Introduction

Psychopathy is characterized by a constellation of personality traits and behaviors of which society typically disapproves (Hare 1993). Individuals with high levels of psychopathy are superficially charming and manipulative. They are often devoid of deep social emotions, especially guilt, empathy, and love (Cleckley 1941, 1988; McCord and McCord 1964) and, according to some influential conceptualizations, characterized by poor impulse control and a proclivity towards aggression (Hare 1991/2003; Hare and Neumann 2009). Nevertheless, the role of impulsivity and antisocial behavior within the nomological network of psychopathy is controversial, as some authors contend that poor impulse control is not an essential feature of psychopathy (see Levenson 1993; Poythress and Hall 2011). Similarly, although psychopathy is linked to a heightened risk for antisocial and criminal behavior, it is unclear whether such behavior is inherent to psychopathy per se or is instead merely a downstream consequence of psychopathy (Cooke et al. 2004; Lilienfeld 1994; Skeem and Cooke 2010).

In recent years, the implications of psychopathy for the workplace have received popularity in the entertainment and news media. Nevertheless, such popularity has greatly outstripped its attention in the academic literature (Smith and Lilienfeld 2013), contributing to a poor understanding of how psychopathy manifests itself on the job. Although previous research has strongly emphasized the implications of psychopathy for financial and emotional damage to the workplace (e.g., Babiak and Hare 2006; Boddy et al. 2015), some adaptive outcomes have also been acknowledged (e.g., being better at crisis management; see Lilienfeld et al. 2012b). In this study, we aimed to elucidate the associations between psychopathy and both adaptive and maladaptive workplace behaviors from the perspective of the relatively new triarchic model of psychopathy (Patrick et al. 2009). In doing so, we placed particular emphasis on the largely under-researched and still poorly understood trait domain of boldness, which has heretofore received little attention in corporate psychopathy research.

The triarchic model of psychopathy integrates recurring themes in the psychopathy literature using three distinct traits of approximately equal relevance (Patrick et al. 2009; Patrick and Drislane 2014). Boldness encompasses the ability to remain calm in threatening situations and recover quickly from stressful experiences (Patrick et al. 2009). Individuals with high levels of boldness tend to be venturesome, socially dominant, persuasive, and unresponsive to punishment cues. Meanness comprises traits of callousness, deceitfulness, and manipulativeness (Patrick et al. 2009). Individuals with high levels of meanness tend to be self-assured, callous, and prone to aggressive and deceitful behaviors (Anderson et al. 2014; Drislane et al. 2014; Patrick et al. 2009; Sellbom and Phillips 2013). Individuals with high levels of disinhibition are prone to impulse control problems such as non-planfulness, impaired affect regulation, and deficiencies in behavioral restraint (Anderson et al. 2014; Drislane et al. 2014; Patrick et al. 2009; Sellbom and Phillips 2013). The triarchic model is particularly apt for this study because it incorporates boldness, a largely under-researched trait in this context that may hold both adaptive and maladaptive implications for workplace behavior. For example, an individual unresponsive to affective-laden punishment cues (high on boldness) could be unresponsive to negative feedback at work.

One further advantage of the triarchic model is its theoretical overlap with certain models of corporate psychopathy. Marshall et al. (2015) explained corporate psychopathy as a tendency to engage in three problem behaviors. Concealment may relate to boldness, whereby individuals are gregarious and grandiose Imprudence describes excessive risk-taking behavior that is presumably akin to disinhibition. Finally, corruption bears notable similarities to meanness given that it is marked by a lack of a moral compass for behavior.

Corporate Psychopathy

Given that psychopathy is a dimensional construct (see e.g., Marcus et al. 2012), it can attain high levels in the general population, even among individuals who have never been incarcerated (Lilienfeld et al. 2014; Neumann and Hare 2008). Non-incarcerated psychopathic individuals similarly display many or most of the essential characteristics of psychopathy and tend to cause harm in ways other than overt criminal conduct (e.g., lying, stealing, threatening employees; see Babiak and Hare 2006). One area of important consideration in this regard is the workplace, given ample opportunity for individuals high on psychopathy to channel their personality pathology into more covert outlets, such as exploiting and manipulating others. Individuals with marked traits of corporate psychopathy appear to be adept at rising to leadership positions in organizations (Boddy 2011a; Chiaburu et al. 2013). Moreover, their often interpersonally destructive behaviors (e.g., lying and exploiting employees) may bear long-term implications for the workplace and employees (e.g., Hogan et al. 1994; Hogan and Kaiser 2005).

The potentially high levels of psychopathic traits in the workplace are concerning because of the possible financial and emotional harm they can cause organizations and its employees. Babiak and Hare (2006) cited specific cases in which psychopathic individuals create chaos by engaging in embezzlement and imposing demands that are impossible to meet. Employees who work alongside these individuals have reported lower job satisfaction and psychological well-being (Mathieu et al. 2014b), as well as bullying and unfair supervision (Boddy 2011b). Some have even suggested that the 2008 Global Financial Crisis, which led to worldwide financial losses and retrenchment, was attributable in part to the actions of corporate psychopaths (Boddy 2011a). Apart from emotional and financial harm, organizations with psychopathic individuals have been reported to be marked by lower corporate social responsibility (Boddy et al. 2010). That is, these organizations appear to engage in corporate behavior that is judged to be unethical or socially irresponsible. Nevertheless, this research is difficult to interpret given its susceptibility to mono-mode bias, as the same individuals who rated their employees on psychopathy also rated their organizations on corporate malfeasance (Smith and Lilienfeld 2013). Considering the severity of these potential consequences, it is imperative for managers to understand how psychopathy is manifested in the workplace.

