The Leadership Quarterly 23 (2012) 1104–1133 Contents lists available at SciVerse ScienceDirect The Leadership Quarterly journal homepage: www.elsevier.com/locate/leaqua Quantifying leader lives: What historiometric approaches can tell us☆ Gina Scott Ligon a,⁎, Daniel J. Harris a, Samuel T. Hunter b a b University of Nebraska at Omaha, USA The Pennsylvania State College, USA a r t i c l e i n f o Article history: Received 14 May 2012 Received in revised form 15 September 2012 Accepted 3 October 2012 Available online 14 November 2012 Keywords: Outstanding leadership Life history Historiometric Q-Sort Multivariate analysis a b s t r a c t Historically notable leaders, such as John F. Kennedy, Mohandas Ghandi, and Rupert Murdoch, serve as exemplars in our field of study. Across the domains of military, business, government, ideology, and the arts, these outstanding leaders have markedly impacted the institutions, fields, and broader social structures in which they worked and lived. To understand the unique styles, developmental experiences, and performance contributions of such leaders, a historiometric approach is encouraged. We define this methodological strategy as the study of multiple eminent leaders that requires the translation of historical, qualitative information into quantitative indices of individual differences in leaders. Best practices, limitations, and implications for this technique as an advancement of leadership theory are reviewed. © 2012 Elsevier Inc. All rights reserved. 1. Introduction The biographic accounts of John F. Kennedy, Mohandas Ghandi, and Rupert Murdoch portray leaders, albeit markedly different types of leaders (Chadha, 1999; Perret, 2002; Shawcross, 1997, respectively). Accounts of each of these and other historically notable leaders show that they had disproportionate impact on the institutions and fields in which they worked and the social structures in which they lived (Bass, 1985), and upon close examination it becomes apparent that they did so differentially (Mumford, 2006). Reviewing historical accounts of the early lives and careers of influential leaders also illustrates that each rose to his respective height of power through unique patterns of life events (Ligon, Hunter, & Mumford, 2008). Post hoc examination of their contributions to society gives some indication of who created more or less sustainable institutions, movements, and/or policies in his or her tenure. In short, biographical and historical accounts of these leaders contain numerous objective indicators of psychological, contextual, and performance variables and patterns. Moreover, qualitative methods such as the examination of historical records allow us to identify leader behaviors that might be considered less salient (e.g., problem solving) or difficult to observe via “typical leadership study” methods (Hunter, Bedell-Avers, & Mumford, 2007). For example, a relatively new theory of leadership, the Charismatic, Ideological, and Pragmatic (CIP) model, based on observable differences in leader cognition and sensemaking across three markedly different styles of outstanding leadership, would not have emerged if not for historiometric methodology (Mumford, 2006). In short, transforming qualitative information into quantitative metrics to build and test theoretical frameworks is an exciting methodology for our field as it provides access to leader characteristics otherwise difficult to tap. So why is it that we do not incorporate the study of these objectively-verifiable accounts of such leaders into our empirical paradigms more systematically? The study of multiple outstanding leaders that requires the translation of historical, qualitative ☆ This material is based upon work supported by the U.S. Department of Homeland Security under Grant Award Number 2010-ST-061-RE0001. The views and conclusions contained in this document are those of the authors and should not be interpreted as necessarily representing the official policies, either expressed or implied, of the U.S. Department of Homeland Security. ⁎ Corresponding author. Tel.: +1 402 554 2972. E-mail address: [email protected] (G.S. Ligon). 1048-9843/$ – see front matter © 2012 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.leaqua.2012.10.004 G.S. Ligon et al. / The Leadership Quarterly 23 (2012) 1104–1133 1105 information into quantitative indices of individual differences in leaders, or leader historiometric analysis, is an underutilized approach to the empirical study of leadership for at least three reasons. First, early uses of historiometric techniques drew criticism due to the inference of psychological variables from hard, objective data (Cox, 1926). However, because more recent operationalizations of psychological variables tend to rely upon concrete, objectively-measured behaviors (as opposed to general traits), contemporary historiometric investigations of such psychological variables may be on the upswing (e.g., Deluga, 2001; Winter, 1987). Second, biographic data, often gathered by historians untrained in leadership theory, are typically incomplete. Moreover, although academic biographies tend to focus on accounts of events that can be verified by multiple parties, such as life events (e.g., schools attended) and public career decisions (e.g., speeches given), rivate interactions (e.g., those between a leader and her spouse) are less often reported as they cannot be verified as readily. While who gathers the record has implications for the type of data that can be garnered from a historical biography, it does not negate the utility of the information that can be obtained, especially when a careful content coding scheme is applied by trained raters. Related to this is the third reason historiometric techniques have been underutilized in the field of leadership research: they are labor intensive and complex to execute. To ensure reliable and valid inferences are drawn from the often heterogeneous information found in historical records, a multi-step data analytic strategy is required to exam multiple cases. These steps, while described in pieces across various domains (McKeown & Thomas, 1988; Mumford & Threlfall, 1992; Simonton, 1988a, 1990), have not been conceptually described together in one place with regard to the study of leadership. Despite reasons for the relative lack of historiometric studies conducted, when the research question depends on defining the patterns of individual differences in the behaviors, developmental experiences, and performance of leaders — and particularly outstanding leaders who serve as exemplars for marked influence and performance — we argue that a historiometric approach is the best available methodological framework to address the issues at hand. In short, historiometric approaches offer an empirical method to test leadership theory in a way that other approaches do not. Thus, the primary purpose of this submission is to provide a detailed account of the use of historiometric approaches to empirically test individual differences in patterns of attributes, style, problem solving, and performance among leaders. Prior to this, we first review a commonly employed sample — outstanding leaders — where application of historiometry is particularly well-suited, and we describe typical frameworks used to study it. 2. Outstanding leaders Outstanding leaders are those individuals who have had a marked influence on their organizations and on the broader social system (Bass, 1985; Gardner, 1993). For example, the laws that govern how we work today can be credited to the Roosevelts, both Teddy and Franklin, through their influence over the passage of fair labor laws. Two important characteristics differentiate outstanding leaders from more normative, day-to-day leaders. First, outstanding leadership is a rare event. Because a central tenet of how we define an outstanding leader lies within having disproportionate impact on the way we live, only few individuals can rise to positions where the expression of such a vision for leadership influence can be realized. Second, these leaders, and the contextual influences around them (e.g., close followers, performance outcomes), are extensively chronicled — by multiple types of sources. What is important to note is that while seemingly positive in connotation, outstanding leaders can exert their disproportionate influence through ethical or unethical means, a distinction that has been described as personalized versus socialized leadership (House & Howell, 1992). Multiple sources exist detailing conditions and characteristics that give rise to both of these types of leadership styles. Implications of these points are clear: outstanding leadership, both in positive and destructive forms, is a low base-rate phenomenon that yields insight into how to achieve large scale outcomes (good and bad). This does not negate the applicability of such lessons to the “rest of us,” it just means that they stand as exemplars for what leadership can be — which is highly influential and impactful. 2.1. Typical approaches used in studying outstanding leadership Most research on outstanding leadership has been conducted through four traditional approaches: 1) dimension validation via normalcy, 2) simulation, 3) case study, and 4) quantitative collective case analysis. The first approach, dimension validation via normalcy, is essentially the validation of outstanding leadership qualities through generalization to normal leader samples. Put differently, the qualities of outstanding leaders are compared against those of normal leaders, and conclusions are drawn regarding the collective difference between outstanding leaders and normal, non-outstanding leaders (for example, see Avolio & Bass, 1999). The benefit of using this approach to study outstanding leadership is that it helps differentiate outstanding leaders from more normative leaders. However, because the foundation of this approach lies in comparisons between groups (i.e., normal versus outstanding leaders), a drawback to this approach is that it does not allow within group comparisons; researchers cannot differentiate between individual outstanding leaders. Another drawback to this approach is that the contexts in which normal leaders typically conduct their affairs may not generalize to those contexts in which outstanding leaders conduct theirs, such as leading a small office division versus leading a country. As leadership, like most psychological variables, operates on a continuum, it makes sense that we study extreme cases who stand outside most leadership analyses in addition to those in more typical methodological approaches (McKelvey et al., 1999). The second typical approach of studying outstanding leadership is to conduct simulations. Those simulations are often achieved by giving participants various scripts based on leader behavior, the purpose of which is generally to determine how followers respond to scripts that exemplify some outstanding leadership style. Likewise, another type of simulation is to have participants take the roles of outstanding leaders. For an example of the latter type, see Hunter, Bedell-Avers, and Mumford (2009). A benefit to using simulations is that, because of their highly-structured nature, they are appropriate for understanding discrete cognitive processes. However, because of that high amount of structure and the common use of student samples, the 1106 G.S. Ligon et al. / The Leadership Quarterly 23 (2012) 1104–1133 results of such studies are often difficult to generalize to workforce samples and the contexts of a normal organization, which are generally not highly structured and are more ambiguous than situations presented in simulations. The third approach, the case study, requires analysis of a single case in significant detail, often held against a particular leadership theory. A case study by Neal and Tansey (2010), for example, thoroughly explored the leadership of Rafik Hariri, the former Lebanese Prime Minister. A benefit to using the case study approach is the level of description that can be gathered about one case, often from multiple sources. Such description is important to better understand those cases, and how they offer evidence for a particular leadership theory, at great depth. In addition, there is much research in organizational behavior (Gibbert, Ruigrok, & Wicki, 2008; Kempster, 2006, 2009) and political psychology literature (Kaarbo & Beasley, 1999) on how to conduct such case analyses systematically. However, in these fields such case research is typically qualitative, making it difficult to statistically compare case studies to one another and within a larger theoretical framework. Also, because only one case is generally examined, the conclusions drawn from the analyses are often difficult to generalize to other cases or be applied to sets of cases. The fourth and final typical approach to studying outstanding leadership is the quantitative collective case study. This approach builds on the case study approach in two important ways. First, quantitative case analysis typically utilizes multiple cases instead of just one (Jauch, Osborn, & Martin, 1980). Second, and more importantly, a coding scheme is applied to the qualitative data to obtain quantitative indices of leader behavior. A benefit of this approach is that it takes into account the context in which each individual leader conducted his or her affairs and, because of the depth of the investigations, the data are often quite rich. However, the conclusions and analyses can only be as good as the data afford, and rigorous controls need to be applied to the quantitative analyses to minimize the possibility of certain biases. The historiometric approach represents a special case of this final approach. Specifically, historiometric research is the application of a content coding scheme to qualitative data across multiple cases, with the further implication that such qualitative data are historically notable in nature. 3. Historiometric approach Historiometric methods date back to some of the oldest analytic techniques in social science research (Woods, 1909, 1911), but it has been refined over the years by Dean Simonton's social psychology research program and Mike Mumford's outstanding leadership research program. This method relies upon available and verifiable historic records (Mumford, 2006; Simonton, 1990, 1998a, 2003). His and others' (Tetlock, 1979; Winter, 1987) work have at least three implications for leadership research. First, this type of data comes in many forms, such as financial documents, school transcripts, and employment history, lending credibility to the reliability and validity of the information it yields. Raw data used in these studies does not come from laboratories or other “live human participant protocols;” instead, data used in these investigations come from historical records of people and situations of notable importance. This last piece is important for studies of outstanding leadership: in order to obtain multiple, historically verifiable markers of behavior and performance, the cases under examination need to have left enough of a “mark” on history that trained historians — and multiple historians — took time to investigate and write about them. The people publishing these works are typically not trained leadership or organizational behavior researchers. Instead, most of the sources used in historiometric research come from academic historians and political scientists. This is related to the second characteristic of historiometric research — it requires quantitative analysis of the qualitative information provided in such historical documents. A multi-step data analytic technique is required for historiometry. First, ratings must be obtained from the qualitative data at hand. Because historical records are not prepared by leadership scholars in most cases, a clear set of variables, time frame, and type of information must be defined a priori (Mumford, 2006; Tetlock, 1979). Second, the scores yielded from these ratings are subjected to rigorous quasi-experimental and advanced multivariate analysis. While the data are correlational in nature, careful selection of sophisticated analytic techniques can afford more certainty around obtained inferences about relationships (Simonton, 1990). Third, and perhaps most important, the goal of historiometric research is to test theory (Mumford, 2006; Simonton, 2003). Whereas quantitative single-case analysis is intended to develop theory based on detailed examination of a single case, the multiple observations afforded in a well-designed historiometric study allow us to investigate highly complex theories of individual differences among outstanding leaders. The aim of the next section is to illustrate the type of information relevant to outstanding leadership theories that can be derived from a historiometric approach. 3.1. Leadership data available from historiometric approaches There are at least four types of leadership data available through historiometry: 1) situational influences/conditions, 2) behaviors, 3) developmental events, and 4) sustainable performance. First, rich contextual data can be used to examine boundary conditions for a given leadership style's effectiveness (Eisenhardt, 1989; Weber, 1926). For example, Tetlock (1979, 1981, 1983). Coding critical events of hundreds of leaders solving international crises, he found low cognitive complexity is associated with taking extremist positions and high cognitive complexity is associated with more moderate approaches to conflict resolution. In a recent study of terrorist organizations, Asal and Rethmeyer (2011) found that only leaders who sought to make alliances across ideological and organizational lines proved to have most threat for terrorism to the United States. By assessing individual differences embedded in the culture and organization at large, leadership scholars are better able to draw inferences about generality and utility of our theories. G.S. Ligon et al. / The Leadership Quarterly 23 (2012) 1104–1133 1107 Second, and as previously discussed, historiometric research yields data on behavioral markers of individual differences. While Cox (1926) and Watson's (1927) original work on intelligence was the pioneering data in this area, recent approaches by Simonton (1983, 1991, 2006) have demonstrated that biographic summaries of eminent leaders can differentiate leaders on rated intelligence and problem solving style. In addition to intelligence, a comprehensive approach in differentiating 120 notable charismatic, ideological, and pragmatic (CIP model) leaders on aspects of the CIP model, Mumford, Bedell, Hunter, Espejo, and Boatman (2006); Mumford, Gaddis, Licuanan, Ersland, and Siekel (2006); Mumford, Gaddis, Strange and Scott (2006); Mumford, Scott, and Hunter (2006); Mumford, Strange, Gaddis, Licuanan, and Scott (2006) extracted behaviors indicative of vision formation, political skill, propensity for violence, creative problem solving, and leader member exchange relationships using historiometric techniques. Other researchers have applied this technique to measure behavioral markers of narcissism (Deluga, 1997, 1998, 2001), Machiavellianism (Bedell, Hunter, Angie, & Vert, 2006), and networking (Bedell-Avers, Hunter, Angie, Eubanks, & Mumford, 2009) in outstanding leaders. Third, historiometric approaches provide ample data about developmental experiences and seminal events as related to leadership style. As the importance of experiential development has taken a front seat in contemporary leadership research (Day, 2000; Ligon & Hunter, 2010; McCall, 2010), examining how leaders develop over time seems like a logical phenomenon to study. In other domains, this adult development literature is abundant. For example, in examining the life histories of eminent creative scientists and artists, Simonton (1975, 1988b). In an examination of the obituaries of leading scientists, Mumford et al. (2005) identified important career events such as early exposure to the field and supervised practice as critical indicators of late-life innovation. Finally, in a review of early career events, clear differences arose between the patterns of development between charismatic, ideological, and pragmatic leaders (Ligon et al., 2008). Table 1 provides an example of the type of developmental event information yielded from historiometric approaches. Finally, and perhaps most important, historiometric approaches allow us to examine the sustainability of performance (Bedell-Avers et al., 2009; Mumford, 2006). Because the approach itself is rooted in the historical significance of these individuals, examining the lasting (and sometimes non-lasting) effects of their decisions provides us with a time-tested gauge of performance. Two examples are relevant here. First, upon examination of the most popular selling bands in 1992, the group Boys II Men clearly looks to be the front-runner, breaking Elvis' record for number of weeks at #1 on the Billboard Top 100 Chart for their song, “End of the Road.” Pearl Jam, another band also at their arguable “height of power” in 1992, however, continues to produce solid albums and perform to sell-out crowds some twenty years later. Archival analysis allows us to determine whose performance was indeed more sustainable in the long-term, albeit slightly less “popular” if performance were measured at only one juncture (Pearl Jam was on the Billboard Top 100 Chart in 1992 for two weeks, but never at #1 spot). Relevant to leadership, temporal stability of criteria also come into play. In a series of studies examining Presidential influence, Simonton (1986a, 1986b, 1987) and Winter (1987) found that the performance criteria of a) contemporary popularity versus b) long-term success was differentially predicted by leadership characteristics. Thus, it is unlikely that studying more normative or run-of-the-mill leaders via contemporary empirical methods (e.g., lab studies, survey research) will afford generalization to outstanding leader samples. However, we can learn much about leadership as a phenomenon from studying cases of extreme leadership, or the leadership that emerges in complex, ill-defined situations and domains (Bass, 1985). Moreover, there are processes — such as communication strategies, sensemaking, and problem solving — that may generalize to all types of leader situations if adapted to the context at hand. For example, a high school football coach can readily use the same techniques deployed by NCAA legends to motivate his team and coaching staff through ideological sensegiving strategies (Hunter, Cushenbery, Thoroughgood, & Ligon, 2011). In terms of understanding formative influences, as these leaders are deemed “outstanding” given their disproportionate influence, seeing what objective career events (e.g., mentoring) are related to subsequent performance may yield insight for leader development in a host of domains. What is Table 1 Example information found in historiometric approaches. Early life events Idi Amin Bill Belichick Osama bin Laden “In 1940 Amin went to live with a maternal uncle in Bombo and tried to register for formal schooling, but he was rejected because Nubians could not then get admission to schools. He is said to have taken part and been injured in the Nubian anti-discrimination riots in the same year. In 1941 he was accepted by the Garaya Islamic School in Bombo and attended until the end of 1944. At Gayara he showed that his was a mind of potential, excelling in and winning honours for his skills in Qur'anic recitation in 1943. After leaving he got a job as door attendant and hat and coat checker at the Imperial Hotel in Kampala. It was here that he impressed a British officer who recruited him into the army” (Moghal, 2010, p. 9). “The man behind the dynasty is himself the son of a football coach. Bill's father was head football coach at the Naval Academy, and young Bill grew up idolizing Navy players such as Roger Staubach, the 1964 Heisman trophy winner, who, after his naval service, won the Super Bowls for the Dallas Cowboys. Bill's mother said the coaches did not mind having young Bill around. ‘He wasn't a best. He was very good at listening and learning and remembering.’ Bill spent a year in prep school and then attended Wesleyan in Middletown, Connecticut. As reserve defensive end, he had questions: ‘What if the offense did that?’ Or ‘Whose responsibility is that?’ He served as team co-captain of the lacrosse team. What he really gained in college was an approach to learning that was both strategic and tactical. ‘You learn how to think and solve problems,’ he says. ‘Here's the problem; you go figure it out.’ He was an economics major.” (Baldoni, 2005, p. 112) “In his youth, Osama regularly met prominent Islamists in social and religious settings arranged by his father and — after Muhammad's death — his older brothers. These events, one observer noted, ‘started forming an Islamic responsibility in him at an early age.’” (Scheuer, 2011, p. 24) “He also accompanied Allia and his step siblings on summer holidays in Syria for his first seventeen years, experiencing a more cosmopolitan atmosphere than that in the Kingdom. Thus in watching his mother and traveling with her during the summers, Osama was further acquainted — as he had been at his father's halaqat — with the diversity of the Muslim world, learning how to get along with devout Muslims who did not embrace the Kingdom's puritanism.” (Scheuer, 2011, p. 27) 1108 Table 2 Historiometric studies involving leadership. Leader(s)/type of leader(s) Research question Conclusion Data source(s) Main constructs of assessment Construct codings Statistical analyses Bass and Farrow (1977) Political figures How are leaders' actions influenced by situational factors? Different leadership styles are used depending on situational factors. – Biographies – Various documents – Ratings – Judgments - Scale reliability – Correlation – Factor analysis Bass, Avolio, and Goodheim (1987) World-class leaders Can biographies be used to differentiate leadership styles of well-known leaders? The use of biographies can be a valid and reliable way to measure leadership qualities. – Biographies – System inputs of the organization, work group, task, attitudes, leader styles, and within system relations – System outputs – Transformational leadership style – Transactional leadership style – Ratings Bedell et al. (2006) Outstanding leaders in the 20th century Different leadership styles and orientations differ in levels of Machiavellianism, and that Machiavellianism is negatively related to leadership performance. – Highly structured academic biographies – Pathways to Outstanding Leadership – Machiavellianism – Leadership performance – Leadership style – Leadership orientation – Various controls – Ratings – Objective indicators (only for the controls) Cell (1974) Various heads of state How do leaders with different outstanding leadership styles (i.e., charismatic, ideological, or pragmatic) and orientations (i.e., personalized or socialized) differ in Machiavellianism? Do situational factors influence the charisma of heads of state? – MANOVA – ANOVA – Interrater agreement – Correlation – Fisher's R to Z transformation – Validity – Reliability – Correlation – Regression – Biographies – Interviews from country nationals – Charisma – National social crisis – Nationalism plus nationalistic movements – Disrupted youth – Denial of access – Prepower following – Judgments – – – – – DeChurch et al. (2011) Leadership during natural disaster responses In what ways are the effects of leadership important in mission critical multiteam environments (e.g., responses to natural disasters)? Certain situational factors, namely nationalism plus nationalistic movements, denial of access, and disrupted youth, are collectively strong indicators of a leaders' charisma. Functional leaders positively influence several aspects associated with multiteam processes, emergent states, and performance. – Critical incidents – Leadership–relevant incidents – Leadership strategy – Leadership coordinating – Ratings – Judgments (regarding incidents) – Interrater agreement Deluga (1997) U.S. presidents How does narcissism relate to charismatic leadership and rated Narcissism was largely positively related to presidential – Center for Army's Lessons Learned – Defense Technical Information Center archives – Various newspapers and books – Various documents available through multiple types of sources (e.g., blogs and communities of practice) – Profiles developed by Simonton (1986a, 1988a) that were derived from – Narcissism – Presidential charismatic leadership – Ratings – Rankings – Interrater agreement – Regression Correlation Regression Partial correlation Path analysis Validity (scale) G.S. Ligon et al. / The Leadership Quarterly 23 (2012) 1104–1133 Reference Reference Leader(s)/type of leader(s) Research question Conclusion Data source(s) Main constructs of assessment Construct codings performance in U.S. presidents? charismatic leadership and rated performance. biographical sources and presidential fact books – DeGregio's (1991) presidential reference work – Same sources as used in Deluga (1997) – Rated presidential performance U.S. presidents How does proactivity relate to charismatic leadership and rated performance in U.S. presidents? Presidential proactivity positively related to presidential charismatic leadership and rated performance. Deluga (2001) U.S. presidents How does a president's Machiavellianism relate to charismatic leadership and rated performance? Presidential Machiavellianism was positively related to charismatic leadership and rated performance. – Profiles developed by Simonton (1986a, 1988a) and Deluga (1997, 1998) – Previous studies Emrich, Brower, Feldman, and Garland (2001) U.S. presidents How does the use of images in presidential speeches influence perceptions of charisma and greatness? Presidents who used more image-based rhetoric were perceived as being greater and more charismatic. – Presidential inaugural addresses – Pivotal presidential speeches – Various sources Etheredge (1978) Various U.S. politicians Does personality relate to a leaders' foreign policy decisions? – Scholarly works – Insiders' accounts – Biographies – Autobiographies Eubanks et al. (2010) Outstanding political leaders of the 20th century With regard to criticism, how do leader response strategies influence follower reactions and resolution of the criticism? Fiol, Harris, and House (1999) U.S. presidents How do charismatic leaders generate social change? Certain personality traits (e.g. introversion/ extroversion) seem to relate to how an American leader regards foreign policy issues. Collaborative and confrontational response strategies seem to be most effective for gathering follower support and continuing with an agenda, whereas response strategies that are avoidant, divert attention, or attempt to be persuasive are less effective. Charismatic leaders seem to use consistent communication strategies for – Proactivity – Presidential charismatic leadership – Rated presidential performance – Various control variables – Machiavellianism – Presidential charismatic leadership – Presidential rated performance – Control variables (friendliness, need for affiliation, proactivity, and need for power) – Image-based rhetoric – Concept-based rhetoric – Public charisma and creativity (based on Simonton, 1988a) – Greatness (based on a previous study) – Dominance over subordinates – Extroversion – Policy disagreements – Personality disagreements – Ratings – Rankings – Objective indicators – Interrater agreement – Regression – Ratings – Interrater agreement – Regression – Ratings – Judgments – Correlation – Regression – ANOVA – Ratings – Judgments – Correlation – Counts and percentages – Highly structured academic biographies – Pathways to Outstanding Leadership – Hostile criticism – Leader appraisal of criticism – Leader response strategy – Others' reactions to leader's response – Resolution of criticism – Various controls – Ratings – Objective indicators (only for controls) – Interrater agreement – Correlation –Regression – Exploratory factor analyses – Various presidential speeches (mostly inaugural addresses or addresses to congress) – Communication acts – Charisma – Judgments – Interrater reliability – ANOVA – Least squares means test 1109 (continued on next page) G.S. Ligon et al. / The Leadership Quarterly 23 (2012) 1104–1133 Deluga (1998) Statistical analyses 1110 Table 2 (continued) Reference Leader(s)/type of leader(s) Research question Conclusion generating social change. Coach succession, and at varying times (e.g., during season and off-season), seems to influence team performance more than owner succession. NBA coaches and team owners Does coach succession influence team performance differently than owner succession? Holmes and Elder (1989) U.S. presidents Which indicators differentiate the best and worst U.S. presidents? House, Spangler, and Woycke (1991) U.S. presidents Does a president's personality and charisma influence his effectiveness? Hunter et al. (2011) Outstanding college and NFL coaches Kenney and Rice (1988) U.S. presidents Do the leadership styles of charismatic, ideological, or pragmatic leadership influence how coaches lead their teams? Which contextual factors best predict presidential greatness? There is evidence that a coach's leadership style affects how that coach leads the team and guides the team members. A variety of contextual factors (e.g., whether the president is assassinated, served during wartime) significantly predict presidential greatness. Kynerd (1971) U.S. presidents How do five presidential rankings compare to one another? Of the five sets of rankings discussed, they do not seem to be statistically different from one another, but rankings may be Various indicators (e.g., publishing books before assuming office, being more assertive on foreign policy issues) significantly differentiate the best and worst U.S. presidents. A president's personality and charisma do indeed influence presidential performance, although different traits are stronger indicators. Main constructs of assessment Construct codings Statistical analyses Unidentified – Coach succession – Owner succession – Team performance (winning percentage) – Controls (prior performance, coaching ability, coaching experience) – Countless indicators – Objective indicators – Correlation – Time series modeling – Regression – Chi-square test for equality – Ratings, judgments, and objective indicators from previous studies – t-test – Fisher's exact test for independence – Institutional age – Motives – Activity inhibition – Crises – Editorial charisma – Behavioral charisma – Presidential performance – Ratings – Judgments – Objective indicators – Factor analysis – Discriminant validity – Regression – Partial least squares technique – Leadership type/style – Leadership orientation – Various traits – Various controls – Ratings – Judgments – Objective indicators (only for controls) – Interrater agreement – Correlation – MANCOVA – Discriminant function analysis – Presidential greatness – Popular vote victories – Term(s) served – Assassination – War – Economy – Success or failure of passing legislature – Scandal – Rankings of presidential greatness – Rankings – Regression – Rankings (based on previous studies) – Correlation – z-score – A variety of sources related to presidential studies – Data from Winter (1987) – Letters and speeches written by presidents – Letters and speeches in autobiographies – Memoirs and diaries written by presidents – Biographies – Editorials from New York Times – Historical biographies – Chicago Tribune – Various sources – Previous studies that ranked presidents on greatness G.S. Ligon et al. / The Leadership Quarterly 23 (2012) 1104–1133 Giambatista (2004) Data source(s) Reference Leader(s)/type of leader(s) Research question Conclusion Data source(s) influenced by raters' political bias. Certain life events are better indicators of particular leader types or orientation than others. Outstanding leaders of the 20th century Do certain developmental events differentially predict leadership type and orientation? Lilienfeld et al. (2012) U.S. presidents How do certain psychopathic personality traits relate to the U.S. presidency? Marmor-Lavie and Weimann (2005) Israeli politicians Can a reliable and valid measure be developed regarding the use of emotional appeals in modern political campaigns? McCann (1992) U.S. presidents Which formula best predicts presidential greatness? McCann (2005) Deceased U.S. presidents How does the relationship between conscientiousness and death age apply to U.S. presidents? Boldness, via fearless dominance, positively related to several factors of presidency (e.g., performance, persuasiveness), whereas impulsive antisociality was related to factors of negative job performance. A reliable and valid measure was developed, with results suggesting that emotional appeals differ depending on certain contextual factors, but also the party that a politician is affiliated with. A 6-variable formula (including 5 personological variables and one zeitgeist variable) predicted a large amount of variance (91%) in presidential greatness. Conscientiousness positively correlated with death age and exercise, and negatively correlated with smoking and drinking. The other Big Five factors were not correlated with death age, exercise, smoking, or drinking. Construct codings Statistical analyses – General history texts – Biographical websites – Academic biographies – Originating event – Turning point event – Anchoring event – Analogous event – Redemption event – Contaminating event – Various thematic constructs – Ratings – Judgments – Objective indicators (only for controls) – Rater expertise – Previous studies about U.S. presidents (e.g., Simonton, 1986a) – Polls – Countless indicators – Ratings – Objective indicators – Televised political spots – Emotional arousal (for various emotions) – Emotional appeal to hope, pride, sympathy, warm-heartedness, anger, and fear – Ratings – Objective indicators – Mean – Median – Mode – Validity – Interrater reliability – t-test – MANOVA – Previous research on presidential greatness – Presidential greatness – Various personological dimensions – Various situational variables – Zeitgeist – Ratings (from previous research) – Objective indicators – Correlation – Regression – Big Five ratings – Health practice assessments – Death age – Inauguration age – Birth year – Big Five personality variables – Health practices – Ratings – Judgments – Objective indicators – Correlation – Regression – Logistic regression – Kappa agreement – Q-sort – Interrater agreement – Chi-square frequency analysis – Correlation – MANCOVA – Discriminant function analysis – Regression – Interrater reliability – Correlation – General linear modeling – Incremental validity 1111 (continued on next page) G.S. Ligon et al. / The Leadership Quarterly 23 (2012) 1104–1133 Ligon et al. (2008) Main constructs of assessment 1112 Table 2 (continued) Leader(s)/type of leader(s) Research question Conclusion Data source(s) Main constructs of assessment Construct codings Statistical analyses Miller and Stiles (1986) U.S. presidential candidates and U.S. presidents Familiarity has increased from 1920 to 1981, and familiarity also increases from acceptance speech to inaugural address. – Coding per verbal response mode – Intercoder reliability – t-test – Correlation Outstanding violent and non-violent leaders of the 20th century – Transcripts of U.S. presidential nomination acceptance speeches and inaugural addresses from a variety of sources – Preliminary sources included historical summaries, biographical listings, and web sites – Biographies – Familiarity Mumford et al. (2007) Does level of verbal familiarity differ across time with respect to presidential nomination acceptance speeches and inaugural addresses? Are ideological leaders more likely to incite violence than non-ideological leaders? – Ratings – Judgments – Objective indicators – Interrater agreement – Correlation – Regression – Discriminant analysis Mumford (2006) Charismatic, ideological, and pragmatic leaders What theoretical dimensions differentiate charismatic, ideological, and pragmatic leaders? Charismatic, ideological, and pragmatic leaders differ across a variety of theoretical dimensions. – Histories – Biographies – Encyclopedias – Ratings – Objective indicators – Q-sort – Discriminant analysis – Regression Murray (1928) Great orators (including religious leaders) Can traits present in early life, especially genius, influence whether one will become a great orator? – The Early Mental Traits of Three Hundred Geniuses – Ratings – N/A O'Connor, Mumford, Clifton, Gessner, and Connelly (1995) Charismatic leaders What personality characteristics best differentiate between personalized and socialized charismatic leaders with regard to their destructiveness? Traits early in life, particularly genius and perseverance, can influence one's status later in life as a great orator. Those traits should be promoted in schools so as to create great orators. The studied constructs of interest seem to influence the amount of destructiveness a charismatic leader engages in. – Leadership type – Violence/non-violence criteria – Performance criteria – Several predictor variables at the individual, group, organizational, and environmental levels – Various controls – Problem solving – Communication – LMX relationships – Developmental influences – Political tactics – Performance – Various personality traits – IQ – General history texts – Almanacs – Biographical listings – Encyclopedias – Historical, scholarly biographies – Outcome uncertainty – Need for power – Object beliefs – Negative life themes – Narcissism – Fear – Self-regulation – Ratings Panagopoulos (2004) U.S. congressional candidates Gender does influence campaign communications strategies, but not in very sizeable ways. – Database of political advertising collected by Campaign Media Analysis Group – Tone of ad – Style of ad – Policy content of ad – Various factors – Judgments Rubenzer, Faschingbauer, and Ones (2000) U.S. presidents Are different campaign communications strategies used between male and female congressional candidates? Can standardized personality tests be used in historical research? – Interrater agreement – Correlation – MANOVA discriminate function analysis – t-test – Cross-validation – ANOVA – Causal analysis – Means – t-test – Multivariate probit analysis By using the NEO PI-R and presidential experts, results support the notion – Rater expertise (raters were presidential experts) – Big Five personality factors – Positive and negative valence – Ratings The attributes associated with ideological leadership influenced the amount and occurrence of institutional and cultural violence. – Modified interrater reliability – Q-sort (including Q analysis and G.S. Ligon et al. / The Leadership Quarterly 23 (2012) 