MSc Thesis written by Wies Maria Vos
11908386
Faculty: Economics and Business
Degree: Master Business Administration – Consumer Marketing Track EBEC approval: EC 20210215110233
Supervisor: Carla Freitas Silveira Netto
University of Amsterdam June 21, 2021
Channel Choice and Eating Habit Changes
–
Role of Stress (COVID-19) and Shopping Frequency
–Statement of originality
This document is written by Student Wies Maria Vos who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Abstract
The COVID-19 pandemic caused one of the largest changes for brick-and-mortar stores, particularly a switch from physical grocery shopping to online grocery shopping. Therefore, channel choice (online vs. offline) is an emergingly relevant topic and has been proven to affect the food choices consumers make. Besides, the COVID-19 lockdown is associated with direct consumption behavior changes and caused several side-effects, such as stress. Also shopping frequency affects the food choices of consumers. This study aimed to investigate the impact of stress (COVID-19) and shopping frequency on eating habit changes among adults in an offline compared to an online shopping environment. The study comprised a questionnaire that inquired demographic information, channel choice, shopping frequency, stress, and dietary habits information (adherence to the Mediterranean diet, number of meals per day, intake of certain foods). The survey was conducted from the 6th to the 18th of May 2021. A total of 361 respondents have been included in the study, aged 18 and above (71.5% females). The study demonstrates that in an offline shopping environment, consumers under stress are more likely to change their eating habits to being healthier, whereas in an online shopping environment, consumers with low stress are more likely to change their eating habits to being healthier.
Concerning the shopping frequency, in an online shopping environment, consumers with a high shopping frequency are more likely to have healthy eating habits, whereas in an offline shopping environment, consumers with a low shopping frequency are more likely to have healthy eating habits. These findings highlight several unexplored differences between the online and offline shopping environment, with important implications for retailers, public policy makers and consumers.
Keywords: Channel choice (online vs. offline), grocery shopping, eating habits, stress, shopping frequency, COVID-19 pandemic, food consumption
Table of Contents
Introduction ... 9
Theoretical Framework ... 14
Online versus offline: advantages & disadvantages ... 14
Online versus offline: shopping environments ... 16
Instinctive choice of vices ... 19
Product presentation: symbolic versus physical ... 20
Mediterranean Diet Adherence ... 22
Consumer Behavior in times of crisis ... 22
Stress (COVID-19) and eating habits ... 23
Shopping frequency ... 24
Contribution ... 26
Conceptual Model ... 27
Figure 1: Conceptual Model ... 27
Research Methodology ... 27
Research Design and Procedure ... 27
Data collection process and sample characteristics ... 28
Table 1: Respondents general characteristics ... 29
Measurement of Variables ... 29
Independent variable: channel choice ... 30
Dependent variable: eating habits change COVID-19 ... 30
Dependent variable: meal change COVID-19 ... 30
Dependent + Moderator variable: healthy food consumption ... 31
Moderator variable: stress (COVID-19) ... 32
Moderator variable: shopping frequency ... 33
Missing Value ... 33
Recoding ... 33
Reliability ... 33
Table 2: Cronbach’s Alpha ... 34
Computing Scale Means ... 34
Table 3: Descriptive Statistics ... 34
Analytical Strategy ... 34
Results ... 35
Table 4: Means, Standard Deviations, Correlations ... 35
Adherence to the MD ... 36
Figure 2: Online Channel: Compliance with items from MEDAS according to high, medium and low adherence to the Mediterranean Diet (MD) ... 38
Figure 3: Offline Channel: Compliance with items from MEDAS according to high, medium and low adherence to the Mediterranean diet (MD) ... 39
Table 5: Positive answers to MEDAS-questionnaire and adherence to the MD ... 40
Eating habits changes during the COVID‐19 pandemic ... 41
Table 6: Eating habits changes offline vs. online shoppers ... 42 Figure 4: Offline Channel: Variation in food consumption during the COVID-19
Figure 5: Online Channel: Variation in food consumption during the COVID-19
pandemic ... 44
Hypothesis 1 ... 45
T-test eating habits change COVID-19 ... 45
Table 7: Group Statistics – Eating Habits Change COVID-19 ... 45
Table 8: Independent Samples Test – Eating Habits Change COVID-19 ... 45
T-test meal change COVID-19 ... 46
Table 9: Group Statistics – Meal Change COVID-19 ... 46
Table 10: Independent Samples Test – Meal Change COVID-19 ... 46
T-test healthy food consumption ... 47
Table 11: Group Statistics – Healthy Food Consumption ... 47
Table 12: Independent Samples Test – Healthy Food Consumption ... 47
Table 13: Regression Results – Eating Habits Change COVID-19 ... 49
Table 14: Regression Results – Meal Change COVID-19 ... 50
Table 15: Regression Results – Healthy Food Consumption ... 51
Table 16: Regression Results – Channel Choice ... 52
Hypothesis 2 ... 53
Figure 6: PROCESS Model 1 ... 54
Table 17: Moderator Analysis for hypothesis 2 Eating Habits Change COVID-19 54 High-adherence group (N=51) ... 55
Table 18: Moderator Analysis for hypothesis 2 – High-adherence ... 55
Hypothesis 3 ... 56
Figure 7: PROCESS Model 1 ... 56
Table 19: Moderator Analysis for hypothesis 3 – Eating Habits Change COVID-19 ... 57
Discussion ... 58
Implications ... 62
Conclusion, limitations and recommendations for future research ... 65
References ... 67
Appendix ... 79
Table 20: Whole sample – Respondents general characteristics ... 79
Figure 8: Whole sample – Compliance with items from MEDAS according to high, medium and low adherence to the Mediterranean diet (MD) ... 80
Table 21: Whole sample – Positive answers to MEDAS-questionnaire and adherence to the MD ... 81
Table 22: Whole sample – Eating habits changes ... 82
Figure 9: Whole sample – Variation in food consumption during the COVID-19 pandemic ... 83
Dependent variable: Meal change COVID-19 ... 84
Table 23: Moderator Analysis for hypothesis 2 – Meal Change COVID-19 ... 84
Dependent variable – Healthy food consumption ... 84
Figure 10: PROCESS Model 1 ... 84
Table 24: Moderator Analysis for hypothesis 2 – Healthy Food Consumption ... 85
Dependent variable: Low-adherence group (N=88) ... 85
Table 25: Moderator Analysis for hypothesis 2 – Low-adherence ... 85
Dependent variable: Medium-adherence group (N=222) ... 86
Table 26: Moderator Analysis for hypothesis 2 – Medium-adherence ... 86
Extra: Moderated moderation analysis ... 86
Figure 11: PROCESS Model 3 ... 87
Table 27: Moderated Moderation Analysis – Eating Habits Change COVID-19 .... 87
Table 28: Moderated Moderation Analysis – Meal Change COVID-19 ... 88
Dependent variable: Meal change COVID-19 ... 89
Table 29: Moderator Analysis for hypothesis 3 – Meal Change COVID-19 ... 89
Dependent variable – Healthy food consumption ... 89
Figure 12: PROCESS Model 1 ... 89
Table 30: Moderator Analysis for hypothesis 3 – Healthy Food Consumption ... 90
Dependent variable: Low-adherence group (N=88) ... 90
Table 31: Moderator Analysis for hypothesis 3 – Low-adherence ... 90
Dependent variable: Medium-adherence group (N=222) ... 91
Table 32: Moderator Analysis for hypothesis 3 – Medium-adherence ... 91
Dependent variable: High-adherence group (N=51) ... 91
Table 33: Moderator Analysis for hypothesis 3 – High-adherence ... 91
Questionnaire ... 92
Introduction
Since 2020, the world has changed. The outbreak of a new virus, named COVID-19, created a different life for everyone. The disease evolved in the Wuhan region of China, and on the 11th of March 2020, the World Health Organization stated the outbreak of COVID-19 as a global pandemic (Liu et al., 2020). Five weeks after the knowledge that the disease is transmittable from person to person, the first positive test was confirmed in the Netherlands.
