Journal of Business Ethics

, Volume 117, Issue 3, pp 615–634

Prediction of Whistleblowing or Non-reporting Observation: The Role of Personal and Situational Factors

Authors

    • Key Centre for Ethics, Law, Justice and GovernanceGriffith University
  • R. Wortley
    • UCL Jill Dando Institute of Security and Crime Science
Article

DOI: 10.1007/s10551-012-1548-3

Cite this article as:
Cassematis, P.G. & Wortley, R. J Bus Ethics (2013) 117: 615. doi:10.1007/s10551-012-1548-3

Abstract

This study examined whether it was possible to classify Australian public sector employees as either whistleblowers or non-reporting observers using personal and situational variables. The personal variables were demography (gender, public sector tenure, organisational tenure and age), work attitudes (job satisfaction, trust in management, whistleblowing propensity) and employee behaviour (organisational citizenship behaviour). The situational variables were perceived personal victimisation, fear of reprisals and perceived wrongdoing seriousness. These variables were used as predictors in a series of binary logistic regressions. It was possible to identify whistleblowers on the basis of individual initiative, whistleblowing propensity (individual and organisational), fear of reprisals, perceived wrongdoing seriousness and perceived personal victimisation. It was concluded that whistleblowers are not markedly dissimilar to non-reporting observers. Based on the two most influential variables (perceived personal victimisation and perceived wrongdoing seriousness), the average Australian public sector whistleblower is most likely to be an ordinary employee making a good faith attempt to stop what they perceived to be a serious wrongdoing that was initially identified through personal victimisation.

Keywords

Australian public sector organisationsNon-reporting observationPersonal victimisationWhistleblowerWhistleblowingWrongdoing seriousness

Whistleblowers are employees who attempt to stop the performance of organisational wrongdoing that may otherwise remain hidden from sight (Rehg et al. 2008; Rothschild 2008). There can be financial, environmental, social and health costs associated with undetected or unreported wrongdoing. Therefore, whistleblowing potentially constitutes a valuable service to an organisation and society in general (Miceli et al. 2009). It is also possible that the whistleblower is not the only employee aware of wrongdoing (Milliken and Morrison 2003). In this research, the intention was to identify predictors of whistleblowing using personal and situational variables that potentially separate whistleblowers from non-reporting observers.

Identifying predictors of whistleblowing is important. Attempts to de-legitimise whistleblower claims often involve use of labels with negative connotations implying the whistleblower has questionable motives (Ashforth and Anand 2003; Miceli et al. 2009; Tsahuridu and Vandekerckhove 2008). Demonstrating whistleblowers are not intrinsically different from non-whistleblowers, who makes it difficult to attribute the whistleblowing to idiosyncratic flaws or malignancy, thus removing justification for ignoring reports of wrongdoing. There is a reason to believe that whistleblowers are not markedly different from non-reporting colleagues. For example, Somers and Casal (1994) found whistleblowers had neither particularly high nor particularly low organisational commitment. There is also evidence that whistleblowers may be more desirable employees than most (Brewer and Selden 1998; Miceli et al. 1991a, b).

Another reason for differentiating whistleblowers from non-reporting observers is to assist organisations that use whistleblowers as an internal anti-corruption resource. If only some employees are willing to report wrongdoing then anti-corruption efforts are hampered as only a portion of those with the potential to provide intelligence are willing to do so. Unreported wrongdoing would continue to have an adverse effect on the organisation. Over time this effect has the potential to increase in severity, becoming more firmly embedded in the organisations repertoire of internally acceptable behaviour (Ashforth and Anand 2003).

Research Background

The data analysed in the current study were drawn from a large Australian whistle-blower research project entitled ‘Whistling While They Work: Enhancing the Theory and Practice of Internal Witness Management in Public Sector Organisations’ (WWTW) (see Brown 2008 for a comprehensive accounting of this project’s purpose, structure and quantitative results). The WWTW project was a collaborative effort between academics from five Australian universities and a steering committee from 14 Australian public sector organisations from four jurisdictions (Queensland, New South Wakes, Australian Capital Territory and Western Australia). The projects broad goals were to measure how much whistleblowing was occurring in Australian public sector organisations, to examine how whistleblowing was being managed and to understand the impact of whistleblowing on employees. Fulfilling this last broad goal involved collection and analysis of data that allowed the authors to examine whether whistleblowers could be distinguished from non-reporting observers on the basis of personal and situational variables.

The researchers conducting the WWTW project regarded whistleblowing as a prosocial behaviour (Brown and Donkin 2008) as conceptualised by the prosocial model of whistleblowing first proposed by Dozier and Miceli (1985) and subsequently revised (Miceli et al. 2001). According to this model whistleblowing is considered prosocial in that whistleblowers are attempting to prevent the performance of behaviour which is detrimental to people or organisations (Dozier and Miceli 1985). The whistleblower does not need to be purely altruistic, reporting can be through internal or external channels and the whistleblower does not have to play a specific role in the wrongdoing (for example the whistleblower does not have to have been complicit in the performance of the wrongdoing and may or may not be personally victimised). The prosocial model of whistleblowing proposes that both personal and situational factors influence the decision-making of a potential whistleblower (Dozier and Miceli 1985; Miceli and Near 2005).

The extent to which behaviour is the result of personal characteristics of the actor on the one hand, and the situational context in which the behaviour is performed on the other is one of the enduring debates in psychology. Commonly referred to as the cross-situational consistency debate, the issue was brought to prominence in Mischel’s (1968) classic Personality and Assessment. In the organisational context, this debate refers to the extent to which behaviours in the workplace derive from the characteristics of the employees or from the organisational environment to which the employees are exposed.

Support for the idea that both personal and situational variables should be considered together when trying to explain human behaviour in organisational contexts can be found in various streams of organisational research. These streams include studies on personality (Mischel 2004), on the various permutations of person-environment fit (Kristof-Brown et al. 2005), on ethical decision-making and behaviour within organisations (Jones and Kavanagh 1996; Trevino 1986), work attitude (Ambrose et al. 2008), on the exercising of organisational voice and on the ‘mum’ effect (Harlos 2010; LePine and Van Dyne 1998; Marler et al. 2012). The relevance of both personal and situational variables specifically to whistleblowing can also been observed in previous research (Keil et al. 2010; Mesmer-Magnus and Viswesvaran 2005; Miceli and Near 1988, 2005; Sims and Keenan 1998). Consistent with this previous research and the parent WWTW project, the current study explored the influence of personal and situational variables on the decision to whistleblow or not.

Accordingly in this study we have classified our variables of interest as either personal or situational. Personal variables refer to stable demographic characteristics, attitudes and behavioural patterns possessed by participants. Situational factors refer to the perceived nature of the specific act of wrongdoing that the participants observe. We acknowledge that our situational variables are not objective measures of the wrongdoing in question but come to us filtered through the perceptions of the observer. Nevertheless, they are situational in the sense that they are specific to the particular decision of whether to report or not report an observed wrongdoing.

The Dozier and Miceli (1985) article suggested a number of personal and situational variables that could be relevant to whistleblowing decision-making, but it did not specify a definitive set. Therefore, the authors chose personal and situational variables and derived hypotheses (introduced in the next section) from a review of existing literature. The next section will present our reasoning for including each predictor in the reported analyses. The research methodology, results and discussion of results will follow in turn.

Personal Predictors

Demography

Demographic variables have proven to be inconsistent differentiators between silent observers and whistleblowers (Vadera et al. 2009). Some studies, for example, Rothschild and Miethe (1999) found no compelling reason to believe a demographic profile could be developed that would differentiate whistleblowers from no-reporting observers. Other studies have, however, found evidence that demographic variables can differentiate between whistleblowers and silent observers (Miceli and Near 2005; Near and Miceli 1996; Mesmer-Magnus and Viswesvaran 2005).

The inconsistent findings regarding the association between demographic variables and whistleblowing was one reason for the inclusion of demographic predictors in the current study. The assumption was that the influence (or lack thereof) of demographic variables in whistleblowing contexts will become clearer as results from individual studies are added to the existing evidence base. The demographic predictors used in the current study (gender, age and tenure (public sector tenure and organisational tenure) were all identified by Vadera et al. (2009) as requiring further research due to inconsistency in differentiating whistleblowers from non-reporting observers. A second reason for including demographic predictors is that academic interest in possible relationships between demography and a range of organisational outcomes appears to be increasing over time (Joshi et al. 2011). A third reason is that identifying demographic variables differentiating whistleblowers from non-reporting observers means organisations could remove barriers that prevent specific demographic categories from reporting wrongdoing. For example, if non-reporting observers tend to be female then reasons for female reticence could be identified and corrective measures taken.