Although most of the literature has emphasized the maladaptive implications of psychopathy for the workplace, there is some evidence that certain psychopathic traits, especially boldness, may be associated with adaptive leadership behaviors, such as superior persuasiveness, public communication, and crisis management (e.g., Lilienfeld et al. 2012b). By operationalizing psychopathy from the triarchic perspective, the study will examine the behavioral correlates of boldness in the workplace. Although boldness may exert adaptive main effects on workplace behavior, it may be linked to maladaptive workplace behaviors in statistical interaction with other psychopathy traits, such as disinhibition. A manager or employee who is bold but who possesses intact impulse control may engage in constructive risk-taking, whereas one who is bold but disinhibited may express his or her risk-taking propensities in socially or even physically destructive outlets. It is important for human resources professionals in organizations to understand this downside when evaluating colleagues for their propensity to cause workplace harm.

A number of workplace behaviors have been associated with psychopathy. These include relatively more maladaptive behaviors such as counterproductive work behaviors (CWB) (e.g., Boddy 2014), hard tactics of influence (e.g., Jonason et al. 2012), unethical business decision-making (e.g., Stevens et al. 2012), and passive leadership behavior (e.g., Mathieu et al. 2014a; Westerlaken and Woods 2013), and relatively more adaptive behaviors such as soft tactics of influence (e.g., Jonason et al. 2012), adaptive leadership behavior (e.g. Mathieu et al. 2014a; Westerlaken and Woods 2013), creativity and team play behavior (e.g., Babiak et al. 2010).

CWB

CWB represents a collection of maladaptive behaviors defined by sabotage, theft, withdrawal, production deviance, and abuse (Bennett and Robinson 2000). “Sabotage” refers to damage to the organization’s physical environment, “withdrawal” refers to absenteeism and lateness, “production deviance” is deliberately doing a job incorrectly, and “abuse” is harm directed to others. O’Boyle et al. (2011) meta-analysis examined the associations between the Dark Triad, operationalized using measures of psychopathy, Machiavellianism, and narcissism, and CWB, and reported that psychopathy was positively, albeit modestly, associated with CWB across 27 studies. Most studies employed police/correctional officers and relied on measures that assessed mainly the disinhibition domain of psychopathy. Similar associations were identified in another study that examined CWB using integrity tests (Connelly et al. 2006). The results showed that the Psychopathic Personality Inventory (PPI; Lilienfeld and Andrews 1996) subscales of Machiavellian Egocentricity, Blame Externalization, and Impulsive Nonconformity (mainly “disinhibition” traits; Sellbom and Phillips 2013) were most negatively associated with self-report overt “integrity” test scores, which are highly saturated with indices of CWB.

In a sample of 304 “white-collar” workers, Boddy (2014) found that employees working under psychopathic managers were more likely to engage in CWB, as compared with employees working under non-psychopathic managers. This study operationalized psychopathy using the Psychopathy Measure—Management Research Version (PM-MRV; Boddy 2009), a measure that has demonstrated provisional but promising psychometric properties in both Australian and British work samples (Boddy 2014), including good predictive validity (Boddy 2011b) and high internal consistency (Boddy 2014). Although the content of the PM-MRV was drawn largely from the established psychopathy literature (Cleckley 1941/1988; Cooke and Michie 2001), this measure focuses primarily on traits geared towards “meanness” (Jones and Hare 2015). As a consequence, Boddy’s study may have underestimated the predictive role of potentially adaptive features of psychopathy, such as boldness.

Tactics of Influence

Individuals in an organization can use either hard (e.g., threats of appeal) or soft (e.g., compliments) tactics to influence others. Jonason et al. (2012) found that psychopathy, as operationalized by the Dirty Dozen measure (a measure of the dark triad; Jonason and Webster 2010), was associated with hard but not soft tactics in a mixed sample of 419 university students and community volunteers. Nevertheless, the psychopathy scale of the Dirty Dozen only reflects “meanness” within the triarchic model (Miller et al. 2012), raising questions about whether these findings extend to other features of psychopathy, such as boldness or disinhibition.

It is unclear if boldness is preferentially associated with soft or hard tactics. Considering that individuals with high levels of boldness are influential and characterized by superficial charm and social poise, it is likely that such characteristics would translate to more prosocial behavior, such as the use of soft tactics. However, bold individuals are also fearless and largely unafraid of offending others, and would therefore presumably have less hesitation with employing hard tactics. The nature of these associations will be considered in the current study.

Unethical Business Decision-Making

Unethical decision-making describes an individual’s propensity to engage in immoral or norm-violating behavior, often in ambiguous situations (Stevens et al. 2012). Such behaviors exert lasting consequences on an organization, beyond the decision made at hand. Research suggests that the ethics of business leaders relate to, and potentially affect, organizations’ ability to attract ethically minded people (Ogunfowora 2014). Psychopathy, as operationalized by global scores on the Self-Report of Psychopathy Scale (SRP-III; Paulhus et al. 2009), was positively associated with unethical decision-making among 272 university students (Stevens et al. 2012). The SRP-III is primarily associated with meanness and disinhibition (although it also has some saturation with boldness; see Drislane et al. 2014), but it is unclear from their findings which of those domains would be most directly relevant to unethical decision-making. Moreover, Heinze et al. (2010) found that the PPI Machiavellian Egocentricity subscale (which is primarily a measure of “meanness”; see Sellbom and Phillips 2013) was positively associated with unethical decision-making. In contrast to the Stevens et al. (2012), Heinze et al. (2010) used a sample that is of greater relevance to business settings (66 Master of Business Administration students) and did not conceptualize psychopathy as a unitary construct. Nevertheless, they used non-business-related vignettes that were not particularly representative of unethical decision-making in most workplaces (e.g., whether to steal food during a famine from a rich man who does not need the extra food and intends to sell it on the black market). Thus, further explication of which psychopathy domains are most linked to these behaviors is necessary. However, in light of the literature just reviewed, it is likely that both meanness and disinhibition would be negatively associated with ethical decision-making.

