Academy of Strategic Management Journal (Print ISSN: 1544-1458; Online ISSN: 1939-6104)

Research Article: 2017 Vol: 16 Issue: 1

A Shift in Leadership

Eyal Eckhaus, Ariel University


Transformational Leadership, Personality, Text Mining


Leadership is the most important role in the organization (Chang, Chiou, & Wang, 2007), and is the most frequently researched area of organizational behavior and management (Micic, 2015). As it became evident that all leaders do not possess the same traits or attributes, the study of leadership began to focus on the behavior of leaders (Peters, 1997). One of the most common concepts in leadership is the transformational-transactional one, which was introduced in the 1970s and was later expanded (Bass, 1985, 1990, 1996; Bass & Avolio, 1994a, 1994b; Burns, 1978; Downton, 1973). The transformational leader was defined as one who serves as a role model, demonstrates an ability to motivate, inspire, and stimulate followers to be creative and innovative, and shows genuine concern for each individual (Bass & Avolio, 1994a; Hassan, Fuwad, & Rauf, 2010). Transformational leaders challenge the way followers think so as to enable them to be more creative in their problem solving (Burns, 1978). This study, therefore, focuses on transformational leadership (TL). And yet every positive trait has a negative side (Hartlep, 2013; Marques, 2008; Toffa, 2004) and personality traits may change or shift over time (Lüdtke, Roberts, Trautwein, & Nagy, 2011). For instance, life events may change a person's stable level of happiness (Lucas, 2007; Robertson & Cooper, 2011: p.43 ). Therefore, traits that constitute TL may also gradually change, and when this change lacks balance it can be negatively manifested. For example, if a leader is too caring it may lead to underperformance (Cooper, 2014). Similarly, too little or too much novelty or communicability is unhelpful (Scott, 2006). However, most research on transformational leadership concentrates on its positive influence on employees and the organization (Breevaart, Bakker, Demerouti, & Derks, 2015). For instance, Schein (1985) and Trice and Beyer (1993) argue that leadership can change in a constructive direction and support the organization’s culture through the institution of a new cultural identity based on shared values and norms. Eckhaus (2016) found a relationship between the manifestation of leader TL traits and company stock price, concluding that TL may change over time. Employing Eckhaus’ (2016) framework, this study focuses on the ability of traits that constitute TL to shift to the point that the leader may even cease to be defined as transformational. Specifically, among known negative traits, Machiavellianism, narcissism and psychopathy are often referred to as the dark triad of personality (Paulhus & Williams, 2002), and have recently attracted interest in management research (Kramer, Cesinger, Schwarzinger, & Gelléri, 2011).

This study makes several contributions. First, it provides empirical evidence for the presence of different manifestation rates of TL. Second, it enhances the research on TL with the perspective that TL is not necessarily constant and may change over time. Third, the study employs a powerful technique of Natural Language Processing (NLP) and reinforces the usefulness of one the most common and basic techniques in the field of text analysis. A case study on Enron is presented in order to demonstrate the research arguments. Managerial implications are then discussed.

The Dark Side of Leadership

The dark side of TL and charismatic leadership has been previously emphasized (Conger, 1990; Howell & Avolio, 1992), focusing on the negative traits and behaviors of the leader (Howell & Shamir, 2005). For instance, one of the most notorious traits of successful leaders is hubris, a trait regarded as a common reason for leadership failure (Kroll, Toombs, & Wright, 2000). Hubristic CEOs engage in more risk-taking (Hollow, 2014; Petit & Bollaert, 2012), biased reporting (Brennan & Conroy, 2013), and often overestimate their ability to extract value from acquisitions (Hayward & Hambrick, 1997). Similarly, over-confidence, which partially rises from self-attribution bias (Leonard & Biberman, 2007), harms the decision making process (James & Gudmundsson, 2011), and can lead to hubris (Sharif & Irani, 2006).

The ‘dark triad’ is a group of three personality traits encompassing Machiavellianism, narcissism, and psychopathy (Kaur & Swati, 2013; Visser, Pozzebon, & Reina-Tamayo, 2014), which are associated with negative oriented leaders (Furtner, Rauthmann, & Sachse, 2011).