Gender

Some studies have found males are more likely to whistleblow than females (Miceli and Near 1988; Miceli et al. 1999; Near and Miceli 1996; Sims and Keenan 1998). It has been suggested that results such as these can be explained by males being less sensitive to the personal risks involved (Vadera et al. 2009). Liyanarachchi and Adler (2011), for example, reported that the higher likelihood of males reporting wrongdoing could be attributed to females being unwilling to face retaliation in some age groups. Kaplan et al. (2009) however, found females would not necessarily refrain from whistleblowing due to perceived personal cost of reporting wrongdoing. Females would, however, be more likely to use an anonymous reporting channel than men in the belief that personal costs would be lower.

Researchers expecting females to be more likely than males to report wrongdoing have attributed their expectation to females being more likely (than males) to feel a personal responsibility to whistleblow as a means of ending the wrongdoing (Vadera et al. 2009). There is little evidence to support the expectation that females are more concerned about unethical workplace behaviour than males (Kish-Gephart et al. 2010). A review by O’Fallon and Butterfield (2005), however, suggests that where gender differences in ethical reasoning are found it will be females who display more concern. Decker and Calo (2007) found evidence that females evaluated unethical executives less favourably than did males and were more favourable in their evaluation of a whistleblower. These researchers did not, however, assess whether this gender difference would translate into a personal intention to blow the whistle or actual whistleblowing.

Some whistleblowing studies have, however, found females are more likely to report wrongdoing than males. For example, Bjorkelo et al. (2010) found that while males were more likely than females to report wrongdoing two or more times, females were more likely than males to report wrongdoing once. Keil et al. (2010) found females were more likely than males to intend reporting wrongdoing. Keenan (2000) found females were more likely to whistleblow than males and the meta-analysis of Mesmer-Magnus and Viswesvaran (2005) found females were more likely than males to actually blow the whistle. The following hypothesis is offered in light of the Mesmer-Magnus and Viswesvaran (2005) meta-analysis.

Hypothesis 1a

Females will be more likely to be classified as whistleblowers than non-reporting observers.

Tenure

According to Vadera et al. (2009), the majority of research examining the association between tenure and whistleblowing has examined whether it influences the choice between internal or external reporting channels. The current research, however, is concerned with whether or not tenure (either organisational or public sector) differentiates whistleblowers from non-reporting observers as these relationships are comparatively underexplored. There are a number of reasons for assuming whistleblowers would have loner tenure than non-reporting observers.

Those with longer tenure could be expected to be more likely to report wrongdoing than those with shorter tenure due to a greater understanding of how the organisations internal control systems (formal and informal) could be used to report wrongdoing (Keenan 2000). Increased tenure could also allow accrual of a greater stock of idiosyncrasy credits (through displays of loyalty and competence over time) which could offer a degree of protection from retaliation to a longer tenured whistleblower (Miceli and Near 1988). It may be that tenure is positively associated with sources of power (for example high quality relationships with other powerful organisational members and rank) so those with longer tenure would be less likely to remain silent (Milliken et al. 2003). Alternatively, it may be that increased tenure decreases the likelihood of whistleblowing. For example, immersion in a professional or organisational milieu may lend credibility to behaviours that those with lesser tenure perceive to be wrongdoing (Ashforth and Anand 2003; Gino and Bazerman 2009).

The majority of studies have reported that organisational tenure does not differentiate whistleblowers from non-reporting observers (Kaptein 2011: Rothschild and Miethe 1999; Sims and Keenan 1998; Singer et al. 1998; Skivenes and Trygstad 2010). Where tenure has been found to distinguish whistleblowers from non-reporting observers those with longer tenure were more likely to report wrongdoing than those with less tenure. For example Stansbury and Victor (2009) created a composite age-tenure dummy variable and found as tenure and age increased so did the likelihood of whistleblowing. Miceli and Near (1988) also combined age and tenure into a composite ‘service’ variable as an indicator of idiosyncrasy credits. They found whistleblowers scored higher on service than silent observers. A review by Near and Miceli (1996) and the Mesmer-Magnus and Viswesvaran (2005) meta-analysis reported that organisational tenure positively predicted whistleblowing. No study examining the relationship between public sector tenure (as distinct from organisational tenure) and whistleblowing could be located. However, Keil et al. (2010) found organisational tenure did not predict whistleblowing intention but years of work experience (as a project manager) did.

It is acknowledged that much of the available research suggests tenure should not be expected to have a predictive relationship in differentiating whistleblowers from non-reporting observers. However, hypothesis 1b was formed on the basis of research demonstrating that when organisational tenure does influence whistleblower decision-making those with more tenure are more likely to blow the whistle than remain silent. Similarly, the authors also expect, on the basis of Keil et al. (2010) that public sector tenure will influence whistleblower decision-making in the same direction as organisational tenure.

Hypothesis 1b

Employees with longer public sector tenure will be more likely to be classified as whistle-blowers than non-reporting observers.

Hypothesis 1c

Employees with longer organisational tenure will be more likely to be classified as whistle-blowers than non-reporting observers.

Age

The relationship between age and whistleblowing is generally predicated in the assumption that age is a proxy for power (Vadera et al. 2009). Therefore, younger people could be expected to have less inclination to whistleblow. In contrast, age could be expected to positively predict whistleblowing if being older was associated with status or accumulation of idiosyncrasy credits (Miceli and Near 1988; Stansbury and Victor 2009).

Previous research has had an inconsistent relationship between age and whistleblowing. MacNab et al. (2007), MacNab and Worthley (2008) and Lee et al. (2004) all failed to find a statistically significant association between age and whistleblowing. In contrast Stansbury and Victor (2009) created a composite age-tenure dummy variable and found as tenure and age increased so did whistleblowing. Similarly, Miceli and Near (1988) combined age and tenure into a composite ‘service’ variable as an indicator of idiosyncrasy credit. They reported that increasing ‘service’ predicted whistleblowing. Age has also been associated with whistleblowing as a variable in its own right. For example Keenan (2000) found a significant positive correlation between age and whistleblowing. Liyanarachchi and Adler (2011) (using hypothetical scenarios as stimuli in a quasi experimental design) found a main effect of age on whistleblowing intention which suggested older accountants were more likely to report wrongdoing than younger accountants. The meta-analysis by Mesmer-Magnus and Viswesvaran (2005) found that older people were more likely to intend to report wrongdoing and a review by Near and Miceli (1996) found whistleblowers tend to be older employees. It should be acknowledged that intent to report wrongdoing does not necessarily translate into actual whistleblowing (Mesmer-Magnus and Viswesvaran 2005). Those studies measuring whistleblowing intention, however, report results similar to studies conducted in applied settings with actual whistleblowers (Keenan 2000: Miceli and Near 1988; Stansbury and Victor 2009) and the review by Near and Miceli (1996). Therefore, the following hypothesis was tested.

Hypothesis 1d

Older employees are more likely to be classified as whistleblowers than non-reporting observers.

Job Satisfaction

Job satisfaction is an attitude based in cognitive evaluation of the job and the job context (Weiss 2002). Job satisfaction was included because whistleblower credibility is sometimes attacked by implying that the whistleblower was an unreasonable, chronically-dissatisfied malcontent (Martin and Rifkin 2004; Miceli et al. 2009; Near and Miceli 1996; Rothschild 2008). If, in the current study, whistleblowers report lower levels of job satisfaction than non-reporting observers then managers may be correct to characterise whistleblowers as malcontents who have responded to a relatively minor wrongdoing more extremely than a ‘reasonable’ employee. Alternatively, if whistleblowers have higher levels of job satisfaction than non-reporting observers it would incorrect to characterise the average whistleblower as a dissatisfied malcontent.

It may be possible to support the view that whistleblowers are dissatisfied malcontents on the basis of meta-analytic evidence suggesting job satisfaction tends to be negatively correlated with dispositional variables such as neuroticism, negative affectivity and trait anxiety (Bruk-Lee et al. 2009). The strength of the employee job satisfaction relationship could therefore be a reflection of a dispositional bias influencing the interpretation of a workplace situation. For example, employees with more negative dispositions could feel less job satisfaction than those with more optimistic dispositions within the same work environment (Bruk-Lee et al. 2009). Similarly, Miceli and Near (2005) suggested job satisfaction should have an association with whistleblowing since job satisfaction fluctuates in response to a specific job situation, and that whistleblowers may feel less job satisfaction than non-reporting observers. For example, an observer of wrongdoing who possesses a more negative disposition could be more likely to feel less job satisfaction (on observing wrongdoing) than those with a more positive disposition and be more likely feel the wrongdoing should be reported. The available research, however, is not consistent with this view.

