Comparing Models with Positive Anticipated Emotions, Food Values, Attitudes and Subjective Norms as Inﬂuential Factors in Fast-Food Purchase Intention during the COVID-19 Pandemic in Two Channels: Restaurants and Mobile Apps

: The purpose of this research is to determine if positive anticipated emotions, food values, attitudes and subjective norms inﬂuence food purchase intention in two different models: a fast food restaurant and a food delivery service via mobile apps. For this study, we utilized a non-experimental, causal, descriptive and cross-sectional design. From October 2020 to January 2021, self-administered online surveys were distributed to a convenience sample of 200 fast-food consumers at restaurants, and users of food delivery services via mobile apps Puebla City, Mexico. IBM–SPSS Statistics and the SmartPLS 3 Partial Least Squares Structural Equation Modeling were used to test our hypotheses. The results underscored a difference in attitudes between the models. The attitude toward the brand positively and signiﬁcantly inﬂuenced purchase intention via mobile apps, whereas attitude toward eating a hamburger positively and signiﬁcantly inﬂuenced purchase intention of visiting a fast-food restaurant. In both models, positive anticipated emotions exhibited the closest relationships with purchase intention, attitude toward the brand and attitude toward eating a hamburger, whereas food values exerted an insigniﬁcant effect on attitudes and purchase intention. Future research should consider performing a face-to-face survey with a random sample while accounting for different demographics, regions and countries, as well as including other brands, food types and restaurants.


Introduction
The tourism sector in general, and restaurants in particular, are especially vulnerable to disease outbreaks, which can threaten firms' financial viability and impose great burdens on workers [1]. Kim et al. (2020) [2] confirmed the negative influence of epidemic disease outbreaks on the restaurant industry.
With the onset of the COVID-19 pandemic, humans have changed their habits and behaviors around food acquisition and consumption [3][4][5][6]. In tandem, the introduction of social distancing as a pandemic containment strategy has disrupted food systems [7]. In Mexico, for instance, there was a 90% decrease in the number of diners sitting at restaurant tables by 18 March 2020 [8]. However, some food and beverage establishments were able to continue operating by adopting food delivery services [8]. Against this background, the present research seeks to determine the factors that positively and significantly influence consumers' food purchasing during the COVID-19 pandemic under two situations: whether people (1) are eating at a fast-food restaurant and (2) are using a food delivery service via mobile apps. To this end, we followed several studies and emphasized purchase intention as an important factor [9][10][11]. The literature review confirmed that several Sustainability 2021, 13, 12857 2 of 16 variables from the Theory of Planned Behavior (hereafter TPB) influence food purchase intention [12][13][14][15][16][17]. According to Ellison et al. (2021) [18], there is an ongoing shift toward online food purchasing, while Choi et al. (2021) [10] affirmed that more people are using food delivery apps. Several studies on such delivery apps have found that they can impact loyalty, service quality, packaging and customer satisfaction [19]. Yet, none of those studies have addressed food values, positive anticipated emotions or TPB variables (such as the influence of food purchase intention) while comparing different channels.
It is important to note that choices around food and food channels is increasingly complex-and COVID-19 has only exacerbated this reality by adding a safety dimension to the choice of distribution channel [20,21]. Various channels are having to modify their sales, supply and satisfaction planning in response to dramatic behavioral shifts from new and current customers [22]. For this reason, this study focuses on detecting the factors that cause a purchase intention which implies making adjustments to the sector, taking purchase intention as a reference. Practitioners would benefit from knowing what factors can predict food purchase intention in this climate. To this end, we draw from the TPB and focus on food values, positive anticipated emotions, attitude toward the brand, attitude toward eating and subjective norms [11]. In this way, the paper seeks to illuminate any meaningful differences between eating at fast-food restaurants and using a delivery service through mobile platforms.
In sum, the present article analyzes the decision-making process behind consuming fast food from two types of channels: in-person restaurants (traditional) and mobile delivery apps (modern). This study aims to examine the influence of: (i) food values and positive anticipated emotions on consumers' attitudes and subjective norms; (ii) both attitudes to purchase intention; and (iii) subjective norms on purchase intention. The paper proceeds as follows: we review the previous literature on food values, positive anticipated emotions, attitudes and subjective norms on purchase intention; after that step, we delineate the research hypotheses and then define the empirical methodology used to test said hypotheses; and in the final section, we describe the main findings and highlight some implications for theory and management. Bagozzi et al. (2006) [23] defined anticipated emotions as a person's belief about the emotional consequences of an action. Mellers and McGraw (2001) [24] suggest that these emotions serve to guide behavior and make decision making easier. Some authors consider emotions to be indicators of an individual's intention to perform a particular behavior, such as purchase behaviors during the information processing phase [25,26]. Ajzen and Sheikh (2013) [27] included TPB with emotions as an influence variable in purchase intention.

