Public Sector Economics



Taxpayers' attitudes toward tax compliance in the Slovenian tax system: differences according to gender, income level and size of settlement

Lidija Hauptman
Berislav Žmuk
Ivana Pavić
Article   |   Year:  2024   |   Pages:  177 - 201   |   Volume:  48   |   Issue:  2
Received:  September 29, 2023   |   Accepted:  February 16, 2024   |   Published online:  June 10, 2024
Download citation       



This study investigates the relationship between certain economic and psychological factors and demographic characteristics of Slovene taxpayers, such as gender, income level and size of settlement as it is becoming important for a country’s tax compliance framework to align with the tax recommendations of global institutions. The results show some gender differences, with males being less likely to feel guilty or bad if taxes are not paid in full than females, whereas females tend to have the opinion that working for cash-in-hand payment without paying tax is not a trivial offence. Taxpayers with low incomes tend to agree that tax evasion is morally acceptable if tax rates are too high. Taxpayers from rural settlements exhibit a higher tendency to feel morally obligated to pay their taxes than taxpayers from urban settlements. The findings indicate that the vast majority of taxpayers feel morally obligated to pay their taxes.

JEL:  M48, H26, D91

1 Introduction

Global institutions (UN, 2015; OECD, 2016; WTO, 2023) are exerting significant pressure on governments to achieve the sustainable development goals, enhance revenues, and strengthen tax compliance, which is why gaining a better understanding of taxpayers’ behaviour has become increasingly important for alignment with the recommendations of global institutions. One way to improve a country’s tax compliance framework is by analysing it from various perspectives. Individual taxpayers exhibit diverse motives, influenced by economic and psychological factors, for either complying with the rules or else engaging in tax evasion. This behaviour is also impacted by gender, income level, settlement size, age, religion, education, employment status, sector of work and other factors. The research question in this paper pertains to understanding whether taxpayers’ attitudes towards the economic and psychological factors of tax compliance in Slovene tax system differ according to their gender, and whether income level and size of settlement impact compliance factors.

In the context of its Tax and Gender Equality Policy, OECD (2022) urged countries to take steps to analyse the impact of gender on tax compliance measures as 70% of the countries in their research stated that they do not collect gender-disaggregated data. Slovenia belongs among those countries in which gender-disaggregated data are available from tax returns for use in policy analysis, but it does not practice gender budgeting and has no access to gender disaggregated non-tax data available for policy analysis or detailed micro data on male and female incomes (OECD, 2022). Gender equality constitutes a core value within the European Union (Stanimirović and Klun, 2021). This fundamental value should be reflected in taxation. Countries’ public policies should be designed in such a way as to take into account gender equality (Stanimirović and Klun, 2021). Taking this argument into consideration, it is not clear whether the proposed OECD measures in the 2022 report on gender equality would enhance the efficiency and effectiveness of a country’s tax system, especially if the system is already gender tax neutral as per existing tax laws in Slovenia. Still, there is some empirical evidence confirming that the country has taken the first steps in gender-responsive budgeting via guidelines and recommendations (Stanimirović and Klun, 2021). However, the impact of gender on tax compliance measures has been confirmed in various studies and reports. Some studies confirm higher female levels of tax compliance globally (Hofmann et al., 2017; OECD, 2019a; D’Attoma, Volintiru and Steinmo, 2017; D’Attoma, Volintiru and Malézieux, 2020; Kangave, Waiswa and Sebaggala, 2021), while others highlighted higher male tax compliance (Grosh and Rau, 2017; Alm and Malezieux, 2021). There are also studies that have not identified any significant gender impact (Vincent, Stevenson and Owolabi, 2023; Nichita et al., 2019).