Sims and Keenan (1998) failed to find a relationship between job satisfaction and external whistleblowing. Other research, however, suggests job satisfaction would be positively related to whistleblowing. Job satisfaction has been found to be higher when supervisors endorse ethical work behaviour (Viswesvaran et al. 1998). Vakola and Bouradas (2005) found a negative relationship between job satisfaction and denial of employee voice. Mesmer-Magnus and Viswesvaran (2005) found a small positive association between job satisfaction and actual whistleblowing. Brewer and Selden (1998) reported whistleblowers to be high in job satisfaction and Miceli and Near (1988) found employees with more positive reactions to their work were more likely to be whistleblowers than silent observers. The cited research suggests the following hypothesis.

Hypothesis 2

Whistleblowers will have higher job satisfaction than non-reporting observers.

Trust in Management

Trust in management was included in the reported analyses because trust is the willingness to be vulnerable to the actions of another (Mayer et al. 1995) and whistleblowing can place an employee in a vulnerable position within the organisation (Rehg et al. 2008). Trust can be considered an attitude based in evaluation of social information pertaining to the probable risk of harm within a relationship (Kramer 1999; Schoorman et al. 2007). Factors that inform evaluations of trustworthiness include affect (Yang and Mossholder 2010), previous interpersonal interactions, shared social categorisation, organisational role, observing whether others trust the target, employee perceptions of an organisations general level of procedural justice (Burke et al. 2007; Choi 2008; Kramer 1999), relative status (Lount and Pettit 2012) and the perceived benevolence, domain-specific ability and integrity of the target (Mayer et al. 1995). Trust exists between and within different levels of an organisation’s authority structure (Burke et al. 2007; Schoorman et al. 2007). Therefore it is feasible (as in the current study) for managerial and non-managerial employees to self-report their level of trust in the management team.

Employee trust has been found to have attitudinal, behavioural and financial consequences for organisations (Colquitt et al. 2007; Dirks and Ferirn 2002; Schoorman et al. 2007). There is also a reason to believe that those with more trust in their management team will be more likely to report wrongdoing. Employees would be able to evaluate the trustworthiness of a management team by attending to whether members of the management team generally treat employees with benevolence and integrity and the level of domain-specific ability displayed by members of the management team (Mayer et al. 1995). On observing wrongdoing, an employee would attempt to evaluate the risk of whistleblowing with reference to their existing level of trust in the management team. Observers of wrongdoing who doubt managerial benevolence, integrity or ability to prevent the wrongdoing may decide personal risk is unacceptably high and so remain silent (Gundlach et al. 2003; Henik 2008; Mayer 1995; Schoorman et al. 2007). Previous research has shown that observers of wrongdoing who doubt a management team’s ability to stop the wrongdoing tend to remain silent rather than blow the whistle (Lewis 2011; Near et al. 2004). Binikos (2008) found silence was more likely when organisational trust was low and internal whistleblowing more likely when organisational trust was high. Keil et al. (2010) reported that the perceived benefit of whistleblowing (relative to cost) increased as trust in supervisors increased (trust in supervisors was the most influential single predictor of whistleblowing intention), which in turn was related to the increasing levels of whistleblowing intention.

There can also be a policy and procedural basis for trust in management. Seifert et al. (2010) found that the presence of organisationally just whistleblowing policies and procedures increased the likelihood of internal whistleblowing and organisational justice can be positively associated with trust in management (Burke et al. 2007). It may be optimistic, however, to rely on the presence of whistleblower policies and procedures when deciding whether or not to a whistleblower could trust management. For example, Lewis (2011) states ‘Organisations can encourage trust by introducing suitable whistleblowing policies and procedures, but these may be of limited effect if top management does not reinforce ethical behaviour’ (p. 80). The implication is that while policy and procedure can build trust, trust in human beings would moderate the influence of fair procedures on employee behaviour (De Cremer and Tyler 2007).

Conversely, it might be that whistleblowers have lower trust in management than non-reporting observers. For example the whistleblower may have detected wrongdoing being committed by a member (or members) of the management team. In this scenario an observer’s level of trust in management could be negatively influenced by factors such as the perceived level of interpersonal closeness between the ‘wrongdoer’ and other members of the management team, whether the management team benefits from continuation of the wrongdoing and to what extent the wrongdoing is institutionalised within the organisation. Rather than staying silent, or reporting to wrongdoing a distrusted management team, a whistleblower may resort to external reporting channels (Kaptein 2011; Lewis 2011; Miceli and Near 1985; Rehg et al. 2008). Therefore there would be a negative association between trust in management and whistleblowing.

The level of trust existing before wrongdoing is observed could, however, predispose an observer of wrongdoing to confer less personal responsibility for untrustworthy or ambiguous behaviour, even if the observer is personally involved in conflict with a manager (Dirks and Ferrin 2001; Korsgaard et al. 1996). Signs of unexpected untrustworthiness (on the part of a trusted member of a management team) could therefore motivate a search for evidence (by the potential whistleblower) that a wrongdoing has actually occurred and was performed by the suspect, rather than the potential whistleblower automatically assuming wrongdoing has occurred and assigning guilt (Kramer 1999). This search for evidence is characteristic of social information processing observable during the whistleblower decision-making process (Gundlach et al. 2003).

This information search could be thought of as monitoring, implying whistleblowers are mistrustful of the management team (Kramer 1999; Langfred 2004). However, monitoring contextualised by higher levels of pre-existing trust can be considered a form of prosocial assistance (DeJong and Elfring 2010). Whistleblowing can be considered a prosocial, if deviant act (Brief and Motowildo 1986; DeMaria 2008; Dozier and Miceli 1985; Spreitzer and Sonenshein 2004). Monitoring preceding whistleblowing could therefore be positively associated with trust in management (rather than being a sign of distrust which would imply a negative relationship between trust in the management team and choosing to report the wrongdoing) since most whistleblowers’ intention is to create beneficial organisational change (Rothschild 2008).

The research cited-above suggests that a whistleblower is likely to have a higher level of trust in management than would a non-reporting observer. Therefore, the following hypothesis is offered.

Hypothesis 3

Whistleblowers will have higher levels of trust in management than non-reporting observers.

Whistleblowing Propensity

Whistleblowing propensity indicates the degree to which whistleblowing is believed to be an appropriate action once an employee becomes aware of wrongdoing (Sims and Keenan 1999). While some research has used single factor measures of whistleblowing propensity (for example Keenan 2002; Liyanarachchi and Adler 2011; Liyanarachchi and Newdick 2009) other studies have determined that whistleblowing propensity has two distinct components. In the current study, the two factored approach to measuring whistleblowing propensity was used. The two factors are ‘individual propensity’ to whistleblow and ‘organisational propensity’ for whistleblowing (Keenan 2000, 2007; Miceli and Near 1985; Rothwell and Baldwin 2007; Sims and Keenan 1998; Tavakoli et al. 2003).

Individual propensity represents the strength of an individual’s belief that they have a duty to report wrongdoing, the belief that whistleblowing is in the best interests of the company and that whistleblowing should be encouraged (Tavakoli et al. 2003). Organisational propensity reflects the strength of an individual’s belief that an organisation values whistleblowing and facilitates the making of a report (Keenan 2000, 2002). The perceived level of organisational approval for whistleblowing can be inferred from the presence of a whistleblowing policy, providing information relevant to the internal report process or unambiguous organisational statements stating that whistleblowing is valued by the organisation (Keenan 2002).

The two factored approach was taken because each factor reflects distinct origins for the level of whistleblowing propensity (personal/internal and external/organisational) and therefore it would be possible to examine which origin was more influential in motivating whistleblowing and non-reporting observation. This was reasonable because even though the two factors can be positively correlated (Keenan 2007; Sims and Keenan 1998) (although Keenan (2000) reported a non-significant correlation), indicating a degree of mutual influence, they can have different relationships with other variables. For example, Keenan (2007) reported a non-significant correlation between individual propensity and fear of retaliation while organisational propensity was found to have statistically significant negative correlation with fear of retaliation. Similarly, Sims and Keenan (1998) reported that variables indicative of individual propensity to whistleblow predicted the use of external whistleblowing channels, whereas variables indicative of organisational propensity did not. Both types of propensity have been found to distinguish whistleblowers from non-reporting observers.