1104–1133 Reference Reference Leader(s)/type of leader(s) Research question Conclusion Main constructs of assessment Construct codings that standardized can be reliably used in psychohistorical research. – Previous studies of presidential greatness – Job performance – Intelligence factors – Behavior as president – Other variables – Presidential success – Eminence – Father's status – Intelligence – Education – Major vocation – Life span – Vocational choice – Birth year – Military success – Casualties – Various individual factors – Various situational factors – Various interaction terms – Leadership ratings (based on a previous study) – Transition into presidency – Various administration events – Various biographical predictors – Intelligence – Morality – Leadership – Eminence – Life span – Reign span Simonton (1976) Eminent creators and leaders Were the analyses conducted by Cox (1926) susceptible to certain artifacts? The analyses conducted by Cox (1926) might have been influenced by data reliability and a time-wise sampling bias. – Data gathered and analyzed by Cox (1926) Simonton (1980) Military leaders of land battles Which factors predict military success during land battles? A military leader's years of experience, winning streaks, willingness to take the offensive, and divided command are significant determinants of a victory. – Military encyclopedia – Military dictionaries Simonton (1981) U.S. presidents Various determinants and biographical variables influence presidential performance and greatness. – Several archival sources of U.S. presidents Simonton (1983) Hereditary monarchs of European nations (circa 900–1800 A.D.) What are the principal determinants of the rated greatness of American presidents, and can presidential performance be predicted using preelection biographical variables? Which intergenerational factors influence monarchs the most, as well as which individual differences are affected by those factors? – General reference works – Biographical dictionaries – Historical encyclopedias – National histories Simonton (1984) European hereditary monarchs To what degree do situational factors, as opposed to individual factors, influence ruler eminence? Simonton (1985) Political leaders (U.S. presidents, vice presidents, and congressmen) Is the vice-presidential effect influenced more by individual factors or situational factors? Intergenerational transfer of traits is influenced by genetic, rolemodeling, cohort, and sociocultural effects, suggesting that no one process is necessarily more impactful than another. Although leader eminence is strongly influenced by the number of significant events during leadership, individual factors still play a large role. The vice-presidential effect seems to be more strongly influenced by situational factors. Statistical analyses Q-factor analysis) – Averages and standard deviations – Correlations – Judgments – Objective indicators – Correlation – Regression – Judgments – Objective indicators – Correlation – Regression – Discriminant analysis – Ratings – Judgments – Objective indicators – Correlation – Regression – Ratings – Objective indicators – Correlation – Regression – Factor Analysis – Same constructs as Simonton (1983) – Significant events – Ratings – Objective indicators – Correlation – Partial correlation – Regression – Various reference works – Almanacs – Statistical books – Presidential Vetoes, 1789–1971 – Countless variables – Judgments – Objective indicators – Correlation – Discriminant analysis – Time-series analysis (continued on next page) 1113 – Same sources as Simonton (1983) G.S. Ligon et al. / The Leadership Quarterly 23 (2012) 1104–1133 Data source(s) Leader(s)/type of leader(s) Research question Conclusion Data source(s) Main constructs of assessment Construct codings Statistical analyses Simonton (1986a) U.S. presidents How do presidents vary in personality based on the Gough Adjective Check List? Intellectual Brilliance was the only factor that significantly predicted presidential greatness. – Personality attributes (per the Gough Adjective Check List) – Ratings – Objective indicators – Correlation – Reliability – Factor analysis – Cluster analysis Simonton (1986b) U.S. presidents When given limited information about leaders (e.g., U.S. presidents), do naïve raters differ markedly from historians? – Semantic components of leader schema (strength, activity, and goodness) – Greatness – Various historiometric predictors – Judgment – Ratings – Objective indicators – Interrater reliability – Correlation – t-test – Regression Simonton (1987) U.S. presidents How does a president's inflexibility interact with situational factors to influence veto behavior? – Presidential Vetoes, 1789–1971 – Previous research – Variables about vetoes – Political variables – Inflexibility – Dogmatism – Ratings – Judgments – Objective indicators – Correlation – Time-series regression analysis Simonton (1988a) U.S. presidents Can the presidents be differentiated based on attributes derived from biological descriptors, do those attributes belong to factors, and do those factors relate to other variables? – Biographical sources – Several previous studies – Presidential style – Various personality traits, biographical experiences, and performance criteria – Ratings – Objective indicators – Correlation – Regression – Alpha reliability – Factor analysis – Cluster analysis Simonton (1988c) Eminent creators, leaders, and celebrities from Chinese civilization (circa 840 B.C. to 1979 A.D.) Do major and minor historical figures co-fluctuate across time, as well whether intra-field generational autoregression occurs regarding achievements? – Histories – Anthologies – Biographical dictionaries – Time series (20 year intervals) – Categories of achievement – Major versus minor figures – Objective indicators – Correlation – Time series/trend analysis (via regression) Simonton (1992) Eminent creators and leaders of Japan (circa 580–1939 A.D.) Which factors influence the number of distinguished women over time in male-dominated cultures? Attributions of greatness to leaders do no significantly differ between naïve raters and historians when given limited information about those leaders. Whether a president vetoes can be the result of an interaction between his inflexibility and certain situational factors (such as electoral mandate and which party controls Congress). The presidents were reliably discriminated on many items concerning presidential style, and those items collapsed into five dimensions. Likewise, those dimensions related to other variables of interest. Major and minor historical figures co-fluctuate across time, and the achievements of previous generations can predict the achievements of subsequent generations. Female literary and nonliterary eminence changed along with male literary activity, power and aggressive behavior, and ideology – Various biographical reference works – Presidential fact books – American history compendia – Data and information from Simonton's previous studies on presidential greatness – Japan Biographical Dictionary & Who's Who (1965) – Kodansha Encyclopedia of Japan (1983) – Time series (20 year intervals) – Female achievement – Male literary creativity – Male machismo activities – Male ideologies – Objective indicators – Alpha reliability – Correlation – Time series analysis G.S. Ligon et al. / The Leadership Quarterly 23 (2012) 1104–1133 Reference 1114 Table 2 (continued) Reference Leader(s)/type of leader(s) Research question Conclusion Eminent creators and leaders of Japan (circa 580–1939 A.D.) Do foreign ideas and peoples affect a nation's achievements? Is there much of a time gap between cultural acceptance and achievement influence? Which domains of achievement are most influenced? Simonton (2001) U.S. presidents Do certain variables adequately predict rankings of presidential performance? Simonton (2006) U.S. presidents How does estimated IQ scores of U.S. presidents relate to their performance? Simonton (2008) Eminent African Americans Spangler and House (1991) U.S. presidents Because previous studies involving the relationship between giftedness and genius have almost solely focused on majority-culture samples, how does that relationship apply to minority-culture samples? How is presidential performance influenced by individual motives? Strange and Mumford (2002) Charismatic and ideological leaders of the 20th century Do different mental models influence different vision-based leadership styles? Foreign influences typically had positive effects on most domains of creative achievement, and leadership domains were positively associated with national openness to those foreign influences. A 6-variable equation (including years in office, war years, scandal, assassination, heroism in war, and intellectual brilliance) successfully predicted rankings of presidential performance. Estimated IQ scores correlate with evaluations of presidential leadership performance. Both adult eminence and creative achievement positively correlated with early giftedness in a minority-culture sample (specifically, African Americans). The number of nots that a president used in speeches and letters measured a style of using power. Construct codings Statistical analyses – Time series (20 year intervals) – Domains of achievement – External influences (outside influences, travel abroad, and eminent immigrants) – Objective indicators – Correlation – Cross-correlation – Reference works – Previous studies by Simonton – Performance rankings – Years in office – War years – Assassination – Scandal – War hero – Intellectual brilliance – Ratings – Objective indicators – Correlation (including validity) – Alpha reliability – Regression – Previous studies and various sources (in particular: Cox, 1926; Rubenzer et al., 2000; Simonton, 1986a) – The Encyclopedia of Black America – The African-American Almanac – The Encyclopedia of African-American Culture and History – Various majority- and minority-culture sources – IQ – Intellectual brilliance – Openness to experience – Leadership performance – Archival eminence – Creative achievement – Giftedness – Birth year – Living contemporary – Gender – Achievement domain – Ratings – Objective indicators – Missing-values estimation methods – Correlation – Ratings – Objective indicators – Regression – Motive scores from Winter (1987) – Letters and speeches from collections, autobiographies, memoirs, and diaries – Biographies – Preliminary sources included general history texts and biographical websites – Biographies – Motive – Activity inhibition – Presidential performance – Ratings – Objective indicators – Correlation – Regression – Charismatic behaviors – Ideological behaviors – Leader performance – Leader contributions – Various controls – Ratings – Objective indicators – Correlation – Regression – MANOVA – Discriminant analysis – Biographical dictionaries – Same sources as Simonton (1992) (continued on next page) 1115 Vision-based leadership styles differ: ideological leaders stress personal values and standards whereas charismatic leaders adhere more to social needs and change requirements. Main constructs of assessment G.S. Ligon et al. / The Leadership Quarterly 23 (2012) 1104–1133 Simonton (1997) Data source(s) 1116 Table 2 (continued) Leader(s)/type of leader(s) Research question Conclusion Data source(s) Main constructs of assessment Construct codings Statistical analyses Suedfeld and Bluck (1988) Political leaders of countries that engaged in, or were recipients of, surprise attacks Do leaders' integrative complexities change before a surprise attack occurs? – Ratings – Interrater reliability – Integrative complexity – ANOVA – t-tests Revolutionary leaders How does a revolutionary leader's conceptual complexity relate to his success? – Proceedings of UN General Assembly or Security Council – Speeches – News conferences – Various sources (e.g., letters to colleagues, theoretical treatises, defense pleas) – Integrative complexity Suedfeld and Rank (1976) – – – – – Judgments – Correlation – ANOVA Suedfeld, Tetlock, and Ramirez (1977) Political leaders of countries involved in the Middle East conflict Do leaders' integrative complexities change before an outbreak of war? – Ratings – Interrater reliability – Integrative complexity – Correlation – ANOVA U.S. senators in the 82nd Congress How does a senator's integrative complexity relate to his or her personality and standing on foreign policy (i.e. isolationist, ambivalent isolationist, or nonisolationist)? – Speeches made in UN General Assembly and Security Council in the months prior to outbreaks of major hostilities – Congressional speeches relevant to foreign policy – Integrative complexity Tetlock (1981) – Integrative complexity – Evaluative assertion – Judgments – Ratings – Interrater reliability – Integrative complexity – Correlation – ANOVA – Discriminant analysis Tetlock (1983) U.S. senators Does a senator's political ideology relate to his or her integrative complexity? From the attackers' perspective, a decrease in complexity occurs as the time of a strategic surprise attack nears. Revolutionary leaders who were more adaptable to the environmental demands they encountered were more successful. Complexity decreases the nearer an outbreak of war is, suggesting that complexity could predict when hostilities may erupt. Foreign policy statements made by isolationists were the least complex, and typically expressed more positive attitudes toward their in-group and more negative attitudes toward an out-group. Conservative senators typically made less complex statements than did moderate and liberal senators. – Speeches from the Congressional Record – Ratings – Objective indicators – Interrater reliability – Integrative complexity – ANOVA – ANCOVA – Regression Thoemmes and Conway (2007) U.S. presidents How do U.S. presidents' integrative complexity differ? Presidents' integrative complexity differ based on a number of factors. – State of the Union speeches – Integrative complexity – Political party – Education – Age – Length of service in the Senate – Integrative complexity – Various environmental measures – Various personality measures – Ratings – Objective indicators Thorndike (1936) Male members of European royal families To repeat Woods's work to determine on Monarchs accuracy of assessment regarding the relationship between intellectual ability and estimability Results replicated the positive and moderate relationship between intellect and morality, suggesting that more intelligent leaders may – Heredity in Royality – IQ – Morality – Ratings – Interrater reliability – Integrative complexity – Correlation – MANCOVA – Regression – ANOVA – ANCOVA – Multilevel modeling – Effect sizes – Correlation Polarized contrast Qualified contrast Integrative comparison Conceptual complexity G.S. Ligon et al. / The Leadership Quarterly 23 (2012) 1104–1133 Reference Reference Leader(s)/type of leader(s) Conclusion Data source(s) Wallace, Suedfeld, and Thachuk (1993) Leaders during the Gulf Crisis of character (i.e., morality). Does a leader's integrative complexity influence their behaviors? Wendt and Muncy (1979) U.S. vice presidents Which personality variables influence the political success of U.S. vice presidents? Winter (1980) Southern African leaders Winter (1987) U.S. presidents Do certain factors influence the type of motive imagery used by Southern African leaders? What are the various motive profiles of U.S. presidents according to three different leadership models, and how do those profiles relate to leader appeal and performance? be more morally upstanding. Changes in integrative complexity, especially during stressful times, can indicate certain behaviors. Certain factors (activity–charisma, likability, and shrewd persistence) influence U.S. vice presidential success. A leader's war disposition and activity correlate with the types of motive imagery used. Yammarino, Mumford, Serban, and Shirreffs (2012) Political leaders Zullow and Seligman (1990) U.S. presidential candidates Which type of leader (personalized, socialized, charismatic, ideological, or pragmatic) is more likely to be the victim of assassination or attempted assassination? Do the ways in which presidential candidates frame explanations about bad events predict defeat? Pessimistic ruminators were much more likely to lose. Construct codings Statistical analyses – Texts from New York Times – Texts from embassies – Integrative complexity – Ratings – Interrater reliability – Integrative complexity – Excerpts from Madmen and Geniuses – Dozens of characteristics – Judgments – Objective indicators – Regression – Factor analysis – Interjudge variability – Verbatim responses to reporters' questions – Verbatim transcripts published in weekly news magazines – Presidents' first inaugural address – Achievement, affiliation, and power motive imagery – War disposition – Activity – Achievement – Affiliation–intimacy – Power motive imagery – Ratings – Judgments – Split-half reliability – Correlation - Validity – Ratings – Correlation – Previous historiometric studies that included political leaders – Various websites (including Wikipedia biographies, university website biographies, and historical biographies) – Nomination acceptance speeches (obtained from published proceedings or the New York Times) – Campaign itineraries – Whether leader was socialized or personalized – Whether leader was charismatic, ideological, or pragmatic – Victim of assassination – Target of assassination attempts – Various controls – Ratings (from previous studies) – Objective indicators – Chi-square analysis – Log linear models – Logit/logistic regression – Discriminant analysis – Causal explanations for bad events – Pessimism – Rumination – Explanatory style – Vote spread – Change in spread – Ratings – Interrater agreement – Z scores – Correlation (zero-order and partial) Q11 1117 Reference: citation. Leader(s)/type of leader(s): the leaders or type of leaders being assessed. Research question: the broad research question being studied. Conclusion: the general conclusion of the study. Data source(s): sources from which data were gathered. Main constructs of assessment: the constructs the leaders were assessed on. Construct codings: how the constructs were coded. Statistical analyses: types of statistical analyses used. A president's greatness in part is associated with his motives, and likewise a higher match between a president's motive profile and contemporary society resulted in higher presidential appeal. Socialized charismatic leaders were most likely to be victims of assassination or attempted assassination, whereas socialized pragmatics were least likely to be victims. Main constructs of assessment G.S. Ligon et al. / The Leadership Quarterly 23 (2012) 1104–1133 Research question 1118 G.S. Ligon et al. / The Leadership Quarterly 23 (2012) 1104–1133 beneficial about the study of outstanding leadership, then, is that by virtue of the type of information available about them, we are able to discern contextual details that would not be afforded if the unit of analysis (i.e., non-outstanding leader) were not under such academic scrutiny. In short, because these individuals are considered “outstanding” in terms of influence, there are copious open-source documents detailing observable, and more importantly quantifiable, differences among them in regard to behaviors, developmental influences, and performance. Given the above examples of what historiometric approaches can provide, it is surprising that we do not see more of these studies published in Leadership Quarterly to test develop, and refine leadership theories. Like all empirical methods, this approach does suffer some limitations, and these disadvantages should be assessed carefully in relation to the advantages afforded. In addition, because few graduate programs offer training in this approach, its underutilization in our field may also be due to lack of clarity around methods required for its execution. Thus, in the remaining sections we will provide an examination of methodological best practices for leadership historiometry, followed by a balanced analysis of the strengths and weaknesses of this technique. 4. Methodological best practices To date, page limits have restricted leadership scholars from fully providing step-by-step details on the methodological rigor required to some degree. While other broader social science outlets have published methodological pieces on variants of this technique (for example, see Simonton, 2010) in relation to other domains, when examining patterns of individual differences in leadership, a particular set of procedures must be employed for at least two reasons. First, because historical data about outstanding leaders is quite substantial and heterogeneous in nature, a content-driven coding scheme must be a priori defined in relation to the theories and sample at hand. While application of standard individual difference assessments has been applied successfully in some studies (Simonton, 1986a,b), we have found particular success in generating a theoretically-driven, psychometrically sound list of behavioral items hypothesized to differ among leaders (Hunter et al., 2011; Ligon et al., 2008; Mumford, 2006). These items, anchored by clear observational markers (e.g., type of University attended), are reliably identifiable in the factual, historical writing style found in most of the documents used in historical analysis. Second, because the subjects under examination are at times well-known and sometimes elicit rater biases (e.g., halo, Simonton, 1990) it is occasionally important to remove identifiers from the data in addition to employ a coding method that scrambles the observations across leaders and across raters. For this reason, we recommend using a multi-step coding technique, where observations are sorted into categories and/or assigned ratings by several iterations and across independent rater teams (McKeown & Thomas, 1988; Mumford, Gaddis, Strange, et al., 2006). In short, application of historiometry to the study of leadership requires six inter-related steps: 1) model specification, 2) sampling plan formation, 3) content coding scheme development, 4) materials preparation, 5) coding logistics, and 6) descriptive and multivariate analysis. Table 2 provides an overview of exemplar studies using variants of these techniques to examine leadership in particular. 