of the shelf, it should in a way be made clear
by retailers to consumers that:
I. The difference in quality between a food item with the shortest shelf life and a
food item with a longer shelf life is actually negligible and almost non-existing
(Aschemann-Witzel et al., 2017)
II. There is still enough time to consume the food item with the shortest shelf life
before the food item reaches its date mark
If retailers could in a way explain this to consumers, consumers might start to re-think and reconsider
their behaviour in retail stores. Ideally, consumers will then realise that date labelling selection
behaviour is undesirable by linking this selection behaviour to food waste. Thus, weakening the
arguments that drive date labelling selection behaviour could indirectly raise consumers’ food waste
awareness and concern.
The question turns into how to turn this into retailer interventions. In this second study, as a first attempt
to intervene with date labelling selection behaviour, it was decided to work with persuasive messages
6 In Dutch: “Weggooien is zonde”
27
aimed at consumers at the moment-of-purchase in retail stores. It is often argued that interventions
applied at the moment-of-purchase are most effective in encouraging behavioural change (Young et al.,
2018). The content of the persuasive messages will be aimed at disempowering either the ‘higher
quality’ motive or the ‘more flexibility’ motive. It is predicted that persuasive messages that challenge
underlying motivations for date labelling selection behaviour will be effective in attenuating the extent
to which consumers engage in this behaviour. Based on this, the following hypothesis is formulated:
• H1: Consumers that are exposed to a message that challenges an underlying motivation for date
labelling selection behaviour will be less likely to engage in date labelling selection behaviour
than consumers that are not exposed to such a message.
Furthermore, it became apparent from the first study that the ‘more flexibility’ motive was by far the
most important consumer motive in favour of date labelling selection behaviour. Because of this, it is
predicted that the message aimed to disempower the ‘more flexibility’ motive will be most effective in
attenuating the extent to which consumers engage in date labelling selection behaviour. Based on this,
the following hypothesis is formulated:
• H2: Consumers that are exposed to the flexibility-related message will be less likely to engage
in date labelling selection behaviour than consumers that are exposed to the
quality-related message.
These hypotheses were tested in an experiment among Tilburg University students.
3.3 – Methodology
3.3.1 – Experimental Design and Participants
The study consists of two parts: The first part consists of a shopping task that takes place in a simulated
supermarket environment (i.e. the ‘actual’ experiment), followed by an online post-experiment
questionnaire. In the shopping task, participants have to make a choice which food item to buy. The
study employs a between-subjects experimental design with four conditions resulting from the
manipulation of the type of message (no message, neutral message, flexibility-related message, and
quality-related message). In the next section, the manipulation of the type of message will be explained
in more detail. The experiment examines the effectiveness of different messages in reducing consumer’s
date labelling selection behaviour. Participants are randomly assigned to one of the four message types.
28
For the data collection, primary data was collected in Tilburg University’s CentERlab7, where
participants were placed in individual cubicles. Dutch speaking Tilburg University students were
recruited and invited to the lab to participate in the study. In total, 96 Tilburg University students
participated (Mage = 21.03, SDage = 2.00; male = 41.7%, female = 58.3%). Each of the four ‘type of
message’ conditions contained 24 participants. In general, the participants indicated to be prone to date
labelling selection behaviour (“tendency to engage in date labelling selection behaviour”, M = 4.98 (SD
= 1.42) on a 7-point scale), and to have the intention to not waste food (“intention to avoid food waste”,
M = 5.23 (SD = 1.32) on a 7-point-scale). As will be explained in more detail in the next sections, it
was chosen to work with yoghurt drink products with respect to the shopping task. The majority of the
participants liked yoghurt drink (94.8 percent) and indicated to sometimes buy yoghurt drink (71.9
percent). The study was conducted between May the 29th and June the 5th. Participants received a
payment of 5 euro in return for their participation.
3.3.2 – Manipulation of the Type of Message
In this study, the type of message was manipulated. Three different messages were used: A neutral
message (‘Pick me’), a flexibility-related message (‘Pick me: you still have sufficient time to drink
me’), and a quality-related message (‘Pick me: I’m just as tasty as my friends behind me’). The neutral
message was added as a control condition in order to rule out the possibility of the effects being solely
attributable to a ‘label’ effect (i.e. any label would reduce selection behaviour). To continue, retailers
often make use of sticky labels attached to food items to communicate (for example) a discount to
consumers. In this study it is decided to use such labels as a way to communicate the aforementioned
three messages to consumers. Therefore, the message manipulation is reflected by different labels stuck
onto food items in front of a shelf with the shortest remaining shelf life (see Figure 6).
3.3.3 – Materials
Based on the three different messages, three brightly coloured labels were designed (see Figure 6).
Attention is often drawn to objects that strongly contrast with the background (Yarbus, 1967). Using
bright colours to contrast an object from its background increases one’s ability to detect the object
(De Vries, Hooge, Wertheim, & Verstraten, 2013). Because the study was conducted in Dutch, the
messages on the labels were also formulated in Dutch.
7 CentERlab is CentER’s facility for conducting economic and business experiments at Tilburg University. CentERlab consists of a computerized laboratory containing 10 individual cubicles.
29
In the shopping task, it was examined whether the different messages printed on labels were effective
in reducing the extent to which consumers engage in date labelling selection behaviour. In order to
examine whether the different messages have an effect on which food items participants pick, it is
necessary to simulate a supermarket environment including associated product shelves and prices.
Figure 7 shows the simulated supermarket yoghurt shelf. In order to increase external validity, it was
attempted to realistically mimic and reconstruct the yoghurt shelf as accurately as possible compared to
how it looks like in actual retail stores. Participants
were taken to a separate individual cubicle in the lab
where they encountered the simulated product shelf
with three different types of yoghurt drink (500 ml).
Each type of yoghurt drink had 6 items (i.e. packs)
placed on the shelf (18 packs in total), with the
associated price situated on top of the shelf. It was
chosen to work with yoghurt drinks because of the
assumption that yoghurt drinks are prone to date
labelling selection behaviour as being a chilled,
limited shelf life product. The three different variants
of yoghurt drink used, together with the date marks on
the six items, were:
- ‘Optimel Raspberry’ the first three on the shelf 05 - 06, the last three on the shelf 12 - 06
- ‘Optimel Lime’ the first three on the shelf 07 - 06, the last three on the shelf 12 - 06
- ‘Vifit Red Fruits’ the first three on the shelf 06 - 06, the last three on the shelf 13 - 06
The yoghurt drinks on the simulated product shelf were an exact copy of a yoghurt shelf in the
supermarket, as were the associated prices. To be able to record choices, each yoghurt drink (per type)
was assigned a small number from one to six written on the bottom of the item (1 = in front of the shelf,
Figure 7 – Yoghurt drink shelf used in experiment
Figure 6 – Labels used in experiment (type of message from left to right: neutral, flexibility-focused, quality-focused)
30
6 = on the back of the shelf). Next to the simulated shelf lay a piece of paper with the message ‘please
place the pack of yoghurt drink of your choice on this paper’.
