Journal of

Intelligence

Article
Emotional Intelligence and Transformational Leadership:
Meta-Analysis and Explanatory Model of Female
Leadership Advantage
Ning Hsu 1, *, Daniel A. Newman 2 and Katie L. Badura 3

1 Department of Psychology, Virginia Polytechnic Institute and State University, Blacksburg, VA 24060, USA
2 Department of Psychology, University of Illinois at Urbana-Champaign, Champaign, IL 61820, USA
3 Scheller College of Business, Georgia Institute of Technology, Atlanta, GA 30308, USA
* Correspondence: [email protected]

Abstract: Emotional intelligence is a second-stratum factor of general intelligence (MacCann et al.

2014) that: (a) has been popularly touted as an essential individual difference for effective leadership
(Goleman 1998), but also (b) exhibits large gender group differences favoring women (Joseph and
Newman 2010). Combining these insights, we propose that emotional intelligence is a key mechanism
in the so-called female leadership advantage (Eagly and Carli 2003—which emphasizes the finding
that women are rated slightly higher in transformational leadership compared to men). The current
study seeks to explain this gender leadership gap by specifying three personality-based theoretical
mechanisms that enhance transformational leadership: (a) emotional intelligence (favoring women),
(b) communion (stereotypical femininity; favoring women; Hsu et al. 2021), as well as an offsetting
effect of (c) agency (stereotypical masculinity; favoring men). Meta-analytic data (including original
meta-analyses among the leader’s ability-based emotional intelligence, transformational leadership,
communion, and agency) are used to test our theorized model. Results confirm the full mediation
model of female leadership advantage. Because the three unique mechanisms operate in different
Citation: Hsu, Ning, Daniel A.
directions, their individual indirect effects are notable, but their cumulative indirect effect is small and
Newman, and Katie L. Badura. 2022.
near-zero. In conclusion, we emphasize incorporating emotional intelligence with other personality-
Emotional Intelligence and
based explanations of gender effects in leadership perceptions.
Transformational Leadership:
Meta-Analysis and Explanatory
Keywords: emotional intelligence; female leadership advantage; transformational leadership; gender;
Model of Female Leadership
Advantage. Journal of Intelligence 10:
agency; communion; meta-analysis
104. https://doi.org/10.3390/
jintelligence10040104

Received: 15 September 2022

1. Introduction
Accepted: 10 November 2022
Published: 14 November 2022
The current work investigates the so-called female leadership advantage, with the goal of
offering a theoretical model to explain the personality and individual difference origins
Publisher’s Note: MDPI stays neutral
of gendered leadership advantages. In particular, the model we propose here differs
with regard to jurisdictional claims in
from previous work on personality-based models of gender gaps in leadership (Badura
published maps and institutional affil-
et al. 2018), because we emphasize the incorporation of one underappreciated explanatory
iations.
mechanism—ability-based emotional intelligence (Mayer and Salovey 1997). Ability-based
emotional intelligence is a second-stratum factor of human cognitive ability, reflecting
intelligence in the emotion domain (MacCann et al. 2014). In short, we specify and test a
Copyright: © 2022 by the authors.
model asserting that women are perceived as more transformational leaders in part because
Licensee MDPI, Basel, Switzerland. they exhibit much stronger ability-based emotional intelligence than men, on average.
This article is an open access article En route to demonstrating how ability-based emotional intelligence can play a unique
distributed under the terms and role in the female leadership advantage, we will: (a) summarize the nuanced evidence for
conditions of the Creative Commons the female leadership advantage, with an emphasis on transformational leadership, and
Attribution (CC BY) license (https:// (b) propose and test an integrated personality-based model of the female leadership advan-
creativecommons.org/licenses/by/ tage, specifying the relative roles of three attributes (emotional intelligence, communion,
4.0/). and agency) in explaining the gender gap in transformational leadership. In order to test

J. Intell. 2022, 10, 104. https://doi.org/10.3390/jintelligence10040104 https://www.mdpi.com/journal/jintelligence

J. Intell. 2022, 10, 104 2 of 20

the model, we provide updated meta-analyses of the relationships between ability-based

emotional intelligence, transformational leadership, communion, and agency.

1.1. Female Leadership Advantage

When thinking of great leaders throughout history, many people might think of
Mahatma Gandhi, Abraham Lincoln, Nelson Mandela, and Winston Churchill, to name
but a few. One thing these leaders have in common is that they all happen to be of one
gender: male. Indeed, the historical prototype or stereotype of a great (or good) leader has
typically been male and masculine (Koenig et al. 2011). Such masculine leader stereotypes
could also create a disadvantage for women seeking or performing in leadership roles,
such that it is more difficult for women to become or be perceived as a good leader, due
to the mismatch or incongruity between the role expectations for women and the role
expectations for leaders (Eagly and Karau 2002). Indeed, various discrimination-based
theories of gender gaps in leadership attainment and leader evaluations have received a
great deal of attention (i.e., Think Manager—Think Male: Schein 1973; Role Congruity
Theory: Eagly and Karau 2002; Lack of Fit Theory: Heilman 2001), as these theories are
typically used to explain the underrepresentation or undervaluing of women in leadership
roles (i.e., female leadership disadvantage).
However, the contention that women might, in fact, have a leadership advantage
in contemporary society has also been popularized in recent years. One version of the
female leadership advantage perspective (cf. Eagly and Carli 2003; Vecchio 2002; Yukl
2002) proposes that women have an advantage in leadership because women have more
skills and experiences in interpersonal relationships, inclusive decision making, caregiving,
and power-sharing (Grant 1988; also see Lipman-Blumen 1983; Vecchio 2002). When
authors (e.g., Eagly and Carli 2003; Eagly 2007) refer to the female leadership advantage,
they often cite meta-analytic empirical evidence of various kinds. This evidence is briefly
reviewed below.

1.1.1. Gender Differences in Leadership Style

Eagly and Johnson (1990) meta-analyzed the gender differences in various leadership
styles, including interpersonal style, task style, interpersonal-task style continuum, and
democratic-autocratic style continuum. Task style refers to a leadership style with an
orientation to task accomplishment, such as coordinating activities to complete assigned
tasks; whereas interpersonal style refers to a leadership style with a socioemotional orienta-
tion, focusing on the maintenance of interpersonal relationships as well as the morale and
welfare of the group members (Bales 1950). Task style and interpersonal style were most
famously developed in the Ohio State studies of leader behavior, where these two factors
are, respectively, known as initiating structure and consideration (e.g., Halpin 1957; Halpin
and Winer 1957; Hemphill and Coons 1957; Stogdill 1963). Eagly and Johnson (1990) found
that on average, women exhibited very slightly higher interpersonal style than men did
(d = .04, CI [.01, .07], k = 136), while women exhibited an identical level of task style as men
did (d = .00, CI [−.03, .03], k = 139). Relatedly, when meta-analyzing the set of primary
studies that measured interpersonal style and task style on a single interpersonal-task
style continuum, men and women scored similarly (d = −.03, CI [−.10, .03], k = 31). The
largest gender effect found in Eagly and Johnson’s (1990) meta-analysis demonstrated that
women exhibited a more democratic style of leadership (on the democratic-autocratic style
continuum) than men on average (d = .22, CI [.15, .29], k = 23; Eagly and Johnson 1990).
In general, however, the gender gaps in leadership styles found in Eagly and Johnson’s
meta-analysis were almost always near-zero.
The construct validity of Eagly and Johnson’s (1990) meta-analysis has also raised
concerns (Vecchio 2002). First, the gender effects were averages across a wide range of
measures, including the Leader Behavior Description Questionnaire (LBDQ; Stogdill 1963;
Stogdill et al. 1962), Leadership Effectiveness and Adaptability Description (LEAD; Hersey
and Blanchard 1977, 1982), Leadership Opinion Questionnaire (Fleishman 1953, 1957, 1960),
J. Intell. 2022, 10, 104 3 of 20

Organizational Climate Description Questionnaire (Halpin 1966), and so on; with these
instruments all designed to assess different constructs. Eagly and Johnson’s (1990) meta-
analysis also included measures such as the Least Preferred Co-Worker scale (Fiedler 1967)
that have not shown adequate convergent validity with more rigorously validated and
established measures of leadership style (e.g., LBDQ; Stogdill 1963; Stogdill et al. 1962).
Second, this meta-analysis was largely based on leaders’ self-ratings (197 out of the total
370 primary studies included in the meta-analysis used leaders’ self-ratings of leadership),
and Eagly and Johnson (1990) did not report the gender effects separately by rating source
(e.g., reporting gender effects for self-rating vs. other-rating separately). Therefore, it is
possible the results from Eagly and Johnson (1990) could have been largely driven by
leaders’ self-stereotype or social desirability, instead of observed leadership behaviors (as
observed by others).
Along the same lines, a more recent meta-analysis on paternalistic leadership style
(Hiller et al. 2019) shows that the gender gaps in authoritarianism (r = −.01, ρ = −.01,
CI [−.06, .04], k = 11, N = 4385), benevolence (r = −.01, ρ = −.01, CI [−.04, .01], k = 11,
N = 5236), and morality (r = .03, ρ = .03, CI [.00, .07], k = 7, N = 2754) were also close to zero.
Altogether, this evidence supports the contention that the gender differences in leadership
styles may be negligible and near-zero.

1.1.2. Gender Differences in Leadership Effectiveness

Besides the gender differences in leadership styles, research has also looked into
whether there is a gender difference in leadership effectiveness (e.g., Paustian-Underdahl
et al. 2014). Measures of leadership effectiveness included in Paustian-Underdahl et al.’s
(2014) meta-analysis consist of: (a) leader performance, (b) leadership ability, (c) measures
of satisfaction with leaders, (d) measures of satisfaction with leaders’ performance, (e)
coding/counting of effective leadership behaviors, and (f) evaluations of organizational
productivity/team performance. Paustian-Underdahl et al.’s (2014) meta-analysis found
that women are perceived to be slightly, but non-significantly, more effective leaders than
men (overall d = −.05, k = 99, N = 101,676). Importantly, this gender difference is larger and
statistically significant when looking at only other-ratings of leader effectiveness (other-
rating d = −.12, k = 78, N = 96,893). Further, when looking at only self-ratings of leader
effectiveness, male leaders exhibited a higher mean than female leaders (self-rating d = .21,
k = 19, N = 4711). In other words, the direction of the gender gap in leader effectiveness
ratings depended entirely on whether the leader effectiveness was self-reported. However,
the construct validity of Paustian-Underdahl et al.’s (2014) meta-analysis is a potential
source of ambiguity (the same as in all past meta-analyses of leadership effectiveness),
because there is no standardized measurement of leadership effectiveness typically used in
the literature.

