2. Consumer Behavior: Influencing Factors
Consumer behavior is dynamic, registering a continuous evolution. The determining motivations that lead to the satisfaction of needs will be continuously pursued, depending on the level, importance and possibilities of the individual [
1]. Consumer behavior, in a specific approach, can be defined as a multidimensional concept par excellence, as a result of a system of dynamic relationships among the processes of perception, information, attitude, motivation and practical manifestation, which characterizes the integration of the individual or group in space, as described by all consumer goods and services existing in society, at a given time, by individual or group decision-making acts regarding them [
2]. According to Roering and Block [
3], one of the reasons why consumer behavior is such an academically rich field is that it continues to be based on the work of other disciplines. Examples of such domains are social psychology, economics, political science, statistics, and philosophy.
Certain authors [
4] define and attribute to marketing the property of meeting the needs of the consumer, which means that the basis of marketing is always the consumer, and studying and understanding their behavior is the key to the success of any marketing strategy.
The role of consumers is also defined by [
5], identifying them as the ones who determine the number of sales and the profit of a company through their purchasing decisions, so that their shares establish the economic viability of a company. When analyzing consumer behavior, it is essential to consider both attitudes as well as their manifested behavior. The notion of perspective can be defined as being an individual, distinct aspect of each individual. Psychologists claim that attitudes are particularly influenced by factors of education and cultural environment. At the opposite pole, manifest behavior is a concrete action, one that is observable with regard to a specific situation. We conclude that attitudes, along with many other factors, determine the subsequent behavior of the consumer [
6].
Kotler and Keller [
7] state that the starting point in studying and understanding the behavior of the consumer can be represented by the stimulus–reaction model. Thus, marketing and environmental stimuli enter the conscious mind of the consumer, resulting in the entry into play of a series of psychological processes, combined with the specific personal characteristics of the individual, that lead to purchasing decisions. The primary goal of marketing is to understand what is going on in the mind of the consumer, between the moment of penetration of external marketing stimuli and the moment of final purchasing decisions. According to Kotler and Keller, four essential psychological processes fundamentally influence and determine the reactions of consumers in the face of various marketing stimuli: motivation, perception, learning and memory.
Perception is seen as a somewhat complicated process that consists of mental activity when finding, understanding, and judging incentives. It is achieved with the help of the nervous system’s sensory receptors. In addition to the fact that elements of a physiological nature can explain perception, psychological aspects also play a defining role. However, the essential feature of perception is that it is selective and characteristic, due to the particularities of consumption, being complementary to the need [
8].
Learning, in turn, is closely related to perception, representing the set of elements by which individuals can get to know the products. In the process of assimilating information, specialists from the field of marketing pay the most attention to the sources of information, which in turn can be classified into two categories: personal resources, such as acquaintances and friends; and impersonal resources, such as product features. In the field of marketing theory, this stimulus has been the subject of several analyses, such as that following Bayesian theory. This is defined as a change to relatively constant behavior that is the result of repeated experience. This learned experience can be accrued through advertising or through the satisfaction obtained as a result of the use of a product [
8].
According to Kotler and Armstrong [
9], all the information and experiences an individual encounters in their lifetime can, therefore, be stored in their memory for a long time. Psychologists distinguish memory as falling into two categories: short-term memory (immediate), which presents comprises the temporary storage of information; and long-term memory (duration), involving the storage of information in what is, to a large extent, a permanent deposit.
Regardless of the circumstances, at any given time, a person has many needs. In their turn, these requirements are classified into biogenic needs, arising from physiological conditions such as hunger, thirst or discomfort, and psychogenic needs, arising from a state of psychological tension, such as the need for recognition, social groups, esteem, and a sense of belonging to a group. These needs turn into motivation when stimulated to a sufficient level of intensity.
The reason behind such motivation is that the need has become pressing enough to cause the individual to act [
9].