Narcissism refers to a sense of grandiosity and a strong need to be seen and admired by others (Thomaes, Brummelman, Reijntjes, & Bushman, 2013). Some manifestations of narcissism include an extravagant sense of self-importance, a sense of superiority, self-centered and self-referential behavior, exaggeration of talents, boastful and pretentious behavior, grandiose fantasies of unlimited success, a willingness to exploit others, and lack of empathy (Arjoon, 2010). It was found that leaders expressing personalized characteristics such as narcissism often had a negative impact on society (Bedell, Hunter, Angie, & Vert, 2006), and their organization (Pryor, Odom, & Toombs, 2014). Some academics argue that the latest global financial crisis was caused, in part, by the narcissistic personality disorder (NPD) of corporate leaders who substituted robust risk management for greed and personal gains by promoting self-serving and grandiose aims (Arjoon, 2010)


Machiavellianism refers to an individual's tendency to deceive and manipulate others for personal gain (Mudrack, 1990). Researchers have extensively examined the relationship between charismatic leadership and Machiavellianism (Deluga, 2001; Gardner & Avolio, 1995; House & Howell, 1992). Transformational and charismatic leadership have been the primary typologies of outstanding leadership. However, studies show that leaders are cognitively, behaviorally, and developmentally different from each other, and that charismatic, ideological, and pragmatic leaders use characteristically different methods of influence (Bedell et al., 2006). This implies that leaders may manifest transformational leadership qualities differently, and to varying extent.


Psychopathy is characterized by traits such as manipulativeness, superficial charm, lack of remorse or shame, egocentricity, exploitation, deceitfulness, irresponsibility, arrogance, and shallow affect (Cleckley, 1988). It can be understood as an extreme variant of common dimensions of personality (Miller, Lynam, Widiger, & Leukefeld, 2001). Extreme traits can negative or generally interfere with socially or personally valued goals (Wakefield, 2008). This indicates that an extreme manifestation of a trait may have a negative impact: i.e., if a transformational leader manifests transformational characteristics in the extreme, they may be expressed negatively.

The study presents a case study using three of the most publicized Enron senior managers. Enron was a case of CEO and corporate greed, fraud, and malfeasance (Ferrell & Ferrell, 2011). Two Enron CEOs were convicted of fraud and will be compared with another well-known high profile senior manager who tried to stand up to the corruption.


Enron was formed in 1985 and within 15 years emerged as one of the nation’s largest revenue producing companies by buying electricity from generators and selling it to consumers (Palus, Bródka, & Kazienko, 2011). One of the strongest traits of transformational leadership is paying attention to follower needs (Washington, 2013), and Enron was featured on Fortune's popular "100 best companies to work for in America" (Petra & Loukatos, 2009). Further, transformational leadership can influence organizational creativity and innovation both directly and indirectly (Hu, Gu, & Chen, 2013), and Enron was named by Fortune magazine as "America's most innovative company" from 1996 to 2000 (Petra & Loukatos, 2009). Therefore, it can be assumed transformational leadership among its senior management team was manifested at least in some degree.

However, on 22 October 2001 the share price of Enron dropped to $20.65, down $5.40 in one day after the announcement by the Securities and Exchange Commission (SEC) that it was investigating several suspicious deals struck by Enron, and characterizing them as "some of the most opaque transactions with insiders ever seen” (Norris, 2001). With a stakeholders deep mistrust in Enron senior managers (Choudhary, 2012), this event marked the beginning of Enron’s downfall and raised many questions. Some argue that Enron, in fact, never possessed a transformational leadership style (Odom & Green, 2003). In contrast, our new perspective is that, in reality, Enron did use transformational leadership. However, as with the manifestation of any trait in an extreme form, over time it resulted in a negative expression. This will be demonstrated drawing on the profiles of three Enron senior executives: Vincent Kaminski, Jeffery Skilling, and Kenneth Lay.

