Influence of value determinants on customers` preferences: to order food online of to go out

As for the method of data analysis, this study uses logistic regression analysis. Logistic regression is used to obtain odds ratio in the presence of more than one explanatory variable. The procedure is quite similar to multiple linear regression, with the exception that the response variable is binomial. Linear regression analysis is very often performed in such studies (Blanchflower & Oswald, 2011), but linear regression does not show the dependence of binary variables on ordinal variables directly, in addition, a dynamic series is not used (1 respondent does not respond daily for a whole year, but 400 respondents at a time). That is, there is no" line" as such, and the probability of one of the events - an online order or eating out-is considered regression, the logistics system is suitable for this. In the case of this study, the dependent variable (human choice) takes two values (make an order or go out to eat), and the task is to find coefficients that would show the direction of its relationship with the independent ones (determinants: freedom, happiness, self-respect), among the well-known machine learning models, logistic regression is the best approach. For example, the decision tree will not give coefficients, but will simply predict the value of a variable based on input data. Linear regression is not suitable because the dependent regression must be continuous.

Consequently, this study will find out whether there is a relationship between value determinants and people's preferences and will also help to develop tips for the restaurant business, based on the obtained results.



Methodology

Research objectives

To start methodology with, the authors represent 2 main aims of the research:

Analyze the influence of key values in social psychology, identified and described by Wagner and Mazzon (1991), on preferences of customers in the food industry, more specifically - to order online through a range of special devices or eat out;

Based on the results of the analysis, provide practically implementing tips for the competitiveness of the restaurant business from the point of view of management and marketing;

Justification and development of the hypotheses and specification of the relationship between these hypotheses and the theoretical foundation of the thesis are placed in the Research Question Statement in order to include in methodology more information about data, data collection and analysis.

Speaking about the goals of the study, the authors aim to identify new trends in the restaurant business, which allow look at the choice of the buyer from a new position. As stated in the Research Question Statement section, if the restaurant team wants to change the company's activities, the employees responsible for this are most likely to check basic points:

quality of dishes;

appearance of a cafe/restaurant or website, talking about delivery;

level of service of consultants, managers, waiters or food deliverers;

service support;

waiting time of the order.

All these basic points were voiced at the National Restaurant Association in 2015 and called outdated (the most innovative and large-scale exhibition in the world about the restaurant business, where all companies show their technological innovations for the first time). At one of the expert sessions of the NRA show, the following results of research on the restaurant business in the world were announced:

75% of visitors expect an individual approach

71% of restaurant visitors are interested in ordering takeaway food

47% of people expect that they can make a pre-order over the phone

78% of Millennials search for restaurant menus online

This statistic is almost fully available on their website, the link is placed in the reference list.

All this suggests that it is not enough to simply meet the standard criteria of a restaurants' performance. Although the NRA primarily demonstrates technological trends in the restaurant business, their research, as well as value determinants researches in other areas, has become the basis of this study. As already mentioned, this research objective is the finding and interpretation of the new alternatives to the usual approach of attracting customers, deeper understanding of reasons for their choice.

Research design

The research design of this study is a combination of explanatory and descriptive style, as the authors are aimed both describe key existing trends in customer behavior, identify new ones, explain them / provide practically useful and implementing advices for restaurant sphere how to improve their activities in order to maintain regular customers and attract new customers.

Five types of research designs can be defined (Bryman & Bell, 2007):

1. Experimental design, which is unusual in management research due to the challenge of accomplishing exact control levels when dealing with organizational behavior;

2. Cross-sectional design or social survey design, which involves data collection on several cases during the same time frame in order to collect qualitative or quantitative data related to two or more variables, in an effort to determine associations between the variables after the data has been analyzed;

3. Longitudinal design is specifically used to monitor changes over a period of time in the relevant research environment;

4. Case study design is an intensive examination of a particular situation or instance;

5. Comparative design is where identical or contrasting cases are studied, and the similarities or differences are reported.

Based on the fact that the authors: collect data only once (collecting data to see the changes over a period of time does not make a sense for the research question of this study), have limited resources, collect the quantitative data related to 1 dependent and 3 independent variables (quantitative data, two or more variables - characteristics of cross-sectional research design), this study applies has definitely cross-sectional research design.

