Theoretical background

Benefits of a general decision-making model to study corruption.

In order to find out how the cognitive-motivational decision-making process leads to corruption, we draw on normative decision-making theory [19] and particularly the decision-making model by Engel, Blackwell [20]. This model describes how individuals make their decisions in a rational way by elaborating on different stages of the decision-making process. This basic model of consumer decision-making has been widely used in marketing research [35]. This model typically explains that there are some stages before consumers chose a specific product: namely, problem recognition, information search, and evaluation. The approach is particularly well suited for shedding light on the questions of why and how individuals conclude that corruption is the best means for reaching their goals or solving their problems. Thus, it goes beyond current decision-making models of immoral behavior [26] and initial research on corrupt decisions [8]. In contrast to the moral decision-making models discussed above [26, 29], the model proposed by Engel, Blackwell [20] is not limited to particular aspects of decision-making such as moral and specific goals. Thus, the model is more appropriate for the present study because we want to focus on more than just the moral aspects of the decision-making process. To that end, the present study utilizes an informed grounded theory design [21] in order to explore the decision-making process underlying corruption. Using the general decision-making model as a theoretical framework will hopefully reveal new or even contradictory insights. By abstaining from moral aspects, this study may minimize social desirability bias in the data collection process because it can avoid feelings of being judged as participants are not specifically concerned with the question whether the corruption they engaged in is moral or immoral.

The chosen model [20] consists of four stages: problem recognition, information search, evaluation, and choice (see Fig 1). Problem recognition involves processes related to identifying and being aware of a problem. The problem is recognized when individuals detect a discrepancy between the current state and a certain desired state [36]. For example, an employee might receive a very low salary (current state), but yearns to buy a house for their family (desired state). The desired state thus becomes a goal that the individual hopes to achieve via a certain behavior [37].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. The stages of the decision making-process (Engel et al., 1986).

https://doi.org/10.1371/journal.pone.0291819.g001

Once the ’problem’ and corresponding goal have been identified, individuals move to the second stage: information search. This stage includes activities focused on finding potential ways to solve the previously identified problem and reach one’s goal. In relation to the previous example, the employee may search for information about the various options available for earning the money to buy a house (e.g., borrowing money from the bank, saving their income, or engaging in corruption). According to Hoyer and MacInnis [35], individuals use a range of information sources when looking for ways to satisfy a certain goal. Sources may include internal sources (e.g., prior knowledge) or external sources (e.g., the internet or colleagues). The content of information is the type of information that an individual obtains before deciding to take action: for example, the costs and benefits associated with a particular solution. When individuals try to obtain information, they may focus on only one specific type of information and elaborate on its attributes, or they could search for several alternative pieces of information [38]. In the case of corruption, an individual could search for information about what is actually understood to constitute corruption, what the risks of getting caught are, or what penalties could be expected as a result of engaging in corrupt action. Information search has, to date, received meager attention in the corruption literature or in general research on unethical behavior. For instance, it is not included in the corrupt action model developed by Rabl and Kühlmann [8], nor models of moral/ethical decision-making [26, 29]. However, research from other fields has shown that the source and content of information are both important elements in the decision-making process [36]. Thus, models related to corruption need to incorporate this factor.

Once individuals have gained a certain amount of information on how to achieve their goal, the evaluation stage follows. This stage is concerned with processes in which individuals compare and contrast the different options to find out what the best option is [35]. In this stage, individuals examine the information about the attributes of the different options that they have gathered [36]. In relation to our previous example, employees will examine which option is the best for achieving their goal of buying a house; for example, whether borrowing money from the bank, saving money, or engaging in corruption will be the easiest, quickest, or safest option. This stage overlaps with the stage of moral judgment in the literature on moral decision-making [26], which refers to mental processes that determine what action is the most appropriate one to take. However, in those models, little (or no) attention has been given to the underlying reasons behind the conclusion that corruption is the best solution. Exploring the reasons that drove convicts of corruption to see corruption as an adequate solution to their ’problem’ is essential for understanding underlying motives behind corrupt action.