3.3.4 – Procedure and Measures
Procedure
Tilburg University students were invited to the lab and placed in an individual cubicle. Participants
were instructed that the experiment was about how consumers make choices between products in the
supermarket. It was explained that they were soon taken, by the lab assistant, to another cubicle where
a refrigerated product shelf of a supermarket was simulated. The participants were instructed to imagine
that it was a real supermarket and that they had the intention to buy a yoghurt drink package.
Furthermore, it was instructed that they should select their preferred yoghurt drink (i.e. the yoghurt
drink that they would buy) out of the eighteen presented packs of yoghurt drink, exactly as they would
do in real-life. Finally, participants were informed that when they were done with the shopping task
they had to fill in a short post-experiment questionnaire online. The exact instruction given to
participants (in Dutch) can be found in Appendix 3. After participants received the instruction, they
were taken by the lab assistant to another cubicle for the shopping task. After participants placed the
yoghurt drink of their choice on the designated spot they were brought back, by the lab assistant, to
their initial cubicle to fill in the post-experiment questionnaire. In the meantime, the lab assistant
recorded (after every participant) the corresponding number of the selected yoghurt drink pack.
Before the actual experiment took place, a small pre-test was conducted on the 28th of May to ascertain
that the online instruction given prior to the shopping task was clear (n = 7; Mage = 23.29, SDage = 2.29;
male = 71.4%, female = 28.6%). After participants in the pre-test completed the shopping task they
were asked whether everything was clear with respect to what was expected from them. It appeared it
was not clear they could choose from all eighteen yoghurt drinks, so this was added for clarification in
the instruction. Additionally, the statement ‘the lab assistant will put everything back in place after you
have finished the shopping task’ was added to the instruction to clarify that it was okay to rumble around
on the shelf.
Measures
In the shopping task, the extent to which consumers engage in date labelling selection behaviour was
measured on the basis of participants’ product choice. If participants chose a yoghurt drink pack with
assigned number 1, 2, or 3 they did not engage in date labelling selection behaviour. On the contrary, if
participants chose a yoghurt drink pack with assigned number 4, 5 or 6 they did engage in date labelling
selection behaviour because these yoghurt drink packs had a longer remaining shelf life.
31
In the post-experiment questionnaire, supplementary background information of participants was
obtained that could be linked to participants’ choice behaviour in the shopping task. The gathered
background information relates to the extent participants define themselves as someone engaging in
date labelling selection behaviour with respect to limited shelf life food products (such as yoghurt
drink), and to the extent participants have the intention not to waste food.
The tendency to engage in date labelling selection behaviour was measured via three items on a 7-
point scale (1 = certainly not, 7 = most certainly): ‘I usually pay close attention to date marks’, ‘I usually
consciously take a product with a long shelf life’, and ‘I usually take the product in front of the shelf or
on top of a pile’8.
Intention to avoid food waste (Visschers, Wickli, & Siegrist, 2016; Stancu, Haugaard, & Lahteenmaki,
2016) was measured via three items on a 7-point scale (1 = certainly not, 7 = most certainly): ‘I think it
is a shame to throw away food’, ‘I am consciously trying to waste as little food as possible’, and ‘I think
food waste is a big problem’.
In addition, the post-experiment questionnaire also contained some control questions to assess whether
participants had noticed the message labels in the first place. Participants were asked whether or not
they had seen a label on some of the packs of yoghurt (‘You have just seen the shelf with packs of
yoghurt drink, have you seen a label on (some of) the packs of yoghurt drink?’). In the case participants
answered that they had seen a label, it was asked what they believed was written on the label in the
form of an open-ended question (‘If so, what do you believe was written on the label?’). Furthermore,
participants were asked whether or not they liked yoghurt drink and whether or not they ever buy
yoghurt drink. The post-experiment questionnaire (in Dutch) can be found in Appendix 4.
3.4 – Results
This results section is subdivided into two separate paragraphs. The first paragraph shows the results of
the main analysis that tests whether the messages were effective in reducing date labelling selection
behaviour, whereas the second paragraph shows the results of several follow-up analyses.
3.4.1 – Main Analysis
In the shopping task, participants selected their preferred yoghurt drink package from a small
assortment. Based on participants’ product choice, it is derived whether or not participants engage in
date labelling selection behaviour. Table 3 provides the descriptive results.
8 Within the overall construct the third item ‘I usually take the product in front of the shelf or on top of a pile’ was reverse-coded.
32
Type of Message (n = 96) Not engaging in date labelling selection behaviour (yoghurt # 1, 2, 3)
Engaging in date labelling selection behaviour (yoghurt # 4, 5, 6)
1. No message 11 (45.9 %) 13 (54.1 %) 2. Neutral 11 (45.9 %) 13 (54.1 %) 3. Flexibility-related 11 (45.9 %) 13 (54.1 %) 4. Quality-related 13 (54.1 %) 11 (45.9 %)
Total: 46 (47.9 %) Total: 50 (52.2 %) Table 3 – Descriptive results Study 2
Overall, slightly more than half of the participants (52.2 percent) engaged in date labelling selection
behaviour. In the intervention-free ‘no message’ condition, 54.1 percent of the participants (in contrast
to 64.3 percent of the participants in Study 1) engaged in date labelling selection behaviour. This finding
is consistent with the first study, where it was detected that higher educated and younger consumers
were less likely to engage in date labelling selection behaviour.
The primary question is whether the extent to which participants engage in date labelling selection
behaviour differs significantly between the type of message conditions. Hypothesis 1 predicted that
participants that are exposed to a message that challenges an underlying motivation for selection
behaviour, are less likely to engage in date labelling selection behaviour than participants that are not
exposed to such a message. Furthermore, hypothesis 2 predicted that participants that are exposed to
the flexibility-related message would be less likely to engage in date labelling selection behaviour than
participants that are exposed to the quality-related message. In order to test these two hypotheses, a
logistic regression was conducted with type of message as the independent variable and whether or not
participants engage in date labelling selection behaviour as binary dependent variable. The logistic
regression shows that the overall effect of the type of message on date labelling selection behaviour is
not significant (Wald χ² (3) = .499, p = .919). In other words, date labelling selection behaviour did not
significantly differ between the message types, rejecting hypothesis 1. Despite the nonsignificant
overall effect of message type, pairwise differences in date labelling selection behaviour between the
message types were examined as follow-up in the next section. In contrast to hypothesis 2, the logistic
regression did not reveal a difference in date labelling selection behaviour between the flexibility-
related message and the quality-related message (Wald χ² (1) = .333, p = .564). Therefore, the second
hypothesis is also rejected with respect to this study. The output of the logistic regression can be found
in Appendix 5.