To slow down the transmission rate of the virus, numbers of countries began to enforce a strict hygiene regime and implemented strong lockdown measures. This imposed restrictions on daily living, such as; social distancing, remote working, home confinement, and temporary closing of schools, businesses and universities (Maliszwenka, Mattoo & Van Der Mensbrugghe, 2020). Eventually, to fight COVID-19, countries were required to enforce national and city-wide lockdown measures. During the first COVID-19 lockdown, started on March 15 in the Netherlands, people were requested to avoid busy places, work remotely if possible, stay at home when having any virus symptoms and travel outside peak hours (Dutch Government, 2020). During this period, the Netherlands imposed a less severe lockdown strategy compared to other countries. In this way, the Dutch government strived to cushion the social, economic and psychological costs of social isolation, while controlling and containing the spread of COVID-19. In this context, restaurants and pubs had to close their doors, while meal delivery services, the bakery, local food shops and supermarkets stayed open (Dutch Government, 2020). Although the infections with COVID-19 decreased, this was not for long.
After the summer of 2020, the number of infections rose to critical levels again (Dutch Government, 2020). This forced the Netherlands to impose a second lockdown, in which we are still present at this time.
However, as the majority of national governments enforced measures to stop the spread of COVID-19 by imposing lockdowns, less attention has been directed to the potential side-
effects of such a lockdown. Previous studies have indicated that self-quarantine, and especially mandatory, can result in psychological distress, which is manifested most frequently by irritability, low mood, anger, emotional exhaustion, insomnia, and depressive symptoms (Sidor
& Rzymski, 2020; Brooks et al., 2020). Furthermore, El Bilali et al. (2019) came to warn of the disruptive impact these measures can have on agri-food systems and food consumption.
COVID-19 has showed the vulnerability of global food systems to crisis and shocks. Besides, prior research has indicated that the COVID-19 lockdown is associated with direct consumption behavior changes including both decreased and increased alcohol consumption (Koopmann et al., 2020; Kim et al., 2020), increased consumption of unhealthy food, snacking between meals, uncontrolled eating and overall higher number of meals (Ammar et al., 2020; Carroll et al., 2020; Sidor & Rzymski, 2020), increased smoking frequency (45%) among smokers (Sidor &
Rzymski, 2020) self-reported weight gain in adults (e.g., due to stress, snaking in response to food cues or no to little sleep) (Zachary et al., 2020), as well as in younger adults (Sidor &
Rzymski, 2020).
Since the starting phase of COVID-19, the retail grocery industry has witnessed several changes in the collective patterns of consumer behavior, and even a radical change in demand for home delivery services, online shopping and certain products (Nie et al., 2010). To a certain extent, these changes are the result of the restrictions and recommendations from global authorities and national governments to protect persons from the fast transmission rate of the COVID-19 virus. During a crisis, retail consumers change their preferences regarding the places where to shop, how often, and what to buy (Nie et al., 2010). Roggeveen and Sethuraman (2020) discovered that due to a crisis like COVID-19 consumers are also likely to develop or learn new shopping routines. In order to have a successful business organization, understanding consumer buying patterns that are emerging in a crisis play an important role (Sharma &
Sonwalkar, 2013).
COVID-19 has affected where and how consumers purchase their food (Cranfield, 2020). The pandemic caused one of the largest changes for brick-and-mortar stores, particularly a switch from physical grocery shopping to online grocery shopping (Eriksson, 2020). Yet, consumer behavior during a pandemic is not always consistent with the actual level of risks.
Thus, the starting stage of this crisis may have caused extreme behavior changes, these changes may eventually turn into more normalized behavior or a possible new-normal behavior (Pennings et al., 2002).
Imagine a consumer who chooses to order groceries from an online grocery store rather than visiting a traditional grocery store to purchase the items in-person. Does this consumer purchase a different assortment of items online compared to offline? More specifically, are there more or less vices – tempting food items that generally provide short-term benefits (e.g., tasty, unhealthy foods), but provide fewer long-term benefits (e.g., healthiness; Khan & Dhar, 2007; Wertenbroch, 1998) – present in a consumer’s shopping basket when the consumer orders their groceries online compared to purchasing groceries offline in-person? Moreover, does a consumer who has a high amount of stress compared to a consumer with a low amount of stress purchase more vices when shopping either online or offline? And does the shopping frequency of a consumer have any influence on this? Despite the practical, societal and theoretical relevance of these questions, there is little research that has addressed them. The relevance of the previous questions is especially important when we consider the increasing and alarming obesity rates worldwide (e.g., more than 50% of adults in the US are obese or overweight;
Ogden et al., 2014; World Health Organization, 2014). Previous studies have demonstrated that assortment size (Sela, Berger, and Liu 2009), shelf arrangements (Van Kleef, Otten, and Van Trijp, 2012) and choice set compositions (e.g., healthy or unhealthy options which are presented in a separate vs. unified choice set; Fishbach & Zhang, 2008) cause substantial impacts on the decisions of consumers to buy vices. Thus, this prior research notes various product
presentations within a single retail channel, but ignores the query of whether particular retail channels influence the choices of consumers, which is, between-channel differences. The product presentations between online and offline shopping environments differ fundamentally, therefore, they are likely to have differential impacts on buying behavior. Furthermore, shopping groceries through an online channel is becoming increasingly common (Bauerová &
Klepek, 2018) This suggests that there is great urgency in the quest of a better understanding of the potential impact this channel might have.
Additionally, prior research has found that consumers engage in less spending the further in advance of delivery they place their online grocery order (Milkman, Rogers &
Bazerman, 2010). Moreover, these researchers have discovered that the shopping basket of consumers contains less unhealthy products and more healthy products, the further in advance the order is placed.
