How Much You See Is How You Respond: The Curvilinear Relationship Between the Frequency of Observed Unethical Behavior and The Whistleblowing Intention

This article uses a sample of 3076 employees working in the USA to examine the relationship between the frequency of unethical behavior that employees observe in their organization and their intention to whistleblow. The results confirm the expected curvilinear relationship based on the Focus Theory of Normative Conduct. This relationship is a combination of a diminishing negative relationship between the frequency of observed unethical behavior and the intention to whistleblow internally and a linear positive relationship between the frequency of observed unethical behavior and the intention to whistleblow externally. The beliefs of employees about how supportive their management is when handling whistleblowing reports moderates the relationship between the frequency of unethical behavior employees observe and their intention to whistleblow.


Introduction
There is a rich literature on whistleblowing in work settings. Because whistleblowing is an important phenomenon (Miceli et al. 1991(Miceli et al. , 2009) and a complex one (Culiberg and Mihelič 2017;Nayir and Herzig 2012), there have been many studies on the characteristics of whistleblowers, the process of whistleblowing, and the consequences of whistleblowing (Culiberg and Mihelič 2017;Near and Miceli 2016). An individual's morality and commitment, the intensity and type of unethical behavior, and the organizational culture and reporting infrastructure (Dungan et al. 2019;Vadera et al. 2009) are some of the individual, situational, and organizational antecedents that explain who, when, how, and to whom employees report unethical behavior.
This article focuses on the frequency of observed unethical behavior as an antecedent of whistleblowing. Although this situational antecedent has not been identified and studied in the literature, we predict that it influences whether and how employees will blow the whistle. More specifically, this article asserts that the more often employees observe unethical behavior in their organization, the weaker is their intention to whistleblow internally and the stronger their intention to whistleblow externally. The Focus Theory of Normative Conduct developed by Cialdini and colleagues (Cialdini et al. 1990(Cialdini et al. , 1991 will be used to explain these relationships and the moderating effect of the employees' beliefs about how supportive their management is in handling their whistleblowing reports.
To test the above-mentioned relationships, we used a sample of 3076 employees working in the USA. Before presenting and discussing the results, we first define the concepts, develop the hypotheses, and explain the method used.

Whistleblowing and Unethical Behavior
A widely used definition of whistleblowing is "the disclosure by organization members (former or current) of illegal, immoral, or illegitimate practices under the control of their employers, to persons or organizations that may be able to effect action" (Near and Miceli 1985, p. 4). Whistleblowing refers to employees reporting unethical behavior they have observed to someone who could stop and correct this behavior, sanction the perpetrator(s), and/or prevent this type of behavior from happening again. Unethical behavior is the 1 3 umbrella term for illegal, immoral, or illegitimate behavior (Jones 1991). This definition means that whistleblowing can be done internally or externally (although there are scholars, like Jubb (1999), who argue that the definition of whistleblowing should be restricted to external reporting). In the case of internal whistleblowing, allegations are brought to the attention of the perpetrator(s), (higher) management, and an organizational ethics hotline (Kaptein 2011). In the case of external whistleblowing, allegations are brought to the attention of a party outside of the organization, for example, a regulator or the press ). This definition of whistleblowing leaves open how employees blow the whistle, for instance, verbally (in person or by phone) or in writing (letter or email), and anonymously, confidentially, or publicly (Kaptein 2011).
The concept of whistleblowing is different from the concept of voice on which there is a growing literature (Liang et al. 2012;McLean et al. 2013). Unlike whistleblowing, voice is not limited to unethical behavior and is broader because it is about "employees recommending [sic] changes in the organization with the intent to improve operations or otherwise benefit the organization." (Near and Miceli 2016, p. 109). This is similar to what Huang and Paterson (2017) call "ethical voice", which concerns not only unethical behavior but also raising ideas to improve the ethics of policies, products, and procedures. Whistleblowing is also more specific than error reporting (Zhao and Olivera 2006), speaking-up (Edmondson 2003), and upward communication (Detert and Edmondson 2011). Whistleblowing is however broader than Treviño and Victor's (1992) concept of peer reporting. Whistleblowing can also include reporting about unethical behavior of people who have higher positions in the organization. Whistleblowing is also different from reporting of grievances; the latter involves employees raising personal issues, such as disputes with their employer (Martin 2014;Vandekerckhove and Phillips 2019).
The studies on unethical behavior as an antecedent of whistleblowing have mainly focused on the moral intensity of the unethical behavior. Jones (1991) developed the construct of moral intensity, which consists of several situational characteristics that define ethical dilemmas, to facilitate understanding of ethical decision-making. These characteristics have also been suggested for unethical behavior as an antecedent of whistleblowing decision-making: i.e., the magnitude of the consequences of the unethical behavior (Cassematis and Wortley 2013;Robinson et al. 2012); the social consensus on the extent to which the behavior is unethical (Valentine and Godkin 2019); the probability of the effect of the unethical behavior (Curtis and Taylor 2009); the temporal immediacy of the effect of the unethical behavior (Singer et al. 1998); the proximity of the unethical behavior (Singer et al. 1998); and the concentration of the effect of the unethical behavior (Alleyne et al. 2013).
Other factors of unethical behavior found to be antecedents of whistleblowing decision-making are the strength of the evidence of unethical behavior (Brink et al. 2013;Miceli and Near 1985) and the type of unethical behavior. Near et al. (2004) found that employees who witness mismanagement, sexual harassment, or unspecified legal violations are more likely to report them than employees who observe stealing, waste, safety problems, or unfair discrimination. Near and colleagues were however unable to explain their results due to a lack of data.
The frequency of unethical behavior is another potential antecedent of whistleblowing that has not been identified either by research on specific antecedents of whistleblowing or by published overviews of potential antecedents. For example, Bowes-Speary and O' Leary-Kelly (2005), Zhang et al. (2009), Miceli et al. (2012, and Valentine and Godkin (2019) mention many (potential) antecedents of whistleblowing but not the frequency of unethical behavior (or related terms such as wrongdoing or immoral, illegal, or illegitimate behavior). Neither does the proposed research agenda on whistleblowing by Miceli et al. (2012) and Culiberg and Mihelič (2017). Most research does take unethical behavior into account because whistleblowing is about how people respond to observed unethical behavior; however, they approach unethical behavior as a unique, isolated, or specific incident that people are confronted with. There are at least two exceptions. One is Lee, Heilmann, and Near's study (2004) that showed that the frequency and duration of respondents' experience of sexual harassment predicted their whistleblowing. Another exception is research by Miceli and Near (2002), where they found that whistleblowers believed that wrongdoing was more likely to stop when it was less frequent. In both studies, the frequency of observed unethical behavior did not include unethical behavior in general but was limited to the specific unethical behavior being reported. As discussed later in this article, the frequency of unethical behavior can be related to the moral intensity of unethical behavior.
This article focuses on the frequency of unethical behavior in general. We expect that the frequency with which employees observe (as a sensory act) unethical behavior in general in their organization influences whether they will blow the whistle once they are confronted with an unethical behavior. What is observed as unethical behavior is not limited to what the employees themselves have experienced as a victim [Lee et al. (2004) have studied this], but it can also include unethical behavior they have witnessed as a bystander. Kaptein (2008a) developed a scale for measuring the frequency of observed unethical behavior in an organization. Such behavior is observed by employees in their work and are morally unacceptable to the larger community. This scale was developed based on the codes of business organizations. It includes a broad spectrum of 37 items ranging from "Providing inappropriate information to analysts and investors" and "Submitting false or misleading invoices to customers" to "Violating environmental standards or regulations" and "Stealing or misappropriating assets (e.g., money, equipment, materials)." We use this scale to examine the frequency with which employees observe unethical behavior in general.
This article also focuses on the intention to whistleblow. There has been much research on the intention, propensity, preference, motivation, or inclination to whistleblow (Alleyne et al. 2019;Cho and Song 2015;Latan et al. 2019;MacNab and Worthley 2008;Mansbach and Bachner 2009;Nayir and Herzig 2012;Nayir et al. 2018;Valentine and Godkin 2019) defined as "the individual's probability of actually engaging in whistle-blowing behaviour" (Chiu 2002, p. 582). The idea behind studying the intention to whistleblow instead of the actual behavior is that intention is a prerequisite for behavior (Alleyne et al. 2013) and even a strong predictor of actual behavior (Ajzen 1991). It is widely recognized that the intention is not identical to actual whistleblowing behavior (Mesmer-Magnus and Viswesvaran 2005), that intentions to whistleblow do not always translate into actual whistleblowing behavior (Park and Blenkinsopp 2009), and that studying whistleblowing behavior has certain advantages (Miceli et al. 2008). However, when studying the relevance of antecedents, such as the frequency of observed unethical behavior, the influence of the antecedent on the intention must be studied first before studying the influence of the antecedent on behavior. This is because it is only when the influence of the antecedent on the intention has been ascertained that it becomes possible to study how the antecedent influences behavior via the intention. In addition, given that whistleblowing is a hypersensitive topic and studying actual whistleblowing behavior is quite complex (Bjørkelo et al. 2008) and inherently difficult to measure (Alleyne et al. 2019), it is therefore useful to study the relationship between the antecedents and the whistleblowing intention (Alleyne et al. 2019;Culiberg and Mihelič 2017). In this regard, we are following a well-established research tradition (Valentine and Godkin 2019).