Next to gender, previous research proposed income level as a significant determinant influencing tax evasion and therefore a robust predictor of tax compliance, occasionally contingent on the geographic region (McGee, 2012; Hofman et al., 2017; Khalil and Yusuf, 2022). Psychological research found that not just level but also the origin of income (earned or endowed) affects taxpayers’ decision making and produces different behaviour (Durham, Manly and Ritsema, 2014). Higher-income taxpayers can exhibit attitudes toward the probability of audits and punishments, fairness, trust in the tax system, tax evasion, and public services that are different from those of low-income taxpayers. Elderly taxpayers and those with low income disproportionately engage in undeclared work (Levenko and Staehr, 2022). Low-income taxpayers tend to exhibit lower tax compliance than high-income taxpayers, and taxpayers with higher incomes are considered more prone to tax evasion than those with lower incomes (Hofmann et al., 2017). The investigations of income level impact on tax compliance in previous studies showed mixed results, with both a positive (Kirchler, 2007; Kirchler et al., 2007) and a negative relationship (Alm and McKee, 2006) being adduced.

Even though there has been some empirical research measuring the impact of urban or rural business location on tax compliance (Dissanayake and Premaratna, 2020; Williams, 2020), there is not sufficient research in measuring the impact of settlement on individual taxpayers with respect to tax compliance. Settlement in rural area can provide stronger social bonds and communal pressure might contribute to increased tax compliance, while in urban areas, settlements with enhanced access to tax-related information and resources, can foster a better understanding of tax regulations and consequently encourage tax compliance. The importance of this demographic factor is still not clear, but some argue (Devos, 2008) that taxpayer location has no impact on compliance.

Our study discusses the impact of three factors (gender, income level, and settlement size) that have already been addressed in previous studies. However, their combined effect on tax compliance decisions has not been examined, especially not among Slovenian taxpayers. Thus, the aim of this research is to examine the impact of taxpayers’ attitudes as the relationships between socio-demographic factors such as gender, income level, settlement size, and tax compliance. The results can provide support for policymakers when enhancing tax compliance policies and adjusting tools based on the impact of these three factors. Furthermore, they can be beneficial as support for the governments’ regulatory impact assessment (RIA) of regulations and non-regulatory alternatives. If tax authorities understand the motives behind tax compliance, they will also understand and easier predict the reasons for evasion. This article makes a valuable contribution to the tax compliance literature by analysing the attitudes and influences of gender, income level, and settlement size on both the economic and psychological factors affecting taxpayers. This way a better understanding of taxpayer motivation and reasoning can be achieved.

This paper consists of six parts. After the introduction is the second section, which summarizes the theoretical literature about tax compliance factors. The third section presents an overview of empirical research, detailing the methodological aspects used in this paper and variable descriptions. In the fourth part, discussion of the main results is presented. The last section contains concluding remarks along with the potential for future research.

2 Theoretical background and research hypotheses

2.1 Tax complinace theories

Tax compliance means that taxpayers fully disclose their income and meet all tax obligations by paying the required taxes as required by law (Alm, 1991). There are two basic approaches to tax compliance; economic approach also called the Allingham and Sandmo model, the deterrence model, the expected utility theory, or the rational choice theory and the behavioural approach also called the moral sentiments theory. The behavioural approach states that taxpayers’ tax compliance behaviour is influenced not only by financial factors but also by non-economic variables such as trust in government, tax morality, perception of tax justice and even by religiousness. Tax compliance is a multifaceted topic, and researchers often define it differently according to the specific nature and scope of their studies.

The basic theoretical model used in much research on tax compliance originates from Becker’s work in 1968, particularly his economics-of-crime model. The conventional conclusion drawn from this model is that individuals fulfil their tax obligations primarily out of fear of being detected and penalized if they fail to report all their income (Alm, 2019). In the economic approach developed by Allingham and Sandmo (1972), taxpayers tend to engage in tax evasion and tax avoidance. Taxpayers typically conduct a cost-benefit analysis, weighing the potential penalties for irregularities and the tax loss they might incur if their tax evasion is discovered during inspections or audits against the amount of tax they would owe if they fully declared their income. If the penalty for tax evasion and detected tax loss exceeds the tax they would pay when declaring their income honestly, they are more likely to opt for full disclosure; otherwise, they may choose not to do so (Güzel, Özer and Özcan, 2019).