Leadership

The association between psychopathy and leadership behaviors, with respect to the Full Range Leadership Model ([FRLM]; Avolio and Bass 1991), has also been examined (e.g., Mathieu et al. 2014a; Westerlaken and Woods 2013). The FRLM consists of factors that capture transformational, transactional, and passive leadership styles. Transformational leadership uses techniques such as idealized influence, inspirational motivation, intellectual stimulation, and individual consideration, to influence others. Transactional leadership relies on tangible rewards and punishment, and consists of two factors (Avolio et al. 1999). Contingent reward refers to using rewards to commend achievements. Active management-by-exception is actively supervising work and providing negative reinforcement. As leaders continually look over to correct mistakes, employees increase their quality of work so as to avoid being corrected. Both transformational and transactional leadership are adaptive leadership styles associated with higher organizational performance (Bass and Avolio 1993; Bass et al. 2003). Passive leadership is associated with a more hands-off approach and is characterized by passive management-by-exception, acting only in response to mistakes, and laissez-faire leadership, in which leaders are uncommitted and lazy (Avolio et al. 1999). This leadership style is maladaptive and inconsistent with leader effectiveness (Judge and Piccolo 2004).

Westerlaken and Woods (2013) examined the relationship between psychopathy and the FRLM using the Self-Report Psychopathy Scale-Revised (SRP-III-R12; Williams et al. 2007) in 115 university students, and found that psychopathy was associated with more passive and less adaptive leadership. Mathieu et al. (2014a) reported similar findings when they examined psychopathy, assessed by the Business-Scan 360 (B-Scan 360; Mathieu et al. 2013), and the FRLM in a mixed sample of 591 employees in financial institutions and the public service. From the triarchic model perspective, these studies suggest that psychopathy measures that are highly saturated with meanness and disinhibition are associated with destructive leadership behaviors. Nevertheless, these studies may have largely neglected boldness and its relation with leadership behaviors.

In contrast, Lilienfeld et al. (2012b) argued that not all leaders with high levels of psychopathic traits are maladaptive in their examination of psychopathy in 42 Presidents of the United States (up to and including George W. Bush) and their performance during their presidential terms. Psychopathy scores were derived from presidential experts/biographers ratings on the five-factor model of personality (Costa and McCrae 1992). Presidential performance was assessed by ratings from panels of presidential historians (e.g., crisis management and public persuasiveness) and other objective measures (e.g., election landslides and initiating new legislation). They found that estimates of PPI Fearless Dominance, which is analogous to boldness, were associated with better presidential performance. These findings provide preliminary evidence that boldness may be positively associated with adaptive leadership (behavior predictive of higher organizational performance) potentially complementing the possibility for maladaptive leadership prediction via meanness and/or disinhibition.

Creative Thinking

Creativity is a process that enables the ability to create new and different effective ideas (Amabile 1983). The process of creativity consists of divergent thinking and cognitive flexibility. Divergent thinking is the ability to form original ideas and conceptualize multiple solutions to a problem (Guilford 1968) and cognitive flexibility is the ability to reconceptualize ideas with respect to varying situational demands (Spiro and Jehng 1990).

Babiak et al. (2010) examined psychopathy, operationalized by the PCL-R, and creativity and found that individuals with high scores on the PCL-R Interpersonal facet tended to exhibit higher creativity, operationalized in terms of a charisma/presentation style factor, which included creativity, strategic thinking, and communication abilities. In triarchic model terms, the PCL-R Interpersonal facet is related primarily to meanness, but is also the only PCL-R facet that contains non-trivial amounts of variance relevant to boldness (Venables et al. 2014; Wall et al. 2015). Past research has provided some evidence that PPI Fearless Dominance (which is a measure of boldness; Drislane et al. 2014; Sellbom and Phillips 2013) is positively associated with cognitive flexibility, as operationalized in an overall executive cognitive functioning score (Sellbom and Verona 2007). As such, boldness may be associated with greater levels of creativity, a largely adaptive characteristic bearing substantial implications for workplace behavior.

Team Play

An effective work team is a group whose performance is greater than the sum of each individual team member’s effort (Robbins et al. 2011). Babiak et al. (2010) found that corporate professionals with high scores on the PCL-R Interpersonal facet received low scores on a responsibility/performance factor, which comprised an individual’s team play abilities, management style, and general performance. Further, Jonason et al. (2012) found that the Dirty Dozen psychopathy scale negatively predicted an individual’s use of team play at work. Taken together, it is likely that individuals with high levels of psychopathy, and in particular meanness, will engage in less team play in the workplace.

It is, however, unclear how boldness is associated with team play. Given that boldness reflects the social dominance traits of psychopathy, it is likely that boldness will predict less team play. In contrast, considering that individuals with high levels of boldness are also characterized by a lack of social anxiety and being superficially gregarious, it would not be surprising for individuals high on boldness to engage in more team play.