Hypotheses
To understand the relationship between the frequency of observed unethical behavior and the intention to whistleblow, it is relevant to distinguish internal from external whistleblowing. As explained below, it is reasonable to expect the frequency of observed unethical behavior to have different relationships with either forms of whistleblowing. So we first develop separate hypotheses on internal and external whistleblowing. From these emerges a hypothesis on whistleblowing in general. The final hypothesis is about a moderating factor.

Internal Whistleblowing
It is reasonable to expect a diminishing negative relationship between the frequency of observed unethical behavior and the intention to whistleblow internally. The Focus Theory of Normative Conduct as developed by Cialdini and colleagues (Cialdini 2003;Cialdini et al. 1990Cialdini et al. , 1991 explains that this relationship can be reasonably expected. According to the Focus Theory of Normative Conduct, there are two kinds of norms relevant for explaining human behavior: injunctive and descriptive norms. Injunctive norms prescribe what ought to be done; they are perceptions about what constitute typically morally approved and disapproved behavior. Descriptive norms describe what is normal; they are characterizations of behaviors that are typically performed. People are influenced in their behavior both by how things should be (the injunctive norms) and how things are (the descriptive norms). Which of these two types of norms has the stronger influence depends on which specific norm is present, most demanding, and most salient for the individual at the time of the behavior; in other words, the norm on which the person focuses at the time. Descriptive norms can influence people's behavior more than injunctive norms can (Reid and Aiken 2013) because descriptive norms tend to motivate behavior in the immediate context and is more easily observable (Cialdini et al. 2006;Reno et al. 1993). The Focus Theory of Normative Conduct has been useful for studying many types of behavior, such as littering (Cialdini et al. 1990), alcohol consumption (Lee et al. 2007), theft (Cialdini et al. 2006), energy conservation (Schultz et al. 2007), sun protection (Reid and Aiken 2013), seat belt use (Litt et al. 2014), and counterproductive work behaviors (Jacobson et al. 2020).
There are at least two relevant injunctive norms that can influence the whistleblowing intentions of employees. The first injunctive norm refers to the unethical behavior that employees observe and the extent to which that behavior is considered unethical. According to Dozier and Miceli (1985) and Gundlach, Douglas, and Martinko (2003), the whistleblowing decision-making process comprises of at least four steps (which corresponds to the ethical decisionmaking models proposed by Rest (1986) and Jones (1991)): observing the behavior, analyzing and judging whether the behavior is unethical or not, forming an intention to whistleblow, and engaging in the act of whistleblowing. When employees observe unethical behavior, the question is whether they also analyze and judge it as being unethical. Whether this happens depends on the injunctive norms. The injunctive norms-established for example in laws and regulations, codes of conduct, policies and procedures-may 1 3 characterize the behavior as unethical. Business codes of ethics are also a source of injunctive norms because they prescribe many different types of behavior as ethical or unethical. Kaptein (2008a) also relied on business codes of ethics to generate the items in his scale for the frequency of observed unethical behavior.
A second relevant injunctive norm refers to the responsibility of employees when they observe unethical behavior at work. Once employees have judged that the observed behavior is unethical, the next question is whether they acknowledge their responsibility to report this (blow the whistle); this is part of the third step in the whistleblowing decision-making process. Many business codes of ethics prescribe that employees who observe unethical behavior in their organization have the moral responsibility to take action to redress the situation (Hassink et al. 2007;Jackson et al. 2010). This is consistent with the literature that considers whistleblowing a positive social behavior (Miceli et al. 2008), altruistic (Dozier and Miceli 1985), as loyalty to society (Andrade 2015), as a moral duty (Ceva and Bocchiola 2020), and as the normative responsibility of employees (Tsahuridu and Vandekerckhove 2008). In many legal jurisdictions, employees also have the legal responsibility to blow the whistle given the law that passive observers can be held liable for not taking adequate action (Tsahuridu and Vandekerckhove 2008). Therefore, when employees observe unethical behavior, injunctive norms do not only prescribe that they should evaluate the behavior as unethical but that they should also disclose and expose it (whistleblow).
Descriptive norms can have an opposite influence from that of injunctive norms in the whistleblowing decisionmaking process. In an ideal situation, descriptive norms are similar to injunctive norms in the sense that employees observe that colleagues blow the whistle when they evaluate an unethical behavior as unethical. However, when employees observe that not every colleague evaluates an unethical behavior as unethical and/or blows the whistle when they ought to, then descriptive and injunctive norms diverge. The more employees observe that their colleagues do not correctly evaluate unethical behavior and/or do not blow the whistle when they ought to, the more likely that the injunctive and descriptive norms diverge. This means, the more likely employees will believe that the unethical behavior is not entirely unethical or not unethical at all and that they should not always or maybe never whistleblow when they observe what they consider as unethical behavior. Instead of evaluating the observed behavior as unethical, it is whistleblowing about it that is characterized as unethical from a descriptive point of view.
When we use the Focus Theory of Normative Conduct to predict the relationship between the frequency of observed unethical behavior and the intention to whistleblow internally, we expect that the more often employees observe unethical behavior, the less inclined they will be to engage in internal whistleblowing once they are confronted with an unethical behavior. Cialdini et al. (1990Cialdini et al. ( , 1991 found that people were more likely to litter themselves when they saw more litters because the descriptive pro-littering norm diverged from the injunctive anti-littering norm. Smith et al. (2012) and Robinson et al. (2014) also found that when injunctive and descriptive norms conflict, people's intentions to follow the injunctive norms become weaker. The present study is not concerned with whether employees who observe unethical behavior are also more likely to behave unethically themselves. Robinson and O'Leary-Kelly (1998) have studied this topic. We are concerned with whether frequently observing unethical behavior leads to a change in employees' intention to whistleblow. If employees do not observe unethical behavior in their work, it is more likely that the descriptive norm does not diverge from the injunctive norm because unethical behavior and the descriptive norm "do not whistleblow when unethical behavior is observed" are not salient. If the frequency of observed unethical behavior increases and becomes more salient, then the descriptive norm might become "unethical behavior is the norm." In addition, the higher the frequency of observed unethical behavior, the more salient this descriptive norm becomes relative to the injunctive norm "ethical behavior is the norm." So the higher the frequency of observed unethical behavior, the more likely employees will think that unethical behavior is normal and that whistleblowing about unethical behavior is abnormal (because one only blows the whistle on behavior that is supposed to be not normal or not accepted). This would then make it less likely that employees will form the intention to whistleblow internally when they are confronted with unethical behavior.
It is reasonable to expect that the relationship between the frequency of observed unethical behavior and internal whistleblowing is diminishingly negative. The study of Cialdini et al. (1990) helps illustrate this point. The study found that the more litter there is in the surrounding (in their case, flyers that were thrown on the ground), the more likely people will also litter. However, this effect was diminishing in that as the amount of litter increased, the less difference each additional piece of litter made on the likelihood that people littered and that given a certain amount of litter (in their case eight flyers lying on the ground), additional pieces of litter did not lead to more people who littered. In the study, everyone already having littered did not explain this last point, because almost 60% of the participants did not litter when there were eight flyers lying on the ground. We can also expect that the more unethical behavior employees observe, the less any incremental increase in observations will influence their intention to blow the whistle internally. The reason for this diminishing effect is that one or a few more observations of unethical behavior has a larger contribution to the salience of the descriptive norm "unethical behavior is normal and blowing the whistle is abnormal" than when more observations of unethical behavior are added to an already large number of observations of unethical behavior. The higher the frequency of observed unethical behavior, the more it confirms to the employees that the descriptive norm is "blowing the whistle internally is not normal" and the less it changes the descriptive norm. Furthermore, we expect that a high frequency of observed unethical behavior does not lead to the situation where nobody in the organization would have any intention to blow the whistle internally. Even when employees observe unethical behavior very frequently, the injunctive norms could still be acknowledged and influential, although they might be less strong than when they converge with the descriptive norms. Jacobson et al. (2020) shows that descriptive and injunctive norms can exert independent effects on behavior even when one of the two norms has a stronger influence because the weaker norm can nevertheless influence a behavior when it triggers within the person an awareness of that norm. Moreover, the personal norms of employees, which differ from injunctive and descriptive norms in that they refer to individual-level standards rather than the standards of social groups (Niemiec et al. 2020), may explain why employees are less or not influenced by their social context and keep to their intention to whistleblow internally even when they very frequently observe unethical behavior. Therefore, we formulate our first hypothesis:

H1
The relationship between the frequency of unethical behavior observed by employees in their organization and their intention to engage in internal whistleblowing is diminishingly negative.

External Whistleblowing
A different relationship holds between the frequency of observed unethical behavior and the intention to whistleblow externally.
An injunctive norm may prescribe that when employees observe unethical behavior, they first have to consider internal whistleblowing before they decide to do it externally. Current laws, business codes of ethics, and whistleblowing policies prescribe that employees should first report the wrongdoing internally (Hassink et al. 2007;Tsahuridu and Vandekerckhove 2008). This prescription is also supported by the literature: i.e., internal whistleblowing is considered to stimulate more self-control (Miceli et al. 1991), selfefficacy (Miceli et al. 2001), and self-diagnosis (Chen and Lai 2014) and tends to be less damaging for the organization and colleagues (Nayir et al. 2018) compared to external whistleblowing. Employees who straightaway go to the press when they observe for the first time a small incident at work cannot be considered ethical from the point of view of the above-mentioned injunction. However, injunctive norms do not forbid external whistleblowing but may consider it as morally justified or even required under certain conditions: e.g., when internal whistleblowing does not lead to the right follow-up action in the organization, the unethical behavior is serious, there is sufficient evidence about the unethical behavior, or when it is reasonably certain that external whistleblowing will stop the unethical behavior (De George 1990;Velasquez 1998).
Descriptive norms can also suggest to employees that they should first report internally before doing it externally. As gathered from the literature (Callahan and Collins 1992;Mansbach and Bachner 2010;Miceli et al. 2008;Miceli and Near 2002;Rothschild and Miethe 1999; Vandekerckhove and Phillips 2019), employees are more likely to blow the whistle internally before doing it externally. Two of the reasons whistleblowers give for going external are their lack of trust in their organization stopping the unethical behavior and their belief that only an external party can stop the unethical behavior (Dworkin and Baucus 1998;Miceli et al. 2008). Employees judge internal whistleblowing more favorably than external whistleblowing (Culiberg and Mihelič 2017). The latter is often seen as an extension of the whistleblowing process in case internal whistleblowing does not bring about the expected responses (Miceli and Near 1992;Vandekerckhove and Phillips 2019). Therefore, employees who decide to reveal unethical behavior are more likely to do it internally first.
We can assume that the more frequently employees observe unethical behavior, the more they will be inclined towards external whistleblowing. The more frequently employees observe unethical behavior in their organization, the more likely they will believe that unethical behavior is (considered) normal in their organization, that internal whistleblowing is less expected, appreciated, or approbated, and thus the less inclined they will be to blow the whistle internally when they witness unethical behavior. Furthermore, the less inclined employees are to report the wrongdoing internally, the more external whistleblowing becomes their only option. The option to externally whistleblow thus becomes more salient the more frequently employees observe unethical behavior. This makes unethical behavior a bigger problem in the organization because employees observe not just a single isolated incident but rather a pattern of different types of unethical behavior (Callahan and Dworkin 1992;Near and Miceli 1995). The more serious the problem of unethical behavior becomes for the organization, the stronger becomes the injunctive norm that employees should reveal the problem externally because the organization appears unable to solve the problem by itself. In this case, the salience of the conflict between the descriptive norm (that unethical behavior is normal within the organization) and the injunctive norm (that employees should expose unethical behavior when they observe it) also becomes more salient. This conflict then becomes a reason for resorting to external whistleblowing. At this point, the reporting is no longer just about one or a few incidents of unethical behavior but concerns a pattern of unethical behavior supported by the descriptive norm in the organization that unethical behavior is normal. We can therefore expect that the more frequently employees observe unethical behavior in their organization, the more they will be inclined towards external whistleblowing. We can also expect this relationship to be linear: every increase in the frequency of observed unethical behavior makes it clearer to employees that external whistleblowing is the only or final option for addressing the unethical behavior and that they have the responsibility to engage in this option. This leads to the next hypothesis: H2 The relationship between the frequency of unethical behavior observed by employees in their organization and their intention to engage in external whistleblowing is positive and linear.

