A Meta-Analytic Review of Quantitative Studies on Emotional Intelligence and Leadership

Ever since Daniel Goleman (1998) maintained that emotional intelligence is the “sine qua non” of leadership, the link between emotional intelligence and leadership has become a topic of widespread interest in leadership research. The overriding focus of the leadership literature has been on emotional intelligence as hypothesized competencies or traits of individual leaders to affect leadership behaviors, effectiveness or emergence (e.g., Leban and Zulauf, 2004; Hawkins and Dulewicz, 2007; Rego et al., 2007). Many studies have found that leaders’ emotional intelligence explains a high proportion of variance in leadership effectiveness and a variety of organizational outcomes (e.g., Carmeli, 2003; Ozcelik et al., 2008). A significant range of literature also provides empirical evidence to support the notion that emotional intelligence is a predictor of transformational leadership (Barbuto and Burbach, 2006; Leban and Zulauf, 2004; Mandell and Pherwani, 2003; Duckett and Macfarlane, 2003). Despite all the evidence supporting the positive value of emotional intelligence on leadership, several studies have produced contradictions regarding the necessity of emotional intelligence for leadership behavior, practices or effectiveness by arguing that more data based on defensible methodologies are needed to prove the validity of the EI/leadership link (Antonakis, 2003; Antonakis et al., 2009; Locke, 2005).