4.1. Model specification Similar to all empirical examinations of leadership, the initial step in this methodology revolves around specifying hypothesized differences between leaders in the population of interest. Because this technique requires a fair amount of historically verifiable secondary-source data, research questions that allow this technique must seek to compare exemplars of similar historical notoriety and timeframe on constructs of interest. Thus, there are three typical types of theoretical models employed in the typical historiometric study: 1) descriptive studies seeking commonalities among leaders, 2) predictive studies assessing differential relationships to performance criteria, and 3) combination of the two for theory development. When the goal is to describe commonalities in a sample of outstanding leaders, researchers ask the question: what repeatable patterns can we see in this group? In a hallmark study representing this type of model specification, Murray (1928) examined a sample of individuals deemed “Great Orators” to determine if a common set of traits in early life were related to later status as an outstanding orator. A goal in this study was to determine what kind of educational activities should be promoted in early education to influence subsequent oration skills. As a result, activities under model specification required copious rating scales to be developed measuring artifacts of “genius” as described in historical sources about the individuals in question. While Murray's work exemplified early efforts to describe commonalities, the second type of theoretical model deployed in historiometry is intended to differentiate sample members based on some common criterion. Simonton's (1988b, 2001) work to differentiate U.S. Presidents in terms of rating and ranking performance data, respectively, represents a model intended to capture performance on a continuum, while Yammarino and colleagues' (2012) recent examination of predictors of assassination dichotomize the performance criterion (i.e., assassinated versus not). Each of these types of studies is intended to assess leadership characteristics related to some subsequent outcome (i.e., performance, assassination) and they provide some evidence as to the strength, direction, and robustness of these relationships. For example, Simonton (2001) examined contextual characteristics such as degree of scandal, war, and conflict during terms as they related to subsequent presidential rankings by subject matter experts (SMEs). Yammarino and colleagues (2012) examined a priori classified leadership style, based on the CIP model, of assassinated leaders when compared to a matched sample (heads of state and world leaders) of non-assassinated leaders in an effort to determine which leadership style might be most vulnerable to assassination. Finally, the third model used in historiometry, rooted in theory development, combines the first two types of research questions above to provide convergent and divergent validity evidence about the inferences that can be made about styles of G.S. Ligon et al. / The Leadership Quarterly 23 (2012) 1104–1133 1119 leadership. To date, Mumford (2006) is the most comprehensive effort in this vein that relies upon a historiometric approach. He executed a series of studies to investigate the homogeneity within the leader styles of charismatic, ideological, and pragmatic leaders (CIP) in terms of developmental influences, problem solving, communication, political tactics, and follower interactions (Mumford, Scott, et al., 2006). In addition, he examined the relationship of these patterns of differential characteristics between each style with performance criteria such as sustainability, valence (i.e., positive or negative), and scope of influence (Mumford, Strange, et al., 2006). This set of research questions required careful sampling to examine exemplars of each of the leader styles, as well as a content coding scheme used to discriminate them on hypothesized theoretical differences and performance criteria. 4.2. Sampling plan formation How the population of outstanding leaders is sampled depends on which of the three general research questions described above are posed. For example, in Murray (1928). As a great deal of Simonton (1988b, 2001, 2008). When the research question seeks to differentiate leaders on a temporal or situational bound set of criteria, however, more targeted sampling must be employed. For example, in DeChurch et al. (2011) study of leaders during natural disasters, leaders were sampled through the Army and the Defense Technical Information Center's archives in order to examine leader behaviors such as coordination, strategy formation, and communication. In a study examining the relationship between conscientiousness and death age, McCann's (2005) sample included only deceased U.S. Presidents. The nature of the research question dictates the desired characteristics of the sample members, and thus sampled populations of outstanding leaders will vary across studies accordingly. However, there are at least four best practices associated with the overall sample plan formation across most historiometric studies: 1) sample membership validation, 2) time period alignment, 3) selecting quality source data, and 4) selecting multiple leaders and types of source data. First, in some cases sample membership is clear; Yammarino and colleagues (2012) sought to examine the leadership style of every leader assassinated in the 20th Century, which helped make sample inclusion a dichotomous, clear cut decision. However, in other cases, particularly when research questions are geared around validating a particular stylistic framework, a priori assignment to such categorical distinctions is more complicated. This can be done via reliance upon previous classification schemes (Eubanks et al., 2010), empirical clustering (Strange & Mumford, 2002), or subject matter expert (SME) ratings of category assignments (Ligon et al., 2010). For example, as Strange and Mumford's (2002) study of ideological leadership was the first that sought to distinguish this style from a charismatic style, a cluster analysis (Aldenderfer & Blashfield, 1984) was first applied to identify what behaviors (e.g., focus on the past in messages — ideological) clustered together empirically. Next, performing a discriminant function analysis (Klecka, 1980), each leader was then assigned a score on these clusters to identify who manifested higher indices of one leadership style or the other. Leaders were then classified as ideological (high scores on ideological functions), charismatic (high scores on charismatic functions), or mixed (high scores on both functions) based on these empirically derived distinctions. In subsequent studies examining these theories (Mumford, 2006), members were sampled who evidenced strong (i.e., high scores on one or the other type of function) ideological or charismatic leader styles. While empirically derived techniques allow for classification of leaders based on clusters of behaviors, this step requires a great deal of resources to execute and is not always the best strategy given the research question at hand. Thus, the reliance of SME ratings to validate leader classifications is often employed in historiometric sampling plans. For example, in a study examining non-violent ideological leaders in regions and organizations relatively comparable to violent ideological leaders, a SME group of religious, geographic, and social movement scholars was used to help identify populations of interest and validate our selected sample assignments (Ligon et al., 2010). Because these area-specific SMEs are typically not also leadership scholars per se, the process of developing carefully constructed scales is critical for valid ratings of leaders into classified groups. To do this, first an initial classification, based on a sincere effort from the study Principal Investigator (PI) and research team in reviewing material for potential leaders to include in the sample, is assigned. Next, evaluations of the validity of these assignments to cells/types are completed via SME ratings on a series of Likert scales about classification scheme. Fig. 1 shows the rating scales we provided for SMEs in making assignments to some categories of interest: leaders of non-violent, ideological organizations whose height of power was during the 20th Century (Ligon et al., 2010). As noted in Fig. 1, another component of the sampling plan is to identify the period of interest from which to sample behaviors of the leader. For study-specific questions about time periods, this is defined by the research question. For example, in Simonton's (2008) study of eminent African Americans, he sampled activities during leaders' early (childhood and education timeframe) developmental periods to examine relationships between formative events and subsequent performance. However, when the research question assesses behaviors related to productive leadership activities (e.g., speeches, leader–member exchange relationships), one approach is to a priori identify the leader's height of power and sample such events during this timeframe. From a developmental standpoint, this is likely when the leaders' most prototypic mode of executing will manifest; events sampled in early career might not afford a full examination of the leader's style that would later dictate performance. In addition, at the end of power, the situational conditions might not afford full expression of leadership style as control gradually passes to others. From a pragmatic standpoint, we have also found that this is when leaders are “watched” the most; studying leaders during the height of power may yield the most information about his/her leadership style and contextual influences associated with it as most academic biographers and journalists will record events from this time frame. Given this predilection, it is also important to note that the model specified should dictate this time period of sampled behavior. For example, studying leader decisions leading up to loss of power might also warrant insight into behaviors to avoid or minimize. Whatever within-subject time period is selected for investigation (e.g., rise to power, peak of power), however, it is also important that other time specific 1120 G.S. Ligon et al. / The Leadership Quarterly 23 (2012) 1104–1133 boundary conditions be considered in accordance with designated temporal reference. For example, in Hunter et al.'s (2011) study of college and NFL football coaches, the aim was to investigate only those coaches who were achieved success at the pinnacle of their profession. Thus, although football was clearly played (and coached) prior to the instantiation of formal championships, only coaches who won national collegiate championships or Superbowls were included in the final study sample. The key takeaway here is that given the complexity of data collection associated with these efforts and the ambiguity such speculation requires, we recommend a priori designating the appropriate within-subject time period as well as time-based boundary conditions for case inclusion. The last two practices in sampling for historiometric studies are related: identifying quality and ensuring reliable source data. The information used to obtain qualitative source material about sample members is quite heterogeneous across studies, but some general guidelines can be inferred. The common thread between all of the historiometric studies we found was the reliance upon domain experts — often academics devoting significant portions of their research programs to objective analysis of these leaders — to assist sample development by using their techniques to gather and report information in a structured, standardized manner. For example, Simonton (1986a, 1986b, 1988a, 1988b) has developed presidential profiles that others have used for studies examining the U.S. Presidents (Deluga, 1997, 1998, 2001), while DeGregorio (1991). Although, examples also abound where less academic resources are used in tandem to these data sources; DeChurch and colleagues (2011) relied upon secondary sources such as blogs and newspaper reports to cross-reference archived accounts of how national crises were handled. Thus, when the events are highly publicized in nature, a variety of sources can be used to draw conclusions. Given the variability in source quality, however, statistical control must be exercised to at least partially account for quality concerns and somewhat standardize the type of information gathered. Ways to handle source quality variance through application of control variables will be discussed in the next section on content coding schemes. Finally, it is imperative that multiple sources, events, and leaders are sampled to assess behavior to draw reliable, generalizable conclusions about relationships. There are copious high quality studies applied to one leader where researchers obtain quantitative metrics about a given case, but we argue these studies are not indicative of a true historiometric approach. For example, in Abe's (2011) study of Alan Greenspan's speeches during economic downturns, conclusions were able to be drawn about that particular leader's use of psychological distancing, but little could be generalized about leadership in general. Thus, we argue that three defining hallmarks of historiometric studies are sampling: 1) multiple members of a given population of leaders, 2) multiple types of behaviors, and 3) multiple sources depicting those leader behaviors. Methodologically related, criterion data are strengthened if they can be gathered from sources independent of the predictor data, limiting single-source biases that can be introduced. In a study examining organizational characteristics as they related to terrorist attacks, Asal and Rethemeyer (2008) relied upon the Global Terrorism Database (National Consortium for the Study of Terrorism and Responses to Terrorism (START), 2012) to gather criterion data regarding attack lethality, while predictor data was sampled from external news accounts and organizational websites. 4.3. Content coding scheme As the central goal in historiometric studies is to transform historically verifiable qualitative source information into quantitative data, this third methodological step requires a particularly stringent set of techniques. Similar to the above section on sampling plan, model specification drives what this content coding scheme will include, and thus also dictates cross-study differences in variables assessed. Despite these cross-study differences, there is a set of best practices associated with our approach to leadership historiometry. Acknowledging that there are other frameworks for content coding (Simonton, 2010), for studies of leadership we advocate a set of techniques that draw upon psychometric principles with a particular focus on biodata, or life history, item generation best practices (Mumford, Stokes, & Owens, 1990). Thus, the goal of this section is to provide a description of the general framework recommended for content coding, followed by a specific discussion of exemplar coding schemes in predictor, criterion, and control data in these studies. First, the coding scheme we recommend requires multiple trained coders for each set of constructs (predictors, criteria, and controls). This is counter to many historiometric studies we reviewed, where approaches largely rely on coding from one individual, albeit an individual familiar with both the sample and variables under examination (and often the study PI). For example, in a study of Please review the following list of leaders and his/her inclusive years for height of power. Next to each name, indicate your rating about each individual’s position (for leaders with whom you are unfamiliar, please write “unfamiliar”): 1– disagree 2 – agree 3 – strongly agree 1. _____ This is a former leader of non-violent organization. 2. _____ This is a former leader of an ideological organization (organizationdevoted to advancing a certain belief system rather than maximizing profits; profits are sought only in effort to advance the belief system). 3. _____ Height of power, or period when this leader exercised the most influence on others, was during designated time-frame (if not, please write in inclusive years and justification). Fig. 1. Example scale for non-leadership SMEs to use in validating sample cell assignments. G.S. Ligon et al. / The Leadership Quarterly 23 (2012) 1104–1133 1121 terrorist leaders' psychological profiles, Sageman (2008) gathered and coded open-source reports of identified leaders of violent groups under the Global Salafi Mujahideen ideology. Sageman, a trained clinical psychologist who now works closely with the intelligence community, was well-suited to identify and quantify evidence of varying personality characteristics of terrorist leaders via secondary source descriptions. However, his ratings were singular in nature; no effort was made to assess the reliability of the ratings he assigned. This clinical, case study approach makes it difficult to replicate findings across researchers, limits the applicability of this approach given the burden on the PI's time, and provides limited reliability and validity evidence for inferences made. While the coding by a single domain expert is useful in efforts for theory development because it yields rich contextually anchored information, for historiometric analysis of leadership in general we recommend training and using multiple coder teams to gain an index of inter-rater reliability for constructs observed. This type of training and set of procedures will be explained in greater detail in section five, Coding Logistics. Second, we advocate for coding schemes to be developed with the same practices used in test development (Osterlind, 1998). That is, constructs of interest should be defined, iteratively reviewed, and edited by a subset of SMEs in item writing to ensure clarity, parsimony, and unidimensionality. For example, in a study examining the communication strategies of outstanding leaders, Mumford, Bedell, et al. (2006); Mumford, Gaddis, Licuanan, et al. (2006); Mumford, Gaddis, Strange, et al. (2006); Mumford, Scott, et al. (2006); Mumford, Strange, et al. (2006) were careful that item stems were generated to measure affect and use of stories as separate constructs, and only assessed in communications with the mass public (see Fig. 2 for how these scales were finalized). This was refined via SME feedback during the item generation phase of this project; earlier iterations were worded, “To what extent does the leader use emotionally evocative stories or analogies to illustrate a point?” This first item was essentially double-barreled, in that some leaders used emotion but not stories, and vice versa, to communicate with the mass public. It follows, then, that validated scales used for other purposes can be adapted to these studies as another unique approach. For example, in an effort to assess the leader–member exchange (LMX) relationships of outstanding leaders and their top management teams (TMTs), Strange (2004) used both internally developed items and an externally validated scale. First, using SMEs familiar with item-writing and LMX theory, Strange and colleagues developed a set of theory-driven (Dienesch & Liden, 1986; Graen & Uhl-Bien, 1998) content items for judges to apply to passages describing exchanges between leaders and followers. Coders also applied a pre-established behavior description inventory, the LMX-7 (Schriesheim, Castro, & Cogliser, 1999), to appraise the quality of leader-follower relationships by assessing these same exchanges. Results showed that personalized and socialized leaders display significant differences on her study-specific relational items, as well as the LMX-7 standard behavioral markers. In addition, using this multi-tiered assessment technique, Strange (2004) also found that relationship strategies such as support, team-based, and power-sharing were related to metrics of sustainable performance. Third, behavioral exemplars, benchmarked against the sample at hand, are used to illustrate differing levels of each construct for at least two reasons. First, the nature of biographical material used in historiometry is heterogeneous, ill-defined, and unstandardized given that it was prepared ipsatively by multiple scholars in diverse domains. To infer levels of psychological constructs (e.g., positive affect), observable and concrete examples must be provided to coder teams as a reference point to