Internal and External Whistleblowing
We may expect a curvilinear relationship between the frequency of observed unethical behavior and the intention to whistleblow when we combine the expected relationships for the intentions to internally and externally whistleblow. The relationship is expected to be U-shaped because employees who have only observed a single incident of unethical behavior will not or less likely have the belief that internal whistleblowing is not appreciated or effective, and therefore they will be more inclined to blow the whistle internally rather than externally. However, as stated above, the more frequently employees observe unethical behavior, the less they are inclined to blow the whistle internally and the more they are inclined to do it externally. We then expect the intention to pursue internal whistleblowing to become weaker more quickly than the intention to whistleblow externally becomes stronger. The reason for this is that not every employee who is no longer inclined to internally whistleblow will necessarily have the intention to do it externally (Kaptein 2011;Vandekerckhove and Phillips 2019). For example, external whistleblowing is riskier and more timeconsuming for employees than internal whistleblowing (Dozier and Miceli 1985;Dworkin and Baucus 1998). A U-shaped relationship implies that the intention to blow the whistle is strongest among employees when their frequency of observed unethical behavior is relatively low (when they are most inclined to whistleblow internally) or high (when they are most inclined to whistleblow externally). Therefore, the third hypothesis is:

H3
The relationship between the frequency of unethical behavior observed by employees in their organization and their intention to whistleblow is curvilinear (U-shaped).

Management Support for Whistleblowing
How the frequency of observed unethical behavior influences the intention of employees to whistleblow depends on how employees interpret that frequency. As explained above, employees infer from the frequency with which they observe unethical behavior what the descriptive norms are. Their beliefs about how management will deal with their whistleblowing reports are expected to influence how they interpret that frequency. In the whistleblowing literature, how management handles whistleblowing is referred to as supervisor or (top) management support (Mesmer-Magnus and Viswesvaran 2005;Sims and Keenan 1998). This concept is related to the broader one of perceived organizational support (Eisenberger et al. 1986), which is about the global beliefs employees develop concerning the extent to which the organization values their contribution and cares about their well-being. In our case, management support is about the extent to which employees believe that management values whistleblowing reports and cares about whistleblowers.
Management support or trust in management (Cassematis and Wortley 2013) is important for understanding the intention of employees to whistleblow. Management support concerns the expectations of employees about the extent to which management will take appropriate action (Miceli and Near 1994a, b), handle the report confidentially (Miceli et al. 2008), protect the whistleblower from retaliation (Cho and Song 2015;Detert and Burris 2007), discipline the perpetrator(s) fairly and consistently (Keenan 1990), and will ensure that the whistleblowing report is useful and effective in the sense that the unethical behavior will be "terminated at least partly…and within a reasonable time frame" (Near and Miceli 1995, p. 681). Research confirms the importance of management support for whistleblowing. A meta-analytic research by Mesmer-Magnus and Viswesvaran (2005) shows that management support positively influences employees' whistleblowing intentions. Near et al. (2004) found among employees of a US military base that the most frequent reason given for not blowing the whistle was the belief that managers could or would not make any changes and that the wrongdoing would continue unabated. The second most frequent reason they found was the fear of reprisal (cf. Near and Miceli 2016). For the whistleblowers, the most common reprisals were poor performance appraisals and tighter scrutiny of their daily activities by management. Keil et al. (2010) report that trust in managers was the most influential predictor of IT employees' intention to whistleblow.
We assume that if employees believe that management supports whistleblowing, then the frequency of unethical behavior will have no impact on their intention to blow the whistle internally. High management support prevents employees from interpreting an increase in the frequency of observed unethical behavior as an absence of appreciation for whistleblowing. However, when management support is low, high frequency of observed unethical behavior makes this lack of support salient, thereby strengthening the descriptive norm that whistleblowing is not appreciated and increasingly weakening the intention of employees to whistleblow internally. Miceli et al. (2001) refer to this as the signaling effect that observations of unethical behavior has regarding the organization's response to unethical behavior and the chances of successful whistleblowing in that context. When management support is high, high frequency of observed unethical behavior makes this support salient, thereby strengthening the descriptive norm that whistleblowing is appreciated and reinforcing the intention of employees to whistleblow internally. Employees may then, for example, interpret the high frequency of unethical behavior as their organization struggling to prevent unethical behavior sufficiently, and they may feel stimulated, due to high management support, to disclose any observed unethical behavior to help their organization and management in this effort. Therefore, we expect management support for whistleblowers to moderate the relationship between the frequency of observed unethical behavior and the intention of employees to engage in internal whistleblowing.
Regarding external whistleblowing, we can expect management support to also moderate the relationship with the frequency of observed unethical behavior. The aboveformulated Hypothesis 2 holds that when the frequency of observed unethical behavior increases, employees will be more inclined towards external whistleblowing. Management support is supposed to moderate this relationship in the sense that when employees believe management support for whistleblowing is low, an increase in observed unethical behavior will influence their inclination to whistleblow externally less than when they believe management support is high. The Focus Theory of Normative Conduct again provides an explanation for this moderation effect. When there is low management support, an increase in the frequency of observed unethical behavior makes more salient the descriptive norm "unethical behavior is normal and blowing the whistle is abnormal." The higher the frequency of unethical behavior when management support is low, the more it is emphasized to employees that management accepts or even appreciates unethical behavior. High management support implies that management would take action to combat unethical behavior because the more unethical behavior takes place, the more it becomes a serious, urgent, and undeniable problem. The belief of employees that their management accepts unethical behavior encourages their belief that whistleblowing is less appreciated or even not accepted in their organization. The more salient this descriptive norm is, the more it discourages the intention of employees to externally whistleblow. When management does not combat unethical behavior internally, employees may not consider themselves in the position-because they are hierarchically lower than management-to combat unethical behavior through external whistleblowing. Employees may also believe that they run more personal risks when management support is low than when it is high. When management support is low, there will presumably be stronger disapproval for external than for internal whistleblowing because the former involves informing a party outside the organization not only about the unethical behavior but also about the low management support within the organization. In this case, management may consider external whistleblowing as disloyalty to management (cf. Nayir et al. 2018) and therefore is more likely to take measures against employees who whistleblow externally (Dozier and Miceli 1985). While low management support restrains the inclination of employees to blow the whistle externally, we still expect that an increase in the frequency of observed unethical behavior will lead to a stronger inclination to externally whistleblow because, as argued above, the more unethical behavior employees observe, the stronger becomes the injunctive norm that they should report externally.
For whistleblowing in general, management support for whistleblowing is expected to moderate the relationship with the frequency of observed unethical behavior. Combining the above-described expected moderating role of management support in the relationships of frequency of observed unethical behavior with both internal and external whistleblowing, we can then infer that management support should also have a moderating role in the relationship between the frequency of observed unethical behavior and whistleblowing in general. This leads to our final hypotheses:

H4a
The relationship between the frequency of unethical behavior observed by employees in their organization and their intention to internally whistleblow is moderated by the employees' beliefs about their management's support for whistleblowing, such that this relationship is weaker if employees perceive high management support for whistleblowing and stronger if employees perceive low management support for whistleblowing.

H4b
The relationship between the frequency of unethical behavior observed by employees in their organization and their intention to externally whistleblow is moderated by the employees' beliefs about their management's support for whistleblowing, such that this relationship is stronger if employees perceive high management support for whistleblowing and weaker if employees perceive low management support for whistleblowing.

H4c
The relationship between the frequency of unethical behavior observed by employees in their organization and their intention to whistleblow is moderated by the employees' beliefs about their management's support for whistleblowing, such that this relationship is stronger if employees perceive high management support for whistleblowing and weaker if employees perceive low management support for whistleblowing.

Sample
To test our six hypotheses, we needed a large and diverse sample of participants. Some types of intended responses, such as external whistleblowing, may be rare, and so the larger the sample, the higher the likelihood of finding such responses (cf. Cassematis and Wortley 2013). A variety of respondents working in different settings were needed to get a higher variance in the number of observed unethical behavior (cf. Culiberg and Mihelič 2017). If the study were conducted in an organization where respondents have the same frequencies of observed unethical behavior, then we would not find any relationships because the independent variable would be constant. A diverse group of respondents also helps in finding widely applicable results. With this in mind, we collected data from the working population of the USA. With the assistance of the panel-database firm EMI-Online Research Solutions and KPMG, we compiled a sample of adults working full-time for organizations that employ at least 200 people in 16 selected industries spread throughout the USA. The privately registered, pre-screened members of the panel received a nominal financial reward for their participation. We excluded respondents who did not answer all questions or who gave very unrealistic answers. An example of unrealistic answer is giving all 37 items of the unethical behavior scale (the independent variable) the same answers (excluding the never and always options). The data collection yielded 3075 usable surveys, with a wide variety of demographics. 1

Intention to Whistleblow
For the dependent variable, intention to whistleblow, we used the same scale used by Kaptein (2011). This scale consisted of one item for external whistleblowing and three items for internal whistleblowing. Respondents were asked what they would do if they observed unethical behavior: "If I observed a violation of my organization's standards of conduct, I would…." This was followed by the three items that measured internal whistleblowing ("Try resolving the matter directly," "Notify my supervisor or another manager," and "Call the ethics or compliance hotline"), and by the one item that measured external whistleblowing ("Notify someone outside the organization"). For each response item, we used a 6-point Likert scale, ranging from 1 = strongly disagree to 5 = strongly agree, with 6 = not applicable as the alternative response and treated as a missing value in the analysis. We computed the variable "intention to internally whistleblow" by averaging the answers of every respondent on their three responses about internal whistleblowing. The variable "intention to whistleblow" was computed by averaging each respondent's score on "intention to internally whistleblow" and their answer to the single item on external whistleblowing. Requesting respondents to indicate what they think they would do if they observe unethical behavior in their organization is similar to what has been done in studies other than Kaptein's, such as those of Victor, et al. (1993), King (2001), Tavakoli et al. (2003), Ayers and Kaplan (2005), and Cho and Song (2015). The advantage of asking this is that every respondent can answer this question, including those who have not observed any unethical behavior. Furthermore, we 1 Of the 3,075 respondents, 51.8% were female. The age distribution was: 3.1% were 18-24 years old, 9.2% 25-29 years, 14.2% 30-34 years, 13.4% 35-39 years, 12% 35-39 years, 13.2% 45-49 years, 14.5% 50-54 years, and 20.5% were 55 years or older. With regard to job tenure, 5.8% have been working for less than a year in their current organization, 10.6% from 1 to 2 years, 22.4% from 3 to 5 years, 22.4% from 6 to 10 years, and 40.1% for more than 10 years.. A little less than a quarter (23.4%) worked for an organization with 200-1,000 employees, 15.8% for companies with 1,000-2,999 employees, 10.9% with 3,000-4,999 employees, 14.4% with 5000-9999 employees, and 35.4% for organizations with 10,000 or more employees. With regard hierarchical level within the organization, 54.1% did not hold managerial position, 13.0% worked as supervisor, 20.9% as mid-level manager, 7.1% as senior manager or junior executive, and 4.8% as senior executive, officer, or director. A wide range of functions were included: 9.1% sales and marketing; 11.6% operations and service; 6.9% manufacturing and production; 10.3% research, development, and engineering; 1.9% purchasing and procurement; 12.2% technology; 2.2% training and education; 3.2% quality, safety and environmental; 8.7% clerical and support; 8.9% general management and administration; 7.2% finance and accounting, 1.3% legal and compliance; 1.0% internal audit and risk management; 0.8% public and media relations; 2.8% government and regulatory affairs; and 12.0% other functions. The respondents also represented a wide range of sectors: 5.1% aerospace and defense; 9.8% automotive; 10.0% banking and finance; 10.0% consumer markets; 4.8% chemicals and diversified industrials; 9.6% electronics, software, and services; 1.7% forestry, mining, and natural resources; 10% government and public sector; 9.5% healthcare; 10.1% insurance; 3.2% media and communications; 1.6% oil and gas; 3.1% pharmaceuticals and life sciences; 1.7% power and utilities; and 9.9% real estate and construction.
Footnote 1 (continued) were interested to learn whether employees who have never observed unethical behavior differed in their intention to whistleblow when they would observe unethical behavior in the future from those who have observed unethical behavior in the past. There are also several reasons we posed the question in terms of observed violation of organizational standards of conduct rather than observed unethical behavior: the former requires less explanation, is more concrete for respondents, less dependent on the respondent's personal ethical norms, and is applicable to others in the same organization. The scale reliability was low, as indicated by Cronbach's alpha of 0.57.

