National culture as a moderator between social norms, religiosity, and tax evasion: Meta-analysis study

Abstract The objective of this research is to present the meta-analysis results of the relationship of social norms and religiosity with tax evasion, with national culture as a moderating variable. Previous researches show mixed results of the relationship between social norms, religiosity, and tax evasion. Meta-analysis is considered a way out of saturated and inconsistent research since it provides an efficient and systematic approach to make a robust conclusion. This research synthesizes 54 results of 14 individual articles published from 1989 to 2017. This study shows three main findings. Although social norms are not effective to fight tax evasion, religiosity is a useful instrument to decrease tax evasion. Theoretically, national culture plays a pivotal role as a variable to moderate social norms with tax evasion. However, this variable can not moderate religiosity with tax evasion. These findings show that the Theory of Planned Behavior inconsistently predicts the influence of social norms on tax evasion. Practically, tax authorities should consider involving religious leaders to encourage the awareness of tax compliance and to decrease tax evasion.


PUBLIC INTEREST STATEMENT
Tax evasion phenomenon is an important topic that needs to be explored comprehensively. This study employs a meta-analysis to synthesize the empirical research results on tax evasion from 1989-2017, particularly on the fiscal paradigm of taxpayers' psychology. This paradigm focuses on the aspects of social and religious norms which influence taxpayers' behaviors and also Hofstede's national culture as the moderator variable. This study presents a robust conclusion that social norms are not effective to reduce tax evasion, but religiosity plays a useful role in reducing tax evasion. This study also finds that Hofstede's cultural dimensions play a fundamental role as the moderator variable between social norms and tax evasion, in contrary to religiosity and tax evasion moderating result. Theoretically, this result may enrich the tax evasion literature and particularly contribute to the formation of a tax evasion model. Practically, the results may be used to evaluate tax policies, specifically the development of an anti-tax evasion system.

Sutrisno T
Muh Dularif evasion. However, this variable can not moderate religiosity with tax evasion. These findings show that the Theory of Planned Behavior inconsistently predicts the influence of social norms on tax evasion. Practically, tax authorities should consider involving religious leaders to encourage the awareness of tax compliance and to decrease tax evasion.