Data collection

As for data collection methods, they are divided into qualitative and quantitative. Qualitative methods include focus groups, in-depth interview, projective techniques and ethnography. Quantitative methods include surveys/questionnaire (face-to-face, by mail, by telephone, online) (Noyes, Booth, Moore, 2019). This study examines the behavior and motives of buyers, and therefore requires a database of quantitative responses from people of different ages, gender, and occupation, so the best way to collect data is to use a survey. Moreover, the authors aim to study the internal motives for customers' choice, which requires data collection that allows capture a wide audience to look at trends in general. Interviews and focus groups are not suitable for this, since they do not allow define the motives of the majority. The advantages of the survey in case of this research are the possibility quickly collect data, maintain anonymity (if necessary), and interview people from different social groups. Interviews, focus groups are aimed at obtaining more narrowly focused and specific data, such a performance of the particular sphere or a company (Rubin & Babbie, 2011). Thus, a quantitative method of data collection is used. Quantitative research - a research methodology that seeks to quantify the data and, typically, applies some form of statistical analysis (Malhotra, 2012).

To find and analyze research papers and books on the basis of which the survey was created for respondents, authors used:

ProQuest Dissertations & Theses, which includes more than 2 million doctoral and master's theses;

Science Direct, which contains more than 1,700 journals published by Elsevier in all thematic areas;

the international analytical search platform Web is Science; the search engine for scientific texts Google Scholar.

The main interest was represented by studies describing the influence of psychological and sociological factors on people's behavior, as well as their choice in relation to something.

The process of forming questions for the survey is based on a survey from an article (Blanchflower & Oswald, 2011) that analyzed international happiness. The word “happiness” here means various characteristics, which include -marital status, self-perception of happiness, as well as personal characteristics, such as self-satisfaction (one of the possible synonyms for self-esteem). The survey has the following questions «Would you say that you are very happy, pretty happy, not too happy?» and «Would you say that you are very satisfied with your life, moderately satisfied, a little dissatisfied, or very dissatisfied with what you do?”. The survey from the article also contains questions about age, gender, and marital status. These questions are also included in this study, as they are necessary for the representativeness of the sample, and a question that allows choose between two alternatives - eating out, or ordering at home, which is the main topic of this study.

Thus, the survey contains the following 5 questions: gender, age (age intervals were selected according to the intervals in the Federal Statistical Service, where data on the population of Saint Petersburg was taken from), questions about self-perception of happiness and self-esteem, and the question of choosing from two alternatives (eat our or order online). The survey is located in the Appendixes 1, 2 and 3. A distinctive feature of this survey from which it was based on, is the use of Likert scale for questions about happiness and self-esteem. This scale contains 5 answers instead of 4, which allows respondents to choose the option "I find it difficult to answer the question", which allows them to avoid forced dishonest answers to questions.

The survey was compiled in the Google forms format, which allows quickly and easily create and edit surveys, track the dynamics of responses online, and download responses in the desired format. Since multiple people can respond to a survey at the same time, Google forms is the most appropriate tool for collecting a large number of responses. Other tools for creating surveys, such as SurveyMonkey, Survio, and Examinare, are paid, which is not justified, since Google forms provides all the same options for free. The survey was sent out both online and in print, in order not to lose respondents who do not actively use the Internet, and therefore do not make orders online, as well as respondents who do not use the Internet at all. The survey was sent to various social networks-VK, Instagram, WhatsApp, Viber (for the more Mature age category). The printed versions of the survey were offered to people near restaurants and cafes.

Sample and its representativeness

The authors are only interested in the population of Saint Petersburg, for justified reasons represented in the chapter “Literature review”. The subjects of the Leningrad region do not participate in the study, since the delivery service, which defines one out of two core directions of the research, does not extend to many territories outside the city.

People whose ages range from 15 to 49 are selected from the entire population as, according to the Federal state statistics service, they mostly constitute a category of economically active population in this city. The economically active population - is a population that has the opportunity to participate in the production process of material values, and also in the provision of necessary services. Belong to this category of citizens: entrepreneurs, employees in employment, and people who currently do not have any work, but want to get it (Dordzhieva, Berikova & Badmahalgaev, 2018). The economically active population is included in the UN System of national accounts.

The methodology of the International Labour Organization defines the economically active population as people from 15 to 72 years of age. Nevertheless, the authors of the study rely on the specifics of the population in a particular country, city, as well as the specifics of the study to provide relevant and practical results. It is assumed and proved by that people over the age of 49 rarely use or do not use at all devices for ordering food in Saint Petersburg , so they are not of interest in the context of the research.

The population data is taken from the Federal state statistics service - the population of Saint Petersburg on January 1, 2019 is 5 383 890 people, of which 2 621 204 are people aged from 15 to 49, and they make up the general population of this study. The Table 1 shows the distribution by gender to determine the number of men and women.

