MAGNITUDE OF CORRUPTION, RATIONALIZATION, AND INTERNAL CONTROL QUALITY: AN EXPERIMENTAL STUDY ON SUBSEQUENT OFFENSE

This research examines the extent to which the rationalization tactics model reflects the relation between magnitude of corruption and attitudes toward corruption. The rationalization model predicts that individuals will justify their corrupt acts and that rationalizations will affect subsequent offense depending upon the levels of internal control quality. Using a 2 X 2 between-subjects experiment with 170 public servants as participants, this research finds that individuals are more prone to accept and rationalize their corrupt acts in the gratuities context than in the bribery context. Further, this research finds that rationalizations lead to subsequent offense. However, this study cannot find support on the mitigating effect of internal control quality. The results point to the importance of considering the implications of organizational culture for future effective combat against corruption.


INTRODUCTION
Indonesia is still in a state of highlevel corruption. Of the 168 countries reviewed, Indonesia ranks 88th in the corruption perception index (Transparency International 2016). Various efforts have been made to eradicate corruption. The Corruption Eradication Commission has successfully completed a large number of high-profile corruption cases. In 2015, the Commission had conducted 183 investigations and 91 prosecutions (KPK 2016). In the same year, 33 court decisions had been taken and Rp198 billions had been saved and returned to the state treasury. Nevertheless, the anticorruption agenda is still far from ideal.
The typical definition of corruption encompasses the misuse or abuse of the office (whether public or corporate) for the sake of private self-interest (Fleming and Zyglidopoulos 2009). Corruption has become the cause of a number of acute problems, such as low economic growth (Mauro 1995), competitiveness decline and inflated business costs (Mistry 2012), and threat to investment climate (Kurtzman et al. 2004). Olken and Pande (2013) assert that corruption may increase the marginal tax rate of firms, significantly lower the quality of public service, induce inefficiencies, and lessens government's ability to correct an externality which can lead to obstructed law enforcement. Research in economics and political science provides explanation upon why a number of countries fall into category of high prevalence of corruption while other countries tend to be free from the problem (Wang 2014). Prior research investigates determinants and impacts of corruption. The main issue that the literature discusses is exogenous factors affecting corruption, which in general are beyond government's control (Liu and Feng 2015).
Meanwhile, research on corruption has sparked discussion regarding human factors. Body of literature has given insights on an aggregate level, but still lacks of resolution on micro level. Few research highlights causes, consequences, and other variables pertaining to corruption at the individual level with relevant organizational contexts. In various cases, corrupttion is usually the result of an agreement or consent of others. This white collar crime could take place in a long time and even is passed throughout generations in organizations. In most instances, the corrupt acts involves honorable community members (such as ministers, judges, and politicians). Hence, the issue about why one is willing to act in an unethical conduct is still an important research focus. An essential question has surfaced concerning the reasons that lead otherwise law-abiding individuals to be involved in corrupt practices.
Corruption, once entrenched, is difficult to eliminate. Although progress in combating corruption and unethical conduct are generally appreciated by public, corruption and unethical conduct still remains a serious problem requiring major efforts. Therefore, it is interesting to know about the economic, political as well as psychological factors underlying the persistence of corruption. Why individuals commit subsequent offenses or repeated wrongdoings becomes a fundamental question. Understanding basic notion causing subsequent offense can serve as an enabler in effective strategy implementation necessary to reduce corruption which remains prevalent in many areas. Combating corruption and unethical conducts has to be supported with "cultural and psychological strategies" as well as legal administrative strategies. Rusch (2016) suggests that subsequent offense of corruption may result from the culture of compliance. An malfunctioning organizational culture may produce individuals' faulty mental shortcuts and furthers law violation. Other people who witness the offense tend not to speak up or act in response to the situation if they perceive that no one else witnessing that situation is speaking up or acting (Latané and Darley 1970). People can assume that others either are responsible for taking action or would do so if they thought it necessary, leading to an unintended diffusion of responsibility within the group and a failure by any group member to take action. This can lead to even further misconducts.