The last stage in the process is choice. In this stage, individuals choose corruption as the best option among various alternatives. In the corruption literature, there are several classifications of corruption, e.g., individual versus interpersonal corruption [39] or individual versus organizational corruption [18]. Since corruption is a complex phenomenon, there is no universal classification of corruption, however. Examples of corrupt behavior include bribery (giving some form of benefit in exchange for preferential treatment), embezzlement (taking or converting money, property or other valuables of public/organizational funds for personal benefit), and favoritism (misuse of authority to favor family, friends, or one’s own party) [40]. Asking individuals convicted of corruption about their concrete actions and thought processes may therefore lead to new, more psychology-oriented insights and potentially the discovery of additional (sub-) forms of corruption.

Aims of the present study and research questions

Using the normative decision-making model [20] as a framework alongside a qualitative approach (i.e., informed grounded theory) [21], our study explores each stage of the decision-making process for corruption (i.e., problem recognition, information search, evaluation, and choice). We aim to advance the current literature on the decision-making around unethical, immoral, and corrupt behavior [8, 26, 29]. This literature has largely ignored the reasons (i.e., goals) and means (i.e., information processes and evaluation thereof) that drive individuals’ conclusion that corruption is the best option for achieving their personal and professional goals. Based on the theoretical model by Engel, Blackwell [20], we formulated the following research questions:

Research question 1: What are the specific goals that individuals convicted of corruption wanted to achieve when engaging in corruption?

Research question 2a: What type of information content did individuals convicted of corruption search for before they decided to engage in corruption?

Research question 2b: Which sources did individuals convicted of corruption consider when searching for information?

Research question 3: What aspects of the different options to act did individuals convicted of corruption consider when they eventually chose corruption as a solution to a specific problem?

Research question 4: Which concrete behaviors did individuals convicted of corruption engage in that eventually led to a sentence of corruption and their subsequent imprisonment?

Design

In this qualitative study, we used an informed grounded theory approach [21] to answer the above research questions. Grounded theory is especially appropriate for research topics about which little is yet known [22]. This approach allows researchers to describe phenomena in a detailed way [41] and is especially useful when studying processes [42]. By connecting stages within a process [43], grounded theory allows for the emergence of a new theory and provides insight into the processes between categories. As the decision-making process of corruption has not been investigated yet, and we aim to study decision-making as a process, grounded theory is particularly appropriate for our study.

There are different variants of grounded theory [see for an overview, 44, 45]. In classic grounded theory, the researcher should delay conducting a literature review until the end of the data analysis [22, 46]. The reasoning behind this delay is to keep the researcher free and open to discovering theory from the data and avoiding bias, such as forcing data into a pre-existing theory that may not fit the data. Delaying the literature review in classical grounded theory is based on the ontological assumption of an objective reality [22] which can best be discovered by researchers if they are free from pre-existing knowledge [22, 42]. A popular later version of grounded theory, constructivist grounded theory [47], takes a different, relativist ontological, perspective that assumes that there is no objective reality, but that realities are socially constructed. Therefore, this perspective assumes that individuals cannot avoid their pre-existing knowledge when constructing reality [44]. Building upon this perspective, Thornberg [21] proposed informed grounded theory that is rooted in constructivist grounded theory but particularly emphasizes how a literature review can aid researchers and benefit the grounded theory research process. Thus, according to Thornberg [21], informed grounded theory represents a research process that is grounded in data by grounded theory methods, while being informed by existing research and theoretical frameworks. Rather than considering existing research and theoretical frameworks as obstacles for developing theory from the data, informed grounded theory considers them as sources of inspiration and tools to help a researcher focus on specific aspects and phenomena.

This study adopted an informed grounded theory approach [21] to explore the decision-making process underlying corruption. We draw on the normative decision-making model [20] as a theoretical framework. Therefore, the development of grounded theory in the current study is based on the decision-making model by Engel, Blackwell [20]. Following that model, we identified aspects of four different stages of the decision-making underlying corruption: goal, information search, evaluation, and behavior. In order to identify the aspects of each stage, we constructed codes, concepts, and theories that were grounded in the data by applying grounded theory methods [21, 41]. We considered the existing corruption literature as important insights to identify new potential concepts or theories in every stage of the decision-making process underlying corruption. Therefore, the development of the decision-making model underlying corruption in this study is based on the decision-making model on newly collected data and considers the existing corruption literature to analyze aspects within the decision-making process.