Introduction
Tax evasion is a serious problem in many countries. Tax evasion is defined as an attempt to reduce tax payment in violation of the law (Chen & Tsai, 2018;Franzoni, 1998). This act causes a loss of state revenue from the tax sector (Mazur et al., 2007). In the long term, tax evasion causes loss of revenue of about 1% of the Gross Domestic Product (GDP) of the member countries of the Organization of Economic Cooperation and Development (OECD) and 1.5% of the GDP of non-OECD countries (Crivelli et al., 2015). A further consequence of reduced tax revenue is the government's weakened ability to provide public goods and social facilities (Marandu et al., 2015).
In the academic realm, the tax evasion phenomenon is an important topic that needs to be explored comprehensively (Abraham et al., 2017;J. Alm et al., 2019;Mahaputra et al., 2018). It is difficult to observe an event of tax evasion, so there is not much empirical evidence that shows the results of research on the determinants of tax evasion. The researchers consider that the process of identifying the determinants of tax evasion is a daunting task given the limitations of highquality data (Khlif & Achek, 2015). In addition, the determinant of tax evasion occurrence and the tax evasion amount is not fully understood (Abraham et al., 2017). As a result, the tax compliance variable is more popularly used by most researchers than the tax evasion variable.
The community and its environment are an essential factor since they influence taxpayers' decision to comply or be involved in tax evasion (J. Alm et al., 2012). This condition called social norms that are the social aspects of psychology that determine whether a person will obey or not. This condition calls for social norms, which are of the social aspect of psychology, which determine whether or not a person will obey. This aspect will influence the behavior of his family, group, and community (Hashimzade et al., 2014(Hashimzade et al., , 2015. Consistently with this trend, some researchers begin to pay more intensive attention to their social and environmental conditions as an essential determinant of tax compliance (Abraham et al., 2017;Hokamp, 2014;Sayidah & Assagaf, 2019). The results show that the relationship between social norms and tax evasion indicates mixed results. Some researchers conclude that social norms play an essential role in reducing tax evasion (Alleyne & Harris, 2017;Richardson, 2006), but the remaining studies show the opposite results (Petee et al., 1994;Wenzel, 2004).
Another psychological aspect which influences the level of tax evasion is the individual's spiritual level (Benk et al., 2016(Benk et al., , 2015. Based on divine authority, such religiosity is unique since it is simultaneously the source of three aspects, namely personal trust, social life, and even a source in formal legal issues, both directly and indirectly (Mackie et al., 2015). In line with social norms, various empirical studies examining the effect of religiosity on tax evasion also do not present direct conclusions. Some studies show that by adopting a particular religion or having specific obedience, there is a tendency for tax evasion to be lower (J. Alm et al., 2016;Grasmick et al., 1991;Richardson, 2008). Nevertheless, some studies indicate the opposite result (Cason et al., 2016), or unclear relationship (Boone et al., 2013;Torgler, 2004).
Starting from the inconsistent results of the relationship between social norms, religiosity and tax evasion, this study uses a moderator variable to strengthen the influence of social norms and religious variables on tax evasion. One potential moderator variable is culture, as reflected in Hofstede's cultural dimensions (Hofstede et al., 2010). In some studies, culture serves as a mediator to understand and explain the differences in use and the effectiveness of practices (Doering et al., 2019;Strielkowski & Cabelkova, 2015), including in taxation practices. Hofstede (1980) finds there are fundamental differences in habits and behaviors, which are specific to each country influenced by their respective cultures. Previous research takes the national culture aspect of Hofstede (1980) as the antecedent which influences tax evasion (Richardson, 2008;Strielkowski & Cabelkova, 2015;Tsakumis et al., 2007). Thus, national culture becomes one interesting topic which influences taxpayers' behaviors. This research employs the four fundamental dimensions of Hofstede's national culture, which are power distance, individualism, masculinity and uncertainty avoidance, to empirically moderate the relationship of social norms and religiosity with tax evasion. One more dimension was added in 2010, namely long term orientation (Hofstede et al., 2010), but is not discussed in this study since it is not really relevant to this topic.
The objective of this study is to synthesize research findings related to the influence of social norms and religiosity level on tax evasion. Based on the search results of many scientific publications, there is no synthesis research focusing on the relationship of social norms and religious with tax evasion. Blackwell's (2007) research employing meta-analysis attempts to synthesize 20 experimental studies, but the factors studied are audit, sanctions and the availability of public goods and their correlation with tax compliance. Other researchers employing the same method also synthesize the effect of socio-demographic factors to determine tax compliance level (Hofmann et al., 2017). Also, there is a tendency that meta-analysis only focuses on primary research with a single research method. For example, Blackwell (2007) focuses on experimentalbased research, while Hofmann et al. (2017) use survey-based research material. To fill the research gap, this research reviews and synthesizes various studies using meta-analysis related to the relationship of social norms and religiosity with tax evasion. We use raw materials in the form of primary research based on experiments, surveys, and secondary data.
As a synthesis research, this study aims at finding a robust conclusion related to the influence of social norms and religiosity on tax evasion. To assess the previous findings which are often inconsistent or conflicting, this study employs a meta-analysis as a statistical method to integrate results from different studies and identify the relationships appearing in the previous studies on the same topic. Therefore, the previous findings, whether they are significant or not, with positive or negative impact, are treated equally as long as they fulfill the inclusion criteria as the research sample. However, weighting samples is conducted to improve accuracy. This study puts higher weight to large sample studies than to the small ones. Consistently with Hunter & Schmidt (Hunter & Schmidt, 2004., p. 82), this frequency-weighted average method is the best to find a robust conclusion regardless of the condition of the variance of population correlation across studies.
By synthesizing the 54 findings of 14 individual articles, this study makes a different conclusion for social norms and religiosity. Social norms do not significantly reduce tax evasion level. On the other hand, the better the religiosity level, the lower the potential of an individual to be involved in tax evasion is. Besides, heterogeneity analysis shows that the two components of high culture, namely individualism and uncertainty avoidance, serve as the moderator variables in the relationship between social norms and tax evasion. Meanwhile, in the relationship between religiosity and tax evasion, the four cultural dimensions, namely power distance, individualism, masculinity, and uncertainty avoidance, do not evidently serve as the moderator variables. Therefore, this research provides a convincing (robust) conclusion related to the influence of social norms and religious level on tax evasion.
Theoretically, this study contributes to the existing literature in several ways. First, the metaanalysis results provide conclusive evidence regarding the influence of social norms and religiosity on tax evasion. These results may simultaneously reduce debate among the previous studies with contradictory results. Second, this result also confirms that the Theory of Planned Behavior does not entirely, consistently predict the influence of social norms and religiosity on tax evasion. Third, the results of heterogeneity studies show that the national culture aspect may strengthen the relationship between social norms and tax evasion. This provides guidance for future research to consider including the Hofstede's national culture aspect into tax evasion related research. Finally, the research contributes to the formation of a tax evasion model, specifically the tendency of individual taxpayers' behavior in fulfilling tax obligations. Practically, the results of the study may be used to evaluate tax policies, specifically the development of an anti-tax evasion system. This is an important concern in policy making to identify the factors behind tax evasion. Regulators should also consider the existing cultures in a country. A policy may be effective in one country, but ineffective when it is applied to another country, given the different cultures.
In the next section, this paper will explain the background of theory and literature review related to tax evasion. Section 3 will describe the research methodology. The results will be presented in section 4, followed with the discussion of and analysis on the results. Finally, the conclusions are presented in section 5.

Theory of planned behavior
One fundamental theory which explains the influence of social norms and religiosity on tax evasion is the Theory of Planned Behavior. The Theory of Planned Behavior (TPB) is popularized by Ajzen (1991). TPB is the development of the Theory of Reasoned Action (TRA) proposed by Fishbein & Ajzen in 1975. TRA states that a person tends to have a behavior based on the results of his evaluation that an act will result in something positive and beneficial. This TRA assumes that individual behavior is not a compulsion and is entirely within the individual self-control. However, further findings show that this behavior is not always voluntary and under control. Therefore, Ajzen (1991) develops TPB, which is the refinement of the TRA.
Planned behavior refers to an individual behavior which is planned based on information he receives or his perception of existing conditions or possibilities that will occur (Malik & Younus, 2019;Owusu et al., 2019;Puspita et al., 2016). Referring to TPB, individual behavior will be determined by the intention depending on three essential things: First, individual attitude related to action (attitude); second, subjective norms inherent in the person (subjective norms); and third, the possibility of supervision of and control over the behavior that he will do (perceived behavioral control; Ajzen, 1991;Langham et al., 2012;Onu & Oats, 2014). Here, social norms and religiosity play a fundamental role in formulating the three aspects to determine the intention. Although there is possible interactions between religiosity and social norms, religiosity tends to shape attitudes and subjective norms. On the contrary, social norms tend to influence the formation of perceived behavioral control.