Table 2

Population of Saint Petersburg, January 1, 2019*

15-19	20-24	25-29	30-34	35-39	40-44	40-49
male	female	male	female	male	female	male	female	male	female	male	female	male	female
106892	106432	119899	125482	212680	210494	256748	265294	225621	234844	189230	203105	174658	189825

*Source: Federal state statistics service

Assuming a general population of 2 621 204 for 100%, the percentage of men and women in each age category (significant activity for the future research - to reflect the characteristics of the general population for the sample) is considered. The results calculated by the authors in Microsoft Excel are shown in Table 2.

Table 3

Percentage of men and women in each age categories, January 1, 2019*

15-19	20-24	25-29	30-34	35-39	40-44	40-49**
male	female	male	female	male	female	male	female	male	female	male	female	male	female
4,1%	4,1%	4,6%	4,8%	8,1%	8,0%	9,8%	10,1%	8,6%	9,0%	7,2%	7,7%	6,7%	7,2%

*Source: Federal state statistics service

** Estimated values

The composition of the elements of the general population (Table 2) is made to understand the representativeness of the sample. Sample is a part of the objects from the general population selected for study in order to make a conclusion about the entire general population (Churchill & Brown, 2007). All samples can be divided into two types (Yadov, 1998):

probabilistic or random;

non-probabilistic, or deterministic.

When forming a deterministic sample, the probability of any element of the general population being included in it is unknown, and it is impossible to answer control questions that require statistical generalizations for the entire general population.

In comparison, when forming a random sample, the probability of any element of the population being included in it is easily determined and is the same for all elements. From this definition, it can concluded that only such a sample allows researchers to conduct a study whose results with a certain degree of confidence reflect the characteristics of the entire population, what is most important for the authors of the study.

Random sampling is usually associated with the collection of data based on a survey, when information about the elements of the sample is made conclusions about the entire population, which, in turn, are the answers to the control questions of the study.

The random sample is formed in 4 stages (Gillard, 2020):

Selection of the sample based on the formulating of research goals;

Determining the optimal sample size;

The choice of method of sampling;

Checking the representativeness of the sample.

1 stage - sample base

At the beginning of this chapter, the authors outlined two main goals of this study: Firstly, analyzing the impact of value determinants of customers on their preferences - eat out or order online; the choice of such alternatives was due to the reasons specified in the chapter “Statement of the research question” (the emergence of new competing forms of interaction with customers in the food industry; new competing methods of organizing restaurants to reach new segments of customers; creating a competition between restaurants with delivery, regular restaurants, and restaurants of the new Dark Kitchen concept, which implies only delivery). Secondly, creating the useful tips for improving the company's performance from the point of view of marketing and management.

The sample base is a list of elements of the General population that is used for forming the sample (Bouchon-Meunier, Yager & Zadeh, 2000). In most studies, the sample is based on reference books, official lists of participants in something, registers (Rubin & Babbie, 2011). In recent years, the number of organizations that specialize in selling lists with the names and addresses of residents of specific regions, based on which researchers conduct surveys, has increased. They are usually compiled in an electronic format that is convenient for working with databases, and the cost of lists is proportional to the number of people in it. However, this information may be outdated, for example, if data was taken from a telephone directory that was updated more than a year ago. The authors of this study decided to create on their own a sample base, which is acceptable for research. The authors of the study also made sure that the compiled database does not bear signs of bias and objectively reflects the general population, is not outdated and sufficiently represents the general population, i.e. the data is appropriate for the representativeness of the sample.

The database of potential participants of the study was created in the following way, randomly chosen people who are:

residents of Saint Petersburg, for reasons justified in the introduction section;

men and women of different ages and genders, according to the percentage of the total population of the city;

people who equally have and do not have access to Internet resources for ordering online;

studying and not studying in schools, universities, institutes, colleges

working/non-working people.

2 stage - optimal sample size

The selection of the sample size is determined by the following factors (Thietart, 2011):

Specified degree of confidence that the data collected from the sample reflects the characteristics of the entire population (confidence level);

Necessary accuracy of estimates made based on the generated sample (acceptable margin of error);

Intended type of analysis of the collected data;

Size of the general population from which the researcher forms a sample.

Before starting to determine the sample size, the researchers decide what the degree of confidence of information about the general population should be. The parameter that determines this degree of confidence, researchers call the confidence level and usually set it at 95%. This means that out of hundreds of samples formed by the chosen method, in 95 cases the sample adequately represents the characteristics of the general population. As for optimal margin of error, most researchers believe that the optimal margin of error is within 3-5% (Rubin & Babbie, 2011). The margin of error is also selected by the researcher.

In this study, the confidence level is 95%, the margin of error is 5%.