Kaminski was the head of Enron's Research Department and designer of tools for the pricing of commodity options and hedging strategies as well as the fixed price and derivative transactions for natural gas contracts (Boje, Gardner, & Smith, 2006). He was considered an extremely bright and well-respected executive at Enron (Nance & Koerwer, 2004). Kaminski protested against the actions that led to Enron's most notorious accounting fraud of income statement manipulation, which resulted in a reduction in his duties (Nance & Koerwer, 2004). Kaminski was lionized by federal prosecutors and in a series of books for challenging improper top executive business practices, and later departing Wall Street for academia (Zuckerman, 2006).

Skilling joined Enron in 1990, rising to become its president and chief operating officer in 1996 and its Chief Executive Officer (CEO) in February, 2001 (Chandra, 2003). He actively cultivated a culture of pushing limits: "Do it right, do it now and do it better" was his motto. He encouraged employees to be independent, innovative, and aggressive (Sims & Brinkmann, 2003). However, his own precocious achievements were enmeshed in immoral and fraudulent tactics. For instance, Skilling's letters to shareholders stated that Enron had hit a record $1.3 billion in net income, while the actual net income in the audited financials was only $978 million (Brewer, 2007). Skilling resigned as President and CEO of Enron and sold shares of his company stock totaling $66 million dollars. Only two months later, Enron restated earnings, stock prices dropped, and the company froze shares in an attempt to help stabilize the company (Ibid.). He was later convicted of 18 counts of fraud and conspiracy and one count of insider trading, but was acquitted on 9 other counts of insider trading (Hueston, 2007).

Outstanding leaders are characterized by their substantial impact on organizations and the broader society (Gardner, 1993). However, although outstanding leaders are powerful influencers, their influence is not necessarily positive and can be exploited for good or bad outcomes (Beyer, 1999; Yukl, 1999). In an study of directionality of influence and impact of outstanding leaders on society, House and Howell (1992) distinguished socialized from personalized leaders. Socialized leaders work towards enhancing others and the broader social system by training subordinates to acquire capabilities that transcend the leader. Personalized leaders, conversely, tend to be self-aggrandizing and seek to enhance their power and control regardless of the cost to others and the broader social system (Ibid), representing a dark side of charismatic leadership (Conger, 1989). Many Enron examples exist demonstrating the executive craving for power, specifically by Skilling (cf. Wheat, 2002). Personalized leaders are motivated by narcissistic deprivation whereas socialized leaders are motivated by other psychological resources such as social dominance and a sense of responsibility (Popper, 2005). Skilling communicated his priorities to his employees that profits are the main company goal and should be achieved at all costs (Sims & Brinkmann, 2003). It is, therefore, easy to see in Enron a personalized style of leadership and the extremes to which core executives went to pursue it.

Since, as expressed above, leaders may manifest transformational leadership qualities differently and to varying extent, and an extreme manifestation of a trait may have a negative impact, it can be assumed that Skilling’s transformational trait manifested in a more negative extreme than Kaminski’s. Thus the following hypothesis:

H1. Skilling’s transformational expressions are higher than Kaminski’s.

Kenneth Lay was Enron’s Chief Executive Officer (CEO) and founder (Giwa & Aghimien, 2014). After Skilling resigned from his post, Lay returned as CEO, promising that there were no "accounting issues, trading issues, or reserve issues" at Enron (Sims & Brinkmann, 2003). Less than 4 months before Enron filed for bankruptcy, Lay remained highly positive about the company’s financial health, with statements such as ‘‘the balance sheet is strong,’’ ‘‘the third quarter is looking great,’’ and ‘‘Enron's stock is an incredible bargain at current prices’’ (Ferrell & Ferrell, 2011). Lay was vilified by federal prosecutors and the media as the key Enron executive behind the massive fraud, but, in reality, was convicted of lack of proactivity, preventive caution, and awareness of internal corporate behaviors (Ibid). Lay, who amassed US $152.7mn in payments and stocks (Ukey, 2013), was found guilty of six counts of fraud and conspiracy in the jury trial and four counts of bank fraud in a separate bench trial (Hueston, 2007). Lay died on July 5, 2006 soon after he was convicted (McLean & Elkind, 2006).As with Skilling, we argue that Lay’s transformational trait manifested to a more negative extreme than Kaminski’s. Thus, the following hypothesis:

H2. Lay’s transformational expressions are higher than Kaminski’s.