The Current Study

The goal of the current study was to elucidate the associations between psychopathic personality traits and both maladaptive and adaptive workplace behaviors from the perspective of the triarchic model, a perspective that has received little explicit attention in the business literature. One advantage of this model is its potential to shed light on both the positive and negative implications of corporate psychopathy. This scientifically balanced perspective is broadly consistent with burgeoning literature on the “bright” and “dark” sides of leadership and of work behavior more generally (Janssen et al. 2004; Judge et al. 2009). We also emphasize the heretofore-underemphasized boldness trait domain and its implications for both adaptive (as a main effect) and maladaptive (in statistical interaction with other psychopathy traits) workplace behaviors. This study also made use of a more generalizable sample to the average worker in the United States as it consisted of a greater range of workers in the community relative to studies on specific professions (e.g., “white-collar” workers, police/correctional staff, and university students).

In light of the literature we have reviewed, we hypothesized that the triarchic psychopathy domains would be differentially associated with maladaptive and adaptive workplace behaviors. First, we hypothesized that meanness and disinhibition would be positively associated with CWB (Boddy 2014; O’Boyle et al. 2011). We further hypothesized that boldness would potentiate the positive association between disinhibition and CWB (i.e., an interaction effect) as individuals with high levels of disinhibition would be more likely to engage in CWB when they are also fearless (not afraid to make serious mistakes or to offend other employees). We further hypothesized that meanness (e.g., seek pleasure from abusing others; exploitativeness) would potentiate the positive association between disinhibition and CWB given that the simultaneous presence of poor impulse control and callousness would be expected to eventuate in greater risk for antisocial behavior in the workplace.

Second, we hypothesized that meanness would be positively associated with hard tactics (Jonason et al. 2012). We also explored the association between boldness and both hard and soft tactics. Moreover, we hypothesized that boldness would potentiate the positive association between meanness and hard tactics, and that individuals high on boldness would exhibit soft tactics when low on meanness. Although individuals with high levels of boldness may exhibit more prosocial behaviors, especially those involving social and physical risk, high levels of callousness will probably attenuate that association (Smith et al. 2013).

Third, we hypothesized that meanness and disinhibition would be negatively associated with ethical decision-making (Heinze et al. 2010; Stevens et al. 2012). We also hypothesized that higher levels of disinhibition would potentiate the negative association between meanness and ethical decision-making, as highly disinhibited individuals are non-planful and prone to acting on their impulses.

Fourth, we hypothesized that boldness would be positively, whereas meanness and disinhibition would be negatively (Mathieu et al. 2014a; Westerlaken and Woods 2013), associated with adaptive leadership (Lilienfeld et al. 2012b). We also hypothesized that boldness would be negatively, but meanness and disinhibition would be positively, associated with passive leadership. We further hypothesized that lower levels of boldness would potentiate the positive association between disinhibition and passive leadership. Individuals with high levels of disinhibition are non-planful and impulsive and should be more likely to exhibit passive leadership behaviors (hands-off leadership style) when lower on boldness (higher anxiety levels and lower social dominance). Conversely, we hypothesized that lower levels of disinhibition would potentiate the positive association between boldness and adaptive leadership.

Fifth, we hypothesized that boldness would be positively associated with the cognitive flexibility aspect of creativity (Babiak et al. 2010; Sellbom and Verona 2007). Finally, we predicted that meanness would be negatively associated with team play (Babiak et al. 2010). We did not advance any clear predictions for boldness and team play; thus, this aspect of the study was exploratory.

Method

Participants

Participants were 510 adults from the United States recruited via Amazon’s Mechanical Turk (MTurk), an online crowd sourcing system for registered users to complete surveys for monetary compensation. MTurk has a reputation for yielding high quality data, with psychometric properties comparable with or exceeding those of published research (Buhrmester et al. 2011). Of this 510, 26 participants did not meet validity benchmarks (i.e., >44) on the Psychopathic Personality Inventory-Revised (PPI-R; Lilienfeld and Widows 2005) Inconsistency Responding Scale, nine had left more than 15 % of items on the PPI-R incomplete, and 139 reported that English is not their native language.Footnote 1 As some participants met more than one exclusion criterion, a total of 167 participants were screened out. The final sample (N = 343) consisted of individuals who ranged in age from 18 to 71 years (M = 35.10, SD = 10.80). Of these 343 participants, 62.68 % completed the survey in North America (214 in United States and 1 in Puerto Rico), 34.11 % in Asia (113 in India, 2 in the Philippines, and 1 each in Singapore and Thailand), 2.04 % in Europe (2 in the United Kingdom and Macedonia, and 1 each in France, Italy, and Turkey), 0.58 % in South America (1 each in Columbia and Mexico), 0.29 % in Oceania/Australia (1 in New Zealand), and 0.29 % in Africa (1 in Kenya). The sample consisted of slightly more males (51.02 %). The average years of education was 15.36 years (SD = 3.02). The average years of work experience was 12.60 years (SD = 9.36). Table 1 presents descriptive statistics of the sample.

Table 1 Frequency statistics of sample
Full size table

Measures

Psychopathic Personality Inventory-Revised (PPI-R)

The PPI-R (Lilienfeld and Widows 2005) is a 154 item self-report measure of psychopathy that captures psychopathic personality traits associated with various psychopathy conceptualizations (e.g., Cleckley 1941, 1988; Karpman 1941). The PPI-R items can be aggregated to score the triarchic domains of psychopathy (Hall et al. 2014), with substantial validity support in inmate, community, and university samples (Hall et al. 2014; Sellbom et al. 2015).