3.4.2 – Follow-up Analysis: Moderators of the type of message effect
Unfortunately, no significant message effect was found; the message interventions turned out not to
have the derised negative effect on date labelling selection behaviour. In order to try to understand why
the messages were ineffective, several follow-up analyses were performed. Note, however, that sample
33
size is relatively small and hence the follow-up analyses are in fact underpowered. Nonetheless, we
believe it is valuable to look for patterns in the data as these insights could serve as potential input for
further research. We identified three potential moderators of the effect of the messages:
1. The time until expiry
2. Participants’ a priori intention to avoid food waste
3. The extent to which participants noticed the message labels
Time until expiry
To start, it could be the case that the messages had an insignificant overall effect on whether or not
participants engaged in date labelling selection behaviour because while each day passed by during the
experiment, the date marks of the yoghurt drinks came closer and closer. It could be that in the beginning
of the experiment the messages were effective, but later on in the experiment they turned ineffective (or
even backfired). It is predicted that over time more participants engaged in date labelling selection
behaviour in the experiment, because the expiry date mark of the yoghurt drinks came closer and closer
over time allowing for less flexibility with respect to when to consume the yoghurt. As such, especially
the flexibility-related message may backfire when flexibility is actually low. Therefore, it is examined
whether ‘time’ moderates the effect the message interventions have on date labelling selection
behaviour of consumers. Time was entered in the logistic regression model as a linear term ranging
from -2.5 (day one) to 2.5 (day six). Potential non-linear effects of time were captured by a quadratic
term. Finally, interactions between the type of message and time (linear and quadratic) were included
in the model, to examine the effectiveness of different types of messages over time. It was chosen to
use ‘time in days’ because of the assumption that the tendency to engage in date labelling selection
behaviour increases by the day. The model results reveal that none of the effects are significant (all ps
> .05), which may be due to power problems. Nonetheless, it might still be valuable to look for patterns.
Figure 8 (on the next page) shows the predicted probability that a yoghurt drink with the shortest
remaining shelf life (i.e. in front of the shelf) is selected for the different message types over time.
Although the effect of ‘time’ is insignificant, the negative time trend is in the expected direction (b =
-.239). Looking at the simple slopes, we find negative (but insignificant) time trends for all types of
messages (no message: b = -.239, flexibility-related message: b = -.529, quality-related message: b =
-.191), except – somewhat surprisingly – for the neutral message (b = +.110). Interestingly, while the
quality-related message seems to mitigate date labelling selection behaviour over the course of time
(compared to no message at all), the flexibility-related message seems to accelerate date labelling
selection behaviour when moving towards the expiration date. This is consistent with the idea that the
flexibility-related message backfires when flexibility is actually low (essentially turning the message
into a false claim). However, note that none of these effects reaches significance. The model results can
be found in Appendix 6A.
34
Intention to avoid food waste
It could also be that the messages did not significantly affect date labelling selection behaviour because
the message interventions were only effective for participants that are concerned about food waste and
have the intention to avoid food waste in the first place. In other words, it could be that the effect of the
messages are more profound for participants who have a stronger (versus weaker) intention to avoid
food waste. Therefore, it is examined whether ‘intention to avoid food waste’ moderates the effect the
message interventions have on the extent to which participants engage in date labelling selection
behaviour. In the post-experiment questionnaire, the self-reported intention of participants to avoid food
waste was measured via three items each on a 7-point scale (Cronbach’s Alpha = .854)9 and this
‘intention to avoid food waste’ was added to the logistic regression model. The model results reveal
that the interaction between ‘type of message’ and ‘intention to avoid food waste’ is insignificant (Wald
χ² (3) = 4.404, p = .221), which may be due to power problems. Nonetheless, it might still be valuable
9 Overall construct was created under the assumption that each item is equally important (i.e. same unit weights)
Figure 8 – Graphical visualization of the effect of type of message over time, on the probability of selecting a yoghurt drink with the shortest shelf life.
35
to look for other patterns and trends in the data. Therefore, a follow-up analysis was conducted to
estimate the simple effects of the type of messages separately for participants with low intention to
avoid food waste and for participants with high intention to avoid food waste (based on a median split).
Results are displayed in Figure 9. However, the model results reveal that the simple effects of the
messages for participants with low intention to avoid food waste and for participants with high intention
to avoid food waste separately, are nonsignificant (all ps > .05). Despite the insignificant effects, the
model results do reveal some interesting patterns. Participants with high intention to avoid food waste
have the tendency to engage less in date labelling selection behaviour in the presence of a message label
(compared to no message at all), particularly with respect to the flexibility-related message (b = 1.099)
and the quality-related message (b = .0693). On the contrary, participants with low intention to avoid
food waste have the tendency to engage more in date labelling selection behaviour in the presence of a
message label (compared to no message at all), especially with respect to the flexibility-related message
(b = -1.281). However, recall that none of these effects reaches significance. The model results can be
found in Appendix 6B.
Figure 9 – Graphical visualization of the effect of type of message, per low or high intention to avoid food waste, on the probability of selecting a yoghurt drink with the shortest shelf life
36
Noticing the message labels
Finally, it could be the case that the message interventions were ineffective in reducing the extent to
which participants engage in date labelling selection behavior, because the messages were only effective
for participants that actually noticed the message labels stuck onto the yoghurt drink packs in front of
the shelf. In the post-experiment questionnaire, participants were asked whether or not they had seen a
label on some of the packs of yoghurt drink. In total, 58 out of the 72 participants who were in fact
exposed to yoghurt drink packages with a message label recalled seeing a message label (72.2 percent).
It is predicted that the message interventions were more effective for participants who actually noticed
the message labels (versus all participants) when there were message labels to be noticed. Is the effect
of the messages stronger if we filter out the participants who should have seen message labels, but did
not see them? In order to test this, we repeated the initial, main analysis solely for participants who
indicated to have noticed the message labels, and tested the effect of the message conditions against the
‘no message’ control condition which still included all participants. The analysis revealed that the
effects of the messages were slightly stronger (compared to the initial main analysis) but still did not
reach significance (Wald χ² (3) = .366, p = .947). The model results can be found in Appendix 6C and
are also graphically visualized in Figure 10 below.
Figure 10 – Graphical visualization of the effect of type of message, for all participants and for solely the participants who noticed the message labels, on the probability of selecting a yoghurt drink with the shortest shelf life.
37
3.5 – Preliminary Conclusion
The study showed that ‘type of message’ had an insignificant effect on whether or not participants
engaged in date labelling selection behaviour. In order to better understand why the message
interventions were ineffective, several follow-up analyses were performed with 1) the time until expiry,
2) participants’ a priori intention to avoid food waste, and 3) the extent to which participants noticed
the message labels. However, none of these follow-up analyses showed significant effects which is most
likely due to the relatively low sample size of this second study and subsequent power problems.
38
39
Conclusions and Recommendations
4.1 – General conclusion and discussion
A substantial amount of food produced for human consumption is wasted nowadays, which entails
negative economic, environmental and ethical consequences. This research revolved around the
supposition that, with respect to limited shelf life food products, consumers engage in date labelling
selection behaviour in retail stores which in turn contributes to food waste. However, up to this point,
it was not empirically studied how big the proportion of consumers exactly is that have the tendency to
select food items with a longer or the longest shelf life, why consumers have the tendency to do so,
when consumers are more/less inclined to do so, what retailers could subsequently do about it etcetera.