1.1.3. Gender Differences in Transformational Leadership

Perhaps the best evidence for an actual female leadership advantage comes from the
domain of transformational leadership (Eagly et al. 2003). Transformational leaders are
defined as leaders who give vision and sense of mission, earn respect and trust (charisma),
convey high expectations and communicate purposes effectively (inspiration), encourage
attentive thinking and problem solving (intellectual stimulation), and offer individual
attention and mentor followers individually (individualized consideration; Bass 1990).
There are several potential reasons why transformational leadership style would be enacted
or perceived to be enacted especially well by women (cf. Eagly and Carli 2003; Yoder
2001). First, transformational leadership behaviors, unlike more traditionally emphasized
leadership behaviors, have sometimes been thought of as more stereotypically androgynous
and/or feminine (not exclusively masculine; Eagly and Carli 2003). This characteristic of
transformational leadership allows women to be more freed from the double-bind created
by the role incongruity between their stereotypical gender roles and their leader roles.
Second, women might be especially adept at executing the more stereotypically feminine
J. Intell. 2022, 10, 104 4 of 20

facets of transformational leadership (e.g., individualized consideration) because of a

spillover effect from their gender roles to their leader roles. Third, women who survived
the double standard and glass ceiling in leader selection in organizations might possess
more leadership skills than their male counterparts (who did not face the same hurdles in
order to be in a leadership position; Eagly et al. 2003; see also Eagly and Carli 2003).
In support of this view, Eagly et al.’s (2003) meta-analysis of gender and transforma-
tional leadership found that, on average, women were rated higher on transformational
leadership than men were (d = −.10, k = 44). This finding of a female leadership advantage
was supported across different measures of transformational leadership: MLQ norming
study (d = −.11); other MLQ studies (d = −.11, k = 26); and studies using other measures (d =
−.09, k = 17). Importantly, these gender effects are larger when only looking at subordinate
ratings (d = −.15, k = 26) compared to leader’s self-ratings of their own transformational
leadership (d = −.06, k = 26). Further, some evidence supports measurement equivalence
between men and women for transformational leadership ratings (Walumbwa et al. 2005),
similar to various other types of leadership (e.g., assessment center leadership ratings,
Anderson et al. 2006; authentic leadership, Caza et al. 2010). Such evidence suggests the
mean difference between men and women in transformational leadership should largely
reflect a true difference in the latent construct of transformational leadership, rather than a
measurement artifact.
Due to the pattern of empirical findings just reviewed, in the current study we will
focus on the gender gap in transformational leadership, as that is the domain where the
female leadership advantage has been most clearly demonstrated (Eagly and Carli 2003).
Specifically, we will begin by focusing on transformational leadership measures that are
not self-reported (i.e., separating self-rated transformational leadership from non-self-rated
transformational leadership). Our initial focus on other-rated leadership is consistent with
common definitions of leadership that emphasize how leaders influence their followers
and others (e.g., Winston and Patterson 2006; Silva 2016).

1.2. Personality-Based Leadership Theories

The theory of leadership categorization (Lord et al. 1984) proposes that people have
implicit theories of leadership, such that they possess certain ideas of what most leaders
are like (prototypes). For example, people find specific traits (e.g., charisma, masculinity,
etc.) to characterize their cognitive schema for leaders (Offermann et al. 1994). This
conceptualization of personality and leadership (Lord et al. 1986) is rooted in a social-
cognitive perspective (e.g., Mischel 1973), which suggests that personality traits should
influence social perceptions because observers tend to use others’ personality traits to help
organize their perceptions of others (Winter and Uleman 1984). According to this view, the
personality traits of leaders should affect how they are perceived as a leader. Zaccaro (2007)
further proposed an integrated model where attributes (e.g., personality and cognitive
abilities, social appraisal and problem-solving skills) operate through leader processes
to predict leadership outcomes (i.e., leader emergence, leader effectiveness, and leader
advancement and promotion). Based on the theory of leadership categorization (Lord et al.
1984) and a model of leader attributes and leader performance (Zaccaro 2007), personality
traits could be one potential reason that helps explain the female leadership advantage—i.e.,
the observed gender gap in transformational leadership. In our attempt to explain the
female leadership advantage, we focus on three traits/individual differences in particular:
emotional intelligence, communion, and agency.

1.2.1. Emotional Intelligence and Leadership

Goleman’s (1995, 1996, 1998; Goleman et al. 2001) best-selling books and most-
downloaded articles have fueled the popularization of emotional intelligence. One reason
for the popularity of this work is Goleman’s (1998) expansive claim that “IQ and technical
skills are important, but emotional intelligence is the sine qua non of leadership” (p. 82)—
i.e., the claim that emotional intelligence is an essential individual attribute for being a
J. Intell. 2022, 10, 104 5 of 20

great leader. To elaborate, Goleman (1998) advanced several components of emotional

intelligence at work—including self-regulation, empathy, and social skill—to advance
his conceptual (and non-empirical) explanation for his claim that emotional intelligence
matters more than general cognitive ability/intelligence. To begin, cognitive function is
worse when people are emotionally upset or disregulated. Further, emotional intelligence
helps people to be more perceptive and attentive to the emotions of others, to be able to
manage conflict, to be able to communicate in a more constructive and helpful way, to be
able to help build a climate where diversity is valued instead of a source of conflict, and to
be able to collaborate with others more effectively (Goleman 1996).
Empirically, emotional intelligence (EI) has been defined and measured in two ways
(Mayer et al. 2000): (a) ability EI, and (b) mixed EI. Ability EI was defined as the ability
to understand emotions and apply emotions and emotional knowledge to improve the
thinking process (Mayer et al. 2008), whereas mixed EI was defined (more broadly) as
noncognitive ability or skill (Baron 1997), socially or emotionally intelligent behaviors
(Bar-On 2004), and emotion-related personality traits (Petrides and Furnham 2003). Much
of the criticism of EI has focused on mixed EI because of its broad definition (EI as “a
grabbag of everything that is not cognitive ability”; Joseph and Newman 2010, p. 72;
Murphy 2006) and its lack of discriminant validity from personality traits (de Raad 2005).
Indeed, meta-analysis shows that mixed EI is largely redundant with a hodgepodge of
other well-known traits including: (a) ability EI, (b) self-efficacy, (c) self-rated performance,
(d) conscientiousness, (e) emotional stability, (f) extraversion, and (g) general mental ability
(Joseph et al. 2015).
In contrast, ability EI has been argued to have a clear theoretical definition and to
therefore be the only conceptualization of EI worthy of future study (Daus and Ashkanasy
2005). In support of ability EI, MacCann et al. (2014) demonstrated empirically that ability
EI fits well within the general hierarchical structure of human cognitive abilities as a
second-stratum factor that marks “the expression of intelligence in the emotion domain”
(p. 358). Ability EI is commonly understood using the Four-Branch Model of EI (Mayer
and Salovey 1997; Salovey and Mayer 1990), where ability EI consists of (a) perceiving
emotion accurately, (b) utilizing emotions to improve thinking processes, (c) making sense
of emotions, and (d) regulating emotions. Given its superior conceptualization as a facet
of actual intelligence, the current study will focus exclusively on the ability EI definition
(i.e., EI measured using performance-based measures—measures that exclusively employ
right-wrong and/or multiple-choice formats, with these tests of ability EI administered to
the leaders themselves).
George (2000) further proposed that the four components of ability EI contribute
to effective leadership. She defined effective leadership using many of the features of
transformational leadership, including (a) providing vision, (b) stimulating intelligence,
(c) motivating confidence, (d) encouraging flexibility, and (e) giving meaning; and she
argued that leaders high in EI would be better equipped with the socioemotional skills
to enact these effective leadership behaviors, because these behaviors are emotion-driven.
Echoing this logic, we propose that ability EI will positively predict transformational
leadership.

Hypothesis 1 (H1). Emotional intelligence has a positive relationship with transformational

leadership (i.e., this hypothesis involves ability-based EI).

1.2.2. Emotional Intelligence and Gender

Given that a major objective of the current study is to explain the mechanisms underly-
ing the female leadership advantage, we note that ability-based EI—which we hypothesized
to predict transformational leadership—is also known to exhibit a large gender gap favor-
ing women (Joseph and Newman 2010). When explaining gender gaps in personality and
individual differences (like the observed gap in ability EI), a useful theoretical framework
is social role theory (Eagly 1987; Wood and Eagly 2012). According to social role theory,
J. Intell. 2022, 10, 104 6 of 20

gender differences in individual attributes are rooted in the gendered division of labor.
Specifically, biological specialization (related to men’s advantage in average upper body
strength and women’s advantage in childcare/nursing) gives rise to gender segregation in
jobs/roles, and this gendered division of labor forms a basis for dispositional inferences
about gender group differences in traits. Social role expectations and gender norms are
then reinforced via the socialization of internal gender identities (Heilman 1983; Eagly and
Wood 2009), as well as social regulation, external pressure or backlash against exhibiting
gender-non-normative attributes (Rudman et al. 2012; Williams and Tiedens 2016; see
review by Hsu et al. 2021).
What are the implications of social role theory for the gender gap in emotional intel-
ligence? To recap, social role theory holds that gendered socialization pressures tend to
correspond to the gendered division of labor in a given society (Wood and Eagly 2002).
As a result of the general tendency toward occupational gender segregation (concen-
tration of women in “domestic work and communally demanding employment”—e.g.,
nursing, teaching, childcare, and other people-oriented occupations; (Su et al. 2009), ob-
servers often “infer that [women] are warm, caring, and socially skilled” (Wood and Eagly
2012, p. 71). However, we contend that these gendered dispositional inferences about
women can be further parsed into two major individual difference components: (a) mo-
tivational/discretionary (“will do”) attributes, and (b) skill- and ability-based (“can do”)
attributes. The prior category (i.e., gendered “will do” attributes) coalesces around the
broad personality trait of communion (see Bem 1974; described in more detail below);
while in the latter category (i.e., gendered “can do” attributes) coalesces around the set
of knowledge, skills, and abilities that facilitate the achievement of communal goals (e.g.,
nurturance, warmth, empathy). Importantly, we contend that this set of gender-normed
individual differences in knowledge, skill, and ability that facilitate communal goals is
well-encapsulated by the concept of ability-based emotional intelligence. In support of
the view that women are socialized into higher mean levels of emotional intelligence (e.g.,
emotion perception, emotion understanding, and emotion regulation) in order to accom-
modate gendered social roles that are often prescribed by the division of labor, Joseph and
Newman (2010) found that women scored substantially higher on tests of ability-based
emotional intelligence than did men (ability-based EI dcorrected = .52, k = 14, N = 2216).
As mentioned earlier, meta-analysis has estimated that women exhibit slightly more
transformational leadership than men, on average (d = −.10, k = 44; Eagly et al. 2003).
Eagly et al. (2003) contend that one reason why women are more transformational leaders
is because women are more communal. However, they did not look at socio-emotional
skills. As such, we offer a novel and complementary potential explanation for the female
transformational leadership advantage—emotional intelligence. Transformational leader-
ship involves more socio-emotional aspects of leadership; thus, women’s higher emotional
intelligence is likely one reason why there is a slight female advantage in transformational
leadership.