Further, by taking Skilling and Lay as a group, it may be assumed that they will show a higher mean of transformational expression than Kaminski. Thus the following hypothesis:

H3. Skilling and Lay’s transformational expressions together as a group are higher than Kaminski’s.

Finally, based on the above, Kaminski should display a steadier rate of TL expression, in contrast to a more volatile rate from the two other senior executives. Therefore:

H4a. Deviation in Lay’s transformational expressions from the mean will be higher than Kaminski’s.

H4b. Deviation in Skilling’s transformational expressions from the mean will be higher than Kaminski’s.


In May 2002, the Federal Energy Regulatory Commission (FERC) publicly released a corpus of actual emails by Enron employees. The FERC took this unusual step in order to improve the public understanding of the various reasons for their investigation of Enron. Some emails specific to certain individuals were removed for privacy and legal reasons (Diesner, Frantz, & Carley, 2005). This corpus is known as a reliable source of data for research since it contains modern conversational language (Lindsey, Veksler, Grintsvayg, & Gray, 2007). The Enron dataset has been used extensively for research, including data mining, text analysis, and natural language processing (Wilson & Banzhaf, 2009). As with similar studies exploiting the same dataset (cf. Ying & Wu, 2011), a Structured Query Language (SQL)-based processed database is used to query the Enron corpus, which was first processed by Shetty and Adibi (2004). This procedure removed a large number of duplicate emails, computer generated folders, junk data, invalid email addresses and blank messages, importing all email messages into a relational database (Ouchi, 2000). This approach enables query analysis using SQL, which takes advantage of the structures of data stored in tables (Özcan, Chamberlin, Kulkarni, & Michels, 2006) with no dependency on the content or text language (Serón & Bobed, 2016). This methodology then utilizes natural language processing, a technique that converts text data into coded form, for computer-based analysis (Toyabe, 2012). NLP is an established domain of computational linguistics that emerged sixty years ago (Alzaben, 2012), and is currently the most popular technique implemented in systems designed for human interaction (Zhao, 2013). Most current language processing approaches are based on the syntactic representation of text, which relies mainly on word occurrence frequencies (Cambria & White, 2014). The most common method for text analysis using NLP techniques is ‘Bag-of-Words’ (BOW) (Razavi, Matwin, De Koninck, & Amini, 2014), in which documents are represented as a collection of words, regardless of word order (Munezero et al., 2014). Text classification based on a binary form means that each word in the vector of words explored in the text is marked with 1 if the word exists and 0 if otherwise (Alabdulmohsin, Cisse, Gao, & Zhang, 2016; Nii, Miyake, Takahama, Uchinuno, & Sakashita, 2013). However, exploring occurrence frequencies of each word in the text is more common (Delany & Bridge, 2006). This study will demonstrate that for the purpose of this study both techniques may be used to prove the research hypotheses.

Utilizing BOW, all emails sent by Skilling, Lay, and Kaminski are extracted, and the number of TL for each email sent are counted, employing the TL word list provided by Salter, Green, Hodgson & Joyner (2013) for transformational leadership expressions.

Statistical Analysis

Data were analysed using independent t-test and univariate analysis. Statistical significance was set at p ≤ 0.05 for all analyses.


Skilling Vs. Kaminski

In order to ascertain the greater more prevalence of transformational expressions in use by Skilling as compared to Kaminski (H1), a t-test was used to compare the two senior executives (Table 1). Results indicate a statistically significant difference between Skilling and Kaminski using both the binary representation (t = 3.06, df = 72.54, p<0.01) and the frequency count (t = 2.70, df = 72.40, p<0.01). Indeed, the mean of Skilling's transformational expressions using both the binary representation (M= 0.92, SD =1.11) and the frequency count (M= 1.42, SD =2.09) was higher than Kaminski’s binary representation (M= 0.51, SD = 0.9) and frequency count (M= 0.75, SD = 1.61), respectively, although the number of Kaminski’s emails is about sixty times higher than Skilling's.