Positive Anticipated Emotions
A consumer who experiences positive emotions toward a brand will be more likely to develop a relationship with the brand, which can then shape their future perceptions, experiences and attitudes toward the brand and its offerings [28]. Similarly, foods generate powerful emotional responses that are fundamental to the satisfaction of consumers' needs and expectations [29]. The study by Pérez-Villarreal et al. (2019) [11] connected food values and positive anticipated emotions with two different attitudes to predict purchase intention for hamburgers. The authors concluded that positive anticipated emotions such as happiness, enthusiasm and satisfaction positively influence attitude toward the brand, attitude toward eating and the intention to purchase a hamburger from a fast-food restaurant.
In short, emotions work via attitudes to substantially affect people's purchase intention toward a specific product or brand [29]. Prinyawiwatkul (2020) [30] affirmed that the emotions stemming from eating are foundational to consumers' satisfaction and thereby significantly influence their attitude toward purchasing from a particular brand.
Several authors have incorporated the variable of positive anticipated emotions into the original TPB model and established its importance [27,31,32]. Accordingly, we propose the following hypotheses: Hypothesis 1 (H1). Positive anticipated emotions will positively influence attitude toward the brand among people (a) eating in a fast-food restaurant and (b) eating fast food via a mobile delivery app.

Hypothesis 2 (H2).
Positive anticipated emotions will positively influence attitude toward eating among people (a) eating in a fast-food restaurant and (b) eating fast food via a mobile delivery app.

Hypothesis 3 (H3).
Positive anticipated emotions will positively influence subjective norms among people (a) eating in a fast-food restaurant and (b) eating fast food via a mobile delivery app.

Hypothesis 4 (H4).
Positive anticipated emotions will positively influence purchase intention among people (a) eating in a fast-food restaurant and (b) eating fast food via a mobile delivery app.

Food Values
According to Martínez-Ruíz and Gómez-Cantó (2016) [33], Izquierdo-Yusta et al. (2020) [34] and Muro-Rodríguez et al. (2021) [35], product attributes can highly influence a product choice. Consumers assign a level of importance to certain product features, which could positively or negatively affect the purchase decision process. Several authors have affirmed that food values reflect the importance of product attributes [11,15]. Such food values are emblematic of Marketing 3.0 approaches, which emphasize treating individuals as full human beings rather than as mere consumers. This is also known as the valuesdriven era because marketing decisions often try to incorporate consumers' personal values [35]. In recent years, scholars have connected food values to emotions and attitude, with the goal of enhancing the relationship between food values and purchase intention [15]. One landmark study in this vein is by Lusk and Briggeman (2009) [36], who synthesized the literature on food preferences and human values to devise a food values scale that can reflect consumers' willingness to purchase. Their efforts led to the values proposed by Lusk (2011) [37] constituting a fundamental contribution to marketing and consumer behavior [33].
According to Rokeach (1973) [38], a value is a belief that defines an individual's behavior. Meanwhile, a value system results when a group of individuals hold the same ideology and preferences for an enduring length of time.
Manan (2016) [39] and Lang and Lemmerer (2019) [40] proved that food-related values effectively influence attitudes, which then impact food purchase intention and behavior [41]. Similarly, Cunha et al. (2014) [42] found that food values influence attitude toward eating a specific food. Likewise, Pérez-Villarreal et al. (2019) [11] proved that food values influence people's attitude toward not only eating a hamburger at a fast-food restaurant, but also toward the brand itself. Nevertheless, none of these authors analyzed whether these same attitudes shift when people use a food delivery service via mobile apps. Because of the COVID-19 pandemic, we expect that consumers have assigned different importance to food values in 2020 compared to 2019 [18]. Based on the above ideas, we propose the following hypotheses:

Attitudes, Subjective Norms and Purchase Intention
Every attitude and intention are stemmed from values, through a hierarchic relationship between values, attitudes, intention and finally, behavior [38,43]. The Theory of Planned Behavior (TPB) has recently become a premier tool for explaining purchase intention and behavior, especially in relation to several facets of eating [44]. TPB argues that consumers' identities-and by extension, their food choices-are rooted in their intentions, experiences, attitudes and subjective norms. Therefore, consumers will never hold the exact same opinions, even if they technically belong to the same market segment [45,46]. TPB contends that behavior is influenced by three determining factors: (1) attitude, (2) subjective norms and (3) perceived control over one's own behavior [47]. Note that the theory treats these three variables as conceptually independent [47,48]. This study specifically focused on attitudes and subjective norms with the possibility of continuing the research by adding perceived control.
Multiple authors have considered attitude to be a relevant factor in interpreting and predicting purchase intention, including toward food [11,13,47,49,50]. Bredahl (2001) [51] argued that a strong attitude can drive an individual's intention to purchase a product; in other words, an individual's attitudes and intentions to perform a specific behavior are intertwined.
Attitude toward the brand may positively and negatively influence purchase intention, as consumers will leverage brand knowledge when evaluating a product [52,53]. Individuals' experiences or recommendations will shape their attitudes toward a brand, and then they will decide to adopt or reject those perceptions [54]. Individuals generally become more familiar with, and positive toward, a brand through repeated exposure to it [55]. Such exposure is strongly related to purchase intention and even post-purchase behaviors [56].
With regard to food, attitude toward eating is a significant psychological factor that may shape purchase intention toward the foods in question [57]. Phrased differently, purchasing a food is reflective of a positive attitude toward it [11,58,59]. Thus, a restaurant should understand consumers' perceptions about the offered food, as well as the cultural, psychological and social motivations that drive consumer behavior [58,59]. Some factors that influence food purchase intention include taste, smell, texture, price, brand and quality, among others [60]. Another important factor is lifestyle, which shapes people's attitude toward what they eat and where they purchase it [12,61]. Naturally, people have had to adapt their lifestyles to the realities of the pandemic. Accordingly, we propose the following hypotheses: Subjective norms describe the social pressure exerted on an individual to perform or refrain from a specific behavior. The individual assesses whether relevant others agree that a behavior should be performed-that he/she "should do it" [47,62]. Fishbein and Ajzen (1975) [63] defined subjective norms as the sum of people's opinions about whether a given individual should (or should not) engage in a behavior.
Several authors have used the subjective norms variable to predict and understand intention and the resulting behavior in different fields, finding that this variable and attitudes are among the strongest predictors [15,64,65]. Scholars have even found a positive relationship between subjective norms toward food consumption and visiting fast-food restaurants [66]. Based on the above, we propose the following hypothesis:

Hypothesis 10 (H10). Subjective norms positively influence purchase intention among people (a) eating in a fast-food restaurant and (b) eating fast food via a mobile delivery app.
In sum, these ten hypotheses reflect ten different effects across two fast-food purchase models (within a restaurant vs. via a mobile delivery app) during a pandemic. The effects are: (1) attitude toward the brand on purchase intention, (2) attitude toward eating a hamburger on purchase intention, (3) food values on attitude toward the brand, (4) food values on attitude toward eating a hamburger, (5) food values on subjective norms, (6) positive anticipated emotions on purchase intention, (7) positive anticipated emotions on attitude toward the brand, (8) positive anticipated emotions on attitude toward eating a hamburger, (9) positive anticipated emotions on subjective norms and (10) subjective norms on purchase intention (see Figure 1).
Several authors have used the subjective norms variable to predict and understand intention and the resulting behavior in different fields, finding that this variable and attitudes are among the strongest predictors [15,64,65]. Scholars have even found a positive relationship between subjective norms toward food consumption and visiting fast-food restaurants [66]. Based on the above, we propose the following hypothesis:

Hypothesis 10 (H10). Subjective norms positively influence purchase intention among people (a) eating in a fast-food restaurant and (b) eating fast food via a mobile delivery app.
In sum, these ten hypotheses reflect ten different effects across two fast-food purchase models (within a restaurant vs. via a mobile delivery app) during a pandemic. The effects are: (1) attitude toward the brand on purchase intention, (2) attitude toward eating a hamburger on purchase intention, (3)

