Acceptance of Corrupt Acts: a Comparative Study of Values Regarding Corruption in Europe

Abstract

The present contribution seeks to explain variation in the degree of acceptance of corrupt acts by taking into consideration both individual characteristics and societal ones. We used a large dataset covering 43 European countries and employed multi-level models in order to disentangle the compositional and contextual effects. Our main findings suggest that young single Europeans with no occupation but with material possibilities are more likely to consider corrupt acts as being acceptable. The presence of a partnership and of children as well as high confidence in the governance bodies of a country makes corrupt acts less acceptable. In addition, the society where one lives is also important: individuals living in the former soviet countries display on average higher acceptance of corrupt acts than individuals living in the former communist block or in long established democracies. This conclusion holds also after controlling for how widespread corruption is in these countries or how high their income inequality is. [PUBLICATION ABSTRACT]

Headnote

Abstract

The present contribution seeks to explain variation in the degree of acceptance of corrupt acts by taking into consideration both individual characteristics and societal ones. We used a large dataset covering 43 European countries and employed multi-level models in order to disentangle the compositional and contextual effects. Our main findings suggest that young single Europeans with no occupation but with material possibilities are more likely to consider corrupt acts as being acceptable. The presence of a partnership and of children as well as high confidence in the governance bodies of a country makes corrupt acts less acceptable. In addition, the society where one lives is also important: individuals living in the former soviet countries display on average higher acceptance of corrupt acts than individuals living in the former communist block or in long established democracies. This conclusion holds also after controlling for how widespread corruption is in these countries or how high their income inequality is.

Keywords: Corruption, Values, Communism, Europe

Introduction

It is a generally accepted idea that corruption is one of the main problems that stands against the social and economic development of societies (Mauro 1995; Bardhan 1997; Li, Xu et al. 2000). For instance bribery, a form of corrupt behavior, is considered so harmful that OECD has initiated in 1997 a formal Convention on Combating Bribery of Foreign Public Officials in International Business Transactions, asking for the member states to enact laws and regulations that forbid tax deductibility of bribe (OECD, 2011).The political discussions around the potential harmful impact of corruption on the social, political and economic realms reflected also on the research agenda, inspiring several initiatives that aimed at quantifying the level of corruption and to evaluate its causes and its social effects (Treisman 2000; Sandholtz & Taagepera 2005; You & Khagram 2005; Tavits 2008; Uslaner 2009). Despite the ardent interest in the topic, preference was given to measuring and explaining levels of corruption at country level (Dreher, Kotsogiannis et al. 2007), and relatively little attention was given to studies on how individuals think about corruption and to what degree they find it acceptable or not. For instance, Gatti, Paternostro & Rigolini (2003) reported only one previous study that investigated the differences between individuals in their attitudes toward corrupt acts (Swamy, Knack et al. 2001). We argue that taking an actor perspective is valuable since the acceptance of corruption by individuals, taken as a value orientation, might have potential consequences for individual behavior and also for society at large. For instance, policies targeted at reducing levels of corruption might be hindered or even rendered inefficient if the population has in fact a high level of acceptance of corrupt behavior. In addition, higher acceptance of corrupt acts might also facilitate the engagement in and decrease the level of informal sanctions of corrupt behavior. If so, identifying which groups of population manifest lower or higher acceptance of corrupt behaviors is a valuable piece of information for targeted policies aimed at reducing corruption.

While the individual differences in the acceptance of corrupt acts received some attention from previous studies, scholars also pointed out that the attitudes toward corruption differ between social contexts. For example, Gatti, Paternostro & Rigolini (2003) show that the average acceptance of corruption varies between countries and the authors argue that the social environment has a significant influence on the individuals' attitudes toward corruption. The same argument was brought forward also by Cameron et al. (2009) who, based on an experimental study carried out in four countries, found evidence for the variation across cultures in the propensity of individuals to punish corrupt acts. However, while the theoretical arguments are compelling, the previous literature lacks systematic empirical evidence based on large samples of countries and high quality comparable data that would support the claim that the social context shapes the attitudes toward corruption of individuals.

In the present study we aim at addressing the lacks in the literature that we briefly mentioned in the previous paragraphs, namely we wish to provide an answer to the question of whom and under what conditions individuals accept corrupt practices. We not only look at the characteristics of individuals but we also develop arguments about the influence of the social context where they live in and how this context might shape their acceptance of corrupt acts. We argue thus for the need to integrate individual and contextual characteristics in a broader model aimed at explaining the variation between peoples' values towards corruption. The research question that guides our investigations is: how can we explain the differences between individuals in their level of acceptance of corrupt acts by looking at characteristics of individuals and of the societies? In order to address this question we use the integrated dataset of European Values Studies, wave 2008, who allows us the largest coverage of European countries. We employ multi-level models on a working dataset that covers 62674 individuals nested in 43 European countries.