We argue that the shopping baskets of consumers who are ordering their groceries online consists of relatively fewer vices compared to consumer who are shopping their groceries offline in-person, by cause of the inherent difference in the presentation of the products: an offline store presents the products physically, whereas an online store presents them symbolically. The sensory distance is increased when presenting the products symbolically, which in turn decreases the vividness of the product and renders immediate gratification as less important (Hoch & Loewenstein, 1991; Loewenstein, 1996; Mischel & Ebbesen, 1970; Shiv &
Fedorikhin, 1999, 2002). Therefore, as consumers experience less gratification when a product is presented symbolically and less vivid, they might buy relatively fewer vices in an online grocery shopping environment comparted to an offline in-person grocery store. Next to this, we argue that consumers who have higher stress will have more vices in their shopping basket compared to consumers who have lower stress, both in an online and offline grocery shopping environment. The reason for this is that stress causes behavior changes, such as increased
consumption of unhealthy food, overall higher number of meals and uncontrolled eating (Ammar et al., 2020; Carroll et al., 2020; Sidor & Rzymski, 2020). Last, we argue that the shopping baskets of consumers who purchase their groceries either online or offline that have a lower shopping frequency will contain relatively fewer vices compared to consumers who have a high shopping frequency, by cause of the reason that a shopping basket of a consumer is said to consist of more healthy and less unhealthy items, the further in advance their grocery order is placed (Milkman, Rogers & Bazerman, 2010).
Not much research has yet been conducted on the socio-demographic differences and individual level changes in food and drink purchases during the lockdown period. Various questions, such as; ‘Does the stress of the COVID-19 crisis impact food choices?’, and, ‘Does online grocery shopping alter the food choices customers make?’ have been researched, yet scarcely. Furthermore, the moderating role of online or in-person grocery shopping on the effect of COVID-19-stress on consumption behavior still remains unanswered. Would this effect of change in consumption behavior due to the stress of COVID-19 be stronger for consumers who are ordering groceries online? Or would this effect be stronger for consumers who are shopping groceries physically? And might shopping frequency influence this? This leads to the research question of this study: “To what extent does channel choice (online vs. offline) influence the eating habit changes among adults and is this effect moderated by stress (COVID-19) and/or shopping frequency?”
To add to existing literature and in order to better understand the impact of channel choice on changes in dietary behaviors as well as individual differences, the aim and purpose of this research is to analyze self-reported consumption behavior and food and drink purchases among a representative sample of adult consumers who are shopping either online or offline, and to assess whether stress (COVID-19), and the frequency of shopping, moderate this effect.
Presenting this unresearched differences between online and offline grocery shopping environments portrays an important theoretical contribution. However, the notion that online compared to offline grocery shopping channels could limit the number of vices in consumers’
shopping baskets also has practical and societal significance. Many consumers are struggling to limit their intake of unhealthy vices, as indicated by the persistence of the obesity epidemic.
The marketing actions by food manufacturers and grocery stores often trigger consumers to select even more unhealthy and indulgent foods (Dubé et al., 2010; Kessler, 2009; Pollan, 2006;
Popkin, 2002). In proving the prediction that shopping groceries online compared to offline decreases the fixation on immediate gratification, and therefore leads consumers to purchase fewer vices, as well as proving that stress and shopping frequency can influence the forementioned relationship, this research presents considerable relevance for consumers, retailers and policy makers.
In the next section, the theoretical framework for the study is presented and hypothesis are developed. Next, we present the method, results, then conclude with a discussion and conclusion of the findings and an outline of the theoretical contributions, practical implications and suggestions for future research.
Theoretical Framework
Online versus offline: advantages & disadvantages
According to Buttle, Francis and Coates (1984), Alyott and Mitchell (1998) and Roberts et al. (2003), in-store grocery shopping is often considered to be tiring and time consuming, frustrating, stressful and un-enjoyable, especially when the store is crowded. Geuens et al.
(2003) mention that although online grocery shopping offers an alternative to the stresses and boredom of in-store grocery shopping, the latter is still preferred by many consumers, because of its perceived superiority in terms of social, experiential and functional aspects. The social
aspects are valued, such as, meeting friends and shopping with family members (Roberts et al., 2003; Prasad and Arysari, 2011), the fun of watching other shoppers and being part of a crowd (Mehta et al., 2014), as well as recreational and experiential elements (e.g., impulse purchases and browsing for new products). The functional aspects include the time saving which results from combining in-store grocery shopping with other chores and the ability to find bargains.
Avoiding the negative aspects of in-store grocery shopping is perceived as major advantage (and an important determinant) of online grocery shopping (Roberts et al., 2003). According to Verhoef and Langerak (2001), an important advantage of online grocery shopping is the reduction of the physical effort while grocery shopping in-store. Several consumers considered the online grocery shopping as a means of reducing the time pressure which is associated with in-store grocery shopping. The time saving and reduction of physical effort which are associated with online grocery shopping are related to convenience. This convenience emerges as a decisive factor for online grocery shopping and is perceived as a major advantage (Verhoef
& Langerak, 2001; Geuens et al., 2003; Roberts et al., 2003). Further advantages of online grocery shopping are the opportunity to find good deals and the greater variety offered (Roberts et al., 2003), as well as avoiding invasive sale people and impulse buying (Ramus & Nielsen, 2005).
However, there are also several disadvantages of online grocery shopping. Verhoef and Langerak (2001) mention that inconveniences (e.g., waiting for deliveries) can offset many perceived advantages of online grocery shopping compared to in-store grocery shopping. The most commonly mentioned disadvantages of online grocery shopping are the concerns about the security of transactions and privacy, delivery charges, not being able to judge the quality of products in-person, the inability to use coupons and take advantage of promotions, perceived complexity and the lack of social contact (Morganosky and Cude, 2000; Verhoef and Langerak, 2001; Roberts et al., 2003; Ramus and Nielsen, 2005). In addition, the loss of recreational and
experiential aspects of online grocery shopping, the lack of personal service, the inconvenience when not all products are received (missing products) or can be bought online, and the stress associated with waiting for a delivery are also mentioned by consumers to be disadvantages on online grocery shopping (Ramus and Nielsen, 2005).
Online versus offline: shopping environments
The environments of online and offline grocery shopping are different in multiple ways.
Regarding the online shopping environment, the costs for information search are lower, as screening information and forming a consideration set can be done easily, the product assortments are broader and more information is available. Additionally, consumers often have a personal shopping list where they can save their previous purchase items, therefore, consumers can easily repurchase these products. Nonetheless, the perceived risk in online shopping tends to be higher compared to offline shopping (Alba et al., 1997; Bart et al., 2005;
Danaher, Wilson & Davis, 2003; Huang, Lurie & Mitra, 2009). A different research by Huyghe et al. (2017) also mentions that perceived risk tends to be higher in an online environment. Next to this, impulse purchases are made less frequently in the online compared to the offline environment. Contrastingly, in an offline shopping environment, consumers are able to feel and touch the food products and even taste several items. This is especially important for acquiring information about the experience attributes of food products, which can only be evaluated after consuming the product (Nelson, 1974). Also, offline shopping might be more fun due to the store atmospherics present in an offline store. However, shopping in a crowded store can also lead to annoyance, irritation, or time-consuming demands. Moreover, offline stores do not demand order lead time. Another study by Pitts et al. (2018), mentions that consumers prefer to purchase perishable, fresh products when shopping in an offline brick-and-mortar store, while they are hesitant to purchase perishable products via online grocery shopping. Pitts et al.