facilitate reliable, valid ratings. Second, outstanding leaders operate in unique environments, and thus the way their behaviors manifest should be operationally defined by exemplars from other outstanding leaders. This approach, the development of benchmark scales, requires an initial oversampling from the population of interest (e.g., CEOs) to array behavioral examples in a way that differentiates low, moderate, and high levels of a given construct. We recommend oversampling by 10%, and then randomly selecting 10% of that sample to a priori identify benchmark behaviors to exemplify constructs. Regardless of whether a Likert rating scale or a Q-Sort strategy (McKeown & Thomas, 1988) is used, these behavioral examples can provide a sample-bound reference for raters to use in assessing leadership constructs relative to those of other outstanding leaders. These are simply illustrative examples of behaviors demonstrative of the construct of interest and can increase reliability indices of agreement between raters significantly when implemented. We recommend using these benchmarks in rater training, but also including them on rating materials to be used in independent coding by rater/judge teams. Finally, in line with best practices of scale development, we suggest that each construct be measured by multiple items to provide some index of reliability (Traub, 1994). There are two approaches we have found successful to do this: 1) multiple Likert-scaled items for each construct, and 2) a modified Q-Sort approach. For the first approach, and similar to test item development, we recommend a coding scheme where raters read through sampled biographical material and then make ratings on multiple (at minimum four) Likert scales assessing related but orthogonal indices of the construct of interest. For example, to assess the long-term impact of communication strategies, raters assessed six components of speech criteria such as 1) extent people still quoted leader speeches, and 2) degree to which speeches are considered landmark events (Mumford, Gaddis, Licuanan, et al., 2006; Mumford, Gaddis, Strange, et al., 2006). Items can then be empirically combined to create scale-level assessment (e.g., sustainability of speech impact), granted that appropriate inter-item correlations are obtained (Traub, 1994). While the generation of multiple items across constructs is a well-established way to transform qualitative information into quantitative ratings, we have also found success using a modified Q-Sort technique (McKeown & Thomas, 1988). When there are copious data points to be evaluated independently, Q-Methodology results in a reliable and unique way to obtain metrics of leadership characteristics. Q-Sort techniques are hardly innovative (Stephenson, 1935), but historically they have been applied to assess an individual's subjective viewpoints or arrayed personality markers (e.g., “this trait is most like me versus that trait”). Once viewpoints are modeled into Q-Sorts, data analysis then examines N Q-Sorts (i.e., the individual/Q-Sort arrays, not the items, serve as the unit of analysis) through some empirical reduction technique. There are multiple ways (e.g., factor analysis, cluster and then discriminant function analysis) these can be analyzed to obtain factor scores to represent “typical patterns of responding,” and then each sample member can be assigned a score on such factors or functions. 1122 G.S. Ligon et al. / The Leadership Quarterly 23 (2012) 1104–1133 While traditional social science research applies Q-Methodology to study constellations of traits, beliefs, or some other subjective characteristic of individuals participating in the Q-Sort, we have applied it in an effort to analyze historiometric data via trained judges as well. In a study examining the developmental themes related to the CIP model of leadership and the personalized versus socialized variants of each, we used models from the adult development literature (McAdams, 2006; Pillemer, 2001) to sample seminal developmental events that occurred in 120 leaders' lives from age 14-early career (Ligon et al., 2008). Our approach yielded roughly 1400 events across the leaders under consideration, and Table 3 illustrates each of these types of events. The next part of this study required thematic coding of markers related to charismatic, ideological, or pragmatic leadership styles in each of these events. After reviewing the relevant leadership literature about each of these styles and likely developmental themes related to each, 28 dimensions were retained to differentiate charismatic, ideological, and pragmatic leaders' early developmental experiences (e.g., risk taking, spirituality, and flexible decision making dimensions, respectively). Each of these dimensions was then illustrated by benchmark behavioral examples. A team of six judges was asked to review each event drawn from the biographies (e.g., career originating event) and assign the event to the thematic dimension categories. Specifically, judges were asked to determine whether each event did or did not reflect relevant thematic content for each of the 28 dimensions under consideration. Scoring then occurred by determining those dimensions to which the majority of the judges assigned an event. Dimension scores were obtained for each leader by determining the number of events identified for a leader assigned to a given dimension, and then dividing by the total number of events identified for the leader under consideration. An example of an anchoring, or belief formative, event from Fidel Castro that scored high on ideological and personalized leadership dimensions can be seen on Table 3. This event scored highest on thematic dimensions of Belief Commitment, Inspirational Communication, Risk-Taking, Power Motives, and Negative Life Themes, as the majority of judges assigned this event to these categories over others. This type of formative life event might help explain how Fidel Castro has been characterized as a personalized ideological leader in multiple studies of his leadership exchange relationships (Strange, 2004), performance (Mumford, Strange, et al., 2006), and problem solving style (Mumford, Bedell, et al., 2006). More detail about best practices in training these judges and the overall rating procedure will be described in the following sections on Coding Logistics and Preparation of Materials. To the best of our knowledge, the study of developmental events is the only historiometric one that has employed a Q-Methodology, but it seems appropriate for any study of leadership where the assignment of multiple, relatively standardized activities or events to a large amount of thematic dimensions is required. For example, McCall's (2010) work on experiential development activities could make use of sorting experiences such as job rotations, project assignments, and mentoring into thematic dimensions of leadership skills developed. Using these best practices of treating leadership historiometric coding schemes with the same consideration required of psychometrically sound tests and scales in general, there are three general categories of content to be coded across all historiometric studies: predictor, criterion, and control variables. Model specification dictates what types of predictor content should be coded, and Table 2 illustrates that constructs of interest range from Machiavellianism markers (O'Connor et al., 1995) to indices of morality (Thorndike, 1936). In short, there is no recommended set of content to be coded in predictor data of historiometric leadership studies; instead, if one follows the above described procedures and generates sample benchmarks of behavioral exemplars for each construct of interest, virtually any leadership theory can be assessed in this type of research paradigm. Related, there is much variability in the types of criteria assessed in these studies, depending on the theoretical model of interest. For example, study-specific criteria assessed in Simonton (2001, 2006), while McCann (2005). In addition to these study-specific criteria, it is recommended that a set of general performance criteria also be assessed to compare leaders across domains and provide some evidence about the relationships between our theories of leadership (e.g., authentic leadership facets) and performance in a social domain. Based on findings from Strange and Mumford (2002), there are 12 general performance criteria that can be assessed regarding overall societal impact. These have been used in numerous historiometric studies to differentiate leaders on short and long-term performance metrics across occupations. The first five criterion measures are based on observations typically found in the Expression of Positive Affect—To what extent does the leader express positive emotions (e.g., happiness, joy) in his or her speeches and communications with the mass public? 1 –The leader never expresses positive emotions in his/her speeches or communications with the mass public. 3 – The leader occasionally expresses positive emotions in his/her speeches or communications with the mass public. 5 – The leader frequently expresses positive emotions in his/her speeches or communications with the mass public. Use of Stories and Analogies—To what extent does the leader communicate his or her points through the use of stories and analogies in his or her speeches and communications with the mass public? 1 – The leader never uses stories or analogies his/her speeches or communications with the mass public. 3 – The leader occasionally uses stories or analogies his/her speeches or communications with the mass public. 5 – The leader frequently uses stories or analogies his/her speeches or communications with the mass public. Fig. 2. Example items and rating scales for positive affect versus the use of stories. G.S. Ligon et al. / The Leadership Quarterly 23 (2012) 1104–1133 1123 prologue chapters of academic biographies, and are quantified via counts examining 1) the number of positive contributions, 2) the number of negative contributions, 3) the number of different types of positive contributions, 4) the number of different types of negative contributions, and 5) the number of institutions established by the leader. What is helpful about these criteria in a content coding scheme is that they require raters to make judgments about relatively objective markers on metrics that should be applicable across the types of eminent individuals in historiometric leadership samples, and information can be readily verified across multiple reference sources. In addition to these counts of criteria mentioned, a SME in leadership applications or a trained judge (if provided with behavioral benchmarks) is required to make ratings on seven additional criteria after reviewing biographic material. Table 4 illustrates the items that these ratings examine, made on a 5-point Likert scale. The final type of items to code in leadership historiometry is intended to provide covariate controls for the inferences being drawn. Similar to the predictor and criterion content coding schemes, each exemplar study in our review applied study-specific controls. For example, when examining LMX relationships with leaders in a dissertation about leaders and top management teams, a number of follower-demographic control measures (e.g., similarity of followers to leader) were formulated to examine the inferences' generalizability to other LMX relationships (Strange, 2004). In addition to these particular content analysis considerations, we identified a number of additional measures meant to assess internal and external validity concerns. These constructs were assessed either via objective counts (e.g., word count), ratings on Likert scales (e.g., amount of documentation provided for objective markers), or categorical assignment (e.g., was the leader a pre- or post WWII leader?). To control for threats to source quality or internal validity, material characteristics regarding author credentials/training, amount of data available (and amount of secondary supporting sources used to obtain that data), and degree of media and author bias are some of the most commonly applied items. To assess external validity concerns, a series of measures are applied to assess the leader's role and the context in which he/she led. For example, most studies controlled for type of leadership role (e.g., business, political, military, non-profit), number of years in leadership power, and time-frame and region of leader's height of power. More detail will be provided in the section on analyses as to varying ways to examine and account for these variables' influence on the relationships of interest. For a full list of recommended internal and external validity controls, see Table 5. Table 3 Examples of developmental life events used in Q-Sort. Event type Benchmark example Originating event “From an early age, the young Rupert [Murdoch] was aware o the power and the glory and the sheer fun which accrued to his father from newspapers. Keith [Rupert's father] used to take his son around the Herald's office on Flinders Street, and Rupert often said later that the smell of the ink, the noise of the presses and the highly charged atmosphere were irresistible. ‘The life of a publisher is the best life in the whole world. When kids are subjected to it there's not much doubt they'll be attracted to it.’” (Shawcross, 1997 p.27) “The most dramatic story concerns Lewis's involvement in the 1903 disaster at the Union Pacific Railroad Company's coal mine in Hanna, WY. Passing through the area by chance, Lewis arrived in time to assist a rescue team in carrying out he torn, charred bodies of 234 miners…‘what ripped his emotions to shreds was the sight of the numb, mute faces of the wives now suddenly widows of the men they loved.’” (Dubofsky & Van Tine, 1986 pp. 14–15) “In what Fidel calls, ‘a decisive moment on my life,’ Angel Castro decided during the boys' summer holiday after the 4th grade that they would not go back to school…But Fidel [Castro] was determined to return to school. As he tells the story, ‘I remember going to mother and explaining to her that I wanted to go on studying; it wasn't fair not to let me go to school. I appealed to her and told her I would set fire to the house if I wasn't sent back…so they decided to send me back. I'm not sure if they were afraid or just sorry for me, but my mother pleaded my case.’ Fidel was learning quickly that absolute and uncompromising stubbornness was a powerful weapon. This may have been the most important lesson he had drawn from his young years at the finca, and he never forgot it.” (Szulc, 1986 p. 112) “Almost forty years later, on the occasion of a commencement address at Fisk, and perhaps under the influence of the occasion, DuBois recalled those three years of “splendid inspiration” and nearly “perfect happiness” with teachers whom he respected, amid surroundings which inspired him. The ten years after Fisk he chronicled as “a sort of prolongation of my Fisk college days.” I was at Harvard, but not of it. I was a student of Berlin but still a son of Fisk. I used my days there to understand my new setting…” (Broderick, 1959 p. 9) “She [Betty Friedan], who had been the ringleader and chief instigator, the one who generated all the excitement, was suddenly alone, abandoned by her friends. The creator of clubs was not chosen for the most exclusive club at all—the high school sorority. She was desolate…The year of loneliness that followed was the lowest point of her life. She blamed it primarily on anti-Semitism…The sight of the car full of friend, a vision that she yearned for, triggered something in her, and she made a promise to herself: ‘They may not like me now, but [someday] they are going to look up to me.’” (Hennessee, 1999 p. 15) [After receiving average marks on his officer's appraisal, [Charles] de Gaulle was given a lackluster assignment.] “Indeed, for a soldier with his innate conviction of his intellectual superiority, the choice of a department concerned with such routine matters as transport and supply was humiliating. At Mayence, in fact, he was put in charge of refrigeration, which must have seemed an insulting punishment for an unwelcome independence of spirit…[de Gaulle after receiving the news] ‘Those c…s of the Ecole de Guerre! I shall only come back to this dirty hole [sale boite] as Commandant of the Ecole! And you'll see how everything will change!’” (Crozier, 1973 p. 39) Turning point event Anchoring event Analogous event Redemption event Contaminating event Adapted from Ligon et al., 2008. 1124 G.S. Ligon et al. / The Leadership Quarterly 23 (2012) 1104–1133 4.4. Materials As historic records provide a rich, contextually-anchored source of descriptive data, successful application of the coding schemes described in the preceding section depends to a large degree on the procedures used to draw behaviors and apply ratings accordingly. In addition, depending on the type of research questions addressed, care must be taken to abstract relevant samples of behavior. In a study examining how political leaders used masculine rhetoric and ideologies as a political strategy, Coe, Domke, Bagley, Cunningham, and Van Leuven (2007) selected direct quotes from George W. Bush from a sample of communications found in the National Archives and Records Administration's Weekly Compilation of Presidential Documents. In Simonton's (1998b) examination of King George III of Great Britain's possible precipitators of poor health, he reviewed academic biographies of King George III and sampled only references to the king's physical and mental health to be coded for stress and health ratings. Finally, to assess Alan Greenspan's changes in communication style in relation to differences in the economic environment, Bligh and Hess (2007) applied content analysis to testimonies and speeches available on the Federal Reserve Board's website. Material to be content coded should be prepared in such a way so as to address three concerns with historical data: 1) reduction of bias from raters, 2) standardization of source material, and 3) ease of use of coding schemes for raters. As noted earlier, the approach to leadership historiometry we recommend is one rooted in obtaining multiple ratings from multiple, trained coders. Thus, a first concern when conducting this type of research is to prepare materials in a way that minimizes bias from raters. As a result, when preparing abstracted source material from historically notable leaders, by definition some caution must be applied to limit coder bias. In our studies of the 120 eminent leaders (e.g., Malcom X, Rupert Murdoch, Michael Collins), rater teams were assessed for bias by asking for confidence ratings from coders as to their a) degree of familiarity with each leader and b) knowledge of positive versus negative reputation of each leader. After numerous trials with our raters, the overwhelming majority of whom were undergraduate psychology majors, we found that fewer than 10% of the leaders were consistently and accurately recognized. While this is somewhat of a disheartening finding in general, for the purposes of reducing rater bias we can be somewhat certain that leaders in our sample were unlikely to be rated with systematic bias given the lack of familiarity with eminent historical leaders in general from undergraduate college students. However, we recommend taking similar precautions to assess rater bias prior to beginning any historiometric study, and particularly so when exemplars at hand carry strong cultural recognition (e.g., Osama bin Laden). In the event such bias is detected, preparing data in a way that removes identifiers is recommended. For example, we conducted our study on outstanding leaders at the same time that Martha Stewart, one of our sample members, received increased media attention due to felony convictions and a string of television documentaries depicting Stewart in an unfavorable light (Ligon et al., 2008). To reduce potential bias, individually identifying information about her and all leaders in our sample was subsequently removed from material to be coded. Second, as biographic information used in historiometric studies is complex, situationally contextualized, and prepared by different authors, standardizing how behaviors are sampled and presented to rater judges can increase the reliability of coding. One way to do this is to create profiles for each type of event or behavioral example for a given leader. Simonton (1986a, 1986b, 1988a, 1988b) has created leader profiles with various types of biographical data that have been used in numerous studies (e.g., Deluga, 1997, 1998), and developmental event cards were generated for the Q-Sort procedure for the study on developmental patterns (Ligon et al., 2008). This extra step of collating and re-organizing content from fact books, biographies, and other sources is time consuming and requires careful attention to detail to ensure judges are able to evaluate all relevant material. However, by providing materials to be rated in a standardized fashion, raters are able to reduce coding time for the study and easily find information upon which to assign ratings. Thus, this approach is recommended when a study employs relatively unskilled raters rather than domain experts who might be more readily able to assess relevant variables in copious, unstructured data. Table 4 Typical performance criteria used in historiometric leadership studies. Count metrics 1. Number of positive contributions made by the leader. 