Explaining differences in the acceptance of corrupt acts

The theories used to derive our hypothesis are derived from two fields: general theories of values (because we define acceptance of corrupt acts as a value orientation) and theories specific to criminology such as social learning theory and desistance theory (because corrupt acts fall within the category of deviant behaviors).

Individual level expectations

When explaining why some individuals have higher levels of acceptance of corruption, the first element that we take into account is religiosity. Corrupt acts imply lie, dishonesty and deceive, breaking thus general accepted rules of conduct in society. Religious denominations agree on the fact that these types of behaviors are not desirable, and such, acceptance of corrupt acts goes against religious norms. Consequently, our first expectation is that individuals with higher level of religiosity will display lower acceptance of corrupt acts (Hypothesis 1).

Another explanation for the variability in the acceptance of corrupt acts between individuals can be derived from the desistance theory, who tries to elucidate the observed curvilinear relationship between age and involvement in criminal acts, i.e., with age the involvement in criminal acts decreases (Laub & Samson 2001). An explanation for this empirical relationship is that "the decline in crime occurs because factors associated with age reduce or change the actors' criminality" (Torgler, 2006, pg. 135). For instance, actors have more to lose as they grow older if the crime is discovered and prosecuted. In addition, research on the relationship between values and age has already shown that older cohorts display higher level of more traditionalist values, such as the religious ones (Inglehart, 1971). Based on these two lines of reasoning we expect that older individuals would have lower level of acceptance of corrupt acts (Hypothesis 2).

The social learning theory (Akers, 1977) provides a framework that emphasizes the role of different associations in shaping the involvement in deviant behavior. According to this theory one of the pathways through which behaviors are learned by individuals is differential reinforcement path which refers to the balance of anticipated or actual rewards and punishments for own acts. We argue that this mechanism might be relevant also for shaping the values of individuals, via a similar type of process based on the anticipated outcomes (potential gain vs. potential loss) of the involvement in corrupt acts of themselves or of their peers. If the own involvement in corrupts acts is taken into account, this comes with the danger of being discovered and of suffering the consequences. The consequences can range from losing status among the peers to being prosecuted, having to pay fines to cover the damage produced, etc. When considering these consequences, individuals from some social categories might perceive these costs as being too high in comparison with the potential win. For instance, individuals that have a higher socio-economic status might perceive a high cost for getting involved in corruption due to the danger of losing the privileges of their position or of losing the esteem of their peers. This cost might be perceived to be not so high for the individuals with low occupational status and low income, who might see the potential gains (e.g., material resources or jobs) as having more weight. These arguments lead to the expectations that individuals with higher socio-economic status in society will display less acceptance of corruption than individuals with lower socio-economic status (Hypothesis 3a).

However, we can also derive competing expectations for the above scenario. Individuals with high socio-economic status have more resources that can be used to unfairly buy advantages for themselves. In opposition, individuals with low socio-economic status are disadvantaged when competing on a "market" where the rule is simple: who can pay wins. In addition, high socioeconomic status people also have resources to mitigate the potential discovery of their corrupt acts, which suggests that involving in this kind of criminal behavior could only have minor negative consequences for them. For the low socio-economic individual the costs of discovery of own corrupt behavior could be substantial in comparison with their resources. Based on these arguments, the competing hypothesis is that high socio-economic status individuals could find corrupt acts more acceptable than low socio-economic status individuals (Hypothesis 3b).

Different expectations can be derived if considering the scenario when the peers are involved in corrupt behavior and not the individual. For the high socio-economic status individuals the fact that other members of society, especially lower status ones, involve in corrupt acts could trigger feelings of threat. If the corrupts acts of others are not discovered it would imply an unfair advantage for the outsiders of acquiring desirable resources (power positions, wealth, jobs, etc.), which in turn could threaten their privileged position. On the other hand, low socio-economic status individuals would also feel threatened by the involvement of their peers in corruption: it would imply even lower access to valuable material and social resources. In this scenario, everybody has to lose when other people engage in corrupt acts and thus we would expect no difference between the high socio-economic status individuals and low socio-economic individuals in the acceptance of corrupt acts (Hypothesis 3c).