Given all these characteristics and the factors influencing consumer behavior that have been scientifically substantiated over long periods, we can consider that these motivations are generally valid and applicable today. However, even on the independently verified scientific basis that most behavioral decisions can be anticipated or explained, in crises, things change and people act differently. One such situation was the epidemiological context generated by the COVID-19 pandemic, in which consumer decisions were no longer influenced only by known endogenous and exogenous factors but also by increased concern and desire for self-protection.
In this context, several specialists have tried to identify the behavioral peculiarities generated by the epidemiological context of COVID-19 and to highlight the factors that triggered these changes. For example, Laato et al. [
10] investigated unusual consumer behavior through a structural model, based on the stimulus-organism framework that connects exposure to information sources with two behavioral responses: notable acquisitions and voluntary self-isolation. The results they obtained explicitly showed that there is a clear relationship between the intention to self-isolate and the intention to buy items that are less common. In other words, the more information about a situation that reaches consumers, the more intensely they feel the need to make unusual purchases.
Moreover, other researchers [
11] have studied new buying habits compared to old ones to determine whether these changes will be permanent, or if consumers will return to pre-pandemic purchase behaviors. It seems that some consumer habits will return to normal, but some objections will disappear because, in conditions of supply blockage, the consumer has discovered a more favorable and accessible alternative for satisfying the same needs as before.
During the pandemic, given the economic and social changes produced, the buying and consuming behavior of consumers has changed. Certain specialists [
12] believe that the temporary closure of many food outlets and the psychological impact of the inaccessibility of consumer goods has led consumers to make purchases differently from the way they normally would. The same authors conducted an experiment to identify consumer reactions in three different scenarios: in the first scenario, the number of cases of illness with the new virus was increasing; in the second scenario, the number of people confirmed positive remained constant; and in the third scenario, the number of cases was decreasing. The results showed that the trend of the COVID-19 pandemic causes significant differences in consumer behavior, as they are less willing to expose themselves to risk and go shopping in physical stores at times when the pandemic situation has not improved.
Considering the behavioral precedent in the case of purchasing consumer goods, we can appreciate that during the pandemic, people choose to give up or postpone medical services, especially if their health problem is not an emergency. Considering all these aspects, it is imperative to analyze the consumer’s behavior in the field of health services and how much the consumer is willing, in the current context, to evaluate the quality of the services they benefit from.
4. Materials and Methods
The research methodology in the context of this article respects the stages of studying a phenomenon through modeling via structural equations. Modeling by structural equations allows more precise analysis of the studied subjects (in this case, the satisfaction of the health service beneficiary in the context of the COVID-19 epidemic) by considering the interdependence relationships between variables as having direct and positive effects on the studied phenomenon. First of all, quantitative research was performed (the tool used was a questionnaire, consisting of approximately 20 questions) in order to build the necessary database for the analysis and proposal of a conceptual model. The data collection took place between 10 July and 25 July 2020. The questionnaire was distributed among patients (beneficiaries of health services) in 4 medical units in the southeastern part of Romania. The questionnaires were completed face to face but with the help of an online platform (practically speaking, the interview operator does not fill in the answers on a piece of paper but with the help of a tablet, inputting the data directly into the online platform). The duration of completing a questionnaire was about 15 min, depending on the degree of understanding of the respondent and the time spent thinking up answers to the questions. The first question was of the filter type, allowing only those respondents who benefited from health services in the context of the COVID-19 epidemic to access the questionnaire for the answers. The last seven questions were for identification and socio-demographic classification, to allow us to establish the profiles of the respondents, and the other questions provided the basis for achieving the behavioral model, thus identifying the variables used, based on the items presented, as answers to the questions that were asked. Therefore, each variable in the model was designed, based on one or more questions, and their image was created using the items used as answer options.
The purpose of the research (modeling by structural equations) was to determine the factors that, in the crisis caused by the COVID-19 pandemic, contribute to the formation of satisfaction in the beneficiaries of health services. In this sense, the model analyzed factors such as the sources of information used not only to identify providers, social influence, and the need felt for health services but also the protection measures used by the healthcare provider in order to prevent the spread of the virus.