(2018) also argue that online grocery shopping offers a potential promise to promote healthier
product choices, which includes fewer impulse purchases, a better access to healthy foods in the online environment, even when these healthy foods are less available in the physical stores.
Yet, consumers may be less likely to purchase perishable food products (e.g., fresh vegetables and fruits) via online grocery shopping, due to concerns about food safety, bruising and freshness. As a result, less healthy purchasing habits could be cultivated via the online shopping experience, as consumers both have healthy and unhealthy processed products easily available in the online environment (Pitts et al. 2018). Next to this, Campo and Breugelmans (2015) discovered that consumers purchased heavier and more bulk items in the online vs. offline stores.
Milkman, Rogers and Bazerman (2010) and Huyghe et al. (2017) stated that online shopping resulted in an avoidance of unhealthy ‘vices’ in favor of more ‘virtuous’ items.
Huyghe et al. (2017) also argue that the shopping baskets of consumers contain fewer vices when they shop online, compared to offline, as the inherent difference in the presentation of the product; an offline store presents the products physically, whereas an online store presents products symbolically, by using pictures. The symbolic presentation of the product induces a sensory distance, which decreases the vividness of the product and renders immediate gratification as less important (Huyghe et al., 2017). As a result, consumers might purchase fewer vices while online shopping, as they experience less gratification from the less vivid, symbolic products compared to an offline grocery store. Another study by Gorin et al. (2007) demonstrated that encouraging participants who were currently trying to lose weight, to buy their groceries online, decreased the total number of foods (and high-fat foods) in their home.
Moreover, the participants mentioned many benefits while ordering their groceries online, such as; making fewer impulse purchases and making healthier choices.
Even though several studies present these differences, a limited number of empirical studies analyzed the impacts of online and offline shopping environments on consumers’
purchase behavior, with a few exceptions that studied price sensitivity and brand loyalty.
Danahar, Wilson and Davis (2003) have discovered that brands with a high market share contain a loyalty advantage when consumers purchase online compared to offline. In addition, households exhibit a lower price sensitivity when they purchase online instead of offline (Chu, Chintagunta, & Cebollada, 2008). Other studies do not present a direct comparison between online and offline shopping environment, but do consider the influence of several elements, such as payment method (Bagchi & Block, 2011; Thomas, Desai, & Seenivasan, 2011) and order lead time (Milkman, Rogers, & Bazerman, 2010), that seem to differ between online and offline stores.
Additionally, online shopping environments require their consumers to pay cashless.
Paying cashless is becoming more and more traditional in the offline shopping environment, however, recent research has investigated if paying with cash or cashless has any influence on food consumption. Park, Lee and Thomas (2021) have investigated why shoppers spend more money on unhealthy food products when they pay cashless. They proposed that negative arousal which is elicited by monetary payments plays a big role. Whereas cash payments increase negative arousal, which in turn increases attention to decision risks, cashless payments reduce negative arousal, and consequently reduce attention to decision risks. Cashless payments are able to increase risky consumption behavior, when the attention to decision risks is reduced.
Two studies were conducted that demonstrate that cashless payments reduce negative arousal and increase purchase intentions of vice food (e.g., hedonic foods), and that the effect of cashless payments is stronger for consumers who are more sensitive to health risks.
Therefore, the product presentation differences (e.g., physical in offline stores, symbolic in online stores) in an online and offline shopping environment are relevant, particularly in the context of grocery shopping. Regarding vices – defined as unhealthy, tasty and attractive products – the way products are presented could determine the purchase behaviors and
preferences of consumers, such that a symbolic (versus physical) presentation might reduce the number of vices in a consumers’ shopping basket.
Instinctive choice of vices
When consumers are given a binary choice context, they have difficulties selecting a virtue option (e.g., vegetables, fruit salad) that is in line with their desire to live a healthy and long life due to the presence of a vice option (e.g., snack, sweets) that is immediately tempting.
This choice dilemma creates a self-control issue; the mental stress consumers experience presents the inherent conflict between their reflective and affective behavioral systems (Loewenstein 1996; Metcalfe & Mischel 1999). Shiv and Fedorikhin (1999) mention that regulating vice choices is a challenge for a lot of consumers due to their naturally strong and automatic visceral responses to vice items. According to Vohs and Heatherton (2000), the cognitive resources which are needed to override these visceral responses might be unavailable.
The extent to which consumers utilize their willpower to alleviate these affective responses, as well as the intensity of affective responses determine the likelihood of selecting vices over virtues. The intensity of these affective responses to a specific food item is dependent on the vividness of the presentation of the product, which applies to the amount of detail that is presented for the sensory inspection of the product to form a mental image (Loewenstein, 1996).
The more vivid the product’s presentation, the more detailed is the mental image of the product.
Next to this, the sensory gratification that emerges from the consumption of the product becomes more desirable. Therefore, imagining the gratification that will emerge from consuming the product will be easier for consumers when the product presentation is more vivid. The intensity of the affective responses to the product will be greater, the more consumers anticipate the awaiting gratification. This will eventually facilitate behavioral actions which are based on visceral responses instead of cognition and, as a consequence, will increase the
tendency of consumers to purchase vices. Thus, studies analyzing the choice of consumers in a binary choice context (vices versus virtues) suggest that product presentations that are less vivid diminish the desire of consumers for immediate gratification. This leads them to being less prone to select vices. The following section will describe how online and offline shopping environments can be distinguished by the differences in the vividness of the presentation of their products.
Product presentation: symbolic versus physical
First, several studies about the effects of physically absent versus physically present products – instead of symbolically present versus physically present products – will be discussed, because the former justification of the product presentations’ vividness is most prevailing in present research. Mischel and Ebbesen (1970) did an experiment in which they asked children to voluntarily delay their reward, such as cookies, pretzels, and marshmallows and discovered how these children expressed considerably less willingness to wait for the preferred reward when there was a less preferred option physically present. An explanation for this is that reward objects trigger a more intense desire when they are present physically (versus absent), as this product presentation makes the reward more vivid the children, which helps them to anticipate and imagine the gratification that is associated with the consumption of the reward (Loewenstein 1996).
Bossert-Zaudig et al. (1991) and Hill, Magson, and Blundell (1984) found that the mere sight of food already intensifies visceral responses, such as increasing hunger and salivation, respectively. Likewise, in the case of offline stores, stockpiled products are visually salient.
Therefore, they elicit more intense visceral responses, which results in an increase of consumption frequency (Chandon & Wansink, 2002; Wansink & Deshpandé, 1994). Thus, delaying gratification is more challenging for customers who encounter a product physically.