Social groups defined by other characteristics might also evaluate different the cost of involvement in corruption. People in a relationship or that have children might evaluate the potential consequences of corruption (both when it comes to their engagement or that of their peers) as being higher than single people or people with no children. If discovered, they might lose the love or respect of their family members, they might be physically removed from them due to imprisonment or they might contribute to the material hardship of their family due to potential fines. In addition, the corrupt behavior of other people might be perceived more negative when considering that it affects the life chances of the whole family. Based on the above we expect that individuals that have a stable partnership and / or that have children will find corrupt acts less acceptable than individuals that are outside a partnership or that do not have children (Hypothesis 4).

Trust in governance bodies that enforce the reign of the law in a country might be another characteristic explaining difference in the acceptance of corrupt acts. Individuals that trust that the government agencies work properly would also trust that the rules of society are reinforced by the state, ensuring that every member has the same chances and opportunities and un-fair behavior is punished. In turn, individuals with high trust in governance bodies would believe to face higher chances to be exposed if engaging in corrupt acts (i.e., due to the capacity to control corruption). In addition corrupt acts go against the normative values that are represented by police or law (i.e., equality and justice) and these ideas could influence the acceptance of corrupt acts in the sense that more trust in governance bodies would go hand in hand with less acceptance of corruption (Hypothesis 5).

Country level expectations

The first characteristic that we take into account at country level is the presence of corruption as a widespread phenomenon. We again turn to the social learning theory (Akers, 1977) but now to the differential association mechanisms, which, in plain words, argues that if everybody is doing it, it might be not so bad: exposure to corrupt act or the perception of widespread corruption might make these behaviors a normal part of the day-to-day life and in turn, more acceptable for individuals. This reasoning has received some support from previous research suggesting that the prevailing social norms might be a factor that contributes to the sustenance of corruption (Cameron et al., 2009). Based on this reasoning we expect that in countries where the level of corruption is higher the acceptance of corrupt acts by individuals will be higher (Hypothesis 6a). However, we can also argue for the opposite: in countries where corruption is more wide-spread, individuals will have higher chances to hear about to the unfairness of other members of society who pay their access to valuable resources (e.g., jobs, diplomas, etc.). In addition, it also increases the chance to directly experience the disadvantage of this kind of unfair competition, and this in turn, can reinforce a negative evaluation of corrupt acts, as the differential reinforcement mechanisms would suggest. If this is the case, we would expect that in countries where the level of corruption is higher the acceptance of corrupt acts by individuals will be lower (Hypothesis 6b).

Previous studies show that ex-communist countries incorporated acceptance of corruption in the norms and values of society (Sandholtz & Taagepera, 2005; Pietrzyk-Reeves, 2006). While the former communist countries are undergoing a process aiming at the democratization of the system, which is at least formally implemented, we expect that the translation of the formal rules into the day-to-day practices of citizens to lag behind. In this respect, in countries with longer history of democratic regimes the norms of fairness and respect of the law are more likely to be also accepted and put into practice by individuals than in the new democracies. Previous studies support this idea and show that countries with a history of democracy also have a stronger civic culture (Muller & Seligson, 1994) and corruption goes against the values and norms that a strong civic culture fosters. Thus, we expect that in former communist countries to find higher levels of acceptance of corrupt acts than in the established democracies (Hypothesis 7).

Income inequality is a last contextual characteristic that we propose to have in mind when explaining the differences in the acceptance of corrupt acts between countries. Higher inequality implies a large majority of people that are on the poor end of the income distribution and a small minority that possess high material wealth. This translates into a strong imbalance between the life-chances of individuals due to their socio economic background, and subsequently, also a strong unbalance in the resources available to "buy" unfair advantages. Individuals are likely to be aware of this unbalance given that in more un-equal countries the visibility of the differences between the social statuses is higher (Wilkinson & Pickett, 2009). Furthermore, the involvement in corrupt acts might be seen as facilitating even stronger income discrepancies as result of the unfair material advantages resulted. Opinion surveys showed that Europeans find income inequality to be unjust and they express more desire toward decreasing the gap between rich and poor (Alesina, Di Tella & MacCulloch, 2004). This could also imply that rejecting income inequalities would go hand in hand with rejecting the involvement in corrupt acts. Based on the above we expect that in more un-equal countries individuals would be more likely to have a lower acceptance of corrupt acts than in countries that are more equal (Hypothesis 8).

Data and methods

In order to test our hypotheses we made use of data from the European Values Study (EVS, 2008), a large-scale, cross-national, and longitudinal survey research program on basic human values. We use most recent information, collected in 2008. From the integrated dataset we excluded North Cyprus, North Ireland, and Azerbaijan due to non-availability of some of the measures used in our analyses. Our working dataset covers 62674 individuals nested in 43 European countries.

