1. Introduction
Since December 2019, the international community has been dealing with COVID-19, which spread to countries around the world and led to 41,104,946 confirmed cases and 1,128,325 confirmed deaths [
1]. On 31 December 2020, Saudi Arabia was the most affected country in the Arab world and was 35th in the world, with 362,601 cumulative cases and 6214 cumulative deaths [
2], despite previously experiencing a similar coronavirus epidemic, Middle East respiratory syndrome (MERS). This global crisis has had consequences at all economic and social levels. It has ravaged the world’s economies and healthcare systems and instilled considerable fear, panic, and uncertainty in people of all countries [
3,
4]. With the spread of the pandemic, the mobility of people and goods was reduced in most countries as authorities were forced to apply a full or partial lockdown, leading to disruptions in the normal functioning of markets and economic and social facilities.
As described in [
5], the COVID-19 pandemic has encouraged countries to diversify their trade to avoid external shocks. Indeed, the volatility in oil prices that we saw at the start of the COVID-19 crisis negatively impacted the oil-based economies of several countries around the world. In addition to drops in their gross domestic product (GDP), several countries, like Azerbaijan, have faced downturns in their oil and gas sectors, which consequently caused further downturns in the wider economy. In addition, in Azerbaijan, oil price developments have led to higher inflation and decreased consumption of goods and services [
6,
7].
Several questions have arisen that need to be answered based on an analysis of purchasing behaviors during the first wave of the COVID-19 pandemic. The relationship between government measures to combat the pandemic and their side effects, such as triggering panic buying, was examined in [
8]. Consumers’ sentiments influence their choice of purchases and shopping channels, and these have evolved along with the changes brought by COVID-19 and the related interventions of public authorities. The crisis has created two types of consumer: (i) consumers who remain pessimistic and continue to stockpile essential items, possibly preferring online shopping, and (ii) consumers who are relatively optimistic and continue to purchase nonessential goods and services through various channels. In a recent study examining the notions of self-regulatory theory and personal well-being, a relationship was established between COVID-19 as a disruptive perceived threat and stockpiling behavior [
9].
Consumers are primarily affected by this pandemic because they represent the ultimate recipients of the various production, marketing, and other efforts of organizations [
10]. This pandemic has had direct and indirect effects on lifestyles, purchasing power, desires, and the general consumption of goods and services [
11,
12,
13]. In this sense, the devastating effects this has had on societies and economies of all sorts was reported in [
14]. The importance of young employees feeling secure and satisfied with their jobs during the COVID-19 pandemic has been raised [
15], along with the effect this has had on performance in organizations [
16,
17].
The world has experienced several health crises, such as the epidemics/pandemics caused by Ebola, severe acute respiratory syndrome (SARS), MERS, swine flu, and dengue [
12], and mitigating actions have been employed to manage the health risks [
18]. These health crises impacted consumer behavior in terms of purchasing behavior to mitigate health risk. In response to the effects of the COVID-19 pandemic, some researchers have speculated that human behavior will be pushed in different directions. For example, as shown in [
13], the new normal for consumer behavior could be explored through several dimensions, such as understanding it through economies of consumption, savings, and health, realigning the current younger generation to new life principles and building a new segment of consumers, and inventing new products or services to attract the conscious generation. In several countries, this health pandemic has led to the limited availability of some goods, with commodity prices reacting strongly and reflecting changes in supply and demand caused by measures aimed at limiting contagion [
19,
20]. Managerial theory shows the destabilization of consumption during crises, particularly health crises, and previous research has also shown that consumer behavior is disrupted.
Several categories of people are more exposed to COVID-19 contamination, such as healthcare workers, who are among the most at-risk groups, especially those who care for COVID-19 patients [
21].
Thus, from a physical and psychological viewpoint, food has never been so important in the daily lives of individuals, yet the COVID-19 pandemic seems to be changing humanity’s relationship with food. Several nutritionists have emphasized the effects of healthy food choices, especially during a pandemic [
22]. Healthy eating behaviors involve eating a variety of foods that provide the nutrients people need to stay healthy, feel good, and have plenty of energy. These nutrients include protein, carbohydrates, fats, water, vitamins, and minerals, and they are very important for human health. When combined with physical activity and maintaining a healthy weight, consuming healthy food is a great way to help the human body stay strong and healthy.