Levenson’s Self-Report Psychopathy Scale (LSRP)

The LSRP (Levenson et al. 1995) is a widely used, 26-item self-report measure of psychopathy. The items can be grouped into three factors, namely egocentricity, callousness, and antisociality (Brinkley et al. 2008); these factors have been validated in several studies (e.g., Brinkley et al. 2008; Salekin et al. 2014; Sellbom 2011).

Counterproductive Work Behavior Checklist (CWB-C)

The CWB-C (Spector et al. 2006) is a 45-item self-report measure of CWB. The CWB-C yields five subscale scores: Sabotage, Withdrawal, Production Deviance, Theft, and Abuse. This measure has received initial psychometric support (Spector et al. 2006) and converges with peer-report versions of the measure (Fox et al. 2007).

Tactics of Influence

The 16 tactics of influence measure (DuBruin 1991) asks participants how often they use a given tactic to influence others at work. Apart from “assertiveness” and “logic or reason,” all other tactics were indexed into hard and soft tactics (see Jonason et al. 2012). Jonason et al. (2012) indexed two composites, hard and soft tactics, by assessing face validity and internal consistencies of these tactics; their study provided initial psychometric support for these aggregates with respect to psychopathy associations.

Business Decision-Making Vignettes

A measure of ethical decision-making was specifically designed for this study. Participants were presented with four business-related scenarios and two decisions to be made for each scenario (one ethical and one unethical) (see “Appendix”). For each decision, participants rate how moral they believe the decision to be on a five-point Likert scale (1 = not at all moral, 2 = a little moral, 3 = somewhat moral, 4 = very moral, 5 = extremely moral), and how likely (a) they would engage in the behavior and (b) others would engage in the behavior on another five-point Likert-type scale (1 = not at all likely, 2 = a little likely, 3 = somewhat likely, 4 = very likely, 5 = extremely likely). Scores on the “unethical decision” were reversed to form three overall scores; how moral an individual thinks the decision is (Ethical), how likely he or she is to engage in that behavior (Self-Behave), and how likely others will engage in that behavior (Others-Behave).

Team Player Inventory (TPI)

The TPI (Kline 1999) is a 10-item self-report questionnaire to assess an individual’s positive predisposition to engage in team play at work. The TPI shows good convergent validity with other measures of similar constructs. For example, in one study, the TPI was positively associated with the Social Participation subscale of the Jackson Personality Inventory at r = .55 (Kline 1999).

Multifactor Leadership Questionnaire (MLQ)

The MLQ Actual (Bass and Avolio 1995) is an 18-item self-report measure of the FRLM. The measure yields nine subscale scores, which index three leadership styles: transformational, transactional, and passive leadership. Adaptive leadership comprises of transformational and transactional leadership styles and maladaptive leadership is represented by the passive leadership style. The structural validity of the MLQ has been validated in these studies (Antonakis et al. 2003; Muenjohn and Armstrong 2008).

Remote Associates Test (RAT)

The RAT (Mednick 1962; Mednick and Mednick 1967) is a widely used measure of creativity. Participants were presented with 25 three-word tasks and asked to find a word that is associated with all three words. For example, the three words “room,” “blood,” and “salts,” are associated with “bath” (i.e., bathroom, bloodbath, bath salts). The RAT has been validated against measures that aim to measure insightful problem solving (Ansburg 2000), which in part reflects cognitive flexibility.

Data Analyses

Structural equation modeling (SEM) was used to estimate the associations between the latent constructs of the triarchic psychopathy domains and various workplace behaviors. Confirmatory factor analysis was carried out using Mplus Version 7.3 (Muthén and Muthén 2014) to specify an 11-factor model (three psychopathy domains; eight criterion domains) and estimate zero-order correlations among latent constructs. In addition, structural regression analyses were conducted to examine the unique contribution of each latent triarchic domain in statistically predicting hypothesized workplace behavior latent constructs, including the testing of hypothesized interaction effects. Maximum likelihood estimation with robust scaling (MLR) was used to estimate the parameters in the model. Indicator variables for boldness, hard tactics, soft tactics, creativity, and team play were constructed by dividing items into two or three parcels, as equally as possible (see Little et al. 2002). The meanness construct was estimated by PPI-R Meanness, LSRP Egocentricity, and LSRP Callous. The disinhibition construct was estimated by PPI-R Disinhibition and LSRP Antisocial. The three indicator variables for the ethical decision-making latent variable were constructed by reverse-scoring items from the unethical decision measure. The indicator variables for latent CWB and passive leadership variables were the subscales associated with the construct. Adaptive leadership was estimated using three indicator variables, two of which represent transformational leadership and one representing transactional leadership. Latent interaction variables of boldness × meanness, meanness × disinhibition, and boldness × disinhibition were estimated using all unique two-way interaction with centered indicator variables.

Results

Table 2 presents the descriptive statistics and internal consistency reliability estimates for all observed variables. Although some indicator variables were quite brief (e.g., Laissez-Faire leadership) and therefore resulted in lower than typically accepted Cronbach’s alphas, all mean inter-item correlations showed acceptable unidimensionality (>.15; Clark and Watson 1995).