Participants

In order to better understand the decision-making process underlying corruption, we studied individuals who had experience with corruption and had been sentenced for their corrupt behavior. Our participants were individuals convicted of corruption in Indonesia. Indonesia’s corruption perception index (CPI) scores are consistently below 50 on a scale of 0 (being the most corrupt) to 100 (being the least corrupt) [23] indicating a high incidence of reported corruption cases. Almost every day, the Indonesian media reports about corruption cases, ranging from the regional level to the national level. After the Suharto era, political power in Indonesia became decentralized, leading to a spread of corruption particularly at the regional level [48]. The Corruption Eradication Commission of Indonesia (KPK), for example, reported that at least 32 persons in the position of the head of the regional government were prosecuted and caught in corruption cases between 2015 and 2018 [49]. Thus, because corruption is deeply ingrained into peoples’ daily lives, studying corruption in this context is particularly relevant. This study was therefore conducted in three regional prisons in Indonesia. Two of the prisons are male prisons, and one is a female prison. Although access to the specific target group (prisoners convicted of corruption) can be considered extremely challenging (in terms of official approval, but also in terms of openly speaking to and establishing rapport with participants), this study benefitted from one of the authors being from Indonesia, making this process feasible.

To gain access to the prisons, we sought a permit letter to conduct interviews by sending a research proposal to the East Java Regional Office of the Ministry of Justice and Human Rights, Indonesia. When the permit was issued, we brought it to the regional prisons with individuals convicted of corruption. In each prison, the prison authority appointed a public officer to interact with the first author. The first author explained the research proposal, including the characteristics of the study participants. A separate room was provided for interview processes in the prisons, to ensure that conversations would not be overheard by the other prisoners or guards. Each participant who meets the inclusion criteria was called by the public officer to the room to answer the interview questions. The interviews were conducted from May 2017 until August 2017.

The researcher interviewed each participant privately, face to face, using the prison’s provided room. Before the interview began, participants were informed about the research including the research context, the purpose of the study, as well as confidentiality of the research process. Then, the written informed consent was signed by each participant who agreed to participate in this study. Participation in this study was fully voluntary. Even though the invited participants were in prison, they had the option to refuse to participate in the interview without any negative consequences. In this study, three invited individuals declined to participate in the interview process.

In grounded theory, the principle of ’theoretical saturation’ is used to justify the sample size. Theoretical saturation refers to the point in the data collection and coding process at which no additional new conceptual categories emerge [22]. Throughout the 38 interviews that we collected in three prisons in Indonesia, our coding dictionary stabilized further and further (while we continuously acknowledged the criteria of conceptual depth suggested by Nelson [50], until we felt that, within the last interviews, a point of data saturation was reached. The participants included 27 men and 11 women who were between 32 and 73 years old (M = 51.7, SD = 9.4) at the time of the interview. Most participants were educated individuals (i.e., 15 participants had a bachelor degree, 13 held a master degree, three were doctors, seven had a senior high school degree, and the rest had obtained another educational level). In terms of organizational employment, 26 participants had worked for public organizations, while 12 had been employed in private organizations. Participants held a variety of positions at the time that they had engaged in corruption (e.g., principal, company owner, lecturer, regional government head, secretary, treasurer, broker, and tax officer). Furthermore, 22 participants had been in a leadership position, while 16 had been subordinates without any supervisory responsibility.

Data collection

We employed semi-structured interviews to explore the specific decision-making stages that individuals engaged in before deciding to behave corruptly. The interview guide was developed based on the stages of the normative decision-making process [20] described above. We attempted to find information on the four main stages of this process. In line with grounded theory, the interview questions were changed and adapted during the process of interviewing based on the insights gained from the previous interviews [22]. For example, the question about the goal that participants wanted to achieve was revised from “What was your goal when you made that decision?” to “What was the benefit for you of carrying out that behavior?” in order to maximize insights into participants’ goals for engaging in corrupt behavior (see S1 Appendix for examples of questions in the interview guide).