2. Number of negative contributions made by the leader. 3. Number of different positive contributions made by the leader. 4. Number of different negative contributions made by the leader. 5. Number of institutions established by the leader. Likert scale rating metrics 6. How much did the leader contribute to society? 7. How long did these contributions last? 8. How many people did the leader affect? 9. Did the leader initiate mass movements? 10. Was the leader's agenda maintained when he/she left power? 11. Were institutions established by the leader still in existence? 12. What was the biographer's evaluation of the leader? Note. A great deal of empirical work has been executed to demonstrate the validity and reliability of the procedures used to apply these metrics to studies of outstanding leadership. For more detail about this validation effort, consult Pathways to Outstanding Leadership, by M.D. Mumford (2006, pp. 71–73). G.S. Ligon et al. / The Leadership Quarterly 23 (2012) 1104–1133 1125 Finally, most of the coding schemes illustrated in Table 2 required judgmental evaluations on multiple different types of constructs. Assigning ratings is cognitively taxing, and assigning ratings on multiple different types of indices can be exhausting and laborious for judges. Thus, we recommend a) keeping rating anchors scaled in a consistent fashion, as there is limited evidence that varying types of response options (e.g., rotating high versus low anchors) provides any psychometric benefit (Osterlind, 1998), b) providing benchmark examples on or near rating scales for ease of reference, c) limiting the number of assessments any one rater makes per event/behavioral example in a given session (e.g., for the Q-Sort we did, pilots indicated that raters could reliably sort up to 50 events into thematic categories in a 2-hour rating block, more or less than that resulted in decreased reliability), and d) organizing all materials fastidiously, either through electronic means or hard copy bindings. This last step, material organization, is often overlooked in project plans of these studies, but our experience indicates that if done well, this step can take graduate student teams up to 2–4 weeks, depending on the sample size and number of coding dimensions. 5. Coding logistics Related to the above section, raters must be trained extensively on how to use these materials in a reliable fashion. In addition, a general framework should be developed to implement the coding process, as multiple independent rater teams are recommended for predictor, criterion, and control dimensions. Thus, in the following section, we will provide detail about such coder training programs, specifically benchmark and calibration training, assessing reliability, and developing a system for rater teams to code material. First, content coders should be comprised of teams separate from those who gathered the materials to be coded. In our work, we have found success using a mix of graduate and undergraduate students pursuing degrees in psychology, but given the training that occurs for these judges, individuals from any major could be successful in this vein. In considering numbers, it is recommended that at least three judges make each rating that requires some level of content judgment (e.g., Q-Sorts, Likert ratings), but for metrics that are more straightforward in nature (e.g., controls such as word count), one judge should suffice. After recruiting judges, we recommend a two-week benchmark training program, which typically is comprised of 4–6 h of face-to-face training, and 4–6 h of independent work by judges. In this program, judges are first familiarized with the nature of the stimulus material (e.g., developmental events, problem solving descriptions, or leader–member exchanges) that will be used in coding. Next, they are presented with operationally defined dimensions which they will code in this material. For example, in Eubanks and colleagues' (2010) study on how leaders responded to criticism, judges were familiarized with definitions of collaborative versus confrontational strategies, and then raters were asked to restate in their own words what key concepts differentiated these approaches. As situating these definitions in the sample at hand is critical for leadership historiometry, the next step in training is to illustrate varying levels of these constructs with benchmark examples abstracted from the population of interest (e.g., U.S. Presidents). This serves two purposes. First, it allows judges to better understand how these expected behaviors manifest in this type of sample, providing a framework to use in comparing similar behaviors from individuals in the study at hand. Second, the use of these benchmarks in training helps judges learn which information is useful for coding (i.e., concrete behavioral descriptions) versus which information should be discounted (e.g., superfluous contextual details, author subjective statements). In short, benchmark training facilitates raters learning how to assess only the incident under consideration when making ratings. Rater training should also cover typical biases (e.g., halo, leniency, central tendency errors) and specific errors made in historiometric studies (e.g., focus on evaluative/subjective statements, imposition of personal biases, and inadequate attention to nuances of academic writing). Following this informational portion of training, rater teams should practice assigning a small subset of ratings in the session with others and the study PI. When ratings are made on a 5-point scale, we recommend discussing differences greater than 2 points. This type of synchronous, real-time discussion of subtle distinctions using the same dataset can increase the reliability of ratings substantially. Table 5 Typical covariate control measures applied in historiometric leadership studies. Internal validity controls 1. Word count. 2. Number of outside sources referenced. 3. Author education level. 4. Author credibility (e.g., multiple publications on leader, graduate training, awards won). 5. Degree of bias (both positive and negative biases should be assessed). 6. Was author raised in subject's (leader's) country of origin? 7. Primary type of publication (e.g., peer reviewed). 8. Amount of objective predictor data available. 9. Amount of objective criterion data available. 10. Was source translated? External validity controls 11. Geographic region of leader (i.e., Western or Non-Western). 12. Development of leader country during height of power (i.e., industrialized or non-industrialized). 13. Munificence of leader country. 14. Post or pre-WWII time period. 15. Length of height of power. 16. Type of organization (i.e., for-profit, military, government, non-profit). 1126 G.S. Ligon et al. / The Leadership Quarterly 23 (2012) 1104–1133 Next, raters are assigned a small subset of leader behaviors (predictor, criterion, or control data) to independently assign ratings over the next week. For Likert items, using procedures outlined by LeBreton and Senter (2008), the study PI then can assess inter-rater reliability of these ratings. For categorical assignments (e.g., Q-Sort procedures), a kappa agreement coefficient should be obtained to assess inter-rater agreement of categorical assignments (McKeown & Thomas, 1988). Given that raters are asked to make assessments about a large amount of variables on relatively complex information, we have found that inter-rater reliability coefficients typically range from .70–.90, and kappa coefficients range from .55–.90 for well-calibrated raters. Until agreement indices fall into these ranges, however, we recommend repeating these iterations of training (particularly the discussion of differences larger than 2 points) with content coders. In addition, at several points during the coding process (particularly for large samples), we have found success with taking “pulse” reliability and agreement assessments. Specifically, we recommend assessing reliability and agreement at 10% increments of the coded material (so for N = 100, assessing coders at n = 10, 20, 30 leaders). When coefficients of agreement fall below acceptable ranges, we recommend conducting “calibration training,” or training intended to re-orient raters to how each other see the differences in the data. In making these Likert ratings, judges are typically presented with either a binder or electronic folder containing a subset of the relevant stimulus materials abstracted from the historical records. The material should be structured in a way where material applying to one leader is distributed across multiple binders or folders and thus not expressly identifiable for the coding teams. In addition, the rating plan should ensure that these binders/folders are rotated across the judges so that each leader was assessed by various coders at different levels of training throughout the process. For the Q-Sorts, cards to be sorted (e.g., non-identifiable descriptions of developmental events, Ligon et al., 2008) were also distributed in a way that raters were not sorting all cards of one leader in a given session. Once all of the data has been coded and acceptable reliability coefficients are obtained, the three independent coders' ratings should be averaged to lessen likelihood of spurious errors in any one individual's scores. For Q-Sorts, assignments can be either averaged across judges or added together to obtain arrays for each unit of analysis. In the following section, we will describe how these metrics are typically evaluated using descriptive and inferential analytic frameworks. 5.1. Analyses Like other methodological best practices, the analyses flow directly from the theoretical model under examination. If the above-detailed practices to gather multiple predictors, criteria, and controls are implemented, the research team is left with a collection of quantitative data to which virtually any analytic technique can be applied depending on the research model specified. Given that multiple analyses exist for historiometric data, the following section is not intended to be prescriptive in nature. Instead, in line with the other sections, we will provide one exemplar set of techniques we have found useful, particularly when handling the internal and external validity controls. When the theoretical model specifies an examination of the patterns of individual differences that drive distinctions between types of leaders (e.g., the CIP model) and performance, a multi-tiered analytic approach based on multivariate analysis of covariance, discriminant-function analysis, and regression can be useful (Mumford, 2006). After running appropriate exploratory statistics to examine (i.e., check for assumption violations) and clean the data (i.e., correct data entry errors), the first set of analyses are typically a series of multivariate analyses of covariance (MANCOVAs). In these analyses, the grouping variables of interest (e.g., personalized versus socialized leader orientations) are treated as independent variables and the dimensions under examination (e.g., degree of collaboration, scores on Big Five personality variables, jargon used in speeches) are treated as dependent variables. The covariates applied in these analyses should include both internal and external validity controls, and retained if those covariates proved significant (p b .05) with the dependent variables of concern and assuming they met the sphericity assumption. In short, inferences about differences (e.g., use of jargon) across grouping variables (e.g., leader orientation) are only made after taking into account important controls. The next set of analyses provide a summary description of how leader groups (e.g., personalized versus socialized, assassinated versus not assassinated) differ from each other in terms of the dimensions under examination (e.g., LMX techniques used). Thus, when even marginally significant differences (p b .10) are obtained across independent variables in the MANCOVAs, a discriminant-function analysis can be conducted to identify the underlying set of variables that best account for these differences. This approach allows researchers to assess packages of variables, as many of these operate in tandem to describe behaviors in domains as complex as leadership (Klecka, 1980). For example, in a study examining problem solving differences of charismatic, ideological, and pragmatic leadership styles, a significant main effect was obtained from the MANCOVA analysis when taking into account general and study-specific controls [F(8, 102)= 33.63; p b .001]. As a result, a set of discriminant functions were obtained to distinguish charismatic, ideological, and pragmatic leaders, and dimensions yielding sizable loadings on that function were information gathering (r = .75), concept selection (r = .42), and idea generation (r = −.35). Because the dimensions underlying this function represented information acquisition, this function was labeled Expertise by the research team (Mumford, Bedell, et al., 2006). The discriminant functions obtained in this second set of analyses provide the basis for the third analytic step: obtaining leader scores on each of these functions. In the above-described example of problem solving, average scores were obtained for each of the three leader types. Pragmatic leaders, known for their analytical approach to leadership (Mumford & Van Doorn, 2001), scored the highest on this Expertise function (M = 1.57) when compared to charismatic (M = −1.73) and ideological leaders (M = .15). Thus, this third step in the analytic framework can illustrate how the different groups of leaders (e.g., styles, assassinated versus not) score relative to important differentiating characteristics. G.S. Ligon et al. / The Leadership Quarterly 23 (2012) 1104–1133 1127 The final set of analyses is designed to examine how these important differences relate to leader performance or some separate criterion of interest (e.g., health ratings). To do this, each leader's scores on a given criterion variable must be obtained (e.g., number of positive contributions). Each criterion score is then correlated with and regressed on the scores on the discriminant-function variables intended to summarize cross-group differences in dimensions under examination from steps 1–3 (e.g., the scores on the Expertise function in the previous example). When conducting these regressions, however, it is important to note that significant (p b .05) control variables identified in the first set of MANCOVAs are entered as the first block of predictors. This helps to ensure that observed differences take into account important internal and external validity concerns by covarying out their influence statistically. While these four steps represent a general framework for how to reduce and interpret the data obtained in leadership historiometry, other techniques associated with study-specific differences can also be incorporated into this scheme. For example, Simonton's (2010) review of time series approaches to examine these variables is quite detailed. Another example can be found in Aldenderfer and Blashfield's (1984) publication, which reviews techniques used in cluster analysis to investigate useful coding schemes and taxonomies. This last approach holds particular promise when the research question is more exploratory in nature than the ones summarized in Table 2. 6. Summary for methodological best practices In the above six sections, we reviewed contemporary approaches for conducting historiometric studies of leadership. When reviewing the studies presented in Table 2, there are three key lessons that can be taken from the scholars involved with this research. First, sampling should be approached rigorously so that inferences can be made about not only outstanding leaders in the population of interest, but also for leadership theory development in general. Studies that sample from heterogeneous populations and control for cross-group differences (e.g., industrialization of country led, munificence of organization) are more readily generalizable to individuals outside of positions of “outstanding” influence. Second, content coding should be approached similarly to test item development, with a focus on the generation of items and benchmarks that allow for assessments to be made on objective, behavioral data. Rather than making subjective judgments based on historical data that is often variable in focus, length, and detail, we delineated a set of item generation and scale validation procedures, based on what is available in the broader psychometric literature, that can be used to translate the qualitative information about these leaders into metrics to be used in multivariate statistical techniques to draw valid, reliable inferences. In addition, a plan to gather and measure indices to statistically control internal and external validity concerns is encouraged given the variability found in historical source data in general. Third, leadership historiometry requires multiple steps, types of personnel (e.g., PI, SMEs, coders), and resources (e.g., biographies, external criteria). A detailed project plan should be defined early in the process. If PIs are aware of sequence and logical ordering of these steps, however, these studies, like other paradigms in the social sciences, can take less than a year to execute from start to finish. It is our hope that defining these methodological best practices in detail for this present effort will facilitate other leadership researchers to use and share this technique in their own programs of research. However, before embarking on this type of study, a careful review of the advantages and disadvantages associated with it should be evaluated. In the following section, we outline a set of pros and cons for using historiometry to study leadership. 7. Evaluation of the technique for assessing leadership: advantages and disadvantages Like any empirical technique, historiometry comes with pros and cons for the study of leadership. There are five noteworthy concerns for this type of research paradigm: 1) generalizability of findings, 2) empirical rigor and reliance upon secondary sources, 3) heterogeneity of performance measures, 4) intensity of labor involved, and 5) inability to draw causal inferences. Before discussing the advantages, we will turn to considerations associated with each of these oft-cited criticisms. 7.1. Generalizability of findings First, by definition historiometric techniques require examination of leaders whose influence is historically notable. As such, they tend to work in pinnacle leadership roles in domains such as government (e.g., Barack Obama), corporate (e.g., Warren Buffett), military (e.g., Dwight Eisenhower), and non-profit and/or religious settings (e.g., Indira Gandhi). This poses a problem on at least two fronts. First, the problems which they face in leadership roles are likely more complex, ill-defined, and critical than what might be expected in other leadership roles. Thus, the skills required to execute them might be unique to those instances (resulting in low base rates of occurrence in general). While this is certainly a concern when generalizing from studies on more normative leaders to eminent leader populations, it causes less concern as understanding phenomenon in higher-ordered settings can help to infer what behaviors, albeit perhaps lower levels of such behaviors, might prove useful for leadership in general. For example, in the study on outstanding collegiate and NFL football coaches, we found the expression of ideological leadership was related to performance of teams at both levels. While leading a high school football team may not require the same degree of leadership skill and exact same behaviors, it is a logical assumption that coaches at these more conventional levels of leadership power would benefit from using some of the same techniques to motivate their teams and coaches (e.g., reflection on past great teams as exemplars for performance). Another external validity concern revolves around the personnel surrounding these leaders; because these leaders are likely to have direct reports or TMTs who are also comprised of high level leaders, it is difficult to infer that their performance 1128 G.S. Ligon et al. / The Leadership Quarterly 23 (2012) 1104–1133 (e.g., sustainability of influence) can solely be attributed to the leader alone. This may not be a similar condition when extending findings to other types of more normative leaders, where a different set of techniques might be critical for performance. For example, it is well-established in a host of empirical studies that pragmatic leaders tend to execute their agendas via elites, or key idea champions and experts in their TMT and networks (Mumford & Van Doorn, 2001). It is difficult to generalize those influential behaviors to leaders who might not have access to such high level leaders around them. Thus, some caution is encouraged before drawing conclusions about leadership regarding the nature of how these leaders operate with and through collaborators. 