Table 2 Means, standard deviations, range, skew, kurtosis, and internal consistency estimates
Full size table

Confirmatory Factor Analysis

The hypothesized model (Model 1), χ 2 (409, N = 343) = 1246.29, p < .001, χ 2/df = 3.05, RMSEA = .077, CFI = .90, TLI = .87, produced RMSEA and CFI/TLI values that indicated marginal acceptable fit. However, a respecified model was considered through the use of empirically derived modification indices, to the extent conceptually defensible. In the subsequent model (Model 2), correlated residuals of LSRP Egocentricity and LSRP Callousness, and Self-Behave and Others-Behave from the ethical decision-making vignettes were added. LSRP Egocentricity and LSRP Callousness were added as it is conceptual sound that there would be residual measurement variance overlap unaccounted for by the latent meanness construct. In addition, it also makes sense to correlate Self-Behave and Others-Behave indicators, as how oneself behaves would be associated with how they expect others to behave, which goes beyond the latent construct of ethical decision-making. Model 2, χ 2 (407, N = 343) = 1145.22, p < .001, χ 2/df = 2.81, RMSEA = .073, CFI = .91, TLI = .89, produced an RMSEA, χ 2/df, and CFI/TLI values with more acceptable fit. Further, Model 2 produced a statistically better fit compared with Model 1, Δχ 2 = 101.07, df = 2, p < .001. In the final measurement model, all observed indicators were sufficiently accounted for by their latent construct (indicator loadings >.45) except for the Others-Behave indicator at .29 (see Fig. 1).

Fig. 1
figure 1

Best fitting final measurement model. All values are standardized and significant at p < .05. Refer to Table 3 for correlations between latent variables. B1 boldness 1, B2 boldness 2, B3 boldness 3, M1 PPI-R meanness; M2 LSRP egocentricity, M3 LSRP callousness, D1 PPI-R disinhibition, D2 LSRP antisocial, Self-B self-behave; Others-B others-behave, C creativity, Team team play, Hard hard tactics, Soft soft tactics; Tf transformational leadership, Tc transactional leadership, PMBE passive management-by-exception, LF laissez-faire, Sa sabotage, With withdrawal, PD production deviance, CWB counterproductive work behavior

Full size image

Table 3 presents the correlations among the 11 constructs. As expected, meanness and disinhibition were positively correlated with CWB. All three psychopathy domains were positively associated with hard and soft tactics of influence. It was not expected, however, that disinhibition would be associated with hard tactics, and meanness and disinhibition to be associated with soft tactics. Further, all three domains were negatively associated with ethical decision-making. Meanness and disinhibition were most correlated with ethical decision-making (r = −.77 and −.73, respectively). As hypothesized, boldness was positively associated with adaptive and negatively associated with passive leadership, and both meanness and disinhibition were negatively associated with adaptive leadership and positively associated with passive leadership. Contrary to our hypotheses, boldness was not associated with creativity. Lastly, boldness was positively and disinhibition (rather than meanness, as predicted) was negatively associated with team play.

Table 3 Zero-order correlations among latent variables for best fitting measurement model
Full size table

Regression Analyses

Because the magnitudes of correlations disattenuated at the latent level (increase correlation magnitude as measurement error is accounted for) and the latent triarchic domains are overlapping (particularly meanness and disinhibition), a latent regression analysis allowed for a better representation of the unique associations of each of the psychopathy domains with the workplace latent constructs. A structural regression model was estimated whereby only the hypothesized main effect associations were specified in the model. The model fit statistics, χ 2 (415, N = 343) = 1257.28, p < .001, χ 2/df = 3.03, RMSEA = .077, CFI = .89, TLI = .87, were similar to the final measurement model and indicated a marginal to acceptable fit. Table 4 presents the associations among the predictor and criterion variables after specifying regression pathways according to our hypotheses. The results showed that disinhibition positively predicted CWB, whereas meanness did not. As expected, meanness positively predicted hard tactics of influence. Boldness positively predicted soft tactics but did not predict hard tactics of influence. Meanness negatively predicted ethical decision-making but disinhibition did not. As hypothesized, boldness positively predicted adaptive leadership, but both meanness and disinhibition did not. Results further indicated that boldness negatively and disinhibition positively predicted passive leadership, whereas meanness did not. Unexpectedly, boldness did not predict creativity. Lastly, boldness positively and meanness negatively predicted team play. All significant regression pathways are shown in Fig. 2.

Table 4 Standardized beta weights for latent triarchic psychopathy domains predicting latent workplace behavior criterion variables
Full size table
Fig. 2
figure 2

Latent regression model. Correlations among predictor and criterion variables, respectively, are estimated but not shown. All values are standardized and significant at p < .05. CWB counterproductive work behavior, Passive MBE passive management-by-exception

Full size image

Finally, we examined non-additive (interactive) effects of psychopathy domains on the prediction of workplace behaviors. Separate models to test each individual interaction term were estimated to avoid issues with multicollinearity. The model fit statistics after adding a boldness × disinhibition term predicting CWB, adaptive and passive leadership, χ 2 (600, N = 343) = 1501.22, p < .001, χ 2/df = 2.50, RMSEA = .066, CFI = .90, TLI = .88, indicated an acceptable fit. None of the hypotheses were supported. The model fit statistics after adding the boldness × meanness term in predicting hard and soft tactics, χ 2 (706, N = 343) = 1759.26, p < .001, χ 2/df = 2.49, RMSEA = .066, CFI = .88, TLI = .86, indicated mediocre to acceptable fit. Neither hypothesis was supported. The model fit statistics after adding the meanness × disinhibition interaction term predicting ethical decision-making and CWB, χ 2 (602, N = 343) = 1644.68, p < .001, χ 2/df = 2.73, RMSEA = .071, CFI = .88, TLI = .86, indicated mediocre to acceptable fit. Higher levels of meanness potentiated the positive association between disinhibition and CWB, as hypothesized (β = .18, p = .01). Higher levels of disinhibition did not potentiate the negative association between meanness and ethical decision-making.