As for the intended type of data analysis, the authors form hypotheses in such a way as to reveal the dependence of one variable on another, so regression analysis of data is chosen among other possible methods of analysis. In statistical modeling, regression analysis is a set of statistical procedures for examining dependencies between variables. It includes many methods for modeling and analyzing relationships between a dependent variable and one or more independent variables, also called predictors or regressors (Nwanganga & Chapple, 2020). Regression analysis helps to understand how the "typical" value of a dependent variable changes when one of the independent variables changes. In the case of this study, the dependent variable (human choice) takes two values (make an order or go out to eat), and the task is to find coefficients that would show the direction of its relationship with the independent ones, so among the well-known machine learning models, logistic regression is the best approach in this case. For example, the decision tree will not give coefficients, but will simply predict the value of a variable based on input data. The logistic regression model is based on the following principle: in contrast to the linear regression model, it predicts not the value of the variable itself, but the probability that the variable takes a value that makes it possible to understand which category the object under study belongs to. Logistic regression is used when the response variable is categorical. For example, yes / no, true / false. In this study, there are two alternatives: ordering online / going to a bar, cafe, or restaurant (Nwanganga & Chapple, 2020).

Returning to the sample forming, it is based on the given confidence level and measurement error in order to obtain a number (sample size) that can reflect the population if the sample correctly conveys its characteristics. The calculation in this case can be carried out according to the formula:

Where:

n - sample size;

z - normalized deviation, determined based on the selected confidence level. This indicator describes the possibility and probability of responses falling into a special confidence interval. In practice, the confidence level is often assumed to be 95% or 99%. Then the z values will be 1.96 and 2.58, respectively;

p - variation for the sample, in fractions. In fact, p is the probability that respondents will choose a particular answer. As the authors mainly have only 2 alternatives - ordering online and eating out, p is considered to be 50%, i.e. 0.5;

q = (1 - p);

e - acceptable error;

It should be noted that there are many sites on the Internet for calculating the optimal sample size based on the specified parameters of the confidence level and the margin of error, some of them are listed below:

https://allcalc.ru/node/100

https://calculatorium.ru/

https://fdfgroup.ru/

The final calculations were the same on the site and according to the formula - 384 people from the general population are the optimal sample for the study.

3 stage - method of sampling

After selecting the necessary sample base and determining its size, you need to decide which method of forming it is the most optimal for your research. There are 5 types of random sampling:

simple;

systematic;

stratified;

clustered;

multistage;

The type of sample is determined primarily by the control questions and goals of the research project, as well as the need for statistical analysis of the data obtained. The choice will also be determined by factors such as the need for personal contact with the study participants, the geographical distribution of participants, and the nature of the sample base. The selection of the sample type depends on whether the data is collected with the help of assistants, as well as on the simplicity of the data collection procedure (Rubin & Babbie, 2011). This study uses a simple random sampling method. Simple random sample (SRS) is a probabilistic sampling method, according to which each element of the general population has a known and equal probability of selection. Each element is selected independently of each other element, and the selection is formed by randomly selecting elements from the selection base, as it correlates with this study's aim. Other sampling methods (Rubin & Babbie, 2011) are excluded for the following reasons:

Systematic random sampling involves the selection of elements at a certain interval, which is not necessary for this study;

Stratified random sample is a type of random sample, in which the general population is divided into two or more layers for some reason, i.e. the sample base is divided into subsets, which is not typical for this study.

Cluster random sample is very similar to a stratified sample, since the general population must be divided into separate clusters This study does not suggest division into clusters.

Multistage sampling is used to overcome the problem caused by heterogeneous distribution of the population, as well as in cases where a wide geographical region is being studied. The value determinants' influence research involves only Saint Petersburg, where population is distributed fairly homogeneously.

4 stage - representativeness of the sample

A representative sample is a sample that contains all the subgroups that are important for the study. For each subgroup, shares with different characteristics of the general population should be provided proportionally. The sample is representative in this study, and the Sample section provides a detailed description. The characteristics of gender, age, type of employment, as well as random selection of respondents in the sample were observed.

Data analysis

After the data has been received (400 responses instead of 384 for ease of analysis), the data is analyzed to identify the relationship between the dependent variable (the choice between ordering online and eating out), and independent value determinants selected and justified by the authors (happiness, self-esteem, marital status). Data analysis (logistic regression analysis, the choice of the type of analysis is described in the section Research Question Statement) for the authors of the study presented a choice of several alternatives: SPSS, MS Excel, R, Jupiter (Python programming language). MS Excel has limited features without an additional tool package. It is possible to provide the logistic regression analysis using Kamakura'a Analytuc Tools. SPSS also allows make the logistic regression analysis, which is quite easy to perform in a few clicks. Jupiter is also a convenient tool, but it requires certain skills to work with the Python programming language. In such type of study (research format, relatively extensive database, relationship between variables, bachelor thesis), RStudio widуspread in use and the authors are familiar with it programming language. The data analysis of this research is provided through RStudio. The significant step is the conversion of variables to binary form before the analysis as it is required in logistic regression. (assign values 0 and 1). The answers before the conversion are placed in Appendix 5.