Hypothesis 2 (H2). Emotional intelligence mediates the effect of gender (favoring women) on
transformational leadership.

1.2.3. Agency and Communion

In discussions of gender and personality, much of the theoretical and research attention
has involved the two concepts of communion and agency (Bakan 1966; Wiggins 1991).
Communion is a female-associated trait (also called femininity), while agency is a male-
associated trait (also called masculinity), although each of the two traits can be exhibited
by both men and women alike (Bem 1974; Spence and Helmreich 1978; also see meta-
analytic review by Hsu et al. 2021). The tradition of studying gender identity via gender-
stereotypical personality traits became the basis for numerous research studies, in which
communion was characterized using terms such as warmth, affection, and sympathy; while
J. Intell. 2022, 10, 104 7 of 20

agency was characterized using terms such as assertiveness, forcefulness, and self-reliance
(Wood and Eagly 2015; cf. Ma et al. 2022).
Traditionally, agency has been an advantageous trait for leaders to possess. For
example, agency (or masculinity) has been found empirically to be one of the strongest
predictors of both leadership emergence and leadership effectiveness (Lord et al. 1986;
Mann 1959), and this finding has been echoed in research on the related trait of dominance
(Hogan 1978; Judge et al. 2002; Smith and Foti 1998; Stogdill 1948). Indeed, one classic study
(Megargee 1969) showed in a lab experiment that the high-dominance individual in a mixed-
dominance dyad tended to emerge as the leader, unless the high-dominance individual
was a woman paired with a low-dominance man, in which case the man was chosen as
leader. A more recent meta-analysis (Badura et al. 2018) has shown that: (a) women are less
likely to be chosen as a leader, and (b) this gender gap in leader emergence can be largely
explained by agency. In other words, agentic personality can serve as a mediator for gender
effects on leadership.
Wiggins (1991) argues that both agentic (i.e., getting ahead) and communal (i.e., getting
along) traits are essential for individuals to be able to function well in human society. An
individual needs to be agentic enough in order to be in charge and be able to protect their
own resources for survival, while they need to be communal enough in order to maintain
relationships, be popular and be able to avoid sabotage from other people. As such, being
agentic and being communal could both be functional. Similarly, we contend that both
traits can be functional in being perceived as an effective leader. However, agency and
communion are gendered traits that have consistently shown mean gender differences
across time. A recent meta-analysis has shown that, on average, men are more agentic
than women (g = .40, k = 928, N = 254,731) and women are more communal than men
(g = −.56, k = 937, N = 254,465; Hsu et al. 2021). Therefore, gendered traits such as agency
and communion could underlie offsetting gender-leadership effects, which would help
explain the slight gender gap in transformational leadership favoring women. That is, the
agentic leadership advantage for men and the communal leadership advantage for women
could largely cancel each other out, resulting in a very slight female leadership advantage
(Eagly et al. 2003). Altogether, based on the circumplex of interpersonal variables (Wiggins
1991), we propose that both agency and communion would have a positive relationship
with transformational leadership, and that both agency and communion would mediate
the effect of gender on transformational leadership, albeit in opposite directions.

Hypothesis 3 (H3). Communal traits have a positive relationship with transformational leadership.

Hypothesis 4 (H4). Communal traits mediate an effect of gender (favoring women) on transforma-
tional leadership.

Hypothesis 5 (H5). Agentic traits have a positive relationship with transformational leadership.

Hypothesis 6 (H6). Agentic traits mediate an effect of gender (favoring men) on transformational
leadership.

1.3. Integrated Model of Female Leadership Advantage

Finally, by combining the predictions specified in Hypotheses 1 through 6, we propose
an integrated personality-based model of female leadership advantage (see Figure 1). In
this model, we highlight the role of ability-based emotional intelligence as a mechanism for
the female leadership advantage. We also emphasize that the model specifies three distinct
mechanisms of gendered leadership advantage, which operate in differing directions:
(a) a female leadership advantage via emotional intelligence, (b) a female leadership
advantage via communion, and (c) a male leadership advantage via agency. It is the fact
that these effects operate in opposite directions that produces a combined cumulative
female leadership advantage that is small in magnitude.
 J. Intell. 2022, 10, x FOR PEER REVIEW 8 of 22

J. Intell. 2022, 10, 104 via communion, and (c) a male leadership advantage via agency. It is the fact that these 8 of 20
effects operate in opposite directions that produces a combined cumulative female leader-
ship advantage that is small in magnitude.

Emotional
Intelligence

Transformational
Gender Communion
Leadership

Agency

Figure 1. A Personality-Based Model of Female Leadership Advantage.

Figure 1. A Personality-Based Model of Female Leadership Advantage.
2. Materials and Methods
2. Materials andInMethods
order to test the integrated theoretical model of the gender gap in transformational
In orderleadership
to test the(Figure 1), we combined meta-analytic correlations into an overall correlation
integrated theoretical model of the gender gap in transformational
matrix, which then served as the basis for path analysis (i.e., meta-analytic SEM;
leadership (Figure
Viswesvaran and combined
1), we Ones 1995). In meta-analytic correlations
order to compile the into correlation
overall meta-analytic an overall correlation
matrix, we used a combination of published meta-analyses and original meta-analyses.
matrix, which then served as the basis for path analysis (i.e., meta-analytic SEM; Viswes-
varan and Ones 1995).Meta-Analyses
2.1. Original In order to compile the overall meta-analytic correlation matrix, we
used a combination of published
For relevant meta-analyses
bivariate relationships that haveand original
already meta-analyses.
been meta-analyzed, we ex-
tracted effect sizes directly from previous meta-analyses (see Table 1). We also update the
Harms and Credé (2010) meta-analysis when estimating the relationship between ability-
2.1. Original Meta-Analyses
based EI and transformational leadership measured from a different source (for which
For relevant bivariate
they reported relationships
k = 4 primary thatexclusively
studies). We focus have already been meta-analyzed,
here on ability-based EI meas- we ex-
tracted effectured sizeson the leader. To estimate this and other bivariate relationships that had not been
directly from previous meta-analyses (see Table
previously meta-analyzed, we conducted eight original meta-analyses in the current
1). We also update the
Harms and Credé (2010)
study (using meta-analysis
the Schmidt and Hunter when estimating
2015, approach). the relationship
These relationships include (a,between
b) ability-
based EI and performance-based
transformational ability emotional intelligence
leadership measured withfrom
agencyaand communion,
different (c, d, e,
source (for which they
and f) transformational leadership [both self-rated and non-self-rated] with agency and
reported k = 4communion,
primaryand studies). We focus exclusively
(g, h) transformational leadership [bothhere on ability-based
self-rated EI measured on
and non-self-rated]
the leader. Towith estimate this andability
performance-based other bivariate
emotional relationships that had not been previously
intelligence.
meta-analyzed, Tablewe conducted
1. Meta-Analytic eightMatrix
Correlation original meta-analyses
of Gender, in Communion,
Emotional Intelligence, the current study (using the
Agency,
Schmidt and and Hunter 2015, approach).
Transformational Leadership. These relationships include (a, b) performance-based
ability emotional intelligence with1.agency and 2. communion, 3. (c, d,
4. e, and 5.f) transformational
1. Gender leadership [both self-rated and non-self-rated] - with agency and communion, and (g, h)
2. Emotional Intelligence (performance-based abil- −.25 b
transformational
ity EI measures only)
leadership [both self-rated
(14/2,216)
- and non-self-rated] with performance-based
ability emotional intelligence. −.30 c .07 a
3. Communion -
(937/254,465) (5/1,172)
4. Agency .23 c .09 −.0005 a
- d
Table 1. Meta-Analytic Correlation Matrix of Gender, Emotional Intelligence, Communion, Agency,
and Transformational Leadership.

1. 2. 3. 4. 5.
1. Gender -
2. Emotional Intelligence (performance-based ability EI −.25 b
-
measures only) (14/2216)
−.30 c .07 a
3. Communion -
(937/254,465) (5/1172)
.23 c −.0005 a .09 d
4. Agency -
(928/254,731) (5/1172) (554/110,243)
−.09 e .13 a .50 a .27 a
5. Transformational Leadership (non-self report) -
(10/2996) (6/618) (8/779) (6/420)
−.15 e .29 a .36 a .31 a .27 f
6. Transformational Leadership (leader’s self-report)
(10/836) (17/1923) (3/589) (4/820) (23/2784)
Note. Each cell contains ρ (the mean sample size-weighted meta-analytic correlation corrected for unreliability
attenuation in both X and Y, but with no correction for the gender variable), followed by k (number of effect sizes)
and N (total sample size), presented as (k/N). a Original meta-analysis. b Joseph and Newman (2010). c Hsu et al.
(2021). d Badura et al. (2018). e Extracted from Eagly et al. (2003, pp. 576–77). f Lee and Carpenter (2018). Gender:
F = 1, M = 2.
J. Intell. 2022, 10, 104 9 of 20

2.1.1. Literature Search

When conducting meta-analysis, there are two main approaches to addressing con-
struct validity. One approach (the “kitchen sink” approach) is to include a set of primary
studies that employ a wide range of both validated and unvalidated measures, as long as
some claim can be generally made that each measure roughly assesses the target construct
of interest (e.g., Barrick and Mount 1991). A second approach is to focus more exclusively
on only primary studies that used validated measures of the construct of interest (e.g.,
Hurtz and Donovan 2000). In the current study, we chose to employ the latter approach.
As such, to identify primary studies to include in our original meta-analyses, we searched
within papers that cited the commonly used, validated measures of the constructs of in-
terest. These primary studies were located using the Google Scholar ‘search within citing
articles’ function.
For the relationships of emotional intelligence with agency and communion, we used
the keywords “MSCEIT” OR “MEIS” OR “STEU” (i.e., the names of commonly used
performance-based EI measures; to avoid the common term STEM, we only used STEU
as one of our keywords), while searching within papers that cited commonly used and
validated agency and communion measures (the Bem Sex Role Inventory—BSRI; Bem 1974;
and personal attributes questionnaire—PAQ; Spence and Helmreich 1978; Spence et al. 1974,
1975), as done in previous meta-analyses (Hsu et al. 2021; Badura et al. 2018; Twenge 1997).
For the relationships of transformational leadership with agency and communion, we used
the keywords “Multifactor Leadership Questionnaire” OR “Global Transformational Leadership”
OR “Leadership Practices Inventory” OR “Transformational Leadership Inventory” (i.e., the names
of the most commonly used transformational leadership measures) to search within papers
that cited commonly used, validated measures of agency and communion (BSRI; Bem
1974 and PAQ; Spence and Helmreich 1978; Spence et al. 1974, 1975). For the relationship
between transformational leadership and performance-based emotional intelligence, we
used the keywords “MSCEIT” OR “MEIS” OR “STEU” (i.e., the names of commonly used
performance-based EI measures) to search within papers citing commonly used, validated
transformational leadership measures, including the Multifactor Leadership Questionnaire
(MLQ; Bass 1985), Global Transformational Leadership scale (GTL; Carless et al. 2000),
Leadership Practices Inventory (LPI; Posner and Kouzes 1988), and Transformational
Leadership Inventory (TLI; Podsakoff et al. 1990). Overall, 845 results were returned from
the above keyword searches for papers that cited seminal works (i.e., Bem 1974; Spence
and Helmreich 1978; Spence et al. 1974, 1975; Bass 1985; Carless et al. 2000; Posner and
Kouzes 1988; Podsakoff et al. 1990).