From a consumer behavior viewpoint, the literature classifies the numerous facets of purchasing decisions into internal and external variables, and they consider the consumption of goods during a pandemic to be a risky phenomenon [
23]. However, to better identify consumer behavior and suggest actions to be taken, some authors insist that these risks must be examined in detail. For example, in [
24,
25], overall risk is divided into performance risk, social risk, physical risk, financial risk, psychological risk, and time-loss risk. For example, the physical risk of making purchases during a pandemic modifies consumers’ purchasing behaviors and habits [
2,
26,
27,
28,
29], and increased financial risk reflects the possibility of losing income following a purchase [
30]. Consumption behavior during a pandemic is, therefore, sensitive to consumer attitudes [
31,
32] and affected by the social distancing measures imposed by public authorities [
11,
31,
33].
Unlike traditional marketing, in experiential marketing, salespeople can facilitate the creation of a special relationship between a brand and potential customers. This makes it possible to involve the buyer and generate feelings of well-being, which, in turn, builds a feeling of customer loyalty toward the brand. Over time, this technique guides the customer’s shopping experience [
34,
35], such as when the customer is considered to act not just rationally but also emotionally based on accumulated experiences. However, recent research has shown that people’s reactions to the COVID-19 pandemic vary according to age and gender [
31,
36]. People have different levels of concern about the effects of the pandemic, and this guides their behavior. For example, from an economic viewpoint, older people feel less affected than younger, working-age people, but from a physiological viewpoint, the health consequences have forced them to change their purchasing habits. Women also tend to be more concerned about the health effects of the pandemic than men. During the COVID-19 pandemic, personal factors such as age and gender are key factors explaining behavior, and they moderate the relationship between the intent to buy and purchasing decisions [
31,
37]. We therefore believe it is necessary to study consumer behavior during the COVID-19 pandemic and learn about the new characteristics of this behavior, so we can uncover the circumstantial specifics related to the purchasing process. This study, therefore, aims to understand some of the determinants and moderators of purchasing decisions by consumers of healthy foods.
The rest of this paper is structured as follows. First, we examine the theoretical background that characterizes the pandemic-related literature to explain consumer behavior. Next, we introduce our research model and propose related hypotheses, and then explain our empirical study and its data-collection methods and results. Next, we discuss these results by detailing all of the key findings and the theoretical and practical implications of this work, as well as its limitations and some suggestions for further research. Finally, we conclude the paper.
6. Discussion
6.1. Key Findings Based on Hypothesis Tests
The structural model (Equation (1)) shows a good fit to the data (
Table 3). Indeed, the level of the indices ranges from good to excellent (GFI > 0.9; AGFI > 0.9; RMSEA = 0.069), and the incremental indices show the good quality of the model. Thus, the results of the estimation of structural equations prove that there is a significant relationship between the intention to buy and overall risk. This risk is, therefore, negatively (γ = −0.136) and significantly (CR = 4.071;
p < 0.05) related to purchasing intention during the COVID-19 crisis (
Table 4), so H1 is supported. For H2, our structural model has a good fit with the data (
Table 3), and all the parsimonious and incremental indices (GFI > 0.975; AGFI > 0.937; RMSEA = 0.074) indeed reinforce the good quality of the model. Structural equation estimations also prove that there is a significant relationship between purchasing experience and intention to buy: intention to buy is positively (γ = 0.210) and significantly (CR = 9.531;
p < 0.05) related to the shopping experience (
Table 2 and
Table 3), so H2 is also supported. For P1, the data have good fitting qualities (
Table 3), and all the indices indicate excellent model quality (GFI > 0.977; AGFI > 0.945; RMSEA = 0.065) (
Table 3). Then, the results of structural equation estimation confirm a significant relationship between intention to purchase and ease of purchase: ease of purchase is positively (γ = 0.308) and significantly (CR = 3.071;
p < 0.05) related to intention to purchase (
Table 4), so P1 is supported.