Social norms and tax evasion
The social aspect of psychology takes part in deciding whether a person will obey or not (Ajzen, 1991), such as social norms. Social norms are defined as social-psychological construct referring to the agreement between related parties in a community or group in relation to commands and prohibitions which bind the community members without formal rules and law enforcement on orders (Bobek et al., 2011(Bobek et al., , 2013. This aspect is influenced by the behavior of their family, group, rust and community where they live towards tax obligations (Benk et al., 2016(Benk et al., , 2015. More specifically, social norms may be classified into subjective norms, injunctive norms, and descriptive norms (Bobek et al., 2013;Mackie et al., 2015). The first construct, subjective norms, refers to individual perception regarding the views or behaviors of friends or family related to tax. The second construct, injunctive norms, is the individual opinion of the conduct or actions of colleagues in the same group with the same income or social class. The third construct, descriptive norms, refers to individual perception of public perspective in general on tax. If a close friend or family member tends to disobey or a group of colleagues generally hides their income, the person will also tend to violate it. If the public in general considers tax evasion efforts as something standard instead of illegal or unethical, there is a tendency of embezzlement rate to increase (J. Alm et al., 2017).
The researches show that the relationship between social norms and tax evasion has mixed results. Some researchers conclude that social norms play an essential role in reducing tax evasion (Alleyne & Harris, 2017;Richardson, 2008), but several other studies show the contrary (Petee et al., 1994;Wenzel, 2004). However, rationality indicates that if a person is surrounded by people who tend to be tax-compliant, or his business colleagues tend to pay taxes appropriately, it is less likely that he will embezzle taxes.
Social norms will form a framework which will influence decision on taxes. Public awareness to pay taxes is an important element to reduce and prevent fraud. Some recent research shows that better and more positive social norms exist in society (Dwenger et al., 2016;Hallsworth et al., 2014). There is an assumption that paying taxes is an obligation that one must obey, and if a person embezzles taxes, he will be considered as an immoral member of society. This assumption will certainly increase tax compliance or may be assumed to reduce tax evasion level. In general, social norms are deemed good and positive if the social conditions encourage people to more obediently fulfill their obligations. From the description above, the hypothesis is: H1: The better the view of social norms on taxes, the lower the tax evasion level is.

Religiosity and tax evasion
Besides the social environment, another psychological aspect that influences individual behavior related to tax is the individual level of religiosity. Religiosity is very instrumental in describing the attitudes and behavior of individuals in an organization and daily life (Farrukh et al., 2016). At the same time, religiosity will shape individual behavior and attitudes for him not to focus solely on gaining profit. In this case, religiosity is a unique concept since it is considered capable of stabilizing social systems in society (Dodd & Gotsis, 2007).
Religiosity consist of three characters, namely elements of personal trust, issues of social life and a source in formal legal problems, both directly and indirectly (Mackie et al., 2015). As a source of personal inspiration, religion consists of commands, prohibitions, rewards and sins for its followers. As an inspiration for the social aspects, the three celestial religions, which are Judaism, Christianity, and Islam, do not only regulate relationship with God but also regulate relationship between humans by banning greed and ordering to help the poor (Kepes, 2016;Kunkle, 2008). As an inspiration in the law sector, several countries such as Iran, Saudi Arabia, and Vatican City use religious codes as their favorable state law (Kishi et al., 2017).
In line with social norms, various empirical studies examining the influence of religiosity on tax evasion also do not present robust conclusions. Some studies show that adopting a particular religion or having unwavering obedience tends to lessen tax evasion (J. Alm et al., 2016;Richardson, 2008). Meanwhile, some other studies show the contrary (Cason et al., 2016) or indicate unclear relationship (Boone et al., 2013;Torgler, 2004). Religiosity may indeed help individuals to get through certain difficult times. However, at the state level, individuals need more than just religiosity. Good quality of social (state) environment such as effective and transparent government services and good efforts to control corruption may reduce tax evasion level. However, logically, when religion teaches honesty and encourages people to do charity and good deeds with others, individual attachment and obedience to his religion should have encouraged him not to be involved in tax evasion. From the description above, the hypothesis is: H2: The higher the level of individual religiosity, the lower the tax evasion level.