Frequency of Observed Unethical Behavior
For measuring the frequency of observed unethical behavior, we used Kaptein's (2008a) scale. This scale is an improvement on the only other existing scale for unethical behavior developed by Newstrom and Ruch (1975). Kaptein's scale consists of 37 items, with a time frame of 12 months (i.e., "In the past 12 months, I have personally seen or have firsthand knowledge of employees or managers…") and uses a five-point frequency scale: 1 = never, 2 = rarely, 3 = sometimes, 4 = often, and 5 = almost always. We calculated a general score for the frequency of observed unethical behavior by averaging the answers to all 37 items of each respondent. The scale reliability was high, as indicated by Cronbach's alpha of 0.99.

Management Support for Whistleblowing
We measured management support for whistleblowing by using five items, following the elements described in the above section Management Support for Whistleblowing. Similar to what we did for the variable intention to whistleblow, we asked what management would do rather than what management has done. The question posed was, "If I reported a violation of my organization's standards of conduct to management, I believe….", followed by the items "appropriate action would be taken," "my report would be handled confidentially," "I would be protected from retaliation," "those involved would be disciplined fairly regardless of their position," and "I would be satisfied with the outcome." We used the Likert scale for each response item, from 1 = strongly disagree to 5 = strongly agree, with the sixth alternative response (6 = not applicable) treated as missing value in the analysis. Since these items were not taken from an existing validated scale, we conducted a principal component analysis. Based on the screen plot and eigenvalues, one factor was extracted. The initial eigenvalue of the first factor was 4.0 and 0.32 for the second factor. The first factor explained 80.84% of the variance. All factor loadings were higher than 0.88. The reliability of the scale was high with Cronbach's alpha of 0.94. The mean of the variable was calculated by averaging the answers on the five items.

Control Variables
To take into account the possibility that the respondents' employment and demographic characteristics might have influenced their responses, the same five control variables as in Kaptein's (2011) study were entered in the analyses: the respondent's gender (with two categories, 1 = male, 2 = female); age (with eight categories, from 1 = 18-24 to 8 = 55 and over); job tenure (with five categories, from 1 = less than 1 year to 5 = 10 years or more); and hierarchical level (with five categories, from 1 = individual contributor (not supervising others) to 5 = Senior Executive/Officer/ Director); and size of the respondent's organization (with 5 categories from 1 = 200-999 employees to 5 = 10,000 employees or more).

Analyses
We used hierarchical regression analysis to test our hypotheses. All predictor variables were grand-mean centered. The interaction of the mean-centered frequency of observed unethical behavior and the mean-centered moderator of management support were calculated by multiplying them. To exclude interference between the dependent variables, the intention to externally whistleblow was included as control variable in the analyses with the intention to internally whistleblow as dependent variable, and vice-versa. Table 1 depicts the means, standard deviations, and intercorrelations of all research variables. The bivariate correlations between the frequency of observed unethical behavior and the intended whistleblowing responses were all significant at p < 0.01: r = 0.09 for internal whistleblowing, r = 0.44 for external whistleblowing, and r = 0.37 for whistleblowing in general (internal and/or external). Correlations between management support and the intended whistleblowing responses were also significant and positive: r = 0.54 for internal whistleblowing, r = 0.08 for external whistleblowing, and r = 0.33 for whistleblowing in general. The correlation between the independent variable, frequency of observed unethical behavior, and the moderating variable, management support, was insignificant (r = − 0.02; ns), indicating discriminant validity between these two variables.

Results
To examine the distinctiveness of the variables, a confirmatory factor analyses (CFA) was conducted. The measurement model consisted of three factors: frequency of observed unethical behavior, intention to whistleblow, and management support for whistleblowing. The results indicated that the three-factor model provided an acceptable fit of the data (CFI = 0.91). The three-factor model had a significantly better fit than the one-factor model (CFI = 0.53) and the two-factor model when whistleblowing was the separate factor (CFI = 0.54). The fit was slightly better than the twofactor model when management support was the separate factor (CFI = 0.87) and when unethical behavior was the separate factor (CFI = 0.90). The three-factor model did not have a better fit because the item external whistleblowing did not load very well in the scale for the intention to whistleblow. This explains the low scale reliability of the intention to whistleblow. Removing the item external whistleblowing from this scale would lead to a slightly better fit of the threefactor model (CFI = 0.93). However, removing this item was not an option because this item is crucial in studying the concept of whistleblowing.
Regarding Hypothesis 1, we expected the relationship between the intention to internally whistleblow and the frequency of observed unethical behavior to be diminishingly    Table 2 and Fig. 1 show this to be the case: the final model (Model 3) has a regression coefficient with the frequency of observed unethical behavior of β = − 1.35 (p < 0.01) and a squared frequency of observed unethical behavior of β = 1.31 (p < 0.01). Entering the quadric term increased the explained variance by 5.2%. Therefore, Hypothesis 1 is supported. Table 3 and Fig. 2 show the results concerning Hypothesis 2. We expected the relationship between the frequency of observed unethical behavior and the intention to externally whistleblow to be positively linear. Model 3 shows that for external whistleblowing only the frequency of observed unethical behavior was significant with β = 0.34 (p < 0.01), whereas the squared frequency of observed unethical behavior was not significant with β = 0.04. Entering the quadric term here did not increase the explained variance, showing that the relationship is linear. Entering the independent variable frequency of observed unethical behavior after the control variables led to an increase of 12.5% in the explained variance. Therefore, the more frequently employees observed unethical behavior, the stronger the intention to externally blow the whistle became. Therefore, Hypothesis 2 is supported. Table 4 and Fig. 3 present the results for Hypothesis 3. We expected the relationship between the whistleblowing intention and the frequency of observed unethical behavior to be curvilinear. In Model 2, the frequency of observed unethical behavior is positively related to the intention to whistleblow (β = 0.31; p < 0.01). Model 3 shows that the quadric term of the frequency of observed unethical behavior was significant (β = 0.91; p < 0.01). As shown in Fig. 3, the relationship between the frequency of observed unethical behavior and the intention to whistleblow is indeed a U-curve. That is, the relationship between the frequency of observed unethical behavior and the intention to whistleblow shows a negative or downward trend at lower frequencies of observed unethical behavior and a positive or upward trend at higher frequencies of observed unethical behavior. Therefore, Hypothesis 3 is supported.
Regarding Hypothesis 4a, we predicted that the relationship between the frequency of observed unethical behavior and the internal whistleblowing intention would be moderated by management support. Table 5 and Fig. 4 present the results. In all three models, the relationship between management support and internal whistleblowing is strong, positive, and significant. In Model 3, the relationship between the interaction term frequency of observed unethical behavior × management support and the intention to internally whistleblow was negative (β = − 0.10; p < 0.01), while that between the interaction term and frequency of observed unethical behavior squared was positive (β = 0.14; p < 0.01). For internal whistleblowing, high management support dampened the negative effect of the frequency of observed unethical behavior on internal whistleblowing (shown in Fig. 1) and even created a positive effect when the frequency levels of observed unethical behavior was medium or higher (Fig. 4, right-hand side of upper line). However, when employees believed that management support was low, the intention to internally whistleblow was negatively related to the frequency of observed unethical behavior (Fig. 4, lower line sloping downwards to the right). This leads to a situation that when the frequency of observed unethical behavior is low, the difference between low and high management support is smaller than when the frequency of observed unethical behavior is high. To conclude, this supports Hypothesis 4a. Regarding Hypothesis 4b, we expected the relationship between the frequency of observed unethical behavior and external whistleblowing to be also moderated by management support for whistleblowing. Table 6 and Fig. 5 show the results. The relationship between management support and external whistleblowing intention was low, negative, and only significant in Models 4 and 5 (but not in the full Model 6). In Model 6, the interaction term with the frequency of observed unethical behavior was significant and positive (β = 0.17; p < 0.01), while the interaction term with the squared frequency of observed unethical behavior was significant and negative (β = − 0.10; p < 0.01). This confirms Hypothesis 4b.
Regarding Hypothesis 4c, we expected the relationship between the frequency of observed unethical behavior and whistleblowing to be moderated by management support for whistleblowing. As shown in Table 7, management support was significantly and positively related to the intention to whistleblow in all three models. The interaction term of the frequency of observed unethical behavior with management support was also significant (β = 0.09 in Model 5, and β = 0.08 in Model 6, with p < 0.01), while the interaction term with the squared frequency of observed unethical behavior was not significant. Instead of the U-curve in Model 3 without the moderator (shown in Fig. 3), Fig. 6 shows that we now have a straight linear relationship. The line was steeper in the case of high than in low management support. This indicates that employees who believe that management will handle well their whistleblowing reports are increasingly more inclined to blow the whistle, whether internally and/or externally, than employees who believe that their management will be less supportive of their whistleblowing reports. Therefore, this supports Hypothesis 4c.