In order to provide answers to these questions, two experimental studies were conducted.
With respect to the extent to which consumers engage in selection behaviour (the first research
question), it can be concluded from Study 1 that almost two-thirds (64.3 percent) engaged in date
labelling selection behaviour when selecting a limited shelf life yoghurt product from the shelf. This
finding is in line with existing literature suggesting that consumer’s date labelling selection behaviour
is omnipresent and therefore problematic. In Study 2, in which Tilburg University students participated,
54.1 percent engaged in date labelling selection behaviour. A possible explanation for this lower
percentage could be, as also became clear from consumer profiling in Study 1, that higher educated and
younger consumers are less likely to engage in date labelling selection behaviour.
As expected, almost all consumers that engaged in date labelling selection behaviour deliberately and
consciously did so. Interestingly, it can be concluded that the most prominent reason for consumers as
why to (deliberately) engage in date labelling selection behaviour (the second research question) is
because a yoghurt pack with a longer remaining shelf life would provide more time for consumption
and therefore more flexibility. In other words, it became clear that the ‘more flexibility’ motive
underlying date labelling selection behaviour is by far the most important motive for consumers to
engage in such selection behaviour. Somewhat surprisingly, the ‘higher quality’ motive turned out to
be most important for only 10 percent of the consumers that engaged in selection behaviour.
Furthermore, the majority of consumers that did not engage in date labelling selection behaviour also
deliberately and consciously did so, while it was predicted that this behavioural act would mainly be
automatic and unconscious since this is the norm-consistent way. It could be, however, that we have
turned this behavioural act in more deliberate and conscious behaviour than it in reality is. By asking
consumers whether their choice was deliberate or not, these consumers might started to think about
their own behaviour more than they would normally do. Yet, only very few consumers reported to
deliberately select the first food item in order to avoid food waste. It could be that consumers are
40
unaware of the fact that date labelling selection behaviour contributes to food waste, or that consumers
simply do not care enough about food waste. A more important motive for consumers not to engage in
date labelling selection behaviour is related to the argument that they still have sufficient time for
consumption. Only a few participants reasoned that the food item in front of the shelf will taste as good
as food items further on the back of the shelf.
Various situations and circumstances were assessed in this research in order to identify when consumers
are more/less likely to engage in date labelling selection behaviour (the third research question).
Noteworthy, only a few situations and circumstances were examined that could potentially impact the
effectiveness of retailer interventions, but there is much more to discover. To start, it can be concluded
that consumers are less likely to engage in date labelling selection behaviour when they have the
intention to consume the food on the same day as the day on which the food item is purchased. In that
situation, the need for flexibility is in a sense eliminated thereby lowering the tendency to select food
items with a longer expiration date. Furthermore, it appeared that consumers are also less likely to
engage in date labelling selection behaviour when they are higher educated, have a higher net monthly
household income, when they have a bigger household, and when they are aged 15 to 24. In addition,
it was examined whether consumers are more likely to engage in date labelling selection behaviour over
time when food items start to approach their date mark, whether consumers with high intention to avoid
food waste are less likely to engage in date labelling selection behaviour, and whether consumers that
are actually aware of the retailer interventions are less likely to engage in date labelling selection
behaviour. However, as will be explained in more detail below, none of these effects reached
significance.
With respect to what the retailer could do to successfully intervene with date labelling selection
behaviour (the fourth research question), Study 2 examined whether message interventions that
challenge the underlying motives for selection behaviour (by means of labels attached to food items in
front of the shelf with the shortest shelf life) were effective in attenuating the extent to which consumers
engage in this behaviour. Unfortunately, this first attempt at finding a solution turned out to be
ineffective. Both hypotheses with respect to the message interventions were therefore rejected; it could
not be concluded that (1) consumers that are exposed to a message that challenges an underlying
motivation for date labelling selection behaviour are less likely to engage in such selection behaviour
than consumers that are not exposed to such a message, and that (2) consumers that are exposed to the
flexibility-related message are less likely to engage in date labelling selection behaviour than consumers
that are exposed to the quality-related message. In an attempt to understand why the message
interventions were ineffective (i.e. did not have a significant effect on the extent to which consumers
engage in selection behaviour), several follow-up analyses were conducted. Potential situations were
assessed in which consumers could in general be more/less inclined to engage in date labelling selection
behaviour (therefore relating to the third research question). Unfortunately, none of the identified
41
potential moderators (time until expiry, participants’ a priori intention to avoid food waste, and the
extent to which participants noticed the message labels) turned out to have a significant effect on the
relationship between the message interventions and selection behaviour, which may be due to power
problems. Although none of the effects reached significance, some interesting patterns were detected
in line with what was expected; a negative time trend was observed, consumers that have high intention
to avoid food waste are slightly more inclined to engage less in date labelling selection behaviour in the
presence of the flexibility-related or the quality-related message, and the messages seem to have a
slightly stronger effect on the consumers who actually noticed the message labels when message labels
were present. To end, a possible reason of why the flexibility-related message turned out to be
ineffective could be related to the persistence of the ‘more flexibility’ consumer motive underlying date
labelling selection behaviour. It became apparent from Study 1 that the ‘more flexibility’ motive is a
hard to disempower motive. So while the ‘more flexibility’ motive is valuable to focus upon with respect
to retailer interventions since this motive is for many consumers the most prominent reason as why to
engage in date labelling selection behaviour, it is at the same time a difficult to disempower motive.
4.2 – Recommendations
What do the findings mentioned above imply, and which recommendations can be derived from this?
First of all, it is recommended to take consumer’s date labelling selection behaviour in retail stores and
its contribution to food waste seriously. This research showed that many consumers opt for and
eventually buy food items with a longer/ the longest remaining shelf life with respect to limited shelf
life food products. Therefore, date labelling selection behaviour is something to consider in the battle
against food waste. By reducing the number of consumers that engage in date labelling selection
behaviour, less food items will turn into food waste in retail stores. Unfortunately, the first attempt in
Study 2 to successfully intervene with consumer’s date labelling selection behaviour failed; the labels
with various messages aimed to weaken consumer motives underlying date labelling selection
behaviour did not have the intended negative effect. All in all, it is recommended to first find an effective
retailer intervention before implementing anything, so further research is necessary.