H3 assumes that the intention to purchase has a positive effect on the decision to purchase healthy foods during the COVID-19 pandemic. An examination of this relationship is possible through a binary logistic regression of all the variables. We then tested the following hypothesis: The greater the intention to buy, the more likely consumers will decide to purchase healthy foods. This makes it possible to verify consumer behavior when deciding between purchasing (1) and not purchasing (0). Empirically, the model’s quality is satisfactory, because 58.31% of respondents′ statements are split between buyers and non-buyers. The goodness-of-fit indicators (χ2 model = 20.810; ddl/χ2 = 2; <0.05) show that the model is significant (
Table 5).
The goodness-of-fit test divided the subjects into deciles based on the predicted probabilities, and then we calculated a chi-square from the observed and expected frequencies. A
p-value of 0.111 was calculated from a chi-square with six degrees of freedom. This value is greater than 0.05, implying that the model’s estimates adjust to the data to an acceptable level [
88]. Cox and Snell’s R
2 and Nagelkerke’s R
2 confirm the relationship between the dependent and independent variables. The model, therefore, explains 8.9% of the variance of the dependent variable. This result is acceptable, if not strong enough, but it proves that the model has a good fit.
Table 5 allows an evaluation of the contribution of each variable (purchase intent, purchase decision) in explaining the choice between purchasing and not purchasing. The β coefficient also makes it possible to compare the relative weights of the variables as follows: when the absolute value of a coefficient is high, the corresponding variable is increasingly important in explaining the model. On the other hand, a negative β means that the variable has more influence on the decision to not make a purchase. However, the analysis of the β coefficients did not allow us to measure the influence of each variable, so, for this, we worked with the odds ratio Exp(β), which facilitates an easier interpretation of the results.
Table 5 shows that purchase intention has a significant and positive effect (β = 0.229; χ2 de Wald = 12.250;
p = 0.000). In addition, the Wald coefficients were significant, allowing us to finally interpret the results and conclude that the logistic regression we carried out fully affirms H3.
Regarding the test for the moderating effect of age and gender on the relationship between the fear of COVID-19 and the intention to purchase healthy foods, the results validate the propositions. Indeed,
Table 6 shows that the calculated difference in chi-square values is not significant at the 5% level. Thus, the correlations between the constructs and their measurement variables are not significantly different from one group to another, proving that the two models are “equivalent” or “stable” because the measurement parameters between the two groups of people are invariant.
Finally, during the confirmatory analysis, we compared the chi-square value of a free model (where one parameter is not equal between the two groups) with the chi-square value of a constrained model (where all parameters are equal between the two groups). If, between the two models, the difference in chi-square value is significant, the variable has a moderating effect.
Examining the regression coefficients and the significance of the link between the standardized coefficients helps to provide a clear idea of the direction of influence for the moderating variable.
Table 7 shows that the chi-square difference test is significant, so the free model has greater explanatory power than the constrained model. This indicates that the fear of COVID-19 has a moderating effect on the relationship between the intention to purchase and the ease of purchase. By applying the “step-by-step downward plausibility” method, the variables that significantly influence the purchasing decision, whether directly or through an interaction with intention to purchase, will appear.
Table 7 summarizes only the significant influence relationships, whether direct or through interactions. The regression of Equation (2) confirms the direct influence of intention to purchase on purchasing decision, and a person’s age and gender were found to be moderators of the relationship between intention to buy and purchasing decision (
Table 8). Therefore, H4 is supported. In addition to the moderating role of these two variables (age and gender), they are also quasi-moderating variables (
Table 9). Indeed, 72.5% of purchases made during the COVID-19 period can be explained through consumers’ intention to purchase, age, and gender (Nagelkerke′s R
2 = 72.1%). Our results, therefore, confirm that age and gender directly influence the completion of a purchase, so these two variables play the role of quasi-moderators.
Table 10 summarizes all of the obtained results.
6.2. Theoretical Implications
We present six theoretical implications of these findings. First, in contrast to recent studies of specific aspects of consumer behavior during the COVID-19 pandemic, we integrated several variables related to consumer behavior during times of crisis and empirically show how these behavioral variables interact to stimulate purchasing decisions and influence the purchasing process during a pandemic. Second, the results emphasize that when the risks of consumption are high for consumers, they have a lower intention to purchase and may no longer risk making a purchase. If the consumer is experienced with and accustomed to shopping in department stores, purchasing healthy foods during the COVID-19 pandemic will be faster and more intensive. Consumers of healthy foods are quick to obtain and consume these desirable goods because their intention to purchase positively influences their purchasing decisions. Indeed, previous researchers mostly investigated the role of such variables in the purchasing process, but with less precision, especially in a time of crisis. Furthermore, to the best of our knowledge, this is the first study to investigate how consumer behavior variables operate during a pandemic, such as the COVID-19 crisis in Saudi Arabia.