Social norms, religiosity, national culture, and tax evasion
One theory of the cultures across countries elaborates through the concept of Hofstede's cultural dimensions. Culture is defined as the collective thinking in the mind of a society which distinguishes the members of a community with other parties (Hofstede, 1980). Culture strongly influences not only the determination of individual perception, psychology and behavior, but also his values and beliefs. Cultural influence determines not only society's daily routine but also the relationship between citizens and government, including taxes (Richardson, 2008;Tsakumis et al., 2007). National culture may also encourage or inhibit tax avoidance or tax evasion behavior, regardless of individual taxpayer's characteristics or other institutional components (Bame-Aldred et al., 2013;Dularif et al., 2019).
In relation to the possibility of a cultural dimension serving as the moderator variable, there is a tendency that in countries with high power distance, the dependence on those who have power tends to be extensive (Hofstede et al., 2010). In countries with high power distance, the social influence on society members tends to be more intensive compared to that in countries with low power distance. In countries with high power distance, the social norms are more positive towards taxes and a higher level of religiosity. The presence of good leaders will encourage society to be more compliant, so tax evasion is lower. On the other hand, in countries with high power distance, there is a tendency that the taxation system favors the ruling party and is deemed to cause injustice to the public in general (Richardson, 2008). Therefore, although the social conditions are generally favorable, disillusionment with the existing system will encourage people to be disobedient and increase tax evasion level. Even in highly religious countries, an unjust system encourages people to be disobedient since they feel that there is an insecurity of the tax management.
Individualism shows how strong the bond between one individual and other individuals in a community is. Individualism is often associated with the opposite, namely collectivism. A high individualism index indicates that a culture is more concerned with individual interest than the common interest, while in a country with a low individualism index, collective interest is considered far more important than personal interest (Hofstede et al., 2010). The respect for individual rights in a country with high individualism tends to encourage the state to emphasize equal right for every individual. This equality of rights helps law to be enforced indiscriminately. Therefore, when existing social norms support stronger obedience to tax regulations and reliability, law enforcement will indiscriminately and reasonably be an essential aspect to reduce the level of tax evasion. On the one hand, law enforcement will better suppress the level of tax evasion. On the other hand, individual interest in a country with high individualism also allows one to make every effort to find a loophole in tax regulations, which may in turn encourage fraud to a higher level.
Masculinity refers to how significant the role and position of gender in society. In a country with high masculinity index, men are deemed to be the superior, performance-oriented and material success. Low masculinity indicates a tendency and orientation towards shared prosperity and concern about others (Hofstede et al., 2010). On the one hand, high appreciation of performance in a country with high masculinity encourages people not to tolerate fraud. When social norms encourage people to obey, and religious teachings prevent people from deviating in a country with high masculinity, tax evasion tends to be lower. On the other hand, the tendency of orientation to material success in countries with high masculinity may encourage people to engage in actions which may harm those around them, even though the social environment and religious forbid it.
High uncertainty avoidance index shows a high level of public anxiety over uncertainty. There is a tendency to make more formal laws to reduce such uncertainty. Consequently, the legal system tends to be complicated (Hofstede et al., 2010). Many official rules, on the one hand, provide the public certainty to fulfill their tax obligations. Therefore, a positive social environment and strong religion may encourage taxpayers to be more obedient. On the other hand complex rules often overlap one another. Consequently, the tax non-compliance level tends to increase (Richardson, 2008;Tsakumis et al., 2007). From the description above, the hypotheses are: H1a: Power distance moderates the relationship between social norms and tax evasion.
H1b: Individualism moderates the relationship between social norms and tax evasion.

H1 c: Masculinity moderates the relationship between social norms and tax evasion.
H1d: Uncertainty avoidance moderates the relationship between social norms and tax evasion.
H2a: Power distance moderates the relationship between religiosity and tax evasion.
H2b: Individualism moderates the relationship between religiosity and tax evasion.
H2 c: Masculinity moderates the relationship between religiosity and tax evasion.
H2d: Uncertainty avoidance moderates the relationship between religion and tax evasion.

Research method
This research employed the method proposed by Hunter and Schmidt (2004) as its primary approach, mainly to calculate the best estimate of the population correlation (� r) and to detect the moderator variable. Following Hunter & Schmidt (2004, p. 82), this research determined the influence of social norms and religiosity as the independent variables on tax evasion, as the dependent variable based on the significance of (� r) value. According to Hunter & Schmidt (2004, p. 415), this study identified the existence of a moderator variable based on the analysis of Q statistic based on a chi-square distribution with k-1 degrees of freedom of ðX 2 kÀ 1 ).

Study criteria
This study employed a meta-analysis to analyze the research hypotheses. Meta-analysis is a powerful instrument to solve a problem, where previous studies provide inconsistent conclusions, particularly related to factors which determine tax evasion (Marandu et al., 2015). The previous findings are summarized, analyzed and integrated to reexamine the relationship between the independent factors and the dependent variable using the meta-analysis.
There were keywords to search for the primary (individual) articles as the study materials, which was the first step in this research. The next step is to utilize electronic databases, such as Proquest, EBSCO, Emerald, Science Direct, JSTOR, and Wiley. For items on the topic of the relationship between social norms and tax evasion, the keywords are "social norms (or subjective norms, injunctive norms and descriptive norms) and tax evasion". To determine the relationship between religiosity and tax evasion, the keywords are "religiosity (or religion, religious and spiritual) and tax evasion". The related articles with keywords were downloaded and arranged systematically. For identification and further processing, the files of the articles were named in a specific system to show the name of the journal, year, volume and publishing, author, and title of the article.
The objects of this study are the results of previous research which examine the relationship of social norms and religiosity level with tax evasion. A judgment sampling was employed by this research to determine the material of this meta-analysis study. In other words, the objects depend on the researcher assessment according to the pre-established criteria for the research sample. This meta-analysis objects must meet the following requirements: (1) This research used tax evasion as the dependent variable.
(2) This research used social norms or religiosity as the independent variable.
(3) This research was based on quantitative empirical data. The results displayed in statistical figures could be converted into correlation coefficients (r-Pearson).
From the process above, eight articles were obtained with 21 findings related to social norms and six articles with 43 results discussing the religious aspects of public trust. Therefore, there are 14 articles in total, with 54 findings from 1989-2017 selected in this study.