2.1.2. Inclusion Criteria

Studies were included in the meta-analysis according to the following rules. First, a
primary study had to report sufficient statistics to calculate an effect size (e.g., correlation
coefficient r) for one of the proposed relationships in the current study. Second, the samples
in the primary studies needed to be adult samples (e.g., undergraduate samples and
working adults). Third, only primary studies written in English were included in the
current meta-analysis.
For the relationships involving emotional intelligence (i.e., the relationships of emo-
tional intelligence with agency, communion, and transformational leadership), a primary
study had to specifically measure the leaders’ EI using a performance-based ability EI
measure (i.e., the MSCEIT; Mayer et al. 2003, MEIS; Mayer et al. 1999, and STEU/STEM;
MacCann and Roberts 2008). That is, the leader needed to have taken a test of performance-
based ability EI (i.e., a multiple-choice test that contains right and wrong response options).
For the relationships involving transformational leadership (i.e., the relationships of trans-
formational leadership with agency, communion, and emotional intelligence), a primary
study had to measure transformational leadership using one of the validated transforma-
tional leadership measures, including the Multifactor Leadership Questionnaire (MLQ;
Bass 1985), Global Transformational Leadership scale (GTL; Carless et al. 2000), Leadership
 measure (i.e., the MSCEIT; Mayer et al. 2003, MEIS; Mayer et al. 1999, and STEU/STEM;
MacCann and Roberts 2008). That is, the leader needed to have taken a test of perfor-
mance-based ability EI (i.e., a multiple-choice test that contains right and wrong response
options). For the relationships involving transformational leadership (i.e., the relation-
ships of transformational leadership with agency, communion, and emotional intelli-
J. Intell. 2022, 10, 104 gence), a primary study had to measure transformational leadership using one of the val-10 of 20
idated transformational leadership measures, including the Multifactor Leadership Ques-
tionnaire (MLQ; Bass 1985), Global Transformational Leadership scale (GTL; Carless et al.
2000), Leadership Practices Inventory (LPI; Posner and Kouzes 1988), and Transforma-
Practices InventoryInventory
tional Leadership (LPI; Posner(TLI;and Kouzes
Podsakoff et 1988), and
al. 1990). Transformational
Applying these criteriaLeadership
resulted In-
ventory
in the(TLI; Podsakoff
inclusion of five et al. 1990).
effect Applying
sizes for these criteria
the EI-Agency resulted
relationship (N =in1172),
the inclusion
five effectof five
effect sizes
sizes for for
the the EI-Agency relationship
EI-Communion relationship (N = 1172), six
= 1172), fiveeffect
effectsizes
sizes
forfor
thethe EI-Communion
Agency-TFL
(non-self report)
relationship relationship
(N = 1172), (N =sizes
six effect 420), for
eight effect
the sizes for the
Agency-TFL Communion-TFL
(non-self (non-
report) relationship
(N self report)
= 420), relationship
eight effect sizes(Nfor
= 779), six effect sizes for the
the Communion-TFL EI-TFL (non-self
(non-self report) relation-
report) relationship (N = 779),
six ship
effect(Nsizes
= 618),
for four effect sizes
the EI-TFL for thereport)
(non-self Agency-TFL (self report)
relationship relationship
(N = 618), (N =sizes
four effect 820),for the
three effect sizes for the Communion-TFL (self report) relationship (N = 589), and 17 effect
Agency-TFL (self report) relationship (N = 820), three effect sizes for the Communion-TFL
sizes for the EI-TFL (self report) relationship (N = 1,923). A summary of the literature
(self report) relationship (N = 589), and 17 effect sizes for the EI-TFL (self report) relationship
searches and exclusion/inclusion process (which adheres to PRISMA and APA guidelines)
(N can
= 1923). A summary of the literature searches and exclusion/inclusion process (which
be seen in Figure 2.
adheres to PRISMA and APA guidelines) can be seen in Figure 2.

Primary studies identified through forward searching of seminal papers in

Google Scholar and full-text articles assesses for eligibility
(k = 845)

Primary studies excluded

(k = 806)

Non-English (k = 154)
Child sample (k = 0)
Did not report statistics on relationships of interest (k = 360)
Did not measure variables of interests using measurement
instruments included in the current study (i.e., BSRI and PAQ
for Agency and Communion, MSCEIT, MEIS, STEU/STEM
for EI, MLQ, TLI, GTL, and LPI for TFL; k = 130)
Duplicate (k = 151)
No full-text access (k = 11)

Primary studies/samples included in quantitative synthesis (meta-analysis)

EI-Agency Relationship = 5 primary studies/samples

EI-Communion Relationship = 5 primary studies/samples
Agency-TFL (non-self report) Relationship = 6 primary studies/samples
Communion-TFL (non-self report) Relationship = 8 primary studies/samples
EI-TFL (non-self report) Relationship = 6 primary studies/samples
Agency-TFL (self report) Relationship = 4 primary studies/samples
Communion-TFL (self report) Relationship = 3 primary studies/samples
EI-TFL (self report) Relationship = 17 primary studies/samples

Figure 2. Flow Chart for Meta-Analytic Literature Search.

Figure 2. Flow Chart for Meta-Analytic Literature Search.

3. Results
3.1. Eight Original Meta-Analyses
Meta-analytic results for the EI-TFL relationship are reported in Table 2. Supporting
Hypothesis 1, EI was positively related to TFL (non-self-reported): (r = .12, ρ = .13, k = 6,
N = 618, 95% CI [.04, .20]). When looking at self-reported transformational leadership, EI
was positively related to self-reported TFL (r = .25, ρ = .29, k = 17, N = 1923, 95% CI [.08, .50]).
Meta-analytic results for the communion-TFL and agency-TFL relationships are reported in
Tables 3 and 4, respectively. Supporting Hypothesis 3, communion was positively related
to TFL (non-self-reported): (r = .44, ρ = .50, k = 8, N = 779, 95% CI [.32, .57]). Agency had a
positive, but not statistically significant relationship with TFL (non-self-reported): (r = .20,
ρ = .27, k = 6, N = 420, 95% CI [−.06, .45]). When looking at self-reported transformational
leadership, both communion (r = .30, ρ = .36, k = 3, N = 589, 95% CI [.11, .50]) and agency
(r = .25, ρ = .31, k = 4, N = 820, 95% CI [.11, .39]) were positively related to self-reported
TFL. Meta-analytic results for the communion-EI and agency-EI relationships are reported
in Table 5. Both communion (r = .05, ρ = .07, k = 5, N = 1172, 95% CI [−.06, .16]) and
agency (r = −.0009, ρ = −.0005, k = 5, N = 1172, 95% CI [−.05, .05]) did not have statistically
significant relationships with ability-based EI.
J. Intell. 2022, 10, 104 11 of 20

Table 2. Meta-analytic results for Emotional Intelligence and Transformational Leadership.

k N r ρ SDρ 95% CI 80% CV

Emotional Intelligence-Transformational leadership 28 2953 .20 .23 .16 [.14, .26] [.03, .43]
Rater
Self-Rated TFL 17 1923 .25 .29 .16 [.17, .33] [.08, .50]
Other-Rated TFL 6 618 .12 .13 .02 [.04, .20] [.11, .16]
Subordinate 5 472 .15 .17 .00 [.06, .23] [.17, .17]
Mix of other raters (i.e., supervisor, peer,
1 146 .03 .03 .00 [.03, .03] [.03, .03]
and subordinate)
Mixed (e.g., self mixed with other) 5 412 .08 .09 .03 [−.02, .18] [.05, .13]
Sample type
Organizational sample 26 2419 .20 .24 .18 [.13, .27] [.01, .46]
Other sample 2 534 .17 .21 .00 [.15, .20] [.21, .21]
MBA working student sample 1 375 .16 .19 .00 [.16, .16] [.19, .19]
Mixed sample (i.e., including both
1 159 .20 .23 .00 [.20, .20] [.23, .23]
students and nonstudents)
Type of industry
Banking 1 138 .08 .09 .00 [.08, .08] [.09, .09]
Education 7 521 .11 .13 .00 [.04, .18] [.13, .13]
Welfare Compensation and Job Search Activities 1 102 .28 .32 .00 [.28, .28] [.32, .32]
Manufacturing 1 133 .08 .09 .00 [.08, .08] [.09, .09]
Hospitality 1 142 .22 .24 .00 [.22, .22] [.24, .24]
Religious Organizations 1 27 .01 .01 .00 [.01, .01] [.01, .01]
Mixed 8 703 .27 .32 .30 [.09, .46] [−.06, .71]
Sample country
US 10 937 .14 .16 .07 [.07, .21] [.07, .25]
Other 12 1430 .22 .26 .21 [.11, .33] [−.01, .53]
Publication type
Published 7 824 .15 .18 .07 [.07, .23] [.09, .27]
Unpublished 23 2333 .22 .26 .17 [.15, .29] [.04, .47]
Note. k = number of effect sizes in the meta-analysis; N = total sample size in the meta-analysis; r = uncorrected
correlation; ρ = correlation corrected for attenuation in the predictor and criterion; SDρ = standard deviation of
the corrected correlation; CI = confidence interval around r; CV = credibility interval around ρ.