Materials and Methods
This research utilizes a non-experimental, causal, descriptive and simple cross-sectional design that has been supported by quantitative empirical evidence. The data were collected via a self-administered online survey distributed to a convenience sample of fast-food consumers over age 18. The survey was distributed via Google Forms and participants had to indicate that they eat at fast-food restaurants and/or have used restaurants' mobile apps. In the first phase, we distributed 250 surveys from October 2020 to January 2021. Of those, 80% responded completely, leaving 200 usable surveys for this research. The final number of the sample is according to the power statistical confidence level and margin of error. For these calculations, we used 11 predictors, with a 95% confidence level and 6.94 margin of error. The geographical scope was Puebla City, Mexico. The survey contained some screening questions that participants completed before answering to ensure that they met the criteria for the research. We used the IBM-SPSS Statistics Base version 22 and the SmartPLS 3 Partial Least Squares Structural Equation Modeling (hereafter PLS-SEM) to statistically test our hypotheses.

Materials and Methods
This research utilizes a non-experimental, causal, descriptive and simple cross-sectional design that has been supported by quantitative empirical evidence. The data were collected via a self-administered online survey distributed to a convenience sample of fast-food consumers over age 18. The survey was distributed via Google Forms and participants had to indicate that they eat at fast-food restaurants and/or have used restaurants' mobile apps. In the first phase, we distributed 250 surveys from October 2020 to January 2021. Of those, 80% responded completely, leaving 200 usable surveys for this research. The final number of the sample is according to the power statistical confidence level and margin of error. For these calculations, we used 11 predictors, with a 95% confidence level and 6.94 margin of error. The geographical scope was Puebla City, Mexico. The survey contained some screening questions that participants completed before answering to ensure that they met the criteria for the research. We used the IBM-SPSS Statistics Base version 22 and the SmartPLS 3 Partial Least Squares Structural Equation Modeling (hereafter PLS-SEM) to statistically test our hypotheses.

Survey Development
The survey featured 55 indicators (items) in total, divided into three sections: one corresponded to food values (33 items) and the other two sections (covering 22 items) corresponded to attitudes toward the brand and attitudes toward eating, respectively; all answers used a Likert scale from 1-5 (1 = not important at all/strongly disagree; 5 = very important/strongly agree). All items were tailored to reference the two models (inside the restaurant versus delivery via mobile app). Regarding attitude toward the Sustainability 2021, 13, 12857 6 of 16 brand, items were adjusted based on the brands used in this study. We used a generic hamburger as the focal food product when assessing the variables. Meanwhile, we applied a Likert scale from 1-5 (1 = strongly disagree, 5 = strongly agree) to measure positive anticipated emotions, used purchase intention and subjective norms. Indicators with a variance inflation factor (hereafter VIF) greater than 3.3 were eliminated [67]. Thus, the items ATEH2Rest and DelightedRest were eliminated from the fast-food restaurant model, while the items ATEH2Delivery, DelightedDelivery and NS2Delivery were removed from the mobile app model.
The item descriptions in Tables 1 and 2 were modified. We want to emphasize that the variable proposal applies to the two channels of interest: (1) the fast-food restaurant model and (2) food delivery service via mobile app model.

Appearance
The extent to which food is appealing.

Convenience
Ease of cooking and consumption of food.
Environmental Impact Impact of food production on the environment.

Equity
The extent to which all parties involved in food production equally benefit.

Organic
The extent to which food production is without modern technologies.

Nutrition
Amount and type of fat, protein, vitamins, etc.

Origin
Where food products were grown.

Price
The price paid for food.

Safety
Food consumption will not cause disease.

Taste
The extent to which food consumption is appealing to the senses.

Tradition
Traditional consumption patterns preservation.
Positive Anticipated Emotions adapted from Bagozzi and Dholakia (2006) [23] HappyRest If I can go eat a hamburger in a fast-food restaurant next month, I feel happy.

DelightedRest
If I can go eat a hamburger in a fast-food restaurant next month, I feel delighted.

ExcitedRest
If I can go eat a hamburger in a fast-food restaurant next month, I feel excited.

ProudRest
If I can go eat a hamburger in a fast-food restaurant next month, I feel proud.