7.2. Empirical rigor and reliance on secondary source material As the unit of analysis requires the use of secondary sources and coding schemes rather than direct observations, the empirical rigor of these studies can be questioned. Given that we are relying upon the social construction of the events leading up to and during such sample members' height of power, we run the risk of making inferences based on the social structures and practices as recounted by the biographers and historical writers rather than more objective, factual accounts from history (Boje, 2001; Fairclough, 2003; Mills, 2003). For example, in an effort to appeal to the American Dream, early biographers of Walt Disney — commissioned by Disney himself — recounted a self-made entrepreneur who fostered a climate of congenial creativity and autonomy (Boje, 2001). Academic historians later described a leader who preferred a hierarchical organization, instilled fear in meetings, and aggrandized his own contributions (Mosely, 1985). In essence, there are multiple realities, and they are each subject to the intentions of the author and to some degree the raters (Boje, 2001). This and other examples illustrate that the inferences made are only as valid as the procedures used to sample and assess them. Thus, this criticism is a valid one if special care is not taken with respect to statistical and inferential concerns when sampling, designing coding, and conducting analyses. As noted above, if researchers comport these studies with psychometric principles as a guide (e.g., multiple measures, methods, and people), then this approach can be as rigorous as other paradigms to gather similar inferences. As Simonton (2010) noted, “the scientific rigor of historiometric research is often underappreciated. Not only can these measures display quite respectable reliability coefficients, but in addition the application of advanced multivariate and quasi-experimental designs can help make some inferences more secure” (p. 282). 7.3. Contextual influences on performance measures A related criticism of using historiometry to examine leadership lies in the social construction of performance measures. Specifically, what is judged as effective leadership is bound by situation, time, and other contextual factors (Mills, 2003). Moreover, the dominant view at the time of the analysis might impact the events, characterizations of the individuals' roles in them, and even the net valence of a given leader's impact. In addition, because these leaders tend to manage large, complex organizations, it is difficult to draw direct linkages to specific behaviors and performance outcomes. We have advocated applying controls to measure the influence of contextual variables such as time, region, munificence, and gender of the leader, but more work needs to be done to ascertain a covariate list specific to application of historiometry to leaders, and particularly in reference to the research question at hand. While we have argued for the need to obtain hard, objective data when possible through historiometry, we cannot lose sight of what subjectivity and richness affords, especially when sampled carefully. As such, we advocate combining more subjective accounts of performance (e.g., author descriptions of performance) with more objective indicators (e.g., number of institutions established), as the social construction of what was outstanding performance can be a useful indicator referent to other leaders in different time periods (Boje, 2001). One example of the benefits of utilizing subjective and objective criteria can be found in a study conducted on the performance contributions of eminent leaders of scientific endeavors (Mumford et al., 2005). Here we examined the academic obituaries of 600 eminent leaders in social, biological, health, and natural scientific domains. When examining the relationship between subjective biographical accounts of performance and more objective indices (e.g., publication counts, impact factors, patent applications), we found author “bias” (i.e., amount of positive hyperbolic descriptors of a given scientist's work contributions by the obituary author) was related to objective performance indices in a given field than it was to other contextual factors (e.g., time period, obituary author credentials). Most interesting is that whereas field norms differed in terms of objective performance indices (e.g., number of publications differed for eminent scientists across natural versus social sciences), the obituary authors' subjective accounts of leader contributions tended to be more stable across fields. Thus, it seems that subjective accounts, instantiated in what the dominant view of “performance” was at a given place and time, may provide the most valuable source of performance information to compare across domains (e.g., government, for-profit business, military) and time. 7.4. Intensity of labor required There are three time-intensive steps associated with the procedures. First, sample selection and gathering requires the use of external judges (e.g., SMEs), an internal logistics team (e.g., graduate students), and multiple trips to the library. For example, in Mumford's (2006) sample of 120 leaders, originally 180 leaders were identified to fit the parameters of interest. Preliminary research was then gathered about each of these individuals from multiple team members to assign them CIP categories. After SMEs (an external team of productive leadership scholars) assessed each of these exemplars, the sample was reduced to 120 G.S. Ligon et al. / The Leadership Quarterly 23 (2012) 1104–1133 1129 leaders. However, in an effort to gain reliable information, two and sometimes three academic biographies were gathered for each of these individuals. This part of the process — the first step in executing a series of studies examining outstanding leadership — lasted three months and required 10 h per week of three doctoral students. Second, rater training is an often overlooked step that dramatically increases the reliability of judge coding. This step not only requires personnel to develop the training, but also a minimum of two weeks for judges to go through the training. Finally, coding of the data, while it does have benefits for raters (e.g., learning about leaders and applying well-established metrics to observe them), can be laborious and time consuming for rater teams. Because we recommend rotating judges across different sets of variables, this step can be tricky to execute if not planned out thoughtfully. For the one year duration of the studies in Mumford's (2006) book on outstanding leaders, twenty-four undergraduates and eleven graduate students were used as coding teams for the various study predictor, control, and criterion data. While other empirical paradigms, particularly studying intact teams in applied settings, can require similar types of investment, it cannot be underemphasized that leadership historiometry is indeed labor intensive if done correctly. 7.5. Inability to draw causal inferences As noted in Section 6 on analyses, the data that can be drawn from this type of technique is correlational in nature. As a result, relationships that can be observed may indeed be caused by spurious, third-level variables. When conducting these studies, it is critical to measure and statistically control confounding variables to provide some safeguards around this. However, similar to other methodological approaches in the field of leadership (e.g., Hunter et al., 2007), it remains that leadership historiometry is a technique that allows inferences about patterns of behaviors, events, and characteristics as they related to varying leadership styles and performance. The five issues identified are important to consider, but at least four advantages make historiometric research a highly viable paradigm for our field: 1) theory testing, 2) real world generalizability, 3) availability of leadership observations, and 4) replicability of procedures. 7.6. Theory testing The most important contribution of historiometric techniques is the capacity to test theory. Given a rigorous, theory-driven sampling procedure, historiometric techniques afford comparison between leader types, nations, organizations, time periods, and domains in ways that augment other approaches in our field. For example, in an effort to assess variables accounting for the most variance in presidential rankings, Simonton (2001) sampled U.S. Presidents and applied a coding scheme to assess degree of scandal while in office, presence of war, heroism in war, and intellectual brilliance. Because he examined this across time periods, he was able to adequately test the robustness of his theories of greatness across all members of the sample of interest: U.S. Presidents. It is our argument, then, that historiometric techniques provide the best way to examine the robustness of our theories of leadership across a host of settings. 7.7. Real world generalizability Because historiometry is not conducted on convenience samples or in a laboratory, it allows us to examine leader behavior in complex environments over variable conditions. And while we sacrifice some of the internal validity that assignment to carefully designed conditions allows, the examination of leaders in actual organizations over time allows for inferences to be made to other leader situations more readily. In addition, this type of methodology allows us to have case exemplars for illustrating leadership principles in training and academic settings. As case-based training has been linked to enhanced learning from novices, historiometric techniques might be a particularly fruitful approach for generalizing leadership principles and illustrating points to students with actual behaviors. For example, the database on developmental events allows fast access of events depicting important characteristics (e.g., presence of spirituality in seminal life events of ideological leaders like the one in Table 3 on Fidel Castro) to use in illustrating similarities and differences among leadership theories (Ligon et al., 2008). 7.8. Availability of leadership observations Life history data of eminent leaders are replete with copious objective markers regarding leadership predictors, controls, and criteria. These markers range from descriptions of educational events that transpired, emotional expression in personal letters and artifacts, to use of the word “I” as an indicator of self-aggrandizement in describing accomplishments (Deluga, 1997). In short, this type of research allows scholars to study variables (e.g., aggression, war heroism) that may not be accessible in other paradigms, given the sensitivity and rarity of their occurrence. In addition, because they are recorded in actual leadership settings, their expression might have more range than one might expect in a laboratory or post hoc self-report study. 7.9. Replicability Because a best practice in preparation of materials is to record and transcribe leadership data with precision, the type of database afforded can yield exact replication of procedures and extension of findings. For example, after conducting our study on 1130 G.S. Ligon et al. / The Leadership Quarterly 23 (2012) 1104–1133 developmental influences on leadership performance (Ligon et al., 2008), Bedell and colleagues were interested to see how important developmental influences predicted a range of other leader behaviors, such as problem solving, political tactics, and leader–member exchange (Mumford, Bedell, et al., 2006). Given the accessibility of our database and events, she was able to conduct a study with these new dependent variables measures with ease, following the procedures we had outlined. In addition, Simonton's leader profiles have been used in several studies examining U.S. Presidents (McCann, 1992; Simonton, 1981, 1986a,b). Thus, given the historical nature of sample material, this technique allows for enhanced and exact replication of results. 8. Conclusions and future directions Before exploring the conclusions and future directions of leadership historiometry, it is important to note that there are at least three limitations to this present effort. First, we chose to focus on a particular application of quantitative analysis of qualitative data — that of outstanding individuals. Another related instantiation of this technique, for example archival data analysis (Asal & Rethemeyer, 2008; Hepworth & West, 1988), may also be a useful tool in our repertoire to advance leadership theory. While the majority of the procedures outlined would also fit archival analysis, one is cautioned about the quality of information archival techniques afford. While historiometric research, and specifically the sampling approach we recommended, focus on coding data from experts in their particular fields, archival data analysis often employ open-source documents such as news reports, website posts, and even material from the organization itself (Sageman, 2008). When reading the reference section of an academic biography, it becomes difficult to equate such source data to that found in some archival studies. For example, in McCullough (2001). In addition, McCullough has received numerous investigative writing awards, such as the Pulitzer Prize for his 1993 book, Truman. The use of such rigorous source information in sampling in historiometric studies may be one reason why many of the internal validity controls prove non-significant (Mumford, 2006). Thus, if employing archival analysis on perhaps non-outstanding sample leaders (e.g., emails, speeches from company presidents), a copious set of controls should be applied in order to be sure findings are not due to some spurious third variable present in this type of non-verified data. Second, we chose to propose a set of techniques, and specifically coding techniques rooted in psychometric theory, that the present authors have found to be successful in this type of examination (Hunter et al., 2011; Ligon et al., 2008; Mumford, 2006). Other techniques also provide useful, albeit different, types of information in biographic analysis. The application of validated measures to assess personality characteristics (Simonton, 2006) or number of objective events (Simonton, 1997) provides rich information about outstanding leaders. For example, in a time series analysis of eminent leaders in Japan, Simonton (1997) coded the number of times such leaders traveled outside the country to examine the relationship between cognitive complexity and exposure to outside influence. In addition, Simonton (2001). Whereas the approach we recommended, relying upon coding multiple data-points and then reducing that data via multivariate analyses such as discriminant-function analysis, allows for the examination of patterns of interacting markers related to performance, Simonton's approach facilitates obtaining important, discrete markers that can be assessed independently to predict specific outcomes. His rigorous, thoughtful techniques have been outlined extensively and should also be considered when deciding which approach to implement (Simonton, 2010). Finally, while the intent of the present effort was to clearly specify the procedures needed for leadership historiometric research, there are some areas we have not been as explicit in describing: Q-Sort, discriminant function, or time series analysis. However, we have made an effort to provide the key resources we have found useful in conducting this part of our studies. For example, the Sage Series in Quantitative Applications in Social Sciences has published editions on Q-Methodology (McKeown & Thomas, 1988) and Discriminant Analysis (Klecka, 1980) that have proven invaluable in our work in this area. Each of these resources provides specific details of the assumptions needed, analytic interpretations, and data requirements to execute them. In addition, Simonton (1990, 2010) has published chapter-length pieces on best practices related to his approach, and specifically for collecting time series data in historiometry. Despite these and other limitations, the present effort has implications for leadership research. First, while there are numerous leadership theories in general, few of them have been empirically tested on outstanding leaders. We are not arguing that this technique should usurp traditional lab approaches or survey research, but instead hope that historiometry will be another approach to examining the complex issues underlying high level leadership. We also hope this submission can aid researchers interested in pursuing such endeavors. Related, another conclusion from this effort is that historiometric research is a multi-step, labor-intensive approach, but also one that is relatively straightforward once executed. If sampling and content coding plans are designed to obtain a comprehensive examination of the formative and height of power years of eminent individuals, resultant data can provide the basis for a significant part of one's research program in leadership for multiple studies. Given the richness of historiometric data and the rigor associated with data collection, there are at least four future directions we hope to see employed. First, historiometric designs should be incorporated in validating leadership theory. For example, models such as authentic leadership and complexity theory could be applied to eminent samples to better understand how these leadership styles relate to performance, as well as how they differ from existing theory. Second, recent advancements in text analysis should be investigated in conjunction with this approach. For example, Abe (2011) used LIWC, a text analysis software program, to count changes in specific phrases or words in speeches of Alan Greenspan during different economic periods. In addition, Bligh and Hess (2007) used a content analysis program called Diction 5.0 to assess the degree of certainty, optimism, immediacy, and jargon on communications from Alan Greenspan during the same period. Using this technology to assess frequencies of other types of psychological markers in larger, historiometric samples should be a way to assess the robustness of these and other findings. We also recommend cross-validating existing content coding schemes (e.g., traditional G.S. Ligon et al. / The Leadership Quarterly 23 (2012) 1104–1133 1131 coder teams) with such automated technologies to determine what kinds of research questions automated software can and cannot reliably answer. Finally, it is still unclear which internal and external validity concerns impact our capacity to draw inferences. Thus, our recommendation is to collect those that we have delineated in this manuscript. However, this takes a considerable amount of time to gather (e.g., ascertaining information about author education/training for each component of source material), and sometimes even doubles the project timeline depending on how many sources per sample member are required for predictor and criterion information. Thus, if historiometry is to become a tool used across a more diverse set of research questions and samples, it would be beneficial to investigate which specific controls are most critical to covary out and for what types of criteria. As our hope is that this type of research becomes more readily accessible to our field through this effort, further investigation of methodological advances such as automated text coding and scaling covariate controls should help make historiometry a more commonly used, widely understood approach to studying leadership in general. Acknowledgements Thanks to Mike Mumford who taught us just about everything we know on this method, even if doing so required far too many weekend trips to the University of Oklahoma Library for all of us. References Abe, J. A. (2011). Changes in Alan Greenspan's language use across the economic cycle: A text analysis of his testimonies and speeches. Journal of Language and Social Psychology, 30, 212–223, http://dx.doi.org/10.1177/0261927X10397152. Aldenderfer, M. S., & Blashfield, R. K. (1984). Cluster Analysis (Sage University Paper Series on Quantitative Applications in the Social Sciences, series no. 007–044). Beverly Hills, CA: Sage. Asal, V., & Rethemeyer, R. K. (2008). The nature of the beast: Organizational structures and the lethality of terrorist attacks. The Journal of Politics, 70, 437–449. Asal, V., & Rethmeyer, R. K. (2011). Wicked web: Social network analysis for combating Islamic and Arab terrorist networks. START Research Review, 12–13. Avolio, B. J., & Bass, B. M. (1999). Re-examining the components of transformational and transactional leadership using the Multifactor Leadership Questionnaire. Journal of Occupational and Organizational Psychology, 72, 441–462. Baldoni, J. (2005). How great leaders get great results. New York: McGraw-Hill. Bass, B. M. (1985). Leadership and performance beyond expectations. New York: Free Press. Bass, B. M., Avolio, B. J., & Goodheim, L. (1987). Biography and the assessment of transformational leadership at the world-class level. Journal of Management, 13, 7–19. Bass, B. M., & Farrow, D. L. (1977). Quantitative analyses of biographies of political figures. Journal of Psychology, 97, 281–296. Bedell, K., Hunter, S., Angie, A., & Vert, A. (2006). A historiometric examination of Machiavellianism and a new taxonomy of leadership. Journal of Leadership & Organizational Studies, 12, 50–72, http://dx.doi.org/10.1177/107179190601200404. Bedell-Avers, K., Hunter, S. T., Angie, A. D., Eubanks, D. L., & Mumford, M. D. (2009). Charismatic, ideological, and pragmatic leaders: An examination of leader– leader interactions. The Leadership Quarterly, 20, 299–315, http://dx.doi.org/10.1016/j.leaqua.2009.03.014. Bligh, M. C., & Hess, G. D. (2007). The power of leading subtly: Alan Greenspan, rhetorical