Dependent variable

Acceptance of corruption was measured as a scale based on 4 questions that asked respondents whether they find justifiable the following corrupt behaviors: accepting a bribe, claiming undeserved state benefits, cheating on tax and paying cash to avoid tax. The 4 items were measured with a scale ranging from 1 "never justifiable" to 10 "always justifiable". We recoded the scales so that all individuals who declared that the 4 corrupt practices were justifiable to some degree received value 1 while individuals who found them "never justifiable" received value 0. We used the recoded variable to compute a summative scale. Reliability tests resulted in a Cronbach alpha of 0.73 in the full sample. The same test conducted within country also resulted in good reliability indices, ranging from a minimum of 0.63 in Belgium to 0.89 in Georgia. The resulted variable measuring the acceptance of corrupt practices ranged from 0 to 4, where a higher score indicated higher acceptance.

Independent variables at individual level

Religiosity was computed as a factor scale for the following 9 items: if respondents believe in God, hell, life after death, sin and heaven, if they consider themselves as religious persons, if they get strength from religion and how important is religion and God in their life. When needed items were recoded so that a higher value to represent higher religiosity. Given that some of these items had quite high level of missing values we first addressed this issue, as presented in the section on missing values, and after that we computed the religiosity scale.

Age was measured by computing the age of the individuals at the date of data collection based on the reported birth year. We centered the age variable at 18 and also included quadratic term in order to allow for curvilinear effects.

Socio-economic status was measured by the income and occupational status of the respondents. Income was measured by the monthly household income (x1000), corrected for PPP in Euros, variable that is provided in the integrated EVS dataset and which is calculated by the EVS team (see the technical documentation for information on the calculations). It provides a measure of disposable income for consumption of the households that is comparable between countries. Note that the original variable had a high level of missing values (see Table 1). See below the section on Missing values covering detailed explanations on how we dealt with this problem. In our analyses we used also a quadratic term to allow for non linear effects of income.

The occupational status of the individuals was derived from the variable measuring the European Socio-Economic Classification (Harrison and Rose 2006). We recoded the original variable with 9 categories to the restricted 5 category form: (1) higher and lower salaried, (2) intermediate employee, (3) small employers and self-employed, (4) low employees in sales and service, (5) lower technical and routine occupations. In addition we also have a category for those individuals that do not have an occupation. In our analyses the lower technical and routine occupations were used as reference category.

Partnership status of the respondents was derived from their legal civil status: those who were married or living together were categorized as having a partnership. In addition we used the measure of the number of children declared by respondents to derive a dummy variable for the individuals that had kids compared to those who did not have.

Trust in governance bodies was measured by a mean scale composed of three questions asking respondents if they had confidence in the police, government and justice. A higher score on this scale indicated higher confidence in governance bodies.

Control variables

We also had two control variables at individual level: the level of education of individuals, introduced in analysis as high and medium education vs. low education (reference category) and gender measured as a dummy for males.

Independent variable at country level

The former communist history was computed by assigning each country in the following categories:

(1) long established democracy, (2) country that belonged to the former soviet bloc and (3) country that belonged to the former communist bloc. In the analyses we introduced the variable as dummies with the category long established democracy as reference group.

The level of corruption of the county was measured by the Corruption Perception Index developed by Transparency International (2008). The index is based on perceptions on the prevalence of corruption as seen by business people and country analysts and not by the population at large. It ranges from 10 (highly clean countries) to 0 (highly corrupt countries).

Income inequality was measured by the Gini Index, a low value indicating low inequality while a high value represents high inequality. It ranges from 0 to 100 and was derived from the Standardized World Income Inequality Database (SWIID) (Solt 2009), dataset that was developed with the purpose of increasing the coverage across country and time while also improving the comparability across observations.

Detailed information on the variables in the analysis can be found in Table 1.

Missing values treatment and data analysis

According to Allison (2001) a good method to deal with the missing values requires to accomplish the following: minimize bias in the parameter estimates, maximize the use of the available data and yield good estimates of uncertainty. Recent studies point out that traditional approaches to address the problem of missing values (i.e., list wise deletion, pair wise deletion, mean substitution, dummy variable adjustment) are not adequate (Graham, 2009). Instead, modern approaches were proposed, known under the name of multiple imputation methods, which satisfy the three criteria.

The main idea of multiple imputation methods is that instead of filling in missing values to create a single imputed dataset, several imputed data sets are created each of which contains different imputed values. The analysis is then conducted on each of the imputed datasets and the estimates are then combined (Rubin 1987). In the present study we used the chained equations method as implemented in ICE (Royston 2005), a user contributed add-on under STATA 11. The advantages of this method is that it provide support for all type of variables, it does not assumes a single multivariate model for all the data and it allows the imputations to be made within the category of a specified variable (in our case within each country).

