Food for thought:
temporal distance
and the purchase –
consumption difference
And the effect of household size
in this food waste context
Food for thought: temporal distance and the
purchase - consumption difference
And the effect of household size in this food waste context
Name: Christiaan Mensink
Address: Aquamarijnstraat 685, Groningen Phone number: +31 627481613
E-mail: d.c.mensink@student.rug.nl Completion date: June 17th, 2019
Student number: 2893940 Qualification: Master Thesis
First Supervisor: Prof. Dr. Jenny van Doorn Second Supervisor: Marit Drijfhout
1
Management Summary
2
Preface
Before you lies the thesis “Food for thought: temporal distance and the purchase – consumption difference”. Writing this thesis was one of the graduation requirements for the MSc Marketing at the University of Groningen. It has been written between February and June.
I would like to thank my supervisor for her guidance and quick responses to questions that I had. Also, I would like to thank the research group that I was part of for their comments. I would like to thank the respondents of my survey as well, because I would not have been able to conduct research without them. Lastly, I would like to thank my family and friends for debating content-related, but also technical issues with me.
I hope that you will enjoy your reading.
Christiaan Mensink
3
Table of contents
1. Introduction ... 4
2. Theoretical framework ... 6
2.1 Food waste and its antecedents ... 6
2.2 Conceptual model ... 9
2.3 Effect of temporal distance ... 10
2.4 Effect of household size ... 12
3. Method ... 14
3.1 Measurement of constructs ... 15
3.2 Testing the effect of temporal distance on food waste ... 16
4. Results ... 17
4.1 The effect of temporal distance ... 17
4.2 Household size and other covariates ... 18
4.3 Household size as a moderator ... 19
5. Discussion... 22
5.1 Summary of findings ... 22
5.2 Implications... 24
5.3 Limitations ... 26
4
1.
Introduction
“Imagine walking out of a grocery store with four bags of groceries, dropping one in the parking lot, and just not bothering to pick it up. That is essentially what we are doing.”
This quote from Dana Gunders in the food waste documentary Just Eat It accurately summarizes the general attitude towards food: we carelessly waste lots of food. FAO (2011) estimates industrialized countries’ food waste at the consumer and retail level to be around 40%. Several studies indicate that food waste at the household level has one of the largest savings potentials (e.g. Jörissen et al., 2015, Hebrok & Boks, 2017, Stenmark et al., 2016, Canali, 2014). This paper will therefore focus on food waste on the household level.
During the last years, investigating food waste in households has been a growing research field (Porpino, 2016). In those papers, purchasing too much is found to be the main determinant of food waste, after which other reasons follow: over-preparation, avoidance of leftovers and inappropriate food conservation or stockpiling issues (Porpino et al., 2015). Evans (2011a) and Williams et al. (2012) find that the excessive purchase may be due to the package size: the purchased volume exceeds the volume needed. This also pops up in the mismatch between the rhythm of daily life and the time to prepare food. Stefan et al. (2013) and Stancu et al. (2016) stress the importance of food provisioning routines in planning and shopping. The more routined consumers are, the less food is wasted. Chandon & Wansink (2006) find that the main cause of food waste is underestimating high levels of inventory. Also, consumers do not learn from previous mistakes.
5 Although the mentioned effect is expected to affect household food waste for households in general, its magnitude may differ, when other variables are taken into account. In the following chapters, especially household size is considered as an interacting variable in this context. This focus follows from findings indicating that the avoidable waste in one-person households is relatively big (Koivupuro et al., 2012, Porpino, 2016). Although this sociologic difference is proven to be relevant in the food waste literature (e.g. Koivupuro et al., 2012, Porpino, 2016), not much research behind this has been conducted yet.
More specifically, previous research lacks reasons and implications to combat the food waste problems arising in this segment. However, the relative increase of single-person households is recognized as a possible driver to increase food waste in the future (Canali, 2014, Porpino, 2016). Therefore, it is important to find reasons behind this phenomenon. Two possible reasons that will be discussed are lifestyle (Comber et al., 2013, Ganglbauer et al., 2013) and the stimulation of over-shopping (Williams et al., 2012, Canali, 2014). These reasons are both relatable to construal level theory, which will be discussed in detail in the following chapter.
6
2.
Theoretical framework
2.1 Food waste and its antecedents
2.1.1 Food waste at the household level
As defined by Ganglbauer et al. (2013), household food waste is “the unintended result of multiple moments of consumption dispersed in space and time across other integrated practices such as shopping and cooking, which are themselves embedded in broader contextual factors and values”. In other words, household food waste is a result of the mismatch between expectation and reality, with respect to all factors at the consumer level. To find out more about this phenomenon, this section divides the food waste problem into three sections: purchasing, storage and consumption.