All interviews were conducted by the first author, who is a native Indonesian. Having an interviewer who is the same nationality of the interviewees is beneficial for helping participants feel comfortable and allowing them to talk in their native language [51, 52]. Likewise, the shared background can incline participants to feel greater trust toward the interviewer, which is crucial when talking about their unethical behaviors [51, 52].

Because corruption is very delicate topic, we tried to minimize any types of undesirable treatment effects and participant reactions [51, 52]. To this end, we started the interview by expressing empathy for their situation. We began the interview by asking participants about their behavior that led them to prison instead of using the term corruption in order to minimize social desirability bias. To create a safe environment for participants where they could honestly and openly discuss their experiences, we carefully ensured their confidentiality and privacy during the process of data collection. Participants were encouraged to talk about their experiences related to the behavior for which they had been convicted as openly as possible. Most of them were enthusiastic about participating in the interview. They were thankful for being heard and enjoyed being able to talk about their experiences, their coping mechanisms with the prison situation, and their personal opinion about the court decisions with a researcher. After the interview, participants completed a demographic questionnaire. All responses to the interview questions were recorded with an audio recorder (average duration was 45 minutes per participant). The internal ethical review committee at the authors’ home university in Europe approved the procedure of this study.

Data analysis

Following Wilhelmy and colleagues [53], we transcribed the interview data until nearly reaching saturation (i.e., until the number of new categories decreased significantly). Thus, we transcribed 23 interviews; the remaining 15 interviews were coded directly from the audio files. As suggested by Urquhart [43], the data obtained from the interviews was coded in the original language (Indonesian). It has been recommended that researchers use the original language as far along in the analysis process as possible, in order to capture the experiences of participants in an unbiased way and avoid loss of meaning [54]. Thus, we did not translate the complete interviews; for illustrative purposes, we only translated the codes and the corresponding excerpts from the transcripts. The coding was conducted by two coders (i.e., the first author and a research assistant) who are native Indonesians with excellent English skills. The first author trained the research assistant in three one-hour sessions. This training included how to assign a code to the text and organize categories. Since not all authors had mastered the Indonesian language, our team discussions revolved around the material (codes and excerpts) that was translated into English.

For the data analysis, we followed the three steps specified by grounded theory: open coding, axial coding, and theoretical coding [41]. Coding is a process of deriving and developing concepts from the data at hand [41]. Coding can be done word by word, phrase by phrase, sentence by sentence, or paragraph by paragraph [41]. Following Wilhelmy and colleagues [53], we coded each possible element that we considered worthy of coding, that is, single words, sentences, or whole paragraphs. Constant comparative analysis, which is the analytic process of comparing different pieces of data and looking for similarities and differences [41] occurred in the three coding steps. The constant comparative analysis was conducted based on the informed grounded theory approach [21]. In this process, we considered the existing literature and how this could be used to identify and label new categories [55]. Through all the coding processes, we made use of the coding software MAXQDA 2018.

First, in the open coding stage, we analyzed and coded the raw data. The purpose of this coding step is to understand the essence of what is being expressed in the raw data and assign a conceptual name (code) to describe that understanding [41]. Following the procedure described by Corbin and Strauss [41], the two coders independently coded the data and then met to compare and discuss the differences in their individual coding. Following previous grounded theory studies [53, 56, 57], we used a coding dictionary to facilitate the coding process. The coding dictionary is an evolving system of categories that is continually modified (e.g., new codes are added; some codes are changed) based on constant comparison between new codes and existing codes [56]. The two coders recorded their consensus on the appropriate use of code in the coding dictionary.