One of the important aspects in the design of marketing research is represented by determining the size of the sample used [
17]. The representativeness of the sample is given primarily by the number of respondents and their characteristics. According to a 2019 EUROSTAT study, 20% of the country’s population never see a doctor (due to the rather high unemployment rate, there are people who, unfortunately, work outside the country and miss out on health services both abroad and at home; this encompasses almost 10–15% of the population registered in Romania). Thus, to determine the sample size, statistical formulas based on the significant takeovers considered for the elaboration of the research were used. Thus, the formula used was [
17]:
where:
n—sample size;
t—the coefficient of the probability of guaranteeing the research results (confidence level), established by the researcher depending on the complexity of the research (the t value is usually taken from statistical tables);
p—non-percentage share of those components in the sample that has a common characteristic or can be characterized by a certain attribute;
q—is equal to (1 − p) and represents the non-percentage share of the components in the sample that do not have a common characteristic or cannot be characterized by a certain attribute.
∆ω2 is also noted and represents the margin of error associated with the representativeness of the research.
According to the governmental statistical studies carried out at the country level (Functional Analysis of the Health Sector in Romania for the General Secretariat of the Government—
http://sgg.gov.ro/docs/File/UPP/doc/rapoarte-finale-bm/etapa-II/MS-RO-FR-Health-Sector-ROM) (19 November 2021), approximately 80% (81%, to be more precise) of the country’s population turn to health services for help. Given the fact that the country’s population is 19 million inhabitants, we calculated the percentage of 80% to identify the number of health-service users (resulting in a figure of approximately 15 million beneficiaries of health services). In this case, applying the formula for calculating the sample size, with a 95% probability of guaranteeing the results, we then have an associated value
t = 1.96, and a margin of error of ± 5%. Taking into account the fact that approximately 80% of the Romanian population is represented by people who use medical services, we considered the value
p as 0.80 (80%), resulting in
q = 0.20 (20%). The sample size can then be expressed as:
n = (1.96
2 × 0.80% population × 0.20)/0.05
2.
Thus, the final sample size for which the research can be considered as representative, with a confidence level of 95% and a margin of error of ± 5%, is 384 people (calculated value confirmed with the help of the automatic platform for calculating the sample dimensions
https://www.infomass.ro/wp-content/uploads/2010/09/caculator_marime_esantion.htm) (19 November 2021). However, taking into account the fact that we are conducting pilot research in the southeastern part of Romania, we can consider the representativeness of the sample of beneficiaries with a final number of 100 respondents who availed themselves of health services during the COVID-19 pandemic. The questionnaire was applied to only 100 respondents for two important reasons. Firstly, as mentioned in the text of this article, it is pilot research—that is, a research idea conducted with a small number of respondents to prove its effectiveness and necessity. Starting from this article, numerous other research studies in the field can be performed later, having already confirmed their relevance and the fact that there are links between the analyzed variables. Secondly, based on the statistical information at a national level, and applying the formula regarding the representativeness of the sample, a result of 384 respondents was obtained (a number that would provide representativeness to such research, based on the ratio between the total population, the number of people who turn to health services on their own initiative, and the probability of guaranteeing the results. Considering the fact that the research was carried out only in some medical units in the southeastern part of the country (i.e., less than half of the entire territory), we thus reduced the number of respondents who could give representativeness to the research to less than half (keeping the proportions-area/number of inhabitants-number of respondents).
Another argument regarding the sample size is related to the analysis method used (PLS-SEM). In the specialist literature, we identified a series of approved opinions regarding the recommendation of using the PLS-SEM analysis in the case of small samples, due to the accuracy of the obtained results.