If we first look into the purchasing stage, consumers buy more than they need. This is called excessive purchasing. The main reasons found for purchasing excessively by Porpino et al. (2015) can be categorized into two subcategories. Firstly, impulse buying and unplanned purchase is found to increase food waste. Stefan et al. (2013) found that moral attitudes drive this: consumers do not always bother about food waste. The second category consists of the large package preference of consumers and the role of promotions in this context. The high frequency of special offers and big package promotions also stimulate purchasing too much (Koivupuro et al., 2012, Cox & Downing, 2007). Because small packages are limited available, consumers are sometimes obliged to choose for (too) big packages (Koivupuro et al., 2012). For both suggested reasons, consumers do not learn from previous mistakes by themselves (Chandon & Wansink, 2006), which makes it a hard problem to combat.
7 A systematic approach would also help in storing the food properly. This increases the likelihood that food is consumed before the expiration date (Farr-Wharton et al., 2014). Also, consumers do not always store their food with care: it is then stored under suboptimal conditions (Koivupuro et al., 2012, Hebrok & Boks, 2017, Cox & Downing, 2007). For instance, people do not always freeze products, while freezing would extend the expiration date. One way to tackle this problem could be to educate people more thoroughly about food storage (Porpino et al., 2015), because environmentally educated households waste less (Williams et al., 2012).
For most products, the storage phase ends when a meal is prepared: the consumption phase starts. In this food waste stage, over-preparation plays a noteworthy role (Cox & Downing, 2007). Porpino et al. (2015) argue that the inability to plan meals is a first reason for this. Secondly, many consumers see having much food as wealth, hence, they have a taste for abundance. Lastly, hospitality towards all people that consume your prepared meal is mentioned as a reason for over-preparation. This is further explained by the good provider identity, which states that people’s evaluation of the food provider depends on the amount of food that is prepared (Visschers et al., 2016, Graham-Rowe et al., 2014). Another contributor to food waste at this level is the avoidance of leftovers (Hebrok & Boks, 2017, Stancu et al., 2018, Farr-Wharton et al., 2014). People have a prejudice against leftovers or have a freshness preference (Porpino et al., 2015). Lastly, the overall consumer awareness about food and waste is related to food waste production (Parizeau et al., 2015, Stefan et al., 2013).
8
2.1.2 Households characteristics
Although food waste on the consumer level has been of increasing interest for researchers (Porpino, 2016), existing research with a focus on household types is limited. In this section, the effect of household size will be considered first. After that, a research of Wenlock et al. (1980) will be discussed, to examine the effect of household composition.
Several studies indicate that food waste is positively related with the number of household occupants (e.g. Koivupuro et al., 2012, Wenlock et al., 1980). This seems reasonable, since the amount of food purchased also increases with the number of household occupants, and more food means more possibilities to waste food. However, the waste per capita decreases when household size increases (Wenlock et al., 1980, Baker et al., 2009). Also, avoidable food waste is estimated to be larger for smaller households than for bigger households (Koivupuro et al., 2012). Avoidable food waste is in this case defined as food or raw materials that could have been consumed if it was stored or prepared differently.
Next to household size, its composition may play a role as well. Wenlock et al. (1980) studied the effect of household composition on food wastage. They measured the amount of energy that was discarded, as a content total and as a percentage of purchased products. To analyze household composition effects, they analyzed the number of adults and the number of children separately. For adults, it was found that the more adults a household consists of, the more food is wasted. However, there was no significant increase in food waste as a percentage of products purchased. Also, the more children in the household, the more absolute food waste, but this effect is smaller than that of more adults. When the number of children increased, the percentage of food wasted as percentage of food purchased decreased. Also, each extra child added less waste to the total than the previous one.
9 2.2 Conceptual model
Before, there has been elaborated on food waste and its antecedents. Household food waste is essentially the difference between purchased and consumed food. This research argues that the bigger the difference in time between the moment of purchase and the moment of consumption is, the bigger the quantified difference between purchased and consumed food is, and hence, the bigger the amount of food waste. The following section will explain the underlying theory, which (among other reasons) stresses the importance of taking into account contextual factors (Griffin & Ross, 1991) and spontaneous purchasing decisions (Cox & Downing, 2007, Farr-Wharton et al., 2014). This temporal distance effect is expected to be even stronger for smaller households, through flexibility (Ganglbauer et al., 2014) and the likelihood of over-shopping (Canali, 2014, Evans, 2011a). The conceptual model below shows the expected relationships between the variables of interest: household size is expected to moderate the effect of temporal distance on the difference between buying and consuming food. The following sections will discuss why these effects are expected to happen, based on theory.
Temporal distance Δ Buying/consuming
10 2.3 Effect of temporal distance
Consumers often do their groceries in advance. Stefan et al. (2013) find that 92% of the consumers does a main shopping trip once a week or less often, where most of the products for that week are bought. They also find that only 16% does a smaller shopping trip each day, hence, consumers most often buy meals in advance. Thereby, they create a temporal distance between the point of purchase and the point of consuming. This temporal distance could be a reason for the throw-away of fresh foods. Temporal distance is one of the psychological distances proposed in construal level theory (Trope & Liberman, 2003). This theory proposes that our mind can predict, remember, imagine and speculate about events from the past and the future by forming abstract mental construals of events that are distal (Liberman & Trope, 1998, Hamilton & Thompson, 2007).