This approach consequently provides a logical and effective outcome, indicating that tax compliance is dependent on the possibility of audit (detection) and penalty rates. According to this, it can be concluded that tax compliance depends on enforcement elements whereas government can achieve higher tax compliance by increasing audit and penalty rates. Alm (2019) pointed that there are some problems related to the conventional expected utility approach to tax compliance since certain key predictions of this approach are not strongly corroborated by empirical evidence. For example, one prediction of the standard theory is that a higher tax rate increases compliance. Conventional approaches to improving tax compliance encompass deterrent measures such as regular audits, diverse penalty structures, and potential criminal charges (Shim, Shim and Shim, 2023). In traditional tax compliance literature, deterrence is regarded as a crucial element in the decision-making process related to tax compliance (Kogler et al., 2013). In accordance with the economic model proposed by Allingham and Sandmo in 1972, robust enforcement through rigorous audits and substantial fines stands as the foremost factor influencing tax compliance. Prior research suggests that deterrence, while important, is not the sole determinant of tax compliance and sometimes may result in negative rather than positive influences on compliance (Shim, Shim and Shim, 2023). According to this perspective, social and psychological factors, including social norms, considerations of fairness, and trust in authorities, are likely to lead to voluntary tax compliance. The slippery slope framework, proposed by Kirchler, Hoelzl and Wahl (2008), which represents the development of the economic theory of tax evasion developed by Allingham and Sandmo (1972), recognizes the interconnectedness deterrent factors that increase enforced tax compliance and social psychological factors that lead to voluntary compliance. This framework examines two dimensions: trust in the tax authority and the authority’s power, and their impact on tax compliance. As a result, numerous subsequent studies drew inspiration from and were built upon this framework (Gangl et al., 2012; Batrancea et al., 2019).

Extension of the economics-of-crime model incorporated specific behavioural elements that have been explicitly addressed in other social sciences. Behavioural economics can be broadly described as an approach that utilizes techniques and insights from various fields, particularly psychology, to enhance the understanding of individual and collective decision-making processes. Behavioural economics has been employed in two overarching and interrelated dimensions. Onedimension concentrates on individual factors, while the other on group considerations. Frame dependence is often associated with an individual’s psychological tendencies or cognitive limitations. Individuals are driven by diverse motivations, encompassing not only self-interest in a narrow sense but also factors stemming from collective considerations. These include concepts such as fairness, altruism, reciprocity, empathy, sympathy, trust, guilt, shame, morality, alienation, patriotism, social norms, social customs, social capital, tax morale, intrinsic motivation, and numerous other objectives (Alm, 2019). Numerous factors have the potential to influence individual decisions regarding compliance. In accordance with gender socialization theory, distinct roles and values are assigned to men and women, leading to variations in their interests, decisions, and practices (Helmy, Dwita and Cheisviyanny, 2020). According to gender socialization theory, men are twice as likely as women to participate in unethical behaviour (Tjondoro, Widuri and Manoy, 2019). Women taxpayers have been shown to exhibit an elevated tax morality (Hartmann, Mueller and Kirchler, 2019).

The theory of planned behaviour (planned behaviour theory), a theory in social psychology, explains human behaviour by considering behavioural intentions influenced by attitudes towards three components (the behaviour, subjective norms, and perceived behavioural control) (Ajzen, 1991). According to the theory of planned behaviour (TPB), taxpayers are expected to engage in conscious decision-making, considering all the available information, they will analyse the potential consequences of their actions, resulting in their decision being influenced by a careful weighing of advantages and disadvantages, as well as the anticipated reciprocity (Sundari, Chariri and Utomo, 2022). The study by Taing and Chang (2020) uses the TPB as a conceptual framework to explore the factors influencing the compliance intentions of Cambodian taxpayers.

2.2 Economic and psycological factors of tax complinace

Tax compliance factors can be grouped into three categories based on the level of control power they possess. The first group of factors includes those that are under the control of tax authorities and include penalty, audit probability and tax rate (Alm, 2012; Ritsatos, 2014). Factors in group two are affected government and include trust in government and the tax service (Ritsatos, 2014). The third group of factors includes factors such as personal norms, social norms, and religiosity, all influenced by the taxpayers and by the community they belong to. There are two different paradigms that can be used to increase tax compliance, one using enforcement (Alm, 2012), known as the deterrence approach, and another approach, known as the non-deterrence approach, which includes trust and the service paradigm (Nurkholis, Dularif and Rustiarini, 2020), as well as the fiscal psychological paradigm (Schmolders, 1959).