Moreover, according to the literature, the general complexity of a structural model has little influence on the sample size requirements for PLS-SEM. The reason is the asynchronous analysis of the relationships in the structural model (two early studies were systematically evaluated for PLS-SEM performance with a small sample size and concluded that it had good results). Moreover, a simulation study cited by Hair et al. (2013) indicated that PLS-SEM is a good choice when the sample size is small because it offers higher levels of statistical power in situations with model structures or smaller sample sizes [
18].
Even so, some authors consider that a calculation rule should still be applied for the sample size used in the SEM-PLS analysis, and the general conclusion approached and adopted in the literature referred to the “rule of 10”. This rule says that the sample size is sufficient if the largest variant was chosen between two multiples of 10, as follows: 10 times the number of links between the variables in the model, or 10 times the number of formative indicators used to measure a single construct [
18]. In the present article, we identified in the model a figure of 7 links between constructs (variables), so a sufficient number would be 70 respondents for the purposes of analysis.
The interviewed patients were clients of medical service providers from both the public and private sectors; however, we selected for participation only those patients who resorted on their own initiative to specialized medical services (and not in an emergency). For this reason, we considered that the influencing factors regarding the choice of a particular supplier are equally representative. When analyzing the socio-demographic information of the respondents, we can identify the following characteristics in the sample: 75% were female and 25% were male; 85% were higher-education graduates (bachelor’s, master’s, or doctoral degrees) and only 15% had merely secondary education; 95% were employees or entrepreneurs; 3% were pupils/students and only 2% were without an occupation; 15% had incomes below RON 2500, while 85% of the respondents registered incomes over RON 2500; 85% come from urban areas and only 15% come from rural areas. Thus, we can state that most of the respondents were women, with higher education, with an above-average income level and who come from urban areas (centralized information is shown in
Table 1).
Before modeling the results obtained through structural equations, some relevant results can be highlighted, namely, that the main reason for receiving medical services in the context of the COVID-19 pandemic was those analyses and consultations performed in an emergency (according to
Figure 1).
Moreover, according to the respondents in this research, the protection measures adopted by the medical service provider that contributed to the client’s decision to choose that particular provider showed that the use of appropriate protection measures by medical staff, and testing patients before hospitalization, are two of the more appreciated actions by the beneficiaries (according to
Figure 2).
We can, thus, assume that there is a link between the perceived need and the appreciation of the protection measures implemented by the medical service provider. In this case, a broader analysis was considered useful, providing data based on which we can analyze the direction and intensity of the influence between the variables. The chosen research method is suitable for the subject of this article due to the possibility of concomitant analysis of the links between several variables that may influence the behavior of health service consumers.
For the construction of the variables within the model, a series of questions were used as follows. For the variable “social influence”, a question structured on the semantic differential scale was used (5 selection variants: to a very large extent (5), 4, 3.2, to a very small extent (1)), in which the respondent was asked to what extent he or she was influenced in choosing the provider of medical services by the recommendations of other persons, such as family members, friends and acquaintances, specialized resources, and other doctors. For the variable “sources of information”, two questions were used: a grid-type question with multiple answers (“Where do you get information when you are looking for a health care provider?” is furnished with five possible answers: “search for information on the Internet”, “I follow the promotional materials in media”, “I follow the recommendations of some public figures”, “I ask the family doctor”, or “I ask friends/family”) and a question structured on a different measurement scale semantics, namely, “To what extent did the following sources of information influence you in the decision to turn to a certain provider of medical services?” The answers from which they could select the degree of influence were: online promotion, reading specialized articles, promotion through social networks, TV commercials, recommendations from public figures, recommendations from friends and family, or recommendations from the family doctor. For the variable “felt need”, a question built on the semantic differential measurement scale (with 5 levels) was used, and the answer options were: exclusively consultations, specialized analyses, medical treatments, operations, or other services. The variable denoting “decision to choose a medical service in the context of COVID-19” was measured via two questions and a set of eight items, including: measurement of temperature, epidemiological triage, the provision of additional protective equipment (gloves, gown, etc.), the