Construal level theory is used to explain the trade-off between feasibility and desirability. Feasibility deals here with the ease or difficulty and the means used of reaching the end state, while desirability brings up the attractiveness or value of the end state (Trope & Liberman, 2010). Furthermore, feasibility is connected to near events of concrete meaning, while distant events that are formulated more abstract are connected to desirability (Liberman & Trope, 1998). Therefore, desirability concerns should be more important than feasibility concerns as psychological distance increases. As reason for the difference between feasibility and desirability decisions, Trope & Liberman (2003) argue that people are often free to change their decisions with regard to events that take place in the future.
11 This reasoning is supported by research into the concept of planning fallacy. This fallacy is a consequence of focusing on the specific issue without taking the distribution of outcomes in comparable cases into account (Kahneman & Tversky, 1977, Buehler et al., 1994). This is what also happens in the food waste context: people do not learn from previous mistakes in purchasing excessively (Chandon & Wansink, 2006). Griffin & Ross (1991) confirm this finding in general: predictions are based on models that often underestimate contextual factors.
An example of a contextual factor is time. Buehler et al. (1994) find in two studies that respondents predicted a few days in advance to be more productive than they actually were on the day itself. Gilovich et al. (1993) also confirm that the more distal the to be performed task is, the more optimistic people on average are about success. Applied to the food waste problem, time is also considered as a contextual factor. One example that is discussed in previous literature is the time to prepare and purchase food on the day itself (Evans, 2011a, Graham-Rowe et al., 2014). This contextual factor is not always taken into account in the predictions that take place during the purchasing stage. It is expected that the underestimation of those contextual factors is strongly related with purchasing inaccurate amounts, or even the wrong products.
Hence, several reasons suggest a positive effect of temporal distance on the difference between purchased and consumed food. Spontaneous purchasing and spur-of-the-moment decisions result in an ignorance of the food that is already in stock (Cox & Downing, 2007, Farr-Wharton et al., 2014). Also, the underestimation of future contextual factors results in purchasing food that is eventually not consumed (Griffin & Ross, 1991, Evans, 2011a). Based on this, it is expected that the bigger the temporal distance between purchase and consumption, the bigger the difference between purchased and consumed food. A higher difference between purchased and consumed food is in the end an increase in food waste.
12 2.4 Effect of household size
It has been found that adults throw more products away than children, and larger households have less waste per person than smaller households (Wenlock et al., 1980). Also, the avoidable food waste is estimated to be larger for smaller than bigger households (Koivupuro et al., 2012). Two possible reasons for this are found in the existing literature.
The first reason for the mismatch between daily life and purchase behavior could be the lifestyle of individuals (Buckley et al., 2007). Lifestyle of households is found to affect food waste (Parizeau et al., 2015). Ganglbauer et al. (2013) argue that one-person households are more flexible: they are relatively more away from home than at home. Hence, shopping for a few days in advance is a rougher estimate of eating at home than for families. Also, after an explorative survey of Bisp (2014), a factor analysis shows several statements where families have benefits to smaller households (e.g. “When I do not really feel like cooking, I can get one of the kids or my husband to do it”) contribute to the factor meal preparation. Lastly, it is found that single-person households waste more because they lack the opportunity to share food (Monier et al., 2010).
The suggested lifestyle effect strengthens the effect of temporal distance on food waste. Because the planning of small households is more flexible, the food that was purchased in advance may not be consumed (Ganglbauer et al., 2014). The larger the difference between the purchase moment and the possible consumption moment, the more unsure consumption will be. Also, the effect of underestimating contextual factors may be applicable to a larger extent to small households, since the task of preparing food cannot be divided over more people if time is limited (Bisp, 2014).
13 The stimulation of over-shopping is also relatable to the effect of temporal distance on the difference between buying and consuming food. If temporal distance is small, there are multiple possibilities to consume food before its expiration date. However, if temporal distance increases, the food will be longer in stock in households, and the longer a product is in stock, the higher the chances of wasting it (Chandon & Wansink, 2006). Hence, too big packages increase the amount of purchased food more than the amount of consumed food.
As argued before, it is expected that temporal distance affects the difference between purchased and consumed food. Although there is no previous research in construal level theory with regard to group versus individual behavior, the suggested disadvantages (lifestyle (e.g. Ganglbauer et al., 2014) and over-shopping (e.g. Chandon & Wansink, 2006)) give enough reasons to investigate a moderating effect of household size in this context. This paper tests whether the reasons for excessive purchase suggested before are applicable to smaller households to a larger extent, which yields the following hypothesis:
14
3.