Fines or tax penalties and tax audits are two instruments that necessarily exist in a country’s tax system (Kirchler, Hoelzl and Wahl, 2008). The presence of these two tools is consistent with the constructivist rationality, a top-down mechanism in which formal institutions create instruments to encourage tax compliance (Górecki and Letki, 2020). Tax penalties are regulatory measures in taxation designed to encourage tax compliance. Tax audit includes controlling taxpayers’ compliance with the tax reporting system by completing, calculating, collecting, and deducting all taxes in accordance with applicable laws, ensuring compliance (Rahmayanti, Sutrisno and Prihatiningtias, 2020). Individuals make decisions about whether to adhere to tax regulations or not based on an evaluation of the potential benefits and costs (Allingham and Sandmo, 1972). Taxpayer awareness refers to the recognition, respect, and adherence to applicable tax regulations, accompanied by a sense of seriousness or willingness to fulfil their tax obligations (Nurkhin et al., 2018). Non-compliant behaviour with tax regulations may be result of a taxpayer’s limited knowledge of and the skills required to adhere to them (OECD, 2019b). Previous research showed diverse findings related to correlation between tax knowledge and the degree of tax compliance (Wadesango et al., 2018). Loo, McKerchar and Hansford (2014) stated that within a self-assessment system, taxation knowledge emerges as the most influential factor in shaping taxpayer compliance behaviour. Timothy and Abbas (2021) identified a noteworthy positive connection between tax knowledge and tax compliance, reinforcing the idea that taxpayers’ inherent motivation should be complemented by sufficient tax knowledge. The findings of Rahmayanti, Sutrisno and Prihatiningtias (2020) indicate that tax penalties, tax audits, and the awareness of taxpayers have a positive impact on corporate taxpayer compliance. From a taxation point of view, non-compliant behaviour of taxpayers refers to actions taken to reduce tax obligations, which may occur unintentionally due to ambiguities in tax laws and are commonly recognized as tax avoidance and tax evasion. The decision for individuals on whether to engage in tax evasion is clearly constrained by the available opportunities for doing so. The final choice on tax evasion will depend on the possibility of tax evasion and the personal desire to engage in it. The decision can be influenced by economic or psychological factors, with psychological factors implying that an individual’s well-being may be affected even in the absence of direct economic consequences (Levenko and Staehr, 2022). Economic factors regarding tax compliance include the possibility of evasion, rational individualistic choice, and behavioural individualistic choice (Levenko and Staehr, 2022). In the context of discussing psychological factors for tax compliance, it is possible to differentiate personal norms, social norms, and norms related to interactions with the government (reciprocity) (Levenko and Staehr, 2022).

Dularif and Rustiarini (2022) propose that enhancing tax services and fostering trust in the government are more efficient and simpler to implement than developing positive taxpayer behaviours. Taxpayers’ perception of equitable treatment by tax authorities can have a positive impact on their willingness to comply with tax obligations (Taing and Chang, (2021). The government should understand how to create a sense of fairness in the taxpayer-government relationship (Kim, 2002). When taxpayers believe that the tax burden is equitably distributed among their peers, they are more inclined to adhere to tax regulations (Taing and Chang, (2021). The quality of tax services can be enhanced through improving tax administration services through cooperation and socialization, and by enhancing the efficiency of the tax system (Alm et al., 2010).

The government should take all possible actions to serve the common good since citizens have to give some portions of their revenue or assets to the government in the form of tax. Effective collaboration between the government and its citizens plays a crucial role in fostering trust and shaping tax compliance. Building trust in government is based on three key factors: enhancing public engagement in the political process, provision of sufficient public goods, and upholding government credibility (Piryaei, Akhlaghi and Saeed, 2015). Most research findings (Cahyonowati, Ratmono and Juliarto, 2023; Nurkholis, Dularif and Rustiarini, 2020) endorse the idea that higher levels of trust in government tend to result in greater tax compliance. Dularif and Rustiarini (2022) conducted a comprehensive study and identified trust in government and tax services as the most important factors in achieving tax compliance while social norms, personal norms, and religiousness were found to contribute to tax compliance, but with lesser impact than initially anticipated. Hartmann et al. (2022) discovered that building trust in government and encouragement of a positive attitude towards taxes have a beneficial effect on compliance.

