Content uploaded by Daniel A. Newman
Author content
All content in this area was uploaded by Daniel A. Newman on Dec 17, 2015
Content may be subject to copyright.
Why Does Self-Reported Emotional Intelligence Predict Job Performance?
A Meta-Analytic Investigation of Mixed EI
Dana L. Joseph
University of Central Florida Jing Jin and Daniel A. Newman
University of Illinois at Urbana-Champaign
Ernest H. O’Boyle
The University of Iowa
Recent empirical reviews have claimed a surprisingly strong relationship between job performance and
self-reported emotional intelligence (also commonly called trait EI or mixed EI), suggesting self-
reported/mixed EI is one of the best known predictors of job performance (e.g., ˆ⫽.47; Joseph &
Newman, 2010b). Results further suggest mixed EI can robustly predict job performance beyond
cognitive ability and Big Five personality traits (Joseph & Newman, 2010b;O’Boyle, Humphrey,
Pollack, Hawver, & Story, 2011). These criterion-related validity results are problematic, given the
paucity of evidence and the questionable construct validity of mixed EI measures themselves. In the
current research, we update and reevaluate existing evidence for mixed EI, in light of prior work
regarding the content of mixed EI measures. Results of the current meta-analysis demonstrate that (a) the
content of mixed EI measures strongly overlaps with a set of well-known psychological constructs (i.e.,
ability EI, self-efficacy, and self-rated performance, in addition to Conscientiousness, Emotional Stabil-
ity, Extraversion, and general mental ability; multiple R⫽.79), (b) an updated estimate of the
meta-analytic correlation between mixed EI and supervisor-rated job performance is ˆ⫽.29, and (c) the
mixed EI–job performance relationship becomes nil (⫽–.02) after controlling for the set of covariates
listed above. Findings help to establish the construct validity of mixed EI measures and further support
an intuitive theoretical explanation for the uncommonly high association between mixed EI and job
performance—mixed EI instruments assess a combination of ability EI and self-perceptions, in addition
to personality and cognitive ability.
Keywords: emotional intelligence, job performance, heterogeneous domain sampling, personality,
self-efficacy
Propelled by the New York Times bestseller of Daniel Goleman
(1995), the concept of emotional intelligence (EI) has gained a
great amount of public popularity and business attention in the past
two decades; EI is currently considered a widely accepted practi-
tioner tool for hiring, training, leadership development, and team
building by the business community. As evidence of this, Gole-
man’s (1995) book has been touted as one of the 25 most influ-
ential business management books of all time by Time magazine
(Sachs, 2011), and Goleman’s (1998) article published in Harvard
Business Review has become the most requested reprint from this
journal in the last four decades (Sardo, 2004). Beyond the popu-
larity of Goleman’s work, a search of consulting firm websites
indicates more than 150 consulting firms offer EI-related products
and services (including two of the largest industrial/organizational
psychology consulting firms, Development Dimensions Interna-
tional and Personnel Decisions International). Indeed, EI services
have become a multimillion-dollar consulting industry (Grewal &
Salovey, 2005), with some estimates suggesting that 75% of For-
tune 500 companies have adopted EI-related products and services
(Bradberry & Greaves, 2009). Despite the commercial expansion
of the concept, some scholars from the organizational sciences
have been skeptical about it, given the lack of consensus with
regard to its definition, measurement, and validity (Landy, 2005;
Murphy, 2006).
For instance, one definitional ambiguity stems from the “emo-
tional intelligence” label having been historically applied to two,
relatively distinct theoretical constructs. The first sort of EI con-
struct has been defined as “the ability to carry out accurate rea-
soning about emotions and the ability to use emotions and emo-
tional knowledge to enhance thought” (Mayer, Roberts, &
Barsade, 2008, p. 511), which emphasizes EI as an actual ability,
or facet of intelligence (Daus & Ashkanasy, 2005;MacCann,
Joseph, Newman, & Roberts, 2014). The second definition of EI
uses the EI label as an umbrella term that encompasses a constel-
lation of personality traits, affect, and self-perceived abilities,
This article was published Online First September 22, 2014.
Dana L. Joseph, Department of Psychology, University of Central Florida;
Jing Jin, Department of Psychology, University of Illinois at Urbana-
Champaign; Daniel A. Newman, Department of Psychology and School of
Labor and Employment Relations, University of Illinois at Urbana-Cham-
paign; Ernest H. O’Boyle, Tippie College of Business, The University of Iowa.
Jing Jin is now at Development Dimensions International, Pittsburgh,
Pennsylvania.
Correspondence concerning this article should be addressed to Dana L.
Joseph, Department of Psychology, University of Central Florida, 4000 Cen-
tral Florida Boulevard, Orlando, FL 32816. E-mail: dana.joseph@ucf.edu
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Journal of Applied Psychology © 2014 American Psychological Association
2015, Vol. 100, No. 2, 298–342 0021-9010/15/$12.00 http://dx.doi.org/10.1037/a0037681
298
rather than actual aptitude (Bar-On, 1997;Goleman, 1995;
Petrides & Furnham, 2001). These two definitions have come to be
called ability EI and mixed EI, respectively. Meta-analytic results
have demonstrated that mixed EI measures and ability EI measures
intercorrelate only moderately (ˆ⫽.26, Joseph & Newman,
2010b;ˆ⫽.14, van Rooy, Viswesvaran, & Pluta, 2005), and they
exhibit distinctive patterns of relationships with job performance.
For example, Joseph and Newman (2010b) found that mixed EI
measures exhibited a strong criterion-related validity coefficient of
ˆ⫽.47, whereas ability EI measures exhibited markedly lower
validity for predicting job performance (ˆ⫽.18). Results of recent
meta-analyses further suggest that mixed EI measures can robustly
predict job performance beyond cognitive ability and Big Five
personality traits (⌬R
2
⫽.142 ⫽14%; Joseph & Newman, 2010b;
⌬R
2
⫽.068 ⫽7%; O’Boyle, Humphrey, Pollack, Hawver, &
Story, 2011), whereas ability EI measures exhibit near-zero incre-
mental validity (⌬R
2
⫽.002 ⫽0.2%; Joseph & Newman, 2010b;
⌬R
2
⫽.004 ⫽0.4%; O’Boyle et al., 2011). Joseph and Newman
(2010b) described this combination of results as “an ugly state of
affairs” (p. 72) because many have considered ability EI (i.e., the
weaker predictor of job performance) to be based upon a stronger
theoretical model (Daus & Ashkanasy, 2005;Matthews, Roberts,
& Zeidner, 2004;Matthews, Zeidner, & Roberts, 2002;Murphy,
2006), whereas mixed EI (i.e., the stronger predictor of job per-
formance) has been at the center of controversy due to theoretical
underdevelopment (Murphy, 2006). The lack of theoretical con-
sensus surrounding what mixed EI is, combined with its superior
predictive power, has created a paradox that we believe deserves
additional clarification. Thus, in responding to previous calls for a
theoretical understanding of the substantive content of mixed EI
(Joseph & Newman, 2010b;Locke, 2005), we sought in the current
study to answer two questions: “What do mixed EI instruments
measure?” and “Why are mixed EI instruments related to job
performance?”
In the current article, we thus propose to make two contributions
to the study of mixed EI and job performance. First, we shed light
into the black box of mixed EI construct validity, to meta-
analytically test past conceptualizations of what content mixed EI
instruments actually measure. Second, in an attempt to explain why
mixed EI is so strongly related to job performance, we illuminate
common covariates of mixed EI and job performance and assess
the extent to which mixed EI demonstrates incremental validity
above and beyond these common covariates.
What Do Mixed EI Instruments Measure?
In order to understand what might be in the black box of mixed
EI instruments, we note that prior authors who have questioned the
construct validity of mixed EI have done so primarily because
many mixed EI items appear to capture well-established constructs
other than emotional intelligence (Joseph & Newman, 2010b;
Mayer et al., 2008;Murphy, 2006). In other words, it appears that
authors of mixed EI measures may have (unknowingly) engaged in
domain sampling (Cronbach & Meehl, 1955;Ghiselli, Campbell,
& Zedeck, 1981;Nunnally, 1967), whereby mixed EI measures
were constructed to sample from various well-known content
domains in the field of psychology. Although domain sampling
typically refers to the process of sampling items from a homoge-
neous content domain (e.g., developing a Conscientiousness scale
by drawing items from the Conscientiousness domain), the devel-
opment of mixed EI measures appears to have involved heteroge-
neous domain sampling, or the sampling of items from a diverse
set of content domains. Whereas heterogeneous domain sampling
may illuminate why these measures appear to capture a “grab bag”
of content domains, the question still remains: What exactly are
these content domains that constitute “mixed EI”? In the follow-
ing, we draw on prior theory and content analysis of popular mixed
EI measures to hypothesize that these measures likely capture the
following content domains: Conscientiousness, Extraversion, self-
related qualities (i.e., general self-efficacy and self-rated perfor-
mance), ability EI, Emotional Stability, and cognitive ability.
We begin by noting that several EI scholars have recently
offered suggestions regarding the content captured by mixed EI
measures in an attempt to clear up the muddied waters of the
construct. Specifically, Mayer et al. (2008) have summarized that
mixed EI covers four content areas: (a) achievement motivation
(which is similar to the industriousness facet of Conscientiousness;
Roberts, Chernyshenko, Stark, & Goldberg, 2005), (b) control-
related qualities such as impulse control and flexibility (which
theoretically overlap with the self-control facet of Conscientious-
ness; Roberts et al., 2005), (c) gregariousness and assertiveness
(which are two facets of Extraversion; Costa & McCrae, 1992),
and (d) self-related qualities (e.g., positive self-appraisals, such as
general self-efficacy). Thus, Mayer et al. (2008) appear to have
suggested that mixed EI overlaps with Conscientiousness, Extra-
version, and self-related qualities such as general self-efficacy. We
will discuss each of these potential overlaps below. Before we do,
we would like to point out that prior theoretical work on the
construct of mixed EI is scant. As a result, when discussing the
construct of mixed EI, we often discuss the measures of mixed EI
rather than the construct (i.e., because it is not clear what the
construct of mixed EI actually is, we tend—by necessity—to
confound the construct with the measure; cf. Arthur & Villado,
2008). This is a natural result of a theoretically underdeveloped
construct, and indeed in the current article, we attempt to help
remedy this very issue by developing an understanding of which
constructs are subsumed by mixed EI.
Conscientiousness and Mixed EI
As previously mentioned, prior theoretical work suggests that
mixed EI taps attributes like achievement-motivation and control-
related qualities such as low impulsiveness (Mayer et al., 2008;
Petrides & Furnham, 2001;Zeidner, Matthews, & Roberts, 2004),
which fall into the personality domain of trait Conscientiousness.
For example, Bar-On’s (1997) mixed EI model includes subfacets
of self-actualization, or striving to achieve one’s personal goals,
and impulse control, or effectively controlling one’s emotions—
which are similar to the industriousness and self-control facets of
Conscientiousness, respectively (Roberts et al., 2005). Similarly,
Goleman’s mixed EI model (Wolff, 2006) includes initiative (i.e.,
“readiness to act on opportunities,” p. 3) and achievement (i.e.,
“striving to improve or meeting a standard of excellence,” p. 3),
which theoretically overlap with Conscientiousness facets.
In addition to the content overlap between Conscientiousness
and mixed EI, a secondary reason that one might expect a positive
relationship between the two constructs is because Conscientious-
ness has been characterized as a tendency to follow socially
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
299
SELF-REPORTED EMOTIONAL INTELLIGENCE
prescribed norms (John & Srivastava, 1999), and this dutifulness
in adhering to norms likely carries over into emotional roles as
well. So conscientious individuals may exert extra effort in adher-
ing to emotion-related norms (i.e., Conscientiousness gives rise to
a motivational state that induces one to be meticulous in his or her
task performance [Emmons, 1989], including emotional tasks such
as perceiving one’s emotion, perceiving others’ emotion, display-
ing appropriate emotions, and so forth). We propose that emotional
skills and abilities develop naturally as a result of increased effort
in adhering to emotion-related norms (e.g., the more one exerts
effort in displaying appropriate emotions, the better one becomes
at doing so). Thus, we expected Conscientiousness to be positively
related to mixed EI, which is supported by prior meta-analytic
estimates indicating a strong relationship between Conscientious-
ness and mixed EI (ˆ⫽.38 in both Joseph & Newman, 2010b, and
O’Boyle et al., 2011).
Extraversion and Mixed EI
Extraversion, a dimension of the Big Five, includes two compo-
nents: social vitality and social dominance (Helson & Kwan, 2000).
Some have argued that the social vitality component reflects an
underlying need or desire for social contact that often results in a
greater number of social relationships for extraverted individuals
(Hotard, McFatter, McWhirter, & Stegall, 1989). In the process of
establishing an extravert’s expansive social network, he or she likely
develops a set of emotion-related skills (e.g., the ability to display
positive affect) that are used to build social bonds. Many of the
emotion-related skills that are likely developed as a result of an
extravert’s desire to form social relationships are dimensions of mixed
EI, including relationship skills, social competence (Petrides & Furn-
ham, 2001), interpersonal relationships, and happiness (Bar-On,
1997). Some mixed models of EI also explicitly include assertiveness
(Bar-On, 1997;Petrides & Furnham, 2003), which directly reflects the
social dominance facet of Extraversion (and the assertiveness facet of
Extraversion in the revised NEO Personality Inventory [NEO–PI–R];
Costa & McCrae, 1992), reiterating the overlap between Extraversion
and mixed EI due to common elements of both constructs. The strong
empirical relationship between Extraversion and mixed EI has also
been well documented (ˆ⫽.46, Joseph & Newman, 2010b;ˆ⫽.49,
O’Boyle et al., 2011), supporting the notion that mixed EI is posi-
tively related to Extraversion because (a) extraverts’ inclination to
establish social bonds results in enhanced emotional and social skills
and (b) the social dominance component of Extraversion explicitly
overlaps with dimensions of mixed EI (e.g., assertiveness; Bar-On,
1997).