Method
Data were collected in May 2019 by using a web-based questionnaire with Qualtrics software. The questionnaire was developed in English and translated to Dutch, to get as many respondents as possible. The survey was firstly distributed via the researcher’s network. Secondly, it was distributed via Amazon Mturk. Two identical surveys were published: one where only small households could take part, and one where families could take part. In total, 218 respondents answered the survey. 5 respondents were excluded because they reported no household size. 1 respondent was excluded for not answering the scenario question. The following table shows the demographics of the remaining 212 respondents. 4 respondents did not report their nationality, and 1 respondent did not indicate his age.
Table 1: Descriptives of demographic questions
Demographic Type Respondents
15 3.1 Measurement of constructs
3.1.1 Manipulation of temporal distance
The questionnaire manipulated temporal distance by including two scenarios, to which the participants were randomly assigned. In the first scenario, people were asked to imagine a situation in which they were buying ingredients for a pasta meal that would take place on that same day. This represents a small temporal distance between the moment of purchase and the moment of consumption. In the other scenario, people were asked to imagine a situation in which they were in the supermarket to buy ingredients for a pasta meal that would take place in three days. This represents a big temporal distance between the moment of purchase and the moment of consumption. Three days were chosen because that seems to be a reasonable number of days for all different types of consumers: from those consumers that divide their main shopping trip of a week into two or three smaller shopping trips to the consumers that do their main shopping trip less than once a week (Stefan et al., 2016).
3.1.2 Ingredients meal
The meal for which consumers needed to do groceries, was a pasta meal. Consumers were able to choose different ingredients (e.g. macaroni, minced meat, tomatoes), different package sizes (e.g. 250 grams or 500 grams) and different types of ingredients (e.g. pre-cut or whole units).
3.1.3 Household size
Household size was measured in two questions: the number of adults (on a scale from 0 to 5+) and the number of children (on a scale from 0 to 5+). In this way, it is possible to include the variable household size in different ways of aggregation when testing its effect. First, it will be included as a continuous variable, and as a robustness check, it will be aggregated into small (one- and two-person) and big (three- or more-two-person) households.
3.1.4 Dependent variables
16 think of by themselves. The second question asked what the chances were of cooking a meal with the products that were bought. Both percentage questions used a slider, to overcome the disadvantages of deciles, that were described by Graham-Rowe et al. (2014). The questionnaire used mainly questions with a proportional scale to gather information about general food waste. Participants were asked to indicate their food waste as a percentage of their total purchases, as in other research (e.g. Stefan et al., 2013 & Stancu et al., 2016). By linking the amount of food waste to the food brought into the home, proportional questions attempt to reduce the biased estimates due to consumer memory (Van Herpen et al., 2016).
Table 2: Dependent variable questions
Questions
What are the chances that you would choose for one of the suggested alternatives above?
What are the chances that you will prepare a meal with the ingredients you chose in the supermarket?
3.2 Testing the effect of temporal distance on food waste
To test the first hypothesis, a one-way ANOVA is used, to see the difference between the two scenarios. In this way, the results will indicate whether the scenarios show a significant different effect on the proposed dependent variables.
To test for moderation (and hence, the second hypothesis), an ANCOVA test with interaction is used. Here, the household size is included as a continuous covariate. An interaction term will show whether or not household size significantly affects this relationship.
17
4.
Results
4.1 The effect of temporal distance
4.1.1 Chances of using the ingredients
Shapiro-Wilk tests for normality show that the expected chance to use the ingredients is for both scenario 1 (W = 0.942, p < 0.001) and scenario 2 (W = 0.926, p < 0.001) non-normally distributed. Levene’s test for homogeneity of variances shows W(1,210) = 0.860, p = 0.355. A one-way ANOVA determined a statistically significant difference between groups (F(1,210) = 3.112, p = 0.079) for the manipulated effect of temporal distance on the chances to use ingredients. Respondents assigned to scenario 1 reported a lower chance to use ingredients (M = 67.78, SD = 23.03) than those in scenario 2 (M = 73.22, SD = 21.45).
4.1.2 Chances of choosing for the alternative
Shapiro-Wilk tests for normality show that the expected chance to choose for one of the alternatives is for both scenario 1 (W = 0.948, p < 0.001) and scenario 2 (W = 0.937, p < 0.001) non-normally distributed. Levene’s test for homogeneity of variances shows W(1,210) = 1.427, p = 0.234. A one-way ANOVA determined no statistically significant difference between groups (F(1,210) = 0.411, p = 0.522) for the manipulated effect of temporal distance on the choice for the alternative instead of cooking a meal with the purchased ingredients. Respondents assigned to scenario 1 reported a higher chance of choosing the alternative (M = 47.88, SD = 27.95) than those in scenario 2 (M = 45.34, SD = 29.84).