Discussion

This current study is the first to examine the association between psychopathy, as operationalized by the triarchic model of psychopathy, and a wide range of workplace behaviors. Our study also explicitly examined the role of boldness, a largely under-researched psychopathy-related trait domain in the workplace literature. In light of the existing the literature, we hypothesized that the triarchic psychopathy domains would be differentially associated with workplace behaviors. In general, we found that boldness positively predicted behaviors like the use of soft tactics of influence, adaptive leadership, and team play, and negatively predicted passive leadership. Meanness negatively predicted ethical decision-making and team play, and positively predicted the use of hard tactics. Disinhibition positively predicted CWB and passive leadership. These findings reinforced the multifactorial nature of psychopathy and its potentially different workplace manifestations.

First, the results indicated that only disinhibition uniquely predicted CWB. Although meanness was also significantly correlated with CWB, this relation diminished after accounting for overlap with disinhibition. This result is supported by other studies that operationalized psychopathy using dimensions that mainly reflect disinhibition (e.g., Connelly et al. 2006; O’Boyle et al. 2011). Nevertheless, our results are inconsistent with those of studies that found the interpersonal and affective aspects of psychopathy (mainly “meanness” traits) to be positively associated with CWB (e.g., Boddy 2014). The measure used by Boddy (2014), the PM-MRV, was developed on the assumption that antisociality is not an essential feature of psychopathy (consistent with the views of Cooke et al. 2004; see also Cooke et al. 2012) and therefore does not include items that explicitly assess antisocial behaviors and impulse control deficits (Jones and Hare 2015). As such, the findings of Boddy (2014) and others do not bear directly on the extent to which PCL-R Factor 2 features (which primarily assess “disinhibition” traits) are associated with CWB. We also found that meanness moderated the association between disinhibition and CWB, in that higher scores on both were associated with exponentially greater CWB. This moderating effect emphasizes that certain configurations of the triarchic psychopathy traits would likely manifest in different levels of maladaptive behavior.

Second, meanness was positively associated with hard tactics and boldness was positively associated with soft tactics. This finding seemingly contrasts with previous research that found no association between psychopathy and soft tactics (e.g., Jonason et al. 2012). This difference could be attributed to how psychopathy was operationalized. The Dirty Dozen measure is an extremely brief and highly content-restrictive measure that captures only the interpersonal and affective aspects of psychopathy (mainly “meanness” traits). As such, their study neglected the possibility that other aspects of psychopathy could be associated with soft tactics. Our finding further suggests that bold individuals are more likely to persuade and dominate social environments in prosocial ways.

Third, results showed that only meanness uniquely contributed to the prediction of unethical decision-making. This finding is conceptually consistent with the fact that meanness (characterized by traits of exploitativeness and manipulativeness) logically aligns with moral- or norm-violating behavior, such as unethical decision-making, as consistent with past research (e.g., Heinze et al. 2010). Additionally, these results clarified previous research (e.g., Stevens et al. 2012) that operationalized psychopathy from a unitary perspective (i.e., just one total score that combined meanness and disinhibition) but did not examine which constituent domains uniquely accounted for the variance in unethical decision-making. Thus, while zero-order associations between disinhibition and unethical decision-making were observed, the former did not uniquely predict unethical decision-making above and beyond meanness. This finding dovetails with the corruption domain outlined by Marshall et al. (2015), which is characterized by the absence of a moral compass to regulate behavior.

Fourth, boldness positively predicted adaptive leadership, and disinhibition positively and boldness negatively predicted passive leadership. These findings are inconsistent with studies that found all aspects of psychopathy, as in the PCL-R four-factor model (primarily meanness and disinhibition), to be positively associated with passive leadership and negatively with adaptive leadership (e.g., Mathieu et al. 2014a; Westerlaken and Woods 2013). Although meanness was significantly associated with passive leadership as in previous findings, this effect was attenuated to non-significance after accounting for its overlap with disinhibition. Previous studies examined the dimensions of psychopathy and leadership behaviors at a zero-order level and did not examine each psychopathy dimension after controlling for variance explained by other dimensions. This methodological approach is common, and other studies examining psychopathy and other operationalizations of leadership, in particular self-leadership (e.g., Furtner et al. 2011), have similarly only examined psychopathy and leadership at the zero-order level. Hence, we encourage researchers to examine the unique contributions of psychopathy dimensions in future research on workplace behaviors.

Finally, boldness was not significantly associated with creativity. This inconsistency with previous research (e.g., Babiak et al. 2010) may be due to how creativity was measured. Babiak et al. (2010) operationalized creativity using the 360-degree feedback form, which relied on observer ratings to assess creativity. As such, their findings may indicate that psychopathy is related to perceptions of creativity rather than actual creative intellectual ability. Alternatively, this discrepancy could be attributable to the fact the RAT assesses a relatively narrow element of creativity (cognitive flexibility) and does not measure divergent thinking, which was presumably better accounted for in Babiak et al.’s (2010) study.