Corruption is a complex phenomenon and usually involves more than just greedy individuals (Fleming and Zyglidopoulos 2009). Ashforth et al. (2008) assert that corruption can be attributed to personal greed, organizational culture, or sometimes is simply difficult to understand. Research finds that individual's characteristics influence his fraudulent acts (Hartmann and Maas 2010;Haines and Leonard 2007;Ford and Richardson 1994). Systemic corruption may also be result of moral behaviors that have developed as schema response mechanism to society social roles (Ntayi et al. 2013). Anand et al. (2004) provide another perspective to explain the stimuli that form wrongdoers' ethical logic. Their concept of rationalization explains why one can engage in corruption and at the same time holds a positive image of himself. Rationalizations are mental strategies that allow individuals to view their corrupt acts as justified (Anand et al. 2004). Rationalizations allow actors to alleviate moral anxiety and to neutralize any regrets or negative feelings that stem from their participation in unethical acts. Thus, rationalization works as a defense mechanism that protects the ego from challenges by a punitive superego (Fleming and Zyglidopoulos 2009). The power of rationalization extends from simple and ordinary wrongdoings to serious forms of whitecollar crimes.
We extend earlier studies on corruption by examining the effects of rationalization on subsequent offense. We specifically turn our attention to subsequent offense since it may lead to systemic and endemic corruption which eventually originate a great deal of loss. Corruption happens when a person is able to act willfully with the intent to disregard or disobey the law. Repeated corruption can create shift of mentality and once such "supplementary incomes" start pouring in, the habit is difficult to stop.
We propose and find that magnitude of corruption affects how individual provides rationalization. Indonesia law recognizes seven categories of corruption, from gratuities to bribery, which carry different severity of punishment. Gratuities and bribery are the most common methods of corruption in Indonesia (KPK 2016). We investigate the effect of these two magnitudes of corruption since they have different consequences. We expect and find that individuals provide more rationalizations when they are involved in gratuities than when they are involved in bribery. Our results suggest that involvement in less serious corruption would be more difficult to fight since perpetrators were more likely to create resistance.
Literature on fraud (e.g. Barra 2010; Rae and Subramaniam 2008) asserts that an effective internal control system is a major resource for preventing, detecting, and correcting fraud. Barra (2010) investigates the relationship between the type of control and penalty and efforts to reduce fraud. He finds that separation of functions within the organizations improves employee's cost of committing fraud. This means that separation of functions plays an important role in fraud prevention. Nevertheless, results of these studies have not been specifically addressed the role of internal control in lessening subsequent offense. Near and Miceli (1995) present a whistleblowing model and argue that favorable organizational climate to facilitate whistle blowing actions is necessary for preventing fraud and other unethical conducts. Kaplan et al. (2015) investigate the role of investment in whistle-blowing development and its relevant managerial training. Unfortunately, the study could not provide empirical evidence for the role of such investment due to low power of test. We extend the research of Kaplan et al. (2015) in an effort to present a more comprehensive explanation pertaining to corrup-tion and its prevention means. We hypothesize that internal control quality miti-gates the effect of rationalization on sub-sequent offense. However, the data do not support the hypothesis.
Overall, our study contributes to the corruption literature in at least two ways. The primary contribution is an understanding of how rationalization process increases people' tendency to engage in corrupt behavior. Specifically, our research suggests that different magnitude of corruption carries different cognitive processes. Our findings indicate that they are also affected by individuals' personal definition regarding corruption. Our research, thus, sheds some light on some of the less obvious factors that are involved in corruption.
Second, we provide initial tests of Anand et al. (2004) proposed rationalization model using a laboratory experiment. Prior research on corruption has been conducted using experiment (see e.g. Olken 2007;Alatas et al. 2009;Rivas 2012). This method produces a more convincing causal relationship compared to other methods. Survey methods may not be able to deliver reliable data on corruption activities due to confidentiality reason. In addition, prior research on corruption could not extract strong causal relationship between corruption eradication strategy and corruption reduction (Peishakin and Pinto 2010). Our experiment differs from surveys and perception indices in that our experiment participants receive actual monetary payments, the amounts of which depend on the decisions they make during the experiments. Hence, through experiment we are able to reveal mechanics that can explain why people engage in corrupt acts. This paper proceeds as follows. In the next section, we introduce the constructs, which are central to our analysis, and we develop our hypotheses. In the third section, we present the design of the experiment. The results of the study are presented in the fourth section. In the fifth section, we discuss our findings and their relevance for future research and for business practice.