The behaviour of taxpayers is not only affected by the risks of formal punishment by state authorities but is also determined by psychological factors. Ethical or normative factors are recognized as factors with a significant influence on tax compliance (OECD, 2010; Dularif and Rustiarini, 2021). These factors are also called psychological factors and include personal norms and social norms. Personal norms can be described as deeply ingrained beliefs regarding what is right or wrong to do. Moreover, personal norms can be defined as the conviction that there exists a moral obligation to adhere to certain standards (Wenzel, 2005). Personal norms can be subdivided into two components: attitude (Ajzen, 1991) and intrapersonal norms (Hur, Roese and Namkoong, 2009). Individual personal norms develop over years and cannot be easily influenced. Social norms refer to the behaviour, ideas, and convictions prevalent within social groups, and the norms of a specific social group have a substantial impact on the actions of individuals who identify with that group (OECD, 2010). Regarding these social norms, a differentiation can be made between descriptive social norms and prescriptive social norms (OECD, 2010). Descriptive social norms reflect what others do or what we think they do while prescriptive social norms reflect what others think about certain behaviour or our perceptions of what others believe (OECD, 2010). Bobek, Hageman and Radtke (2015) have classified social norms in four categories: subjective norm, personal norm, injunctive norm, and descriptive norm. Subjective norm pertains to a taxpayer’s perception of their friends’ or family’s opinions or actions, injunctive norm is connected to a taxpayer’s viewpoint regarding the actions of individuals in a comparable situation or within the same group while the descriptive norm refers to a taxpayer’s perception of tax evasion based on prevailing public opinions (Dularif and Rustiarini, 2021). The behavioural approach acknowledges that individuals’ behaviour is significantly shaped by the collective behaviour of the group (Alm, 2019).

2.3 Hypothesis

The Helmy, Dwita and Cheisviyanny (2020) gender socialization theory argues that the socialization of individuals of different genders differs. This results in the formation of unique values, roles, and beliefs that further impact their interests, decisions, and norms (Helmy, Dwita and Cheisviyanny, 2020). Gender can influence the attitude and perception of taxpayers on tax compliance, as indicated in previous research, depending on context. Interaction between gender and economic or psychological determinants have confirmed that, in some cases, females are more moral than male (Kastlunger et al., 2010; Hofmann et al., 2017). This is because females do not break the laws, females’ personal and social norms have a stronger focus on tax compliance than male norms (Hofmann et al., 2017). Females are also less risk-seeking, more engaged in social and societal activities, and perceive sanctions as more threatening than males (Croson and Gneezy, 2009; Prasetyo, Adi and WoroDamayanti, 2020). To avoid tax audits, females tend to comply with the tax laws meticulously (Hai and See, 2011). However, according to the Filippin and Crosetto (2014) meta-analyses, 90 per cent of the published papers either found no differences between the female and male tax compliance behaviour, or only insignificant variations (Kangave, Waiswa and Sebaggala, 2021). Building on the presumption of the previous research, we propose the following hypothesis:

H1: There are no statistical differences in attitudes toward the economic and psychological factors of tax compliance among taxpayers of different genders.

In different geographic regions, income exhibited varying degrees of influence on tax compliance, with some northern hemisphere countries (Western and Eastern Europe, North America, Central Asia) showing a stronger predictive relationship between income and tax compliance, while certain southern hemisphere countries (East Asia and Pacific, Latin America and Caribbean, Sub-Saharan Africa) displayed a weaker predictive association (Hofmann et al., 2017). Dissanayake and Premaratna (2020) proved that the taxpayer’s level of income is positively correlated with voluntary tax compliance, while enforced compliance shows a negative association, whereby power and perceived trust in tax administration are important. The findings of Durham, Manly and Ritsema (2014) suggest that the interaction between income source and context does not have a significant impact on overall compliance, but this combined influence does shape the connection between income level and compliance, as well as the evolution of compliance behaviour over time. Income level positively correlates with compliance in the estimation. Earned income interacts with the tax context to contribute to a stabilizing influence on compliance behaviour as time progresses (Durham, Manly and Ritsema, 2014). Income level significantly influences tax compliance as individuals with high incomes are more likely to engage in non-compliance behaviour, while low-income individuals are less prone to it (McGee, 2012). Prior research on the income level’s effects on tax compliance has identified contrasting views, positive (Durham, Manly and Ritsema, 2014; McGee, 2012) and negative (Alm and McKee, 2006; Trivedi and Chung, 2006). Based on the facts presented and the circumstances that Slovenia belongs geographically to the European countries, it could be assumed that a relationship exists between gross income and tax compliance. However, due to the fact that the Slovene tax system is one of the smallest within the European Union according to number of resident taxpayers, we tested the following hypothesis:

H2: Taxpayers with different income levels do not have statistically significantly different attitudes as far as concerns the economic and psychological factors of tax compliance.

Research concerning the connection between the settlement size (location, rural and urban area) variable of the individual taxpayer and taxpayer compliance is notable while there is some empirical evidences for the location (area) variable of businesses and tax compliance (Dissanayake and Premaratna, 2020; Williams, 2020). According to that research, rural and urban locations are in correlation with trust and voluntary compliance, and among the rural business community, the power of authority reduces tax compliance (Dissanayake and Premaratna, 2020). Williams (2020) even argues that at the European level, those in urban areas were significantly more compliant than those in rural areas. Due to the geographic specifics of the Republic of Slovenia and the size of its population, around two million, we assume, that settlement size has no impact on compliance factors. Therefore, we frame the last hypothesis as:

H3: Taxpayers residing in different settlements size (urban/rural) do not have statistically different attitudes toward the economic and psychological factors of tax compliance.

3 Data and methodology

3.1 Sample and data collection

Survey data about Slovene taxpayers’ attitudes toward tax compliance were collected based on a written questionnaire that was administered. Due to the sensitive nature of the topic and individuals’ reluctance to participate in studies involving taxes and income, the snowball sampling approach was chosen for selecting respondents. The snowball sampling technique is suitable for research when the population is challenging to reach, as is the case when the general population rarely chooses to participate in surveys about their tax payments and income. The process begins by identifying a set of initial respondents for interviews. These respondents then recommend others to be contacted and interviewed. This cycle continues until a sufficient number of respondents is reached (Dragan and Isaic-Maniu, 2013). The target population was individual taxpayers in Slovenia. The sampling process relied on participants’ willingness and their availability to complete the questionnaires. The survey was conducted in the period from March to May 2023. In that period overall 390 respondents participated in the survey and filled out the questionnaire.

In the empirical study three research hypotheses were tested; as explained previously, we used only the segments out of the questionnaire that are presented in table 1. The questionnaire was designed based on the previous relevant research conducted by Hasseldine and Hite (2002), Gangl et al. (2013), Braithwaite (2001), Mohamed et al. (2013), Filippin, Fiorio and Viviano (2013), Palil (2010), McGee and Linge (2008). For the purpose of this paper, in addition to the selected demographic variables, overall, 17 questions were observed.

Table 1
List of observed questions and variables

All 17 questions belong to the 5-point Likert scale type of question. Therefore, the respondents could give an answer in the range from 1 (strong disagreement) to 5 (strong agreement). Answer 3 represents the neutral answer by which respondent points out that he does not disagree or agree with the presented statements in the survey questions. The main variable under the study is the variable related to the tax compliance of the respondents (variable TC). This variable shows how much the respondents feel morally obligated to pay their taxes. The other 16 questions can be separated into two groups: economic (EF) and psychological (PF) variables (factors). Included in the first group are the variables which cover the economic motives of tax compliance whereas the second group includes variables that cover the psychological aspects of tax compliance. In that way, the first, economic group includes five (EF1–5) questions whereas the second, psychological group is composed of 11 questions (PF1–11). To analyse the data, Microsoft Excel, R 4.3.0 and RStudio 2023.06.1+524 were utilized. Microsoft Excel was primarily used for data editing, while RStudio was used for statistical analyses.