3.2. Moderator Analyses

When testing continuous moderators (i.e., date of study and age of sample), we
found no statistically significant results (all p > .05; n.s.). Further, when testing categorical
moderators, we only interpreted moderators when at least 3 primary studies were available
at each level of the moderator (see Tables 2–4). Only 4 moderator tests were statistically
significant. We found that the EI-TFL relationship was weaker when TFL was rated by
others (r = .12, ρ = .13), compared to when TFL was rated by the leaders themselves (r = .25,
ρ = .29; ∆ρ = .16; 95% CI [.02, .26], p < .05; see Table 2). The communion-TFL relationship
was weaker in US samples (r = .27, ρ = .31), compared to non-US samples (r = .47, ρ = .53;
∆ρ = −.22, 95% CI [−.47, −.01], p < .05; see Table 3), while the agency-TFL relationship
was stronger in US samples (r = .36, ρ = .43), compared to non-US samples (r = .10, ρ = .13;
∆ρ = .30, 95% CI [.10, .54], p < .05; see Table 4). Finally, the agency-TFL relationship was
weaker in published papers (r = .15, ρ = .21), versus unpublished papers (r = .36, ρ = .43;
∆ρ = .22, 95% CI [.06, .47], p < .05; see Table 4).
J. Intell. 2022, 10, 104 12 of 20

Table 3. Meta-analytic results for Communion and Transformational Leadership.

k N R ρ SDρ 95% CI 80% CV

Communion-Transformational leadership 11 1368 .38 .44 .20 [.27, .50] [.19, .69]
Rater
Self-Rated TFL 3 589 .30 .36 .18 [.11, .50] [.13, .59]
Other-Rated TFL 8 779 .44 .50 .19 [.32, .57] [.26, .75]
Subordinate 8 779 .44 .50 .19 [.32, .57] [.26, .75]
Sample type
Organizational sample 10 1154 .37 .43 .21 [.24, .50] [.15, .70]
Student sample 1 214 .45 .51 .00 [.45, .45] [.51, .51]
Type of industry
Banking 2 76 .62 .67 .00 [.59, .65] [.67, .67]
Education 2 116 .31 .33 .30 [−.12, .75] [−.06, .71]
Mixed 5 862 .35 .41 .21 [.18, .52] [.14, .68]
Sample country
US 4 548 .27 .31 .22 [.06, .49] [.03, .59]
Other 5 628 .47 .53 .00 [.41, .52] [.53, .53]
Publication type
Published 7 820 .46 .53 .12 [.36, .55] [.37, .68]
Unpublished 4 548 .27 .31 .22 [.06, .49] [.03, .59]
Note. k = number of effect sizes in the meta-analysis; N = total sample size in the meta-analysis; r = uncorrected
correlation; ρ = correlation corrected for attenuation in the predictor and criterion; SDρ = standard deviation of
the corrected correlation; CI = confidence interval around r; CV = credibility interval around ρ.

3.3. Mediator Analyses

To test the hypothesized mediating relationships, we conducted path analyses in
LISREL 10.1 using the meta-analytic correlation matrix shown in Table 1, with the minimum
sample size N. We chose to use the minimum sample size when running the path analyses
in order to be conservative (Badura et al. 2020), because of the considerable difference in the
sample sizes among the meta-analytic correlations, and the fact that several of the proposed
mediator bivariate relationships had modest sample sizes (ranging from 420 to 1923; see
Table 1). As a supplemental analysis, all path analyses and mediation tests were also run
using the harmonic mean N, and the direction and statistical significance of the results were
identical to the results reported here using the minimum N. Six theoretical models were
run and compared (Model 1a compared with 1b, and 2a compared with 2b, 2c, and 2d;
see Table 6). Model comparisons were made on the basis of change in CFI, with a model
difference being interpreted as meaningful when ∆CFI > .01 (Cheung and Rensvold 2002).
The hypothesized full mediation model (Model 1a: shown in Figure 1) displayed good
overall fit (χ2 (df=1) = 1.327, RMSEA = .016, CFI = 1.00, NNFI = .996, SRMR = .006). Model 1b
(partial mediation, including the direct effect from gender to TFL) was a saturated model
with zero degrees of freedom, and therefore displayed perfect fit by design. We compared
Model 1a against Model 1b using change in CFI (Cheung and Rensvold 2002) and retained
Model 1a as the better model. Path coefficients for Model 1a are shown in Figure 3. All
hypothesized paths were statistically significant in the expected directions.
J. Intell. 2022, 10, 104 13 of 20

Table 4. Meta-analytic results for Agency and Transformational Leadership.

k N R ρ SDρ 95% CI 80% CV

Agency-Transformational leadership 10 1,240 .23 .30 .25 [.10, .37] [−.02, .61]
Rater
Self-Rated TFL 4 820 .25 .31 .15 [.11, .39] [.12, .51]
Other-Rated TFL 6 420 .20 .27 .37 [.−06, .45] [−.20, .74]
Subordinate 6
J. Intell. 2022, 10, x FOR PEER REVIEW
420 .20 .27 .37 [.−06, .45] [−.20, .74]14 of 22
Sample type
Organizational sample 9 1026 .21 .27 .27 [.06, .36] [−.07, .61]
Student sample 1 and Other
Model 2a b,c: 2 DVs (Self 214
TFL): .36 .42 .00 [.36, .36] [.42, .42]
Type of industry Gender-EI-Communion-Agency (Full 2.513 2 -- .0247 .999 -- .990 .0127
Banking Mediation) 2 76 .43 .49 .00 [.42, .43] [.49, .49]
Education Model 2b b,d: 2 DVs (Self
1 and Other52TFL): .42 .46 .00 [.42, .42] [.46, .46]
Mixed (add direct path from Gender
4 to Other-
567 .322.073 .42 1 .44 .16 .0506 [.19,
.997
.45] .002 .958 .63] .0116
[.21,
rated TFL)
Sample country Model 2c b,e: 2 DVs (Self and Other TFL):
US 3
(add direct path from Gender 484
to Self- .36.434 .43 1 2.08.00 .0000 [.33, .43] .001
1.000 [.43,
1.022.43] .0052
Other rated TFL) 5 564 .10 .13 .25 [−.10, .31] [−.19, .45]
Publication type Model 2d : 2 DVs (Self and Other TFL):
b,f
0
Published (add 2 direct paths from
7 Gender to Self-
756 .21 0
.15 model) 2.51.29 .0000 [−.03,
1.000.34] .001 1.000
[− .17, .58].0000
(saturated
Unpublished and Other-rated TFL) 3 484 .36 .43 .00 [.33, .38] [.43, .43]
Note. Models judged to fit best are in boldface. ΔCFI > .01 (Cheung and Rensvold 2002). Analyses
Note. k = number of effect sizesoninminimum
based the meta-analysis; N = total
N = 420. a Includes sample
only size
non-self in the
report r = uncorrected
meta-analysis;leadership
transformational as the DV.
correlation; ρ = correlation corrected
b Includes bothfor attenuation
non-self in the predictor
and self-report and criterion;
transformational SDρas=two
leadership standard deviation
DVs. c Does of
not include
the corrected correlation; any
CI =direct
confidence interval
paths from to anyr; transformational
around
gender CV = credibilityleadership
interval around
outcomeρ.variables. d Includes a
direct path from gender to non-self-report transformational leadership. e Includes a direct path from
gender to self-report transformational leadership. f Includes direct paths from gender to both non-
Table 5. Meta-analyticself
results for Emotional
and self-report Intelligence
transformational with Agency and Communion.
leadership.

k N TheR hypothesized SDρ

ρ full mediation model95% CI 1a: shown in80%
(Model CV1) displayed
Figure
good overall fit (χ2(df=1) = 1.327, RMSEA = .016, CFI = 1.00, NNFI = .996, SRMR = .006). Model
Emotional intelligence
1b (partial mediation, including the direct effect from gender to TFL) was a saturated
Communion 5 1172 model.05 .07 .15 [−.06, .16] [−.12, .25]
with zero degrees of freedom, and therefore displayed perfect fit by design. We
Agency 5 1172 compared−.0009Model−1a against Model 1b using change in CFI (Cheung and−Rensvold
.0005 .00 [ − .05, .05] [ − .0005, .0005] 2002)
Note. k = number of effect and retained
sizes Model 1a as the
in the meta-analysis; N better
= totalmodel.
samplePath coefficients
size in for Modelr =1auncorrected
the meta-analysis; are shown in
Figure
correlation; ρ = correlation 3. Allfor
corrected hypothesized
attenuationpaths
in thewere statistically
predictor significant
and criterion; SDρin= the expected
standard directions.
deviation of
the corrected correlation; CI = confidence interval around r; CV = credibility interval around ρ.

Figure 3. Path Model Figure

Results3. Path Model
for the Results for the Personality-Based
Personality-Based Model of FemaleModel ofLeadership
Female Leadership Advantage
Advantage
(Model 1a). Note. Standardized path coefficients (β’s) are presented. N2 = 420; χ2 (1) = .434; RMSEA =
(Model 1a). Note. Standardized
.000, CFI = path
1.000; coefficients
NNFI = 1.023;(β’s)
SRMR are * p < .05.NTransformational
presented.
= .006; = 420; χ (1) =leadership
.434; RMSEA =
non-self-re-
.000, CFI = 1.000; NNFIported.
= 1.023; SRMR
Female = 1, = = 2.* p < .05. Transformational leadership non-self-reported.
.006;
Male
Female = 1, Male = 2.
J. Intell. 2022, 10, 104 14 of 20

Table 6. Fit statistics for alternative theoretical models.

Models χ2 df ∆χ2 RMSEA CFI ∆CFI NNFI SRMR

Model 1a a,c :
Gender-EI-Communion-Agency-TFL
.434 1 – .0000 1.000 – 1.023 .0061
(Full Mediation)
Model 1b a,d : Gender-EI-Communion-Agency-TFL 0
0 .43 .0000 1.000 .000 1.000 .0061
(Partial Mediation) (saturated model)

Model 2a b,c : 2 DVs (Self and Other TFL):

2.513 2 – .0247 .999 – .990 .0127
Gender-EI-Communion-Agency (Full Mediation)
Model 2b b,d : 2 DVs (Self and Other TFL): (add
2.073 1 .44 .0506 .997 .002 .958 .0116
direct path from Gender to Other-rated TFL)
Model 2c b,e : 2 DVs (Self and Other TFL): (add direct
.434 1 2.08 .0000 1.000 .001 1.022 .0052
path from Gender to Self-rated TFL)
Model 2d b,f : 2 DVs (Self and Other TFL): (add 2
0
direct paths from Gender to Self- and Other-rated 0 2.51 .0000 1.000 .001 1.000 .0000
(saturated model)
TFL)
J. Intell. 2022, 10, x FOR PEER REVIEW 15 of 22
Note. Models judged to fit best are in boldface. ∆CFI > .01 (Cheung and Rensvold 2002). Analyses based on
minimum N = 420. a Includes only non-self report transformational leadership as the DV. b Includes both non-self
and self-report transformational leadership as two DVs. c Does not include any direct paths from gender to
any transformational leadership Foroutcome
the modelvariables.
comparisons among Models
d Includes 2a,path
a direct 2b, 2c,from
and 2d (models
gender including both
to non-self-report
transformational leadership.transformational
e Includes a leadership (non-self
direct path report) to
from gender andself-report
transformational leadership (self
transformational report)
leadership.
f Includes direct paths fromas 2 DVs), Model 2a (hypothesized full mediation model: shown in Figure 4) displayed
gender to both non-self and self-report transformational leadership.
adequate overall fit (χ (df=2) = 11.517, RMSEA = .0384, CFI = .995, NNFI = .961, SRMR = .013).
2