Self-Related Qualities and Mixed EI
The third content area that Mayer et al. (2008) suggested is cap-
tured by mixed EI measures is self-related qualities. The idea that
self-related qualities may account for the relationship between mixed
EI and job performance has been similarly articulated by Newman,
Joseph, and MacCann (2010), who theorized that mixed EI measures
capture self-efficacy and self-assessments of past job performance.
First, general/generalized self-efficacy represents one’s perception of
his or her ability to cope with life challenges and task demands across
a variety of different situations (e.g., Chen, Gully, & Eden, 2001;
Judge, Locke, & Durham, 1997;Sherer et al., 1982). Self-consistency
theory suggests that individuals have a desire to behave in a way that
is consistent with their own image (Korman, 1970). When consider-
ing emotional and social behavior, it is likely that individuals who
have a desire to maintain a positive self-image (i.e., individuals with
high general self-efficacy) have cultivated emotional and social skills
that allow them to display appropriate social behaviors to maintain
their self-image. We propose that these emotional and social skills are
represented in the construct of mixed EI; for example, the display of
appropriate social behaviors requires dimensions of mixed EI such as
social responsibility (i.e., the ability to cooperate with others), empa-
thy (i.e., the ability to understand and appreciate the feelings of
others), and interpersonal relationships (i.e., the ability to establish
and maintain relationships; Bar-On, 1997). Therefore, individuals
high in general self-efficacy likely have high mixed EI in order to
display social behaviors that are consistent with their self-views,
whereas those low in general self-efficacy may shy away from social
relationships because doing so is consistent with their self-views (and
as a result, these individuals fail to develop emotional skills and
abilities for maintaining social relationships). In addition, an exami-
nation of the content of mixed EI measures reveals overlap between
the constructs of general self-efficacy and mixed EI, including the
self-regard facet of Bar-On’s Emotional Quotient Inventory (EQ-i;
Bar-On, 1997), which represents the propensity to regard oneself as
generally competent, and Goleman’s (1998) self-confidence dimen-
sion, which also represents one’s sense of self-worth (Wolff, 2006).
Thus, we expected general self-efficacy to be positively related to
mixed EI because mixed EI is one avenue through which an individ-
ual can maintain his or her self-image and because of the content
overlap between general self-efficacy and mixed EI.
Second, from looking at the content of mixed EI scales, it also
appears that these mixed EI instruments tap into something akin to
self-rated performance. Unfortunately, these mixed EI measures
are largely proprietary (thus, the mixed EI items cannot be pre-
sented here in any way), or else a few example items might easily
support the notion that mixed EI scales capture self-rated perfor-
mance. These types of items are similar to the items “I feel I can
produce a lot of good work,” “I perform well in teams,” “I have
accomplished many things in the last year,” and “I have performed
well under pressure” (although these are not actual items on any
mixed EI measure, they are very similar). We note that these items
(and their original counterparts present in actual mixed EI mea-
sures) are conceptually closer to self-ratings of general perfor-
mance rather than self-ratings of job performance per se (e.g., a
respondent may evaluate his or her performance as a member of a
sports team when answering the item “I perform well in teams”).
In the current article, we argue that self-ratings of job performance
are a component of mixed EI because they are a key aspect of
one’s perceptions of performance in general (e.g., perceived ex-
cellence in public speaking at work would likely lead to perceived
strength in public speaking in any context). This is because: (a)
self-ratings of general performance are likely estimated via a
process where one’s broad perceptions of performance are formed
as a mental average of his or her specific performance across
various life domains, and (b) as a mental average of performance
across all life domains, self-ratings of performance likely over-
sample from the work domain because work plays a central role in
most individuals’ lives (e.g., Kreiner, Hollensbe, & Sheep, 2009;
Wanberg, 2012). Therefore, we argue that self-perceptions of job
performance are an indicator of the domain of self-perceived
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
300 JOSEPH, JIN, NEWMAN, AND O’BOYLE
general performance, and as such, we expect self-rated job perfor-
mance to be positively related to mixed EI.
Ability EI and Mixed EI
Beyond the conceptual overlaps between mixed EI and
Conscientiousness/Extraversion/self-rated qualities that were pro-
posed by Mayer et al. (2008), an additional variable that may add
insight into the construct validity of mixed EI is ability EI itself.
Although prior work has shown only a modest relationship between
ability EI and mixed EI (ˆ⫽.26; Joseph & Newman, 2010b), this is
likely due to the content breadth of mixed EI (i.e., emotional abilities
only constitute a fraction of mixed EI content). Self-perception theory
would suggest that one’s self-perceptions are inferred from one’s
behavior (Bern, 1972), and given that mixed EI involves one’s self-
perceptions of his or her emotional abilities, we would expect these
self-perceptions to be drawn from one’s actual emotional abilities
(i.e., ability EI, which includes behaviors such as emotion expression,
voice inflection, and emotion-related gestures; Salovey & Mayer,
1990; see also, Brackett, Rivers, Shiffman, Lerner, & Salovey, 2006).
It has been claimed in prior work that mixed EI includes self-
perceived emotional abilities (Petrides & Furnham, 2001), and a
perusal of items from the EQ-i (Bar-On, 1997), for example, shows
that some of these items clearly reflect self-ascribed emotion regula-
tion and emotion perception abilities. In particular, the emotional
self-awareness and empathy facets of Bar-On’s EQ-i appear to ad-
dress emotion perception ability and emotion understanding (two
facets of ability EI; Salovey & Mayer, 1990), and the emotional
awareness and emotional self-control facets of Goleman’s (1998)
model appear to capture emotion perception ability and emotion
regulation ability (also facets of ability EI; Salovey & Mayer, 1990).
Therefore, it is likely that actual emotional ability (i.e., ability EI) is
part of the content that is sampled within mixed EI measures.
Emotional Stability and Mixed EI
Popular markers of Emotional Stability include low levels of trait
negative affect (Gross, Sutton, & Ketelaar, 1998) and dampened
emotional reactions to daily stressors (Marco & Suls, 1993;Suls,
Green, & Hillis, 1998). These characteristics of emotionally stable
individuals likely reflect an enhanced ability to manage emotions and
use effective emotion regulation strategies (e.g., reappraisal; Gross &
John, 2003). Therefore, we expected Emotional Stability to be posi-
tively related to mixed EI because Emotional Stability involves the
use of emotion regulation skills that mixed EI comprises (e.g., stress
tolerance;Bar-On, 1997). In addition, De Raad (2005) has conducted
empirical analyses on the content validity of several mixed EI mea-
sures and shown that for six mixed EI measures, 42% of the items
were classified by content experts as direct measures of Emotional
Stability. This content validity evidence is consistent with the large
meta-analytic relationship between Emotional Stability and mixed EI
instruments (ˆ⫽.53, Joseph & Newman, 2010b;ˆ⫽.54, O’Boyle
et al., 2011), and the conceptual overlap between several facets of
mixed EI scales and Emotional Stability (e.g., stress tolerance,Bar-
On, 1997;optimism,Goleman, 1998). Thus, it appears that part of the
content “mix” in mixed EI measurement is the well-known concept of
Emotional Stability.
Cognitive Ability and Mixed EI
At this point, we note that any attempt by us to consider
cognitive ability as a content domain that is captured in measures
of mixed EI would be largely antithetical to the philosophy upon
which many mixed EI measures were founded. That is, cognitive
ability is explicitly excluded from most mixed models of EI. For
example, Bar-On’s (1997) mixed model of EI is said to include “an
array of noncognitive capabilities, competencies, and skills that
influence one’s ability to succeed in coping with environmental
demands and pressures” (italics added, p. 14). Interestingly, how-
ever, this very model also includes facets of apparent cognitive
ability components such as problem solving and reality testing
(Bar-On, 1997). In addition, cognitive ability is theorized to pro-
mote individual adaptability, primarily due to the additional infor-
mation processing that is required in novel situations (LePine,
Colquitt, & Erez, 2000). Because adaptability is a component of
mixed EI (i.e., flexibility, or one’s ability to adapt to unfamiliar and
dynamic circumstances; Bar-On, 1997), we expected cognitive
ability to be related to mixed EI—that is, individuals high in
cognitive ability can handle the additional information processing
demands of unfamiliar situations. Because it appears that mixed
models of EI may actually include cognitive ability components
(i.e., some mixed models are theorized to include abilities as part
of the mixture of constructs; Boyatzis, 2009;Mayer et al., 2008;
Petrides & Furnham, 2001) and because mixed EI models involve
adaptability, which is related to cognitive ability via improved
information processing in novel situations (LePine et al., 2000), we
expected to find empirical overlap between measures of general
mental ability and measures of mixed EI.
In sum, we have proposed that mixed EI measures have sampled
from several well-established construct domains, including Conscien-
tiousness, Extraversion, general self-efficacy, self-rated performance,
ability EI, Emotional Stability, and cognitive ability. Because mixed
EI measures appear to sample so heavily from these seven construct
domains, we expected that individual variation in mixed EI will be
largely accounted for by these seven components.
Why Are Mixed EI Instruments Related to
Job Performance?
Previous meta-analyses of mixed EI suggest a strong relation-
ship between mixed EI and job performance (Joseph & Newman,
2010b;O’Boyle et al., 2011), with estimated criterion validities as
strong as, or stronger than, any other personality trait. To illumi-
nate why mixed EI has such a robust relationship with job perfor-
mance, we demonstrate that the proposed content domains from
which mixed EI measures are sampled (see previous section) are
also related to job performance. In other words, mixed EI taps into
a mix of constructs that have well-established relationships with
job performance, which explains why mixed EI predicts job per-
formance.
Why the Seven Knowledge, Skills, Abilities, and Other
Characteristics (KSAOs) Relate to Job Performance
For example, Conscientiousness (a proposed construct domain
from which mixed EI measures are sampled) has a known positive
relationship with job performance (Barrick & Mount, 1991;Bar-
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
301
SELF-REPORTED EMOTIONAL INTELLIGENCE
rick, Mount, & Judge, 2001;Hurtz & Donovan, 2000; a link
theoretically due to Conscientious employees’ accomplishment
striving, status striving [Barrick, Stewart, & Piotrowski, 2002] and
goal setting [Barrick, Mount, & Strauss, 1993]). Similarly, evi-
dence suggests Extraversion can have reasonable predictive valid-
ity for job performance, especially for success in management and
sales jobs (Barrick & Mount, 1991;Vinchur, Schippmann, Swit-
zer, & Roth, 1998; due in part to status striving; Barrick et al.,
2002), and Emotional Stability also has an established positive
relationship with job performance (Barrick & Mount, 1991;Bar-
rick et al., 2001;Hurtz & Donovan, 2000; a relationship explained
by the fact that Neurotic individuals exhibit poorer emotional
coping skills; Connor-Smith & Flachsbart, 2007;Joseph & New-
man, 2010b). Thus, these three Big Five variables help explain the
relationship between mixed EI and job performance, because they
are common antecedents to both constructs.
In addition to these Big Five personality constructs, general
self-efficacy is thought to predict work performance by way of
motivation, goal-setting (Erez & Judge, 2001), and job engage-
ment (Rich, LePine, & Crawford, 2010). In other words, individ-
uals with high general self-efficacy should maintain both direction
and persistence of effort toward the job at hand. Therefore, if
mixed EI measures are sampled from the general self-efficacy
domain, then self-efficacy should partly explain the mixed EI–job
performance relationship. Further, because past performance is the
best predictor of future performance (see meta-analysis by Stur-
man, Cheramie, & Cashen, 2005; as well as seminal discussions by
Corballis, 1965;Humphreys, 1960;Jones, 1962; and Wernimont &
Campbell, 1968), we propose that another key mechanism by
which mixed EI scales predict job performance is that mixed EI
measures ask respondents to report, in part, how well they have
generally performed on projects in the past. Accordingly, we
expect self-rated performance to be considered a common covari-
ate of both mixed EI and supervisor-rated job performance.
Finally, cognitive ability appears to contribute to mixed EI mea-
sures, and it is a fundamental antecedent of job performance (Schmidt
& Hunter, 1998), largely due to the tendency for high-ability employ-
ees to acquire job knowledge (Schmidt, Hunter, & Outerbridge,
1986). Moreover, ability EI has been theorized to relate to job per-
formance via enhanced social interactions, advanced understanding of
the emotional demands on the situation (O’Boyle et al., 2011), and
increased attentional resources (because emotion regulation skill can
slow cognitive resource depletion; Joseph & Newman, 2010b). The
relationship between ability EI and job performance has been sup-
ported via meta-analytic evidence (Joseph & Newman, 2010b;
O’Boyle et al., 2011), and thus, it appears that cognitive ability and
ability EI are common antecedents to both mixed EI measures and job
performance, aiding in the explanation of why mixed EI and job
performance are strongly related.
Heterogeneous Domain Sampling Model
In summary of our arguments, the various constructs tapped by
self-report mixed EI measures (i.e., Conscientiousness, Extraversion,
general self-efficacy, self-rated performance, ability EI, Emotional
Stability, and cognitive ability) also appear to be antecedents of job
performance. Therefore, these seven constructs should explain the
relationship between mixed EI and job performance. One conse-
quence of this state of affairs is that the incremental validity of mixed
EI for predicting job performance should be quite limited once these
constructs are controlled. In other words, we are advancing a theo-
retical model of the mixed EI–job performance relationship that we
refer to as the heterogeneous domain sampling model (see Figure 1,
Model A; Cronbach & Meehl, 1955;Nunnally, 1967). According to
our hypothesized model, mixed EI measures will fail to account for
incremental validity in job performance after we have controlled for
Conscientiousness, Extraversion, general self-efficacy, self-rated per-
formance, ability EI, Emotional Stability, and cognitive ability. In
Figure 1. Model A. Heterogeneous Domain Sampling Model (no incremental validity, no mediation). This is our
hypothesized model. Standardized estimates. All predictors were allowed to intercorrelate.