Summary and Implications
This study examined the relationship between the frequency of unethical behavior observed by employees in their organization and their intention to whistleblow. We found a curvilinear relationship (supporting Hypothesis 3) that is a combination of a diminishing negative relationship between the frequency of observed unethical behavior and the intention to internally whistleblow (supporting Hypothesis 1) and a linear positive relationship between the frequency of observed unethical behavior and the intention to externally whistleblow (supporting Hypothesis 2). We also found that the beliefs employees have about how supportive their management would be in dealing with their whistleblowing reports moderates  Fig. 3 Relationship between frequency of observed unethical behavior and intention to whistleblow the relationship between the frequency of employees' perception of unethical behavior and their intention to whistleblow in general, i.e. internally and/or externally (supporting Hypotheses 4a, 4b, and 4c). Therefore, Miceli et al. (2008) claim that research findings about whistleblowing are often counterintuitive is not true for the findings of our study. The findings of this study are consistent with the expectations we developed based on the Focus Theory of Normative Conduct. Whether and how employees form the intention to blow the whistle when they observe unethical behavior depends not only on what the injunctive norms are but also on what the descriptive norms are. The norms refer to both the evaluation of whether the observed behavior is unethical (descriptive) and, if so, whether employees have the responsibility to blow the whistle (injunctive). In this regard, this study presents a new area for the Focus Theory of Normative Conduct. This theory on how injunctive and descriptive norms influence people in complying with these norms can  also be used to examine how these norms influence people's responses to others who do not comply with the norms. Apart from results that support our hypotheses, the study also has other interesting secondary results. For instance, the intention to whistleblow internally is quite strong when the frequency of observed unethical behavior is low (as shown in Fig. 1); employees are more inclined towards internal instead of external whistleblowing but the difference becomes smaller when their frequency of observed unethical behavior increases (shown in Figs. 1 and 2); some employees had the intention to externally whistleblow even when the frequency of observed unethical behavior was low (shown in Fig. 2); some employees indicated that they had the intention to internally whistleblow even when management support was low (shown in Fig. 4); between high and low management support for whistleblowing, there was a substantial difference in the strength of the intentions of employees to internally whistleblow (shown in Fig. 4); only when there was a low frequency of observed unethical behavior did the employees indicate that they were more inclined to blow the whistle externally when they believed that management support was low than when they believed it was high (shown in Fig. 5); and some control variables were only significant (i.e., hierarchical level) for internal whistleblowing, and some control variables were only significant (i.e., age and gender) for external whistleblowing, while the size of the organization was significant for both (as shown in Tables 2  and 3).