Figure 1: Chances to use ingredients Figure 2: Chances to go for the alternative
18 4.2 Household size and other covariates
In table 3, the results of an ANCOVA test are summarized, with the chances to use the ingredients as independent variable, the scenario as factor and several covariates. Levene’s test shows
W(1,209) = 0.238, p = 0.626. The model is significant (F(6,204) = 7.100, p < 0.001).
Table 3: ANCOVA on chances to use ingredients
Scenario Chances to use ingredients
Observed mean Adjusted mean SD n
Now 67.78 68.06 23.03 117 In three days 73.15 72.80 21.56 94 Source SS df MS F p Scenario 1160.60 1 1160.60 2.693 .102 Household size 1544.45 1 1544.45 3.584 .060 American 13384.90 1 13384.90 31.058 .000 Indian 3314.03 1 3314.03 7.690 .000 Male 394.82 1 394.82 0.916 .340 Age 647.69 1 647.69 1.503 .222 Error 87917.61 204 430.97
Note. R2 = 0.173, Adj. R2 = 0.148, Estimated regression coefficients: Scenario(=1) = -4.75, Household size = 1.65, American = -20.32, Indian = -10.12, Male = 2.83, Age = 0.19.
This ANCOVA test shows that the scenarios do no longer differ significantly (p = 0.102), while people in scenario 1 still report lower expectancies of using the ingredients. The chances to use the ingredients increase when household size increases (p = 0.060), and decrease for respondents from American countries (p < 0.001) and India (p < 0.001), when they are compared to European respondents. The covariates for gender and age are not significant, and will therefore be excluded in further analyses. As a robustness check, household size was aggregated into small and big households. The continuous variable was excluded, and a dummy for big households was included to test this. This dummy affected the model significantly (F = 4.275, p = 0.040), with an estimated effect of B = 6.749. The significance of other variables in the model did not change noteworthy.
Table 4 shows an ANCOVA test on the chances to go for the alternative. Levene’s test shows
W(1,209) = 1.225, p = 0.270. The model is significant (F = 20.180, p < 0.001). Results indicate
19 household size was checked. The effect of the dummy for household size was estimated as F = 0.318, p = 0.318, with B = -3.636. The significance of other variables in the model did not change noteworthy.
Table 4: ANCOVA on chances to go for the alternative
Scenario Chances to go for alternative
Observed mean Adjusted mean SD n
Now 47.89 47.69 27.94 117 In three days 45.60 45.86 29.88 94 Source SS df MS F p Scenario 172.71 1 172.71 0.323 .571 Household size 4.31 1 4.31 0.008 .929 American 22106.62 1 22106.62 41.314 .000 Indian 53144.90 1 53144.90 99.321 .000 Male 11.17 1 11.17 0.021 .885 Age 235.30 1 235.30 0.440 .508 Error 109156.60 204 535.08
Note. R2 = 0.372, Adj. R2 = 0.354, Estimated regression coefficients: Scenario(=1) = 1.83, Household size = 0.09, American = 26.11, Indian = 40.53, Male = 0.48, Age = -0.11
4.3 Household size as a moderator
An ANCOVA test was performed to test for the moderating effect of household size in the relationship between scenario and chances to use ingredients. Insignificant covariates of table 3 were not included for this analysis. Results are summarized in table 5. The model is significant (F = 7.850, p < 0.001). Levene’s test shows W(1,210) = 0.444, p = 0.506. The results indicate that both the scenario (p = 0.525) and the interaction effect between scenario (p = 0.877) are insignificant. The covariates are still significant, where the chances to use the ingredients increase when household size increases (p = 0.093), and decrease when respondents were not from Europe, but from American countries (p < 0.001) or India (p = 0.010).
Again, as a robustness check, aggregated household size was included in the model instead of the continuous variable. The effects of scenario (F = 2.265, p = 0.134) and the interaction effect (F = 0.463, p = 0.497) were still insignificant but had lower p-values, while household size (F = 3.539,
20
Table 5: ANCOVA on chances to use ingredients with interaction
Scenario Chances to use ingredients
Observed mean Adjusted mean SD n
Now 67.78 68.18 23.03 117
In three days 73.15 72.71 21.56 94
Source SS df MS F p
Scenario 176.24 1 176.24 0.406 .525
Household size 1238.73 1 1238.73 2.856 .093
Scenario * Household size 10.35 1 10.35 0.024 .877
American 12089.20 1 12089.20 27.872 .000
Indian 2956.32 1 2956.32 6.816 .010
Error 89349.08 206 433.73
Note. R2 = 0.160, Adj. R2 = 0.140, Estimated regression coefficients: Scenario(=1) = -3.74, Household size = 1.58, Scenario(=1)*Household size = -0.245, American = -18.71, Indian = -9.37
Table 6 summarizes the results of an ANCOVA test on chances to go for the alternative with an interaction term between the scenario and household size to test for moderation. The insignificant covariates from table 4 (Male & Age) are not included in this model. The model is again significant (F = 24.146, p < 0.001), and Levene’s test for equality of variances shows W(1,210) = 1.090, p = 0.298. A significant interaction effect was not found (p = 0.471). The covariates for nationality are still significant (both p < 0.001). The robustness check on household size revealed no significant interaction effect (F = 0.115, p = 0.735) as well, which also holds for the effects of scenario (F = 0.012, p = 0.912) and household size (F = 1.086, p = 0.298).