ⴱ
p⬍.05;
2
(df ⫽1) ⫽
0.19 (p⬎.05), root-mean-square error of approximation ⫽.00, comparative fit index ⫽1.00, Tucker–Lewis index ⫽
1.00, standardized root-mean-square residual ⫽.001 (model fit is good). Perf ⫽performance.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
302 JOSEPH, JIN, NEWMAN, AND O’BOYLE
other words, we believe these seven KSAOs represent all the essential
constructs that constitute the “mix” in mixed EI that is responsible for
the large observed criterion-related validity of mixed EI.
An expert reviewer pointed out that our hypothesized heteroge-
neous domain sampling model can be thought of as one model, in a
set of alternative models, that can each explain why mixed EI relates
to job performance. This set of alternative models includes (a) our
heterogeneous domain sampling model (Figure 1, Model A), which is
ano mediation model, in which mixed EI exhibits no incremental
validity beyond the seven KSAOs, and there is no mediation of the
KSAOs by mixed EI, (b) a partial mediation model, labeled the
“incremental validity model” (Figure 2, Model B), in which mixed EI
predicts job performance partly because it transmits the effects of the
seven KSAOs and partly because mixed EI represents some addi-
tional content that relates to job performance beyond the seven
KSAOs, and (c) a full mediation model (Figure 3, Model C), in which
mixed EI fully captures all of the generative mechanisms by which the
seven KSAOs relate to job performance. As stated previously, in the
current study, we are hypothesizing the first model (Figure 1, Model
A), which offers a simple heterogeneous domain sampling explana-
tion for why mixed EI relates to job performance. We tested this
model (Figure 1, Model A) by comparing it against the two alternative
models suggested by the expert reviewer (cf. incremental validity
[partial mediation] model [Figure 2, Model B], and full mediation
model [Figure 3, Model C]).
If our heterogeneous domain sampling model is accurate, then it
implies that a combination of traits—Extraversion, Emotional Sta-
bility, Conscientiousness, general self-efficacy, self-rated perfor-
mance, cognitive ability, and ability EI—together explain why
mixed EI measures predict job performance so well. To expand
upon this point, individuals who possess these traits should have
motivational tendencies and goals characterized by high status
striving and accomplishment striving (i.e., Extraversion and Con-
scientiousness; Barrick et al., 2002), as well as elevated perfor-
mance expectations (i.e., high self-rated performance and general
self-efficacy). These individuals should further be equipped to
attain these goals and motivational agendas via their heightened
emotional coping skills, emotion regulation skills, and emotional
understanding (low Neuroticism, Connor-Smith & Flachsbart,
2007; high Ability EI, Joseph & Newman, 2010b), as well as their
ability to more quickly absorb job knowledge (cognitive ability;
Schmidt et al., 1986). Mixed EI thus offers a high-utility mixture
of individual traits to predict job performance.
Defining Job Performance
Before we move on to describe the methods used in the current
study, we first briefly expound on our definition of the criterion, job
performance. Indeed, past discrepancies in criterion definition have
led to some inconsistency in prior meta-analytic estimates of the
relationship between mixed EI and job performance (i.e., ˆ⫽.47,
Joseph & Newman, 2010b;ˆ⫽.28, O’Boyle et al., 2011). That is, in
past meta-analyses, O’Boyle and colleagues used an inclusive defi-
nition of job performance that incorporated both subjective ratings
and objective results performance measures (in addition to student
academic performance and self-rated job performance measures),
whereas Joseph and Newman used a narrower definition of the
criterion to include only supervisor-rated job performance (see Table
1). As such, it remains unclear how the mixed EI-job performance
relationship might change across different criterion measures.
With regard to the distinction between subjective ratings versus
objective results measures (e.g., sales, number of widgets produced)
of the criterion, researchers have long lamented that objective mea-
sures of performance tend to be contaminated by factors external to
Figure 2. Model B. Incremental Validity Model (partial mediation). Standardized estimates. All predictors
were allowed to intercorrelate. Model is saturated (df ⫽0), so model fit cannot be estimated (i.e., fit is perfect,
by design).
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
303
SELF-REPORTED EMOTIONAL INTELLIGENCE
the individual (e.g., sales markets, sick leave policies, and equipment
malfunctions; Campbell, 1990;Landy & Farr, 1983;Murphy &
Cleveland, 1995;Smith, 1976), suggesting that objective results mea-
sures reflect both employee performance behavior and environmental
factors that constitute a psychometric nuisance. We here have adopted
Campbell, McCloy, Oppler, and Sager’s (1993) definition of job
performance as employee behavior, and we focused on supervisor
ratings of performance as our primary measure of job performance
behavior (see J. W. Johnson, 2001;Rotundo & Sackett, 2002). For our
own theoretical view on how subjective performance ratings and
objective criterion measures, respectively, relate to mixed EI, we have
borrowed from Aguinis (2013, p. 95) and Grote (1996, p. 37), who
specified that employee KSAOs/traits (e.g., mixed EI) give rise to
employee job performance behaviors, which in turn give rise to
objective results measures of productivity (i.e., a mediation model).
As such, we propose that the effects of mixed EI on results (e.g., sales,
productivity) are downstream from (and explained by) the effects of
mixed EI on rated employee performance behaviors. Therefore, we
predicted that the effect of mixed EI on objective results criteria is
mediated by supervisor ratings of job performance. Unfortunately,
there is a paucity of available primary studies connecting objective
results to several of the KSAOs, which precludes us from testing the
complete multistep mediation model (KSAOs ¡Mixed EI ¡Sub-
jective job performance ¡Objective results). Therefore, we can only
test the final three steps of this mediation sequence in the current study
(i.e., Mixed EI ¡Subjective job performance ¡Objective results;
see Figure 4).
Method
To test our hypothesized models, we first updated the corre-
lations of both mixed EI and ability EI with job performance.
Table 1 lists the primary studies that were originally coded in
the meta-analyses of Joseph and Newman (2010b) and O’Boyle
et al. (2011), as well as the primary studies uniquely included in
the current analysis. We also conducted 16 original meta-
analyses, estimating the bivariate relationships of both general
self-efficacy and self-rated job performance with mixed EI,
ability EI, Emotional Stability, Conscientiousness, Extraver-
sion, and cognitive ability (shown in Table 2). Then, by com-
bining published meta-analyses with our original meta-analyses, we
formed a meta-analytic correlation matrix (Table 3). We used this
meta-analytic correlation matrix as the basis for a series of
structural models to test (a) the amount of variance in mixed EI
measures captured by a set of seven predictors and (b) the effect
of these predictors on the mixed EI–job performance relation-
ship (see Figures 1,2, and 3). Although some scholars have
advocated the combination of meta-analysis with structural
equation modeling (Shadish, 1996;Viswesvaran & Ones,
1995), others have pointed out potential limitations of the
approach because this process (a) uses a pooled correlation
matrix instead of a covariance matrix, (b) lacks a definitive
sample size for the meta-analytic correlation matrix, (c) as-
sumes the elements in the meta-analytic correlation matrix
represent a common population, and (d) ignores second-order
sampling error (see Cheung & Chan, 2005;Landis, 2013;
Newman, Jacobs, & Bartram, 2007). Unfortunately, the only
alternative procedure for testing a structural model with meta-
analytic data (i.e., two-stage structural equation modeling, or
TSSEM; Cheung & Chan, 2005) requires at least one primary
study to measure all of the constructs included in the model, and
because no primary study in the current meta-analytic database
met this requirement, we instead used meta-analytic SEM. In
doing so, we followed Landis’s (2013) set of recommendations
(i.e., we drew the elements in the matrix that were not estimated
as part of the current study from published meta-analyses rather
than conducting mini-meta-analyses, and we warn the reader
that causal inferences cannot be drawn from these analyses). As
for the problem of failing to specify a particular target popu-
Figure 3. Model C. Full Mediation Model. Standardized estimates. All predictors were allowed to intercor-
relate.
ⴱ
p⬍.05;
2
(df ⫽7) ⫽232.84 (p⬍.05), root-mean-square error of approximation ⫽.22, comparative
fit index ⫽.88, Tucker–Lewis index ⫽.37, standardized root-mean-square residual ⫽.07 (model fit is poor).
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
304 JOSEPH, JIN, NEWMAN, AND O’BOYLE
Table 1
Primary Studies of the Relationships of Mixed EI and Ability EI With Job Performance (Comparing Current Meta-Analysis to Joseph & Newman, 2010b, and O’Boyle et al.,
2011)
Study Sample size EI Measure Performance measure Effect size Meta-analysis
a
Mixed EI and job performance
Austin, Evans, Goldwater, & Potter
(2005) 154 Austin, Saklofske, Huang, &
McKenney (2004) Academic performance .22 B
Bachman, Stein, Campbell, &
Sitarenios (2000) 36 EQ-i (Bar-On, 1997) Success in debt collection .30 (original t⫽1.848) B
Brizz (2004) 32 ECI (2nd ed.; Wolff, 2006) Parishioner support (sacramental support
plus financial support) .12 B
F. W. Brown, Bryant, & Reilly (2006) 95 EQ-i (Bar-On, 1997) Subordinate-rated leader effectiveness ⫺.02 B
Budnik (2003) —— — —B
Byrne (2003) 325 ECI (Sala, 2002) Supervisor-rated performance based on
Managerial Skills Questionnaire
(Smither & Seltzer, 2001)
.27 A, C
Byrne, Dominick, Smither, & Reilly
(2007) 161 ECI (2nd ed.; Wolff, 2006) Coworker (e.g., peers, supervisors,
subordinates) rating of managerial skills .27 B
Carmeli (2003) 98 Schutte et al. (1998) Self-rated job performance .32 B
Carmeli & Josman (2006) 215 Schutte et al. (1998) Supervisor-rated task performance .47 A, B, C
Cavins (2005) 73 EQ-i (Bar-On, 1997) Director-rated student leader performance .29 (original F⫽6.287) B
Chipain (2003) 120 Success Tendencies Indicator (STI;
(Taccarino & Leonard, 1999)Objective sales performance .42 B
Drew (2007) 40 EQ-i (Bar-On, 1997) Student teacher performance (mixture of
other-rating and self-rating) .31 B
Dulewicz, Higgs, & Slaski (2003) 53 EIQ (Dulewicz & Higgs, 1999,2000) Supervisor-rated management performance .32 A, B, C
Gabel, Dolan, & Cerdin (2005) 59 EQ-i Spanish version (Ugarriza,
2001)Supervisor-rated job performance .06 C
Goldsmith (2008) 24 EQ-i (Bar-On, 1997) Supervisor-rated workplace performance .20 A, C
Government Accounting Office (1998) —— — —B
Hader (2007) 129 EQI (Rahim et al., 2002) Supervisor-rated job performance .29 A, C
Hanna (2008) 46 ECI (Sala, 2002) Supervisor-rated residence hall assistants
job performance .21 A, C
Higgs (2004) 289 EIQ-G (Dulewicz & Higgs, 2000) Performance assessment by the personnel
department .22 B
Hopkins & Bilimoria (2008) 75 (Male) ECI (Boyatzis & Goleman, 2001,
composite of other ratings) Supervisor-rated success (annual
performance plus annual potential) .23 B
30 (Female) ECI (Boyatzis & Goleman, 2001,
composite of other ratings) Supervisor-rated success (annual
performance plus annual potential) .27 B
Jennings & Palmer (2007) 40 360-degree Genos Emotional
Intelligence Inventory (Gignac,
2010)
Objective performance .43 B
Kostman (2004) 147 Bedwell Emotional Judgment
Inventory (Bedwell, 2002)Supervisor-rated job performance .31 A, C
Lii & Wong (2008) 152 Emotional Intelligence Quotient
Inventory (based on Salovey &
Mayer, 1990)
Self-rated oversea adjustment .18 B
Perlini & Halverson (2006) 79 EQ-i (Bar-On, 1997) Hockey player performance ⫺.16 B
Prati (2004) 209 Schutte et al. (1998) Supervisor-rated job performance .15 C
Rozell, Pettijoh, & Parker (2004) 103 Schutte et al. (1998) Self-rated sales performance .20 B
Sardo (2005) —— — —B
Schumacher (2005) 35 ECI-U (Boyatzis & Sala, 2004) Supervisor-rated performance .35 A, C
(table continues)
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
305
SELF-REPORTED EMOTIONAL INTELLIGENCE
Table 1 (continued)
Study Sample size EI Measure Performance measure Effect size Meta-analysis
a
Semadar, Robins, & Ferris (2006) 136 SUEIT (Palmer & Stough, 2001) Supervisor-rated job performance .25 A, C
Sergio (2001) 134 ECI (Sala, 2002) Supervisor-rated job performance
2
⫽34.27 B
Slaski & Cartwright (2002) 224 EQ-i (Bar-On, 1997) Supervisor-rated management performance .22 C
Stone, Parker, & Wood (2005) 383 EQ-i (Bar-On, 1997) Supervisor-rated task-oriented leadership
abilities .14 B, C
Tombs (2005) 60 EQ-i (Bar-On, 1997) Objective performance .28 B
Vieira (2008) 145 Leadership competency inventory
designed to measure Goleman’s
(1995) EI competencies
Supervisor-rated job performance ⫺.07 C
M. B. Wu (2008) 36 EQ-i (Bar-On, 1997 Overall self-rated resident advisor
performance .35 B
Zizzi, Deaner, & Hirschhorn (2003) 21 pitchers Schutte et al. (1998) Objective baseball performance .34 B
40 hitters .01 B
Ability EI and job performance
Ashkanasy & Dasborough (2003) 119 MSCEIT (Mayer et al., 2002) Overall course assessment .20 B
Blickle et al. (2009) 210 TEMINT (Schmidt-Atzert & Bühner,
2002)Supervisor-rated overall performance .15 C
Bryant (2005) 62 MSCEIT (Mayer & Salovey, 1997) Objective sales performance ⫺.09 B
Byron (2007) 58 DANVA2 (Nowicki, 2000) Supervisor-rated managerial performance .22 B, C
Christiansen, Janovics, & Siers (2010) 69 MSCEIT (Mayer et al., 2000) Supervisor-rated job performance .21 C
Cobêro, Primi, & Muniz (2006) 119 MSCEIT (Mayer et al., 2002) Supervisor-rated job performance .18 A, C
Collins (2002) 52 MSCEIT (Mayer et al., 2000) Multirater feedback of executive success ⫺.08 B
Côté & Miners (2006) 175 MSCEIT (Mayer et al., 2002) Supervisor-rated job performance .32 A, B, C
Farh, Seo, & Tesluk (2012) 212 MSCEIT (Mayer et al., 2002) Supervisor-rated job performance .08 C
Goldsmith (2008 24 MSCEIT (Mayer et al., 2002) Supervisor-rated job performance .11 A, C
Graves (1999) 150 EKT (short version of MEIS; Mayer
& Salovey, 1997)Performance in simulated activities .10 A
69 MSCEIT (Mayer et al., 1999) Performance in simulated activities .24 A
Hanna (2008) 46 MSCEIT (Mayer et al., 2002) Supervisor-rated residence hall assistants
job performance ⫺.12 A,C
Herbst, Maree, & Sibanda (2006) 138 MSCEIT (Mayer et al., 2002) Transformational Leadership Practices .05 B
Kerr, Garvin, Heaton, & Boyle (2006) 38 MSCEIT (Mayer et al., 2000) Subordinates’ rating of supervisory
leadership effectiveness .39 B
Kluemper (2006) 66 MSCEIT (Mayer et al., 2002) Supervisor-rated job performance .25 A, B, C
Kluemper, DeGroot, & Choi (2013) 102 MSCEIT (Mayer et al., 2002) Supervisor-rated task performance .22 C
85 MSCEIT (Mayer et al., 2002) Supervisor-rated task performance .22 C
Law, Wong, Huang, & Li (2008) 102 MSCEIT (Mayer et al., 1999) Objective performance measures ⫺.13 A
Muniz & Primi (2007) 80 MSCEIT (Mayer et al., 2002) Supervisor-rated job performance ⫺.01 A,C
Rosete & Ciarrochi (2005) 41 MSCEIT (Mayer et al., 2002) Supervisor-rated job performance .20 A, B, C
Note. In column headed “Source,” A ⫽studies included in Joseph & Newman (2010b);B⫽studies included in O’Boyle et al. (2011);C⫽studies included in the current article. DANVA2 ⫽
Diagnostic Analysis of Nonverbal Accuracy-2; ECI ⫽Emotion-Competence Inventory; ECI–U ⫽Emotional Competence Inventory–University Version; EKT ⫽Emotion Knowledge Test; EQ-i ⫽
Emotional Quotient Inventory; EQI ⫽Emotional Quotient Index; EIQ-G ⫽Emotional Intelligence Questionnaire–General; MEIS ⫽Multifactor Emotional Intelligence Scale; MSCEIT ⫽
Mayer-Salovey-Caruso Emotional Intelligence Test; SUEIT ⫽Swinburne University Emotional Intelligence Test; TEMINT ⫽Test of Emotional Intelligence.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
306 JOSEPH, JIN, NEWMAN, AND O’BOYLE
lation to which the correlation matrix corresponds—this ap-
pears to be a ubiquitous limitation that plagues the vast majority
of studies in organizational research and is not unique to
meta-analytic SEM.