4.3 – Limitations and suggestions for further research
First of all, a limitation of the follow-up analyses in Study 2 had to do with sample size. Although Study
2 only represented a first attempt at finding an effective solution against date labelling selection
behaviour, the follow-up analyses were underpowered because of this relatively small sample size. It is
plausible that the three identified moderators did not have a significant effect on the relationship
42
between the messages and the extent to which consumers engaged in date labelling selection behaviour
because the analyses were underpowered. Therefore, as a suggestion for further research, one could re-
investigate the observed trends of (1) time until expiry, (2) participants’ a priori intention to avoid food
waste, and (3) the extent to which participants noticed the message labels. By doing so, we might be
better able to understand why the message interventions were ineffective in reducing date labelling
selection behaviour. Besides, the better we understand in which situations or circumstances consumers
are more/less inclined to engage in date labelling selection behaviour, the better we can judge whether
other potential retailer interventions will be effective in attenuating the extent to which consumers
engage in selection behaviour. Therefore, it is also suggested to identify and examine more situations
and/or circumstances in which consumers are more/less likely to engage in selection behaviour.
Furthermore, since Study 2 was only a first attempt at finding a solution against consumer’s date
labelling selection behaviour, a suggestion for further research would be to investigate other types of
(message) interventions. The solution to date labelling selection behaviour could also (partially) lie in
creating more awareness about the effect date labelling selection behaviour has on food waste in retail
stores, while at the same time intervening with message labels. In addition, it is often argued that
consumers do consider suboptimal food (i.e. food with the shortest shelf life) when offered at a discount
(Aschemann-Witzel et al., 2015). Therefore, it could be further researched whether this would
successfully reduce date labelling selection behaviour while at the same time being feasible in terms of
money for retailers. As a final remark, another suggestion for future research could be to re-examine
the message label interventions in a study that does not solely contain students as participants.
43
References Ariely, D., & Shampan'er, K. (2004). Tradeoffs between costs and benefits: Lessons from 'the price of
0'. CeB Research Brief, 1-5.
Arrow, K. (1986). Rationality of self and others in an economic system. Journal of Business, 385-399.
Aschemann-Witzel, J., de Hooge, I. d., Amani, P., Bech-Larsen, T., & Oostindjer, M. (2015). Consumer-related food waste: causes and potential for action. Sustainability, 6457-6477.
Aschemann-Witzel, J., de Hooge, I., & Normann, A. (2016). Consumer-Related Food Waste: Role of Food Marketing and Retailers and Potential for Action. Journal of International Food & Agribusiness Marketing , 271-285.g
Aschemann-Witzel, J., Jensen, M., Jensen, M., & Kulikovskaja, V. (2017). Consumer behaviour towards price-reduced suboptimal foods in the supermarket and the relation to food waste in households. Appetite, 246-258.
Centraal Bureau Levensmiddelenhandel. (2018). De Supermarktbranche. Retrieved from http://www.cbl.nl/de-supermarktbranche/feiten-en-cijfers/
Cialdini, R., Kallgren, C., & Reno, R. (1991). A focus theory of normative conduct: A theoretical refinement and reevaluation of the role of norms in human behavior. Advances of Experimental Social Psychology, 201-234.
Cicatiello, C., Franco, S., Pancino, B., & Blasi, E. (2016). The value of food waste: An exploratory study on retailing. Journal of Retailing and Consumer Services, 96-104.
Cicatiello, C., Franco, S., Pancino, B., Blasi, E., & Falasconi, L. (2017). The dark side of retail food waste: Evidences from in-store data . Resources, Conservation & Recycling, 273-281 .
De Vocht, A. (2018). Minder verspillen, meer besparen! NoFoodWasted. Retrieved from https://www.voordewereldvanmorgen.nl/duurzame-projecten/nofoodwasted
De Vries, J., Hooge, I., Wertheim, A., & Verstraten, F. (2013). Background, an important factor in visual search. Vision Research, 128-138.
Dijksterhuis, A., Smith, P., van Baaren, R., & Wigboldus, D. (2005). The Unconscious Consumer: Effects of Environment on Consumer Behaviour. Journal of Consumer Psychology, 193-202.
Eriksson, M. (2015). Supermarket Food Waste. Uppsala: Swedish University of Agricultural Sciences.
European Commission. (2015). Flash Eurobarometer 425: Food waste and date marking. Brussels: European Union.
European Commission. (2018). Market study on date marking and other information provided on food labels and food waste prevention. Brussels: WRAP.
Ferguson, M., & Ketzenberg, M. (2006). Information Sharing to Improve Retail Product Freshness of Perishables . Production and Operations Management, 57-73.
Food and Agriculture Organization. (2013). Food wastage footprint: Impacts on natural resources - Summary Report. FAO.
Frantzich, S. (2008). Citizen Democracy: Political Activism in a Cynical Age. Plymouth: Rowan & Littlefield Publishers, INC.
44
Garrone, P., Melacini, M., & Perego, A. (2014). Opening the black box of food waste reduction. Food Policy, 129-139.
Graham-Rowe, E., Jessop, D., & Sparks, P. (2014). Identifying motivations and barriers to minimising household food waste. Resources, Conservation and Recycling, 15-23.
Gustavsson, J., Cederberg, C., & Sonesson, U. (2011). Global Food Losses and Food Waste. Gothenburg: Swedish Institute for Food and Biotechnology (SIK).
Albert Heijn. (2018). Voedselverspilling. Retrieved from https://www.ah.nl/over-ah/meer-doen/voedselverspilling
Hug, F., Asnani, S., Jones, V., & Cutright, K. (2005). Modeling the influence of multiple expiration dates on revenue generation in the supply chain . International Journal of Physical Distribution & Logistics Management , 152-160.
Lebersorger, S., & Schneider, F. (2015). Food loss rates at the food retail, influencing factors and reasons as a basis for food waste prevention measures. Waste Management, 1911-1919.
Lehman, P., & Geller, E. (2004). Behavior Analysis and Environmental Protection: Accomplishments and Potential for More. Behavior and Social Issues, 13-32.
Lyndhurst, B. (2011). Consumer Insight: Date Labels and Storage Guidance . London: WRAP.
Miller, D. (2015). The norm of self-interest. The next phase of business ethics: integrating psychology and ethics, 193-210.
Milne, R. (2012). Arbiters of waste: date labels, the consumer and knowing good, safe food. Sociological Review, 84-101.
United Nations. (2017). Progress towards the Sustainable Development Goals. Brussels: Department of Economic and Social Affairs, United Nations.
Newsome, R., Balestrini, C., Baum, M., Corby, J., Fisher, W., Goodburn, K., . . . Yiannas, F. (2014). Applications and Perceptions of Date Labeling of Food. Comprehensive Reviews in Food Science and Food Safety, 745-769.
Nielsen. (2017). Nielsen vrijgave marktaandelen, 2016 t.o.v. 2015. Retrieved from http://www.nielsen.com/nl/nl/insights/news/2017/nielsen-release-market-shares.html
Parfitt, J., Barthel, M., & Macnaughton, S. (2010). Food waste within food supply chains: quantification and potential for change to 2050. Philosophical Transactions of The Royal Society , 3065-3081.
Payne, J., Bettman, J., & Johnson, E. (1991). Consumer decision making . Handbook of Consumer Behaviour, 50-84.
Ratchford, B. (1982). Cost-benefit models for explaining consumer choice and information seeking behaviour. Management Science, 197-212.