Table 6: ANCOVA on chances to go for the alternative with interaction
Scenario Chances to go for alternative
Observed mean Adjusted mean SD n
Now 47.89 47.70 27.94 117
In three days 45.60 45.67 29.88 94
Source SS df MS F p
Scenario 53.02 1 53.02 0.099 .753
Household size 2.28 1 2.28 0.004 .948
Scenario * Household size 279.25 1 279.25 0.522 .471
American 21725.05 1 21725.05 40.639 .000
Indian 55128.93 1 55128.93 103.125 .000
Error 110124.45 206 534.59
21
Figure 3 and 4 show plots of the interaction effect of scenario and household size on both dependent variables. For the purpose of clarity, the aggregated version of household size (small vs. big) was used. For the chances to use ingredients (figure 3), the difference between households is smaller in scenario 2 than in scenario 1, although this difference is not significant. For the chances to go for the alternative, the difference between households is bigger in scenario 2, but this effect is also not significant.
Figure 3: Interaction effect in the chances to Figure 4: Interaction effect in the chances to use the ingredients go for the alternative
63.32 70.29 71.4 74.34 56 58 60 62 64 66 68 70 72 74 76 1 2 Cha nc es to us e in gr edie nts > Scenario >
Small households Big households
46.6 43.8 49.23 48.66 41 42 43 44 45 46 47 48 49 50 1 2 Cha nc es to go for a lte rn ative > Scenario >
22
5.
Discussion
5.1 Summary of findings
5.1.1 Temporal distance and chances of using ingredients
The preceding findings show a significant effect of temporal distance on the chance to use the ingredients that respondents chose in a previous part of the survey. When people in the sample do their groceries for a meal that they expect to prepare in three days, they expect a higher chance of consuming what they bought than when they do their groceries for today. This does not implicitly mean that food waste increases when time between purchase and consumption increases. However, it suggests that consumers indeed underestimate the contextual factors of the event that will take place in the future (Buehler et al., 1994, Griffin & Ross, 1991). Previous research has shown that the products that are bought in advance decrease in value on the day itself: top up shops and the associated spur-of-the-moment decisions lead to the usage of new products, instead of the products that are in stock (Farr-Wharton et al., 2014, Cox & Downing, 2007). The products in stock then stay in stock, and this process can repeat itself several times before the food goes to waste, dependent on the expiration date. Hence, underestimating contextual factors of future events is the first domino in line that falls.
5.1.2 Temporal distance and the chance to choose for the alternative
23
5.1.3 The effect of household size
Next to the main effect of temporal distance, an interaction effect with household size was tested. For both the chances to cook with the ingredients that were bought, as well as the possibility to choose for an alternative, no significant interaction effect was found. Hence, it cannot be concluded that temporal distance effects are bigger for smaller households, although lifestyle (Ganglbauer et al., 2014, Parizeau et al., 2015) and over-shopping because of too big packages (Evans, 2011a, Evans, 2011b) suggested that this might have been true. A possible reason for this is that small households in this sample are efficient, because they have learnt from previous mistakes. A mean age of 32.8 gives reasons to believe that.
Although the interaction effect is not significant, household size covaries significantly with the chances to use the ingredients. The bigger the household, the bigger the chances to use the ingredients. Here, lifestyle can be considered as an argument: small households know that they are relatively more from home than households with children (Parizeau et al., 2015). For the chances of choosing an alternative, no significant effect of household size was found. The aforementioned causes of insignificance for this dependent variable may be applicable to the current model as well.
5.1.4 Nationality as covariate
24 5.2 Implications
5.2.1 Implications for policy makers
It is assumed that policy makers want to reduce food waste. The main finding of this research is that people overestimate their use of ingredients for cooking in future situations. Also, small households have smaller expectancies of using ingredients than their bigger counterparts. This last cause of food waste is likely to be the least changeable through policy implications (Monier et al., 2010), but the implications are consumer-oriented, meaning that it will be applicable to all households, irrespective of size. What can then be done to overcome this planning fallacy effect?
Firstly, people should have more attention for future contextual factors, because underestimation of those factors results in a higher difference between purchased and consumed food (Griffin & Ross, 1991). One efficient way to do that is by urging consumers to make a shopping list (Ganglbauer et al., 2013, Chandon & Wansink, 2006). Thereby, consumers will consider which products are already at home and for which future meals they should buy products. This would also prevent negative effects of frequent top-up shopping, that were described by Cox & Downing (2007). Eventually, this will result in effective planning and shopping routines, which will reduce food waste (Stefan et al., 2013, Farr-Wharton et al., 2014). Also, more research into these contex-tual factors and to what extent they affect purchasing food should be conducted.