We followed the criteria set by Allison (2001) and Graham (2009) for implementing the imputation models. First, missing values on the dependent variable were deleted. Our imputation models include all the individual level variables in our models. In order to preserve the nested structure of the data the imputations are computed within the country. We derived a standard number of 5 alternative imputed datasets which were then analyzed using the standard multiple imputation module in STATA.

The data that we are using is characterized by the fact that individuals are nested in countries, which have specific structural characteristics. In order to adequately address the nested structure of the data and be able to differentiate the contextual from the compositional effects we used multilevel techniques (Snijders & Bosker 1999). In addition, in order to facilitate the interpretation of the effects we standardized all continuous variables, at individual and country level.

Results

We start by presenting some exploratory analyses of the data. As seen in Table 1, the average acceptance of corrupt acts in the 43 European countries is relatively low (1.54 on a scale from 0 to 4). However, as seen in Figure 1 there is quite some variation between the 43 European countries, ranging from as low as a mean of 0.53 in Turkey to as high as 2.67 in Belarus. These findings suggest that the specific contexts of the countries matter for the variation in the acceptance of corrupt acts in Europe.

In Figure 2 and 3 we present the relationships between the contextual characteristics that were brought forward by our theoretical reasoning. We found a strong relationship between the average acceptance of corrupt acts and income inequality (-0.40, p<0.05), which suggests that in countries with more income inequality the acceptance of corrupt acts is lower. However, when looking at the relationship between how widespread corruption is and the acceptance of corrupt acts we found only a weak relationship (0.13, p>0.05) that tentatively suggest that in countries where corruption is more widespread the acceptance of corruption is lower (remember that a higher value on the corruption index indicates less perceived corruption). Regarding the differences in the acceptance of corruption between established democracies and countries that experienced communist regime, we find a significantly higher mean acceptance of corrupt acts in the countries that were part of the former soviet bloc (mean for the soviet bloc: 1.74; mean for the established democracies: 1.54). However, the former communist bloc countries have similar (or even slightly lower) average acceptance of corrupt acts than the established democracies (mean for the former communist bloc: 1.51).

While these exploratory analyses shed some light on the factors that relate to the different levels of acceptance of corrupt acts in the European countries, we cannot exclude the possibility that the observed differences might be due to compositional effects due to individual level characteristics. In order to control for the compositional effects and formally test our hypotheses we turn to the results of multilevel models which are presented in Table 2. Our modeling strategy was the following. First we estimated a null model that allowed us to calculate the inter class correlation, thus determining how much variation in the acceptance of corrupt acts is due to the clustering in countries. The inter-class correlation coefficient indicated that only 9% from the variance in the acceptance of corrupt acts is due to the contextual influence, which suggests that differences between individuals have more weight than differences between countries. We then estimated a model where all individual level characteristics were introduced (Model 1) in order to find support for our individual level expectations and in Model 2 we added the country level characteristics in order to estimate the contextual effects.

Turning to the results of our estimations, based on hypothesis 1 we expected that individuals with higher religiosity will display lower acceptance of corrupt acts. Based on Model 1 (Table 2) we found support for this hypothesis: one standard deviation on the religiosity scale was associated with a decrease of 0.13 in the acceptance of corrupt acts scale. However, the effect size of this parameter indicated that the relationship was a weak one (0.13/1.47=0.08). We also expected that older individuals display lower acceptance of corrupt acts because they have more to loose from involvement in corrupt behavior. As our results in Model 1 show, with every standard deviation increase in age the acceptance of corrupt acts decreased with 0.01 and in addition, for older people this effect was stronger. However, in the absolute terms of the effect size this effect was negligible (0.01/1.47=0.007).

Regarding socio-economic status, we derived alternative scenarios, suggesting contradictory expectations. Our results provided unconsistent evidences depending on the measure of socioeconomic status that we took into account. If we looked at income as a measure of socio-economic status, our results (Model 1, Table 2) suggested that individuals with higher income report a higher acceptance of corrupt acts, supporting the argument of having superior resources that can "buy" more and that can also mitigate the potential costs of being discovered to be involved in corruption. However, we observed that as we go up in the income hierarchy this effect was weaker, suggesting that the alternative mechanisms of the fear of losing the prestige associated with higher socio-economic status might also be at work.