We next specified a series of partial mediation models, for the sake of model comparison.
Model 2b (adding
For the model comparisons amonga directModels
path from2a,
gender
2b,to2c,
non-self-reported
and 2d (models TFL) displayed
including adequate
both
overall fit (χ2(df=1) = 9.503, RMSEA = .0665, CFI = .995, NNFI = .930, SRMR = .012), as did
transformational leadership (non-self report) and transformational leadership
Model 2c (adding a direct path from gender to self-reported TFL; χ2(df=1) = 1.990, RMSEA =
(self report)
as 2 DVs), Model 2a (hypothesized
.0227, CFI = .999, NNFI full= .992,
mediation model:
SRMR = .0052). Modelshown in Figure
2d (saturated 4) displayed
model, with both direct
2
adequate overall fit (χpaths—from = 11.517,
gender RMSEA
to self- and= .0384, CFI =
non-self-reported .995,
TFL) NNFI =
displayed .961, SRMR
perfect fit by =design
.013).
(df=2)
(zero degrees of freedom). When comparing Models 2a, 2b, 2c, and 2d using change in
We next specified a series of partial mediation models, for the sake of model comparison.
CFI, we retained Model 2a as the best fitting model. Path coefficients for Model 2a are
Model 2b (adding a directshownpath from
in Figure gender
4. All paths wereto non-self-reported TFL)
statistically significant in displayed
the expected adequate
directions.
2
overall fit (χ (df=1) = 9.503, RMSEA = .0665, CFI = .995, NNFI = .930, SRMR = .012),and
To test the indirect effects specified in the mediation hypotheses (H2, H4, asH6),
did
we next used Monte Carlo confidence intervals (Preacher
Model 2c (adding a direct path from gender to self-reported TFL; χ (df=1) = 1.990, RMSEA and
2 Selig 2012; see Table 7). In=
particular, we tested whether the product of the path coefficients of X → M (e.g., Gender
.0227, CFI = .999, NNFI →= EI).992,
and MSRMR
→ Y (e.g.,= .0052).
EI → TFL) Model 2d (saturated
was statistically model,
significant. with
As shown both7,direct
in Table three
paths—from gender indirect
to self-effects:
and(a) non-self-reported
Gender → EI → TFL, (b) TFL)
Gender displayed
→ Communion perfect
→ TFL,fitandby (c) design
Gender
→ Agency
(zero degrees of freedom). When→ TFL were tested,Models
comparing and all three
2a,were
2b, statistically
2c, and 2d significant⎯suggesting
using change in that CFI,
EI, communion, and agency jointly mediate the relationship between gender and trans-
we retained Model 2aformational
as the best fitting model. Path coefficients for Model 2a are
leadership (supporting H2, H4, and H6, as well as the overall hypothesized
shown in
Figure 4. All paths were statistically
model shown in Figure 1).significant in the expected directions.

Figure 4. Path Model Results for the Personality-Based Model of Female Leadership Advantage
Figure 4. Path Model Results for the Personality-Based Model of Female Leadership Advantage
(Model 2a). Note. Standardized path coefficients (β’s) are presented. N = 420; χ2 (2) = 2.513; RMSEA
(Model 2a). Note. Standardized N= =1, 420; 2
= .025, CFI =path coefficients
.999; NNFI (β’s)= .013;
= .990; SRMR are *presented.
p < .05. Female Male =χ2. (2) = 2.513; RMSEA
= .025, CFI = .999; NNFI = .990; SRMR = .013; * p < .05. Female = 1, Male = 2.
J. Intell. 2022, 10, 104 15 of 20

To test the indirect effects specified in the mediation hypotheses (H2, H4, and H6),
we next used Monte Carlo confidence intervals (Preacher and Selig 2012; see Table 7). In
particular, we tested whether the product of the path coefficients of X → M (e.g., Gender
→ EI) and M → Y (e.g., EI → TFL) was statistically significant. As shown in Table 7,
three indirect effects: (a) Gender → EI → TFL, (b) Gender → Communion → TFL, and
(c) Gender → Agency → TFL were tested, and all three were statistically significant—
suggesting that EI, communion, and agency jointly mediate the relationship between
gender and transformational leadership (supporting H2, H4, and H6, as well as the overall
hypothesized model shown in Figure 1).

Table 7. Test of indirect effects for gender and transformational leadership (non-self-reported TFL
and self-reported TFL).

Product of Indirect 95% Monte Statistically

Path
Coefficients Effect Carlo CI Significant
Gender→Emotional Intelligence→TFL (non-self) a (−.250)*(.097) −.0243 [−.0483, −.0041] Yes
Gender→Communion→TFL (non-self) a (−.300)*(.473) −.1419 [−.1945, −.0948] Yes
Gender→Agency→TFL (non-self) a (.230)*(.228) .0524 [.0268, .0837] Yes
Gender→Emotional Intelligence→TFL (self-report) b (−.250)*(.268) −.0670 [−.1012, −.0376] Yes
Gender→Communion→TFL (self-report) b (−.300)*(.316) −.0948 [−.1361, −.0597] Yes
Gender→Agency→TFL (self-report) b (.230)*(.282) .0649 [.0348, .0998] Yes
Note. a Indirect effects tests are based on path coefficients from Model 1a (non-self-reported TFL), and are also
identical to the path coefficients from Model 2a for non-self-reported TFL. b Indirect effects tests are based on path
coefficients from Model 2a (self-reported TFL). TFL = transformational leadership.

4. Discussion
There has been growing academic discussion of the female leadership advantage over
the past decades (Eagly and Carli 2003; Paustian-Underdahl et al. 2014; Vecchio 2002; Yukl
2002), largely motivated by the finding of a small advantage for women in non-self-reported
measures of transformational leadership (Eagly et al. 2003). However, it is unclear why
women might be better transformational leaders compared to men. Therefore, one main
goal of the current study was to identify and test potential personality-based individual
differences that might give women an advantage in leadership.
Results confirm two, offsetting sets of phenomena. First, emotional intelligence and
communal personality (i.e., stereotypical femininity, marked by warmth and compassion)
both help women to be perceived as better leaders than men. Second, agentic personality
(i.e., stereotypical masculinity, marked by assertiveness and dominance) helps men to be
perceived as better leaders than women. These three mechanisms (emotional intelligence,
communion, and agency) operate simultaneously and result in a near-zero cumulative
gender effect on transformational leadership.

4.1. Theoretical Contributions

Although the idea of female leadership advantage (Paustian-Underdahl et al. 2014; cf.
Eagly and Carli 2003; Vecchio 2002; Yukl 2002; Grant 1988; also see Lipman-Blumen 1983;
Vecchio 2002) has been increasingly noted in recent decades, many of the classic and better-
known theories of gender gaps in leadership attainment and leader evaluations—those that
favor men—have continued to receive a great deal of attention (i.e., Think Manager—Think
Male: Schein 1973; Role Congruity Theory: Eagly and Karau 2002; Lack of Fit Theory:
Heilman 2001). These classic theories are typically used to explain the underrepresentation
or undervaluing of women in leadership roles (i.e., female leadership disadvantage).
The currently proposed personality-based explanatory model of the gender gap in
transformational leadership supports the existence of two sets of countervailing theoretical
mechanisms, articulating and supporting both a female leadership advantage (via emotional
intelligence and communion) and a female leadership disadvantage (via agency). The cur-
rent study also shows that these mechanisms not only help explain the gender gap in
J. Intell. 2022, 10, 104 16 of 20

transformational leadership, but that the same mechanisms also explain the gender gap in
leadership self-efficacy/self-rated transformational leadership.

4.2. Practical Implications

The double-bind (created by the role incongruity of gender roles and leader roles)
for (potential) female leaders has been one mainstream concept used to describe the
challenges women in leadership roles may face in contemporary organizations. In this
female leadership narrative, women face the dilemma of choosing either to be agentic
(congruent with their leader role, but incongruent with their stereotypical gender role), or
choosing to be communal (incongruent with their leader role, but congruent with their
stereotypical gender role). However, the results from the current study suggest that agency
and communion are both beneficial to being perceived as a transformational leader. We
echo that women and men need not choose between agency or communion (Bem 1974).
Results from the current study suggest that agency and communion help people in general
to be perceived as a leader—suggesting that one way to be perceived as a transformational
leader is to be both assertive and compassionate.
The current study also found that emotional intelligence helps people to be perceived
as a more transformational leader. Organizations hoping to strengthen their leadership
could consider implementing emotional intelligence training (Cherniss and Adler 2000).
Research using experimental designs has shown it is possible to increase emotional intel-
ligence through training (Nelis et al. 2009), which could be one effective way to increase
overall transformational leadership in organizations.

4.3. Limitations and Future Research Directions

First, the original meta-analyses in the current study focused exclusively on validated
measurement instruments (e.g., EI was measured using performance-based ability EI
measures only; leadership was measured using TFL measures only), in order to enhance
construct validity. As a result of this attempt to strengthen construct validity, some of
the original meta-analyses in the current study were based on a small number of primary
studies/samples (k’s ranged from 4 to 17; N’s ranged from 420 to 1923). Future meta-
analytic research could update the current results after more primary data accumulate, to
increase the sample size.
Second, all of the primary studies included in the original meta-analyses in the current
study were correlational studies (i.e., non-experimental designs). Accordingly, we could
not draw causal inferences from the current study. Third, one limitation of all meta-analyses
is that researchers can only study relationships that have been investigated by previous
studies. In the current study, the agency-EI and communion-EI relationships were based
on particularly small numbers of primary studies/samples, because little past research had
examined the relationships between agency and EI, and between communion and EI.
With respect to future research, the current set of findings raises additional ques-
tions about role congruity theory (Eagly and Karau 2002). In particular, it is less clear
that the leadership role is entirely masculine. Rather, the current set of results high-
lights that the transformational leadership role in particular may contain elements of
communion/stereotypical femininity and emotional competence. The leadership role is
multifaceted, in a manner that includes both stereotypically masculine and stereotypically
feminine features simultaneously.
Additionally, the current design was unable to assess interaction effects. For instance,
we could not assess whether agency predicted TFL better for men (see Kim et al. 2020), or
whether communion and EI predicted TFL better for women (as implied by role congruity
theory). Another limitation of the current study is that potential interaction effects between
the mediators cannot be tested. For example, we could not test whether being both
high in communion and EI at the same time would lead to an even greater leadership
advantage. In other words, we could not test whether communion is more effective for
J. Intell. 2022, 10, 104 17 of 20

leadership when people have high EI (i.e., a multiplicative effect of “will do” and “can do”
communal attributes).
Methodologically, all meta-analytic SEM studies potentially have the issues of using a
mixture of subpopulations (Newman et al. 2007). Therefore, it could be unclear whether
results generalize to any particular subpopulation. Another common issue for such studies
is construct validity. Almost always, multiple measurement instruments are used to
measure one single construct. One effort we made to minimize construct validity concerns
in the current study was to be selective with the measurement instruments by including
primary studies with validated measures of emotional intelligence, communion, agency,
and transformational leadership.