Literature Search
In order to estimate the structural models, we compiled a cor-
relation matrix based on meta-analytic estimates from 20 pub-
lished meta-analytic correlations plus 16 original meta-analyses. If
multiple meta-analyses had been published on a particular bivari-
ate relationship, we used the most recent (which was also the most
comprehensive) one. The 16 original meta-analyses included up-
dates of the relationships of both mixed EI and ability EI with
supervisor-rated job performance, as well as the relationships of
both general self-efficacy and self-rated job performance with
cognitive ability, personality traits, and EI. Several strategies were
used to locate primary studies included in the original meta-
analyses. First, we conducted a literature search in the databases
PsycINFO, ERIC, Social Science Citation Index, Google Scholar,
and Dissertation Abstracts International for published and unpub-
lished studies, using combinations and variations of the following
keywords: emotional intelligence,cognitive ability, self-efficacy,
and self-rated job performance. Second, we also cross-checked
reference lists from previous meta-analyses and reviews on similar
topics as well as studies that cited the original scale development
articles for general/generalized self-efficacy (Chen et al., 2001;
Judge, Erez, Bono, & Thoresen, 2002;Judge, Locke, Durham, &
Kluger, 1998;Schwarzer, Bassler, Kwiatek, Schröder, & Zhang,
1997;Sherer et al., 1982).
In accordance with our a priori construct definitions and
research interests, several rules were established for the inclu-
sion of primary studies. First, the analysis was limited to adult
participants (ages 16–70 years, excluding young adolescents
and institutionalized populations). Second, any studies that did
not operationalize general self-efficacy in a manner consistent
with the definition of general self-efficacy from Sherer et al.
(1982), as a trait-like construct that represents global mastery
expectancies, were excluded (e.g., measures of task-specific or
state self-efficacy were excluded, mimicking the procedures of
Judge & Bono, 2001). Composite measures of confidence in
performing tasks across several, specific domains (e.g., Ber-
nard, Hutchison, Lavin, & Pennington, 1996), or self-efficacy
measures that were specific to a particular setting (e.g., Jones,
1986) were also excluded. In addition, measures that claimed to
assess general self-efficacy but appeared to represent another
construct (e.g., the personal mastery measure from Pearlin &
Table 2
Results From Original Meta-Analyses
Variable kNrˆSD
95% CI 80% CI
LL UL LL UL
Job performance (supervisor-rated)
Mixed EI 15 2,168 .23 .29 .13 .21 .38 .13 .46
Ability EI 13 1,287 .17 .20 .03 .13 .26 .15 .24
General self-efficacy 13 2,703 .10 .13 .00 .09 .18 .13 .13
Self-rated job performance
Mixed EI 10 1,601 .36 .41 .09 .34 .49 .29 .54
Ability EI 3 219 .00 .00 .09 ⫺.19 .20 ⫺.12 .12
Conscientiousness 8 2,621 .25 .31 .09 .23 .39 .19 .43
Extraversion 8 2,621 .19 .23 .06 .16 .29 .14 .31
Emotional Stability 8 2,621 .22 .26 .13 .16 .37 .09 .43
Cognitive ability 4 3,298 .03 .04 .05 ⫺.03 .10 ⫺.02 .10
General self-efficacy 3 686 .41 .51 .11 .36 .66 .37 .65
General self-efficacy
Mixed EI 9 1,847 .37 .45 .13 .35 .54 .28 .61
Ability EI 5 709 .30 .36 .40 ⫺.01 .72 ⫺.15 .87
Conscientiousness 30 10,027 .45 .54 .26 .44 .63 .21 .87
Extraversion 23 8,479 .42 .51 .20 .42 .59 .25 .76
Emotional Stability 46 12,510 .48 .56 .12 .52 .59 .40 .71
Cognitive ability 13 4,085 .07 .09 .06 .04 .13 .01 .16
Note. k ⫽number of effect sizes in the meta-analysis; N⫽total sample size in the meta-analysis; r⫽sample-size weighted mean correlation; ˆ⫽
correlation corrected for attenuation in predictor and criterion; SD
⫽standard deviation of corrected correlation; mixed emotional intelligence (EI) and
ability EI correlations with supervisor-rated job performance are also corrected for range restriction; 95% CI ⫽95% confidence interval; 80% CI ⫽80%
credibility interval; LL ⫽lower limit; UL ⫽upper limit.
Figure 4. Mediation model for objective results criteria. Estimates were standardized: N⫽1,846,
2
(1) ⫽7.69
(p⬍.05), root-mean-square error of approximation ⫽.060, comparative fit index ⫽.99, Tucker–Lewis Index ⫽
.96, standardized root-mean-square residual ⫽.02.
ⴱ
p⬍.05. EI ⫽emotional intelligence.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
307
SELF-REPORTED EMOTIONAL INTELLIGENCE
Schooler, 1978) were also excluded. The only exception to this
inclusion rule was in regard to self-efficacy’s correlation with
ability EI. Because there were no general self-efficacy primary
studies available to estimate this effect, we used primary studies
of specific self-efficacy for this particular cell in the correlation
matrix. Third, with regard to job performance measures, we
invoked a set of conservative standards: (a) only the job per-
formance of employed individuals was included; performance
of specific cognitive or noncognitive tasks, lab experiments,
assessment center ratings, and training performance were ex-
cluded; (b) student academic performance and grade point
averages (GPAs) were excluded; (c) studies measuring only
contextual performance or organizational citizenship behavior
were excluded; and (d) studies that provided objective measures
or third-party evaluations of job performance were excluded
because, to be consistent with other meta-analyses in our cor-
relation matrix, we were only interested in supervisor ratings of
job performance. Primary studies of self-rated job performance
were selected according to the same inclusion rules, with one
exception. In order to obtain an adequate sample size for the
relationships between personality/cognitive ability and self-
rated job performance, we chose to include two studies (Os-
wald, Schmitt, Kim, Ramsay, & Gillespie, 2004;Schmitt et al.,
2007) that used behaviorally anchored rating scales across 12
dimensions of college performance (these studies were included
in effect size estimates for the relationships between personal-
ity/cognitive ability and self-rated performance). Results with
and without these two studies were very similar; removing these
studies did not change the relationships by more than .03.
Fourth, any performance-based (e.g., multiple-choice/right–
wrong) measure of EI based on Salovey and Mayer’s (1990)
ability model was coded as ability EI, and all self-report mea-
sures of EI (excluding self-report measures of ability EI; e.g.,
Wong & Law, 2002) were coded as measures of mixed EI.
(Note: We classified the Schutte et al. [1998] measure of EI as a
self-report mixed EI measure; although the original measure is pur-
portedly based on Salovey and Mayer’s [1990] model, the dimensions
of this self-report scale—empathy, self-management of emotions,
utilization of emotions, and management of others’ emotions [Chan,
2003]—do not align with the dimensions of Salovey and Mayer’s
ability EI model, and the items on the scale appear to capture content
much broader than ability EI [e.g., the item “I expect that I will do
well on most things I try” appears to measure general self-efficacy]).
Fifth, studies that used student GPA or ACT scores to represent
cognitive ability were excluded. We also deleted studies that did not
measure Emotional Stability directly but instead measured a related
trait such as the Sensitivity facet from the California Personality
Inventory (e.g., Baker, 2007) or negative affectivity. Finally, studies
that did not provide enough information to calculate the hypothesized
correlations or did not provide sample sizes were excluded. All
primary studies that were identified as part of the original search, but
subsequently excluded for any of the above reasons, are listed in
Appendix A.
Data Analysis
Following Hunter and Schmidt (2004), we calculated sample-
size-weighted mean correlations, with all effect sizes corrected for
unreliability in both the predictor and criterion. For longitudinal
Table 3
Correlation Table From Meta-Analytic Results
Variable 1 2 3 4 5 6 7 8
1. Mixed emotional intelligence —
2. Ability emotional intelligence .26
a
(10/1572) —
3. Conscientiousness .38
a
(31/5591) .13
a
(21/4155) —
4. Extraversion .46
a
(30/5552) .18
a
(23/4269) .00
c
(632/683001) —
5. Emotional Stability .53
a
(30/5386) .20
a
(22/4401) .26
c
(587/490296) .19
c
(710/440440) —
6. Cognitive ability .11
a
(19/2880) .25
a
(28/5538) ⫺.04
d
(56/15429) .02
d
(61/21602) .09
d
(61/21404) —
7. General self-efficacy .45
b
(9/1847) .36
b
(5/709) .54
b
(30/10027) .51
b
(23/8479) .56
b
(46/12510) .09
b
(13/4085) —
8. Self-rated job performance .41
b
(10/1602) .00
b
(3/219) .31
b
(8/2621) .23
b
(8/2621) .26
b
(8/2621) .04
b
(4/3298) .51
b
(3/686) —
9. Job performance (supervisor-rated) .29
b
(15/2168) .20
b
(13/1287) .21
e
(64/12434) .09
e
(56/9664) .11
e
(53/9184) .44
f
(425/32124) .13
b
(13/2703) .34
g
(115/37752)
Note. Each cell contains the correlation corrected for attenuation in the predictor and criterion, followed by knumber of effect sizes and Nsample size. Correlations of supervisor-rated job performance
with mixed emotional intelligence (EI), ability EI, Big Five traits, and cognitive ability were also corrected for range restriction.
a
Joseph & Newman (2010b).
b
Original meta-analyses from current study.
c
Ones (1993).
d
Judge, Jackson, Shaw, Scott, & Rich (2007).
e
Joseph & Newman (2010b), updated from Hurtz &
Donovan (2000).
f
Hunter & Hunter (1984; see Joseph & Newman, 2010b, p. 63).
g
Heidemeier & Moser (2009).