Directions for Future Research
This study points to several possible research directions to build a richer model for a better understanding of the whistleblowing decision-making process.
One possible research is to examine whether the frequency of different types of unethical behavior has different relationships with the intentions of employees to whistleblow. The current study examined general unethical behavior and used an aggregated score for the frequency of observed unethical behavior. While we found a significant relationship between the frequency of observed unethical behavior in general and the intention to whistleblow, the relationship may be different for different kinds of unethical behavior. For example, do the frequency of observed unethical behavior towards the organization, such as stealing company assets, and the frequency of observed unethical behavior towards external stakeholders, such as engaging in false or deceptive sales and marketing practices, have different relationships with the intention of employees to whistleblow? We may find that a higher frequency of observed unethical behavior towards the organization leads to stronger intentions to whistleblow internally because the interests of the organization are at stake, while   6 Relationship between frequency of observed unethical behavior and whistleblowing moderated by management support a higher frequency of unethical behavior towards external stakeholders leads to stronger intentions to whistleblow externally because the interest of those stakeholders are at stake. Another interesting research direction concerns the relation between the frequency and moral intensity of unethical behavior. Jones' (1991) six characteristics of the moral intensity of ethical dilemmas for understanding ethical decision-making have been applied to unethical behavior to understand whistleblowing decision-making. One of the contributions of our study is the introduction of frequency of (unethical) behavior as another relevant characteristic. Having studied the frequency of unethical behavior as a separate characteristic does not imply that it is not related to or part of any of Jones' six characteristics. In this article, we already argued that a higher frequency of observed unethical behavior in an organization may make unethical behavior a more serious issue in that organization (this is the characteristic of the "magnitude of consequences") and more accepted (this is the characteristic of the "social consensus"). In this regard, future research can examine whether the frequency of unethical behavior is indeed related to these two characteristics or perhaps to the other characteristics as well (i.e., probability of effect, temporal immediacy, proximity, and concentration of effect) of the moral intensity of unethical behavior. A related question is whether the frequency of unethical behavior has also a direct influence on the intention to whistleblow or maybe only an indirect influence through Jones's six characteristics. Even if the results would show that the frequency of unethical behavior is not a separate characteristic but only part of the other six characteristics, the current study still makes this contribution: i.e., the concept of moral intensity should not be restricted to one single incident of unethical behavior, but it should be related to any and all unethical behavior observed by an employee.
Another possible research direction concerns the recipients of whistleblowing reports. Our study included four types of such recipients. It would be interesting to include more specific recipients of external whistleblowing reports, such as a regulator, professional associations, or the press, and other internal recipients, such as different layers of management and supporting functions as HR, Legal, and Security. By including more recipients, the explanatory power of the model increases. In this respect, the response scale used in this study for the intention to whistleblow could also be enriched. Because we asked the respondents their intentions to whistleblow to each type of recipient, we were not able to establish the sequence by which employees would approach the different recipients; this is important, for example, to understand whether employees will first whistleblow internally before they do so externally. This sequential process of reporting to different recipients is well examined by Vandekerckhove and Phillips (2019). When redoing the current research, respondents could also be asked about their intended sequence of whistleblowing recipients.
More moderating variables may also be included to explain the relationship between the frequency of observed unethical behavior and the intention to whistleblow. The current study found that management support for whistleblowing, operationalized as the belief of employees of how managers respond to reports of unethical behavior, had a significant moderating effect. It could be investigated whether other factors, for instance preventive ones like a corporate ethics program, are relevant for how employees respond to the frequency of observed unethical behavior.
An interesting research direction pertains to the use of time-frames, other than the twelve months used in the current study, for observed unethical behavior. Because unethical behavior is not observed very frequently-its mean in our study was 1.55 on a five-point scale for 37 items-extending the time frame to two or more years would increase the variance and provide more room to build rich models. A longer time frame could also take into account the experience employees have with whistleblowing better than a one-year time frame could. As the follow-up and completion of a whistleblowing report can take a long period (Vandekerckhove and Phillips 2019), employees will be able to evaluate their management's support for whistleblowing better over a longer period than over a one-year period. Furthermore, employees could use the period after completing their whistleblowing report to get an idea about management support for whistleblowing: if after reporting, the frequency of observed unethical behavior does not change or even increases, then this could undermine their belief in their management's support for whistleblowing; if the reported unethical behavior does not happen again over a long period, this could improve their belief in their management's support for whistleblowing.
Another research direction for building a richer model to help us understand better the whistleblowing decisionmaking process is to examine the relationship between the frequency of observed unethical behavior and the intention of employees to whistleblow in the opposite direction than the one taken by the current study. The current study examined the influence of the frequency of observed unethical behavior on the intention of employees to whistleblow but not on the whistleblowing behavior itself. We have good reasons for this, as discussed above in the section Whistleblowing and Unethical Behavior. At the same time, we should not expect that intentions to whistleblow always translate into actual whistleblowing behavior (cf. Mesmer-Magnus and Viswesvaran 2005; Park and Blenkinsopp 2009). In cases when the intention to whistleblow leads to whistleblowing behavior, an interesting research question is whether whistleblowing behavior also leads to less unethical behavior. This relationship could be explained by the idea that successful whistleblowing does not only lead to the termination of the reported unethical behavior but also to the confirmation or strengthening of the descriptive norm in the organization that unethical behavior is not allowed and even disapproved, which stimulates employees to not behave unethically. The same may be asked about the moderating variable in this study: whether management support for whistleblowing leads to less unethical behavior. This relationship could be explained by the idea that employees who consider behaving unethically are deterred by their belief that their management will not accept unethical behavior and will take adequate actions towards the perpetrator when they find out about these unethical behaviors. Employees may also expect that their management is more likely to be informed about unethical behavior in the organization because others in the organization who observe unethical behavior and believe that management will handle their whistleblowing reports well are more willing to report such behavior to management.
A final research direction to be mentioned here is the search for other theories that can enrich our understanding about the relationship between the frequency of unethical behavior and whistleblowing. This article used the Theory of Normative Conduct (Cialdini et al. 1991) to explain this relationship. Although this theory helped develop our hypotheses, other theories may add explanatory power. One possible theory is the moral approbation theory developed by Jones andVerstegen Ryan (1997, 1998). Jones and Verstegen Ryan argue that moral approbation, the desire of moral agents to be seen as moral by themselves or others, plays a critical role in moral decision-making. This moral approbation has two facets. One is a desired level of moral approbation, which consists of, for example, a desire for moral approval from the agent's referent group. Second is an anticipated level of moral approbation, which depends on the level of moral responsibility the agent anticipates will be attributed to him/ her by the agent's referent group. A desired level of moral approbation could be used to explain why an increase in the frequency of observed unethical behavior leads to weakening of employees' intention to blow the whistle internally. For instance, that the unethical behavior continues or is not corrected indicates that it is approbated. An anticipated level of moral approbation could be used to explain why employees are more intent to whistleblow externally when they observe unethical behavior more frequently. For instance, that the more frequently employees do not stop or correct the unethical behavior they observe, the more others outside the organization will see them as complicit, thus stimulating their sense of responsibility to stop and correct the behavior by reporting it to people outside their organization. Another potentially useful theory is the Corporate Ethical Virtues Theory (Kaptein 2008b) because it can help understand better how employees interpret unethical behavior in terms of the current ethical culture in their organization. The Neutralization Theory (Maruna and Copes 2005) or Moral Disengagement Theory (Bandura 1999) can explain how employees may increasingly rationalize or justify their inaction when they observe unethical behavior more frequently.

Managerial Implications
This study showed that the more frequently employees observe unethical behavior, the less they have the intention to whistleblow internally and the more they have the intention to whistleblow externally. These findings provide further reasons that strengthen the importance of management preventing unethical behavior (Paine 1994;Treviño and Weaver 2003). First, unethical behavior is not only undesirable in and of itself but it also has negative consequences. When employees' intention to whistleblow internally becomes weaker or disappears due to a higher frequency of observed unethical behavior, it becomes less likely that employees will actually inform management when they witness unethical behavior. It then also becomes less likely that management will be able to stop and correct unethical behavior, whereas this becomes even more urgent with higher frequency of observed unethical behavior. Second, a higher frequency of observed unethical behavior also means that employees' intention to whistleblow externally becomes stronger. Because unethical behavior would be reported not to management but to recipients outside of the organization, this would mean management losing control or charge of stopping and correcting unethical behavior. Quite possibly, as a result, external agencies would question and investigate management for its failure to adequately prevent and redress such unethical behavior.
This study also showed that employees' beliefs about how management will handle their whistleblowing reports influence whether they will blow the whistle when they frequently observe unethical behavior and whether they would do it internally or externally. Management support for whistleblowing was operationalized in terms of the expectations of employees about the extent to which their management would take appropriate action when it receives whistleblowing reports, would handle reports confidentially, protect whistleblowers from retaliation, discipline the perpetrator(s) fairly and consistently, and would ensure the usefulness and effectiveness of whistleblowing reports, say, by eliminating reported unethical behavior within a reasonable time frame. An important managerial implication of our study is that convincing employees that their management meets these expectations prevents the weakening of employees' internal whistleblowing intention when they increasingly observe unethical behavior in their organization. This is the good news. The bad news is that this particular intention becomes weaker if employees are not convinced. So hopefully managers will respond well to the findings of this study by responding appropriately to their employees' reports about unethical behavior in organization.
Funding The author works part-time as an equity partner at KPMG that funded the survey ran by the panel company.

Compliance with Ethical Standards
Consent to Participate Informed consent was obtained from all participants included in the study.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.