Logit model

The positive class in the built model is 1 (ordering food online). As it can be seen, Age's coefficient is statistically significant on 5% confidence level, gender is not even on 10%, relationship status, happiness and esteem level are significant on 1% confidence level. The lower the p-value, the more significant the test statistics are, that is why coefficients that relate to marital status, happiness and self-esteem. The coefficients suggest that having higher happiness level, higher self-esteem and being single increase the respondents' probability of eating out. Getting older decreases probability of a respondent of going to a restaurant, bar, cafe.

Average Marginal Effects (AME)

For continuous independent variables in logit model average marginal effects show the probability of being attributed to a positive class will change on average with a small change of the regressor.

For categorical variables it shows how changing the category relative to the base one (in this case categories are 0 or 1 hence the base category is 0) will affect the probability of belonging to a positive class of the dependent variable. The results suggest that having higher self-esteem decreases the probability of a person ordering food by 21.62% on average; higher happiness decreases this probability by 28.42% on average and living single-by 26.88% on average. Getting older increases this probability of making an order online by 0.52% on average, and this fact can be interesting for the future research.

Based on this logistic regression analysis, hypotheses 1, 2 and 3 were identified in the study are confirmed, which is a significant result for the second task of the study - providing practical tips for the restaurant business based on hypotheses. The hypothesis 0 was rejected as the denying of other hypotheses that were accepted.

Descriptive statistics

The most important descriptions of the variables are: sample size is 400 observations, 203 of which prefer eating out, 195 are male, 233 are not single, 146 are unhappy and 201 have low self-esteem. Therefore, among all categories, the proportions of both classes are roughly the same, which makes results more stable and reliable as opposed to having overpopulation of one class over the other. As for the age of the respondents, 75% are older than 27 and median age is 32, which means that results are valid for the middle-aged people.

An interesting fact is that a correlation of 0.60361 was found between the regressors happiness and self-esteem, and in this case we can talk about the possible multicollinearity of the regressors (Table 3). This can be interpreted as follows: low happiness is combined with low self-esteem, and high happiness is more common in people with high self-esteem. In any case, it may also be important facts for future research in this field.

Correlation between collected primary data*

Table 3

Correlation	y	x1	x2	x3	x4	x5 *
y	1
х1	-0,0897	1
х2	0,09342	0,01372	1
х3	-0,4791	-0,006	0,05939	1
х4	-0,5262	0,0851	-0,0282	0,37219	1
х5	-0,5	0,03801	0,05041	0,43687	0,60361	1

*Gathered by authors

**Estimated values

Where:

y - food choice;

x1 - gender;

x2 - age;

x3 - marital status;

x4 - happiness;

x5 - self-esteem;

To summarize, this study confirmed its 1, 2, 3 hypotheses that such value variables as happiness, self-esteem and marital status influence the preferences of customers in the restaurant business and rejected the hypothesis 0. The study identified two additional facts that can be useful for research in the restaurant business, or other spheres - the older people are, the less they are likely to choose to eat out; there is a fairly high correlation between self-esteem and happiness, which can be checked and explained.

Description of results

Psychology and marketing are closely connected and cannot be split. Psychologists have always been deeply interested in people`s emotions and how they affect choices that people make. Many papers on human values have been conducted, however no researches on finding out how happiness, self-esteem and marital status affect human choices in term of choosing whether to eat out or order food online have been conducted. This is the reason for the authors to proclaim that a research gap in this area exists. Stuppy et al. (2020) wrote about emotions affecting buying behavior. Researchers stated that attractive products make customers feel good about themselves. Moreover, Yrjцlд et al ( 2019) also wrote about e-commerce for online restaurants and commerce for offline restaurants. They deeply discussed the importance of customers` emotions for getting satisfaction and stimulate repurchase behavior. Nevertheless, in previous researches other values have been tested and discussed; among them can be named, food quality, dishes appearance, service quality, portions size, restaurant design and etc. These values, for sure, are highly valued by customers; however, this paper is aimed to reveal other values that are “created” in humans` head. Restaurant employees cannot create happiness, marital status, or level of self-esteem. These values can be created only by the human itself.

Discovering the ways, they affect customers` choice is attractive and valuable for business owners as it can provide a guideline on how to improve its business to satisfy customers better and gain bigger profits. Nevertheless, Rani, (2014) discussed the importance of family and personal factors, which seems to be quite similar to this paper, yet the author described mostly demographic variables like economic circumstances, lifecycle stage, occupation etc. It proves that this paper provides unique research on human values that have not been discussed in term of customers` preferences to food ordering or offline restaurant visiting.