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
308 JOSEPH, JIN, NEWMAN, AND O’BOYLE
studies that contained multiple measurements, only the effect size
from the initial measure was kept. For a sample with multiple,
facet-level effect sizes of one relationship, we computed a com-
posite correlation according to the formula provided by Nunnally
(1978), or if inadequate information was available to calculate a
composite, we calculated a simple average. In cases where no
reliability information was provided, we adopted estimates from
Viswesvaran and Ones (2000, p. 231) for reliability of Big Five
personality or imputed the average reliability from all available
studies (Hunter & Schmidt, 2004) for non–Big Five measures. For
estimating the reliability of single-item measures of job perfor-
mance, we followed previous approaches (McKay & McDaniel,
2006;Roth, Huffcutt, & Bobko, 2003) using the Spearman–Brown
formula to downwardly correct the average reliability reported
across other primary studies. Following Hunter and Schmidt
(2004), when the standard deviation of the population estimates ()
was smaller than zero, we used zero instead. Also, to maintain
consistency with other job performance meta-analyses in Table 3,
we based range restriction corrections for the relationships be-
tween ability EI/mixed EI and supervisor-rated job performance
upon average ratios of restricted to unrestricted standard deviations
(i.e., .95 for mixed EI and .99 for ability EI, which suggest range
restriction was very minor for the studies included in the current EI
meta-analyses). Duval and Tweedie (2000) trim-and-fill publica-
tion bias analyses were also conducted (no bias was found; results
are available upon request).
Based upon the meta-analytic correlation matrix in Table 3,we
then conducted multiple regression analyses, with mixed EI as the
dependent variable, to test the extent to which mixed EI measures
are sampling the content domains of Conscientiousness, Extraver-
sion, Emotional Stability, ability EI, cognitive ability, and self-
rated qualities. (We also included ability EI as a second dependent
variable, in response to a reviewer comment.) We also conducted
relative importance analyses (J. W. Johnson, 2000;J. W. Johnson
& LeBreton, 2004) to determine which constructs (e.g., Consci-
entiousness, Extraversion, Emotional Stability, ability EI, cogni-
tive ability, general self-efficacy, or self-rated job performance)
contributed the most variance to mixed EI.
Next, we estimated three structural equation models to test the
effects of the KSAOs (common covariates) of mixed EI and job
performance (see Figures 1,2, and 3; note that Figure 2, Model B,
is mathematically equivalent to estimating two multiple regression
models in this case). Model A is our hypothesized heterogeneous
domain sampling model (no-mediation model; Figure 1, Model A),
which specifies no path from mixed EI to supervisor ratings of job
performance. Model B is a fully saturated model (partial-
mediation model; Figure 2, Model B) in which Conscientiousness,
Extraversion, Emotional Stability, ability EI, cognitive ability, and
self-rated qualities predict mixed EI and supervisor ratings of job
performance, and mixed EI also incrementally predicts job perfor-
mance. Model C is a fully-mediated model (Figure 3, Model C)
that is similar to Model B, except the direct effects of all seven
KSAOs are removed so that mixed EI transmits all the KSAO
effects onto supervisor ratings of job performance. Finally, the
fourth model estimates a mediation model from mixed EI to
supervisor ratings of job performance, which in turn lead to
objective results criteria (Figure 4).
Results
Results of the original meta-analyses conducted in the current study
are presented in Table 2 (primary studies included in these original
meta-analyses are presented in Table 4). Regarding the relationship
between mixed EI and job performance, several major adjustments
were made to improve upon the statistical validity and construct
validity of previous meta-analyses. In particular, seven primary stud-
ies were added beyond Joseph and Newman’s (2010b) meta-analysis,
11 primary studies were added beyond O’Boyle et al.’s (2011) meta-
analysis, and 24 primary studies were removed from O’Boyle et al.’s
(2011) analysis (see list of primary studies in Table 1). This update
and refinement resulted in a corrected mean mixed EI-job perfor-
mance correlation of .29, which is considerably smaller than what
Joseph and Newman (2010b) reported (ˆ⫽.47), and closer to the
estimate reported by O’Boyle et al. (2011;ˆ⫽.28). The relationship
between ability EI and job performance was also updated, with a
mean corrected correlation of .20. This is larger than the estimate from
Joseph and Newman (2010b;ˆ⫽.18) but smaller than the O’Boyle
et al. (2011) estimate (ˆ⫽.24). The estimated population correlation
between general self-efficacy and job performance was only .13,
which is smaller than that reported in a previous meta-analysis (Judge
& Bono, 2001,ˆ⫽.23), although this newer estimate is based on
more than twice as much data.
For self-rated job performance, there was a high correlation with
general self-efficacy (ˆ⫽.51) and mixed EI (ˆ⫽.41), but near-zero
relationships with both cognitive ability (ˆ⫽.04) and ability EI (ˆ⫽
.004). With regard to general self-efficacy, results showed that it is
highly correlated with all three personality traits: ˆ⫽.56 with
Emotional Stability, ˆ⫽.54 with Conscientiousness, and ˆ⫽.51
with Extraversion, and it strongly relates to mixed EI (ˆ⫽.45),
whereas it has only a small relationship with cognitive ability (ˆ⫽
.09).
After combining the original meta-analyses we have described
above with the 20 previously published meta-analyses, we created
the final meta-analytic correlation matrix, which we present in
Table 3. On the basis of this correlation matrix, we estimated the
multiple regression models presented in Table 5. Results indicate
62% of the variance in mixed EI is captured by Conscientiousness,
Extraversion, Emotional Stability, ability EI, cognitive ability,
general self-efficacy, and self-rated job performance, suggesting
that a majority of the mix in mixed EI covers content from
well-established psychological concepts (in contrast, only 23% of
the variance in ability EI is captured by these constructs). As an
aside, we note that general self-efficacy has a strong negative
regression coefficient for mixed EI (and for job performance, as
we show later), due to a suppression effect (Cohen, Cohen, West,
& Aiken, 2003;Tzelgov & Henik, 1991) coming from high mul-
ticollinearity of general self-efficacy with the three Big Five fac-
tors and self-rated job performance. Results from the relative
importance analysis, which partitions R
2
and assigns percentages
of R
2
contributed by each predictor (displayed in Table 6), indicate
that the most important predictors of mixed EI, in order, are
Emotional Stability (29.5%), Extraversion (26.5%), Conscien-
tiousness (16.1%), self-rated performance (14.2%), general self-
efficacy (6.8%), and ability EI (5.5%). Thus, the answer to our
research question—What proportion of the variance in mixed EI is
accounted for by Conscientiousness, Extraversion, general self-
efficacy, self-rated job performance, ability EI, Emotional stability,
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
309
SELF-REPORTED EMOTIONAL INTELLIGENCE
Table 4
Primary Studies Included in the 16 Original Meta-Analyses
Study NPredictor measure r
xx
Criterion measure r
yy
r
Adeyemo & Ogunyemi (2005) 300 General Self-Efficacy .82 Mixed EI .76 .32
Best (2002) 819 General Self-Efficacy .85 Emotional Stability .92 .56
Bledow & Frese (2009) 77 General Self-Efficacy .79 Job performance (supervisor-rated) .96 .28
Blickle et al. (2009) 210 Ability EI .81 Job performance (supervisor-rated) .84 .15
Boyar & Mosley (2007) 123 General Self-Efficacy .88 Emotional Stability .79 .22
Boyce, Zaccaro, & Wisecarver (2010) 327 General Self-Efficacy .95 Conscientiousness .85 .49
Boyce, Zaccaro, & Wisecarver (2010) 327 General Self-Efficacy .95 Cognitive Ability .88 ⫺.05
R. F. Brown & Schutte (2006) 167 General Self-Efficacy .86 Mixed EI .85 .58
T. J. Brown, Mowen, Donavan, & Licata (2002) 249 Conscientiousness .73 Self-rated job performance .82 .18
T. J. Brown, Mowen, Donavan, & Licata (2002) 249 Extraversion .86 Self-rated job performance .82 .12
T. J. Brown, Mowen, Donavan, & Licata (2002) 249 Emotional Stability .88 Self-rated job performance .82 .14
Bryan (2007) 57 General Self-Efficacy .67 Mixed EI .95 .55
Burke, Matthiesen, & Pallesen (2006) 460 General Self-Efficacy .85 Conscientiousness .71 .25
Burke, Matthiesen, & Pallesen (2006) 460 General Self-Efficacy .85 Extraversion .70 .36
Burke, Matthiesen, & Pallesen (2006) 460 General Self-Efficacy .85 Emotional Stability .82 .43
Byrne (2003) 325 Mixed EI .92 Job performance (supervisor-rated) .73 .27
Byron (2007) 58 Ability EI .70 Job performance (supervisor-rated) .91 .22
Byron, Terranova, & Nowicki (2007) 109 Ability EI .77 Self-rated job performance .80 .12
Byron, Terranova, & Nowicki (2007) 51 Ability EI .76 Self-rated job performance .80 ⫺.23
Carmeli (2003) 98 Mixed EI .90 Self-rated job performance .87 .32
Carmeli & Josman (2006) 215 Mixed EI .83 Job performance (supervisor-rated) .85 .47
Chan (2004) 158 General Self-Efficacy .80 Mixed EI .61 .33
Chang (2008) 874 Conscientiousness .78 Self-rated job performance .82 .20
Chang (2008) 874 Extraversion .78 Self-rated job performance .82 .19
Chang (2008) 874 Emotional Stability .78 Self-rated job performance .82 .35
G. Chen, Gully, & Eden (2004) 267 General Self-Efficacy .86 Conscientiousness .82 .29
G. Chen, Gully, & Eden (2004) 267 General Self-Efficacy .86 Emotional Stability .82 .41
G. Chen, Gully, & Eden (2004) 148 General Self-Efficacy .82 Conscientiousness .73 .46
G. Chen, Gully, & Eden (2004) 148 General Self-Efficacy .82 Emotional Stability .69 .42
G. Chen, Gully, Whiteman, & Kilcullen (2000) 158 General Self-Efficacy .88 Cognitive Ability .90 .05
G. Chen, Gully, Whiteman, & Kilcullen (2000) 127 General Self-Efficacy .86 Cognitive Ability .90 .08
G. Chen & Klimoski (2003) 70 General Self-Efficacy .88 Job performance (supervisor-rated) .99 ⫺.01
S. X. Chen & Carey (2009) 113 General Self-Efficacy .91 Conscientiousness .83 .27
S. X. Chen & Carey (2009) 113 General Self-Efficacy .91 Extraversion .76 .41
S. X. Chen & Carey (2009) 113 General Self-Efficacy .91 Emotional Stability .86 .37
Christiansen, Janovics, & Siers (2010) 69 Ability EI .78 Job performance (supervisor-rated) .92 .21
Chu (2007) 666 General Self-Efficacy .84 Conscientiousness .78 .47
Chu (2007) 666 General Self-Efficacy .84 Emotional Stability .78 .30
Clemmons (2008) 231 General Self-Efficacy .86 Conscientiousness .78 .34
Clemmons (2008) 231 General Self-Efficacy .86 Extraversion .78 .28
Clemmons (2008) 231 General Self-Efficacy .86 Emotional Stability .78 .29
Cobêro, Primi & Muniz (2006) 119 Ability EI .78 Job performance (supervisor-rated) .89 .18
Converse, Steinhauser, & Pathak (2010) 90 General Self-Efficacy .84 Conscientiousness .78 .31
Côté & Miners (2006) 175 Ability EI .92 Job performance (supervisor-rated) .91 .32
DeRue & Morgeson (2007) 143 General Self-Efficacy .92 Job performance (supervisor-rated) .95 .13
Devonish & Greenidge (2010) 175 Mixed EI .85 Job performance (supervisor-rated) .92 ⫺.03
Dulewicz, Higgs & Slaski (2003) 53 Mixed EI .77 Job performance (supervisor-rated) .58 .32
Durán et al. (2006) 373 General Self-Efficacy .86 Mixed EI .89 .25
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
310 JOSEPH, JIN, NEWMAN, AND O’BOYLE
Table 4 (continued)
Study NPredictor measure r
xx
Criterion measure r
yy
r
Ebstrup, Eplov, Pisinger, & Jorgensen (2011) 3215 General Self-Efficacy .90 Conscientiousness .79 .48
Ebstrup, Eplov, Pisinger, & Jorgensen (2011) 3215 General Self-Efficacy .90 Extraversion .82 .51
Ebstrup, Eplov, Pisinger, & Jorgensen (2011) 3215 General Self-Efficacy .90 Emotional Stability .85 .51
Eissa & Khalifa (2008) 178 General Self-Efficacy .82 Mixed EI .91 .32
Elfenbein, Curhan, Eisenkraft, Shirako, & Baccaro (2008) 149 Negotiation Self-Efficacy .80 Ability EI .88 ⫺.03
Erez & Judge (2001) 124 General Self-Efficacy .78 Job performance (supervisor-rated) .61 .22
Erez & Judge (2001) 124 General Self-Efficacy .78 Conscientiousness .80 .52
Erez & Judge (2001) 124 General Self-Efficacy .78 Emotional Stability .79 .69
Erez & Judge (2001) 473 General Self-Efficacy .90 Emotional Stability .88 .33
Erez & Judge (2001) 112 General Self-Efficacy .80 Emotional Stability .89 .47
Fan, Meng, Billings, Litchfield, & Kaplang (2008) 255 General Self-Efficacy .88 Cognitive Ability .78 .10
Farh, Seo, & Tesluk (2012) 212 Ability EI .88 Job performance (supervisor-rated) .88 .08
Feng, Lu, & Xiao (2008) 513 General Self-Efficacy .88 Job performance (supervisor-rated) .78 .09
Fortunato & Goldblatt (2006) 268 General Self-Efficacy .90 Conscientiousness .84 .54
Foti & Hauenstein (2007) 81 General Self-Efficacy .85 Cognitive Ability .90 .12
Frese et al. (2007) 123 General Self-Efficacy .88 Cognitive Ability .69 .31
Frese et al. (2007) 80 General Self-Efficacy .79 Cognitive Ability .67 .02
Fuller et al. (2011) 405 General Self-Efficacy .89 Extraversion .81 .24
Gabel, Dolan, & Cerdin (2005) 59 Mixed EI .77 Job performance (supervisor-rated) .86 .06
García-lzquierdo, García-lzquierdo, & Ramos-Villagrasa