Coding procedure
The information included in coding generally contained identity, information on the dependent variable and information on each independent variable, similarly to previous researches (Lipsey & Wilson, 2001). The general identities included the study identity code, researchers, year of publication, the total number of samples, types of tax, unit of analysis, objects of study and population. The dependent variable included the variable and how it was measured. Meanwhile, the independent variables included the measurement, type of effect size published in the article, samples used to test the independent variables and some statistical data such as correlation coefficients, standard errors or t-statistics/F-statistics/chi-square values. These data would eventually be converted into r-Pearson.

Meta-analyses procedures
In this study, a meta-analysis was employed to synthesize the results of previous studies. Similarly to previous researches (Ellis, 2010;Hunter & Schmidt, 2004), after collecting and coding of the articles, which were the research samples, the next steps were as follows: (1) Transform the relevant statistical results of each primary study into a standardized effect size for synthesis. The standardized effect size in this study employed the Pearson coefficient (r-Pearson) (Card, 2015;Hunter & Schmidt, 2004;Littell et al., 2008). The r-Pearson was selected as the measure of conversion under two important reasons. The first reason was that r-Pearson ranges from −1 to 1, thus intuitively, the r-Pearson value may be easily understood. The second reason was the ease of conversion, both from other forms of statistic to r-Pearson and vice versa. If a study displays the r-Pearson value, then that number may be directly used. However, only if it shows other statistical measurements, whether in the form of t, F, p or others, the step taken was to convert the statistical size data into r-Pearson with a formula commonly used in meta-analysis (Card, 2015). (2) Determine the population mean of correlation (r ̅ ) using the formula: Where n i denoted the number of samples, and r i denoted r-Pearson of study i.
(3) Calculate the statistical significance of the mean, which was obtained by calculating the variance of the sample of correlations (V r ), standard error (SE r ), and Z score using the following formula: (4) Test the main hypotheses (H1 and H2). If the probability of obtaining Z score was less than 0.05 (using the excel formula " = NORM.S.DIST (Z, FALSE)") and within 95% confidence interval, it was to calculate using the method below: If it was not contain zero, the mean effect size was statistically significant.
(5) Test the possibility of heterogeneity using a Q-Statistic test with the following formula: When Q value exceeded the critical value of X 2 kÀ 1 , where k was the total number of findings used in the study, we might conclude that the distribution of effect size was heterogeneous. The existence of this heterogeneity encouraged researchers to find moderator variable.
(6) Identify moderator variables. Potential moderator variables might be obtained based on theoretical justification and rationality (Laroche & Soulez, 2012). In this study, the cultural aspect was one important factor assumed to influence taxpayers' involvement in tax evasion (Richardson, 2008;Tsakumis et al., 2007). To test whether cultural factors played a significant role as moderator, this study performed a regression using OLS of r-Pearson of each finding with potential moderator variable with the following econometric model: Where r showed the r-Pearson coefficient of each finding, x showed a set of continuous variables based on the justification of the theory, and it may be the cause the variability of research results, such as cultural aspects in Hofstede's theory. To control the influence of noncultural factors, the Z notation, which represented the control variable was included, for example, the type of tax or difference in variable measurement. Besides, there was a tendency for publication to be biased where studies which made significant conclusions were quite likely to be accepted and published in journals (Abdullah et al., 2015;Anderson et al., 2018). To minimize this, Anderson et al. (2018) proposed to include the Standard Error (SE) of each finding as a precision adjustment instrument. Significant regression results would be an indication that the concerned variable was likely to be a moderator variable (Anderson et al., 2018).

Results and discussion
One common thing in tax research is that many papers have multi-independent variables and multiple findings. If one study contains more than one finding, all of the findings will counted as separate research findings (Dochy et al., 2003). The details of each paper used along with the number of results are presented in Table 1.
The results of the meta-analysis are designed to obtain robust conclusions between social norms and religious norms, as presented in Table 2. Table 2 shows positive but insignificant relationship between social norms and tax evasion (r = 0.0696; p (Z) = 0.064; 95% CI = (-0.001; 0.140)). In other words, the first hypothesis (H1) is not sufficiently supported by the meta-analysis results in this study. Therefore, we may  conclude that positive social norms do not necessarily reduce the level of tax evasion. On the other hand, Table 2 shows significant, negative relationship between religiosity level and tax evasion (r = −0.0226; p (Z) = 0.006; 95% CI = (−0.037; −0.007)). The synthesis results show fairly convincing (robust) support for the second hypothesis (H2). Therefore, we may conclude that taxpayers' higher spirituality level is an effective inhibitor from tax evasion attempts.
The existence of heterogeneity encourages researchers to investigate the source of heterogeneity through the moderator variables analysis. In other words, some variables moderate the variability of the relationship of social norms and religiosity with tax evasion. To further test possible potential moderator variables, r-Pearson of each finding will be regressed with the potential variables. The results of the regression related to these cultural aspects are presented in Tables 4 and 5. Table 4 shows that the coefficient of individualism and uncertainty avoidance is significant, while that of power distance and masculinity is insignificant. Therefore, only H1b and H1d hypotheses are supported by the results of this meta-analysis. Meanwhile, by using religiosity aspects, Table 5 shows that there is no significant coefficient among the four dimensions. It means that the results of this study do not support the H2a, H2b, H2 c, and H2d hypotheses. In other words, cultural aspect is not a moderator variable which may influence the relationship between public trust and tax evasion.  The interesting thing in this research is related to social norms. The influence of positive social norms on the reduction of tax evasion is not only insignificant but also not as predicted by the Theory of Planned Behavior. On the contrary, the effect of religiosity on tax evasion is in line with the theory of planned behavior's predictions and also significant.