When we looked at occupation as a measure of socio-economic status, we found that there were no significant differences in the acceptance of corrupt acts between respondents that were employed. Only the persons that did not have an occupation appeared to have a higher acceptance of corrupt acts than the lower technical and routine occupations. In addition, individuals employed in the service sectors seemed to be (only marginally for á<0.10) more inclined to accept corrupt acts than the lower technical and routine occupations. This said, we found only partial evidence in support for our expectations regarding the relationship between socio-economic status and acceptance of corrupt acts which did not fully support any of the alternative scenario derived. Our next expectation regarded the marital status and the presence of children, both these factors presumably being associated with lower acceptance of corrupt acts, expectation that was supported by our analyses as seen in Model 1 Table 2. In addition, we also expected that individuals with high level of trust in governance bodies will have lower acceptance of corrupt acts, and again, this expectation was supported by our data.

Turning to the effects of the country characteristics, we derived contrasting expectations about the relationship between corruption levels of the country and the acceptance of corrupt acts by individuals. However our results did not provide conclusive results, since the coefficient did not reach significance. It is thus possible that both mechanisms are at work cancelling each other out. Regarding the former communist history of the country we expected to find lower acceptance of corrupt acts in the established democracies. Out results provided partial support for this expectation, since the difference was only significant between countries from the ex-soviet bloc and the established democracies. In countries from the former ex-communist bloc, the average level of acceptance of corrupt acts was not significantly different than in countries with established democracies. Not lastly, we expected that in countries with higher income inequality to find lower acceptance of corrupt acts, and this hypothesis received support from our estimations presented in Model 2.

Although we did not formulated hypotheses about the control variable it is worth noting that, in line with Gatti, Paternostro & Rigolini (2003), we also found that males have a higher acceptance of corrupt acts than females and there are no significant differences between individuals with different educational levels in their acceptance of corrupt acts.

Conclusions

In the present contribution we set out to contribute to the better understanding of the differences in the acceptance of corrupt acts in a large sample of 43 European countries. We employed multi-level models in order to disentangle the individual and contextual effects and explain the variability in the acceptance of corrupt acts between individuals and between societies. Based on our analyses we reached the following conclusions.

Firstly, we found that religiosity and gender had the strongest effects in relation to the acceptance of corrupt acts. These findings are in line with previous research pointing out that women are less inclined to accept corruption and that religious participation decreases the acceptance of corrupt acts (Gatti, Paternostro & Rigolini, 2003, Swami et al., 2001). Regarding religiosity, our argument was that involvement in corrupt acts goes against the principles promoted by all religious denomination. We note that the norms of honesty and fairness are not exclusive to religious institutions. Other institutions are founded on and try to promote the same ideas, e.g., civic organizations that embrace these values in their aims and objectives. Subsequently future research can investigate to what extent non-religious individuals that are active in civic organizations also report lower acceptance of corrupt acts. A confirmation of this expectation would offer an instrument that can be actively used for influencing the level of tolerance toward corrupt behaviors.

Secondly, we found that young individuals without occupation and with more material resources are more likely to find corrupt acts acceptable. These results are again in line with those of Gatti, Paternostro & Rigolini (2003). If indeed the acceptance of corrupt acts has an influence on the actual corrupt behavior, these individuals would have less to lose and more to gain from involving in corruption, transforming them into a risk group. However, the presence of children and the presence of a stable partnership as well as the higher confidence in the government bodies reduce the acceptance of corrupt acts. These results support our line of reasoning that looked at the balance of anticipated or actual rewards and punishments potentially following the involvement in corrupt behavior. Thus, all together we found support for the idea that individuals adjust their acceptance of corrupt acts so that when there is more to lose for themselves or for their loved ones, they report less tolerance of corruption.

Thirdly, we found support for the fact that the variation in the acceptance of corrupt acts is also due to the characteristics of the societies. We found that the acceptance of corruption is higher in the countries that were part of the former soviet bloc, but we found that the individuals living in the former communist bloc have on average similar levels of acceptance of corrupt acts as the individuals living in European countries with established democracies. Because the former communist countries also have higher levels of income inequality and corruption, we tested the previous differences while controlling for the two variables. Even so, we found that the difference in the acceptance of corrupt acts between the ex-soviet and the established democracies is net of their levels of income inequality and corruption. This is a surprising result because the previous literature argued that the acceptance of corruption was something specific to all former communist regimes. Based on our results it seems that, even if corruption in the former communist bloc is still perceived as being widespread, the change has been made toward less tolerant values concerning corrupt acts. Future research is warranted in identifying why the difference in the values toward corruption exists between the former soviet and the former communist bloc countries.