Zhang et al. (2009) found that the perceived ethicality of whistleblowing positively predicted the intention to whistleblow. Sims and Keenan (1998) and Keenan (2000) reported that the likelihood of whistleblowing increased as individual ideals favouring whistleblowing also increased. The Mesmer-Magnus and Viswesvaran (2005) meta-analysis found a positive association between intent to blow the whistle (but not actual whistleblowing) and individual approval for whistleblowing. Other research has found individual approval for whistleblowing distinguishes actual whistleblowers from inactive observers (Miceli and Near 1984, 1985). The following hypothesis is based on the cited research.

Hypothesis 4a

Whistleblowers will have a higher individual propensity to whistleblow than non-reporting observers.

Whistleblowing is more likely in organisations perceived to be more open to reports of wrongdoing (Miceli and Near 1988; Mesmer-Magnus and Viswesvaran 2005) and the presence of whistleblowing policies signals to employees that the organisation is open to reporting wrongdoing (Keenan 1990). In contrast, perceived organisational disapproval of reporting wrongdoing has been found to silence employees (Stansbury and Victor 2009; Vakola and Bouradas 2005). Experimental scenario based research (using a sample of internal auditors and management accountants) found the likelihood of whistleblowing was higher when participants perceived whistleblower management procedures to be fair rather than unfair (Seifert et al. 2010). Sims and Keenan (1998) reported the likelihood of whistleblowing was positively predicted by the perception that policies (formal and informal) supported whistleblowing. Similarly, Keenan (2000) reported that organisational propensity positively predicted likelihood of whistleblowing. Miceli and Near (1985) reported inactive observers could be differentiated from whistleblowers on the basis of not knowing organisational reporting channels (knowledge or reporting channels being a facet of organisational propensity). The following hypothesis is suggested by the cited research.

Hypothesis 4b

Whistleblowers will perceive a higher organisational propensity to whistleblow than non-reporting observers.

Organisational Citizenship Behaviour (OCB)

Organ (1997) defined OCB as ‘performance that supports the social and psychological environment in which task performance takes place’ (p. 95). OCB was included in the current study as a mean of examining whether whistleblowers were more prosocial than non-reporting observers. Research shows OCBs have a positive relationship with work performance and desirable attitudes such as job satisfaction, perceived justice climate and organisational commitment (Hoffman et al. 2007; Nielsen et al. 2009; Podaskoff et al. 2009). OCB can be positively associated with counter-productive behaviours if the OCB was made necessary for reasons given a negative valence by the employee (Spector and Fox 2010), but it is generally related to a decrease in the performance of counter-productive work behaviours (Dalal 2005).

Antecedents of OCB include having a proactive personality in interaction with high quality leader–member exchange (Li et al. 2010) and being prosocially motivated interacting with a desire to make a good impression (Grant and Mayer 2009). Proactivity and being prosocial have been positively associated with whistleblowing (Bjorkelo et al. 2010; Miceli and Near 2005). Whistleblowers tend to feel a sense of duty towards their employer, initially believing organisational authorities will correct the wrongdoing (Rothschild 2008). Voluntarily performed OCBs are prosocial in nature as are most whistleblower reports (Miceli and Near 2005). In addition, a positive association has been found between perceived organisational virtue and the performance of OCB (Rego et al. 2010). Turnipseed (2002) reported a positive association between employee ‘ethicality’ and social advocacy since an OCB and whistleblowing is more likely when an organisations is perceived to have a more ethical culture (Kaptein 2011; Rothwell and Baldwin 2007; Zhang et al. 2009). These studies suggest OCB should be higher amongst whistleblowers than non-reporting observers.

It may be possible for OCB to be negatively related to whistleblowing. For example, the affiliative aspect of OCB could predispose people to preserve peace and stability or to work without complaining rather than create conflict (Choi 2007; Van Dyne et al. 2003). However, the cited research more firmly suggests there will be a positive association between whistleblowing and OCB. OCB is measured in this study as interpersonal helping, individual initiative, personal industry and loyal boosterism (Moorman and Blakely 1995). Each facet is expected to have a positive association with whistleblowing (LePine et al. 2002; Hoffman et al. 2007).

Hypothesis 5a

Whistleblowers will have a higher level of interpersonal helping than non-reporting observers.

Hypothesis 5b

Whistleblowers have a higher level of individual initiative than non-reporting observers.

Hypothesis 5c

Whistleblowers will have a higher level of loyal boosterism than non-reporting observers.

Hypothesis 5d

Whistleblowers will have a higher level of personal industry than non-reporting observers.

Situational Predictors

Fear of Reprisal

Fear of reprisals was measured because there is ongoing debate about the influence of an employees fear of reprisals on the likelihood of whistleblowing. Retaliation comes in many forms including ad hominem attacks, increased monitoring of work performance, demotion or denial of promotion, social ostracism, referral to a mental health professional, being fired, counter accusations, and professional blacklisting (Martin and Rifkin 2004; McDonald and Ahern 2000; Rehg et al. 2008; Rothschild and Miethe 1999).

The percentage of whistleblowers suffering reprisals varies between studies. Rothschild and Miethe (1999) reported approximately 66 % of their internal whistleblower sample suffered some form of reprisal. Miceli et al. (1999) reported an increase in retaliation against whistleblowers from 1980 (16 %) to 1983 (21 %) to 1992 (33 %). In a recent survey of Norwegian public sector organisations 17 % of whistleblowers reported experiencing reprisals (Skivenes and Trygstad 2010). These figures suggest reprisals are possible but are not inevitable.

The fear of reprisals, even if the cited frequencies suggest this fear may be unnecessarily high, could still influence whether silence or whistleblowing follows wrongdoing detection. For example, Casal and Bogui (2008) found employees would rather quit than whistleblow if they expected retaliation but found a situation personally unacceptable. According to Rothschild (2008) the most frequently cited reason reported by non-reporting observers for their silence is fear of retaliation while whistleblowers underestimate the likelihood of retaliation. Other research, however, suggests reprisals do not deter whistleblowers.

Mesmer-Magnus and Viswesvaran (2005) reported threatened retaliation negatively correlates with intention to whistleblow, but does not prevent actual whistleblowing. Most whistleblowers, even knowing the costs they would face, appear willing to report wrongdoing again (Near and Miceli 1986; Rothschild 2008). Reprisals seem more likely to encourage external reports than discourage whistleblowing (Near and Miceli 1996; Rothschild and Miethe 1999; Rehg et al. 2008). It appears that once the decision to make a report is reached the likelihood of retaliation is not a deterrent.

Hypothesis 6

Whistleblowers will have a lower fear of reprisals than non-reporting observers.

Perceived Wrongdoing Seriousness

Perceived wrongdoing seriousness was measured because non-reporting observers could potentially be differentiated if the decision to report (or not) was a reflection of whether an observed activity was sufficiently serious to warrant reporting (Miceli and Near 2005). Almond (2009) suggests perceived wrongdoing seriousness is based on evaluations of harmfulness (a judgement of threat consequential to an action) and wrongfulness (a judgement of ‘moral gravity’ which incorporates a judgment of wrongdoer intent). Behaviour can be seen as wrongful but not harmful, harmful but not wrongful or both harmful and wrongful. Behaviours perceived to be both harmful and wrongful would be considered more serious than behaviours that are only harmful or only wrongful (Alter et al. 2007; Usoof-Thowfeek et al. 2011).

The perceived harmfulness and wrongfulness of a behaviour (the perception of wrongdoing seriousness) can be influenced by socially shared (Ashforth and Anand 2003; Escartín et al. 2009) and personally subjective standards of conduct (Miceli and Near 2005). Subjective perception of wrongdoing seriousness appears to be more influential in whistleblower decision-making than social influences. For example, social consensus regarding the ethicality of behaviour appears unrelated to the intention to whistleblow (Henik 2005; Singer et al. 1998). In contrast, Rothschild and Miethe (1999) reported 79 % of whistleblowers were moved to report by their personal convictions that the behaviour was wrong, harmful (implying a subjective perception of wrongdoing seriousness) or illegal. Taylor and Curtis (2010) reported that perceived seriousness of wrongdoing reflects a personal commitment to subjective moral judgements that increases both likelihood of whistleblowing and perseverance of whistleblowers. Note, in the current study participants were not asked to report the seriousness of the same type of wrongdoing. Therefore, we have measured situational appraisal of the seriousness of a wrongdoing a participant had observed in their workplace rather than an individual’s innate response to a specific wrongdoing type.