In accordance with the research conducted, it can be proclaimed that all three hypotheses were confirmed. In previous sections the authors supposed what the results could be. Formulating hypotheses, the authors conjectured that all the hypotheses would be confirmed. It is worth mentioning that authors` surmise was not baseless. Based on psychological literature an assumption for every hypothesis was created. Getting back to hypotheses of the paper they are the following:

H1: Single people eat out more often than people who have a family. Determinant: freedom (independence, free choice)

H2: People who consider themselves happy eat out more often than others, who prefer to order food online. (Determinant: happiness)

H3: People with low self-esteem order food online more often than others, who prefer to eat out. Determinant: self-respect (self-esteem)

According to Demo (1985) self-esteem is about inner feeling of a person. This is how we evaluate ourselves and feel inside and justify ourselves. The higher our self-esteem is, the more confident we are. This is a significant driver for a thought that high-esteemed people are more likely to go out than low-esteemed. Psychologists also claim about self-serving people, people who focus attention on themselves and crave it. It is believed that self-serving people always have high self-esteem and crave people around them (Greenberg, 2008). They invariably need to be surrounded with society and have offline communicate. Highly self-esteemed people are open-minded and confident which makes them seek to go out. Eating out in a restaurant or cafй is a perfect way for people with high esteem to fulfill their desire for offline communication and social interaction as being in a cafй or restaurant they need to make a face-to-face order, to face other guests (customers) and be at one place with them. Psychological articles state that people with low esteem are more likely to be introverts and homebodies, they evade from overcrowded places, big companies, and noisy places. The reason for having low self-esteem mostly goes from childhood and academic findings support this point of view. Being a kid, we all are easy driven by external factors and others` opinion. We all are also psychologically weak and cannot fully protect ourselves (Waddell, 2006). Inappropriate words can easily hurt a child and provoke low self-esteem to appear. Children face new challenges every day and each challenge outcome affects its self-esteem (Ojanen & Perry, 2007) and this is where it goes from. Becoming adults low self-esteem does not disappear, these complexes grow up together with children (Halliday, 1999).

As it was supposed the hypothesis was confirmed and proven not only empirically, but also by quantitative methods.

Another hypothesis was dedicated to happiness and its affection on customers` choice. Interest in correlation between happiness and consumers` behavior has been on agenda for many researchers. Creating this paper correlation has been found. Happy people tend to eat out more often than those who feel unhappy, and this can be easily explained. Some researchers say that happiness and self-esteem are closely connected, however not every highly self-esteemed person is happy and vice versa. Another point of view holds that self-esteem and happiness are discriminable terms (Lyubomirsky et al., 2006). Nevertheless, Lyubomirsky et al.,(2006) stated happiness as something that is different for everybody; a lasting sense that makes life meaningful and complete. When we are happy, we are more likely to communicate with others. Researchers also define happiness as something that brings joy to our life. Strack et al. (1991) highlighted social contacts as one out of seven aspects of joy. It was also proposed that social cooperation is an important issue of joy and it brings positive effect. Happiness and extraversion are correlated and happy people experience more social interactions than unhappy(Argyle & Lu, 1990). Moreover, a greater extravert` social activity was found. As it was supposed this hypothesis was also confirmed and approved both empirically and quantitatively.

Third hypothesis was dedicated to marital status and freedom (freedom of choice). When we live alone, we are free to choose what do, what to eat and where to go, but when we live together with a family, we need to take into consideration their opinion. Of course, families differ but no one wants to be neglected while deciding on something. Generally, family meal patterns are about gathering at one table and eating together (Fiese et al., 2006). A family person seeks to be together with his/her family and mealtime is an occasion when everybody can attend. On the contrary, single people are freer to choose, they are not limited by the family. While producing hypotheses for this paper, it was rather difficult to predict the result of the first hypothesis whether it would be confirmed or refuted. Nevertheless, it was confirmed and proven.

Many researches have been conducted dedicated to topics which are related to food marketing and online food delivering services, however this paper differs and provides a research that is supposed to fulfill the existing research gap. Nevertheless, as far as the research was based on various literature there are significant similarities that have also been confirmed qualitatively.

Ladhari et al. (2008) wrote about differences between people in terms of sex, age, lifestyle, personal background, and other factors. It is impossible to deny the clam that has been proved by thousands of psychologists and the authors did not have an aim to do that. During this research lifestyle, personal background and other specific factors were not taken into consideration, due to the fact that aforementioned points do not fit the methodology of this research and cannot be measured via logistic regression. These factors are much more complex and require another measurement system that were not planned to be reflected in this paper.