Deng and Srinivasas (2013) mention that when the product presentation mode varies between physical and symbolic (instead of absent vs. present), similar effect on decision making and food resistance emerge. For instance, by presenting visually attractive food in an opaque instead of a transparent package, this will lead customers to eat less of the food (Deng and Srinivasan, 2013). Symbolic product presentations (e.g., pictures) have a higher sensory distance, leading to a decrease in their vividness, compared to physically presented products (Kardes, Cronley & Kim, 2006). Therefore, it is more difficult for customers to experience the sensorial aspect of a product in an online shopping environment, due to the symbolic presentation of products in this channel. Moreover, Degeratu, Rangaswamy and Wu (2000) stated that sensory search attributes, such as visual cues (e.g., packages), have less impact in an online compared to an offline shopping environment. Milkman, Rogers and Bazerman (2010) mention that consumers buy fewer vices the further ahead they plan their purchase. However, as online retailers are constantly trying to decrease order leading times, this finding might be growing less relevant.
Regarding this evidence, we predict that the proportion of vices in the shopping baskets of consumers differs depending on the shopping channel they use (online vs. offline).
Specifically, as symbolic presentations are less vivid and lead to a decrease in sensory imagery, they should diminish the desire of consumers for immediate gratification. This will lead them to purchase relatively fewer vices online compared to offline (Huyge et al, 2017).
This leads to hypothesis 1: Channel choice has a direct relationship with healthy eating habit changes (COVID-19), so that this relationship is stronger for consumers who buy their groceries online.
Mediterranean Diet Adherence
Furthermore, it is extremely important to follow a healthy diet, because our bodies are influenced by the type of food we eat (Di Renzo et al., 2019). However, estimating the overall quality of a diet and determining its link to specific health outcomes is a challenge. Therefore, a compound scale was created including all food items which are considered as characteristic of the Mediterranean diet in order to estimate an individuals’ adherence to this healthy diet. To control for compliance with this diet, a 14-point Mediterranean Diet Adherence Screener (MEDAS) was developed (Schröder et al., 2011). A number of researches have proven that there exists an inverse association between the adherence to the Mediterranean Diet (MD) and the overall mortality rate of cancer related diseases. The healthy MD (De Lorenzo et al., 2017) is a combination of several quality food items, which are based on nutritious contents and in which contaminating substances are absent. Regarding current knowledge, the MD is the key element against immune-mediated inflammatory responses, which are occurring in cancer. In particular, scientific evidence has indicated that the increasing adherence to this specific Mediterranean diet is linked to favorable physical and mental health outcomes (Martinez- Gonzalez et al., 2009; Schroder, 2007). The aim of this research is to analyze and explore changes in the consumption behavior of participants and their adherence to the MD during the COVID-19 pandemic. Thus, on the basis of the 14-point Mediterranean Diet Adherence Screener, participants were divided between three groups; (1) low adherence (score 0-5), (2) medium adherence (score 6-9), and (3) high adherence (score≥10). In this way, three groups were created who have an increasing level of healthiness.
Consumer Behavior in times of crisis
Sharma and Sonwalkar (2013) mention that consumer buying behavior is a collection of the decision process and subsequent behavior, either planned or unplanned, and determined by external and internal factors. The key components in the decision-process are risk and
uncertainty and different factors drive consumer behavior in a risk-related crisis (Pennings et al., 2002). These drivers of consumer behavior in times of crisis are commonly observed from the perspective of consumers’ risk attitude and risk perceptions. Amalia and Ionut (2009) segmented consumers in the categories; “The panicked consumers”, “The rational consumers”,
“The concerned consumers”, and “The prudent consumers”. They mention that the panicked consumer tends to overreact and drastically cut spending in a time of crisis, whereas a prudent consumer postpones major purchases and plans spending carefully. Similarly, concerned consumers plan their spending carefully and are ready to try innovative and new products.
Contrastingly, rational consumers’ behavior remains unchanged (Amalia and Ionut, 2009).
Besides, Dutu (2020) argues that panic may trigger considerable changes in consumer behavior.
Dedeoglu and Ventura (2017) have studied the consumers responses to the swine flu threat, and discovered that fear was a key predictor in behavior change. However, consumer behavior in times of a crisis is not always consistent with the actual level of risks. Therefore, the beginning stage of a crisis may cause extreme behavior changes, which eventually turn into more normalized behavior or a possible new-normal behavior (Pennings et al., 2002).
Stress (COVID-19) and eating habits
People under stress change their diets in both quantity and quality (Ogden and Mitandabari, 1997; Stone and Brownell, 1994). Ingledew, Hardy, Cooper and Jemal (1996) and Wardle, Steptoe, Oliver and Lipsey (2000) found that even though most people eat less when they are under stress, a large minority of eaters who are externally motivated react by eating more than usual. Therefore, stress alters the type of food people are attracted to. These negative emotions rising from stress could cause ‘emotional eating’ (Evers et al., 2018), and overeating, especially of so-called ‘comfort foods’, which are high in sugar, fats and salt (Moynihan et al., 2015; Yilmaz & Gökmen, 2020). Scarmozzino and Visioli (2020) discovered in their study on
food consumption in Italy that 19.5% of respondents gained weight and 46.1% indicated to be eating more during quarantine. Especially, they indicated an increase of consumption of
‘comfort food’, particularly chocolate, salty snacks, ice cream and desserts. 42.7% of respondents linked this significant increase to a higher level of anxiety. Moreover, Ammar et al. (2020) also highlighted that meal patterns and food consumption were unhealthier during the lockdown period. These changes in the lifestyle behavior of people could create a vicious cycle and can even further increase the risk of those with diseases.
Contrastingly, Rodíguez-Pérez et al. (2020) mentioned that the lockdown period led to the adoption of a healthier consumption pattern among an adult population. COVID-19 caused a change in the rise of home-prepared meals. Since cafes, lunchrooms and restaurants are closed, people are more often baking and cooking at home. (The Food Industry Association, 2020).
This leads to hypothesis 2: The positive relationship between channel choice and healthy eating habit changes (COVID-19) is moderated by stress (COVID-19), so that this relationship is stronger for consumers who have lower stress.
Shopping frequency
Previous studies about intrapersonal conflict, which is also known as the multiple selves phenomenon, has documented a tension between the behavior people find themselves hedonically wanting to exhibit and regularly deciding to exhibit due to the short-term rewards (e.g., eating cake, watching television, spending more) and the behaviors they feel they should exhibit given their long-term interests (e.g., starting a diet, going to the gym, saving more) (Schelling, 1984). This tension is described as stemming from two different selves; the want self and the should self, which have clashing preferences (Bazerman et al., 1998). According to Shefrin and Thaler (1998), people often live in a state of internal conflict between a ‘planner
self’, which represents the ‘should self’ described by Bazerman et al. (1998) and a ‘doer self’, which represents the ‘want self’ of Bazerman et al. (1998). This ‘multiple selves framework’
predicts that consumers will be more likely to make ‘want’ choices that have short-term benefits in situations where the outcomes are more immediate, such as indulging in more unhealthy
‘want’ foods and fewer healthy ‘should’ foods. Loewenstein (1996) mentions that the closer the reward, the more likely a consumers’ visceral desires will overwhelm their cooler cognitive systems. Individuals are assumed to have a steep short-term discount rate and a flat long-term discount rate; this leads them to overvalue present- compared to future utility. Therefore, individuals will favor ‘want’ options (e.g., spending and eating unhealthy foods) over ‘should’
options (e.g., eating healthy foods) at a higher rate the sooner their decisions will take place.