Second, in the axial coding stage, we organized codes into categories in order to elevate them to a more abstract level that is relevant for the research questions [43]. The purpose of this process is to find higher-level concepts called themes [41]. We constantly compared the codes to codes that had already been classified into categories or subcategories based on their similarities and differences. For example, all codes related to corrupt behavior could be categorized using the main code behavior, which covered possible subcategories such as bribery, embezzlement, favoritism, and manipulation of information. At this stage, the two coders also met to discuss differences in their reasoning for classifying sub-codes into main codes. These categorizations were documented in the coding dictionary.

The third and final step is theoretical coding. In this stage, the goal is to link various categories to a core category and reveal an underlying theory [41, 43]. A core category is a conceptual idea that could cover all other categories and represent the core theme of the research topic [41]. In order to investigate the relationships between our categories, we compared categories to each other and discussed the links between them [43]. In this process, we tried to identify categories that occurred together across each stage of the decision-making process. For example, we tried to discover whether any specific goal or information search activity was related to a certain type of corrupt behavior. To reveal these relationships, we used the code relations browser tool of the MAXQDA 2018. This tool is able to identify the relationships between codes by examining codes that were reported together by the participants. Finally, we integrated our findings, identified core categories and links between categories, and developed a diagram that illustrates our emerging theory, grounded in the data [41].

Overview

The aim of this study was to shed further light on the decision-making processes of individuals who had been sentenced for corruption. Following the normative perspective on decision-making [20], we first explored the last stage of the decision-making process, that is, participants’ behaviors that had been identified as corrupt and for which they had been sentenced. We then proceeded with the first three stages: the goals that our participants wanted to achieve by engaging in corruption; the information that they searched for before deciding to engage in corruption, and the aspects that they considered when choosing corruption as a solution to a certain problem. We chose this specific order, as we first wanted to familiarize participants with the interview setting and the topic [58] and let them explain their point of view before asking more detailed questions about their underlying motives and considerations. Fig 2 presents an overview of our findings. In Table 1, we provide more detailed information about higher-level categories (axial codes), and lower-level categories (open codes).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Conceptual model of decision making-process underlying corruption.

https://doi.org/10.1371/journal.pone.0291819.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Axial code, open code, and example quote.

https://doi.org/10.1371/journal.pone.0291819.t001

In relation to behavior, we found seven categories that we then organized into two broad categories. The first covers behaviors that represent typical types of corruption in the literature (i.e., embezzlement, bribery, manipulation of information, and favoritism). The second is behaviors that are not generally considered to be typical examples of corruption that we will call atypical types of corruption (i.e., behaviors related to assisting the corruption process, behaviors associated with the false application of policy/administrative procedure, and participants’ perception of being the victim of a conspiracy). Regarding participants’ goals, the data analysis revealed three different categories: personal goals, organizational goals, and social goals. In terms of the information search stage, we explored two types of information content that participants searched for before deciding to engage in corrupt behavior: corruption-focused and open content. Corruption-focused content is a category of information that only focuses on the attributes of corruption as a behavioral option, whereas open content is a category that focuses on alternative options. Participants searched for information by considering several information sources, namely: intrapersonal sources, interpersonal sources, and impersonal sources. Finally, participants chose corruption as a solution to a specific problem because of two main reasons, which we labeled pull and push reasons. Pull reasons include decisions to participate in corruption that were motivated by a positive or normative evaluation of the corrupt behavior itself, e.g., not considering the behavior to be corrupt, or considering the behavior to be ’safe’, an easy solution to the problem, or as accepted and enacted by others. On the other hand, push reasons referred to participants’ decisions to participate in corrupt behavior because others (e.g., authorities) had involved them in the process of corruption or because their corrupt behavior was considered to be the only solution to a problem.

Considering that corruption is a complex phenomenon and takes on various forms in different contexts [59], we describe our findings on the decision-making process based on the category of corrupt behavior that we found. For every corruption category, we describe the specific behavior and the associated decision-making stages (i.e., goals, information search, and evaluation aspects).