5. Conclusions
The current study showed that the gender gap in transformational leadership could
be jointly explained by three personality-based mechanisms (i.e., emotional intelligence,
communion, and agency). Specifically, emotional intelligence and communion could help
explain the female leadership advantage, whereas agency could help explain a countervailing
female leadership disadvantage. This set of findings expands upon Eagly’s (1987) social role
theory and explanations (i.e., sex roles of agency/masculinity and communion/femininity)
to also incorporate knowledge and skills-based tests of emotional intelligence (i.e., social
and emotional skill) simultaneously in the exhibition of transformational leadership. Alto-
gether, these three mechanisms, working in different directions, aggregate to a near-zero
total gender gap in transformational leadership. The near-zero total effect hides a set of
more substantial underlying gendered leadership phenomena (female leadership advan-
tage based on emotional intelligence and communion, and male leadership advantage
based on agency) that work in opposing directions to offset each other.

Author Contributions: Conceptualization, N.H. and D.A.N.; methodology, N.H., D.A.N. and K.L.B.;
formal analysis, N.H. and K.L.B.; writing—original draft preparation, N.H.; writing—review and
editing, D.A.N., K.L.B. and N.H.; supervision, D.A.N. All authors have read and agreed to the
published version of the manuscript.
Funding: This research received no external funding.
Data Availability Statement: The data presented in this study are available from the first author.
Acknowledgments: We are grateful to Tingyan Liu for their assistance as the second coder of all
primary studies.
Conflicts of Interest: The authors declare no conflict of interest.

References
Anderson, Neil, Filip Lievens, Karen van Dam, and Marise Born. 2006. A construct-driven investigation of gender differences in a
leadership-role assessment center. Journal of Applied Psychology 91: 555–66. [CrossRef] [PubMed]
Badura, Katie L., Emily Grijalva, Benjamin M. Galvin, Bradley P. Owens, and Dana L. Joseph Joseph. 2020. Motivation to lead: A
meta-analysis and distal-proximal model of motivation and leadership. Journal of Applied Psychology 105: 331–54. [CrossRef]
Badura, Katie L., Emily Grijalva, Daniel A. Newman, Thomas Taiyi Yan, and Gahyun Jeon. 2018. Gender and leadership emergence: A
meta-analysis and explanatory model. Personnel Psychology 71: 335–67. [CrossRef]
Bakan, David. 1966. The Duality of Human Existence: An Essay on Psychology and Religion. Chicago: Rand Mcnally.
Bales, Robert F. 1950. Interaction Process Analysis; A Method for the Study of Small Groups. Boston: Addison-Wesley.
Baron, Reuven. 1997. BarOn Emotional Quotient Inventory: Technical Manual. Toronto: Multi-Health Systems.
Bar-On, Reuven. 2004. The Bar-On Emotional Quotient Inventory (EQ-i): Rationale, Description and Summary of Psychometric Properties.
Hauppauge: Nova Science Publishers, pp. 115–45.
Barrick, Murray R., and Michael K. Mount. 1991. The big five personality dimensions and job performance: A meta-analysis. Personnel
Psychology 44: 1–26. [CrossRef]
Bass, Bernard M. 1985. Leadership and Performance beyond Expectations. New York: The Free Press.
Bass, Bernard M. 1990. From transactional to transformational leadership: Learning to share the vision. Organizational Dynamics 18:
19–31. [CrossRef]
Bem, Sandra L. 1974. The measurement of psychological androgyny. Journal of Consulting and Clinical Psychology 42: 155–62. [CrossRef]
J. Intell. 2022, 10, 104 18 of 20

Carless, Sally A., Alexander J. Wearing, and Leon Mann. 2000. A Short Measure of Transformational Leadership. Journal of Business and
Psychology 14: 389–405. [CrossRef]
Caza, Arran, Richard P. Bagozzi, Lydia Woolley, Lester Levy, and Brianna Barker Caza. 2010. Psychological capital and authentic
leadership: Measurement, gender, and cultural extension. Asia-Pacific Journal of Business Administration 2: 53–70. [CrossRef]
Cherniss, Cary, and Mitchel Adler. 2000. Promoting Emotional Intelligence in Organizations: Make Training in Emotional Intelligence Effective.
Alexandria: American Society for Training and Development.
Cheung, Gordon W., and Roger B. Rensvold. 2002. Evaluating goodness-of-fit indexes for testing measurement invariance. Structural
Equation Modeling 9: 233–55. [CrossRef]
Daus, Catherine S., and Neal M. Ashkanasy. 2005. The case for the ability-based model of emotional intelligence in organizational
behavior. Journal of Organizational Behavior 26: 453–66. [CrossRef]
de Raad, Boele. 2005. The trait-coverage of emotional intelligence. Personality and Individual Differences 38: 673–87. [CrossRef]
Eagly, Alice H. 1987. Sex Differences in Social Behavior: A Social-Role Interpretation. Hillsdale: Erlbaum, Inc.
Eagly, Alice H. 2007. Female leadership advantage and disadvantage: Resolving the contradictions. Psychology of Women Quarterly 31:
1–12. [CrossRef]
Eagly, Alice H., and Blair T. Johnson. 1990. Gender and leadership style: A meta-analysis. Psychological Bulletin 108: 233–56. [CrossRef]
Eagly, Alice H., and Linda L. Carli. 2003. The female leadership advantage: An evaluation of the evidence. The Leadership Quarterly 14:
807–34. [CrossRef]
Eagly, Alice H., and Steven J. Karau. 2002. Role congruity theory of prejudice toward female leaders. Psychological Review 109: 573–98.
[CrossRef] [PubMed]
Eagly, Alice H., and Wendy Wood. 2009. Sexual selection does not provide an adequate theory of sex differences in aggression.
Behavioral and Brain Sciences 32: 276–77. [CrossRef]
Eagly, Alice H., Mary C. Johannesen-Schmidt, and Marloes L. Van Engen. 2003. Transformational, transactional, and laissez-faire
leadership styles: A meta-analysis comparing women and men. Psychological Bulletin 129: 569–91. [CrossRef] [PubMed]
Fiedler, Fred E. 1967. A Theory of Leadership Effectiveness. Mcgraw-Hill Series in Management. New York: McGraw-Hill.
Fleishman, Edwin A. 1953. The measurement of leadership attitudes in industry. Journal of Applied Psychology 37: 153–58. [CrossRef]
Fleishman, Edwin A. 1957. The Leadership Opinion Questionnaire. In Leader Behavior: Its Description and Measurement. Edited by Ralph
M. Stogdill and Alvin E. Coons. Columbus: Bureau of Business Research, Ohio State University, pp. 120–33.
Fleishman, Edwin A. 1960. Manual for the Leadership Opinion Questionnaire. Park Ridge: Science Research Associates.
George, Jennifer M. 2000. Emotions and leadership: The role of emotional intelligence. Human Relations 53: 1027–55. [CrossRef]
Goleman, Daniel. 1995. Emotional Intelligence. New York: Bantam Books.
Goleman, Daniel. 1996. Emotional Intelligence: Why It Can Matter More Than IQ. London: Bloomsbury Publishing.
Goleman, Daniel. 1998. What makes a leader? Harvard Business Review 76: 82–91.
Goleman, Daniel, Richard Boyatzis, and Annie McKee. 2001. Primal leadership. Harvard Business Review 79: 42–50. [CrossRef]
Grant, Jan. 1988. Women as managers: What they can offer to organizations. Organizational Dynamics 16: 56–63. [CrossRef]
Halpin, Andrew W. 1957. Manual for the Leader Behavior Description Questionnaire. Columbus: Bureau of Business Research, College of
Commerce and Administration.
Halpin, Andrew W. 1966. Theory and Research in Administration. New York: Macmillan Company.
Halpin, Andrew W., and B. James Winer. 1957. A factorial study of the leader behavior descriptions. In Leader Behavior: Its Description
and Measurement. Edited by Ralph M. Stogdill and Alvin E. Coons. Columbus: Bureau of Business Research, Ohio State University,
pp. 39–51.
Harms, Peter D., and Marcus Credé. 2010. Emotional intelligence and transformational and transactional leadership: A meta-analysis.
Journal of Leadership & Organizational Studies 17: 5–17. [CrossRef]
Heilman, Madeline E. 1983. Sex bias in work settings: The lack of fit model. Research in Organizational Behavior 5: 269–98.
Heilman, Madeline E. 2001. Description and prescription: How gender stereotypes prevent women’s ascent up the organizational
ladder. Journal of Social Issues 57: 657–74. [CrossRef]
Hemphill, John K., and Alvin E. Coons. 1957. Development of the Leader Behavior Description Questionnaire. In Leader Behavior: Its
Description and Measurement. Edited by Ralph M. Stogdill and Alvin E. Coons. Columbus: Bureau of Business Research, Ohio
State University, pp. 6–38.
Hersey, Paul, and Kenneth H. Blanchard. 1977. Management of Organizational Behavior: Utilizing Human Resources, 3rd ed. Hoboken:
Prentice-Hall.
Hersey, Paul, and Kenneth H. Blanchard. 1982. Management of Organizational Behavior: Utilizing Human Resources, 4th ed. Hoboken:
Prentice Hall.
Hiller, Nathan J., Hock-Peng Sin, Ajay R. Ponnapalli, and Sibel Ozgen. 2019. Benevolence and authority as WEIRDly unfamiliar:
A multi-language meta-analysis of paternalistic leadership behaviors from 152 studies. The Leadership Quarterly 30: 165–84.
[CrossRef]
Hogan, Joyce C. 1978. Personological dynamics of leadership. Journal of Research in Personality 12: 390–95. [CrossRef]
Hsu, Ning, Katie L. Badura, Daniel A. Newman, and Mary Eve P. Speach. 2021. Gender, “masculinity”, and “femininity”: A
meta-analytic review of gender differences in agency and communion. Psychological Bulletin 147: 987–1011. [CrossRef]
J. Intell. 2022, 10, 104 19 of 20