Moreover, these theories predict that decisions that are made for the nearer future will heighten overall spending (‘want’ behavior) and increase spending on products that are preferred by the
‘want self’. Contrastingly, spending decisions made for the distant future will lead to less overall spending (‘should’ behavior) and increased spending on products that are preferred by the ‘should self’.
Milkman, Rogers and Bazerman (2009) found that the further in advance of delivery consumers place an online grocery order, the less spending they engage in. Furthermore, they found that orders placed between 2 and 5 days in advance of delivery, the percent of an order consisting of ‘want’ groceries decreases and the percent consisting of ‘should’ groceries increases for each additional day in advance of delivery the order is placed.
This leads to hypothesis 3: The positive relationship between channel choice and healthy eating habit changes (COVID-19) is moderated by shopping frequency, so that this relationship is stronger for consumers who have a lower shopping frequency.
Contribution
Regarding these predictions, this study aims to resolve several gaps in existing literature into stress, shopping frequency, channel choice and food consumption. Previous studies have mainly focused on binary choice tasks, such that participants had to select either a vice or a virtue (Dhar & Wertenbroch, 2000; Shiv and Fedorikhin, 1999, 2002). However, consumers seldomly trade off these vices and virtues in real-world grocery shopping environments.
Therefore, we looked at the overall health of participants in order to find any differences in vice purchases between online and offline grocery shoppers. The positive relationship between shopping through an online channel and healthy food consumption is already established in literature, now we will add to this relationship the effect of stress (including COVID-19 stress) and shopping frequency. The findings of this research will contribute to theory and practice, as the research will discover the impact of channel choice (online vs. offline) on healthy consumption behavior among adults, and if stress (including COVID-19 stress) and/or shopping frequency have a moderating effect. This is particularly relevant and interesting for the scientific community and practitioners as they will gain knowledge about how the choice of channel in the grocery retail sector during a crisis impacts consumption behavior. The study will provide a deeper understanding of the different consumption patterns of consumers (online vs. offline) during times of a deep external impact. This knowledge can be used to better understand the possibly changed consumption behavior during a crisis to be able to respond to the different needs of these consumers in the current and future environment. Therefore, to be a successful business organization, it is important to understand the changing consumer buying patterns that are emerging in a crisis, and possibly develop into a new standard routine.
Figure 1: Conceptual Model Conceptual Model
Research Methodology
Research Design and Procedure
This research was a comparative analysis between online and offline grocery shopping consumers. In order to answer the research question: “To what extent does channel choice (online vs. offline) influence eating habit changes among adults and is this effect moderated by stress (COVID-19) and/or shopping frequency?”, a quantitative study, by means of a survey, was performed. The data-collection instrument was a survey, which was administered online.
The questionnaire of the survey, made available on Qualtrics, is in the Appendix. First, a pre- test was performed, to ensure that the questions were clear and unambiguous. The questionnaire asked the participants to answer questions that were divided into 5 different sections; (1) channel choice, (2) stress (COVID-19), (3) dietary habits information, (4) shopping frequency and (5) demographics (age, gender, educational level).
Channel Choice
Online vs. Offline Eating Habits
Change COVID-19 Stress
(COVID-19)
Shopping Frequency H1
H2
H3
Low-adherence
Healthy Food Consumption Online vs. O
Meal Change COVID-19 Online vs. O
Medium-adherence High-adherence
Data collection process and sample characteristics
The population of interest for this study were adults. The research was conducted by using a non-probability convenience sample, as the sampling frame was unknown and the population was large. The inclusion criteria of this study were; being aged 18 years or older.
Respondents were reached through Facebook, e-mail and personal networks. Any similarities in the sample were be assessed by using the demographic information (gender, age, educational level) that was gathered in the surveys. The survey included detailed information about the purpose and objectives of this research (e.g., information about anonymity, aim, estimated time required for completion) to the participants before the commencement of the research. During the data collection period, the researcher attempted to collect as many respondents as possible.
In order for the data to be analyzable, the minimum number of respondents was 300. Due to the distribution of the survey via Facebook, the response rate was difficult to predict.
On the 18th of May 2021, the questionnaire was concluded, and the collected data were analyzed. A total of 408 participants started the questionnaire, and, after validation of the data, 361 respondents have been included in the study, aged 18 and above. The 47 respondents that have not been considered in the analysis did not complete the survey (more than 90% of the answers were missing). The female respondents represent 71.0% of the offline shoppers and 76.7% of the online shoppers. The average ages for offline and online shoppers are 30.64 years old and 34.90 years old, respectively. 55.7% of the respondents have a Dutch nationality. 28.3%
of the remaining respondents is European (83.9% when including Dutch respondents) and 16.1% is non-European. General characteristics of the respondents are reported in Table 1.
Table 20 (see Appendix) shows the general characteristics of the sample overall. In terms of employment status, 111 (30.7%) respondents have a paid work, 232 (64.3%) are students, 9 (2.5%) are unemployed, and 9 (2.5%) are retired. Regarding the education, 299 (82.9%) are highly-educated, 54 (14.9%) are low-educated and 8 (2.2%) did not want to disclose this
information. In terms of their current living situation, 84 (23.3%) live with their family, 90 (24.9%) live with roommates, 60 (16.6%) live on their own, 79 (21.9%) live together with partner without kids, 46 (12.7%) live together with partner and kids, and 2 (0.6%) live without partner and kids. Most of the respondents purchases food at the grocery store (91.7%), and 8.3%
uses online delivery.
Table 1: Respondents general characteristics
Offline shoppers (n = 331) Online shoppers (n=30)
Age (years) 23.46* [20.68-34.66]
30.64 ±13.78
31.17* [22.27-45.75]
34.90 ± 14.12 Age groups (years)
18-24 182 (55.0) 11 (36.7)
25-34 66 (19.9) 6 (20.0)
35-44 16 (4.8) 5 (16.7)
45-54 27 (8.2) 4 (13.3)
55-64 35 (10.6) 3 (10.0)
65+ 5 (1.5) 1 (3.3)
Gender (F) 235 (71.0) 23 (76.7)
Values are expressed as median and IQR in square brackets (M [IQR]) as well as mean and standard deviation (M ± SD) for continuous variables or as numbers and percentages (n (%)) for categorical variables
*The Shapiro–Wilk test was performed to evaluate variables distribution. Variables are considered non-normally distributed for p < 0.05
Measurement of Variables
For the examination of the hypotheses the variables channel choice, eating habits change COVID-19, meal change COVID-19, healthy food consumption, stress (COVID-19) and
Independent variable: channel choice
Channel choice is a categorical variable consisting of two different groups; shopping groceries via an online channel or via an offline channel. Channel choice was measured by asking participants one question about how they purchase their groceries, either online or offline (in-person). All respondents were classified among the two groups: online channel or offline channel. For the analysis of the hypothesis two dummy variables were created. Online channel with the value (0) and offline channel with the value (1).