Value determinants are the central issue of the paper. Thousands of researches on this topic were conducted, yet it is worth mentioning that plenty of diverse values were found. Among them there are quality of food, quality of service, quality of physical environment, economic value, functional value, emotional value, word-of-mouth, symbolic value, satisfaction, repurchase intention suggested by (Yrjцlд et al., 2019); excellence and satisfaction, emotional payoffs, time, and perceived service represented by Jensen & Hansen ( 2007) are also of high importance, yet again do not fit frameworks of our paper. Choosing values was a critical point for authors as while choosing it is essential to pay attention on how chosen values are going to be measured. Aforementioned values seemed to be too general and impossible to be measured in terms of this paper. Value determinants that are represented in this paper are closely related to emotional payoffs as Jensen & Hansen (2007) defined “emotional payoffs” as overall consumers` values related to individual emotions and feelings. Happiness and self-esteem, that are highlighted in this paper are also related to individual emotions and feelings, as self-esteem is the way how we feel and evaluate ourselves inside, and happiness is the way we generally feel about life. And here comes the significant difference among aforementioned authors and this paper: this paper depicts more precise and focused value determinants than aforementioned researchers. Nevertheless, these researches are similar in a way that explain how particular values affect human`s choice.

What is choice? An action that we make based on decision making process. It includes selecting one or several options. The represented research explains how value determinants affect choice among eating out and food ordering. Auty (1992) wrote about consumers` choice and its drivers. She described offline restaurants, cafйs, bars etc. She also talked about factors that drive consumers to choose between these options. The crucial point that distinguishes Auty`s research from this is that she did not mention online food delivery and whole research was dedicated to offline restaurants only. This paper includes not only offline food venues` description, but also a research component dedicated to online food delivery services. Another distinctive feature of this paper is that authors do not differentiate bars, restaurants, cafйs, authors claim about them in general as about offline food, the same thing happens with online food delivery services- they are not distinguished by types, they are discussed generally. Basically, two options are compared: online food ordering ang going out.

In literature review rather big passage was dedicated to innovations and its influence on business. Even though online food delivery itself is not an innovation, as big venues have been delivering food for quite a long period, there is one innovation that should be mentioned. It is called “dark kitchen”, or it is also known as “cloud” or “ghost” kitchen. This business model is a so-called subfield of online food delivery, that does not represent a physical restaurant and it possible only to order food online. In our research we did not distinguish dark kitchens and usual online food deliveries since the core aim was to focus on value determinants and its influence on choosing between two options only. The reason why dark kitchen was not represented as a separate option is that dark kitchen itself is new concept and not that widely known among people, which means that most respondents might have not even heard about it.

As far as this paper is based on values` classification suggested by Kamakura et. al. 1991 quite a lot is supposed to be taken from that research, yet Kamakura`s paper and this differ in many ways. He used RVS measuring while here logistic regression is represented. Furthermore, nineteen values were mentioned in the article. In the modern world we all are extremely busy and hardly can we find time to cook- this is what Saad Andaleeb & Conway (2006) proclaimed and it is hard to overestimate the meaning of these words. They found out that meal patterns are different for different people according to level of their busyness. Working women have less time to cook- this is what they said. And this is true, the more we work, the less time we have. This article was published before online food delivery spread worldwide and became available for nearly everyone, which makes this research limited in some ways. On the contrary, this paper contain comparison between OFD and offline venues. In addition, hypotheses tested in that paper are related to reliability of the venue and food quality. As for methods used to test the hypotheses authors used secondary sources and a survey just like authors of this study did. Overall, questionnaire is widely used in similar researches as it is considered to be one of the most convenient way of gathering information when a lot of respondents are needed.

Practical implementation

It is worth mentioning that this paper provides not only theoretical contribution, but also practical value. As far as public catering and online food delivery are fast-developing and promising fields, they require recommendations for even better development, which this paper provides. Pursuant to the research conducted the following recommendations can be stated:

As far as the authors discovered correlation between self-esteem, happiness, and extraversion it means that extraverted people are more likely to go out in general (Halliday, 1999). According to the results it can be stated that the unhappy and lowly self-esteemed (and as a consequence mostly introverted) people tend to be home sitters it can be suggested for public caterings to organize more private areas inside its venue. It can be remote tables or tables for one person only, cozy corners or small tables for one guest. This implementation will definitely be approved by those who prefer privacy.