Magnitude of Corruption
Corruption refers to a concept that is not easy to define. A broad definition of corruption includes the misuse or abuse of the office with special focus on the improper use of the public good by private self interest (Bratsis 2003;Fleming and Zyglidopoulos 2009), where misuse in turn refers to departures from acceptable societal norms (Anand et al. 2004). Public office can be misused by receiving and offering bribes, embezzlement, extortion, private appropriation and fraudulent use of funds, nepotism, patronage, or theft of state assets. Other scholars use the word corruption as "incidents where a bureaucrat (or an elected official) breaks a rule for private gain" (Olken & Pande 2013). All of the above-mentioned definitions are wellmatched to that of Indonesia law. Indonesia Anticorruption Law recognizes seven groups of corruption that encompass all illegal acts to create private gain at the expense of public or state interest. In this study, we specifically investigate two recognized corruption forms that have the most frequent incidences: gratuities and bribery.
We refer the magnitude of corruption to the significance level of corruption. Thus, magnitude covers not only the size, but also the intensity of corruption. In this study, we classify gratuities as an act of corruption which has a lower magnitude than bribery. Indonesia anticorruption law recognizes the difference magnitude of the two types of corruption. Bribery is the more serious offense of the two, and conviction carries penalties of up to life imprisonment and a maximum fine of Rp1 billion. Illegal gratuity, by contrast, carries a penalty of up to three years' imprisonment and a maximum fine of Rp50 millions (KPK 2016).
Commonly, the public has a negative view on corruption. Nevertheless, perceptions regarding corruption can be influenced heavily by cultural bias and vary across countries (Paldam 2002). People in some countries regard corruption as a serious crime and unethical activity while other individuals have higher tolerance and even treat corruption as a part of life.
Several research studies across disciplines have suggested that individuals and groups are influenced in their behavior by social norms or culture. Cameron et al. (2009) examine the effect of cultural differences on tolerance of corruption. They conducted experiments in four countries: Australia, India, Indonesia, and Singapore by administering two treatments: welfare reducing corruption or welfare enhancing corruption. They find significant cross-country differences. The results from India and Australia suggest that greater exposure to corruption in daily life may build a greater tolerance of corruption. Indians, as compared to Australians, have a lower propensity to punish and are more willing to accept bribes. Alatas et al. (2009) examine gender differences in behavior when confronted with a common bribery problem in four countries: Australia, India, Singapore, and Indonesia. They specifically investigate whether women are less likely to offer bribes and more likely to punish corrupt behavior. The results show that Australian men are more likely to engage in and be more tolerant of corruption than are Australian women. However, they do not find significant gender differences in India, Singapore, and Indonesia. Indonesia Central Statistics Agency surveyed 10,000 households to measure perceptions of corruption especially in public services at the lowest level of government bureaucracy, such as the police and teachers. The survey reveals that 32% of people polled viewed bribing police officers as normal, while 33% of respondents considered it normal to give extra money to civil registry officials (Ismar and Husna 2013). Fisman and Miguel (2007), using a field experiment design and diplomats as subjects, investigate whether people from different cultural backgrounds have different propensities for corrupt acts. They find that when there is no penalty for breaking the law, diplomats follow the behavior norms of government officials in their own countries, suggesting that penalty enforcement can control individual corruption.