Dependent variable: eating habits change COVID-19
To measure ‘eating habits change COVID-19’, respondents filled out one question measured on a 5-point Likert scale by Di Renzo et al., (2020). According to the scale, participants were asked to indicate if and how their eating habits had changed during the COVID-19 pandemic (e.g., “I eat a lot unhealthier”; “I eat a lot healthier”). The scale is organized from a negative outcome to a positive outcome; meaning that respondents who score low have changed their habits to being a lot unhealthier, and respondents who score high have changed their habits to being a lot healthier.
Dependent variable: meal change COVID-19
To measure ‘meal change COVID-19’, respondents filled out one question measured on a 5-point Likert scale by Di Renzo et al., (2020). According to the scale, participants were asked to indicate if and how their numbers of meals a day had changed during the COVID-19 pandemic (e.g., “Yes, I added one or more of the main meals”; “Yes, I skip one or more of snacks between meals”). The scale is organized from a negative outcome to a positive outcome;
meaning that respondents who score low have a negative meal change (e.g., added snacks) and respondents who score high have a positive meal change (e.g., skipped snacks).
Dependent + Moderator variable: healthy food consumption
To measure healthy food consumption, respondents filled out the validated 14-items Mediterranean Diet Adherence Screener (MEDAS) by Schröder et al., (2011), which has a range from 0 to 14 points. For each question, respondents had to choose whether they adhered to the statement or not (e.g., “Do you use olive oil as the principal source of fat for cooking?”;
“Do you consume 3 or more servings (30 g) of nuts per week?”). The MEDAS questionnaire consists of 2 questions about food intake habits and 12 questions about food consumption frequency that are considered to be characteristic of the Mediterranean diet. Every question was given a score, either 0 or 1. One point was given when the participant uses olive oil as the principal source of fat for cooking, when the participant preferred white meat over red meat, or when the participant consumed: (1) 4 or more tablespoons (1 tablespoon = 13.5 g) of olive oil per day (including that used in frying, salads, meals eaten away from home, etc.); (2) 2 or more servings of vegetables per day; 3) 3 or more pieces of fruit (including fresh-squeezed juice) per day; 4) less than 1 serving of red meat, hamburger or sausages per day; 5) less than 1 serving of animal fat per day; 6) less than 1 cup (1 cup = 100 mL) of sugar-sweetened beverages per day; 7) 7 or more cups (100 ml) of wine per week; 8) 3 or more servings of pulses per week; 9) 3 or more servings of fish/seafood per week; 10) less than 2 commercial (not homemade) pastries per week; 11) 3 or more servings of nuts per week; or 12) 2 or more servings per week of a dish with a traditional sauce of onion, tomatoes, garlic, or leeks sautéed in olive oil. Zero points were given for the question when the condition was not met. The final score had a range from 0 to 14. This final score was used for the analysis.
For an extra analysis, the dependent variable healthy food consumption was categorized in three different groups. On the basis of this MEDAS score, participants were divided among three classes: (1) low adherence (score 0–5), (2) medium adherence (score 6–9) and (3) high (score≥10) adherence to the MD.
Next to this, an adapted structured questionnaire packet validated by Di Renzo et al.
(2020) with several questions about the consumption of certain foods, e.g., junk food consumption, sweet beverages, dressing sauces, salty snacks and baked products were asked.
Participants were asked to indicate which of these foods they ate either more or less since the COVID-19 pandemic, if they changed their eating habits (e.g., “I eat a little healthier”; “I eat a lot unhealthier”), and if they changed their number of meals per day (e.g., “Yes, I skip 1 or more of the main meals”; “Yes, I added 1 or more of the main meals”).
Moderator variable: stress (COVID-19)
To measure stress (COVID-19), participants filled out 34 questions measured on a 4- point Likert scale. The first 4 questions (e.g., “Over the last year, how often have you been bothered by the following problems? - Feeling nervous, anxious, or on edge”) were measured by using to the validated 4-point Likert scale by Kroenke, Spitzer, Williams and Lowe (2009) to measure stress and anxiety levels caused by COVID-19 and has a Cronbach’s alpha of 0.90.
The following 30 questions (e.g., “For each question, choose the option that best describes your situation over the last year. - You feel lonely or isolated”) were measured by using ‘The Perceived Stress Questionnaire’ by Levenstein et al. (1993) and has a Cronbach’s alpha of 0.90.
Last, participants were asked one question about how their stress had changed since the start of the COVID-19 pandemic.
Moderator variable: shopping frequency
To measure shopping frequency, participants filled out one question measured on a 4- point Likert scale by Wong et al., (2018). According to the scale, participants were asked to indicate how often they purchased groceries per week (e.g., “Once a week”; “2-3 times per week”).
Missing Value
All variables under investigation were checked for missing data. A frequency test was run for all variables. The amount of missing data was < 11.5% for all variables.
Recoding
The recoding of reverse items applied to eight items of the stress variable. Moreover, we needed to adapt the scale of gender from 1 (Male), 2 (Female) to 0 (Male), 1 (Female). Last, we recoded the items of ‘meal change COVID-19’ to order the scale from negative to positive.
Reliability
Reliability enables to examine the consistency of measurements. Reliability checks were run for stress. The Cronbach’s alpha, which represents the estimator of the internal consistency, has been tested to verify if all the items in one scale measure the same, or if some questions should not be used for analysis. As exhibited in Table 2, the variable has a Cronbach’s alpha > .7, which indicates high level of internal consistency. The other variables; channel choice, shopping frequency and healthy food consumption could not be checked for reliability, as these were nominal variables measured on a categorical scale.
Table 2: Cronbach’s Alpha
Variable Cronbach’s Alpha
Stress (COVID-19) 0.958
Computing Scale Means
As the last step new variables as a function of existing variables were created for the testing of the hypothesis. The mean of all items was calculated. Means and standard deviations of the variables are shown in Table 3.
Table 3: Descriptive Statistics
Variable N Mean Median Variance Range Min Max
Healthy Food
Consumption
361 7.22 7.00 5.76 0-14 0.00 14.00
Stress (COVID-19) 361 2.27 2.26 0.32 1-4 1.09 3.88
Shopping Frequency Per Week
361 1.9 2.00 0.60 1-4 1.00 4.00
Age 361 31.00 27.35 173.23 18->65 18.00 >65
Stress Change COVID-19 361 3.75 4.00 0.86 1-5 1.00 5.00
Eating Habits Change COVID-19
361 3.12 3.00 0.86 1-5 1.00 5.00
Meal Change COVID-19 361 2.89 3.00 1.04 1-5 1.00 5.00
Analytical Strategy
Data was collected online using the Qualtrics software. For the statistical analyses the data was copied into SPSS. As discussed before, a dummy variable was created for the channel
choice variable and a categorical one for the healthy food consumption variable. Normality tests, descriptive statistics, kurtosis and skewness have been computed. Furthermore, no outliers have been found. Regression analyses were computed to establish the main effect of channel choice on the different dependent variables. In order to test the moderating role of stress (COVID-19), shopping frequency, and healthy food consumption, an SPSS macro of Hayes (2012) was used (PROCESS).