Social norms as determinant of tax evasion
The data in Tables 2 and 3 related to the influence of social norms on tax evasion show that when social norms are associated with tax evasion aspect, the results are evidently not following the Theory of Planned Behavior. The results of this study indicate that better social norms do not significantly influence the efforts to reduce tax evasion. The average r-Pearson coefficient shows a positive number. The role of social norms in improving compliance or reducing tax evasion is not reliable. This is an anomaly that better social norms should have reduced the level of tax evasion, but in this case, the study shows different results.
Theoretically, according to the Theory of Planned Behavior, social norms play a significant role in increasing tax compliance or reducing tax evasion. This is confirmed by many psychologists' statements that most individuals tend to need other people and do not like solitude (alienation). One effort to keep acknowledged in one's environment is to avoid different attitudes from others around him. It is this tendency not to be different, which makes the social environment influence an individual attitude, including tax compliance. In a positive social environment (family, group, community), tax compliance is an obligation. Thus, tax evasion behavior tends to be unacceptable (Alleyne & Harris, 2017;J. Alm et al., 2017;Mackie et al., 2015).
However, this study result is not consistent with the Theory of Planned Behavior. There are several underlying possibilities. First, methodologically, the survey articles mapped in this study are likely to experience biased responses. In this meta-analysis study, most of the research samples use a questionnaire or survey to measure social norms. The use of a survey or poll indeed may present data from the intended respondents directly. However, their answers are not pursuant to the reality or truth, but considered useful or correct by the public in general. As a result, there is a possible subjectivity or even bias. This finding is what makes Brizi et al. (2015) and Takeuchi et al. (2015) concerned, that biased response may allow anomaly, thus the study result does not support the theory underlying the research.
Second, as an external factor, it is possible that social norms can not be a stand-alone factor to reduce tax evasion. Social norms may reduce tax evasion only when there are other actions implemented by tax authority such as auditing and imposition of penalty. In a research which uses questionnaire to measure one's level of closeness or social concern or uses non-reported income as a measure of tax evasion, Wenzel (2004) shows that when social norms serve as a stand-alone independent variable, either using a correlation measure or regression, social norms consistently show a positive coefficient. In other words, social norms increase tax evasion. However, in interaction with strict sanctions, social norms function well as an aspect to reduce tax evasion level.
In most democratic countries that use a self-assessment system, what is paid and reported by a taxpayer is considered correct, unless the tax authority has proof that the taxpayer is guilty. In other words, tax evasion can be identified only if an audit has been carried out by the tax authority. Although environmental and social norms consider tax evasion to be unethical or even illegal, it will not be identified by or known to the public without supporting actions by tax authority such as auditing. This finding supports Wenzel (2004) conclusion.
Third, social norms do not influence tax evasion because of taxpayer's independent attitude. When deciding to do risky actions such as tax evasion, a person is influenced more by his own beliefs and values. Social norms are external factors that need to be internalized in taxpayer when making decisions. It is also possible that when a respondent is asked about social aspects, his answer is something that is actually of his personal interest. Sometimes, when taking illegal actions, one's attitude is evident from the beginning, and social factors only complement his consideration without influencing much of his original attitude. This finding is consistent with what is emphasized by Sigala et al. (1999), warning that personal and social aspects often are closely linked and influence each other.
The test result shows that culture is a moderating variable in the relationship between social norms and tax evasion. Although deemed ineffective, there are two cultural dimensions moderating the relationship between social norms and tax evasion, namely individualism and uncertainty avoidance (see Table 4). The cultural factors reflected in individualism play a significant role in shaping the relationship between social norms and tax evasion. Table 4 shows that the coefficient is negative. High individualism index is characterized by people's attitudes that tend not to tolerate fraudulence, which may harm other community members (Hofstede et al., 2010). Moreover, when combined with more strict controls and positive social norms, that attitude will be stricter (Rustiarini et al., 2019). As a result, in a country with high individualism, better and positive social norms will pose even more significant obstacles for someone who intends to evade tax. It is indicated by the negative individualism coefficient, that when social norms are more positive about tax, tax evasion is lower in a country with high individualism index.
The next cultural factor which serves as a moderator variable in the relationship between social norms and tax evasion is uncertainty avoidance. The uncertainty avoidance aspect, as shown in Table 4, shows a positive moderating role. A high uncertainty avoidance index shows a high level of public anxiety over the threat of uncertainty. Besides, in a country with a high index of uncertainty avoidance, it is also characterized by a high distrust in the government in general (Hofstede et al., 2010) and complicated regulations. As a result, the tax non-compliance level tends to increase (Richardson, 2008;Tsakumis et al., 2007). Therefore, when social norms get better but not improvements at the government level, uncertainty will increase. The taxpayers will tend to be involved in tax evasion to reduce uncertainty (Richardson, 2008;Tsakumis et al., 2007).
This study result has theoretical (academic), practical and policy implications. At the theoretical level, the result of the meta-analysis showing the inconsistency of empirical results with the Theory of Planned Behavior implies that researchers need to carefully map survey research. Although this type of research has the advantage of collecting direct answers from respondents, we can not deny that the answers given may be subjective and biased (Takeuchi et al., 2015). Moreover, it is difficult to measure social norms variable. Respondents often find it is difficult to distinguish whether the answers they give are their personal opinions or are influenced by social norms in the society. This leads to different results from expectations or theories.
In practice, the results encourage the government to create programs that are adapted to taxpayers' social environment. For example, when conducting socialization to raise taxpayers' awareness, the government should show how taxes are used for social security, health insurance, or education subsidies. This step may effectively make people aware that tax is intended for social programs so that taxpayers will fulfill their tax obligations.
In addition, the government must establish positive social norms in the community. It is not only under the control of the tax authority or the government but must be fully supported by the wider community. This process takes a long time, maybe even through several generations. The long history of taxes, especially one that tends to have negative connotations (Adams, 2001), poses the toughest challenge for the government to establish positive social norms. Therefore, the policies made must consider taxpayers' psychological aspects to create better and positive social norms. Besides social aspects, national culture also is an important factor in explaining the relationship between social norms and tax evasion. This is a major concern for policymakers to always consider national cultural values in designing tax administration systems.