3.2 Data analysis

The respondents’ structure according to main demographic characteristics is shown in table 2.

Table 2
Respondents’ structure, main demographic characteristics

In the survey, 187 males (48%) and 203 females (52%) participated. The median age of the respondents was 42 years, which means that 50% of the respondents were 42 years old or younger, while the remaining 50% were 42 years old or older.

Slightly over 50% of the participants reported a monthly gross income ranging between 1,001 and 2,000 euros. This specific range is categorized as a medium income level. 12% of respondents fell into the low-income bracket, indicating an income below 1,001 euros, while approximately 17% of respondents belonged to the high-income category, signifying an income exceeding 2,000 euros. Out of all respondents, 21% respondents opted not to disclose their gross income information. Consequently, those respondents, who did not share their gross income information, will be excluded from subsequent analyses involving the gross income variable.

Forty per cent of the respondents lived in an urban settlement, and 60% in a rural settlement. An urban settlement is considered to be a settlement with more than 3,000 inhabitants. On the other hand, a rural settlement is a settlement with fewer than 3,000 inhabitants.

To answer the research question, the results of the survey data were further analysed as follows. In the first step, the structure of responses to the 17 observed questions was analysed. For a better understanding of the structure of responses, in addition, main descriptive statistics indicators were calculated as well. To measure the relationship between the variables, polychoric correlations were calculated. Polychoric correlation is used to calculate the correlation between two ordinal variables. The 17 observed questions have only five levels and therefore it is more appropriate in this case to use the polychoric correlation instead of the usually used Pearson linear correlation. However, the interpretation of polychoric correlation corresponds to the interpretation of Pearson’s linear correlation. If the polychoric correlation indicator is equal to -1, a perfect negative correlation is present between the observed variables. On the other hand, a polychoric correlation indicator value of 1 indicates a perfect positive correlation. In the absence of any correlation, the polychoric correlation indicator is equal to 0.

In the second step, the Chi-square tests of independence were conducted (see McHugh, 2013). The Chi-square test of independence is used to inspect whether two categorical variables are likely to be related or not. In this paper, the relationship between disagreements and agreements on the observed 17 survey questions with respondents’ gender, gross income, and settlement were tested. To conduct the Chisquare test of independence, answers to the observed survey questions were grouped as follows. Answers 1 and 2 in a category 1 and 2 were treated as disagreement whereas answers 4 and 5 were considered as agreement. For all 17 questions, category 3, the neutral category, was not observed and was omitted in the Chi-square test of independence. The null hypothesis of the Chi-square test of independence is that the two observed variables are not related, not associated, and independent. If the null hypothesis of the Chi-square test of independence is rejected it can be concluded that the two observed variables are related, associated, and dependent. The test statistic used in the Chi-square test of independence is as follows (McHugh, 2013):


where oij is the observed cell count in the i-th row and j-th column of the contingency table, eij is the observed cell count in the i-th row and j-th column of the contingency table (McHugh, 2013). The calculated χ2 value is compared to the χcritical value for (r-1)(c-1) degrees of freedom and chosen significance level. The decisions for the Chi-square test of independence will be brought by considering a significance level of 0.05.

4 Results

4.1 Descriptive statistics

In this chapter, the collected attitudes on tax-related questions were analysed. In table 3 the structure of responses on the observed 17 questions is shown.

According to the results provided in table 3, it can be concluded that only 9% of respondents do not feel morally obligated to pay their taxes (variable TC). On the other hand, 27% of respondents agree and 42% of respondents completely agree that they feel morally obligated to pay their taxes. This results in the highest mean value among the observed variables. According to this, it can be concluded that the vast majority of respondents do comply with tax regulations.

However, if the economic variables related to taxes are observed, in general, it can be concluded that the majority of respondents disagree with the economic dimension of compliance. The only economic variable at which the majority of respondents agreed is variable EF4 (mean = 3.74).

Table 3
Structure and main descriptive statistics of responses on tax-related questions, n = 390

If the psychological variables are observed it can be noticed that the highest average agreement is achieved at variables PF11 (mean = 3.92), PF1 (mean = 3.79) and at variable PF9 (mean = 3.46). The most disagreement level respondents have shown at variables PF7 (mean = 2.60) and PF8 (mean = 2.64).