Introduction
Ever since Daniel Goleman (1998) maintained that emotional intelligence is the "sine qua non" of leadership, the link between emotional intelligence and leadership has become a topic of widespread interest in leadership research. The overriding focus of the leadership literature has been on emotional intelligence as hypothesized competencies or traits of individual leaders to affect leadership behaviors, effectiveness or emergence (e.g., Leban and Zulauf, 2004;Hawkins and Dulewicz, 2007;Rego et al., 2007). Many studies have found that leaders' emotional intelligence explains a high proportion of variance in leadership effectiveness and a variety of organizational outcomes (e.g., Carmeli, 2003;Ozcelik et al., 2008). A significant range of literature also provides empirical evidence to support the notion that emotional intelligence is a predictor of transformational leadership (Barbuto and Burbach, 2006;Leban and Zulauf, 2004;Mandell and Pherwani, 2003;Duckett and Macfarlane, 2003). Despite all the evidence supporting the positive value of emotional intelligence on leadership, several studies have produced contradictions regarding the necessity of emotional intelligence for leadership behavior, practices or effectiveness by arguing that more data based on defensible methodologies are needed to prove the validity of the EI/leadership link (Antonakis, 2003;Antonakis et al., 2009;Locke, 2005).
EI constructs have been generalized into two competing models: the ability-based model and the trait-based (or mixed) model (Conte, 2005;Day et al., 2002). The ability-based model conceptualized by Mayer and Salovey (1997) defines emotional intelligence as a type of intelligence reflecting the ability to process emotional information. The trait-based model of emotional intelligence was endorsed by Goleman and Bar-On (Goleman, 1998;Bar-On, 1997); emotional intelligence is defined as a set of non-cognitive attributes, encompassing five broad skill areas: intrapersonal skills, interpersonal skills, adaptability, stress management and general mood (Conte, 2005;Van der Zee and Wabeke, 2004). Previous studies and anecdotal evidence have verified the claim that transformational leadership style could be predicted from trait-based emotional intelligence (Harms and Credé, 2010;Barbuto and Burbach, 2006;Brown and Moshavi, 2005;Mandell and Pherwani, 2003). A number of studies have also been carried out, indicating the ability-based EI to be a determinant of transformational leadership (Daus and Ashkanasy, 2005;Coetzee and Schaap, 2004;Walter and Bruch, 2007). Despite all the above, further doubt has been casted on validation issues with the various EI constructs prior to operational uses in scientific investigations (Cartwright and Pappas, 2009;Landy, 2005;Walter et al., 2011).
With a wide range of literature supporting the significant role of EI in leadership effectiveness and various leadership outcomes, developing EI skills and competencies is thus crucial to leadership development. Empirical studies on interventions of emotional intelligence have provided convincing evidence relating to the positive results of teaching emotional intelligence on individuals' life success and work performance. Accordingly, EI can be learned and developed successfully among employees and managers through welldesigned EI training activities and programmes (Perks and Bar-On, 2010;Meyer et al., 2004;Latif, 2004); however, speculation does exist on the content of such interventions and the nature of EI training outcomes (Zeidner, Roberts, & Matthews, 2002). The issue whether emotional intelligence can be taught is as important as the question of whether emotional intelligence competencies relate to leadership in the research community.
Meta-analytic studies which have examined literature on the relationships between various facets of leadership and emotional intelligence have been found over the period of 2001 to 2010. These synthesized studies, such as Mills' (2009) meta-analysis on whether a consistent, research-driven link can be established between the concept of emotional intelligence and effective leadership, have provided evidence that emotional intelligence may have immense significance and relevance for leadership effectiveness. Another meta-analytic investigation is a dissertation performed by Whitman (2009) on potential process mechanisms that may account for the EI-leadership effectiveness relationship further ascertained that mixed model EI appears to be a better predictor of leadership than ability EI. Similar results were found in Harms and Credé's (2010) meta-analysis containing 62 independent samples derived from existing literature from 1982 to 2009. It was revealed that trait-based (mixed) model of EI has more promise as a predictor of leadership effectiveness than did abilitybased measures. Despite the fact that EI may contribute to successful leadership at some level, results of Harms and Credé's meta-analytic estimate suggested that EI being the core determinant of transformational leadership were overstated. The aforementioned metaanalytic studies which quantitatively reported their findings in terms of standardized effect sizes have continued to spur research interest in whether emotional intelligence had a significant effect on leadership.
Two meta-analytic reviews were found to qualitatively examine the linkage of emotional intelligence and nurse leadership. The two integrative reviews undertaken by Severinsson (2008, 2010) revealed that emotionally intelligent nurse leadership is associated with favorable work climate characterized by resilience, innovation and change. However, controversies do exist in knowledge regarding the nature of EI and the current state of the art in the conceptual development of EI as abilities, skills and personality dispositions Severinsson, 2008, 2010).
Meta-analytical investigation is therefore needed as a common practice to resolve conflicts found in similar research approach (Rosenthal and DiMatteo, 2001;Dixon-Woods et al., 2005). Through systematic review of literature on emotional intelligence and leadership, the study aims at providing a detailed synthesis of what have been investigated and what are in www.intechopen.com need of further exploration in EI and leadership research. The methodology employed in this qualitative meta-analytic review is inspired by Kun and Demetrovics (2010), Gooty et al. (2010) and Walter et al (2011). In the qualitative investigation conducted by Gooty et al. (2010), a selective, qualitative review of affect, emotions, and emotional competencies in leadership studies was adopted to examine theory, methods and quality of affect-based scholarship in leadership. The present study was also informed by a comprehensive attempt by Kun and Demetrovics (2010) who qualitatively reviewed and critically discussed literature on the relationship of emotional disregulation and addictive disorders. In an effort to ascertain whether EI is the sine qua non of leadership, Walter et al. (2011) critically reviewed recent empirical studies aiming at constructively framing EI's role in three criteria of leadership research: leadership emergence, leadership effectiveness and leadership behavior. In a similar vein, the present study follows research procedure employed by Gooty et al. (2010), Kun and Demetrovics (2010) and Walter et al. (2011) in the design and reporting of this systematic review.
The primary objective of the study is to assess research methods/designs and data analysis procedures employed by researchers investigating the relationship between EI and leadership over the period 2001-2010. The contributions of research on the linkage between the two constructs using quantitative approach were reviewed in terms of data analysis techniques, research design, measures used to assess EI, and subject matters in relation to leadership. The purpose was to provide the breadth of knowledge available on the potential utility of EI for predicting leadership effectiveness. The second purpose of the study was to identify research gaps and formulate research questions that could drive further endeavor in exploring topics relevant to the EI-leadership relationship.