Another surprising result was the finding that the level of corruption of a country was not related to the acceptance of the corrupt acts by individuals. Regarding this relationship we had a dual argument: on the one hand, we argued along the "if everyone is doing, it might be not so bad" line, when widespread corruption becomes a norm and thus we expected that the acceptance of corrupt acts will be higher in more corrupt countries. On the other hand, we argued that the detrimental effects of corruption for individuals in highly corrupt countries might make is less acceptable for individuals. Based on our results, we could not rule in favor of any of the two arguments. A possible explanation for this result might be that the two mechanisms work at the same time or they work differentially for different population groups, overall canceling themselves out. If this is the case, future research is needed in order to determine under which conditions and for what social groups the two mechanisms are working.

Not lastly, we also found that income inequality is significantly related to the acceptance of corrupt acts: in countries with higher income inequality individuals report on average lower acceptance of corrupt acts. This is a new piece of evidence that adds to the large body of literature investigating the social and individual effects of income inequality (Wilkinson & Pickett 2009). While previous literature on social and individual effects of income inequality argues for damaging effects of income inequality on a large range of social aspects, we found in our study a more positive effect: it seems that a more unequal distribution of income makes individuals less tolerant toward another social problem, namely corruption. We note that previous studies have shown that in samples of countries the corruption level and income inequality are positively related (Uslaner, 2009). Our analysis suggests the fact that the ecological relationship between income inequality and the level of corruption is not identical to the contextual relationship between income inequality and the individual's acceptance of corrupt acts. While at societal level higher inequalities are associated with perceptions of more widespread corruption, the effect of income inequality on the individuals' values regarding corrupt behavior is the opposite.

To sum up, the present analysis had as main aim to shed more light on the complex relationship between micro and macro characteristics and how they relate to the acceptance of corrupt acts by individuals. Besides the strengths of this study, such as the high quality data and appropriate methods of analysis in a large sample of European countries, we also mention that this analysis could only explain around 7% of the variation between individuals in our dependent variable. In addition, the effects sizes of the coefficients were very small. This indicates that there are other relevant individual characteristics that might be able to explain more from the differences in the acceptance of corrupt acts and future research is needed in order to identify these characteristics. As already suggested, the involvement of individuals in civic organizations or the exposure to civic education that promotes values such as equity and fairness (e.g., via nongovernmental organizations or school curricula) might be factors that could shape the acceptance of corrupt acts. Furthermore, other contextual characteristics such as the wealth of the society, the equality of opportunity or the rule of law might be taken into account as predictors of the degree of acceptance of corruption by individuals.

Other interesting lines of research might also focus on the relation between other value orientations and the acceptance of corrupt acts. The attitudes toward corruption might be only one part of a more general value orientation of individuals that might also include ideas of trust, fairness or equity. These value structures might be context specific, as resulting from the arguments and the research of Inglehart (1971) among others, who showed that the values of individuals are (at least partially) shaped by the exposure to some specific structural characteristics of the society at large (e.g., economic development). If this is the case, it would imply that the attitudes and the values of individuals might have a common exogenous cause. The consequence would be that explaining values through other values might be an exercise plagued by the third variable problem, and a more useful exercise would be to identify how values cluster and what are the structural, exogenous factors that shape them. Not lastly, we mention that in this study we assumed that individuals with the same characteristics will have the same acceptance of corrupt acts regardless of the context where they live and were socialized. However, this might not be the case, and another future line of research might focus on identifying more precisely how individual and contextual factors interact in shaping the acceptance of corrupt behaviors.

References

References

1. Akers, R. (1977). Deviant behavior: a social learning approach. Belmont, Calif: Wadsworth Pub.

2. Alesina, A., Di Tella, R., & MacCulloch, R. (2004). Inequality and happiness: are Europeans and Americans different? Journal of Public Economics, 88(9-10), pp. 2009-2042. http://dx.doi.org/10.1016/j.jpubeco.2003.07.006

3. Allison, P. (2001). Missing Data. Thousand Oaks, CA: Sage Publications.

4. Bardhan, P. (1997). Corruption and development: A review of Issues. Journal of Economic Literature, XXXV, pp. 1320-1346.

5. Cameron, L., Chaudhuri, A., Erkal, N., & Gangadharan, L. (2009). Propensities to engage in and punish corrupt behavior: Experimental evidence from Australia, India, Indonesia and Singapore. Journal of Public Economics, 93(7-8), pp. 843-851. http://dx.doi.org/10.1016/j.jpubeco.2009.03.004

6. Dreher, A., Kotsogiannis, C., & McCorriston, S. (2007). Corruption around the world: Evidence from a structural model. Journal of Comparative Economics, 35 (3), pp. 443-466. http://dx.doi.org/10.1016/j.jce.2007.07.001

7. EVS (2008). European Values Study: Integrated data file. Retrieved November 25, 2011 from http://www.gesis.org/en/services/data-analysis/survey-data/european-values-study/4thwave- 2008/.