Typical types of corruption

This category refers to those typical forms of corruption that have already been identified in the literature. This category includes behaviors like bribery, embezzlement, favoritism, and manipulation of information. Bribery is a behavior that involves offering someone money, services, or other valuables in exchange for preferential treatment [13]. As can be seen in Table 1, the data analysis showed a number of bribery behaviors. For example, a school principal said:

I was only coordinating them. The thing was, in the circle of the Ministry of A, if we want to get some projects, we have to be bold, giving some [money] to individuals in the head office. So, every budget needs their approval. (Participant 6)

Embezzlement is characterized by taking or converting money, property, or other valuables for personal benefit [40]. For example, a tax officer reported: "I took it [the money] all. They did not know that. They only knew the data… From the individuals’ taxes, I gathered it. Then I made the false report" (Participant 10). In the data analysis, we also explored behaviors categorized as favoritism: the misuse of authority to favor certain individuals [60]. An example of behavior that could be categorized as favoritism is giving a loan intended for poor people to ineligible people, i.e., people who are not poor. Finally, manipulation of information refers to the intended or unintended abuse of (access to) information, such as cheating, violation of secrecy rules, disregarding the confidentiality of information, or concealing information [60]. For instance, some of our participants reported that they had engaged in behaviors such as falsifying a document, the creditor’s identity, or a financial report.

Participants who engaged in typical forms of corruption were mostly ’active’ decision makers and, thus, responsible for their behavior. They had different positions in their organization, such as school principal, tax officer, manager and secretary of an empowerment program, village head, and government official. They reported a variety of goals underlying their corrupt behaviors, including personal, organizational, and social goals. The majority of them either explicitly or implicitly reported personal goals underlying their corrupt behaviors. For example, a tax official said that he took much money from the tax account for his personal benefit: "First of all, honestly, I bought a house for my wife. I bought it for 800 million [rupiah], taken from that account" (Participant 10). The personal goals were not only related to money, but also to psychological benefits, such as a boost of self-esteem. For instance, a treasurer of the village empowerment program who manipulated the financial report said: "The point is, what I wanted, was to get the acknowledgment that I was able to do it, to do a good job" (Participant 32). She also stated that she manipulated the financial report to reach a social goal, namely for the village’s sake, as she said: "I only wanted to get the grant again, so the village gets the grant, that was my intention, even though, maybe, I took the wrong way" (Participant 32). Besides that, some other participants reported that organizational goals motivated them to engage in corruption, such as goals related to organizational reputation, organizational profit, and employee compensation (e.g., passing on the benefits arising from the corrupt action to the employees).

In terms of the information search process, all participants who engaged in typical forms of corruption made social comparisons before they finally acted corruptly. They typically searched for information about how other people, particularly their colleagues, behaved in the same situation. For example, a school principal who engaged in bribery stated: "There were several [schools]. I knew the other public schools did it [bribery] as well. I have a colleague. He was the one who told me that the others got the projects in the same way" (Participant 6). Some of the other participants engaging in typical forms of corruption also searched for information regarding the best procedure for engaging in corruption. In addition, some participants engaging in typical forms of corruption had searched for information on the safety and the legal consequences of their behavior. For instance, a tax officer who engaged in embezzlement told us, "Beforehand, I was, huh!… I have read all the books about corruption, I have read the news… how long the sentence would be if I get arrested" (Participant 10). Besides self-knowledge, informal relationships, and impersonal sources, colleagues from the other organizations were the most common information source who participants in this category consulted for information related to corrupt behavior.

Regarding the evaluation aspects of their decision, participants in this category chose corruption as a solution to a specific problem because of several reasons. The reason most frequently cited by participants was that they had considered the behavior to be safe. They made sure that the behavior would not create any problems for them in the future and would not lead to them getting caught. Participants commonly evaluated the behavior as safe based on their own prior experience or on information they had received from close others. For instance, a manager of an empowerment program who engaged in favoritism reported: "It was because in the first, second, third, and fourth-year, it was okay; there were no issues. Even in 2007, I got the award, sir. So, I continued to dare to do so" (Participant 33). A branch manager of a national company who engaged in embezzlement similarly said:

In my organization, many other individuals did it [took the money] for their own sake, but nobody went to jail. It encouraged me; nobody got caught. Maybe many individuals took more [money] than me, the director maybe, but all of them were free. (Participant 21)

Most of these participants also reported additional reasons, such as considering the corrupt behavior as common and enacted by many others, as an easy solution to the problem, or even as the only solution to a problem.