The authors are not suggesting perceived seriousness will always be positively predictive of whistleblowing. For example Casal and Bogui (2008) did not find an association between wrongdoing seriousness and whistleblowing. Rather, seriousness is one factor that is likely to be considered along with other situational factors when deciding whether to report or not. For example, a whistleblower typically would not make a report without balancing personal risk against the seriousness of the wrongdoing (Gundlach et al. 2003; Henik 2008; Keil et al. 2010). However, where perceived seriousness has been found to have an association with whistleblowing there has been a positive association (Brewer and Selden 1998; King 1997; Mesmer-Magnus and Viswesvaran 2005; Miceli and Near 1985; Robinson et al. 2012; Taylor and Curtis 2010). The cited research suggests the following hypothesis.

Hypothesis 7

Whistleblowers will perceive a higher level of wrongdoing seriousness than non-reporting observers.

Perceived Personal Victimisation

There are many ways to become aware of organisational wrongdoing. Research using ‘evidentiary quality’ as a predictor has included personal observation, finding written evidence and being able to produce corroborating witnesses (Near and Miceli 1986; Near et al. 2004). Another means by which to become aware of wrongdoing is to be personally targeted by wrongdoers. Perceived personal victimisation was examined since the possibility of a relationship between personal victimisation and whistleblowing has received almost no direct research attention.

A cost-benefits approach could be used to understand a relationship between whistleblowing and personal victimisation. For example, the choice to remain silent could incur more personal cost than whistleblowing if silence was thought likely to result in continual victimisation. Therefore, the benefit (stopping the victimisation) may outweigh anticipated costs (for example retaliation) that could be incurred through whistleblowing.

Indirect evidence suggests the benefit of whistleblowing outweighs the costs incurred if the employee is being personally harmed by the wrongdoing. For example, information relevant to personally experienced injustice receives more cognitive attention than information about injustice experienced by others (Ham and van den Bos 2008) and engagement in intentional deliberation about wrongdoing is positively associated with willingness to take action against wrongdoers (Usoof-Thowfeek et al. 2011). Organisational justice research indicates that personally experiencing interactional injustice is positively associated with experiencing anger and externalised blame attribution, which tends to motivate the aggrieved party to take defensive action against the source of injustice (Barclay et al. 2005).

Only one study could be located where a variable approximating personal victimisation (egoistic motive) was used to differentiate whistleblowers from non-reporting observers. Specifically, Miceli and Near (1985) reported that the personal impact of a wrongdoing motivated employees to blow the whistle. The following hypothesis is suggested.

Hypothesis 8

Whistleblowers will be more likely than non-reporting observers to claim personal victimisation.

In summary this research examined whether it is possible to differentiate whistleblowers from non-reporting observers using personal and situational perception variables. The personal predictor variables were demography (gender and tenure), individual work attitudes (job satisfaction, trust in management, whistleblowing propensity) and employee behaviour, specifically OCB (interpersonal helping, loyal boosterism, individual initiative, personal industry). The situational predictors were fear of reprisal, perceived wrongdoing seriousness and perceived personal victimisation.

Method

Participants

The participants were employees from 118 Australian public sector organisations who took part in the WWTW project. The organisations differed by size (small- under 150 employees; medium—up to 1,000 employees; large—in excess of 1,000 employees and very large—more than 5000 employees), location (Queensland, New South Wakes, Australian Capital Territory and Western Australia), level of government (local councils, state government and Commonwealth government) and area of responsibility (including health, education, policing, defence, utility management and policy administration). The sample size in the first analysis was n = 3232. There were more (n = 1692) females than males (n = 1541). Age ranged between 17 and 68 years (M = 42 years), public sector tenure ranged between 2 months and 53 years (M = 15 years) and organisational tenure ranged between 1 month and 45 years (M = 11 years). There were more non-reporting observers (n = 2243) than whistleblowers (n = 989).

Materials

The current data were drawn from the Workplace Experiences and Relationships questionnaire developed for the WWTW project using original and existing scales (Brown and Donkin 2008). This 44 question survey consisted of a blend of Likert scales, vignettes, free text response options and categorical responses to measure demographic, attitudinal, behavioural and situational characteristics as perceived by members of Australian public sector organisations. Safeguards were taken against common methods bias by including negatively worded items in some scales (Lindell and Whitney 2001), anonymous participation, using a different response formats and mixing items from different variables within individual questions (Podaskoff et al. 2003).

Measures

Dependent Variable

For the dependent variable participants were classified as either a whistleblower or non-reporting observer. Respondents were asked to nominate an incident (if they were aware of multiple activities they were directed to think about the one that they thought was most serious) for which they had direct evidence. Note that the direction to think about the most serious incident does not imply that the wrongdoing was extremely serious, only that it was more serious than other observed wrongdoings. The decision to report or not was established by the ‘yes’ or ‘no’ response to the following single item. ‘Did you formally report the activity to any individual or group? (Talking about it with your family or co-workers is not a formal report)’. This variable was coded ‘0’ for non-reporting observers and ‘1’ for whistleblowers.

Demographics

The demographic variables were gender (female = 0, male = 1), age (in years), organisational tenure (in years) and public sector tenure (in years).

Job Satisfaction

‘Job satisfaction’ was measured with a six item measure of general job satisfaction (α = 0.86) of proven validity and reliability (Agho et al. 1992). A representative item from the scale is ‘I feel fairly well satisfied with my job’. Job satisfaction, like all predictors (unless otherwise specified) was measured on a five point (1 = strongly disagree–5 = strongly agree) Likert index.

Trust in Management

Trust in management (α = 0.90) was measured using an adaptation of the seven item global ‘trust in employer’ scale of Robinson and Rousseau (1994). An exemplar item is ‘I am not sure I fully trust my management team’.

Whistleblowing Propensity

In the WWTW study, whistleblowing propensity was treated as a global measure using a 10 item adaptation of a scale developed by Keenan (2000). Previous research using an eight item version of this scale suggested the possibility of two sub-scales (individual approval and organisational approval of whistleblowing) (Keenan 2000; Tavakoli et al. 2003). Factor analysis of the current data (two factor solution specified using varimax rotation) revealed an acceptable two factor solution (KMO = 0.82, Bartlett’s test of sphericity significance < 0.001) that accounted for 61.03 % percent of the total variance. One factor (α = 0.84), which explained 30.70 % of the total variance, reflected the individual’s propensity for whistleblowing. This is represented by items such as ‘I personally approve of employees reporting illegal activities within the organisation’. The other factor (α = 0.83) which explained 30.3 % of the total variance, reflected an individual’s belief that the organisation approves of whistleblowing and has clearly articulated reporting procedures (organisational propensity). A representative item is ‘The organisation I work for encourages employees to report wrongdoing’.

OCB

OCB was measured using the four factor, 19 item scale from Moorman and Blakely (1995). The ‘interpersonal helping’ (willingness to assist co-workers as needed) dimension (α = 0.74) includes five items such as ‘I go out of my way to help co-workers with problems’. ‘Individual initiative’ (offering suggestions intended to increase work performance) was measured with five items (α = 0.80), including ‘For issues that may have serious consequences, I express opinions honestly, even when others disagree’.The ‘Personal industry’ factor (doing more than is formally required when performing tasks) was measured with four items (α = 0.67) including ‘I rarely miss work, even when I have a legitimate reason for doing so’. ‘Loyal boosterism’ (protection of and expressing pride in the organisation and its products) was measured with five items (α = 0.83) similar in theme to ‘I defend the organisation when other employees criticise it’.

Fear of Reprisal

This scale used four items from a 13 item ‘positive expectations of management’ scale developed for the WWTW research project. In the original WWTW study these four items were reverse scored and so indicated the absence of fear of reprisals. In the present analysis these items were scored positive to become indicators of fear of reprisals. This four items fear of reprisals scale possessed acceptable internal consistency (α = 0.81). A sample item is ‘I would feel intimidated if I pursued a concern about wrongdoing in this organisation’.

Perceived Wrongdoing Seriousness

Perceived wrongdoing seriousness was measured using a single item (‘How serious did you regard this activity to be?’) scored on a five point Likert index (1 = not at all serious–5 = extremely serious).

Personal Victimisation

Personal victimisation was derived from a categorical variable (Directed at me, I was invited to participate in it, I observed it, I came across direct evidence (e.g. documents), it was reported to me in my official capacity, other) originally to determine how the participant first became aware of the wrongdoing they regarded as most serious (note that this is not a measure of seriousness). Originally multiple responses could be circled. All participants that chose to circle multiple responses were excluded from the current analysis. A dummy coded personal victimisation variable was created by assigning all those whose response was ‘directed at me’ to a code of ‘1’. All other responses were assigned a code of ‘0’.