Table 1:Transformational Expressions of Skilling And Kaminski  
  N Mean Std.Div.
Binary representation      
Skilling 72 0.92 1.11
Kaminski 4367 0.51 0.9
Frequency count      
Skilling 72 1.42 2.09
Kaminski 4367 0.75 1.61

From Table 1 we also observe the deviation from the means (H4b). Using the binary representation, Skilling’s deviation from the mean (1.11) is higher than Kaminski’s (0.9). Similarly, using the frequency count, Skilling’s deviation from the mean (2.09) is higher than Kaminski’s (1.61).

Lay vs. Kaminski

In order to ascertain that transformational expressions are more prevalent for Lay than Kaminski (H2), a t-test was used to compare the two senior executives (Table 2). Results indicate a statistically significant difference between them using both the binary representation (t = 3.76, df = 34.14, p<0.01) and frequency count (t = 3.12, df = 34.31, p<0.01). Indeed, the mean of Lay's transformational expressions using both the binary representation (M= 1.66, SD =1.8) and frequency count (M= 1.89, SD = 2.15) is higher than Kaminski’s binary representation (M= 0.51, SD = 0.9) and frequency count (M= 0.75, SD = 1.61), respectively, although the number of Kaminski’s emails is about sixty times higher than Skilling's.

Table 2: Transformational Expressions Of Lay And Kaminski
  N Mean Std.Div.
Binary representation      
Lay 35 1.66 1.8
Kaminski 4367 0.51 0.9
Frequency count      
Lay 35 1.89 2.15
Kaminski 4367 0.75 1.61

From Table 2 we also observe the deviation from the means (H4b). Using the binary representation, Lay’s deviation from the mean (1.8) is higher than Kaminski’s (0.9). Similarly, using the frequency count, Skilling’s deviation from the mean (2.15) is higher than Kaminski’s (1.61), as displayed in Figure I.

Figure I. Mean differences among the three Enron senior executives.

Skilling & Lay Vs. Kaminski

A univariate analysis was performed in order to examine differences between the three Enron senior executives, and found statistical significance among them, with Kaminski having the lowest mean in both binary representation (M = 0.51) and frequency count (M=0.75). Next, Duncan’s multiple comparison test was carried out in order to examine differences between Skilling and Lay together as one group vs. Kaminski (H3) in the use of transformational keywords. The test indicated statistical significance between the two groups, with Kaminski having the lower mean.


All hypotheses were supported. Skilling’s transformational leadership expressions are higher than Kaminski’s, as expected. Skilling's aggressive email behavior towards critics who dared confront him places him in the category of personalized leaders. After building a promising career, he used immoral and fraudulent tactics which, as noted, later led to numerous counts of fraud and conspiracy (Hueston, 2007). He actively cultivating a culture pushing limits and striving for self-improvement, as well as employee independence and innovation (Sims & Brinkmann, 2003), which ultimately negatively manifested in extreme form as fraudulent behaviors.

Similarly, the second hypothesis was clearly supported. Lay’s transformational leadership expressions are higher than Kaminski’s. From a bold beginning of entrepreneurship as company founder, Lay's Enron went on to become one of the nation’s largest corporations. Disastrously, he later turned to fraud and sent false positive statements on company financial health to shareholders (Ferrell & Ferrell, 2011). Lay serves as another example of a negative manifestation resulting from pushing a trait to an extreme while losing moral direction.

The third hypothesis was also definitively supported. Grouping both Enron CEOs together as compared to Kaminski shows, in the former, a higher use of transformational leadership expressions. This demonstrates their mutual directionality or perspective in contrast to Kaminski. Kaminski displayed unusual behavior among Enron senior managers by challenging improper top executive business practices (Zuckerman, 2006). As such, he demonstrated well-respected leadership qualities that can serve as an example.

As noted, manifestation of certain traits may change over time, and this study provides ample evidence that leadership traits are also capable of shifting, depending on the individual. Similar to any trait that is taken to the extreme, the presence of TL can also undergo transformation from a positive directionality, veering off into negative and catastrophic outcomes. Hypothesis H4a and H4b presented a higher deviation from the mean in both CEOs than Kaminski, which implies a higher volatility than the stable line represented by the well-respected Kaminski.