In graph 1 the mean response values between the tax compliance variable and economic and psychological variables are presented. The economic factor (EF) mean value was calculated by taking into account all answers on five economic variables. To calculate the psychological factor (PF) mean, all respondents’ answers on the 11 psychological variables were observed.

Graph 1
Comparison of mean response values on the tax compliance variable and on economic and psychological variables, n = 390

According to graph 1, it is noticeable that the highest average agreement level is present at the tax compliance (TC) variable. On the other hand, the respondents have shown the lowest average agreement level with the economic variables.

In table 4 polychoric correlations among the tax-related questions are shown.

Strong correlations exist where the polychoric correlations are under -0.50 or above 0.50 or are present only in four cases. Variable TC has a strong positive correlation with variable PF1 (polychoric correlation = 0.72) and with variable PF9 (polychoric correlation = 0.68), respectively. Also, there is a strong positive correlation between variables PF1 and PF9 (polychoric correlation = 0.59). In the fourth case, the strong positive correlation is between variables PF3 and PF4 (polychoric correlation = 0.52).

Table 4
Structure and main descriptive statistics of responses on tax-related questions, n = 390
Correlation matrix, polychoric correlation, tax-related questions, n = 390 respondents

4.2 CHI-square test of independence

In this chapter, the Chi-square tests of independence between the observed taxrelated questions and selected demographic variables were conducted. The sociodemographic variables taken into account were respondents’ gender, gross income, and settlement type. In that way, it was possible to inspect whether the respondents’ tax compliance attitudes towards related questions were different according to their socio-demographic characteristics. The relationship between the tax compliance attitudes on observed questions and respondents’ gender is observed in table 5.

Table 5
Chi-square test of independence, attitude on tax-related questions vs. gender

According to the results in table 5, the null hypothesis of the Chi-square test of independence was rejected in only two cases at a significance level of 0.05. This occurred with variables PF1 and PF2. Therefore, it appears that there is a relationship between variables PF1 and gender, as well as variables PF2 and gender. To better understand these relationships, mosaic plots for these two cases are presented in graph 2.

Graph 2
Mosaic plots, attitude on tax-related questions vs. gender

According to graph 2, males tend to disagree more with PF1 (I would feel guilty-bad if I did not pay my full share of taxes) than females. That would, at the same time, mean that females tend more to agree with PF1 than males. So, females tend more to feel guilty if the taxes are not paid in full than males. Additionally, females tend to disagree more with PF2 (Working for cash-in-hand payment without paying tax is a trivial offence) than males. Due to the fact that the Chi-square tests of independence conducted have shown no statistically different attitudes toward the economic factors of tax compliance but they have shown statistically different attitudes toward two psychological factors, the first research hypothesis H1, that there are no statistical differences in attitudes toward the economic and psychological factors of tax compliance between taxpayers of different genders, can be only partially accepted.

In table 6 the relationship between observed tax-related questions and respondents’ gross income is inspected.

In two cases the null hypothesis of the Chi-square test of independence, at a significance level of 0.05, was rejected. The first case appeared at variable EF3 whereas the second case appeared at variable PF9.

Mosaic plots for cases when the null hypothesis of the conducted Chi-square test of independence was rejected, when attitudes on tax-related questions and gross income were compared, are shown in graph 3.

Table 6
Chi-square test of independence, attitude on tax-related questions vs. gross income
Graph 3
Mosaic plots, attitude on tax-related questions vs. gross income

According to the mosaic plot, given in graph 3, it seems that respondents with medium gross income tend more to disagree with EF3 (If tax rates are too high tax evasion is morally acceptable). On the other hand, the respondents with low gross income tend more to agree with EF3. If a relationship between variable PF9 (I would pay taxes even if there were no tax audits) and the gross income is observed, based on the mosaic plot in graph 4, it can be concluded that respondents with low gross income tend more to disagree with PF9 whereas respondents with high gross income tend to agree with PF9. In that way, the second research hypothesis H2 that taxpayers with different income levels do not have statistically different attitudes toward the economic and psychological factors of tax compliance, should be rejected. Namely, variable EF