Procedure

A total of 23,000 questionnaires were posted to participating agencies. The number of questionnaires posted to each organisation varied depending on the size of the organisations. For example, in small agencies sufficient questionnaires were sent for 50–100 % of employees. In very large agencies sufficient questionnaires for 1–2 % were posted. Any current employee could participate. Participation was voluntary and individual participants were anonymous. In order to protect participant anonymity no signed consent forms were collected. Participation was interpreted to be the expression of informed consent. Completed surveys were returned to the researchers in reply-paid envelopes. We examined a subset of participants who met either our definition of a whistleblower or non-reporting observer.

The definition of a whistleblower employed in the present analyses was adapted from Miceli and Near (1984, p. 689). Specifically, a whistleblower was an employee of an Australian public sector organisation who had disclosed employers’ (or fellow employees’) illegal, immoral, or illegitimate practices that are under the control of their employers to a person or organisations who may be able to effect action. In line with the WWTW project definition the reporting of wrongdoing must not have been part of their normal work role. The wrongdoing must have occurred within the 2 years previous to research participation. Non-reporting observers were those who would have been classified as whistleblowers if they had decided to report wrongdoing instead of staying silent. All participants who did not observe wrongdoing were excluded from the current analyses on the assumption that they, unlike whistleblowers or non-reporting observers, were not whistleblowers simply because they had not observed any wrongdoing. As the definition of a whistleblower used by the researchers specifies that reporting should not be part of their normal work role all those whose normal responsibilities required whistleblowing were also excluded from the current analyses.

Results

Binary logistic regression was used to ascertain which predictors were associated with whistleblowing. No hypotheses specified the relative importance or order of entry of any predictor so all were entered as a single block (Tabachnik and Fidel 2007). An initial analysis was performed which included all hypothesised predictors. A second analysis included only the statistically significant predictors from the first analysis.

Descriptive Statistics

Table 1 includes means, standard deviations and intercorrelations for the full set of predictors. There is a general tendency for equivalence between whistleblowers and non-reporters. Neither group appears to be particularly different demographically, or to have markedly different attitudes. Both groups report similar levels of OCB. All participants have favourable attitudes towards whistleblowing while organisations are thought to be neutral towards whistleblowing. Whistleblowers appear more likely to believe they were personally victimised and are closer to seeing the focal wrongdoing as ‘very serious’. Non-reporters are more neutral about wrongdoing seriousness.
Table 1

Means, standard deviations and intercorrelation between predictor variables and whistleblower versus non-reporting observer

Variable

M

SD

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

1. Gender

0.48

0.50

                

2. Age

41.97

10.48

0.17**

               

3. Tenure with organisation

10.06

9.05

0.23**

0.47**

              

4. Tenure with public sector

14.60

10.50

0.25**

0.65***

0.72**

             

5. Trust in management

3.21

0.83

−0.04

−0.10**

−0.08**

−0.08**

            

6. Job satisfaction

3.39

0.75

0.00

0.09**

0.08**

0.07**

0.38**

           

7. Interpersonal helping

3.94

0.47

−0.08**

0.05**

0.04*

0.02

0.06**

0.23**

          

8. Individual initiative

3.72

0.55

0.02

0.08**

0.08**

0.08**

0.06**

0.18**

0.43**

         

9. Personal industry

3.81

0.57

−0.10**

0.10**

0.04*

0.03

0.04*

0.22**

0.38**

0.31**

        

10. Loyal boosterism

3.48

0.68

−0.07

0.02

0.02

0.00

0.32**

0.45**

0.32**

0.29**

0.30**

       

11. Individual approval

4.07

0.56

−0.07**

0.05**

−0.01

0.00

0.10**

0.10**

0.21**

0.26**

0.20**

0.20**

      

12. Org. approval

3.27

0.77

0.07**

0.10**

0.09**

0.12**

0.29**

0.23**

0.12**

0.19**

0.10**

0.24**

0.24**

     

13. Fear reprisals

2.97

0.74

0.04*

0.08**

0.04*

0.05**

−0.57**

−0.27**

−0.03

−0.08**

−0.02

−0.26**

−0.18**

−0.39**

    

14. Personal victimisation

0.18

0.38

−0.08**

0.03

0.02

0.02

−0.16**

−0.10**

0.07**

0.00

0.01

−0.05**

0.03

−0.01

0.15**

   

15. Wrongdoing seriousness

3.55

0.93

−0.06**

0.03

−0.01

0.00

−0.24**

−0.09**

0.11**

0.10**

0.11**

−0.03

0.10**

−0.04*

0.23**

0.21**

  

16. Whistleblower/non-reporter

0.31

0.46

−0.05**

0.06**

0.02

0.04*

−0.05**

0.00

0.08**

0.11**

0.04*

0.04*

0.14**

0.11**

−0.01

0.26**

0.29**

 

Gender coded 0 = female, 1 = male. Whistleblower/non-reporter coded 0 = non-reporting observer, 1 = whistleblower. n = 3232

p < 0.05, ** p < 0.001

First Logistic Regression

The full model was statistically significant [χ2 (15, 3232) = 519.99, p < 0.001] with correct classification of 73.2 % of the sample. Non-reporting observers were correctly classified more frequently (90.8 %) than whistleblowers (33.4 %). The value of Nagelkerke R2 indicates that 21 % of the variance in the decision to report wrongdoing was explained by the model. Table 2 contains the regression coefficients, odds ratios and 95 % confidence intervals for each predictor.
Table 2

First logistic regression predicting whistleblowing or non-reporting observation

Predictor

β

SE

Odds ratio

Gender

−0.14

0.09

0.87

Age

0.01

0.01

1.01

Tenure with organisation

0.00

0.01

1.00

Tenure with public sector

0.01

0.01

1.01

Trust in management

−0.08

0.07

0.92

Job satisfaction

0.03

0.07

1.03

Interpersonal helping

−0.02

0.11

0.98

Individual initiative

0.26*

0.09

1.29

Personal industry

−0.14

0.08

0.87

Loyal boosterism

0.01

0.07

1.01

Individual approval

0.29**

0.08

1.34

Org. approval

0.22**

0.06

1.25

Fear reprisals

−0.25**

0.07

0.78

Personal victimisation

1.19**

0.10

3.28

Wrongdoing seriousness

0.69**

0.05

1.99

Gender coded 0 = female, 1 = male. Whistleblower/non-reporter coded 0 = non-reporting observer, 1 = whistleblower

p < 0.05, ** p < 0.001

No demographic variables emerged as significant predictors and therefore there was no support for hypotheses 1a, 1b or 1c. Hypothesis 2 predicted whistleblowers could be distinguished from non-reporting observers by reporting more job satisfaction. This hypothesis was not supported. Hypothesis 3 proposed whistleblowers would be distinguishable from non-reporting observers on the basis of higher trust in management than non-reporters. This was not supported. Hypothesis 4a (individual propensity) and hypothesis 4b (organisational propensity) predicted whistleblowers could be distinguished from non-reporting observers by virtue of a higher whistleblowing propensity. Both hypotheses were confirmed. Individual propensity was slightly more influential [Exp(B) = 1.34, p < 0.001] than organisational propensity [Exp(B) = 1.25, p < 0.001].

Whistleblowers were expected to be distinguishable from non-reporting observers on the basis of higher interpersonal helping (hypothesis 5a), individual initiative (hypothesis 5b), loyal boosterism (hypothesis 5c) and personal industry (hypothesis 5d). Only hypothesis 5b was confirmed [Exp(B) = 1.30, p < 0.01]. Hypothesis 6 was confirmed as whistleblowers could be distinguished from non-reporting observers on the basis of having slightly less fear of reprisals [Exp(B) = 0.78, p < 0.001]. Hypothesis 7 was confirmed as whistleblowers believed the observed wrongdoing was more serious than wrongdoing observed by non-reporting observers [Exp(B) = 2.00, p < 0.001]. Hypothesis 8a was confirmed with whistleblowers more likely [Exp(B) = 3.28, p < 0.001] to report being personally victimised than non-reporting observers.

Second Logistic Regression

The second logistic regression included the significant predictors from the first analysis. Note that use of the ‘exclude’ missing responses sub-command (SPSS version 17) was specified as the means of handling missing data. This means the sample size was larger (n = 3357) than the first analysis as fewer variables meant there was less missing data.