Lastly, our results indicated that meanness negatively, as expected, and boldness positively predicted team play. Individuals with high levels of boldness are characterized by a lack of social anxiety (e.g., PPI-R Boldness includes items such as not being bothered by talking in large groups and finding it easy to introduce oneself to strangers). Considering that engaging in team play requires individuals to work with others, individuals who are less socially anxious may be more willing to engage in team play. Individuals with high levels of boldness may also be particularly skilled at concealing their negativity and appearing harmonious when engaging with others, although this intriguing possibility awaits further research. In addition, individuals with high levels of meanness may engage less in team play given that they are untrustworthy and likely to perform poorly in team-based tasks.

Implications and Future Directions

This research showed that the different aspects of psychopathy are differentially predictive of various workplace behaviors. Even though meanness and disinhibition were more predictive of maladaptive workplace behaviors, they appear to relate to different kinds of maladaptive behaviors. These findings indicate that operationalizing psychopathy unidimensionally would only confuse rather than clarify how psychopathy manifests itself in the workplace. One limitation of the present analyses was our exclusive focus on the three broad domains of Boldness, Meanness, and Disinhibition. In accord with the time-honored bandwidth-fidelity distinction in psychometrics (Cronbach 1960; Hogan and Roberts 1996), further research would benefit from the development of triarchic scales that provide lower-order facets within these three domains. These facets may yield finer resolution of specific subtraits and thereby allow more precise predictions of workplace behaviors. Indeed, research on the five-factor model of personality reveals that differing facets (e.g., assertiveness, gregariousness, warmth) within broad trait domains (e.g., Extraversion) sometimes relate in differing, even opposing, ways with behavior (e.g., Ashton 1998).

In addition, the significant and non-significant interaction effects observed also suggest that, in some cases, specific triarchic psychopathy trait combinations would likely predict different levels, or even types, of behavior (see also Kastner and Sellbom 2012; Rock et al. 2013; Venables et al. 2015, for such findings with boldness and disinhibition). Future research should expand upon such findings and examine how different aspects and combinations of psychopathy traits manifest in different occupational (e.g., among professional athletes and high-risk occupations) and forensic settings (e.g., recidivism, crime, and violence). Although certain psychopathy traits in isolation, such as boldness, may be advantageous for certain occupations, it is plausible that the simultaneous presence of all three triarchic traits will predispose to highly maladaptive workplace behaviors. Nevertheless, larger sample sizes will be needed to elucidate such three-way interactions.

Importantly, the current findings also indicate that high levels of certain psychopathy traits are not always maladaptive in the workplace. Interestingly, this divergence between the maladaptive and adaptive aspects of psychopathy has led to debate regarding whether boldness is a defining feature of psychopathy. For example, Miller and Lynam’s (2012) meta-analytic review of psychopathy seemed to suggest that boldness is largely irrelevant to psychopathy considering its non-association with maladaptive behavior. Nevertheless, this meta-analysis provided a truncated nomological network of psychopathy, and a careful examination of the literature indicates that boldness is moderately related to many well-validated psychopathy measures and behaviors known to be associated with psychopathy (Lilienfeld et al. 2012a; Strickland et al. 2013). Our study reinforces the importance of the boldness aspect in psychopathy research and how neglecting it would paint an incomplete picture of how psychopathy manifests itself in the workplace.

Apart from examining specific behaviors that are associated with psychopathic personality traits, future research could also aim to understand how psychopathy relates to overall organizational performance. Although here is some qualitative evidence available on this issue (e.g., interviews with management staff who have experience with “corporate psychopaths”; see e.g., Boddy et al. 2015), more quantitative research along these lines would be helpful. Incorporation of boldness into such studies would be helpful in light of the current findings.

Limitations

Our conclusions should be considered in light of several limitations. First, we excluded 129 participants, which limits generalizability to countries in which English is the native language. Future research could explore the potential cross-cultural differences in psychopathic traits from English and non-English speaking countries, and ascertain whether psychopathy expresses itself differently in the workplace cross-culturally. Nevertheless, to do so, psychopathy will need to be validly operationalized in diverse cultures.

Second, all constructs were operationalized using self-report questionnaires. Considering that pathological lying is a hallmark feature of psychopathy, highly psychopathic individuals may sometimes misreport their socially undesirable attributes on self-report questionnaires. Interestingly, however, research has found that psychopathic individuals are no better at dishonest reporting even though they were more likely to attempt such reporting (Marion et al. 2013; Ray et al. 2013; Verschuere et al. 2014). Apart from dishonest responding, psychopathic individuals also experience affective deficits and may lack the insight to self-report accurately. These deficits could be problematic as this study required individuals to report on their behaviors in the workplace. Future research could consider more objective measures of workplace behaviors, such as workplace disciplinary reports and informant reports.

The overall study survey was somewhat lengthy (318 items). As such, some participants might have been fatigued or had significant lapses in attention and thereby provided careless responses. Nevertheless, we placed the PPI-R, which includes a measure of random responding, at the tail end of the protocol, and we excluded participants who failed to meet a minimal threshold for consistent responding. Hence, it is likely that the remaining protocols are relatively free of careless responding.

Furthermore, our exclusive reliance on self-report questionnaires renders our findings vulnerable to mono-method bias (i.e., upwardly biased associations between self-reported variables). Nevertheless, some scholars contend that the broader problem of shared method variance has been overstated and that the use of different measurement modalities may not guarantee a more accurate estimate of these associations (Conway and Lance 2010). Even so, it will be important to conceptually replicate our findings using alternative indicators of both psychopathy and CWB, especially observer reports, which may help circumvent some of the potential responses biases associated with questionnaires.