use of appropriate protection measures by medical staff, the implementation of social distancing measures by the medical unit, the implementation of measures for the disinfection of spaces, surfaces and aeroflora by the medial unit, testing patients before hospitalization (PCR testing, pulmonary CT, pulmonary X-ray), etc. For the variable “Protective measures addressed by the medical service provider “the semantic differential scale was used by which the respondents were asked to mention the extent to which the implementation of protection measures by the medical unit contributes to the decision to use the services offered by that unit. Among the available answers are “measures imposed on medical staff”, “measures imposed on patients”, “time for disinfection between appointments”, “devices with disinfectant for patients”, “protective equipment for medical staff”, “protective equipment for patients”, “epidemiological triage at entry into the unit”. The last variable used in the proposed conceptual model, namely “satisfaction of the beneficiary of medical services in the context of COVID-19” was measured using a set of nine items that included not only available information and the quality of services provided (for example, the location of the medical unit is convenient, the building is attractive, it is easy to find a parking space, the reception is modern, the staff are friendly, the services of the medical unit are easy to access, etc.) but also the changes as a result of the COVID-19 pandemic (for example, the medical unit has implemented an online programming system, the medical unit offers online consultations/online analysis interpretation for cases that do not require physical consultation, the unit respects the scheduling hours and allocates time between appointments for space disinfection, etc.).
Using
Figure 3, we can observe the causal relations proposed in the case of modeling the satisfaction of the beneficiary of health services in the context of COVID-19, from which it can be seen that the following objectives were considered: O1—social influence has a direct and positive effect on sources of information used by beneficiaries; O2—social impact has an immediate and positive impact on the beneficiary’s decision to choose a medical service in the context of COVID-19; O3—the sources of information used have a direct and positive effect on the need felt by the beneficiary; O4—the need felt by the consumer has an immediate and positive impact on his decision to choose a medical service in the context of COVID-19; O5—the protection measures implemented by the health care provider directly and positively influence the decision of the beneficiary to choose a medical service in the context of COVID-19; O6—protection measures implemented by the medical service provider directly and positively influence the satisfaction of the beneficiary of health services, in the context of COVID-19; O7—the decision to choose a medical use in the context of COVID-19 has a direct and positive effect on the satisfaction of the beneficiary of medical services, in the context of COVID-19.
The proposed hypotheses confirm the existing relationship between two variables, starting from the premise that each has a direct and positive influence on another with which it forms a connection.
After collecting the results on a sample of about 100 people, the data were analyzed using the statistical program SPSS and the WarpPLS program to determine the size of the effect produced between two variables.
In order to develop a representative SEM analysis, it is necessary to evaluate the accuracy of the data used, respectively, the complexity of the variables used (the quality of the items used and how complete each analysis is), their consistency and validity. In this sense, evaluation indicators, such as Cronbach’s alpha, reliability coefficients and the extracted average variants, are used (
Table 2).
The reliability of the measurements is defined as the extent to which they are error-free and provide consistent results; they are measured using Cronbach’s alpha indicator. In order to provide a measure of internal consistency, the indicator must be expressed by a number between 0 and 1 [
19]; it can be seen in the table that this is met.
Within the WarpPls software, the relationships between the variables are highlighted, as well as their validity and confirmation from the point of view of the correctness of the items on the basis of which they were built. Thus, the average validity test extracted from the AVE is used; if it meets the criterion for classification in coefficients, this demonstrates the quality of the measurements, and these can be used to validate the convergence. According to the literature, the values related to the reliability coefficients must exceed the threshold of 0.5 and be lower than any other values recorded on each column [
20].
The results highlighted in
Table 3 demonstrate that the discriminant validity has been met, which shows that the measurements performed are representative of the definition and use of variables within the proposed conceptual model.
The main hypotheses of the research are based on descriptions of the relationships between the latent components of the proposed model. These are tested by calculating the binding coefficients (Beta-standardized coefficients) corresponding to each causal relationship in the model. The value of the Beta coefficients indicates the strength and the direction of the correlation between the variables; the validation of the hypotheses materializes when the value of the related Beta coefficient is higher than 0.1 at a significance threshold of p < 0.05.