Hurtz, Gregory M., and John J. Donovan. 2000. Personality and job performance: The Big Five revisited. Journal of Applied Psychology
85: 869–79. [CrossRef]
Joseph, Dana L., and Daniel A. Newman. 2010. Emotional intelligence: An integrative meta-analysis and cascading model. Journal of
Applied Psychology 95: 54–78. [CrossRef]
Joseph, Dana L., Jing Jin, Daniel A. Newman, and Ernest H. O’Boyle. 2015. Why does self-reported emotional intelligence predict job
performance? A meta-analytic investigation of mixed EI. Journal of Applied Psychology 100: 298–342. [CrossRef] [PubMed]
Judge, Timothy A., Joyce E. Bono, Remus Ilies, and Megan W. Gerhardt. 2002. Personality and leadership: A qualitative and quantitative
review. Journal of Applied Psychology 87: 765–80. [CrossRef] [PubMed]
Kim, Jun-Yeob, Ning Hsu, Daniel A. Newman, P. D. Harms, and Dustin Wood. 2020. Leadership perceptions, gender, and dominant
personality: The role of normality evaluations. Journal of Research in Personality 87: 103984. [CrossRef]
Koenig, Anne M., Alice H. Eagly, Abigail A. Mitchell, and Tiina Ristikari. 2011. Are leader stereotypes masculine? A meta-analysis of
three research paradigms. Psychological Bulletin 137: 616–42. [CrossRef]
Lee, Angela, and Nichelle C. Carpenter. 2018. Seeing eye to eye: A meta-analysis of self-other agreement of leadership. The Leadership
Quarterly 29: 253–75. [CrossRef]
Lipman-Blumen, Jean. 1983. Emerging patterns of female leadership in formal organizations. In The Challenge of Change Perspectives on
Family, Work, and Education. Edited by Martina Horner, Carol Nadelson and Malkah Notman. New York and London: Plenum
Press, pp. 61–91.
Lord, Robert G., Christy L. De Vader, and George M. Alliger. 1986. A meta-analysis of the relation between personality traits and
leadership perceptions: An application of validity generalization procedures. Journal of Applied Psychology 71: 402–10. [CrossRef]
Lord, Robert G., Roseanne J. Foti, and Christy L. De Vader. 1984. A test of leadership categorization theory: Internal structure,
information processing, and leadership perceptions. Organizational Behavior and Human Performance 34: 343–78. [CrossRef]
Ma, Anyi, Ashleigh Shelby Rosette, and Christy Zhou Koval. 2022. Reconciling female agentic advantage and disadvantage with the
CADDIS measure of agency. Journal of Applied Psychology, advance online publication. [CrossRef]
MacCann, Carolyn, and Richard D. Roberts. 2008. New paradigms for assessing emotional intelligence: Theory and data. Emotion 8:
540–51. [CrossRef]
MacCann, Carolyn, Dana L. Joseph, Daniel A. Newman, and Richard D. Roberts. 2014. Emotional Intelligence Is a Second-Stratum
Factor of Intelligence: Evidence from Hierarchical and Bifactor Models. Emotion 14: 358. [CrossRef]
Mann, Richard D. 1959. A review of the relationships between personality and performance in small groups. Psychological Bulletin 56:
241–70. [CrossRef]
Mayer, John D., and Peter Salovey. 1997. What is emotional intelligence? In Emotional Development and Emotional Intelligence: Implications
for Educators. Edited by Peter Salovey and David Sluyter. New York: Basic Books, pp. 3–34.
Mayer, John D., David R. Caruso, and Peter Salovey. 1999. Emotional intelligence meets traditional standards for an intelligence.
Intelligence 27: 267–98. [CrossRef]
Mayer, John D., Peter Salovey, and David R. Caruso. 2000. Models of emotional intelligence. In Handbook of Intelligence. Edited by
Robert J. Sternberg. Cambridge: Cambridge University Press.
Mayer, John D., Peter Salovey, David R. Caruso, and Gill Sitarenios. 2003. Measuring emotional intelligence with the MSCEIT V2.0.
Emotion 3: 97–105. [CrossRef] [PubMed]
Mayer, John D., Richard D. Roberts, and Sigal G. Barsade. 2008. Human abilities: Emotional intelligence. Annual Review of Psychology
59: 507–36. [CrossRef]
Megargee, Edwin I. 1969. Influence of sex roles on the manifestation of leadership. Journal of Applied Psychology 53: 377–82. [CrossRef]
Mischel, Walter. 1973. Toward a cognitive social learning reconceptualization of personality. Psychological Review 80: 252–83. [CrossRef]
Murphy, Kevin R. 2006. Four conclusions about emotional intelligence. In A Critique of Emotional Intelligence. Edited by Kevin R.
Murphy. New York: Psychology Press, pp. 345–54.
Nelis, Delphine, Jordi Quoidbach, Moïra Mikolajczak, and Michel Hansenne. 2009. Increasing emotional intelligence:(How) is it
possible? Personality and Individual Differences 47: 36–41. [CrossRef]
Newman, Daniel A., Rick R. Jacobs, and Dave Bartram. 2007. Choosing the best method for local validity estimation: Relative accuracy
of meta-analysis versus a local study versus Bayes-analysis. Journal of Applied Psychology 92: 1394–413. [CrossRef]
Offermann, Lynn R., John K. Kennedy Jr, and Philip W. Wirtz. 1994. Implicit leadership theories: Content, structure, and generalizability.
The Leadership Quarterly 5: 43–58. [CrossRef]
Paustian-Underdahl, Samantha C., Lisa S. Walker, and David J. Woehr. 2014. Gender and perceptions of leadership effectiveness: A
meta-analysis of contextual moderators. Journal of Applied Psychology 99: 1129–45. [CrossRef]
Petrides, Kostantinos V., and Adrian Furnham. 2003. Trait emotional intelligence: Behavioural validation in two studies of emotion
recognition and reactivity to mood induction. European Journal of Personality 17: 39–57. [CrossRef]
Podsakoff, Philip M., Scott B. MacKenzie, Robert H. Moorman, and Richard Fetter. 1990. Transformational leader behaviors and their
effects on followers’ trust in leader, satisfaction, and organizational citizenship behaviors. The Leadership Quarterly 1: 107–42.
[CrossRef]
Posner, Barry Z., and James M. Kouzes. 1988. Development and validation of the leadership practices inventory. Educational and
Psychological Measurement 48: 483–96. [CrossRef]
J. Intell. 2022, 10, 104 20 of 20

Preacher, Kristopher J., and James P. Selig. 2012. Advantages of Monte Carlo confidence intervals for indirect effects. Communication
Methods and Measures 6: 77–98. [CrossRef]
Rudman, Laurie A., Corinne A. Moss-Racusin, Julie E. Phelan, and Sanne Nauts. 2012. Status incongruity and backlash effects:
Defending the gender hierarchy motivates prejudice against female leaders. Journal of Experimental Social Psychology 48: 165–79.
[CrossRef]
Salovey, Peter, and John D. Mayer. 1990. Emotional Intelligence. Imagination, Cognition and Personality 9: 185–211. [CrossRef]
Schein, Virginia E. 1973. The relationship between sex role stereotypes and requisite management characteristics. Journal of Applied
Psychology 57: 95–100. [CrossRef]
Schmidt, Frank L., and John E. Hunter. 2015. Methods of Meta-Analysis: Correcting Error and Bias in Research Findings. London: Sage.
Silva, Alberto. 2016. What is leadership? Journal of Business Studies Quarterly 8: 1–5.
Smith, Jeffrey A., and Roseanne J. Foti. 1998. A pattern approach to the study of leader emergence. The Leadership Quarterly 9: 147–60.
[CrossRef]
Spence, Janet T., and Robert L. Helmreich. 1978. Masculinity and Femininity: Their Psychological Dimensions, Correlates, and Antecedents.
Austin: University of Texas Press. [CrossRef]
Spence, Janet T., Robert Helmreich, and Joy Stapp. 1975. Ratings of self and peers on sex role attributes and their relation to self-esteem
and conceptions of masculinity and femininity. Journal of Personality and Social Psychology 32: 29–39. [CrossRef]
Spence, Janet T., Robert L. Helmreich, and Joy Stapp. 1974. The Personal Attributes Questionnaire: A Measure of Sex Role Stereotypes and
Masculinity-Femininity. Austin: University of Texas. [CrossRef]
Stogdill, Ralph M. 1948. Personal factors associated with leadership: A survey of the literature. Journal of Psychology 25: 35–71.
[CrossRef] [PubMed]
Stogdill, Ralph M. 1963. Manual for the Leader Behavior Description Questionnaire-Form XII: An Experimental Revision. Columbus: Bureau
of Business Research, College of Commerce and Administration, Ohio State University.
Stogdill, Ralph M., Omar S. Goode, and David R. Day. 1962. New leader behavior description subscales. Journal of Psychology 54:
259–69. [CrossRef]
Su, Rong, James Rounds, and Patrick Ian Armstrong. 2009. Men and things, women and people: A meta-analysis of sex differences in
interests. Psychological Bulletin 135: 859–84. [CrossRef] [PubMed]
Twenge, Jean M. 1997. Changes in masculine and feminine traits over time: A meta-analysis. Sex Roles 36: 305–25. [CrossRef]
Vecchio, Robert P. 2002. Leadership and gender advantage. The Leadership Quarterly 13: 643–671. [CrossRef]
Viswesvaran, Chockalingam, and Deniz S. Ones. 1995. Theory testing: Combining psychometric meta-analysis and structural equations
modeling. Personnel Psychology 48: 865–85. [CrossRef]
Walumbwa, Fred O., Bani Orwa, Peng Wang, and John J. Lawler. 2005. Transformational leadership, organizational commitment,
and job satisfaction: A comparative study of Kenyan and U.S. financial firms. Human Resource Development Quarterly 16: 235–56.
[CrossRef]
Wiggins, Jerry S. 1991. Agency and communion as conceptual coordinates for the understanding and measurement of interpersonal
behavior. In Thinking Clearly about Psychology: Essays in Honor of Paul E. Meehl. Edited by Dante Cicchetti and William M. Grove.
Minneapolis: University of Minnesota Press, pp. 89–113.
Williams, Melissa J., and Larissa Z. Tiedens. 2016. The subtle suspension of backlash: A meta-analysis of penalties for women’s implicit
and explicit dominance behavior. Psychological Bulletin 142: 165. [CrossRef] [PubMed]
Winston, Bruce E., and Kathleen Patterson. 2006. An integrative definition of leadership. International Journal of Leadership Studies 1:
6–66.
Winter, Laraine, and James S. Uleman. 1984. When are social judgments made? Evidence for the spontaneousness of trait inferences.
Journal of Personality and Social Psychology 47: 237–52. [CrossRef]
Wood, Wendy, and Alice H. Eagly. 2002. A cross-cultural analysis of the behavior of women and men: Implications for the origins of
sex differences. Psychological Bulletin 128: 699–727. [CrossRef]
Wood, Wendy, and Alice H. Eagly. 2012. Biosocial Construction of Sex Differences and Similarities in Behavior. In Advances in
Experimental Social Psychology. Amsterdam: Elsevier, vol. 46, pp. 55–123. [CrossRef]
Wood, Wendy, and Alice H. Eagly. 2015. Two traditions of research on gender identity. Sex Roles 73: 461–73. [CrossRef]
Yoder, Janice D. 2001. Making leadership work more effectively for women. Journal of Social Issues 57: 815–28. [CrossRef]
Yukl, Gary. 2002. Leadership in Organizations. Hoboken: Prentice-Hall.
Zaccaro, Stephen J. 2007. Trait-based perspectives of leadership. American Psychologist 62: 6–16. [CrossRef] [PubMed]