Literature search
To provide a composite knowledge of the diversity of the theories, measures, samples, and contexts that have been employed by research studies on emotional intelligence and leadership, this study made extensive search for relevant research articles from the following six academic databases: Academic Search Premier, Show allEducation Research Complete, ERIC, PsycARTICLES, Psychology and Behavioral Sciences Collection, PsycINFO. The focus was on scholarly articles employing quantitative approaches to examine studies exploring the relevance of emotional intelligence to leadership from January 2001 through December 2010. Emotional intelligence and leadership were the two keywords used to execute the electronic search. The electronic search was supplemented by a manual search for further studies found on reference lists shown up by the electronic search. Up to 116 articles were found with topics related to emotional intelligence and leadership. Inclusion and exclusion criteria formulated to identify articles for further examinations were: (1) Papers should be quantitative research studies related to emotional intelligence of leaders, supervisors, or managers in all professions; (2) Papers included should go through peerreview process and published between 2001 to 2010; (3) Articles focusing mainly upon multiple intelligence, general intelligence or social intelligence were excluded; (4) The present analysis is restricted to English language publications. For the purpose of this research, all articles using pure quantitative research methods focusing on the relevance of EI to leadership were reviewed. A total of 43 articles were included in the final data analysis. Major categories being analyzed include: Subject matters investigated in relation to leadership

Statistical/mathematical techniques & research design used
In 1980s', commonly employed data analysis procedures to study educational psychology and educational phenomena were ANOVA, correlation, t-test, chi-square, non-parametric and factor analysis. The trends and shifts of data analysis procedures during the last decades were well summarized by Hsu (2005). In general, there is a continuous drop of experimental research and the trends of frequently used data analysis procedures remain steady. By surveying articles published by the American Educational Research Journal (AERJ), Journal of Experimental Education (JEE) and Journal of Educational Research (JER), Hsu (2005) synthesized that the most frequently used statistical procedures in educational research during 1971 to 1998 in rank order were descriptive statistics, ANOVA/ANCOVA, correlation, regression, t-test, and psychometric statistics. ANOVA, chi-square, correlation, multiple regression and T test were found to be the most frequently employed statistical techniques to assess statistical power in applied psychology and management research published between 1992 and 1994.
To explore current trends of quantitative studies in emotional intelligence and leadership, categorization for plotting statistical techniques was based on reviews of literature on applied psychology, management and educational research designs (Elmore and Woehlke, 1998;Hsu, 2005;Gooty et al. 2010). The coding process included statistical techniques used in each article and categorizing techniques used in each article. Since a single article may employ more than one statistical technique, the total coded techniques would exceed the total number of article reviewed. The focus is on inferential statistical techniques used in conducting research on the relevance of emotional intelligence to leadership; therefore, when an article employed a specific technique more than once as different data analysis procedures, it was coded as one technique. For instance, if a research used 3 independent sample t-tests in one study, it was counted as one statistical technique. However, subsets under a category were treated as separate techniques. Therefore, if an article used both linear regression and hierarchical regression analyses, it was coded as two statistical techniques under the category of regression. Another instance was that if an article employed both one-sample t-test and independent sample t-test, they were counted as two subsets of techniques under the category of t-test.
After coding and categorizing statistical techniques used in articles employing quantitative approach, we found that during the period of 2001 to 2010 the most frequently used technique was correlational analysis; 28 analyses (33.7%) were coded and categorized into that category. Regression analysis ranked the second frequently used techniques, with a total number of 24 analyses (28.9%). Factor analysis ranked the third, with a total of 11 analyses (13.2%), followed by t test (7 analyses; 8.4%), ANOVA/ANCOVA and others (4 analyses respectively; 4.6%), meta-analysis and psychometric theory (2 analyses respectively; 2.3%). The results are presented in Tables 1, & 2. www.intechopen.com  In studying emotional intelligence in relation to leadership, more emphasis has been given to non-experimental research designs than experimental ones. After classifying the 43 research studies included in the final analysis, only 2 out of the 43 (4.5%) were experimental research. Forty-one research studies reviewed in the study (95.0%) employed nonexperimental approach, which obviously outweighed the number of experimental research.
The two experimental exploratory studies confirmed EI can be learned and developed through deliberate training and interventions (Groves et al., 2008;Meyer et al., 2004). Fortyone studies using non-experimental designs are tabulated below in Table 3. Since an article may involve the application of multimethod analysis, the total number of research designs may exceed the total number of articles reviewed in the present study. Among the 40 non-experimental research articles, 47 research designs were found. Correlational predictive (23 entries, 48.9%) and correlational analyses (9 entries, 20.0%) were the most popular research designs, with a total of 32 studies, followed by descriptive study, with a total of 5 studies (10.6%). It is important to note that descriptive statistics was excluded if they were used to present frequencies, percentages, averages of demographic variables; only those involved with analyzing data collected from surveys and reported in findings were counted as studies using descriptive statistics.