Third, unlike in previous studies, which investigated the roles of age and gender in people’s responses to the COVID-19 pandemic and showed that people have different levels of concern about the disease and modify their behavior according to their gender and age [
33,
37], we empirically show that these two variables play a role in moderating the relationship between the fear of COVID-19 and intention to purchase. Indeed, consumers of healthy foods do not take long to obtain and consume these goods, because their intention to buy positively influences their purchasing decisions, and their gender and age are “quasi-moderators” of the relationship between the intention to buy and actually buying healthy foods in supermarkets. However, several research studies focused on the mental health of individuals during the COVID-19 crisis, and psychologists in particular have examined anxiety disorders as a major risk related to epidemic situations (fear that they or loved ones will catch the disease) and living conditions in lockdown (social isolation, loss of income, frustration, etc.). This study expands the existing research into consumer behavior.
Fourth, we used rigorous measurement instruments and justified the choice of the different scales used from a conceptual viewpoint. These scales were relatively short so as to avoid respondents becoming weary, and the expressions were carefully adapted to the Saudi context through a PCA conducted among a sample of 100 people. This made it possible to adapt the items to the actual context, which is characterized by the excessive consumption of sweet and fatty foods. Fifth, our qualitative study is globally important, because it enabled us to determine the moderating role of personal sociodemographic variables like age and gender [
37]. Sixth, our research contributes to the managerial literature on developing marketing strategies, so actors involved in the mass distribution of healthy foods can learn how consumers of those foods behave and modify their behavior.
Our results, therefore, help to understand consumer behavior during a health crisis, and they show that it is important for retailers to develop their points of sale and adapt their assortments with logical category management, because these will facilitate consumers’ purchasing process, and therefore help them to reduce the time they spend shopping.
In addition, these results extend previous research into consumption risks [
2,
23,
24,
25,
26,
27,
28,
29,
30] and changes in attitudes during times of crisis [
30,
31,
32]. They also enrich the literature about the effects of experiential marketing on the purchasing experience [
34,
35] and purchasing intentions [
38,
39,
47,
48], the impact of ease of purchase on purchasing decisions [
66,
67,
68,
69,
70], and the effects of contamination fear on consumer behavior [
21,
22,
67] and the purchasing experience [
31,
34,
35,
36,
43,
46,
60,
64,
65,
83,
84]. In this sense, this research helps to understand the various elements of consumer behavior in the face of the COVID-19 crisis, and contributes to the development of an important debate about which marketing strategies and tactics retail companies should adopt. Finally, this research provides reliable measurement tools that will be useful for marketers during the design phase of a marketing strategy during a crisis.
6.3. Practical Implications
It is expected that Saudi Arabia will soon enter a second wave of the COVID-19 pandemic. As of 31 December 2020, it recorded 362,601 cumulative cases and 6214 cumulative deaths [
1], which are both significantly more than previously expected [
19]. As in many other places in the world, the Saudi public authorities imposed strict limitations on the movements and gatherings of its citizens to contain and suppress the spread of COVID-19 [
3,
4,
8,
21]. However, aside from some compulsory closures, the country has not imposed strict measures on sales areas, especially supermarkets. For example, how many people should be allowed in a particular area so that buyers can maintain sufficient social distancing has not clearly been defined. Our results demonstrate that the most cautious consumers confirmed that their purchasing process under such conditions had been disrupted by the fear of COVID-19 during the first wave of the pandemic. Most said that they would continue to adapt their behavior when the country enters a second wave.
This study’s results support previous work about changes in the purchasing decision process during a health crisis, consumer behavior during purchasing, and health-risk mitigation behaviors [
23,
24,
25,
31,
32]. They also support previous contributions about consumer caution amid rising contagion because consumers are confronting goods in stores [
10,
19,
21]. This health pandemic has had direct and indirect effects on consumers’ lifestyles, purchasing power, desires, and general consumption of goods and services [
11,
12,
13]. Previous studies have included changes in people’s relationships with food quality and food use and a growing preference for healthy foods [
22,
36]. The rising consumption risks have effects on purchasing [
23,
24,
25], such as physical risk [
2,
26,
27,
28,
29] and financial risk [
30], and this influences consumers’ attitudes and behavior in a time of crisis [
30,
31,
32]. Experiential marketing can also contribute to the purchasing experience [
34,
35].