(2007) 127 General Self-Efficacy .81 Mixed EI .90 .45
Gardner & Pierce (1998) 145 General Self-Efficacy .86 Job performance (supervisor-rated) .94 .11
Gardner & Pierce (2010) 230 General Self-Efficacy .93 Emotional Stability .81 .17
Goldsmith (2008) 24 Mixed EI .78 Job performance (supervisor-rated) .79 .20
Goldsmith (2008) 24 Ability EI .63 Job performance (supervisor-rated) .88 .11
Hader (2007) 129 Mixed EI .68 Job performance (supervisor-rated) .58 .29
Hadley (2003) 151 General Self-Efficacy .84 Extraversion .78 .14
Hanna (2008) 46 Mixed EI .82 Job performance (supervisor-rated) .83 .21
Hanna (2008) 46 Ability EI .87 Job performance (supervisor-rated) .83 ⫺.12
Heggestad & Morrison (2008) 240 Social Self-Efficacy .74 Ability EI .88 .10
D. M. Higgins (2009) 77 Cognitive Ability .83 Self-rated job performance .97 .36
D. M. Higgins, Peterson, Pihl, & Lee (2007) 77 Conscientiousness .81 Self-rated job performance .97 .28
D. M. Higgins, Peterson, Pihl, & Lee (2007) 77 Extraversion .88 Self-rated job performance .97 .28
D. M. Higgins, Peterson, Pihl, & Lee (2007) 77 Emotional Stability .84 Self-rated job performance .97 .20
H. R. Higgins (2001) 175 General Self-Efficacy .82 Conscientiousness .84 .56
H. R. Higgins (2001) 175 General Self-Efficacy .82 Extraversion .81 .29
H. R. Higgins (2001) 175 General Self-Efficacy .82 Emotional Stability .90 .43
R. E. Johnson, Rosen & Djurdjevic (2011) 129 General Self-Efficacy .84 Emotional Stability .84 .59
R. E. Johnson, Rosen & Djurdjevic (2011) 138 General Self-Efficacy .82 Emotional Stability .85 .52
R. E. Johnson, Rosen & Djurdjevic (2011) 223 General Self-Efficacy .83 Emotional Stability .89 .51
R. E. Johnson, Rosen & Djurdjevic (2011) 170 General Self-Efficacy .85 Emotional Stability .86 .64
R. E. Johnson, Rosen & Djurdjevic (2011) 140 General Self-Efficacy .82 Emotional Stability .84 .53
R. E. Johnson, Rosen & Djurdjevic (2011) 132 General Self-Efficacy .84 Emotional Stability .87 .48
R. E. Johnson, Rosen & Djurdjevic (2011) 135 General Self-Efficacy .84 Emotional Stability .88 .27
Judge, Bono, Erez, & Locke (2005) 183 General Self-Efficacy .85 Emotional Stability .89 .49
Judge, Bono, & Locke (2002) 348 General Self-Efficacy .86 Emotional Stability .90 .60
Judge, Erez, Bono, & Thoresen (2002) 702 General Self-Efficacy .94 Conscientiousness .74 .32
Judge, Erez, Bono, & Thoresen (2002) 702 General Self-Efficacy .94 Extraversion .72 .29
Judge, Erez, Bono, & Thoresen (2002) 270 General Self-Efficacy .88 Conscientiousness .91 .49
Judge, Erez, Bono, & Thoresen (2002) 270 General Self-Efficacy .88 Extraversion .88 .53
(table continues)
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
311
SELF-REPORTED EMOTIONAL INTELLIGENCE
Table 4 (continued)
Study NPredictor measure r
xx
Criterion measure r
yy
r
Judge, Erez, Bono, & Thoresen (2002) 124 General Self-Efficacy .88 Conscientiousness .90 .46
Judge, Erez, Bono, & Thoresen (2002) 124 General Self-Efficacy .88 Extraversion .75 .35
Judge, Erez, Bono, & Thoresen (2002) 72 General Self-Efficacy .87 Conscientiousness .90 .12
Judge, Erez, Bono, & Thoresen (2002) 72 General Self-Efficacy .87 Extraversion .75 .29
Judge, Erez, Bono, & Thoresen (2002) 440 General Self-Efficacy .80 Conscientiousness .84 .58
Judge, Erez, Bono, & Thoresen (2002) 440 General Self-Efficacy .80 Extraversion .79 .48
Judge, Erez, Bono, & Thoresen (2002) 277 General Self-Efficacy .85 Conscientiousness .87 .45
Judge, Erez, Bono, & Thoresen (2002) 277 General Self-Efficacy .85 Extraversion .78 .42
Judge, LePine, & Rich (2006) 131 Conscientiousness .80 Self-rated job performance .83 .60
Judge, LePine, & Rich (2006) 131 Extraversion .85 Self-rated job performance .83 .22
Judge, LePine, & Rich (2006) 131 Emotional Stability .81 Self-rated job performance .83 .21
Judge, Locke, Durham, & Kluger (1998) 164 General Self-Efficacy .90 Emotional Stability .93 .67
Judge, Locke, Durham, & Kluger (1998) 122 General Self-Efficacy .83 Emotional Stability .86 .49
Judge, Locke, Durham, & Kluger (1998) 122 General Self-Efficacy .81 Emotional Stability .85 .33
Judge, Thoresen, Pucik, & Welbourne (1999) 514 General Self-Efficacy .75 Job performance (supervisor-rated) .61 .08
Kirk, Schutte, & Hine (2008) 92 Emotional Self-Efficacy .85 Ability EI .91 .34
Kluemper (2006) 66 Ability EI .77 Job performance (supervisor-rated) .90 .25
Kluemper, DeGroot, & Choi (2013) 102 Ability EI .78 Job performance (supervisor-rated) .86 .22
Kluemper, DeGroot, & Choi (2013) 85 Ability EI .78 Job performance (supervisor-rated) .90 .22
Kostman (2004) 147 Mixed EI .79 Job performance (supervisor-rated) .80 .31
Ladebo & Awotunde (2007) 156 General Self-Efficacy .81 Self-rated job performance .76 .22
Langendörfer (2008) 122 General Self-Efficacy .88 Conscientiousness .85 .35
Langendörfer (2008) 122 General Self-Efficacy .88 Extraversion .80 .46
Langendörfer (2008) 122 General Self-Efficacy .88 Emotional Stability .85 .67
Law (2003) 88 General Self-Efficacy .83 Emotional Stability .85 .21
Lee, Stettler, & Antonakis (2011) 460 General Self-Efficacy .84 Job performance (supervisor-rated) .81 .12
Lee, Stettler, & Antonakis (2011) 460 General Self-Efficacy .84 Conscientiousness .78 .45
Lee, Stettler, & Antonakis (2011) 460 General Self-Efficacy .84 Extraversion .78 .39
Lee, Stettler, & Antonakis (2011) 460 General Self-Efficacy .84 Emotional Stability .78 .55
Lee, Stettler, & Antonakis (2011) 460 General Self-Efficacy .84 Cognitive Ability .90 .1
Lindley (2001) 301 General Self-Efficacy .87 Mixed EI .90 .54
Lu, Chang, & Lai (2011) 310 General Self-Efficacy .93 Self-rated job performance .81 .48
Lu, Chang, & Lai (2011) 220 General Self-Efficacy .77 Self-rated job performance .74 .46
Luthans, Avolio, Avey, & Norman (2007) 404 Conscientiousness .78 Self-rated job performance .82 .20
Luthans, Avolio, Avey, & Norman (2007) 404 Extraversion .78 Self-rated job performance .82 .05
Luthans, Avolio, Avey, & Norman (2007) 404 Emotional Stability .78 Self-rated job performance .82 .01
McElroy, Hendrickson, Townsend, & DeMarie (2007) 153 General Self-Efficacy .80 Conscientiousness .90 .59
McElroy, Hendrickson, Townsend, & DeMarie (2007) 153 General Self-Efficacy .80 Emotional Stability .93 .52
McElroy, Hendrickson, Townsend, & DeMarie (2007) 153 General Self-Efficacy .80 Extraversion .91 .36
McKinney (2003) 306 General Self-Efficacy .88 Emotional Stability .91 .46
McKinney (2003) 114 General Self-Efficacy .88 Emotional Stability .91 .39
McNatt & Judge (2004) 57 General Self-Efficacy .84 Job performance (supervisor-rated) .93 ⫺.06
Meier, Semmer, Elfering, & Jacobshagen (2008) 96 General Self-Efficacy .80 Emotional Stability .77 .51
Mirsaleh, Rezai, Kivi, & Ghorbani (2010) 127 General Self-Efficacy .85 Conscientiousness .61 .54
Mirsaleh, Rezai, Kivi, & Ghorbani (2010) 127 General Self-Efficacy .85 Extraversion .76 .39
Mirsaleh, Rezai, Kivi, & Ghorbani (2010) 127 General Self-Efficacy .85 Emotional Stability .79 .52
Muniz & Primi (2007) 80 Ability EI .78 Job performance (supervisor-rated) .81 ⫺.01
Oh & Berry (2009) 239 Conscientiousness .92 Self-rated job performance .88 .27
Oh & Berry (2009) 239 Extraversion .95 Self-rated job performance .88 .32
Oh & Berry (2009) 239 Emotional Stability .93 Self-rated job performance .88 .28
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
312 JOSEPH, JIN, NEWMAN, AND O’BOYLE
Table 4 (continued)
Study NPredictor measure r
xx
Criterion measure r
yy
r
Okech (2004) 180 Teaching Self-Efficacy .77 Ability EI .90 .87
Ono, Sachau, Deal, Englert, & Taylor (2011) 38 Mixed EI .79 Job performance (supervisor-rated) .95 .45
Oreg (2003) 134 General Self-Efficacy .93 Conscientiousness .84 .36
Oreg (2003) 134 General Self-Efficacy .93 Extraversion .87 .49
Oreg (2003) 134 General Self-Efficacy .93 Emotional Stability .79 .21
Oswald et al. (2004) 611 Conscientiousness .83 Self-rated job performance .80 .30
Oswald et al. (2004) 611 Extraversion .88 Self-rated job performance .80 .24
Oswald et al. (2004) 611 Emotional Stability .84 Self-rated job performance .80 .15
Oswald et al. (2004) 611 Cognitive Ability .83 Self-rated job performance .80 ⫺.01
Owens (2009) 104 General Self-Efficacy .84 Conscientiousness .78 .40
Owens (2009) 103 General Self-Efficacy .84 Cognitive Ability .90 .29
Parker (2007) 58 General Self-Efficacy .69 Job performance (supervisor-rated) .61 .05
Petrides & Furnham (2006) 87 Mixed EI .84 Self-rated job performance .80 .33
Petrides & Furnham (2006) 80 Mixed EI .89 Self-rated job performance .80 .03
Piccolo, Judge, Takahashi, Watanabe, & Locke (2005) 271 General Self-Efficacy .80 Emotional Stability .86 .41
Pierro (1997) 98 General Self-Efficacy .84 Conscientiousness .86 .58
Pierro (1997) 98 General Self-Efficacy .84 Extraversion .82 .48
Pierro (1997) 98 General Self-Efficacy .84 Emotional Stability .79 .20
Platt (2010) 97 General Self-Efficacy .84 Job performance (supervisor-rated) .98 ⫺.10
Prati (2004) 209 Mixed EI .89 Job performance (supervisor-rated) .94 .15
Ramassini (2000) 204 General Self-Efficacy .84 Emotional Stability .78 .49
Reece (2007) 150 General Self-Efficacy .96 Emotional Stability .91 .55
Robinson (2009) 160 General Self-Efficacy .78 Cognitive Ability .90 .08
Rode et al. (2008) 59 Ability EI .88 Self-rated job performance .80 ⫺.01
Rosete & Ciarrochi (2005) 41 Ability EI .78 Job performance (supervisor-rated) .89 .20
Rozell, Pettijohn, & Parker (2004) 103 Mixed EI .83 Self-rated job performance .85 .20
Schendel (2010) 48 Counselor Activity Self-
Efficacy .95 Ability EI .82 .10
Schimtt et al. (2007) 2488 Cognitive Ability .83 Self-rated job performance .74 .03
Schumacher (2005) 35 Mixed EI .68 Job performance (supervisor-rated) .74 .35
Semadar, Robins, & Ferris (2006 136 Mixed EI .94 Job performance (supervisor-rated) .92 .25
Sevinc (2001) 69 Mixed EI .80 Self-rated job performance .80 .20
Shahzad, Sarmad, Abbas, & Khan (2011) 100 Mixed EI .82 Self-rated job performance .73 .43
Sjoberg, Littorin, & Engelberg (2005) 45 Mixed EI .76 Self-rated job performance .80 .25
Slaski & Cartwright (2002) 224 Mixed EI .79 Job performance (supervisor-rated) .80 .22
Smith & Foti (1998) 160 General Self-Efficacy .88 Cognitive Ability .90 .06
Sovern (2008) 206 General Self-Efficacy .73 Emotional Stability .89 .44
Stewart, Palmer, Wilkin, & Kerrin (2008) 110 General Self-Efficacy .86 Conscientiousness .81 .41
Stewart, Palmer, Wilkin, & Kerrin (2008) 110 General Self-Efficacy .86 Emotional Stability .88 .64
Stone, Parker, & Wood (2005) 383 Mixed EI .79 Job performance (supervisor-rated) .89 .39
Stone, Parker, & Wood (2005) 412 Mixed EI .93 Self-rated job performance .83 .37
Strobel, Tumasjan, & Sporrle (2011) 180 General Self-Efficacy .85 Job performance (supervisor-rated) .77 .15
Strobel, Tumasjan, & Sporrle (2011) 180 General Self-Efficacy .85 Conscientiousness .81 .37
Strobel, Tumasjan, & Sporrle (2011) 180 General Self-Efficacy .85 Extraversion .75 .43
Strobel, Tumasjan, & Sporrle (2011) 180 General Self-Efficacy .85 Emotional Stability .86 .54
Stumpp, Muck, Hulsheger, Judge, & Maier (2010) 199 General Self-Efficacy .87 Conscientiousness .83 .51
Stumpp, Muck, Hulsheger, Judge, & Maier (2010) 199 General Self-Efficacy .87 Extraversion .80 .45
Stumpp, Muck, Hulsheger, Judge, & Maier (2010) 199 General Self-Efficacy .87 Emotional Stability .82 .61
Sturman (2011) 119 General Self-Efficacy .84 Conscientiousness .78 .54
Sturman (2011) 119 General Self-Efficacy .84 Extraversion .78 .34
(table continues)
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
313
SELF-REPORTED EMOTIONAL INTELLIGENCE
Table 4 (continued)
Study NPredictor measure r
xx
Criterion measure r
yy
r
Sturman (2011) 119 General Self-Efficacy .84 Emotional Stability .78 .60
Tews, Michel, & Noe (2011) 265 General Self-Efficacy .81 Job performance (supervisor-rated) .77 .15
Tews, Michel, & Noe (2011) 265 General Self-Efficacy .81 Cognitive Ability .90 .06
Timmerman (2008) 293 General Self-Efficacy .78 Emotional Stability .84 .54
van Hooft, van der Flier, & Minne (2006) 122 Cognitive Ability .83 Self-rated job performance .82 .05
Vieira (2008) 145 Mixed EI .58 Job performance (supervisor-rated) .46 ⫺.07
Wang (2002) 186 General Self-Efficacy .82 Mixed EI .76 .23
M. B. Wu (2008) 36 Mixed EI .93 Self-rated job performance .96 .35
M. B. Wu (2008) 36 Conscientiousness .82 Self-rated job performance .96 .46
M. B. Wu (2008) 36 Extraversion .72 Self-rated job performance .96 .23
M. B. Wu (2008) 36 Emotional Stability .82 Self-rated job performance .96 .42
Y. Wu (2011) 571 Mixed EI .88 Self-rated job performance .86 .44
Xie, Roy, & Chen (2006) 1786 General Self-Efficacy .89 Cognitive Ability .90 .06
Yamkovenko & Holton (2010) 252 General Self-Efficacy .88 Conscientiousness .81 .58
Yamkovenko & Holton (2010) 252 General Self-Efficacy .88 Extraversion .77 .43
Yamkovenko & Holton (2010) 252 General Self-Efficacy .88 Emotional Stability .86 .35
Note. When reliability information was not available in the primary study, the average reliability of all available measures included in the original meta-analyses was substituted. EI ⫽emotional
intelligence; r
xx
⫽reliability of the predictor; r
yy
⫽reliability of the criterion.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
314 JOSEPH, JIN, NEWMAN, AND O’BOYLE
and cognitive ability?—is that a majority of variance in mixed EI
(62%; multiple R⫽.79) is accounted for by these constructs, and
the most important predictors of mixed EI are personality traits and
self-perceptions.