Atypical types of corruption

Theoretical implications and contributions

This study makes several theoretical contributions. First, we contribute to the literature on corrupt decision-making by further exploring the intra-individual, multi-stage process of corruption that goes beyond the previously identified underlying mechanisms and causes for unethical behavior outlined in ethical, moral, and corrupt decision-making models [8, 26, 29]. By exploring additional aspects at every stage of the decision-making process leading to corruption (e.g., how do individuals search for information?, Why do individuals choose corrupt behavior over another kind of behavior as a solution to a specific problem?) as well as the relationships between them, this research furthers our understanding of the cognitive-motivational mechanisms that lead to corrupt actions. Our study thereby extends previous decision-making models focusing on corruption, such as the corrupt action model [8] which draws on the Theory of Planned Behavior [34]. This model explains that corrupt action is driven by the desire and intention to achieve personal and professional goals through corruption [8]. This desire and intention are affected by several individual factors, including attitudes, subjective norms, and perceived behavioral control [8]. While the corrupt action model [8] also focuses on goals, it does not consider other stages that are typically covered in classical decision making theories. Therefore, our study provides a valuable alternative model for explaining the decision-making process underlying corruption. For example, we explored the information process aspects that were not addressed in the corrupt action model [8] and other ethical decision-making models [26, 29]. We found that individuals searched for information regarding the safety and legality of their actions, and compared how others behaved in the same situation. Furthermore, our findings extend previous studies that focus on personal and organizational causes of corruption [5–7]. This research has for example explored the interaction effect of ethical leadership and followers’ dark personality trait Machiavellianism on corruption. This study found that ethical leadership could reduce followers’ corruption, particularly when followers’ trait Machiavellianism is low [6]. Consistent with those findings, our study found that some participants engaged in corruption because they followed unethical instructions from their leader, indicating the important role leaders may play in followers’ corrupt behavior.

Second, using an informed grounded theory approach [21] and interviewing a sample of individuals who had actually been convicted of corruption enriches our understanding of how real-life actors of corruption came to the decision to act corruptly in greater conceptual depth. Most corruption studies have failed to generalize their findings to real-life contexts, as they are mostly conducted in laboratory settings with student samples and using scenarios or games [28]. By investigating the process of corruption with actual convicts of corruption, we were, for example, able to discover atypical forms of corruption (e.g., assisting the corruption process) alongside affirming more typical types (e.g., bribery). We also found that participants engaged in corruption because of push reasons, such as getting involved in corruption by following the instructions of one’s supervisor. Thus, the findings of our qualitative approach with real-life actors add to the external validity of corruption research and provide a deeper understanding of the mechanisms involved in making such decisions. Future studies may build on these findings and investigate the corruption process with a stronger systemic perspective (e.g., with social network analysis) that accounts for the dynamic interplay between active and passive actors and the associated intra- and interpersonal mechanisms. Previous corruption studies applying a network approach have indicated that corruption involves multiple actors [69, 70]. Each actor within the network has a different role, engages in different activities, and has individual connections [69]. Analyzing the intra-individual decision-making processes within such a corrupt network may further our understanding of the interplay between intra- and inter-individual factors in corrupt decision-making processes.

Practical implications and contributions

Our findings may help decision-makers in designing anti-corruption interventions or developing new policies. More specifically, our model provides detailed information about the stages of goal identification, information search, evaluation, and corrupt actions, in addition to the interrelations between these stages. Our results show that the type of corruption the person engaged in determines the decision-making process underlying corrupt behavior. For example, the most frequently reported reason for why participants engaged in bribery was that it was common practice. On the other hand, the most frequently mentioned reason for why participants applied policy/administrative procedures incorrectly was because they did not consider their behavior to be corrupt. Thus, decision-makers should tailor their interventions to the unique aspects of decision-making involved in a specific form of corruption [71], rather than utilize a one-size-fits-all solution.