EI measures used
As noted in the introduction section, the ability-based model and the trait-based (or mixed) model are the two most prominent theoretical constructs for creating EI measures. Higgs 1999). The ability-based model has been under criticism with low reliability and with slight correlations with cognitive ability; whereas the trait-based (or mixed) model was found to lack discriminant validity and overlap with existing personality trait measures (Conte, 2005;Oboyle et al., 2011).

Researchers in Mayer and
Tables 4 and 5 present a summary of 46 EI measures used by the 43 articles reviewed in the present study. Since one article may use more than one EI measure, the sum of EI measures used is not expected to equal the total number of articles examined in the study. The traitbased (mixed) model of EI appears to be predominant in articles published from 2001 to 2010, with an overall percentage of 39% (18 articles). Twelve entries of EI measures (26%) were classified as ability-based measures of EI, whereas 16 EI measures (35%) were "others". Among all the measures used, the most frequently used ones were the ability-based MSCEIT, the trait-based EQ-i and SUEIT , with 6 entries each (13%), followed by the abilitybased WLEIS (5 entries, 11%   (Hawkins and Dulewicz, 2007;Dulewicz et al., 2005). The LDQ was used to measure 15 leadership competences clustered under the three dimensions of intellectual competencies (IQ), managerial competencies (MQ), and emotional competencies (EQ) of leaders or managers in the two studies. The above two studies provided empirical evidences to support the positive effect of EQ on leadership performance.
New models of EI have emerged during 2001 to 2010. For example, a nonverbal measure of the ability to recognize emotional expressions displayed by others was developed in 2001 by Morand. Psychometric properties of the emotional intelligence self-description inventory (EISDI), a measure developed based on the ability-based model of EI, was examined and improved for operational use with fully employed business students in the United States (Groves et al., 2008). In 2008, a newly developed 16-item measure of EI was validated for use with supervisor-subordinate dyads in Hong Kong. The issue of translation and psychometric validation of a measure before adapted across cultures was investigated in two studies. In 2004, the Trait Meta-Mood Scale (TMMS; Salovey, Mayer, Goldman, Turvey) was cross-culturally validated for use with management students in Hong Kong (Law et al., 2004). In 2010, a standardized forward-backward translation procedure was conducted ensure the quality of the translated version of Emotional Skills Assessment Process (ESAP) developed in the United States before administering it on academic leaders in Taiwan (Tang et al., 2010).