Next, we estimated the models in Figures 1,2, and 3. The
sample size for these models was set at 2,168, which is the sample
size for the mixed EI–job performance bivariate relationship.
When no common covariates were taken into consideration, there
was a statistically significant direct effect (⫽.29; standardized
coefficient) from mixed EI to job performance (i.e., the bivariate
correlation). When the theorized antecedents (ability EI, Emo-
tional Stability, cognitive ability, Conscientiousness, Extraversion,
general self-efficacy, and self-rated job performance) were speci-
fied as common covariates of both mixed EI and job performance
(Figure 2, Model B), the mixed EI effect on job performance
dropped from ⫽.29 to near zero (⫽–.02, ns). Indeed, our
hypothesized model, which specified no incremental validity for
mixed EI in the presence of the seven KSAOs (i.e., the heteroge-
neous domain sampling model; Figure 1, Model A), displayed
nearly perfect model fit indices [
2
(df ⫽1) ⫽0.19 (p⬎.05),
RMSEA ⫽.00, CFI ⫽1.00, TLI ⫽1.01, SRMR ⫽.001]. These
results support our expectation that mixed EI fails to exhibit
incremental validity when a set of common causes of mixed EI and
job performance are controlled. Consistent with these results, the
full mediation model (Figure 3, Model C) yielded poor model fit
[
2
(df ⫽7) ⫽232.84 (p⬍.05), RMSEA ⫽.22, CFI ⫽.88, TLI ⫽
.37, SRMR ⫽.07]. Note that Model B is saturated (df ⫽0), and
thus, the fit indices are meaningless (all fit indices take their
maximum values, by design).
Finally, a meta-analysis of the relationship between mixed EI and
objective results measures of performance was conducted (see Ap-
pendix B), in order to compare the bivariate mixed EI-performance
relationship across different criteria (i.e., supervisor ratings of perfor-
mance vs. objective results criteria). The meta-analytic relationship
between mixed EI and objective results performance measures was
ˆ⫽.17 (k⫽11, N⫽1,846), which is smaller than the estimated
relationship between mixed EI and subjective supervisor ratings of
job performance (ˆ⫽.29, k⫽15, N⫽2,168). This finding was
consistent with our theoretical expectation that mixed EI (as an
employee KSAO/trait) would affect objective/results performance by
way of supervisor-rated job performance behavior (see Figure 4). To
test this assertion, we entered the previously described meta-analytic
correlations into a mediation model (for the correlation between
objective results and subjective performance ratings, we used Bom-
mer, Johnson, Rich, Podsakoff, & MacKenzie’s [1995] meta-analytic
estimate of ˆ⫽.39). The practical fit of this mediation model
[
2
(df ⫽1) ⫽7.69 (p⬍.05), N⫽1,846, RMSEA ⫽.060, CFI ⫽
.99, TLI ⫽.96, SRMR ⫽.02] was deemed adequate, and the indirect
effect of mixed EI on objective results performance was statistically
significant (95% Monte Carlo confidence interval [.09, .13]; Preacher
& Selig, 2012; see Figure 2). If we had additionally estimated the
direct effect from mixed EI to objective results performance (df ⫽0;
saturated model), the direct path coefficient would have been
small (⫽.06; p⬍.05), and the path from supervisor-rated job
performance to objective results would have fallen a negligible
amount, from ⫽.39 to ⫽.37. Altogether, these results
support our assertion that mixed EI primarily relates to objec-
tive results criteria by way of its relationship with supervisor-
rated job performance (Figure 4).
Table 5
Meta-Analytic Regression Predicting Mixed EI, Ability EI, and Job Performance
Predictor
Dependent variable
Mixed EI Ability EI Job performance Job performance
Ability EI .20
ⴱ
— .18
ⴱ
.19
ⴱ
Conscientiousness .45
ⴱ
⫺.07
ⴱ
.33
ⴱ
.34
ⴱ
Extraversion .56
ⴱ
⫺.04 .20
ⴱ
.21
ⴱ
Emotional Stability .52
ⴱ
⫺.03 .09
ⴱ
.11
ⴱ
Cognitive ability .06
ⴱ
.21
ⴱ
.43
ⴱ
.42
ⴱ
General self-efficacy ⫺.61
ⴱ
.54
ⴱ
⫺.52
ⴱ
⫺.53
ⴱ
Self-rated performance .31
ⴱ
⫺.25
ⴱ
.41
ⴱ
.42
ⴱ
Mixed EI — — — ⫺.02
R
2
.62
ⴱ
.23
ⴱ
.3948
ⴱ
.3950
ⴱ
Adjusted R
2
.61
ⴱ
.20
ⴱ
.3928
ⴱ
.3927
ⴱ
⌬R
2
.0002
Note. Standardized regression coefficients. For mixed emotional intelligence (EI), harmonic mean N⫽2,127;
for ability EI, harmonic mean N⫽2,006; for job performance, N⫽2,168 (i.e., the sample size for the mixed
EI–job performance bivariate relationship).
ⴱ
p⬍.05.
Table 6
Relative Importance Analysis
Variable
Mixed emotional intelligence
Raw relative
weights % of R
2
Ability EI .034 5.5
Conscientiousness .100 16.1
Extraversion .166 26.5
Emotional Stability .183 29.5
Cognitive Ability .007 1.1
General self-efficacy .042 6.8
Self-rated performance .088 14.2
R
2
.62
Note. EI ⫽emotional intelligence.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
315
SELF-REPORTED EMOTIONAL INTELLIGENCE
Discussion
The link between emotional intelligence and work outcomes such
as job performance has been an area of major controversy (Cherniss,
2010;Murphy, 2006). Despite ever-growing attention from both the
public and academia, and despite the well-known hyperclaims regard-
ing the criterion-related validity of mixed EI in predicting workplace
success (e.g., Goleman, 1995), it has heretofore been unclear what
mixed EI instruments measure, and why these instruments predict job
performance so well. The current study contributed to the existing
literature in two ways. First, we opened the black box of mixed EI
construct validity by examining the extent to which mixed EI mea-
sures capture content from the following constructs: Conscientious-
ness, Extraversion, general self-efficacy, self-rated performance, abil-
ity EI, Emotional Stability, and cognitive ability. Results demonstrate
that a majority of the variance in mixed EI measures is captured by
these constructs (i.e., 62%; multiple R⫽.79), suggesting these
measures tend to sample content from various well-established con-
struct domains in psychology.
Second, based on a combination of original and published meta-
analytic results, we estimated the extent to which mixed EI demon-
strates incremental validity over the seven well-established constructs
(Figure 1) in hopes of answering the question, “Why does mixed EI
strongly predict job performance?” Our results indicated that after
controlling for these constructs, the relationship between mixed EI
and job performance dropped to near zero (⫽⫺.02; ns). Based
upon these findings, the current study offers the unique insight that the
predictive merit of mixed EI can be almost fully explained after one
considers ability EI, self-perceptions (i.e., general self-efficacy and
self-rated job performance), personality, and cognitive ability. This
result differs from the results of previous analyses (Joseph & New-
man, 2010b;O’Boyle et al., 2011), which demonstrated sizeable
incremental validity for mixed EI beyond the Big Five and cognitive
ability but which did not control for self-perceptions or for ability EI.
En route to the previously stated result (i.e., answering why mixed
EI predicts job performance), we also updated the meta-analytic
correlation of mixed EI with job performance by including more
studies than previous meta-analyses and by applying a strict opera-
tional definition of job performance that focused only on supervisor
ratings of performance. Our result (ˆ⫽.29) was notably smaller than
the .47 estimate reported by Joseph and Newman (2010b) but quite
similar to the effect size (ˆ⫽.28) reported by O’Boyle et al. (2011).
However, we note that O’Boyle et al. (2011) had defined job perfor-
mance very broadly, to include academic performance, sports perfor-
mance, self-rated performance, work adjustment, and other criterion
content (see Table 1). Thus, although the current effect size is similar,
the construct relationship being estimated here is quite different from
that of O’Boyle et al.
Theoretical Implications
We now have a theoretical explanation for why mixed EI predicts
job performance—and it turns out to be largely a psychometric
explanation. Mixed EI measures reflect a heterogeneous combination
of traits that have long been known to predict job performance. That
is, mixed EI measures appear to have been developed (perhaps unin-
tentionally) through a process of heterogeneous domain sampling
from seven well-established content domains.
One implication of the heterogeneous domain sampling model
of mixed EI is that mixed EI researchers can now borrow substan-
tive theory from the constituent constructs of mixed EI. To elab-
orate, because we now know what mixed EI is, we can use theory
from the nomological networks of the seven constituent construct
domains to explain additional outcomes of mixed EI beyond job
performance. For example, the large portion of Emotional Stabil-
ity, Extroversion, and Conscientiousness content in mixed EI
could help explain why mixed EI would be a robust predictor of
job satisfaction (see Judge, Heller, & Mount, 2002) and leadership
(Harms & Credé, 2010;Judge, Bono, Ilies, & Gerhardt, 2002).
Another theoretical implication raised by our study involves
the standards for construct validity itself and the general ques-
tion of whether heterogeneous domain sampling should be
considered a legitimate method for establishing “new” con-
structs. On the one hand, some critics might raise the objection
that discriminant validity is a cornerstone of construct validity
(Campbell & Fiske, 1959), and heterogeneous domain sampling
prevents discriminant validity, by definition (i.e., if mixed EI
directly reflects its constituent constructs, then it cannot be
considered distinct from them). As one example of this, heter-
ogeneous domain sampling might help explain why the discrim-
inant validity of EI ratings from Big Five personality domains
is sometimes weak (see multitrait–multimethod evidence from
Joseph & Newman, 2010a)—because EI ratings explicitly con-
tain some Big Five content. On the other hand, proponents of
heterogeneous domain sampling might contend that creating
novel composites of established constructs is itself a meaning-
ful contribution. Macey and Schneider (2008) made this sort of
argument when they characterized the employee engagement
construct as, “a new blend of old wines” (p. 10), despite the fact
that employee engagement was rather clearly developed via
heterogeneous domain sampling by borrowing content from job
satisfaction, organizational commitment, job involvement, and
job affect (Newman & Harrison, 2008;Newman, Joseph, &
Hulin, 2010). The question of whether heterogeneous domain
sampling can be considered a legitimate new method for scale
development is a major theoretical conundrum that emerges
from the current article, but this question is, as yet, unanswered.