The previous research renders a baseline for hypothesizing that the bureaucrats in Indonesia have different levels of tolerance on corruption, depending on the magnitude of the corrupt acts. Public perception of corruption may be related to cultural practices or traditions. In Indonesia, since forty percent of the population is Javanese, and that Javanese people have majority control in the government, Javanese culture is often blamed for the practices of corruption (Magnis and Suseno 1984). Javanese people may tend to tolerate corrupt practices rather than creating conflicts with the perpetrators by reporting the illegal acts. It is generally considered a good-natured prosocial thing to give and receive gifts, but bribing is considered to be antisocial and negative. Hence, if a citizen wants to provide gifts as a token of appreciation to public servants, many people tend to agree to such action. However, research shows that there's actually a fine line between the socially acceptable gratuities and the immoral act of bribing (Torfason et al. 2012). Porter (2012) argues that cultures have widely become less ethical. He finds that 75 percent of Americans believe that corruption has increased over three yearperiod (2008-2011), while 62 percent of Americans believe corruption to be widespread across corporate America. Recent research also shows the existence of ''unethical pro organization behavior'' (Vadera and Pratt 2013). Umphress and Bingham (2011) provide evidence that individuals may conduct unethical proorganizational behavior because they strongly identify with the organization or because of an expectation of reciprocity from the organization. So, even when an organization's top management believes an ethical or values-based atmosphere, this perception may not be shared elsewhere within the organization. These facts likely encourage wrongdoing behavior in subtle ways that are not easily observable. Formal policies may state an expectation of ethical behavior but informally encourage different behaviors.
Extant literature on whistleblowing research indicates that individuals will react differently depending on the type of wrongdoings (Near and Miceli 1995;Gundlach et al. 2003;Kaplan et al. 2015). Robinson et al. (2012) argue that people's actions are influenced by how they use information to arrive at causal explanations for events. If a violation is perceived as beyond the control of the perpetrator, then people will tend to be more tolerant to such offense, and vice versa. According to Robinson et al. (2012), observers of fraud are more likely to report misappropriation of assets than fraudulent financial reporting. Observers will see a person engaging in misappropriation of assets as having more responsibility compared to a person en-gaging in fraudulent financial reporting. Fraudulent financial reporting is more attributable to external pressures such as protecting employees and satisfying equity holders. The other reason is about who potentially benefits from the fraud. Misappropriation of assets only benefits the person engaging in fraud and obviously and directly harms the firm and its shareholders while fraudulent financial reporting is seen as potentially benefiting the firm and its shareholders.
Following the logic, people will tend to treat small gratituties as a common practice while bribery as unethical act that should be avoided. Small gratuities will not harm anybody as severe as bigger bribery. Moreover, the practice of accepting and giving small gratuites is a more common practice than bigger bribery, as suggested by the preceeding discussion regarding culture. Thus, accepting small gratuities may be seen as more attributable to external climate and pressures than accepting bigger bribery. The discussion leads to the following hypothesis.

H1
: Small gratuities will be more acceptable than bigger bribery.

Rationalization
Big ambition combined with greed can be a common answer to the question about the motivation of criminals. One can attribute unethical behavior and corrupt acts to the personal characteristics of individuals. The underlying assumption is that one is eventually responsible for his own moral choices, ethical conduct, and subsequent consequences (Trevino and Youngblood 1990). Other studies suggest that individuals' character (Haines and Leonard 2007), traditional values and empathy (Hobson et al. 2011), honesty (Rankin et al. 2008), and level of Machiavellianism (Hartmann and Maas 2010) as determinants of individuals' tendency to act illegally or unethically. However, the previous dispositional approach fails to recognize the role of context that may lead individuals to act unethically if placed in opposed conditions (Fleming and Zyglidopoulos 2009). Research finds that social pressure affects dysfunctional behavior (Hartmann and Maas 2010) and that effectiveness of a code of ethics will improve manager behavior to the extent that it activates social norms that control opportunistic behavior .
According to psychology literature, rationalization plays an important role in explaining corruption. Rationalization is a psychological mechanism that people use to protect themselves from moral anxiety caused by their unethical conduct or behavior. Individuals rationalize to neutralize or at least reduce their feelings of distress after committing unethical acts. Rationalization is a form of appeasement to oneself. Bad actions are justified because, for example, "others do the same thing" or "we have to do to avoid a greater risk" (Anand et al 2004). The logic is in line with the theory of cognitive dissonance (Festinger 1957). The theory explains that a person's actions can affect his attitude and subsequent belief. The main proposition of the theory is that if a person has two inconsistent cognitions or when the behavior is not in accordance with his thoughts, then he will experience psychological dissonance, an uncomfortable condition. Consequently, he will attempt to reduce this dissonance. Rationalization is one of the ways that people can use to reduce cognitive dissonance. Cognitive dissonance theory provides the foundation for rationalization. The theory presents answer to the question of why people are increasingly trapped in a condition of bounded ethicality (Fleming and Zyglidopoulos 2008).