Religiosity as determinant of tax evasion
The results of the test on the influence of religiosity and tax evasion are shown in Tables 2 and 3. The results of synthesis in this meta-analysis show that religiosity positively influences tax evasion. These results are as predicted by previous studies which make religiosity as their main focus of research (based on religiosity recognition or obedience to perform religious orders; Benk et al., 2015;Eiya et al., 2016;Palil et al., 2013;Strielkowski & Cabelkova, 2015). This phenomenon is intended to emphasize that religion plays an important role in preventing its followers from breaking the rules, both in Western countries and in Asia (Brauer et al., 2013). Therefore, religiosity plays a relatively important role in reducing tax evasion.
Reviewing at the theoretical level, religiosity means one's sentimental attitude towards religion and its influence on human existence and reality in general. From this perspective, religion and religiosity are factors that determine the conception of society and humans within it (Akah & Ajah, 2020). Religious aspects may be seen from the level of religiosity. A person's involvement in a particular religion may be measured with his obedience to perform his religion orders (Benk et al., 2016;Boone et al., 2013;Torgler, 2006;Weaver & Agle, 2002). Religion contains commands, prohibitions, rewards and sins. Besides offering its followers hope and heaven, religion also imposes threats in the form of sin and hell on offenders. This divine authority is a very important internal motivator for individual behavior (Mackie et al., 2015).
With regard to the influence of religion on individual behavior, religion offers hope to its followers. If religions are implemented in daily life (including in social life), religion will give its followers a noble identity (Akah & Ajah, 2020;Weaver & Agle, 2002). One who believes in and does good deeds according to what God commands will get happiness in the form of entering heaven. This will encourage someone to obey the law, including the obligation to pay taxes. Then, the sinner who breaks the prohibition will go to hell. This doctrine will certainly be able to prevent a religious person from violating the law (Weaver & Agle, 2002), including tax evasion. Ideally, the higher one's religiosity level, the greater the influence of religion in shaping his characteristics and behavior, including taxation. Religion that teaches to share with others, avoid selfishness and condemn fraud or violation of the law will significantly reduce efforts to get involved in tax evasion.
Analyzing the moderating factor in the relationship between religiosity and tax evasion is an interesting thing. The results of the heterogeneity analysis in Table 5 show the absence of national culture as a moderating variable between religiosity and tax evasion. One answer to this anomaly is religion's unique character, position and function. Religion does not only serve as a personal norm, but also influences social, legal, and political norms (Mackie et al., 2015). As a personal norm, religion offers hope in the form of rewards for good deeds and the threat of sin for evildoings and wrongdoings. Besides the offer of heaven and threat of hell in return for personal merit and sin, many religions also have universal teachings in the form of caring for others and disagreement over greediness. Moreover, various religious explicitly order their followers to share with others, for example, zakat in Islam or a tenth in Christianity. In other religions, such as Judaism, Hinduism or Buddhism, there are also teachings of sacrifice not only for God but also for others (Kepes, 2016;Kunkle, 2008;Maham & Bhatti, 2019).
In the field of law, in some countries, religious are taken as an essential source of and inspiration for the law. There are even countries that make religion a positive law, such as Saudi Arabia and Iran (Kishi et al., 2017). Making religious teachings a positive law means there is religion's direct involvement in state politics. Certainly, the influence of belief as personal, social, legal and political norms will be understood and responded differently, even by the same followers. This causes differences in the followers' behavior of the same religion even though the knowledge taught tends to be the same (Weaver & Agle, 2002).
The uniqueness of religion and its followers' different responses may be the cause; religion becomes independent of the influence of any other factors, including culture. Religious elements are strong enough to influence their followers' behaviors. The research findings are in line with some other studies (Alm & Torgler, 2006;Benk & Budak, 2012;Mackie et al., 2015). The difference in the level of faith may be the only one to lead to different behaviors (Weaver & Agle, 2002). In other words, the clearer a person's identity is related to a particular religion and the higher individual obedience, the lower the level of tax compliance.
These research have theoretical, practical and policy implications. In general, religiosity level may reduce tax evasion, but the absence of culture as a moderator variable shows that religion is a unique and independent variable. Theoretically, this research implies that although religiosity may serve as an independent variable, researchers must carefully construct and measure this variable. Respondents often experience social desirability syndrome when answering questionnaires (Alleyne & Harris, 2017;Luke & Grosche, 2018). As a result, respondents tend to give normative answers instead of the real ones, especially in religiosity matters. This is not only related to a personal or social relationship with human beings but also related to God. For example, in some religious countries, a person may find it difficult to say that he is not religious despite the fact that he does not believe in the power of God. Likewise, in religious countries, researchers may find it is difficult to find an honest answer that a person will admit that he seldom visits a place of worship. Therefore, a person's religiosity level may be more appropriate to measure with social behavior or attitude, or what is commonly referred as social piety (Bisri, 2016).
In the practical level, although the results indicate that religion is important to reduce tax evasion, literature states that some religious figures still question about the obligation to pay tax. Even in Islam, until now, there are debates related to the fulfillment of obligations other than zakat (Abdullah, 2017;Pakeeza & Qadoos, 2017). Therefore, when religious become an important aspect of daily life, the government needs to cooperate with religious leaders to encourage people fulfilling their tax obligations.
In the policy level, the moderator variable test results show that religion is an independent variable that is not influenced by culture, or it is deemed to be influenced by religiosity variable itself. This implies that the society's religious factor should be the government's main consideration to implement a tax rule, especially when the government is going to adopt a tax rule applicable in other countries. For example, tax policy in a western country, which tends to separate religion from the state, will certainly have different results when applied in a country with a strong religious base.