SatisfiedRest
If I can go eat a hamburger in a fast-food restaurant next month, I feel satisfied.
Self-assuredRest If I can go eat a hamburger in a fast-food restaurant next month, I feel self-assured.

ATB2Rest
Admire the brand.

ATB3Rest
The brand fits my lifestyle.

ATEH2Rest
I would enjoy eating a hamburger. ATEH3Rest If I eat a hamburger, it would be satisfying for me.

ATEH4Rest
I eat a hamburger because it tastes good.

Latent Variable Observed Items Items Description
Subjective Norms (SN) adapted from Brinberg and Durand (1983), Izquierdo-Yusta, Martínez-Ruiz and Jiménez-Zarco (2011) [13,65] SN1Rest Most of the people important to me think I should eat in a fast-food restaurant.

SN2Rest
My friends think I should eat in a fast-food restaurant.

SN3Rest
If my friends eat in a fast-food restaurant, I will probably do it too.

SN4Rest
My family thinks I should eat in a fast-food restaurant.

SN5Rest
If my family eats in a fast-food restaurant, I will probably do it too.
PI2Rest I would consider buying a fast-food product if I have the need.

PI3Rest
It is possible to buy a product in fast-food restaurants.

PI4Rest
The probability of considering buying a product in a fast-food restaurant is high.

Latent Variable Observed Items Items Description
Positive Anticipated Emotions adapted from Bagozzi and Dholakia (2006) [23] HappyDelivery If I can eat a hamburger with a food delivery service via mobile app next month, I will feel happy.

DelightedDelivery
If I can eat a hamburger with a food delivery service via mobile app next month, I will feel delighted.

ExcitedDelivery
If I can eat a hamburger with a food delivery service via mobile app next month, I will feel excited.

ProudDelivery
If I can eat a hamburger with a food delivery service via mobile app next month, I will feel proud.

SatisfiedDelivery
If I can eat a hamburger with a food delivery service via mobile app next month, I will feel satisfied.
Self-assuredDelivery If I can eat a hamburger by food delivery service via mobile app next month, I will feel self-assured.

ATB2Delivery
Admire the brand.

ATB3Delivery
The brand fits my lifestyle.

ATEH2Delivery
I would enjoy eating a hamburger.

ATEH3Delivery
If I eat a hamburger, it would be satisfying for me.

ATEH4Delivery
I eat a hamburger because it tastes good.

SN2Delivery
My friends think I should eat by food delivery service via mobile app.

SN3Delivery
If my friends eat by food delivery service via mobile app, I will probably do it too.

SN4Delivery
My family thinks I should eat by food delivery service via mobile app.

SN5Delivery
If my family eats by food delivery service via mobile app, I will probably do it too.

Latent Variable Observed Items Items Description
Purchase Intention (PI) adapted from Chiu, Hsieh and Kuo (2012), Diallo (2012) [70,71] PI1Delivery I would probably buy products by food delivery service via mobile app.

PI2Delivery
I would consider buying products by food delivery service via mobile app if I have the need.

PI3Delivery
It is possible to buy a product by food delivery service via mobile app.

PI4Delivery
The probability of considering buying a product by food delivery service via mobile app is high.

Results
Henseler et al. (2016) [72] suggested that model goodness of fit should be assessed by using the standardized root mean squared residual (hereafter SRMR) and normed fit index (hereafter NFI). According to Hair et al. (2011) [73] and Hu and Bentler (1999) [74], a good fit parameter for SRMR is 0.05 to 0.08, and for NFI, 0 to 1, with numbers closer to 1 being better [73]. The fast-food restaurant model criteria were SRMR = 0.061 < 0.08 and NFI = 0.748, while the mobile app model criteria were SRMR = 0.062 < 0.08 and NFI = 0.813 (Table 3). Therefore, both models demonstrated acceptable fit. To assess model reliability, we used Cronbach's alpha (α), coefficient rho_A and composite reliability (hereafter CR). These three indicators should have a measurement criterion above 0.7 [75,76]. It is relevant to mention that reliability tests should be applied only to latent variables with reflective indicators-hence, food values are not present in Table 4 [77]. Regarding average variance extracted (hereafter AVE), the index should be above 0.5 [75]. Table 4 shows α, rho_A and CR values above 0.7 and AVE values above 0.5. Thus, both models fulfill established criteria, and they are reliable (Table 4). Next, we applied techniques to establish the variables' convergent and discriminant validity. According to Fornell and Larcker (1981) [78], an adequate score for convergent validity is when AVE is fewer than variance and α is above 0.7. According to Bagozzi and Yi (1988) [79], discriminant validity occurs when the AVE square root exceeds the correlations  Tables 5 and 6 show the convergent and discriminant validity of both models. In both tables, numbers in the diagonal are the AVE square root, while numbers outside the diagonal are correlations between constructs.