Schripsema, A., van der Burgh, M., van der Sluis, A., & Bos-Brouwers, H. (2015). Verwaarding van voedselreststromen uit supermarkten. Wageningen : Wageningen University & Research.
Soethoudt, H., Vollebregt, M., & van der Burgh, M. (2016). Monitor Voedselverspilling (Update 2009-2014). Wageningen: Wageningen UR Food & Biobased Research.
Soethoudt, J., Sluis, A. v., Waarts, Y., & Tromp, S. (2013). Expiry dates: a waste of time? Wageningen: Wageningen UR Food & Biobased Research.
45
Stancu, V., Haugaard, P., & Lahteenmaki, L. (2016). Determinants of consumer food waste behaviour: Two routes to food waste. Appetite, 7-17.
Steg, L., & Vlek, C. (2009). Encouraging pro-environmental behaviour: An integrativ review and research agenda. Journal of Environmental Psychology , 309-317.
Stenmarck, A., Hanssen, O. J., Silvennoinen, K., Katajajuuri, J., & Werge, M. (2011). Initiatives on prevention of food waste in the retail and wholesale trades. Stockholm : Swedish Environmental Research Institute.
Stenmarck, A., Jensen, C., Quested, T., & Moates, G. (2016). Estimates of European food waste levels. Stockholm: FUSIONS.
Storup, K., Mattfolk, K., Voinea, D., Jakobsen, B., Bain, M., Reverte, & Oliveira, P. (2016). Combating Food Waste: An opportunity for the EU to improve the resource-efficiency of the food supply chain . Luxembourg: European Union - European Court of Auditors.
Theotokis, A., Pramatari, K., & Tsiros, M. (2011). Effects of Expiration Date-Based Pricing on Brand Image Perceptions. Journal of Retailing, 16.
Van der Vossen, W., & van Dooren, C. (2016). Bewaren van Voedsel Factsheet. Voedingscentrum.
Visschers, V., Wickli, N., & Siegrist, M. (2016). Sorting out food waste behaviour: A survey on the motivators and barriers of self-reported amounts of food waste in households. Journal of Environmental Psychology, 66-78.
Wedel, M., & Kamakura, W. (2012). Market Segmentation: Conceptual and Methodological Foundations. Springer Science & Business Media.
Williams, H., Wikstrom, F., Otterbring, T., Lofgren, M., & Gustafsson, A. (2012). Reasons for household food waste with special attention to packaging. Cleaner Production, 141-148.
WRAP. (2010). Preparatory study on food waste across EU 27: Technical Report 2010. Brussels.
WRAP. (2018). Why take action: The commercial case . Retrieved from http://www.wrap.org.uk/content/why-take-action-commercial-case
Yarbus, A. (1967). Eye movements during perception of complex objects. . Eye movements and vision, 171-211.
Young, C., Russel, S., Robinson, C., & Chintakayala, P. (2018). Sustainable Retailing - influencing consumer behaviour on food waste . Business Strategy and the Environment, 1-15.
46
47
Appendix 1 – Questionnaire Study 1 Vragenlijst “Boodschappen doen in de supermarkt”
Respondenten willekeurig toewijzen aan één van twee condities:
Instructie = 1 (meteen opeten); (condition 2)
Instructie = 2 (onzekerheid) (condition 1)
Scherm 1:
Deze vragenlijst gaat over boodschappen doen. We zijn benieuwd naar hoe u bepaalde keuzes maakt in de supermarkt. Er zijn geen goede of foute antwoorden.
Scherm 2:
Zo meteen krijgt u een foto te zien van een supermarktschap met pakken yoghurt.
Als Instructie = 1: Stelt u zich voor dat u een toetje gaat maken voor vanavond waarvoor u 500 ml yoghurt nodig hebt. U bent in de supermarkt om onder andere een pak yoghurt te kopen.
Als Instructie = 2: Stelt u zich voor dat u boodschappen aan het doen bent in de supermarkt. U hebt een boodschappenlijstje, maar kijkt daarnaast ook gewoon een beetje rond voor andere dingen. Omdat u ’s ochtends wel eens trek hebt in yoghurt, besluit u een pak yoghurt in huis te halen.
Voor het slagen van dit onderzoek is het van belang dat u zich voorstelt dat het vandaag 3 maart is.
Scherm 3:
Q1. Als u van plan zou zijn een pak yoghurt te kopen [Indien Instructie = 1: voor het toetje voor vanavond], welke van deze pakken zou u dan nemen? De pakken zijn genummerd van 1 tot en met 7. Selecteer hieronder het nummer van het pak dat u zou nemen.
1 2 3 4 5 6 7
48
Scherm 4:
Indien Q1 = 1 OF 2:
Q2. Er stonden pakken yoghurt met een houdbaarheidsdatum van 17 maart in het schap. Hebt u expres gekozen voor een pak met een houdbaarheidsdatum van 15 maart?
1 Nee, ik heb gewoon een pak dat vooraan stond gekozen
2 Ja, ik heb expres voor een pak yoghurt met een kortere houdbaarheid gekozen
Scherm 5:
Indien Q1 = 1 OF 2 en Q2 = 2:
Q3. Waarom koos u expres voor een pak met een kortere houdbaarheid?
Kies het antwoord dat het meest voor u geldt.
1 De yoghurt is nog zeker 12 dagen houdbaar, wat me genoeg tijd zou geven om de yoghurt op te eten
2 Ik verwacht dat de yoghurt even goed smaakt als de yoghurt met een langere houdbaarheid
3 Ik zou het zonde vinden als pakken met een kortere houdbaarheidsdatum uiteindelijk weggegooid zouden moeten worden
4 Anders, namelijk …
Scherm 4:
Indien Q1 = 3, 4, 5, 6 OF 7:
Q2. Er stonden pakken yoghurt met een houdbaarheidsdatum van 15 maart in het schap. Hebt u expres gekozen voor een pak met een houdbaarheidsdatum van 17 maart?
1 Nee, ik heb zo maar een pak gekozen
2 Ja, ik heb expres voor een pak yoghurt met een langere houdbaarheid gekozen
Scherm 5:
Indien Q1 = 3, 4, 5, 6 OF 7 en Q2 = 2:
Q3. Waarom koos u expres voor een pak met een langere houdbaarheid?
Kies het antwoord dat het meest voor u geldt.
1 Het geeft me meer tijd om de yoghurt op te eten voor het over de datum gaat
2 Ik verwacht dat de yoghurt beter smaakt dan de yoghurt met een kortere houdbaarheid omdat het iets verser is
3 Anders, namelijk …
49
Appendix 2 – Results Study 1
• With Y = ‘engaging in date labelling selection behaviour’
Block 1: Method = Enter
Omnibus Tests of Model Coefficients
Chi-square df Sig.
Step 1 Step 71,809 1 ,000
Block 71,809 1 ,000
Model 71,809 1 ,000
Model Summary
Step -2 Log likelihood
Cox & Snell R
Square
Nagelkerke R
Square
1 2711,033a ,035 ,047
a. Estimation terminated at iteration number 3 because parameter
estimates changed by less than ,001.