The reduced model was statistically significant [χ2 (6, 3357) = 527.68, p < 0.001, Nagelkerke R2 = 0.21]. The correct classification rate was nearly identical to that achieved by the full model (overall correct = 73.3 %, non-reporting observer = 90.9 %, whistleblowers = 33.4 %). As can be seen in Table 3 all predictors retained statistical significance. Personal victimisation was the most influential variable [Exp(B) = 3.37, p < 0.001], wrongdoing seriousness was next most important [Exp(B) = 1.98, p < 0.001], followed by individual whistleblower propensity [Exp(B) = 1.33, p < 0.001], organisational propensity [Exp(B) = 1.27, p < 0.001], individual initiative [Exp(B) = 1.25, p < 0.01] and fear of reprisals [Exp(B) = 0.82, p < 0.001]. All predictor variables were associated in the hypothesised direction with the outcome.
Table 3

Second logistic regression predicting whistleblowing or non-reporting observation

Predictor

β

SE

Odds ratio

Individual initiative

0.23**

0.08

1.25

Individual approval

0.29**

0.08

1.33

Org. approval

0.24**

0.06

1.27

Fear reprisals

−0.20**

0.06

0.82

Personal victimisation

1.22**

0.10

3.37

Wrongdoing seriousness

0.69**

0.05

1.98

Gender coded 0 = female, 1 = male. Whistleblower/non-reporter coded 0 = non-reporting observer, 1 = whistleblower

p < 0.05, ** p < 0.001

Discussion

The present study demonstrates whistleblowers were very similar to non-reporting observers in terms of their personal characteristics. Whistleblowers did not appear to fall into a specific demographic category (gender, age or tenure) or be particularly satisfied or dissatisfied with their job. They do not seem to be particularly trusting or distrusting of management. Nor were they, with the exception of individual initiative, particularly noteworthy for their OCB (or lack thereof). Consistent with previous research situational variables (perceived personal victimisation and perceived wrongdoing seriousness) were more influential than any of the individual characteristics that were measured (Miceli and Near 2005; Vadera et al. 2009). Variables distinguishing between whistleblowers and non-reporting observers are discussed below.

Personal Predictors

As found by Miceli and Near (1985), whistleblowers reported higher individual propensity and organisational propensity than non-reporting observers. Individual propensity had a slightly higher association with whistleblowing than did organisational propensity (see Table 3). This pattern has been found in a number of studies across national cultures (Keenan 2002; Keenan 2007; Sims and Keenan 1999; Tavakoli et al. 2003). The current study, when considered together with the cited previous research, suggests individuals are prepared to act in congruence with their individual attitude towards whistleblowing.

Items in the organisational propensity scale refer to the provision of information relevant to the operation of, and access to, an internal whistleblowing process in addition to statements indicating whistleblowing is valued by the organisation (Keenan 2002). Social norms can be developed from attending to organisational policy and communications (Moss and Wilson 2010). The attitude-behaviour relationship is stronger when both are consistent with contextual norms (Smith et al. 2007). The implication is whistleblowers believe reporting wrongdoing was more normative within the organisation and more likely to be appreciated than did non-reporting observers.

As reported above, individual propensity was found to have a slightly higher association with categorisation as a whistleblower than organisational propensity. The possibility a non-hypothesised interaction between individual and organisational propensity1 would influence participant differentiation was examined with an additional analysis (identical to the second logistic regression except for the inclusion of this interaction term). The interaction between individual propensity and organisational propensity proved to be statistically non-significant [Exp(B) = 1.05, p = 58], implying individual propensity and organisational propensity make largely independent positive contributions to whistleblower decision-making with individual propensity being slightly more influential. This result might not be surprising considering the following.

Individual propensity reflects a positive evaluation of whistleblowing, personal values (reflecting belief about what it is right to do) influence situational decision- making and behaviour in organisations in a manner consistent with the individuals values (Meglino and Ravlin 1998). A number of studies suggest whistleblowers often report wrongdoing because doing so is a means of preventing activities that conflict with values held by the whistleblower (Brewer and Selden 1998; Liyanarachchi and Newdick 2009; Miceli et al. 1991a, b; Rothschild 2008; Rothschild and Miethe 1999; Singer et al. 1998; Taylor and Curtis 2010; Tsahuridu and Vandekerckhove 2008). Whistleblowing, albeit less likely than if the culture is supportive of whistleblowing (Miceli and Near 2005; Zhang et al. 2009) (and most likely to an external agency) can occur even when whistleblowing counter-cultural by organisational standards (Kaptein 2011; Miceli and Near 1985; Rehg et al. 2008).

The only significant OCB predictor was individual initiative. Proactive people are willing to act in ways that will bring about behavioural change in others (Grant and Ashford 2008) and whistleblowing is intended to create change (Miceli and Near 2005). This suggests individual initiative motivates whistleblowers to report wrongdoing because individual initiative is positively associated with proactive traits (Grant and Ashford 2008) and proactive traits predict whistleblowing (Bjorkelo et al. 2010).

Situational Predictors

Fear of reprisals (shown on Table 3 to be the least influential statistically significant predictor) was more of a concern for non-reporting observers than of whistleblowers. This result is consistent with the meta-analysis of Mesmer-Magnus and Viswesvaran (2005) who reported intent to whistleblow is reduced by fear of reprisals while actual whistleblowers are relatively unconcerned. One possible reason for the failure of reprisals to prevent whistleblowing is the decision-making process incorporates an assessment of the risks relative to costs of reporting (Keil et al. 2010). Once the likelihood of retaliation has been accepted as part of the cost-benefits assessment, and subjectively judged to be less important than expected benefits (Gundlach et al. 2003), a report will be made even though reprisals are possible. This reasoning is consistent with research finding whistleblowers have high self-efficacy (MacNab and Worthley 2008) as self-efficacy instils confidence in their self-perceived ability to cope with a negative organisational response (Webster et al. 2010). This is not to say that a whistleblower accurately estimates the level of reprisal risk. Rather, once the decision to report wrongdoing is made the tendency is to persist even when an unexpectedly hostile response occurs (Rothschild and Miethe 1999).

Whistleblowers perceived the observed wrongdoing as more serious than did non-reporting observers. Wrongdoing seriousness has previously been found to have a positive association with both whistleblowing intent (Mesmer-Magnus and Viswesvaran 2005) and actual whistleblowing (Brewer and Selden 1998; King 1997; Miceli and Near 1985). Previous research suggests the current results can be explained with reference to personal principles.

Employees who evaluate observed workplace behaviours with reference to strongly held principles are more likely to become whistleblowers than employees who do not share the same principles or do not hold as strongly to these principles (Henik 2008; Rothschild and Miethe 1999; Taylor and Curtis 2010). Therefore it is possible non-reporting observers, even if they observed a similar wrongdoing, would not share the whistleblowers subjective belief that a wrongdoing was serious enough to report (Miceli and Near 2002, 2005). However, in the whistleblower’s mind, consequences of inaction (such as continuation of the wrongdoing) are sufficiently problematic to create an intense internal conflict which the whistleblower attempts to resolve by reporting the wrongdoing (Henik 2008).

Being personally affected by wrongdoing has previously distinguished whistleblowers from non-reporting observers (Miceli and Near 1985). In the current study, perceived personal victimisation was the most influential predictor (see Table 3). Categorisation as a whistleblower was slightly over three times more likely when the participant believed they were personally victimised. The results could reasonably be explained in terms of personal relevance decreasing the perceived cost of whistleblowing relative to risk incurred by whistleblowing.

Indirect exposure to wrongdoing, for example by witnessing harm to a co-worker, is less personally relevant than direct victimisation. Being directly harmed (or threatened with direct harm) would increase the personal relevance of the wrongdoing to the whistleblower (Giacalone and Promislo 2010; Ham and van den Bos 2008) and presumably the perceived cost of allowing the victimisation to continue. Personal relevance is likely to be positively related to the amount of perceived un-ethicality of the wrongdoing (Giacalone and Promislo 2010; Ham and van den Bos 2008). Personal relevance motivates consideration of voice behaviour (Chiaburu et al. 2008), a behaviour similar to whistleblowing (Miceli and Near 2005) and whistleblowing is more likely if the perceived benefits of whistleblowing exceed the risks (Gundlach et al. 2003; Henik 2008).

Considering that the two most influential predictors were situational (perceived personal victimisation and perceived wrongdoing seriousness) the current study suggests a typical could be described as follows. Whistleblowers are normal employees who find themselves in the wrong place at the wrong time (Miceli and Near 2005) and make a good faith attempt to help themself and their organisation by reporting a serious wrongdoing (Rothschild and Miethe 1999).

Limitations

This was a cross-sectional study, and therefore, although inferences can be made, it is inappropriate to state firm conclusions about causality. We do not know if pre-report responses of whistleblowers would be different from the supplied post-report responses. For example trust in management may have been higher pre-report than was reported at the time of the study (Robinson 1996). Longitudinal research would provide better evidence of causality than has been provided by this or previous whistleblower research (Miceli and Near 2005). In a longitudinal study data could be collected before any wrongdoing is observed or reported. Participants could be measured periodically. Those who do not see wrongdoing could serve as a control group. Changes in measured variables following observation of wrongdoing could be more confidently attributed to awareness of the wrongdoing. Differences found between whistleblowers, non-reporting observers and non-observers would indicate what influences in situ whistleblowing decisions.