Hypothesis Testing
To verify the hypotheses, we used path coefficient (β), standard error, t-value and p-value, employing bootstrapping method and a subsample of 10,000. The hypotheses are statistically significant when β is close to −1 or +1; p-value ≤ 0.000 and p ≤ 0.001 mean the results are statistically significant [72]. However, after analyzing both models, we could not confirm support for all the hypotheses.
Regarding the fast-food restaurant model, Table 7 shows the influential variables, such as positive anticipated emotions and attitude toward eating a hamburger. It also establishes that subjective norms are important to purchase prediction. There are three hypotheses with the highest association level: (1) H2a with β = 0.696, t = 17.670 and p ≤ 0.000; (2) H3a with β = 0.655, t = 13.794 and p ≤ 0.000; (3) H4a with β = 0.645, t = 14.345 and p ≤ 0.000. Consequently, this model supported H1a, H2a, H3a, H4a, H9a and H10a, whose t-values ≥ 1.960, p-values ≤ 0.000 and β indicate significant results. In contrast, H5a, H6a, H7a and H8a were rejected due to an insignificant or even negative influence from food values to subjective norms, attitude toward the brand and attitude toward eating a hamburger, and to attitude toward the brand to purchase intention.
Regarding the mobile app model, Table 8 shows that attitude toward the brand and positive anticipated emotions were influential, but with a low association level. As before, subjective norms were important to purchase prediction. There are three hypotheses with the highest association level: (1) H2b with β = 0.662, t = 13.669 and p ≤ 0.000; (2) H1b with β = 0.495, t = 7.376 and p ≤ 0.000; and (3) H3b with β = 0.367, t = 5.097 and p ≤ 0.000. In other words, this model supported H1b, H2b, H3b, H4b, H8b and H10b, whose t-values, ≥, 1960, p-values ≤ 0.050 and β indicate significant results. In contrast, H5b, H6b, H7b and H9b were rejected due to an insignificant or even negative influence from food values to subjective norms, attitude toward the brand and attitude toward eating a hamburger, and to attitude toward eating to purchase intention.  R 2 was calculated to assess the structural model; the higher the value, the better the constructs explain the model. Recommended R 2 values may start from above 0.10, 0.75, 0.50 or 0.25, equivalent to typical, substantial, moderate or weak, respectively [73,80]. According to Hair et al. (2019) [81], another measurement option besides R 2 is predictive relevance (Q 2 ). Recommended Q 2 values are above 0.02, 0.15 or 0.35, which means the model has a weak, moderate or significant predictive relevance, respectively.
Regarding the fast-food restaurant model, Table 9 shows the R 2 attitude toward the brand = 0.473 (weak-moderate) and R 2 attitude toward eating a hamburger = 0.585 (moderate), meaning that positive anticipated emotions explain attitudes. R 2 purchase intention = 0.681 (moderate-substantial), meaning that attitude toward the brand, attitude toward eating a hamburger, positive anticipated emotions and subjective norms explain purchase intention in 68%. R 2 subjective norms = 0.520 (moderate), meaning that positive anticipated emotions explain subjective norms in 52%. As for Q 2 , the four results support that the model has significant predictive relevance. Regarding the mobile app model, Table 9 shows values that are lower than the other model, yet still meet both criteria. R 2 attitude toward the brand = 0.236 (weak) and R 2 attitude toward eating a hamburger = 0.521 (moderate-weak), meaning that positive anticipated emotions explain attitudes in 23% and 53%, respectively. R 2 purchase intention = 0.310 (weak), meaning that attitude toward the brand, attitude toward eating a hamburger, positive anticipated emotions and subjective norms explain purchase intention in 31%. R 2 subjective norms = 0.174 (weak), meaning that positive anticipated emotions and food values explain subjective norms in 18%. As for Q 2 , the four results support the correct operation of these models. Attitude toward the brand had a moderate yet significant relevance, while attitude toward eating had a marginal but still significant predictive relevance.
Traditional reliability and validity assessments do not apply to formative variables, as they do not have to match each other, but food values theory should support the construct as formative. Additionally, the PLS algorithm exposed loadings for reflective indicators and weights for formative indicators.