Following the results obtained, as highlighted in
Table 4 and
Figure 4, it can be seen that most of the proposed hypotheses have been validated, except for the finding that the need felt by beneficiaries to use specialized medical services does not directly and positively influence their decision to choose specific medical services in the context of COVID-19.
Determination coefficients R
2 represents “the percentage in which independent latent variables explain (determine/influence) the variation of dependent variables” [
21]. These were calculated using the WarpPls program for each component-dependent variable of the proposed model and are shown in
Table 5. It should be noted that a predictor variable can be considered to have a substantial explanatory power when the value of the coefficient of determination R
2 is more significant than 0.1.
In the case of the present research, we can observe that most of the values of R are close to 1, which indicates that the predictor variables can be considered to have significant explanatory power.
6. Conclusions
Following the results obtained both from exploratory research of the literature and based on this research, we can conclude by saying that consumer behavior has undergone considerable changes in the context of the COVID-19 pandemic, even in the case of health services. According to the literature, the marketing of health services is not necessary because people feel the need for specialized medical services without their being promoted, but it seems that, in the current context, people have significantly reduced the number of visits to the doctor.
Moreover, following the results obtained via modeling by structural equations, we could see that social influence has a direct impact on the sources of information used by consumers in the case of health services and on their decision when choosing a medical service in the context of COVID-19. In addition, it seems that the more informed they were about the services offered by a particular provider, the greater the need they felt. In this context, it was demonstrated that before seeking health services in the context of COVID-19, consumers felt the need to learn more in two respects: first, if the need they felt required a mandatory visit to the doctor’s; the second time, if the provider of the service has implemented appropriate protection measures. It also appears that the protection measures addressed by the health care provider have significantly contributed to the decision to choose a particular health care service in the context of COVID-19 and to the satisfaction felt afterward (according to hypotheses 5 and 6).
According to the literature, over the years there have been several epidemic outbreaks that have brought behavioral changes in humans (Ebola, SARS, and swine fever). In all these situations, human behavior has changed in two respects, namely, consumption behavior [
24] and behavior to reduce the risk of disease [
25].
In these cases, the obtained results also highlighted the directions of change in consumer behavior, namely, attempts to reduce the use of medical services (behavioral change in decision-making) and increased attention to protection measures implemented by healthcare providers, in case of need (behavioral change, in the sense of reducing the risk of illness).
The results show that the variables used (constructed using the questions and based on the set of items mentioned and listed in the previous section) in the proposed conceptual model were measured accordingly, thus highlighting a number of links not only among factors that contribute to beneficiaries’ decisions regarding health services during the COVID-19 pandemic but also among factors that contribute to their satisfaction.
Moreover, it could be highlighted that the digitization of medical procedures (making online appointments and providing online consultations) represented an evolutionary framework not only for traditional medical services but also to achieve a favorable outcome for beneficiaries.
We can conclude by saying that the advantages brought by the literature consist in analyzing the behavior of the consumer of health services from the perspective of their satisfaction, as felt in the context of COVID-19, and the variables that contribute to its formation. It seems that as consumers adapt to the use of other means to meet their needs, the new means may be considered more convenient and replace the old habits of using health services permanently. Besides this, unusual decisions about behavior and disease prevention measures will continue to be made by consumers, probably until the pandemic is over.
The limits of this research can be considered primarily as being related to the small number of respondents and the area of the research (the southeastern region of Romania). We consider that for a series of representative results at the level of the entire country, such research should be extended to a larger number of respondents and with wider coverage of all regions of the country.
Recommendations for future research can be exemplified by extending such research across the country, but other variables can be added to this proposed conceptual model (such as risk perception, risk attitude, and those factors that determine risk perception, depending on socio-demographic characteristics). We consider that the present research represents a pillar in the studied subject, and any development of it can bring significant information to enhance the specialized literature.