leadership, and monetary policy. The Leadership Quarterly, 18, 87–104, http://dx.doi.org/10.1016/j.leaqua.2007.01.001. Boje, D. M. (2001). Narrative methods for organizational and communication research. London: Sage. Broderick, F. L. (1959). W. E. B. Dubois: Negro leader in a time of crisis. Stanford, CA: Stanford University Press. Cell, C. P. (1974). Charismatic heads of state: The social context. Cross-Cultural Research, 9, 255–305. Chadha, Y. (1999). Gandhi: A life. John Wiley & Sons, Inc.: United States of America. Coe, K., Domke, D., Bagley, M. M., Cunningham, S., & Van Leuven, N. (2007). Masculinity as political strategy: George W. Bush, the "War on Terrorism," and an echoing press. Journal of Women, Politics & Policy, 29, 31–55, http://dx.doi.org/10.1300/J501v29n01-03. Cox, C. (1926). The early mental traits of three hundred geniuses. Stanford, CA: Stanford University Press. Crozier, B. (1973). De Gaulle. New York: Charles Scribner's Sons. Day, D. V. (2000). Leadership development: A review in context. The Leadership Quarterly, 11, 581–613. DeChurch, L. A., Burke, C., Shuffler, M. L., Lyons, R., Doty, D., & Salas, E. (2011). A historiometric analysis of leadership in mission critical multiteam environments. The Leadership Quarterly, 22, 152–169. DeGregorio, W. A. (1991). The complete book of U.S presidents (2nd ed.). New York: Dembner Books. Deluga, R. J. (1997). Relationship among American presidential charismatic leadership, narcissism, and rated performance. The Leadership Quarterly, 8, 49–66. Deluga, R. J. (1998). American presidential proactivity, charismatic leadership, and rated performance. The Leadership Quarterly, 9, 265–292. Deluga, R. J. (2001). American presidential Machiavellianism: Implications for charismatic leadership and rated performance. The Leadership Quarterly, 12, 339–363. Dienesch, R. M., & Liden, R. C. (1986). Leader–member exchange model of leadership: A critique and further development. Academy of Management Review, 11, 618–634, http://dx.doi.org/10.5465/AMR.1986.4306242. Dubofsky, M., & Van Tine, W. (1986). John L. Lewis: A biography. Urbana, IL: University of Illinois Press. Eisenhardt, K. M. (1989). Building theories from case study research. Academy of Management Review, 14, 247–250. Emrich, C. G., Brower, H. H., Feldman, J. M., & Garland, H. (2001). Images in words: Presidential rhetoric, charisma, and greatness. Administrative Science Quarterly, 46, 527–557. Etheredge, L. S. (1978). Personality effects on American foreign policy, 1898–1968: A test of interpersonal generalization theory. American Political Science Review, 78, 434–451. Eubanks, D. L., Antes, A. L., Friedrich, T. L., Caughron, J. J., Blackwell, L. V., Bedell-Avers, K. E., et al. (2010). Criticism and outstanding leadership: An evaluation of leader reactions and critical outcomes. The Leadership Quarterly, 21, 365–388, http://dx.doi.org/10.1016/j.leaqua.2010.03.003. Fairclough, N. (2003). Analyzing discourse: Textual analysis for social research. London: Routledge. Fiol, C. M., Harris, D., & House, R. (1999). Charismatic leadership: Strategies for effecting social change. The Leadership Quarterly, 10, 449–482. Gardner, H. (1993). Creating minds: An anatomy of creativity seen through the lives of Freud, Einstein, Picasso, Stravinsky, Eliot, Graham, and Gandhi. New York, NY US: Basic Books. Giambatista, R. C. (2004). Jumping through hoops: A longitudinal study of leader life cycles in the NBA. The Leadership Quarterly, 15, 607–624. Gibbert, M., Ruigrok, W., & Wicki, B. (2008). What passes as a rigorous case study? Strategic Management Journal, 29, 1465–1474, http://dx.doi.org/10.1002/smj.722. Graen, G. B., & Uhl-Bien, M. (1998). Relationship-based approach to leadership: Development of leader–member exchange (LMX) theory of leadership over 25 years: Applying a multi-level multi-domain perspective. In F. Dansereau, & F. J. Yammarino (Eds.), Leadership: The multi-level approaches (pp. 103–134). Stamford, Connecticut: JAI Press. Hennessee, J. (1999). Betty Friedan: Her life. New York: Random House. 1132 G.S. Ligon et al. / The Leadership Quarterly 23 (2012) 1104–1133 Hepworth, J. T., & West, S. G. (1988). Lynchings and the economy: A time-series reanalysis of Hovland and Sears 1940. Journal of Personality and Social Psychology, 51, 239–247. Holmes, J., & Elder, R. E. (1989). Our best and worst presidents: Some possible reasons for perceived performance. Presidential Studies Quarterly, 19, 529–557. House, R. J., & Howell, J. M. (1992). Personality and charismatic leadership. The Leadership Quarterly, 3, 81–108, http://dx.doi.org/10.1016/1048-9843(92)90028-E. House, R. J., Spangler, W. D., & Woycke, J. (1991). Personality and charisma in the U.S. presidency: A psychological theory of leader effectiveness. Administrative Science Quarterly, 36, 364–396, http://dx.doi.org/10.2307/2393201. Hunter, S. T., Bedell-Avers, K. E., & Mumford, M. D. (2007). The typical leadership study: Assumptions, implications, and potential remedies. The Leadership Quarterly, 18, 435–446. Hunter, S. T., Bedell-Avers, K. E., & Mumford, M. D. (2009). Impact of situational framing and complexity on charismatic, ideological and pragmatic leaders: Investigation using a computer simulation. The Leadership Quarterly, 20, 383–404. Hunter, S. T., Cushenbery, L., Thoroughgood, C., & Ligon, G. S. (2011). First and ten leadership: A historiometric investigation of the CIP leadership model. The Leadership Quarterly, 22, 70–91. Jauch, L. R., Osborn, R. N., & Martin, T. N. (1980). Structured content analysis of cases: A complementary method for organizational research. Academy of Management Review, 5, 517–527, http://dx.doi.org/10.5465/AMR.1980.4288954. Kaarbo, J., & Beasley, R. K. (1999). A practical guide to the comparative case study method in political psychology. Political Psychology, 20, 369–391, http://dx.doi.org/10.1111/0162-895X.00149. Kempster, S. J. (2006). Leadership learning through lived experience: A process of apprenticeship? Journal of Management & Organization, 12, 4–22. Kempster, S. J. (2009). How managers have learnt to lead: Exploring the development of leadership practice. UK: Palgrave Macmillan. Kenney, P. J., & Rice, T. (1988). The contextual determinants of presidential greatness. Presidential Studies Quarterly, 18, 161–169. Klecka, W. R. (1980). Discriminant Analysis (Sage University Paper Series on Quantitative Applications in the Social Sciences, series no. 07–044). Beverly Hills, CA: Sage. Kynerd, T. (1971). An analysis of presidential greatness and “presidential rating”. Southern Quarterly, 9, 309–329. LeBreton, J. M., & Senter, J. L. (2008). Answers to 20 questions about interrater reliability and interrater agreement. Organizational Research Methods, 11, 815–852. Ligon, G. S., & Hunter, S. T. (2010). Putting the development into experiential development. Industrial and Organizational Psychology, 3, 28–32. Ligon, G. S., Hunter, S. T., & Mumford, M. D. (2008). Development of outstanding leadership: A life narrative approach. The Leadership Quarterly, 19, 312–334, http://dx.doi.org/10.1016/j.leaqua.2008.03.005. Ligon, G. S., Leahy, Z., Versella, M., Troyan, C., Gibson, P., Hunter, S. T., et al. (2010). Structure differences between violent and non-violent ideological organizations. Poster presented at the 2010 Society of Industrial and Organizational Psychology Annual Conference, Atlanta, GA. Lilienfeld, S. O., Waldman, I. D., Landfield, K., Watts, A. L., Rubenzer, S., & Faschingbauer, T. R. (2012). Fearless dominance and the U.S. presidency: Implications of psychopathic personality traits for successful and unsuccessful political leadership. Journal of Personality and Social Psychology, http://dx.doi.org/10.1037/a0029392 (Advance online publication). Marmor-Lavie, G., & Weimann, G. (2005). Measuring emotional appeals in Israeli election campaigns. International Journal of Public Opinion Research, 18, 318–339. McAdams, D. P. (2006). The redemptive self: Stories Americans live by. New York, NY: Oxford University Press. McCall, M. W. (2010). Recasting leadership development. Industrial and Organizational Psychologist, 3, 3–21. McCann, S. J. H. (1992). Alternative formulas to predict the greatness of U.S. Presidents: Personological, situational, and zeitgeist factors. Journal of Personality and Social Psychology, 62, 469–479. McCann, S. H. (2005). Longevity, Big Five personality factors, and health behaviors: Presidents from Washington to Nixon. Journal of Psychology, 139, 273–286. McCullough, D. (2001). John Adams. NY, NY: Simonton & Schuster. McKelvey, B., Mintzberg, H., Petzinger, T., Prusak, L., Senge, P., & Shultz, R. (1999). The gurus speak: Complexity and organizations. Emergence: A Journal of Complexity Issues in Organizations and Management, 1, 73–91. McKeown, B. F., & Thomas, D. (1988). Q methodology: Sage University Paper series on Quantitative Applications in the Social Sciences, No. 66–001. Beverly Hills and London: Sage Publications. Miller, N. L., & Stiles, W. B. (1986). Verbal familiarity in American nomination acceptance speeches and inaugural addresses (1920–1981). Social Psychology Quarterly, 49, 72–81. Mills, S. (2003). Michel Foucault. New York: Routledge. Moghal, M. (2010). Idi Amin: Lion of Africa. Central Milton Keynes: Author House. Mosely, L. (1985). Disney's World. New York: Stein and Day. Mumford, M. D. (2006). Pathways to outstanding leadership: A comparative analysis of charismatic, ideological, and pragmatic leaders. New Jersey: Lawrence Erlbaum Associates, Inc. Mumford, M. D., Bedell, K. E., Hunter, S. T., Espejo, J., & Boatman, P. R. (2006). Problem-solving. In M. D. Mumford (Ed.), Pathways to outstanding leadership: A comparative analysis of charismatic, ideological, and pragmatic leadership. Mahwah, NJ: Erlbaum Press. Mumford, M. D., Connelly, M., Scott, G., Espejo, J., Sohl, L. M., Hunter, S. T., et al. (2005). Career experiences and scientific performance: A study of social, physical, life, and health sciences. Creativity Research Journal, 17, 105–129, http://dx.doi.org/10.1207/s15326934crj1702&3_1. Mumford, M. D., Espejo, J., Hunter, S. T., Bedell-Avers, K. E., Eubanks, D. L., & Connelly, S. (2007). The sources of leader violence: A comparison of ideological and non-ideological leaders. The Leadership Quarterly, 18, 217–235, http://dx.doi.org/10.1016/j.leaqua.2007.03.005. Mumford, M. D., Gaddis, B., Licuanan, B., Ersland, B., & Siekel, K. (2006). Communication strategies. In M. D. Mumford (Ed.), Pathways to outstanding leadership: A comparative analysis of charismatic, ideological, and pragmatic leadership. Mahwah, NJ: Erlbaum Press. Mumford, M. D., Gaddis, B., Strange, J. M., & Scott, G. (2006). General method. In M. D. Mumford (Ed.), Pathways to outstanding leadership: A comparative analysis of charismatic, ideological, and pragmatic leadership. Mahwah, NJ: Erlbaum Press. Mumford, M. D., Scott, G., & Hunter, S. T. (2006). Theory. In M. D. Mumford (Ed.), Pathways to outstanding leadership: A comparative analysis of charismatic, ideological, and pragmatic leadership. Mahwah, NJ: Erlbaum Press. Mumford, M. D., Stokes, G. S., & Owens, W. A. (1990). Patterns of life history: The ecology of human individuality. Hillsdale, NJ England: Lawrence Erlbaum Associates, Inc. Mumford, M. D., Strange, J. M., Gaddis, B., Licuanan, B., & Scott, G. (2006). Performance. In M. D. Mumford (Ed.), Pathways to outstanding leadership: A comparative analysis of charismatic, ideological, and pragmatic leadership. Mahwah, NJ: Erlbaum Press. Mumford, M. D., & Threlfall, V. (1992). Quantifying genius. Psyccritiques, 37, 216–218, http://dx.doi.org/10.1037/031362. Mumford, M. D., & Van Doorn, J. R. (2001). The leadership of pragmatism: Reconsidering Franklin in the age of charisma. The Leadership Quarterly, 12, 279–309, http://dx.doi.org/10.1016/S1048-9843(01)00080-7. Murray, E. (1928). An historiometric study of the early traits of great orators. The Quarterly Journal of Speech, 14, 502–508. National Consortium for the Study of Terrorism and Responses to Terrorism (START) (2012). Global Terrorism Database. Retrieved from. http://www.start.umd. edu/gtd Neal, M. W., & Tansey, R. (2010). The dynamics of effective corrupt leadership: Lessons from Rafik Hariri's political career in Lebanon. The Leadership Quarterly, 21, 33–49. O'Connor, J., Mumford, M. D., Clifton, T. C., Gessner, T. L., & Connelly, M. (1995). Charismatic leaders and destructiveness: An historiometric study. The Leadership Quarterly, 6, 529–555, http://dx.doi.org/10.1016/1048-9843(95)90026-8. Osterlind, S. J. (1998). Constructing test items: Multiple-choice, constructed-response, performance, and other formats (2nd edition). Boston, MA: Kluwer. Panagopoulos, C. (2004). Boy talk/girl talk: Gender differences in campaign communication strategies. Women & Politics, 26, 131–155. Perret, G. (2002). Jack: A life like no other. New York: Random House, Inc. Pillemer, D. B. (2001). Momentous events and the life story. Review of General Psychology, 5, 123–134. Rubenzer, S. J., Faschingbauer, T. R., & Ones, D. S. (2000). Assessing the U.S. presidents using the revised NEO Personality Inventory. Assessment, 7, 403–420, http://dx.doi.org/10.1177/107319110000700408. Sageman, M. (2008). Leaderless Jihad. Philadelphia, Pennsylvania: University of Pennsylvania Press. G.S. Ligon et al. / The Leadership Quarterly 23 (2012) 1104–1133 1133 Scheuer, M. (2011). Osama bin Laden. New York, New York: Oxford University Press, Inc. Schriesheim, C. A., Castro, S. L., & Cogliser, C. C. (1999). Leader–member exchange (LMX) research: A comprehensive review of theory, measurement, and data-analytic practices. The Leadership Quarterly, 10, 63–114. Shawcross, W. (1997). Murdoch. New York, NY: Touchstone. Simonton, D. K. (1975). Sociocultural context of individual creativity: A transhistorical time series analysis. Journal of Personality and Social Psychology, 32, 1119–1133, http://dx.doi.org/10.1037/0022-3514.32.6.1119. Simonton, D. K. (1976). Biographical determinants of achieved eminence: A multivariate approach to the Cox data. Journal of Personality and Social Psychology, 33, 218–226, http://dx.doi.org/10.1037/0022-3514.33.2.218. Simonton, D. K. (1980). Land battles, generals, and armies: Individual and situational determinants of victory and casualties. Journal of Personality and Social Psychology, 38, 110–119. Simonton, D. K. (1981). Presidential greatness and performance: Can we predict leadership in the White House? Journal of Personality, 49, 306–323. Simonton, D. K. (1983). Intergenerational transfer of individual differences in hereditary monarchs: Genetic, role-modeling, cohort, or sociocultural effects? Journal of Personality and Social Psychology, 44, 354–364, http://dx.doi.org/10.1037/0022-3514.44.2.354. Simonton, D. K. (1984). Leaders as eponyms: Individual and situational determinants of monarchal eminence. Journal of Personality, 52, 1–21. Simonton, D. K. (1985). The vice-presidential succession effect: Individual or situational basis? Political Behavior, 7, 79–99. Simonton, D. K. (1986a). Presidential personality: Biographical use of the Gough Adjective Checklist. Journal of Personality and Social Psychology, 51, 149–160. Simonton, D. K. (1986b). Dispositional attributions of (presidential) leadership: An experimental simulation of historiometric results. Journal of Experimental Social Psychology, 22, 389–418, http://dx.doi.org/10.1016/0022-1031(86)90042-9. Simonton, D. K. (1987). Presidential inflexibility and veto behavior: Two individual-situational interactions. Journal of Personality, 55, 1–18. Simonton, D. K. (1988a). Presidential style: Personality, biography, and performance. Journal of Personality and Social Psychology, 55, 928–936, http://dx.doi.org/ 10.1037/0022-3514.55.6.928. Simonton, D. K. (1988b). Quality and purpose, quantity and chance. Creativity Research Journal, 1, 68–74, http://dx.doi.org/10.1080/10400418809534289. Simonton, D. K. (1988c). Galtonian genius, Kroeberian configurations, and emulation: A generational time-series analysis of Chinese civilization. Journal of Personality and Social Psychology, 55, 230–238, http://dx.doi.org/10.1037/0022-3514.55.2.230. Simonton, D. K. (1990). Creativity and wisdom in aging. In J. E. Birren, K. Schaie, J. E. Birren, & K. Schaie (Eds.), Handbook of the psychology of aging (pp. 320–329). (3rd ed.). San Diego, CA US: Academic Press. Simonton, D. K. (1991). Latent-variable models of posthumous reputation: A quest for Galton's G. Journal of Personality and Social Psychology, 60, 607–619. Simonton, D. K. (1992). Gender and genius in Japan: Feminine eminence in masculine culture. Sex Roles, 27, 101–119. Simonton, D. K. (1997). Foreign influence and national achievement: The impact of open milieus on Japanese civilization. Journal of Personality and Social Psychology, 72, 86–94, http://dx.doi.org/10.1037/0022-3514.72.1.86. Simonton, D. K. (1998a). Historiometric methods in social psychology. European Review of Social Psychology, 9, 267–293. Simonton, D. K. (1998b). Mad King George: The impact of personal and political stress on mental and physical health. Journal of Personality, 66, 443–466. Simonton, D. K. (2001). Predicting presidential performance in the United States: Equation replication on recent survey results. Journal of Social Psychology, 141, 293–307. Simonton, D. K. (2003). Qualitative and quantitative analyses of historical data. Annual Review of Psychology, 54, 617–640. Simonton, D. K. (2006). Presidential IQ, openness, intellectual brilliance, and leadership: Estimates and correlations for 42 U.S. Chief Executives. Political Psychology, 27, 511–526, http://dx.doi.org/10.1111/j.1467-9221.2006.00524.x. Simonton, D. K. (2008). Childhood giftedness and adulthood genius: A historiometric analysis of 291 eminent African Americans. Gifted Child Quarterly, 52, 243–255. Simonton, D. K. (2010). Emotion and composition in classical music: Historiometric perspectives. In P. N. Juslin, J. A. Sloboda, P. N. Juslin, & J. A. Sloboda (Eds.), Handbook of music and emotion: Theory, research, applications (pp. 347–366). New York, NY US: Oxford University Press. Spangler, W. D., & House, R. J. (1991). Presidential effectiveness and the leadership motive profile. Journal of Personality and Social Psychology, 60, 439–455, http://dx.doi.org/10.1037/0022-3514.60.3.439. Stephenson, W. (1935). Technique of factor analysis. Nature, 136, 295–305. Strange, J. M. (2004). Exceptional leaders and their followers: An in-depth look at leader follower interactions. Unpublished Dissertation. Strange, J. M., & Mumford, M. D. (2002). The origins of vision: Charismatic versus ideological leadership. The Leadership Quarterly, 13, 343–377, http://dx.doi.org/ 10.1016/S1048- 9843(02)00125-X. Suedfeld, P., & Bluck, S. (1988). Changes in integrative complexity prior to surprise attacks. Journal of Conflict Resolution, 32, 626–635. Suedfeld, P., & Rank, A. (1976). Revolutionary leaders: Long-term success as a function of changes in conceptual complexity. Journal of Personality and Social Psychology, 34, 169–178, http://dx.doi.org/10.1037/0022-3514.34.2.169. Suedfeld, P., Tetlock, P. E., & Ramirez, C. (1977). War, peace and integrative complexity: UN speeches on the Middle East problem, 1947–1976. Journal of Conflict Resolution, 31, 123–132. Szulc, R. (1986). Fidel: A critical portrait. New York: William Morrow and Company. Tetlock, P. E. (1979). Identifying victims of groupthink from public statements of decision makers. Journal of Personality and Social Psychology, 37, 1313–1324. Tetlock, P. E. (1981). Personality and isolationism: Content analysis of senatorial speeches. Journal of Personality and Social Psychology, 41, 737–743. Tetlock, P. E. (1983). Cognitive style and political ideology. Journal of Personality and Social Psychology, 45, 118–126. Thoemmes, F. J., & Conway, L. (2007). Integrative complexity of 41 U.S. presidents. Political Psychology, 28, 193–226, http://dx.doi.org/10.1111/j.1467-9221.2007.00562.x. Thorndike, E. (1936). Relation between intellect and morality in rulers. The American Journal of Sociology, 42, 321–334. Traub, R. E. (1994). Reliability for the Social Sciences: Theory and Applications, Vol. 3, Thousand Oaks, CA: Sage. Wallace, M. D., Suedfeld, P., & Thachuk, K. L. (1993). Political rhetoric of leaders under stress: Findings from the Gulf crisis. Journal of Conflict Resolution, 37, 94–107. Watson, G. B. (1927). Post mortem mental tests for eminent men. Journal of Educational Psychology, 18, 202–206, http://dx.doi.org/10.1037/h0065262. Weber, M. (1926). The theory of social and economic organizations. New York: Free Press. Wendt, H. W., & Muncy, C. A. (1979). Studies of political character: Factor patterns of 24 U.S. vice-presidents. Journal of Psychology, 102, 125–131. Winter, D. G. (1980). Measuring the motives of southern Africa political leaders at a distance. Political Psychology, 2, 75–85. Winter, D. G. (1987). Leader appeal, leader performance, and the motive profiles of leaders and followers: A study of American presidents and elections. Journal of Personality and Social Psychology, 52, 196–202. Woods, F. A. (1909). A new name for a new science. Science, 30, 703–704. Woods, F. A. (1911). Historiometry as an exact science. Science, 33, 568–574. Yammarino, F. J., Mumford, M. D., Serban, A., & Shirreffs, K. (2012). Assassination and leadership. Unpublished manuscript. Zullow, H. M., & Seligman, M. P. (1990). Pessimistic rumination predicts defeat of presidential candidates, 1900 to 1984. Psychological Inquiry, 1, 52–62.
© Copyright 2026 Paperzz