Subject matters investigated in relation to leadership
Leadership was first conceptualized into leadership emergence and leadership effectiveness, the two criteria based on which the bulk of leadership researchers in favor of trait theory have made investigations linking personality traits to leadership. Accordingly, traits within a Big Five framework may yield differential associations with leadership across the study settings (Bono and Judge, 2004;De Hoogh et al., 2005;Judge et al., 2002). In model of the aforementioned studies, subject matters in relation to leadership, or leadership criteria, are used in the present study as an organizing framework to estimate leadership-EI relations. As mentioned earlier, the critical review by Walter et al. (2011) was of assistance in informing the present review in classifying subject matters in relation to leadership. For the purpose of offering more generalizability as to subject matters investigated, quantitative studies reviewed in the present study will be classified based on three distinct leadership criteria employed by Walter el al. (2011): leadership emergence, behavior, and effectiveness. Leadership emergence refers to the degree an individual is viewed by others as a leader when limited information about that www.intechopen.com individual's performance is known. Leadership effectiveness, on the other hand, represents a leader's performance in exerting influence on and giving guidance to the activities toward achievement of goals. Performance of a leader is often measured in terms of team, group, or organizational effectiveness. Studies on leadership behavior are described as predominately focusing on transformational leadership behaviors of idealized influence, inspirational motivation, individualized consideration, and intellectual stimulation. Bernard Bass's assessment tool, the Multifactor Leadership Questionnaire (MLQ), is a typical measure used to assess transformational leadership behaviors. However, several studies could hardly be classified into the three leadership criteria. Examples include the two experimental studies examining the possibilities and impacts of EI interventions, validation studies on measurement tools developed for leaders, and meta-analytic studies on EI's role in leadership. The present study therefore suggests that there is a forth criterion classified as "others" used to accurately code experimental, construct validation and meta-analytic studies on EI-leadership relationships.
Leadership criteria  In Tables 6 and 7, subject matters in relation to leadership coded into the four leadership criteria are presented, showing the number of articles in each criterion and the total percentage of the articles they represent, for each year and overall. Articles coded under the four criteria appear to be evenly distributed throughout the past ten years. The most commonly studied criterion is leadership behavior, with a total of 15 articles (33.3%), followed by leadership effectiveness (12 articles; 26.6%), leadership emergence (9 articles; 20%) and others (9 articles; 20%). Among those coded as others, two meta-analytic studies were found providing syntheses of the literature on the theoretical and empirical basis of emotional intelligence and it's linkage to leadership. The two experimental studies were conducted to support the notion that EI can be learned, enhanced and developed through proper training techniques (Groves et al., 2008;Meyer et al., 2004). New EI constructs or measures were established with a focus on particular groups of leaders, such as servant leadership and managers in Hong Kong (Barbuto and Wheeler, 2005;Wong and Low, 2002). In 2001, one way of measuring emotional intelligence validated by Morand (2001) is to combine psychological with perception measures. It is worth pointing out that Morand, in developing a composite measure of emotional intelligence (2001), used Ekman and Friesen's (1975) The Face of Emotion encompassing a set of 17 photos and Merabian and Epstein's empathy scale to conceptualize a nonverbal perception measure of skill at nonverbal communication to assess individuals' recognitions of emotional expressions displayed by others. This perspective is somewhat in contrast to the prevailing psychological-based Likert-scale EI measurements developed to conceptualize and assess emotional intelligence.

Conclusions
This study identified and tabulated research methods and data analysis procedures from studies on leadership and EI relationships over the past 10 years. In addition, it also identified the types of EI measures used and subject matters in relation to leadership investigated. Percentages of frequently used statistical/mathematical techniques, research designs, types of EI measures used and subject matters focused were presented annually so that their trends and practices can be assessed. The results were synthesized from 43 studies conducted between 2001 to 2010 retrieved from six academic databases.
Correlational analysis, regression analysis and factor analysis were identified as frequently employed techniques in research on leadership-EI research. Correlational predictive and correlational analyses research, two most dominant research designs used, overweight casual-comparative research design using ANOVAs or t tests frequently seen in educational and management studies. In addition, research studies on leadership and EI relations employing quantitative methodologies show preference of non-experimental research designs to experimental ones and this preference has remained steady in the research community.
With regard to EI measured used, the majority of them, based on either ability or traitbased models, are heavily reliant on the assessment of self-perceived EI. Despite the increasing number of particular EI instruments developed during the decade, selfreported approach has been widely adopted in framing the newly created constructs of EI.
Only a few studies reviewed in the present studies incorporate multi-rater assessments, such as LDQ, ECI and WLEIS, to provide a more accurate and complete evaluation of an individual's EI competences (Hawkins and Dulewicz, 2007;Wang and Huang, 2009;Barbuto and Wheeler, 2006). The vast dominance of self-reported measures developed for use in exploring the links between leadership and EI may be seen as problematic since socially desirable self and self-knowledge may cause faking goods, which in turn would bring distortion.

www.intechopen.com
With regard to leadership criteria researched, the research community has more or less neglected the subject of leadership development, such as the effects of incorporating EI training in leadership development programs. The impact of EI interventions on leadership development is still at the exploratory stage. The two experimental studies reviewed in the present study have thus far provided empirical evidence suggesting that EI can be heightened among groups of student samples. More research efforts are needed to demonstrate whether EI abilities can be taught, learned, and developed among those holding leadership or management roles. In addition, the empirical shortening in the area of potential effects of an EI training program on leaders' work-related outcomes will provide an avenue for further investigation.