Discussion and Conclusions
The main objective of this research was to compare consumer purchase intention behavior by differentiating the channel used for consuming hamburgers from a fast-food restaurant. To this end, we analyzed the differences between two purchase contexts: an in-store channel (traditional or model A) versus a mobile delivery app (modern or model B). For this study, we assessed variables such as positive anticipated emotions, food values, attitude toward the brand, attitude toward eating a hamburger and subjective norms.
For model A, we observed that positive anticipated emotions had a strong effect on attitudes, subjective norms and purchase intention. This resulted in a top view of the model where positive emotions are evoked prior to determining purchase intention, both directly and indirectly (e.g., through attitudes and subjective norms). Managers need to analyze the emotional typology and take actions that arouse those emotions at the point of sale. Consumer communications, point-of-sale locations, prices and products should emphasize the most relevant emotions in order to drive consumer behavior. On the other hand, the attitude toward eating a hamburger represents a very efficient relationship toward the intention, i.e., the utilitarian aspect of consumption is reinforced in this relationship. This can make a strategic synergy, raising the hedonic (emotional) versus values (utilitarian) aspect. Likewise, it is proven that although the relationship between subjective norms and purchase intention has a positive aspect, it does not form a strong relationship; thus, it must be taken into consideration, but not in the first stage.
In model B, the affective emotional component is observed as the main one. We also found a robust relationship between positive anticipated emotions, attitude toward eating a hamburger and attitude toward the brand. In this model, the attitude toward the brand has a notable impact on purchase intention. Additionally, subjective norms and positive anticipated emotions exert a medium-strength influence on purchase intention. In conclusion, this model provides some initial insights, but future work is needed to establish more conclusive evidence.
Both models constitute a method for explaining purchase intention through some TPB variables as well as food values. In both schemes, we established the transcendental role of positive anticipatory emotions in explaining the purchase intention of fast-food consumers. Likewise, we observed that food values did not perform remarkably in the models; thus, there would be value in exploring other adjacent variables that have more explanatory power. The traditional model demonstrates significant relevance of 68.1% versus 31% for the mobile application one. In other words, the first model features some solid directions, while the second model lacks an optimal level of explanation. There may be other variables that drive food purchases when using an app. For instance, whether the app is directly tied to the restaurant or instead to a third party may play a significant role. In addition, it makes a significant difference between the brand of the app and that of the restaurant.
The fast-food industry needs to invest in more research to understand consumer behavior and values. Managers need more in-depth knowledge about using some channels and the consequences of the purchase intention. More and more consumers are growing closer to the opinions of different external actors, which would be interesting to investigate who they are, for making strategies that possibly impact subjective norm and consumer purchase intention. In order to endure, fast-food companies need to invest more in food values, positive anticipated emotions, attitude toward eating, attitude toward the brand and subjective norms.

Limitations and Future Research
The first limitation of this research is that we used an online survey rather than a face-to-face survey, which is recommended for ensuring a representative sample of participants. Future work should consider population differences and select individuals by proportional allocation to obtain a representative sample by age, gender, socioeconomic status, schooling and region of Puebla State or other states in Mexico. In this way, scholars can classify opinions according to demographic profiles and, more broadly, perform cross-cultural comparisons.
Furthermore, future research could increase the reliability level by performing random sampling and using a larger sample size, as well as analyzing each factor separately. Research should also consider other brands, food types and restaurants, such as full-service restaurants.
Likewise, we recommend adding other TPB-related variables, including predecessor variables to food purchase intention, such as word-of-mouth recommendation, loyalty or satisfaction. Additionally, scholars should more deeply evaluate the importance and relevance that consumers assign to nutritional food values and food safety issues.
Finally, there would be value in performing a longitudinal study in order to chart the evolution of people's priorities during different periods of the pandemic or in response to a different health crisis.