It is important to note that participants who engaged in typical forms of corruption most often considered the issue of safety. Most of the participants concluded that their behavior was safe and that they would not be caught. They came to this conclusion based on their consultation with others and on their own prior experiences. This finding is consistent with the idea that the ethical climate within an organization plays a crucial role in ethical decision-making [72]. Specifically, individuals are more likely to behave corruptly when they work in a context where unethical behavior (e.g., corruption) is not punished, but is instead a socially accepted norm. Based on these findings, we emphasize the importance of considering the ethical climate when designing intervention programs to reduce or prevent corruption.

Another of our findings that may prove useful for policymakers is that some participants believed that their behavior was not corrupt. This indicates that they did not know which behaviors can and cannot be classified as corrupt. Consequently, we highlight the importance of distributing information about corruption laws in order to increase individuals’ awareness and understanding of what is and is not legal. In addition, interventions need to strengthen individuals’ personal responsibility for their own actions, their vigilance toward doubtful supervisory behaviors, and their self-esteem for withstanding orders from authorities that they are critical about. These may be useful complementary strategies in addition to fostering an ethical climate in organizations. Increasing awareness, vigilance, and self-responsibility in individual actors may also help to reduce automatic responding. Following a dual process logic, people could be encouraged to engage in more effortful information processing before proceeding with corruption. This could help to prevent people from unintentionally engaging in corrupt processes due to a restricted decision-making process, as described above.

Study limitations

Despite the contributions that this study makes, as outlined above, it also has several limitations. One such limitation is that our participants were all from Indonesia. It is possible that the results for individuals from other countries (e.g., Western countries in North America or Europe) will differ because previous studies have shown that corruption in one country is related to cultural aspects [73]. Furthermore, our participants’ corrupt behavior was only enacted at a regional level, and we did not investigate corrupt behavior at a national level. Including acts of corruption at a national level might have given us a more comprehensive understanding of large-scale acts of corruption (e.g., grand corruptions). Nevertheless, our sample was still quite diverse and included both men and women who worked in different types of organizations (private/public) and in a variety of positions (e.g., principal, lecturer, and regional government head, secretary, treasurer, and tax officer). In addition to the heterogeneity of our sample, the fact that we collected data until reaching theoretical saturation [22] helped to ensure that our insights may generalize to other samples and contexts.

Another potential limitation of our study design is that participants reported their corrupt behavior retrospectively. Because corruption is a socially undesirable act, it can create cognitive dissonance (i.e., discomfort arising from the idea that one is a good person but has committed a bad act). Therefore, we cannot rule out the possibility that participants engaged in retrospective rationalization in order to reduce cognitive dissonance [74], leading them to reframe their past corrupt behaviors as normal and acceptable [1]. Specifically, the denial of responsibility–one form of rationalization where individuals view circumstances beyond their control as responsible for their corrupt actions [74]–may explain why several participants reported that they only contributed in a small way to the corruption process, were victims of a conspiracy, simply followed orders, or did not have any other choice. In summary, we cannot rule out the possibility that cognitive reappraisal and justification processes influenced participants’ answers, such that they did not accurately reflect their original decision-making processes at the time of their corrupt involvement. However, in an attempt to limit this kind of desirability bias, we ensured complete confidentiality during the research process in order to make participants feel safe and allow them to talk honestly about their experiences [75]. Furthermore, as many participants in our sample did report intentional corrupt behaviors and did not deny responsibility, we believe that our findings may rather reflect an empirical reality rather than simply retrospective rationalization. Nonetheless, future studies should examine the decision-making process more directly and with less delay to avoid the issue of retrospection bias (e.g., in an event sampling study, if possible). Finally, we acknowledge that our intra-personal decision making focused approach may have limited the range of possible underlying mechanisms for corruption that could be identified in our research. Research in adjacent disciplines such as law, social psychology, or sociology could fruitfully employ our method to identify discipline specific mechanisms that could extend the model presented in this research.