Further, Anand et al. (2004) classify rationalization strategies into several types: denial of responsibility, denial of injury, denial of victim, social weighting, appeal to higher loyalties, and balancing the ledger. Denial of responsibility can be described as a condition in which the offender perceives that he is just one of the perpetrators and does not assume the main responsibility of the violation. Denial of injury refers to the condition in which perpetrators believe that no damages arising from their actions. Denial of victim is a rationalization strategy that puts violated party as the one who deserves the unethical treatment. Social weighting is a form of rationalization in which the perpetrator makes selective social comparisons. Appeal to higher loyalties is the position assumed offenders as they argue that their violation of norms is due to their attempt to realize a higher order value. Balancing the ledger is the type of rationalization in which the actors rationalize that they are entitled to do corrupt acts because of their accrued credits (deed or sacrifice) in their jobs.
Consistent with the theory of cognitive dissonance, anxiety and guilt arising from a corrupt act are closely related to cognitive dissonance. Consequently, the perpetrator will evaluate and rationalize his actions to reduce the dissonance. In the case of small gratuities, the perpetrator will use rationalization that leads to a new cognitive equilibrium. In a bribery situation, rationalization will not be as easy as that in the case of the gratuities. The actor will perceive consequences of bribery outweigh those of gratuities. In short, it is harder to rationalize bribery than small gratuities. Thus, the following hypothesis is formulated.

H2
: Small gratuities will be easier to rationalize than bigger bribe.
When a person violates public trust by committing illegal and unethical acts such as corruption, he has a possibility to lie in order to conceal the shame. He can lie by giving false information (act of commission) or withhold the true information (act of omission) (Fleming and Zyglidopoulos 2008). Both types of lie will be mutually reinforcing. This escalation process requires rationalization at every stage. The rationalizaton of repeated corruption will decrease one's sensitivity in assessing moral standards. The person who experience this condition can justify his corrupt acts because "everyone does it". In whistleblowing literature, rationalization takes place in explaining why people behave differently when they are exposed to different contexts of essentially the same offense or fraud (Robinson et al. 2012;Kaplan et al. 2015). Moreover, rationalization has become common in shaping culture that is more accommodating to unethical conduct (Vadera and Pratt 2013). Tenbrunsel and Messick (2004) suggest that repeated rationalization will result in distorted moral reality. Individuals will tend to commit subsequent offense and they will find that rationalization gets easier in the future because of social conformity (DeZoort and Lord 1997). Corruption which originally initiated by some people could transform into a collective norm (Spicer 2009). The discussion leads to the following hypothesis.

Internal Control Quality
Internal control has two categories: accounting control and administrative control (Rodgers et al. 2015). The first category relates to the safeguarding of an organization's assets and the reliability of accounting information while the latter deals with efficiency of operations. Internal control includes a number of elements, such as fraud risk assessment, separation of duties, operation policies, and internal audit. Strong internal control promotes organizational culture that has the ability to detect and prevent dysfunctional behaviors.
Research results show the importance of investment in internal control improvement (Kaplan et al. 2015). Rodgers et al. (2015) find that training and education are effective to change individuals' behavior and reduce fraud action. Improving internal control quality will result in a more responsible management, skilled employees, as well as reliable systems. Kaplan et al. (2015) assert that managerial training can increase management's sensitivity in detecting, preventing, and reporting fraud. Furthermore, they argue that improving internal control provide signal that an organization has strong values to oppose fraud. The signal relates positively with the efforts of the prevention and reduction of unethical acts within organizations and their organizational trust (Seifert et al. 2014).