As an aside, we note that proprietary measurement—which is a
useful way to protect intellectual property and recoup the costs of
measurement research and development—is nonetheless a barrier to
scientific progress here, because proprietary measurement hides the
survey items and thereby can hide the fact that a measure was derived
via heterogeneous domain sampling. This practice gives short shrift to
the long-established constituent constructs, which are the predictive
workhorses in newer compound concepts like mixed EI but which are
forced into anonymity by measurement copyrights.
Finally, another natural consequence of the heterogeneous domain
sampling model is the need to ensure more valid construct labeling.
For mixed EI, the question is whether this composite construct should
really be called “emotional intelligence,” or even “emotional compe-
tence” (cf. Cherniss, 2010). Although we do not feel authorized to
supplant the widely adopted “emotional intelligence” label, the im-
plication of the current study for conceptual construct labeling is that
mixed EI measures reflect mixed competence traits (i.e., “mixed EI”
describes individuals who are emotionally stable, outgoing, conscien-
tious, with a high estimation of their own past and future performance,
and [to a lesser extent] emotionally intelligent).
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
316 JOSEPH, JIN, NEWMAN, AND O’BOYLE
Limitations and Future Research
The current research is also vulnerable to certain limitations, which
leaves room for additional corresponding future research. One partic-
ular Big Five trait that deserves further discussion here is Agreeable-
ness. Ample research evidence has supported the overlap between
Agreeableness and mixed EI (e.g., De Raad, 2005;Joseph & New-
man, 2010b;Petrides & Furnham, 2001); however, we did not include
Agreeableness in our model (Figures 1,2, and 3), primarily because
this is a model of the theorized common causes of mixed EI and job
performance. Agreeableness has a negligible relationship with job
performance (Barrick & Mount, 1991), and it has been noted that
qualities such as empathy and interpersonal sensitivity might even
impair job performance when the work situation demands ruthless-
ness and toughness (Zeidner et al., 2004). However, we recommend
that future researchers who investigate the links between mixed EI
and contextual performance (Ilies, Fulmer, Spitzmuller, & Johnson,
2009), counterproductive work behavior (Berry, Ones, & Sackett,
2007), or team performance (Bell, 2007) consider the role of Agree-
ableness as a common cause. We should also note that whereas the
current study controlled for some broad Big Five traits (e.g., Extra-
version, Conscientiousness), Mayer et al. (2008) specifically de-
scribed mixed EI content in terms of narrower facets of these traits
(e.g., gregariousness, assertiveness, impulse control). Future research-
ers should attend to whether these particular personality subfacets can
more parsimoniously explain the mixed EI–job performance relation-
ship.
As suggested by some researchers (Cherniss, 2010;Jordan,
Dasborough, Daus, & Ashkanasy, 2010), future studies could also
explore the influence of the work context on EI. Depending on the
type of job, specific situation, or various kinds of people involved,
different profiles inside the mixed EI “grab bag” may potentially
have different effects. As a meta-analysis, the current study only
speaks to average effects that were obtained across jobs.
It is also worth noting that whereas the current study focused on
how mixed EI appears to demonstrate a lack of incremental va-
lidity after controlling for a linear combination of personality,
self-perceptions, ability EI, and cognitive ability; some proponents
of mixed EI might argue that mixed EI is actually a profile of
various psychological constructs, rather than a simple linear com-
bination, and this profile could demonstrate incremental validity in
predicting job performance. Although this may be the case, the
current study focused on how mixed EI is currently measured (i.e.,
as a linear combination), and additional research would be neces-
sary to investigate the issue of mixed EI profiles. As another issue,
we mention that EI need not have uniformly positive effects. There
could also be a dark side of EI, in which emotionally intelligent
individuals are capable of deviant behavior when motivated (Côté,
DeCelles, McCarthy, Van Kleef, & Hideg, 2011;Kilduff, Chia-
buru, & Menges, 2010).
As one final direction for future research, we note that the relation-
ship between mixed EI and job performance may vary across dimen-
sions of mixed EI. Based on a reviewer’s suggestion, we meta-
analyzed the relationships of mixed EI facets with both job
performance and the covariates shown in Figure 1 (see Appendix C;
note that no primary study correlations were available between gen-
eral self-efficacy and mixed EI facets, therefore specific self-efficacy
was used as a substitute here). Although we could only estimate our
structural models using the facets of Bar-On’s EQ-i (due to a lack of
facet-level data for other mixed EI measures; e.g., see Tables A and
B), Table C shows that the covariates explain between 35% and 56%
of the variance in each mixed EI facet; and Table C2 demonstrates
that after including the covariates, no mixed EI facet retains positive
incremental validity for job performance (although some EI facets
exhibit incremental validity with a negative regression coefficient, due
to suppressor effects). In essence, these facet-level examinations
largely replicate the results found for overall mixed EI: the covariates
explain much of the mixed EI variance (helping to answer the ques-
tion of what mixed EI is), and the covariates also explain the rela-
tionship between mixed EI and job performance (helping to answer
the question of why mixed EI predicts job performance; although we
caution these EI facet-level results are based on a relatively small
amount of data).
Practical Implications
In addition to the currently proposed theoretical enhancement to
our understanding of the mixed EI construct (i.e., our new explanation
for what mixed EI is and why mixed EI predicts job performance), the
findings of the current article have several practical implications.
First, our findings reiterate previous meta-analytic conclusions that
suggested mixed EI predicts supervisor ratings of job performance
rather well—at least as strongly as any other personality construct
(Joseph & Newman, 2010b;O’Boyle et al., 2011; cf. Barrick et al.,
2001). Thus, for practitioners who have little concern about the
overlap between mixed EI and other, well-established psychological
constructs, these results suggest that mixed EI measures may be used
as part of a selection system because they tap into a diffuse, com-
pound construct of personality and self-perceptions that exhibits rea-
sonable criterion-related validity. This conclusion is markedly differ-
ent from Joseph and Newman’s (2010b) admonition to, “exercise
extreme caution when using mixed EI measures” because it was “not
clear why” mixed EI predicts job performance (p. 72). In other words,
despite the fact that mixed EI does not appear to increase scientific
parsimony in the construct space of the organizational sciences, the
current meta-analytic results suggest that practitioners could use a
single mixed EI measure to capture a portion of the criterion-related
validity that could otherwise be captured by using a battery of seven
KSAOs.
However, we note that the criterion-related validity of mixed EI
(r
2
⫽.29
2
⫽.08) falls notably short of the criterion-related validity
for the composite of seven KSAOs (R
2
⫽39; see Table 5)—revealing
that although mixed EI offers no incremental prediction beyond the
seven KSAOs, the seven KSAOs do offer considerable incremental
prediction beyond mixed EI. As such, and given that the majority of
mixed EI measures are proprietary and require fees to administer,
practitioners will likely be faced with a choice between a shorter,
more expensive mixed EI measure with lower criterion-related valid-
ity versus a much longer battery of personality, cognitive ability, and
self-concept measures with notably higher criterion-related validity.
Managing this tradeoff will depend upon practitioners’ judgments
about applicants’ time, willingness, and capability to complete a
lengthy battery of seven KSAOs. Another practical implication of the
current article is that it illustrates a difficult decision practitioners must
make once they have determined they want to assess EI. Practitioners
must choose between ability EI measures, which show a weaker
relationship with job performance but more precisely capture the
notion of EI as an intelligence (MacCann et al., 2014), versus mixed
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
317
SELF-REPORTED EMOTIONAL INTELLIGENCE
EI measures, which show a stronger relationship with job perfor-
mance but broadly measure many constructs in addition to emotional
competencies.
Conclusion
The current study attempted to help unravel the mix of what
mixed EI actually is. According to current results, the active
ingredients in mixed EI—which make it one of the strongest
known personality-based predictors of job performance—include
Conscientiousness, self-efficacy, self-rated performance, and Ex-
traversion (confirming the conjectures of Mayer et al., 2008, and
Newman et al., 2010), in addition to ability EI, Emotional Stabil-
ity, and cognitive ability. These results illustrate that developers of
mixed EI measures may have engaged in heterogeneous domain
sampling (Cronbach & Meehl, 1955;Ghiselli et al., 1981;Nun-
nally, 1967), whereby mixed EI measures were constructed to
sample from various well-known psychological content domains.
Armed with new knowledge of which psychological fundaments
constitute mixed EI measures, the current article aids in the process
of establishing the construct validity of mixed EI. In answer to the
work that questioned whether mixed EI measures should be used
in personnel selection because it was not clear why mixed EI
predicted job performance (Joseph & Newman, 2010b), the current
results suggest that practitioners might be using measures of mixed
EI as a practical, shorthand alternative to a lengthy battery of
several more traditional KSAOs.
References
Adeyemo, D. A. (2007). Moderating influence of emotional intelligence on
the link between academic self-efficacy and achievement of university
students. Psychology and Developing Societies, 19, 199–213. doi:
10.1177/097133360701900204
ⴱ
Adeyemo, D. A., & Ogunyemi, B. (2005). Emotional intelligence and
self-efficacy as predictors of occupational stress among academic staff
in a Nigerian university. E-Journal of Organizational Learning and
Leadership, 4. No. 1. Retrieved from http://www.leadingtoday.org/
weleadinlearning/da05.htm
Aguinis, H. (2013). Performance management (3rd ed.). Upper Saddle
River, NJ: Pearson–Prentice Hall.
Ahmetoglu, G., Leutner, F., & Chamorro-Premuzic, T. (2011). EQ-nomics:
Understanding the relationship between individual differences in Trait
Emotional Intelligence and entrepreneurship. Personality and Individual
Differences, 51, 1028–1033.
Aremu, A. O., & Lawal, G. A. (2009). A path model investigating the
influence of some personal-psychological factors on the career aspira-
tions of police trainees: A perspective from Oyo State, Nigeria. Police
Practice & Research, 10, 239–254. doi:10.1080/15614260802381059
Arthur, W., Jr., & Villado, A. J. (2008). The importance of distinguishing
between constructs and methods when comparing predictors in person-
nel selection research and practice. Journal of Applied Psychology, 93,
435–442. doi:10.1037/0021-9010.93.2.435
Ashkanasy, N. M., & Dasborough, M. T. (2003). Emotional awareness and
emotional intelligence in leadership teaching. Journal of Education for
Business, 79, 18–22. doi:10.1080/08832320309599082
Austin, E. J., Evans, P., Goldwater, R., & Potter, V. (2005). A preliminary
study of emotional intelligence, empathy, and exam performance in first
year medical students. Personality and Individual Differences, 39, 1395–
1405. doi:10.1016/j.paid.2005.04.014
Austin, E. J., Farrelly, D., Black, C., & Moore, H. (2007). Emotional
intelligence, Machiavellianism and emotional manipulation: Does EI
have a dark side? Personality and Individual Differences, 43, 179–189.
Austin, E. J., Saklofske, D. H., & Egan, V. (2005). Personality, well-being
and health correlates of trait emotional intelligence. Personality and
Individual Differences, 38, 547–558.
Austin, E. J., Saklofske, D. H., Huang, S. H. S., & McKenney, D. (2004).
Measurement of trait emotional intelligence: Testing and cross-
validating a modified version of Schutte et al.’s (1998) measure. Per-
sonality and Individual Differences, 36, 555–562. doi:10.1016/S0191-
8869(03)00114-4
Avery, D. R. (2003). Personality as a predictor of the value of voice. Journal
of Psychology, 137, 435–446. doi:10.1080/00223980309600626
Bachman, J., Stein, S., Campbell, K., & Sitarenios, G. (2000). Emotional
intelligence in the collection of debt. International Journal of Selection
and Assessment, 8, 176–182.
Baker, B. A. (2007). Maximizing multisource feedback: The use of goal
setting to facilitate performance improvement (Unpublished doctoral
dissertation). North Carolina State University, Raleigh.
Barchard. (2003). Does emotional intelligence assist in the prediction of
academic success? Educational and Psychological Measurement, 63,
840–858.
Barfoot, D. S. (2007). Antecedents of leader–follower trust in a Christian
church organization (Unpublished doctoral dissertation), Regent Uni-
versity, Virginia Beach, VA.
Bar-On, R. (1997). Bar-On Emotional Quotient Inventory: Technical man-
ual. Toronto, ON, Canada: Multihealth Systems.
Barrick, M. R., & Mount, M. K. (1991). The Big Five personality dimen-
sions and job performance: A meta-analysis. Personnel Psychology, 44,
1–26. doi:10.1111/j.1744-6570.1991.tb00688.x
Barrick, M. R., Mount, M. K., & Judge, T. A. (2001). Personality and
performance at the beginning of the new millennium: What do we know
and where do we go next? International Journal of Selection and
Assessment, 9, 9–30. doi:10.1111/1468-2389.00160
Barrick, M. R., Mount, M. K., & Strauss, J. P. (1993). Conscientiousness
and performance of sales representatives: Test of the mediating effects
of goal setting. Journal of Applied Psychology, 78, 715–722. doi:
10.1037/0021-9010.78.5.715
Barrick, M. R., Stewart, G. L., & Piotrowski, M. (2002). Personality and
job performance: Test of the mediating effects of motivation among
sales representatives. Journal of Applied Psychology, 87, 43–51. doi:
10.1037/0021-9010.87.1.43
Bedwell, S. (2003). Emotional Judgment Inventory (EJI): Administration
and technical manual. Champaign, IL: Institute for Personality and
Ability Testing.
Bell, S. T. (2007). Deep-level composition variables as predictors of team
performance: A meta-analysis. Journal of Applied Psychology, 92, 595–
615. doi:10.1037/0021-901