Results
A table of correlation coefficients was obtained through SPSS, called correlation matrix, for all of the combinations of variables. According to the correlation matrix in Table 4, stress (COVID-19) is negatively correlated with age (r=-0.414, p=<0.01) and eating habits change COVID-19 (r=-0.126, p<0.05), and positively correlated with stress change COVID-19 (r=0.525, p=<0.01). Healthy food consumption is positively related to eating habits change COVID-19 (r=0.167, p<0.01). Age is negatively related to stress change COVID-19 (r=-0.217, p<0.01). Stress change COVID-19 is negatively related to eating habits change COVID-19 (r=- 0.130, p<0.01). Last, eating habits change COVID-19 is positively related to meal change COVID-19 (r=0.170, p<0.01).
Table 4: Means, Standard Deviations, Correlations
Variables M SD 1 2 3 4 5 6 7
1. Healthy Food Consumpti on
7.22 2.400 1
2. Stress (COVID- 19)
2.27 .566 .038 1
3. Shopping Frequency
1.94 .777 -.021 -.105* 1
4. Age 31.0 13.160 .041 -.414** .019 1 5. Stress
Change COVID-19
3.75 .926 -.075 .525** .012 -.217** 1
6. Eating Habits Change COVID-19
3.12 .927 .167** -.126* .026 .067 -.130* 1
7. Meal Change COVID-19
2.89 1.019 .066 .005 .012 .021 -.062 .170** 1
** Correlation is significant at the 0.01 level (2-tailed)
* Correlation is significant at the 0.05 level (2-tailed)
Adherence to the MD
In order to assess the compliance to the MD recommendations during the COVID-19 pandemic, the MEDAS questionnaire was included in the survey. Participants were divided among 3 classes on the basis of the MEDAS values. For each food, the difference in compliance rates were calculated. These differences were depicted in a radar chart to illustrate the gap between the current state (percentage of respondents currently adherent to each dietary recommendation and the ideal situation (100% compliance) (Figure 2+3).
As expected, among the three classes of adherence to the MD, there were significant differences for most of the items. In particular, in the highest adherence to the MD, the intake of fruit, vegetables, nuts and legumes was respectively: 40.0%, 100.0%, 80.0% and 100.0% for the online shoppers and 91.3%, 87.0%, 82,6% and 87.0% for the offline shoppers (Figure 2+3), underlining the improvement of the consumption of typical components of the dietary pattern in our Mediterranean population. Moreover, the consumption of foods not included in the MD profile seems to be reduced. When comparing the online to the offline group, we can see several differences; in the high-adherence group, fruit intake for the online shoppers was 40.0%, compared to 91.3% of the offline shoppers; in the medium-adherence group, online shoppers had a significantly higher intake of legumes (57.9%) and nuts (68.0%), compared to the offline shoppers (37.9% and 43.0%); in the low-adherence group, the outcomes are quite similar. The radar chart of the overall sample is depicted in Figure 8 (see Appendix).
Figure 2: Online Channel: Compliance with items from MEDAS according to high, medium and low adherence to the Mediterranean Diet (MD). The radar chart plots the values of each item of the MEDAS score along a separate axis that starts in the center of the chart (0%
compliance) and ends at the outer ring (100% compliance). The values are the percentage of the respondents adherent to each recommendation
0,0%
25,0%
50,0%
75,0%
100,0%
Olive oil, main dressing
Olive oil, >= 4 ts/day
Vegetables, >= 2 s/day
Fruits, >= 3 s/day
Red meat, < 1 s/day
Butter, < 1 s/day
Sweet beverage, < 1 s/day Wine, >= 7 s/week
Legumes, >= 3 s/week Fish and seafood, >= 3
s/week Sweets, < 2 s/week
Nuts, >= 3/week White meat over red
‘Sofrito’
MD ADHERENCE PROFILES
Low Adherence (n=6) Medium Adherence (n=19) High Adherence (n=5)
Figure 3: Offline Channel: Compliance with items from MEDAS according to high, medium and low adherence to the Mediterranean diet (MD). The radar chart plots the values of each item of the MEDAS score along a separate axis that starts in the center of the chart (0%
compliance) and ends at the outer ring (100% compliance). The values are the percentage of the respondents adherent to each recommendation
Table 5 shows the results of the positive answers to the MEDAS questionnaire and the adherence to MD. A significant difference between gender was found for the MEDAS score (r
= 0.105, p < 0.05). In particular, females resulted to have a higher MEDAS score when compared to males (r = 0.105, p = 0.047). Finally, no difference between age, employment
0,0%
25,0%
50,0%
75,0%
100,0%
Olive oil, main dressing
Olive oil, >= 4 ts/day
Vegetables, >= 2 s/day
Fruits, >= 3 s/day
Red meat, < 1 s/day
Butter, < 1 s/day
Sweet beverage, < 1 s/day Wine, >= 7 s/week
Legumes, >= 3 s/week Fish and seafood, >= 3
s/week Sweets, < 2 s/week
Nuts, >= 3/week White meat over red
‘Sofrito’
MD ADHERENCE PROFILES
Low Adherence (n=82) Medium Adherence (n=203) High Adherence (n=46)
status groups, education level groups and living situation groups was found for the MEDAS score. Table 21 (see Appendix) shows the results of the whole sample.
Table 5: Positive answers to MEDAS-questionnaire and adherence to the MD
Offline shoppers (n = 331) Online shoppers (n=30)
Olive oil, main dressing 254 (76.7) 22 (73.3)
Olive oil, >= 4 ts/day 83 (25.1) 7 (23.3)
Vegetables, >= 2 s/day 220 (66.5) 21 (70.0)
Fruits, >= 3 s/day 133 (40.2) 11 (36.7)
Red meat, < 1 s/day 245 (74.0) 23 (76.7)
Butter, < 1 s/day 227 (68.6) 18 (60.0)
Sweet beverage, < 1 s/day 245 (74.0) 22 (73.3)
Wine, >= 7 s/week 66 (19.9) 7 (23.3)
Legumes, >= 3 s/week 128 (38.7) 16 (53.3)
Fish and seafood, >= 3 s/week 59 (17.8) 6 (20.0)
Sweets, < 2 s/week 156 (47.1) 17 (56.7)
Nuts, >= 3/week 146 (44.1) 19 (63.3)
White meat over red 180 (54.4) 15 (50.0)
‘Sofrito’** 238 (71.9) 22 (73.3)
MEDAS score 7 [6-9]* 7 [6-9]*
Adherence to MD
Low 82 (24.8) 6 (20.0)
Medium 203 (61.3) 19 (63.3)
High 46 (13.9) 5 (16.7)
Positive answers to the MEDAS questionnaire. Compliance rates of at least 50% are indicated in italics. Data are expressed as number and percentage in parenthesis (n (%)) for categorical