Fraud Triangle model (Cressy 1953) suggests that corruption can occur due to three conditions: pressure to engage in corrupt acts, opportunity, and rationalization. Pressure can take forms as need to achieve performance targets, personal satisfaction, or fear of failure (Murphy and Free 2016). Opportunity to commit fraud can be realized when an organization's control system is weak or when organizational culture and management style supports corruption (Laxman et al. 2014). Rationalization is associated with one's willingness to commit corruption and his ability to shift responsibility to other parties.
Strong internal control decreases the opportunity to commit corruption (Rodgers et al. 2015). Absence of the opportunity will discourage a person from engaging in corrupt acts. Internal control can activate social norms that can control individuals' opportunistic behavior. Thus, the above discussion brings expectations of the mitigating effect of internal control quality. The expectation can be formulated in the following hypothesis.

H4
: Internal control quality moderates the relationship between rationalization and subsequent offense; the higher internal control quality, the weaker the positive effect of rationalization on subsequent offense.

Research Design
To examine the hypotheses we use a 2 X 2 between-subjects factorial experiment design. Magnitude of corruption and internal control quality were manipulated by varying the experimental case scenarios. We develop an instrument to measure rationalization based on rationalization tactics model (Anand et al. 2004). The instrument covers six types of rationalization tactics. The scale contains six items and asks participants to what extent they agree with statements that are indicative of rationalizations. The statements are: I do not assume responsibility because giving or receiving gifts is a common practice (denial of responsibility), The party that provides the gift does not suffer losses (denial of injury), I do not ask for the gift, it is the giver's idea (denial of victim), Other people do worse acts (social weighting), If I turned down the gift, I would have disturbed social order (appeal to higher loyalties), and I deserve the gift(balancing the ledger). The items were scored on a five-point Likert scale ("Strongly disagree" to "Strongly agree").

Participants
One hundred seventy participants volunteered to participate in this experiment. All participants were public servants working in a large state university in Indonesia and each had at least 3 years of full time work experience. The subjects are selected because of their duties in protecting the interests of the State as civil servants. They may have different characteristics than other civil servants. However, it is not the intent of this research to pursue high external validity. As long as the subjects provide sufficient ecological validity and, thus, support the endeavor to achieve high internal validity, they can serve as justifiable subjects in this research.
Twenty five participants failed manipulation tests and were later excluded from the analysis, resulting in 145 usable responses. Participants included 85 females and 60 males whose ages ranged from 24 to 58 years old. Participants' mean age was 40.7 years and mean work experience was 16.1 years. The cell sizes for each of the four treatment groups varied between 34 and 39. Statistical tests did not reveal any significant differences in participant responses across there demographic characteristic.

Research Instrumental and Procedures
The subject participated voluntarily in the experiment during a university convention. We conducted the experiment without specifying the exact purpose of the study to the participants. Participants received a token of appreciation plus earnings depending upon their decision in the experiment. We randomly assigned participants to treatment conditions. The participants spent approximately 30 minutes reading and filling out the instrument questionnaire. The research instrument was pre-tested in a separate group of 30 full time students enrolled in a Master of Management program, most of who were working in a management position. This pilot test confirmed that there was substantial variance in the dependent variable and that a significant amount of this variance could be explained by the study's independent variables. Moreover, the pilot test showed that participants found the case realistic and interesting.
The research instrument contained a case scenario that was specially developed for this study. The scenario describes a situation in which each participant acts as the person in charge in an important procurement process in the university. In the process, each participant must decide whether to accept gifts from a supplier. In the condition of low magnitude of corruption (i.e. gratuities), each participant receives an offer to accept a monetary gift in amount of Rp5 millions after the completion of the procurement project. In high magnitude of corruption (i.e. bribery) condition, each participant is offered a monetary gift in amount of Rp50 millions from a supplier to make the procurement decision in favor of the supplier. In the two conditions, we asked the participants to indicate the likelihood that they would accept the gifts. At this time, we informed them that they would get an opportunity to win prize money in amount of Rp1.5 millions should they decided to accept the gift. Otherwise, they would only get the token of appreciation. Next, we measured their rationalization using the instrument discussed above.