These practical implications indicate that retailers can anticipate unusual consumer behavior during a health crisis, and these changes in behavior can be determined by different levels of fear of contamination according to the consumer’s age and gender. Although our focus was not primarily on creating or adapting measurement scales, the variables we used were developed based on the extensive literature. We also performed exploratory factor analyses followed by confirmatory factor analysis in the validation phase to verify convergent and discriminant validity.
6.4. Limitations and Future Research
We would like to emphasize four limitations of our research. First, the collected data reflected the situation in a country that was heavily impacted by active COVID-19 cases. The country had prepared for the situation by closing schools, restaurants, and public services, and this may have sparked fears of upcoming disruption and triggered people’s behavioral responses. Second, from a theoretical viewpoint, another limitation concerns the conceptual model and the survey method. Indeed, we questioned consumers in front of large stores in a time of crisis, and the answers provided may be biased and may not accurately reflect real consumer behavior. Therefore, the observation or “list before, list after” method might have better reflected the reality of consumer behavior. Third, from a methodological viewpoint, the data collection was time-constrained, and we were unable to survey more people in the various regions of Saudi Arabia, which makes the results difficult to generalize. Fourth, we only considered situational variables that could have given a better explanation of consumer behavior during the COVID-19 crisis. Thus, the failure to consider other variables linked to consumer choice, such as time pressure, the attractiveness of available options, the flexibility of the environment, and so on, limit the usefulness of our results. Consequently, these limitations should be considered when interpreting the results, and for all of these reasons, we recommend that future research take the indicated limitations and research gaps into consideration.
7. Conclusions
Research continues to explain the profound changes in consumer behavior since the onset of this crisis from a psychological perspective. From the same perspective, we collected data from Saudi respondents (n = 360) regarding their consumption of healthy foods during the first wave of the COVID-19 pandemic. In addition, we developed a quantitative and qualitative approach with the objective of modelling consumer behavior. We performed double modelling through (i) a structural equation model to explain the impact of three variables (purchasing experience, ease of purchase, and intention to purchase) on purchasing decisions and (ii) a binary logistic regression model to test the relationship between the intention to buy and purchasing decisions.
The empirical results suggest that women are more concerned than men about catching COVID-19. Indeed, most female respondents considered the novel coronavirus to present an imminent danger, and because they are so afraid of contamination, they take a lot more precautions than men do, especially when shopping. Furthermore, older people have a greater fear of infection when shopping; the youngest respondents, both male and female, were less afraid of coronavirus, so they took fewer precautions against contamination. However, all respondents reported a drop in their visits to shops, restaurants, and gathering places. Therefore, gender and age act as quasi-moderators in the relationship between the fear of COVID-19 and the intention to purchase. The fear of COVID-19 stems from the fact that these people care much more about their health than others’ health, so they are more cautious about avoiding contamination.
It is, therefore, a priority for public decision-makers to take actions to maintain a minimum level of well-being and prevent the development of disorders in the population, because in the short term, this will reduce the demand on the health system and hospitals caused by an influx of people with symptoms of acute anxiety or stress, especially as Saudi Arabia prepares for a second wave of the pandemic. Indeed, during the first wave, consumption risks, such as performance, social, physical, financial, psychological, and time-loss risks, were amplified. The risks became greater and greater, so they negatively influenced the intention to purchase, and consumers gradually decreased their visits to stores. However, their buying experiences and intention to buy meant that they increased their purchases of healthy foods compared with other goods.
While our contributions support and build upon the results of previous research into strategic and tactical marketing decision-making, the lockdown and other government actions meant we investigated fewer people and therefore collected less information. Accordingly, we encourage scholars to take the opportunity to collect further empirical data about purchasing behavior during this crisis period, especially as the country will likely enter a second wave of the pandemic. Finally, while the number of positive COVID-19 cases is now low in Saudi Arabia, on 7 January 2021, public authorities discovered the first 10 cases of a new variant of the virus.