8. Gatti, R., Paternostro, S., & Rigolini, J. (2003). Individual Attitudes toward Corruption: Do Social Effects Matter? World Bank Policy Research working Paper 3122, World Bank. Retrieved July 3 2012, from https://files.nyu.edu/jpr227/public/Research/Corruption.pdf.

9. Graham, J. (2009). Missing Data Analysis: making it work in the real world. Annual Review of Psychology, 60, pp. 549-576. http://dx.doi.org/10.1146/annurev.psych.58.110405.085530

10. Harrison, E., & Rose, D. (2006). The European Socio-economic Classification. (ESeC) User Guide, Institute for Social and Economic Research, University of Essex, Colchester, UK. Retrieved April 9, 2011 from https://www.iser.essex.ac.uk/files/esec/guide/docs/UserGuide.pdf

11. Inglehart, R. (1971). The Silent Revolution in Europe: Intergenerational Change in Post-Industrial Societies. The American Political Science Review, 65(4), pp. 991-1017. http://dx.doi.org/10.2307/1953494

12. Laub, J., & Samson, R. (2001). Understanding Desistance from Crime. Crime & Justice, 28(1), pp. 1-70.

13. Li, H., Xu, L. C., & Zou, H. (2000). Corruption, Income distribution and Growth. Economics and Politics, 12(2), pp. 155-181. http://dx.doi.org/10.1111/1468-0343.00073

14. Mauro, P. (1995). Corruption and Growth. The quarterly journal of economics, 110(3), pp. 681-712. http://dx.doi.org/10.2307/2946696

15. Muller, E., & Seligson, M. (1994). Civic culture and democracy: the question of causal relationships. The American Political Science Review, 88(3), pp. 635-652. http://dx.doi.org/10.2307/2944800

16. OECD (2011). Convention on Combating Bribery of Foreign Public Officials in International Business Transactions. Retrived December 10, 2011, from http://www.oecd. org/document/20/0,3343,en_2649_34859_2017813_1_1_1_1,00.html

17. Pietrzyk-Reeves, D. (2006). Corruption and democratization: a civic republican view. Acta Politica, 41, pp. 370-388. http://dx.doi.org/10.1057/palgrave.ap.5500137

18. Royston, P. (2005). Multiple imputation of missing values: update. The Stata Journal, 5, pp. 1-14.

19. Rubin, D. (1987). Multiple imputation for nonresponse in surveys. New York: John Wiley.

20. Sandholtz, W., & Taagepera, R. (2005). Corruption, Culture and Communism. International Review of Sociology, 15(1), pp. 109-131. http://dx.doi.org/10.1080/03906700500038678

21. Snijders, T., & Bosker, R. (1999). Multilevel analysis. An introduction to basic and advanced multilevel modeling. London: Sage Publications.

22. Solt, F. (2009). Standardizing the World Income Inequality Database. Social Science Quarterly, 90(2), pp. 231-242. http://dx.doi.org/10.1111/j.1540-6237.2009.00614.x

23. Swamy, A., Knack, S., Lee, Y., & Azfar, O. (2001). Gender and corruption. Journal of Development Economics, 64(1), pp. 25-55. http://dx.doi.org/10.1016/S0304-3878(00)00123-1

24. Tavits, M. (2008). Representation, Corruption, and Subjective Well-Being. Comparative Political Studies, 41(12), pp. 1607-1630. http://dx.doi.org/10.1177/0010414007308537

25. Torgler, B., & Valev, N.T. (2006). Corruption and age. Journal of bioeconomics, 8(2), pp. 133-145. http://dx.doi.org/10.1007/s10818-006-9003-0

26. Transparency International (2008). Corruption Perception Index. Retrieved December 10, 2011, from http://archive.transparency.org/policy_research/surveys_indices/cpi/2008

27. Treisman, D. (2000). The causes of corruption: a cross-national study. Journal of Public Economics, 76(3), pp. 399-457. http://dx.doi.org/10.1016/S0047-2727(99)00092-4

28. Uslaner, E. (2009). Corruption. In G.T. Svendsen & G.L.H. Svendsen (Eds.), The Handbook of Social Capital: The Troika of Sociology, Political Science and Economics (pp. 127-142). London: Edward Elgar.

29. Wilkinson, R., & Pickett, K. (2009). The spirit level. Why more equal societies almost always do better. London: Penguin Books.

30. You, J., & Khagram, S. (2005). A comparative study of inequality and corruption. American Sociological Review, 70(1), pp. 136-157. http://dx.doi.org/10.1177/000312240507000107