We provide two scenarios for high or poor internal control quality. In high (low) internal control quality condition, participants were informed that the university has (lacks of) clear separation of functions, (has no) existing competent internal audit unit, and (has no) whistle-blowing mechanism against fraud. The three characteristic is based on studies by Near and Miceli (1995) and Kaplan et al. (2015). Next, we asked the participants to indicate the likelihood that they would commit the same act in the future. Finally, we administered an exit questionnaire, debriefed the participants, and distributed the prize money.

Manipulation Checks
The manipulation of magnitude of corruption was checked with the following post-experimental questionnaire item scored on a five-point Likert scale: "The gift was a form of corruption." The mean score in the low (high) magnitude conditions was 1.71 (3.42). The difference between the scores is significant (t = 12.44, p < 0.000). The manipulation of internal control quality was checked with the following item scored on a five-point Likert scale: "The university has strong internal quality systems." The mean score in the low (high) quality conditions was 1.68 (3.66). This difference is also significant (t = 19.44, p = 0.000).

Hypothesis Testing
The descriptive statistics for corruption acceptability, rationalizations, and subsequent offense are summarized in Table 1, a multivariate model with corruption acceptability and rationalization as the dependent variables of magnitude of corruption is presented in Table 2, and a 2 X 2 ANOVA model, with subsequent offense as the dependent variable, is presented in Table 3. As can be seen from Table 1, and consistent with H1, the average corruption acceptability is higher for gratuities (6.18) than for bribery (2.36). This difference, as presented in Table 2, is statistically significant (F = 5.39, p < 0.05), thus H1 is supported.
H2 predicts that smaller gratuities is easier to rationalize than bigger bribe. Table  1 further presents that the average rationalization is higher for gratuities (2.85) than for bribery (2.46), consistent with H2. The difference, as presented in Table 2, is statistically significant (F = 5.80, p < 0.05), and therefore H2 is supported.
H3 examines the positive effect of rationalization on individuals' subsequent offense. The overall ANOVA model as shown in Table 3 confirms our expectation (significant main effect for rationalization, F = 3.34, p = 0.00), and thus H3 is supported.
Table 3 also shows that internal control quality affects subsequent offense (moderately significant, F = 3.71, p = .057). However, we do not find the moderating effect of internal control quality as predicted in H4. Table 3 shows that there is insignificant interaction effect of rational-ization and internal control quality (F = .83, p = .614), and therefore H4 is not supported.

SUMMARY AND DISCUSSION
In this study, we investigate whether magnitude of corruption affects rationalization and corruption acceptability. Specifically, we hypothesize that gratuities are more socially accepted than bribery. Our data analysis supports the hypothesis. The finding is consistent with the literature, especially research that has conjecture that a violation is perceived as beyond the control of the perpetrator and as a product of external pressures, then the wrongdoing is more acceptable than that in control of the wrongdoers (Robinson et al. 2012;Kaplan et al. 2015). The study results indicate that gratuities are considered prosocial but bribery is antisocial. This finding is particulary is antisocial. This particularly interesting since generally corruption is considered as a heinous crime (Cameron et al. 2009). The results suggest that people are more tolerant of a particular type of corruption but at the same time condemn other kinds of corruption. This implies that the benefits of gratuities are likely to outweigh the costs of the corruption. Our study supports the view that individual attitudes toward corruption are complicated and contingent on myriad factors, such as cultural bias (Paldam 2002).  We also predict that smaller gratuities is easier to rationalize than bigger bribe. The results show that the study subjects are able to rationalize smaller gratuities easier than big bribe. The results are in line with the literature that supports the significant role of context beyond individuals dispositional factors in explaining why individuals act unethically if placed in opposed conditions (Fleming and Zyglidopoulos 2009;Hartmann and Maas 2010;. Our study supports the conjecture that rationalization plays an important role in explaining corruption. Our study also provide evidence for the theory of cognitive dissonance (Festinger 1957). In a bribery situation, rationalization becomes more difficult than in that of gratuities. Our study subjects perceive that consequences of bribery outweigh those of gratuities.