Limitations and overall implications
Several important limitations must be considered. First, studies included in the present meta-analytic review are highly selective. Quantitative studies published from 2001 to 2010 that explicitly focused on leadership and emotional intelligence were collected and coded; therefore, many peripheral themes, such as emotional labor and emotional literacy, are left out. Such narrow focus helped clearly delineate a predictor of leadership criterion. To expand the present research effort, further studies can adopt a broad literature search using keyword combinations around leadership (leaders, team leadership, management/leadership effectiveness, management/leadership outcomes, leader-member exchange) and emotional intelligence (emotion, emotional labor, emotional literacy, emotional learning).
The current study is also limited by the focus on elite journals from databases in management, psychology and education. To increase the number of studies and avoid publication bias, future studies may identify possible sources of data via searches of Dissertation Abstracts and Internet searches for additional unpublished data sources. A third limitation of the current study lies in the lack of inter-observer agreement on specific coding variables (Clarke et al., 2002). To ensure a high level of accuracy and coding consensus, coding variables could be performed by experienced experts or trained graduate students working independently on coding sheets provided by the authors (Clarke et al., 2002;Van Rooy and Viswesvaran, 2004).
In 2010, Harms and Crede conducted a meta-analysis on transformational leadership and emotional intelligence and found that the claim that EI serves as a core competence for transformational leadership was overstated. It was noted that the majority of studies used in their investigation were from unpublished sources; the lack of methodological rigor in data source may yield weak results in their findings. Findings from the present analysis on published peer-reviewed quantitative studies lead to the similar conclusion: the 15 research studies coded under the subject area of "leadership behavior" using transformational leadership measures indicate that EI contribute to transformational leadership at some level, but different studies found significant relationships between distinct areas of EI and distinct components of transformational leadership across cultures and research settings (e.g. Corona, 2010;Tang et al., 2010;Polychroniou, 2009;Sunindijo et al., 2007).
In conclusion, there was obviously still a pressing need among the leadership research community for valid experimental research designs in the area of EI interventions for leaders. The potential application of EI as a pedagogical tool into leadership education may offers a new approach to improving various leadership outcomes. It is also recommended that multiple ratings or 360-degree feedbacks on leaders' EI and leadership competences should be applied to provide a more holistic conceptualization of the nature of this relationship. With regard to cross-cultural adaptation of EI or leadership measures, dominant among the EI measures in use are those developed in English and later translated for use with samples in cultures other than English speaking countries. Appropriate translation and rigorous validation process therefore may have detrimental effects on the study results (Wang et al., 2006;Guillemin et.al., 1993;Jones and Kay, 1992). EI measures reviewed in the present studies have been widely applied to Hong Kong, Greece, Portugal, Singapore, Taiwan and Thailand (Law et al., 2004;Polychroniou, 2009;Rego et al., 2007;Tang et al., 2010;Sunindijo et al., 2007). However, there is some lack of clarity regarding appropriate translation and rigorous validation process before those measures were operationally used in other cultures. Trans-cultural validation of EI assessment tools is therefore required in leadership research in order to develop more refined measures of the construct utilized to investigate the full potential of EI on leadership practices in different countries.

Acknowledgement
This study is an extension of a research project sponsored by National Science Council, Taiwan  Emotional intelligence is an emerging construct for applied research and possible interventions, both in scholastic, academic and educational contexts, organizational contexts, as well as at an individual level in terms of people's well-being and life satisfaction. From the presented contributions, it emerges how this volume is characterized by an interest to give an international overview rich of stimuli and perspectives for research and intervention, in relation to a promising variable of current interest, such as emotional intelligence. The goal is that this book further contributes to the affirmation of a particularly promising variable, such as emotional intelligence, which requires a greater interest and attention in both research and application field.

How to reference
In order to correctly reference this scholarly work, feel free to copy and paste the following: Hui-Wen Vivian Tang and Mu-Shang Yin (2012). A Meta-Analytic Review of Quantitative Studies on Emotional Intelligence and Leadership, Emotional Intelligence -New Perspectives and Applications, Prof. Annamaria Di Fabio (Ed.), ISBN: 978-953-307-838-0, InTech, Available from: http://www.intechopen.com/books/emotionalintelligence-new-perspectives-and-applications/a-meta-analytic-review-of-quantitative-studies-on-emotionalintelligence-and-leadership