There are, however, practical difficulties that make longitudinal whistleblower studies particularly difficult. For example, whistleblowing cannot be guaranteed to occur during the study period, the amount of time, money and goodwill which would have to be generated and maintained for the length of the study could be prohibitive and there is the issue of participant attrition (Miceli and Near 2005). It is unlikely that a longitudinal study could maintain a sample as large and representative of the population as was possible in the current research. The impracticality of a longitudinal program of field research is why some researchers measure intended rather than actual whistleblowing (Tavakoli et al. 2003; Sims and Keenan 1999). Unfortunately, intended whistleblowing and actual whistleblowing cannot be assumed to have equivalent relationships with predictors (Mesmer-Magnus and Viswesvaran 2005). By a large cross-jurisdictional sample of actual whistleblowers this study has removed this known source of error (the difference between intent and actual behaviour).

The use of a questionnaire as the sole means of gathering data raises the issue of common methods bias. However, one of the strengths of this study was the inclusion of a number of procedural controls against common methods bias during the design phase (Podaskoff et al. 2003) of the study which minimise the influence of common methods bias. Negatively worded items were used in some scales (Lindell and Whitney 2001), participation was anonymous (Podaskoff et al. 2003), the analysed variables were part of a much longer survey which included different response formats and items relating to different variables were intermixed in the questionnaire (Podaskoff et al. 2003). Further, Doty and Glick (1998), on the basis of structural equation modelling and meta-analysis, concluded that final interpretation is not substantially influenced by common methods bias. The cited research suggests common methods variance had minimal influence, if any, on the interpretation of the present results.

Finally, this study was not grounded in an existing whistleblower theory which may be considered a limitation (DiMagigo 1995; Sutton and Staw 1995). It is suggested, however, that approach taken in the current research was a strength rather than a limitation. This study took conceptual guidance from the prosocial model of whistleblowing (Dozier and Miceli 1985) and the macro-theoretical principle of considering the person located within a situation. The authors contend that this was temporally appropriate given the developing state of whistleblower research (Miceli and Near 2005; Weick 1995). A sign that whistleblower researchers are engaged in an ongoing process of developing the basis for a generally accepted theoretical framework explaining whistleblower behaviour is the adaptation of existing theoretical frameworks from other areas of research for explanation of whistleblowing (Weick 1995). For example Miethe and Rothschild (1994) suggested using micro-level theories of deviance (social learning theory, rational choice theory and social bond theory). Gundlach et al. (2003) developed a social information processing model of whistleblower decision-making model (by integrating prior whistleblower conceptual perspectives with research into emotional responding, attribution and impression management). Henik (2008) proposed a whistleblowing model which added the cognitive appraisal approach to emotions and the social-functional value pluralism model to the social information processing model and those that existed before Gundlach et al. (2003). There is scope to integrate the present results with future theory building research. For example, the current data suggest a theory of whistleblower decision-making needs to go beyond stating ‘a wrongdoing is observed’ to explicitly examining how the situation surrounding wrongdoing detection (for example by being personally victimised) influences variables proposed by Gundlach et al. (2003) or Henik (2008). Therefore, the current study makes an important contribution to the evolution of a generally accepted theory of whistleblowing.

Practical Implications and Future Research

The most general implication that can be drawn from this study is that whistleblowers are notable for not being particularly notable. Whistleblowers did not differ from non-reporting observers on the basis of tenure. They were not particularly good or bad organisational citizens (overall), neither satisfied or dissatisfied with their jobs, nor particularly trusting nor distrustful or management. The most influential variables (perceived personal victimisation and perceived wrongdoing seriousness) reflected the situational appraisal of the event more than characteristics of an individual. This finding has a number of practical implications as ad hominem reprisals are routinely experienced by whistleblowers (Rothschild 2008).

First, reprisals are more likely to encourage the use of external reporting channels than whistleblower desistance (Rehg et al. 2008; Rothschild 2008). External reporting can be assumed to be less desirable (from the perspective of the organisation, at least) than keeping internal problems in house. Second, personal attacks may undermine an organisation’s defensive capacity by creating a perception of the whistleblower as a courageous underdog (Martin and Rifkin 2004). Third, there seems to be no justification for forcing whistleblowers into psychiatric treatment, denial of promotion or extra monitoring of work performance (McDonald and Ahern 2000; Smith and Brown 2008; Rothschild and Miethe 1999). Furthermore, all complaints need to be taken seriously and investigated objectively (Martin and Rifkin 2004). Doing so makes the organisation’s response to whistleblowing appear procedurally just which increases the likelihood of internal whistleblowing (Vadera et al. 2009) and compliance with decisions (Murphy and Tyler 2008).

A positive implication that can be drawn from the current results is that the average whistleblower in an Australian public sector organisation can expect to be treated fairly by the organisation. There was no sign of discrimination based on demographic category. In addition, previous organisational research suggests that if severe mistreatment was the norm we could expect whistleblower categorisation to be predicted by distrust of management or job dissatisfaction (Bowling and Beehr 2006; Connell et al. 2003; Kramer 1999; Rhoades and Eisenberger 2002). Since these variables were not significant predictors of participant classification it can be inferred that the average whistleblower was not in an abusive post-report work situation. Finding less fear of reprisals amongst whistleblowers than non-reporting observers would also fit with this observation.

Future research could delve further into why personal victimisation predicts whistleblowing. The current results suggest participants who became aware of wrongdoing through personal victimisation were particularly likely to become whistleblowers. Personal relevance provides a reasonable explanation. However, other unmeasured variables potentially associated with personal victimisation or wrongdoing seriousness may also influence the decision to report or remain silent. For example, research has identified wrongdoing type as a situational variable that influences whether employees report wrongdoing or remain silent (Near et al. 2004; Vadera et al. 2009). In the current study wrongdoing type was not included in the analyses. Therefore, the current data cannot provide information about the influence wrongdoing type might have on participant decision-making, whether wrongdoing type would moderate perceived seriousness, or whether personal victimisation would interact with type of wrongdoing and perceived seriousness.

Similarly, the perceived ethicality of the wrongdoing was not measured. Ethicality is used here to refer to a judgment about correctness of a behaviour or process with reference to contextually normative values (Trevino et al. 2006). This is distinct from ‘seriousness’ which may have been interpreted by current participants as referring to harmfulness and wrongfulness of consequences (Usoof-Thowfeek et al. 2011). It should be noted that Mesmer-Magnus and Viswesvaran (2005) found ethical judgment to be associated with whistleblowing intent rather than actual whistleblowing. However, a wrongdoing could be perceived as less ethical if the whistleblower believed they were personally victimised (Giacalone and Promislo 2010) and whistleblowers appear to be motivated by judgements about the ethicality of what they observe (Rothschild and Miethe 1999). Potentially, perceptions of personal victimisation, wrongdoing seriousness and ethicality could be significant positive predictors of deciding to report wrongdoing. Future research is necessary to explore whether these variables influence decision-making via main effects or interactions.

Conclusions

This study examined whether or not personal and situational variables could be used to predict classification as a whistleblower or non-reporting observer within the Australian public sector. The current results suggest any employee could be a whistleblower if they find themself in the wrong place at the wrong time. The two most influential variables predicting whistleblower classification were more reflective of the perceived context around the reported wrongdoing (personal victimisation and wrongdoing seriousness) than personal characteristics. Some individual level variables, however did predict participant classification. These were whistleblowing propensity and individual initiative. Results suggest the typical Australian Public sector whistleblower is a normal employee who discovers wrongdoing through personal victimisation and takes the proactive step of whistleblowing in an effort to stop a serious wrongdoing. Fear of reprisals is a non-issue once the decision to whistleblow is made, but does contribute to the decision of non-reporting observes to remain silent.

Acknowledgments

The authors thank an anonymous reviewer for suggesting that organisational propensity and individual propensity might influence each other and the decision-making of observers of wrongdoers. This research was made possible by funding from the Australian Research Council (ARC Linkage Project LP0560303) and the various industry partners to this project, as listed on the project website: www.griffith.edu.au/whistleblowing. The authors thank their colleagues on the Whistling While They Work Project Team as well as these industry partners and the participants for their assistance with this research. The findings and views expressed in this article are those of the authors and do not necessarily represent the views of the Australian Research Council or the project industry partners.

Copyright information

© Springer Science+Business Media Dordrecht 2012