Secondly, there should be more emphasis on what people will feel while they cook, but also while they waste. This could be an important way to decrease the temporal distance effect. Farr-Wharton et al. (2014) argue that consumers’ negative experiences in previous situations with preparing food change their consumption pattern. Hence, consumers should be remembered to past mistakes. It is also found that consumers feel guilty when they waste food and that they want to prevent this guilty feeling (Parizeau et al., 2015, Stefan et al., 2013). Therefore, preventing guilty feelings could be used in campaigns to promote food saving behavior. To know what effect emphasizing other feelings has, policy makers should conduct more research.
25 and not always aware of the food waste problem (Canali, 2014, Parizeau et al., 2015). To have the biggest impact, campaigns should focus on saving money as key reason to combat food waste in households, because that is found to be the most efficient message (Baker et al., 2009). Consumers can be educated more thoroughly about food waste. Educational sessions on how to store products in the most optimal way and on how to cook efficient with ingredients could be initialized to increase food usage efficiency.
Lastly, existing devices that combat food waste could be used more. For example, use of a food waste app could be stimulated, through which households and restaurants can share food that would have been discarded otherwise. Policy makers could play a role here by highlighting the existence of these devices, since it is not yet known by the public at large. Also, subsidies could be given to producers, to be able to supply these devices for a lower price.
5.2.2 Implications for companies
Also, companies can improve their strategy in combatting household food waste. There are several things they can do if we look into the whole food process in households. The policy implications are also applicable to companies, since a company is one of the channels policy makers could use to reach the consumer. Companies can make a difference, especially in the purchasing phase.
One implication here is that they could try to highlight feasibility considerations. Consumers often undervalue contextual factors such as time (Parizeau et al., 2015) or they just not feel like cooking at the day of consumption (Evans, 2011a). With highlighting feasibility considerations, consumers may go for a convenient option. Another option is offering more small packages, to assist small households in their battle against food waste.
26 5.3 Limitations
As other academic literature, this research has its limitations. First of all, self-reported items may be biased estimates of the real behavior of respondents. Other research argues that this is often the case in the food waste context: people think that they discard less food than they report (e.g. Van Herpen et al., 2016). This could be a reason for a less significant effect of the scenarios.
Secondly, the sample might not be representative. One thing to consider is that it had more young than old respondents, with a mean age of 29.89. Reasons for this are its distribution via internet and via the network of the researcher. Also, the Netherlands, USA and India were over-represented if nationality is considered. Because of that, this sample is not representative for all countries in the world, and the results cannot be generalized to all of them. Including nationality as a covariate.
27
6.
References
Baker, D., Fear, J., & Denniss, R. (2009). What a waste - An analysis of household expenditure on food. Policy Brief No, 6, 2009.
Bisp, S. (2014). Food-related lifestyle: development of a cross-culturally valid instrument for market surveil-lance. Values, lifestyles, and psychographics, 337. Buckley, M., Cowan, C., & McCarthy, M. (2007). The
convenience food market in Great Britain: Convenience food lifestyle (CFL) segments. Appetite, 49(3), 600-617.
Buehler, R., Griffin, D., & Ross, M. (1994). Exploring the "planning fallacy": Why people underestimate their task completion times. Journal of personality and social psychology, 67(3), 366.
Canali, M. (2014). Drivers of Current Food Waste Generation, Threats of Future Increase and Oppor-tunities for Reduction. FUSIONS report.
Chandon, P., & Wansink, B. (2002). When are stockpiled products consumed faster? A convenience–salience framework of post-purchase consumption incidence and quantity. Journal of Marketing Research, 39(3), 321-335.
Chandon, P., & Wansink, B. (2006). How biased house-hold inventory estimates distort shopping and storage decisions. Journal of Marketing, 70(4), 118–135. Comber, R., Hoonhout, J., Van Halteren, A., Moynihan,
P., Olivier, P. (2013). Food practices as situated action: exploring and designing for everyday food practices with households. In: Proceedings of the SIGCHI Conference on Human Factors in Computing Systems. ACM, pp. 2457-2466.
Cox, J., & Downing, P. (2007). Food behaviour consumer research: quantitative phase. Retail Programme–Food Waste: Final Report. Material change for better environment. Brook Lyndhurst.
Evans, D. (2011a). Blaming the consumer–once again: the social and material contexts of everyday food waste practices in some English households. Critical Public Health, 21(4), 429-440.
Evans, D. (2011b). Beyond the throwaway society: ordinary domestic practice and a sociological approach to household food waste. Sociology 46, 41-56.
FAO (2011). Global food losses and food waste. Extent, causes and prevention. Conducted for the International Congress SAVE FOOD! At Interpack2011.