Another recommendation can be to install automatic ordering and payment systems so that the customer would be able make an order privately and pay it by the same way. This innovation can also stimulate people who doesn`t feel happy and confident enough to attend venues as they will feel good and calm being in the cafй or restaurant. These payment systems are widely used in McDonalds venues and “Bushe”.

One more recommendation that can be implemented for online services is to allow more special offers for companies. As it was discovered happy and highly self-esteemed people are generally extraverts which indicates they have many friends and they are more likely to spend time with them. Being able to get discount for a group order will definitely encourage them to order online and the company will be able to gain bigger profits.

A recommendation that can also be applied in offline venues to encourage even more happy people to attend it is to pay more attention on social networks (VK/ Instagram) of the venue. The core feature that can help to increase amount of people attending the venue is interactive posts and engaging visuals with the image of happy people. It will help to attract the target audience and increase profits

Another implementation is to organize events that include social interaction. It can be workshops, lectures, small concerts and etc. These kinds of events is supposed to attract happy people who are ready to communicate with others and share knowledges.

These recommendations can be implemented for any type of venue, no matter if it is a fancy restaurant or a middle-priced cafй. Of course, aforementioned recommendations should somehow be altered according to the type of venue, but generally the ideas can stay the same. The research team believe that venues of high price category, which position itself as a fancy place might take into consideration advice about more isolated tables and cozy corners. Sometimes it feels uncomfortable to always be in sight, that is why some customers who feels less self-confident and happy may avoid such places. However, if there would be a possibility to seclude these customers would be encouraged to visit a restaurant like this. The same tip can work out with middle-priced venues (restaurant or cafйs).

However, it is rather difficult to provide a remote or isolated place in a bar, the authors still believe this tip can also be somehow modified for bars too.



Conclusion

Discussion of main results

Results have shown that there is significant difference between people`s personal background and preferences. A strong correlation between living alone or with a family, feeling happy or not, having low or high self-esteem and ordering food online or eating out has been found.

Living with a family limits you in some ways, no matter how many people there are in a family. If a person lives alone, he or she is willing to choose where to go, what to eat etc., however, if a person lives with somebody ( wife/husband, children, parents, grandparents or all together), they limit his/ her choice and other family members` opinion must be taken into consideration.

Happy people are more likely to go out- this is what has been discovered Happiness is about how we feel inside about the life in general. According to the research it was uncovered that people who feel happy usually are more likely to seek for social contacts and social interactions. As a following, they are also more likely to eat out on cafйs, restaurants of any other offline venues.

Self-esteem affects human`s behavior in many ways, however, it also strongly affects our choices not only in interpersonal issues, but in food issues too. The higher human`s self-esteem is the more likely human is to go out and visit restaurants and cafйs and vise versa. The lower self-esteem is, the more human is likely to stay at home and order food online.

Limitations

Even though the authors carefully thought about factors that can possibly limit the study, some limitations still exist. One of those limitations can be people who does not use internet and basically do not order food. As it was mentioned before, printed versions of the survey were distributed to respondents, however, due to the lack of information about these people it was impossible to figure out whether they use internet or not.

Another significant limitation was that the area of study was limited by one city only. The reason for choosing particularly Saint Petersburg was deeply discussed and justified in literature review, nevertheless, answers of limited group of people was studied, which can be assumed as a limitation.

Another crucial point is the size of the sample. It cannot be said that the sample was large enough to consider it to be fully representative. The number of people using online food delivery services and eating out regularly in much bigger, it means that not all possible respondents were included into the research and this issue also limited the research.

One more critical issue is that respondents who took part into the research and answered the questions might have probably been dishonest. Even though the survey was absolutely confidential so that respondents can be honest and feel free to answer in any way they feel, there is a possibility that still there might be some incorrect answers. Human`s psychology is one of the most complex and controversial science. Not all human`s behavioral patterns have been fully studied, some of them are still uncovered and mysterious for scientists. That is why this paper can also be limited to present one hundred per cent representative study due to the factors that cannot be controlled by the authors.

Future researches

Aforementioned limitations provided a strong basis for future researchers to elaborate more on this topic and investigate it deeper. As it was stated before, psychology is complex and controversial, and authors recommend adding more questions that will help to reveal respondents` identity. Future research may also contain precise interviews that would be aimed to figure out more personal details about the respondent.

Future researches can also be more precise in terms of values. Authors also stated that many classifications of values exist, and it is also possible to include other values that have not been included in this research. Those values can be more global or more precise. Global values can be security, freedom/ independence etc. Among more focused personal values social recognition, pleasure, comfortable life etc. Values that relate to venues` service can also be tested, these values can be food quality, appearance of dishes, quality of physical environment (design), noise, service quality etc. It would also be great to compare aforementioned values with demographic data as age, sex and income.

...