Conclusions, limitation, and future research
This study employs a meta-analysis to synthesize the results of empirical studies from 1989 to 2017, which discuss the influence of social norms and religiosity on tax evasion. The results of the review and synthesis of the 54 results from 14 individual articles show that social norms do not effectively reduce tax evasion. The meta-analysis result fails to confirm the Theory of Planned Behavior which emphasizes the importance of social norms in influencing taxpayers' behaviors. In the tax evasion context, this shows that the taxpayer's behavior is not influenced by the other (social) behavior. This is inseparable from the role of tax authority that promotes honest behavior in fulfilling tax obligations.
On the other hand, this study presents a robust conclusion that a better level of religiosity plays a useful role in reducing tax evasion. This shows that religious values can effectively reduce individual negative behavior. Religiosity is considered to motivate taxpayers to comply with tax laws. The heterogeneity analysis results show that cultural factors, as reflected in Hofstede's cultural dimensions, serve as the moderator variable which influences the relationship between social norms and tax evasion. In contrast, this study shows that the four cultural aspects do not affect the relationship between religiosity and tax evasion.
This study has several limitations. First, this research only focuses on robust findings related to the influence of social norms and religiosity on tax evasion. These two factors are assumed to influence tax evasion independently. These factors are likely to be interrelated, or one or more variable may serve as the moderator of the other variables in influencing tax evasion. Although experimental research, surveys or secondary data have certain technique or method which may be used to explain this, to the researchers' knowledge, there is no technique or method in a metaanalysis research to accommodate it. It is a future research opportunity to identify other factors which determine tax evasion or identify a variable which serves as the moderator variables.
Second, this research synthesizes a variety of primary studies, one of which measures variables using a survey or questionnaire. One weakness of this type of research is possible response bias, especially related to variables which are sensitive to the social environment where the respondent is located, such as social norms and religiosity variables. Although this research has tried to minimize this type of research by using researches published in reputable journals, the criteria cannot completely eliminate this limitation. Therefore, future researchers should prefer primary researches with a low level of response bias.
Third, this study examines cultural aspects as the moderating variable. The variables which represent the cultural aspects are the four cultural dimensions developed by Hofstede, namely, power distance, individualism, masculinity and uncertainty avoidance. There are several criticisms of the use of Hofstede's cultural dimensions as a representation of cultural variables. The first criticism is related to the time the concept is made in 1980 or about 40 years ago (Orr & Hauser, 2008). The second criticism is related to the use of survey and also those surveyed to measure the index of each cultural dimension (McSweeney, 2002a(McSweeney, , 2002b. Despite these criticisms, until now, there is no cultural concept which comprehensively explains the difference in characteristics between countries and presents quantitative data of Hofstede's cultural dimensions. Therefore, in case of other more complete or better cultural concepts, it will provide future research with the opportunity to use new cultural concepts.