Next, our hypothesis that rationalization positively affects subsequent offense is also supported. This result implies that whenever individuals escalate their unethical conduct, they need to rationalize at every stage of the escalation process. The repeated rationalizaton decreases one's sensitivity in assessing moral standards and, thus, results in distorted moral reality (Kaplan et al. 2015;Vadera and Pratt 2013;Tenbrunsel and Messick 2004). Individuals find it easier to commit repeated offense since "everyone does it", "the offense does not harm anybody", "the wrongdoing is just a small matter", or they claim that they deserve the right to do so. Many would suggest that hypocrisy may be a sound argument; perpetrators of corruption may preaching virtue while practicing vice. While extant literature provides explanations from the perspective of types of people that are prone to corruption (Trevino and Youngblood 1990), the current study shed some light on how contextual factors make people behave in such a manner. The current study departs from a more traditional assumption that brings to the fore the individual and his deeds, and specifically shows that rationalization has an important effect on people attitudes toward corrupt behavior. When people think that their behavior falls short of common ethical standards, they justify the legitimacy of their actions.
Our last hypothesis is the moderating role of the quality of internal control on the relationship between rationalization and subsequent offense in that high quality of internal control weakens the relationship. The research fails to find support for such hypothesis. While internal control quality can reduce subsequent offense directly, we cannot find its mitigating effect on rationalization. Thus, our study does not provide empirical evidence to support the conjectures of Kaplan et al. (2015) and Rodgers et al. (2015).
Research finds that improving internal control quality by training and education will result in a more effective system and reduce fraud action (Rodgers et al. 2015). High quality of internal control also provides signal that an organization has strong values to oppose fraud (Kaplan et al. 2015) and the signal affects positively on the prevention and reduction of unethical acts within organizations (Seifert et al. 2014).
However, our research failure to find support for the hypothesis suggests that high quality of internal control may not automatically mitigate individuals intent to corrupt since it does not carry severe penalty on the behavior. We also speculate that the the mitigating effect of internal control is minimum since individuals already have strong rationalizations to justify their behavior. According to the Fraud Triangle Model (Cressy 1953;Murphy and Free 2016), corruption can occur because of pressure to engage in corrupt acts, opportunity, and rationalization. Strong internal control may decrease the opportunity to commit corruption (Rodgers et al. 2015;Laxman et al. 2014) but the internal control may fail to prevent individuals from shifting responsibility to other parties.
Our study has important implications for both researchers and practitioners. Prior research has shown that individuals will react differently depending on their types of wrongful acts (e.g. Robinson et al. 2012;Kaplan et al. 2015). We extend this line of research by examining how the differences in attitudes toward corruption are affected by the magnitude of the act. Specifically, our study demonstrates that rationalization reinforces people to behave in unethical ways.
Understanding this driver of corruption is important, as the fight against corruption can lead to prolonged and too costly process without proper and comprehensive understanding regarding the issue. To effectively reduce the effect of rationalization, organizations must provide opportunities for actors to contemplate and find the true causes behind a potentially corruption. Executives should create an environment where actors are at least encouraged to do the right thing. Employee education and the establishment of independent supervisory body could go a long way toward avoiding rationalization tactics.
Our study has a number of limitations that may encourage future work. First, we note that our findings' generalizability is limited by our use of a single scenario, along with testing only two factors of corruption (rationalization and internal control quality). Additionally, our subjects were from one large university in Indonesia. Despite the assumption that the subjects should have similar practices and experiences as the other large organizations, differences still do exist among organizations. More research is required to determine whether the results of this study are reproducible, and the limits of their generality. Second, our study focused on the implicit theory about causality and its effect relative to only the two factors. It is possible that other cultural values or factors may be more relevant. Future research may investigate these possible factors. Third, the present research was conducted in the context of morality inferences. Whether the results are generalizable to other inferences and judgments is unknown. It is important for the future researcher to consider whether other contexts can provide more valid measures of underlying relationships. We agree with the vast majority of the ethics and morality research community that continuing caution should be exercised down the path we have followed. Another major limitation of this study is the length of time subjects were exposed to the treatment conditions. In reality, people working in organizations may have established relationships from previous interactions. A useful extension of this research would be to conduct a similar study over an extended period of time.