Farr‐Wharton, G., Foth, M., & Choi, J. H. J. (2014). Identifying factors that promote consumer behaviours causing expired domestic food waste. Journal of Consumer Behaviour, 13(6), 393-402.
Ganglbauer, E., Fitzpatrick, G., Comber, R., (2013). Negotiating food waste: using a practice lens to inform design. ACM Trans. Comput. Hum. Interact. 20, 1-25.
Gehrt, K. C., & Yale, L. J. (1993). The dimensionality of the convenience phenomenon: A qualitative reexami-nation. Journal of Business and Psychology, 8(2), 163-180.
Gilovich, T., Kerr, M., & Medvec, V. H. (1993). Effect of temporal perspective on subjective confidence. Jour-nal of persoJour-nality and social psychology, 64(4), 552. Graham-Rowe, E., Jessop, D.C., Sparks, P. (2014).
Iden-tifying motivations and barriers to minimising house-hold food waste. Res. Conserv. Recycl. 84, 15-23. Griffin, D. W., & Ross, L. (1991). Subjective construal,
social inference and human misunder-standing. In M. Zanna. (Ed.), Advances in experimental social psychology (Vol. 24, pp. 319-359). New York: Academic Press.
Hamilton, R. W., & Thompson, D. V. (2007). Is there a substitute for direct experience? Comparing con-sumers' preferences after direct and indirect product experiences. Journal of Consumer Research, 34(4), 546-555.
Hebrok, M., & Boks, C. (2017). Household food waste: Drivers and potential intervention points for design– An extensive review. Journal of Cleaner Production, 151, 380-392.
Jörissen, J., Priefer, C., Bräutigam, K.-R. (2015). Food waste generation at household € level: results of a survey among employees of two European research centers in Italy and Germany. Sustainability 7, 2695-2715.
Kahneman, D., & Tversky, A. (1977). Intuitive prediction: Biases and corrective procedures. Management Science, 12, 313-327.
Koivupuro, H. K., Hartikainen, H., Silvennoinen, K., Katajajuuri, J. M., Heikintalo, N., Reinikainen, A., & Jalkanen, L. (2012). Influence of socio‐demographi-cal, behavioural and attitudinal factors on the amount of avoidable food waste generated in Finnish house-holds. International Journal of Consumer Studies, 36(2), 183-191.
Liberman, N., & Trope, Y. (1998). The role of feasibility and desirability considerations in near and distant future decisions: A test of temporal construal theory. Journal of Personality and Social Psychology, 75, 5– 18.
Monier, V., Escalon, V. and O'Connor, C. (2010) Preparatory study on food waste across EU 27, Brussels: European Commission-DG Environment, Technical Report - 2010 – 054.
28
Parizeau, K., von Massow, M., & Martin, R. (2015). Household-level dynamics of food waste production and related beliefs, attitudes, and behaviours in Guelph, Ontario. Waste Management, 35, 207-217. Porpino, G. (2016). Household food waste behavior:
avenues for future research. Journal of the Association for Consumer Research, 1(1), 41-51.
Porpino, G., Parente, J., & Wansink, B. (2015). Food waste paradox: antecedents of food disposal in low income households. International journal of consumer studies, 39(6), 619-629.
Ryan, I., Cowan, C., McCarthy, M., & O'sullivan, C. (2004). Food-related lifestyle segments in Ireland with a convenience orientation. Journal of International Food & Agribusiness Marketing, 14(4), 29-47. Stancu, V., Haugaard, P., & Lähteenmäki, L. (2016).
Determinants of consumer food waste behaviour: Two routes to food waste. Appetite, 96, 7-17.
Stefan, V., van Herpen, E., Tudoran, A. A., & Lähteen-mäki, L. (2013). Avoiding food waste by Romanian consumers: The importance of planning and shopping routines. Food Quality and Preference, 28(1), 375-381.
Stenmark, A., Jensen, C., Quested, T., Moates, G. (2016). Estimates of European food waste levels. Commis-sioned by the European Commission in the FUSION project.
Trope, Y., & Liberman, N. (2003). Temporal construal. Psychological review, 110(3), 403.
Trope, Y., & Liberman, N. (2010). Construal-level theory of psychological distance. Psychological review, 117(2), 440.
Van Herpen, E., van der Lans, I., Nijenhuis-de Vries, M., Holthuysen, N., Kremer, S., & Stijnen, D. (2016). Consumption life cycle contributions-Assessment of practical methodologies for in-home food waste measurement. Final report, Refresh Project, Wage-ningen, 1.
Visschers, V. H., 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, 45, 66-78.
Wansink, B. (1996). Can package size accelerate usage volume? Journal of marketing, 60(3), 1-14.
Wenlock, R. W., Buss, D. H., Derry, B. J., & Dixon, E. J. (1980). Household food wastage in Britain. British Journal of Nutrition, 43(1), 53-70.