Classification Tablea
Observed
Predicted Datum_Duiken Percentage
Correct 0 1
Step 1 Datum_Duiken 0 537 370 59,2
1 450 666 59,7
Overall Percentage 59,5
a. The cut value is ,500
Variables in the Equation
B S.E. Wald df Sig. Exp(B)
Step 1a conditie ,765 ,091 70,520 1 ,000 2,148
Constant -,941 ,143 43,132 1 ,000 ,390
a. Variable(s) entered on step 1: conditie.
50
• With Y = ‘high flexibility motive’
Block 1: Method = Enter
Omnibus Tests of Model Coefficients
Chi-square df Sig.
Step 1 Step 18,219 1 ,000
Block 18,219 1 ,000
Model 18,219 1 ,000
Model Summary
Step -2 Log likelihood
Cox & Snell R
Square
Nagelkerke R
Square
1 744,973a ,017 ,033
a. Estimation terminated at iteration number 5 because parameter
estimates changed by less than ,001.
Classification Tablea
Observed
Predicted FLEX Percentage
Correct 0 1
Step 1 FLEX 0 0 122 ,0
1 0 964 100,0
Overall Percentage 88,8
a. The cut value is ,500
Variables in the Equation
B S.E. Wald df Sig. Exp(B)
Step 1a conditie ,826 ,195 17,916 1 ,000 2,285
Constant ,811 ,298 7,426 1 ,006 2,250
a. Variable(s) entered on step 1: conditie.
51
Appendix 3 – Instruction Study 2
52
53
Appendix 4 – Post-Experiment Questionnaire Study 2
54
55
Appendix 5 – Results Study 2
• Reference category: ‘no message’
Categorical Variables Codings
Frequency
Parameter coding
(1) (2) (3)
Conditie 0 (geen sticker 24 ,000 ,000 ,000
1 (pak mij) 24 1,000 ,000 ,000
2 (pak mij, flexibiliteit) 24 ,000 1,000 ,000
3 (pak mij, kwaliteit) 24 ,000 ,000 1,000
Block 1: Method = Enter
Omnibus Tests of Model Coefficients
Chi-square df Sig.
Step 1 Step ,501 3 ,919
Block ,501 3 ,919
Model ,501 3 ,919
Model Summary
Step -2 Log likelihood
Cox & Snell R
Square
Nagelkerke R
Square
1 132,417a ,005 ,007
a. Estimation terminated at iteration number 2 because parameter
estimates changed by less than ,001.
Classification Tablea
Observed
Predicted Oudste_datum_pak_Y
Percentage Correct 0 1
Step 1 Oudste_datum_pak_Y 0 39 11 78,0
1 33 13 28,3
Overall Percentage 54,2
a. The cut value is ,500
56
Variables in the Equation
B S.E. Wald df Sig. Exp(B)
Step 1a Conditie ,499 3 ,919
Conditie(1) ,000 ,579 ,000 1 1,000 1,000
Conditie(2) ,000 ,579 ,000 1 1,000 1,000
Conditie(3) ,334 ,579 ,333 1 ,564 1,397
Constant -,167 ,410 ,166 1 ,683 ,846
a. Variable(s) entered on step 1: Conditie.
• Reference category: ‘quality-related message’
Categorical Variables Codings
Frequency
Parameter coding
(1) (2) (3)
Conditie 0 (geen sticker 24 1,000 ,000 ,000
1 (pak mij) 24 ,000 1,000 ,000
2 (pak mij, flexibiliteit) 24 ,000 ,000 1,000
3 (pak mij, kwaliteit) 24 ,000 ,000 ,000
Block 1: Method = Enter
Omnibus Tests of Model Coefficients
Chi-square df Sig.
Step 1 Step ,501 3 ,919
Block ,501 3 ,919
Model ,501 3 ,919
Model Summary
Step -2 Log likelihood
Cox & Snell R
Square
Nagelkerke R
Square
1 132,417a ,005 ,007
a. Estimation terminated at iteration number 2 because parameter
estimates changed by less than ,001.
57
Classification Tablea
Observed
Predicted Oudste_datum_pak_Y
Percentage Correct 0 1
Step 1 Oudste_datum_pak_Y 0 39 11 78,0
1 33 13 28,3
Overall Percentage 54,2
a. The cut value is ,500
Variables in the Equation
B S.E. Wald df Sig. Exp(B)
Step 1a Conditie ,499 3 ,919
Conditie(1) -,334 ,579 ,333 1 ,564 ,716
Conditie(2) -,334 ,579 ,333 1 ,564 ,716
Conditie(3) -,334 ,579 ,333 1 ,564 ,716
Constant ,167 ,410 ,166 1 ,683 1,182
a. Variable(s) entered on step 1: Conditie.
58
59
Appendix 6A – Follow-up Analysis Study 2 with ‘time’
Categorical Variables Codings
Frequency
Parameter coding
(1) (2) (3)
Conditie 0 (geen sticker 24 ,000 ,000 ,000
1 (pak mij) 24 1,000 ,000 ,000
2 (pak mij, flexibiliteit) 24 ,000 1,000 ,000
3 (pak mij, kwaliteit) 24 ,000 ,000 1,000
Block 1: Method = Enter
Omnibus Tests of Model Coefficients
Chi-square df Sig.
Step 1 Step 7,761 11 ,735
Block 7,761 11 ,735
Model 7,761 11 ,735
Model Summary
Step -2 Log likelihood
Cox & Snell R
Square
Nagelkerke R
Square
1 125,156a ,078 ,104
a. Estimation terminated at iteration number 4 because parameter
estimates changed by less than ,001.
Classification Tablea
Observed
Predicted Oudste_datum_pak_Y Percentage
Correct 0 1
Step 1 Oudste_datum_pak_Y 0 31 19 62,0
1 18 28 60,9
Overall Percentage 61,5
a. The cut value is ,500
60
Variables in the Equation
B S.E. Wald df Sig. Exp(B)
Step 1a Conditie 1,509 3 ,680
Conditie(1) ,886 ,860 1,061 1 ,303 2,426
Conditie(2) ,892 ,862 1,071 1 ,301 2,440
Conditie(3) ,844 ,847 ,995 1 ,319 2,327
Dag_MC -,239 ,334 ,514 1 ,473 ,787
Conditie * Dag_MC 1,789 3 ,617
Conditie(1) by Dag_MC ,349 ,436 ,642 1 ,423 1,418
Conditie(2) by Dag_MC -,290 ,531 ,297 1 ,585 ,748
Conditie(3) by Dag_MC ,048 ,441 ,012 1 ,913 1,049
Dag2 ,221 ,193 1,306 1 ,253 1,247
Conditie * Dag2 2,253 3 ,522
Conditie(1) by Dag2 -,252 ,252 1,000 1 ,317 ,777
Conditie(2) by Dag2 -,418 ,288 2,103 1 ,147 ,659
Conditie(3) by Dag2 -,167 ,253 ,436 1 ,509 ,846
Constant -,903 ,628 2,071 1 ,150 ,405
a. Variable(s) entered on step 1: Conditie, Dag_MC, Conditie * Dag_MC , Dag2, Conditie * Dag2 .