This research offers significant contributions. First, it extends Eckhaus’ (2016) framework by providing evidence of the presence of poles of a given trait. Second, since email is the most common, and comparatively speaking the most efficient means of exchanging written information in today's world (Khan, Mizan, Hasan, & Sprague, 2014), it is easier to implement our analytical technique in any organizational context. By using it, a firm’s decision-makers or auditors can supervise in real time company leaders to ensure they stay on the right track of corporate ethics. Third, this study enhances the research on TL, with the perspective that TL is not necessarily constant and may evolve over time and adapt to circumstances. Forth, Salisbury (2008) argues that “Further research is needed into the development of new methods for using performance objectives to ‘measure’ and track these [knowledge that individuals bring to bear on the problems of the organization] contributions.” Since TL promotes intelligence, rationality, problem solving, and encourages followers to find innovative solutions (Hoffman & Frost, 2006), the findings of this study support the use of the proposed framework to measure and track changes in the contribution of leaders to the organization, in the form of the manifestation of TL qualities.

The research employed BOW, which is the most popular approach for modelling text in sentiment analysis (Singh, Varade, Sanap, & Khairnar, 2016). Future research should extend the proposed technique with other NLP techniques. Statistical analyses were evidently elementary, based on the available data set, hence future studies should extend analysis by examining additional variables from a richer sample extracted from an organization. Several beneficial future directions are proposed as follows.

Leadership Cognition

Cognition can be defined as “all mental activities that are involved in acquisition, processing, storage and retrieval of information” (Mittal, Verma, Jain, Khatter, & Juyal, 2012: 165). Although cognition is a significant force in the leader’s emergence and performance (Connelly et al., 2000), leadership cognition has yet to receive much attention (Mumford, Watts, & Partlow, 2015). Since people’s natural language carries important information about their personalities, social situations, and ongoing emotional and cognitive coping processes (Cohn, Mehl, & Pennebaker, 2004), the extended framework proposed in this study may help to fill the literature gap. For instance, future studies may employ Nerb & Spada's (2001) coherence model of cognition and emotion to measure changes in TL manifestation due to cognitive events. Their model makes predictions on the cognitive evaluation of environmental problems, the development of emotions, and action tendencies. They have investigated ten variables: anger, sadness, damage, human agency, controllability, higher goal, knowledge, boycott, help, and typicality. These variables match ten questions meant to measure the estimated damage that was described in the newspaper report. Future studies may extend this work and measure changes in TL after periods in organizations that can be characterized by each of these variables.

Leader Integrity and Change Management

Building on the foundation of behavioral integrity research, the concept of the leader’s integrity, which includes notions of the leader’s moral or ethical behavior, has lately been shown to significantly interest scholars (Simons, Palanski, & Treviño, 2013). Simons (1999) argues that the maintenance of behavioral integrity is complicated and depends on successful management of change. While successful leaders can motivate followers to implement internal changes (Yang, Tsai, & Liao, 2014), the extent to which employees are willing to be influenced by their managers is a key for effective management and change implementation (Simons, 2002), and depends on how the employee perceives the leaders as capable of keeping promises and implementing values (Ibid). Studies emphasize the positive relationship between TL and change management (McKnight, 2013). However, the results of this study stress that the traits manifested by the TL leader may change their impregnability. Future studies may therefore employ the extended framework proposed in this work to examine the link between TL trait manifestation and the leader's integrity, thus promoting the study of leadership integrity and the voluminous research conducted to examine consequences to followers of the leader's misalignment of words and deeds (Simons, Leroy, Collewaert, & Masschelein, 2015).

Leadership and Individual Differences

Research on individual differences has a strong link to leadership research (Dinh & Lord, 2012). The measurement of individual differences or traits is required for the ability to predict leadership outcomes (Antonakis, Day, & Schyns, 2012). This research was recently revived with studies showing that a high rate of variance in leadership ratings depends on person-level effects (Ibid). Furthermore, Zaccaro (2012) argues that “more research attention needs to be directed toward identifying and categorizing the important parameters of leadership situations and how they influence leadership” (p.722), where the situation influences the leaders and interacts with their attempt to influence their followers (McLaurin, 2006). This study’s extended framework may serve as a valuable tool for future research that will use it for the enhancement of individual differences research and the identification of the influence of situations on leadership.