Variables in the Equation
B S.E. Wald df Sig. Exp(B)
Step 1a D2 ,886 ,860 1,061 1 ,303 2,426
D3 ,892 ,862 1,071 1 ,301 2,440
D4 ,844 ,847 ,995 1 ,319 2,327
D1XDag_MC -,239 ,334 ,514 1 ,473 ,787
D2XDag_MC ,110 ,281 ,154 1 ,695 1,116
D3XDag_MC -,529 ,414 1,637 1 ,201 ,589
D4XDag_MC -,191 ,288 ,440 1 ,507 ,826
D1XDag2 ,221 ,193 1,306 1 ,253 1,247
D2XDag2 -,031 ,162 ,037 1 ,847 ,969
D3XDag2 -,197 ,213 ,849 1 ,357 ,822
D4XDag2 ,054 ,164 ,107 1 ,744 1,055
Constant -,903 ,628 2,071 1 ,150 ,405
a. Variable(s) entered on step 1: D2, D3, D4, D1XDag_MC, D2XDag_MC, D3XDag_MC, D4XDag_MC,
D1XDag2, D2XDag2, D3XDag2, D4XDag2.
61
Appendix 6B – Follow-up Analysis Study 2 with ‘intention to avoid food waste’
Categorical Variables Codings
Frequency
Parameter coding
(1) (2) (3)
Conditie 0 (geen sticker 24 ,000 ,000 ,000
1 (pak mij) 24 1,000 ,000 ,000
2 (pak mij, flexibiliteit) 24 ,000 1,000 ,000
3 (pak mij, kwaliteit) 24 ,000 ,000 1,000
Block 1: Method = Enter
Omnibus Tests of Model Coefficients
Chi-square df Sig.
Step 1 Step 5,646 7 ,582
Block 5,646 7 ,582
Model 5,646 7 ,582
Model Summary
Step -2 Log likelihood
Cox & Snell R
Square
Nagelkerke R
Square
1 127,271a ,057 ,076
a. Estimation terminated at iteration number 4 because parameter
estimates changed by less than ,001.
Classification Tablea
Observed
Predicted Oudste_datum_pak_Y Percentage
Correct 0 1
Step
1
Oudste_datum_pak_Y 0 37 13 74,0
1 23 23 50,0
Overall Percentage 62,5
a. The cut value is ,500
62
Variables in the Equation
B S.E. Wald df Sig. Exp(B)
Step 1a Conditie 3,921 3 ,270
Conditie(1) -2,822 2,417 1,363 1 ,243 ,059
Conditie(2) -5,899 3,136 3,539 1 ,060 ,003
Conditie(3) -3,246 2,460 1,742 1 ,187 ,039
Intention_to_avoid_food_waste -,489 ,322 2,300 1 ,129 ,614
Conditie *
Intention_to_avoid_food_waste 4,404 3 ,221
Conditie(1) by
Intention_to_avoid_food_waste ,533 ,444 1,437 1 ,231 1,704
Conditie(2) by
Intention_to_avoid_food_waste 1,085 ,556 3,805 1 ,051 2,959
Conditie(3) by
Intention_to_avoid_food_waste ,687 ,461 2,215 1 ,137 1,987
Constant 2,428 1,787 1,846 1 ,174 11,338
a. Variable(s) entered on step 1: Conditie, Intention_to_avoid_food_waste, Conditie *
Intention_to_avoid_food_waste .
Block 1: Method = Enter
Omnibus Tests of Model Coefficients
Chi-square df Sig.
Step 1 Step 4,767 7 ,688
Block 4,767 7 ,688
Model 4,767 7 ,688
Model Summary
Step -2 Log likelihood
Cox & Snell R
Square
Nagelkerke R
Square
1 128,151a ,048 ,065
a. Estimation terminated at iteration number 3 because parameter
estimates changed by less than ,001.
63
Classification Tablea
Observed
Predicted Oudste_datum_pak_Y
Percentage Correct 0 1
Step 1 Oudste_datum_pak_Y 0 25 25 50,0
1 14 32 69,6
Overall Percentage 59,4
a. The cut value is ,500
Variables in the Equation
B S.E. Wald df Sig. Exp(B)
Step 1a Intention_to_avoid_binary -1,386 ,894 2,402 1 ,121 ,250
D2XLow_intention -,693 ,886 ,611 1 ,434 ,500
D3XLow_intention -1,281 ,901 2,023 1 ,155 ,278
D4XLow_intention -,442 ,868 ,259 1 ,611 ,643
D2XHigh_intention ,288 ,847 ,115 1 ,734 1,333
D3XHigh_intention 1,099 ,847 1,684 1 ,194 3,000
D4XHigh_intention ,693 ,894 ,601 1 ,438 2,000
Constant ,000 ,447 ,000 1 1,000 1,000
a. Variable(s) entered on step 1: Intention_to_avoid_binary, D2XLow_intention, D3XLow_intention,
D4XLow_intention, D2XHigh_intention, D3XHigh_intention, D4XHigh_intention.
64
65
Appendix 6C – Follow-up Analysis Study 2 with ‘noticing the message labels’
• Main, initial analysis but with the following filter: IF Conditie=1 OR Conditie > 1 & Label_Awareness=1
Categorical Variables Codings
Frequency
Parameter coding
(1) (2) (3)
Conditie 0 (geen sticker 24 .000 .000 .000
1 (pak mij) 16 1.000 .000 .000
2 (pak mij, flexibiliteit) 22 .000 1.000 .000
3 (pak mij, kwaliteit) 20 .000 .000 1.000
Block 1: Method = Enter
Omnibus Tests of Model Coefficients
Chi-square df Sig.
Step 1 Step .367 3 .947
Block .367 3 .947
Model .367 3 .947
Model Summary
Step -2 Log likelihood
Cox & Snell R
Square
Nagelkerke R
Square
1 113.309a .004 .006
a. Estimation terminated at iteration number 3 because parameter
estimates changed by less than .001.
Classification Tablea
Observed
Predicted Oudste_datum_pak_Y Percentage
Correct 0 1
Step 1 Oudste_datum_pak_Y 0 13 28 31.7
1 11 30 73.2
Overall Percentage 52.4
a. The cut value is .500
66
Variables in the Equation
B S.E. Wald df Sig. Exp(B)
Step 1a Conditie .366 3 .947
Conditie(1) .167 .646 .067 1 .796 1.182
Conditie(2) .167 .591 .080 1 .778 1.182
Conditie(3) .368 .608 .366 1 .545 1.444
Constant -.167 .410 .166 1 .683 .846
a. Variable(s) entered on step 1: Conditie.