Group Emotional Competence: A Review and Evaluation Study: September 2009
Group Emotional Competence: A Review and Evaluation Study: September 2009
Group Emotional Competence: A Review and Evaluation Study: September 2009
net/publication/282914301
CITATIONS READS
0 609
1 author:
Margarida Truninger
Center for Creative Leadership
13 PUBLICATIONS 16 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Margarida Truninger on 01 April 2016.
ABSTRACT
The present paper offers a literature review and evaluation of empirical research on
group emotional competence (GEC). GEC is a relatively recent research avenue
within the field of emotional intelligence. It is defined by Druskat & Wolff (1999,
2001, 2008), as the ability of a group to generate a shared set of behavioural norms
that guide and support group members’ awareness and management of emotions.
Druskat & Wolff identify nine emotional competent group norms (9 ECGNs),
proposed to enhance the development of group social capital and heighten the levels
of team trust, collaboration and task engagement that characterize a highly performing
team. The present work begins by reviewing the origins of research on emotional
intelligence and covers its behavioural approach as in emotional competences. Then,
it presents and discusses the recent research endeavour of group emotional
intelligence, group emotional competence, and the GEC’s measurement model. Our
overall objective is twofold: first, to unveil some of the gaps in the literature, notably
through a literature review and a critical assessment of the GEC’s measurement model
and its empirical findings; second, to point out potential new paths for future research.
2
“Feelings, along with the emotions they come from, are not a luxury.
They serve as internal guides, and they help us communicate to others signals that
can also guide them. And feelings are neither intangible nor elusive. Contrary to
traditional scientific opinion, feelings are just as cognitive as other percepts.”
1. Introduction
In 1990, the seminal article by two psychologists, Salovey and Mayer, introduced the
concept of emotional intelligence (EI), offering a new perspective on the traditional
rivalry between emotion and reason. There, they asserted that people could reason
about emotions and use emotions and emotional knowledge to assist thought (Salovey
& Mayer, 1990; Mayer, Roberts & Barsade, 2008).
Seen the prevalence of group work structures in most organizations, wherein personal
interactions abound, Druskat and Wolff (1999, 2001) questioned whether emotional
intelligence would have an application for explaining differences in groups’
performance. As such, they pioneered the inception of a group emotional competence
(GEC) model based on Goleman’s (1998, 2001) emotional competences theory of EI.
Unlike with individual EI, the emotional intelligence of groups comes from the
patterns of behaviour, i.e. norms2, that emerge as group members work together. In
fact, these norms guide the group’s interaction at multiple levels. At each level of
interaction, i.e. at the individual, group and intergroup levels, Druskat & Wolff (1999,
2001) propose that emotionally competent group norms (ECGNs) can be directed at
the awareness and management of group emotions. Using this structure, the authors
identify nine emotionally competent group norms that represent the set of behaviours
usually found in highly performing teams3.
The present review intends to cover the recent advances in research that are relevant
for the study of group emotional competence as a behavioural approach to emotional
intelligence in team contexts. Two research questions underpin this review: (1) How
to measure emotional intelligence in team contexts? and (2) Does group emotional
competence positively affect team performance? The first question is of special
importance since empirical findings have not yet fully validated the GEC
measurement model. For this reason, we will devote a last part of this review to the
measurement model of GEC and offer a critical account of some recent empirical
findings. This way, the overall objective of this paper is to unveil both theoretical and
empirical gaps in research, which may serve as guidance to future research.
2
Druskat & Wolff (1999) refer to team norms to designate the formal or informal practices
and patterns of behavior that are sustained through time while team members work together.
3 We will use the terms group and team interchangeably in this paper. In this vein, we define
group or team as “a collection of individuals who are interdependent in their tasks, who share
responsibility for outcomes, who see themselves and who are seen by others as an intact
social entity embedded in one or more larger social systems and who manage their
relationships across organizational boundaries” (Cohen & Bailey, 1997: 241).
4
This paper is organized in 6 sections. The next section offers a brief historical
overview of research on emotional intelligence and covers its behavioural approach in
emotional competences; then, section 3 presents and discusses the recent research
endeavour of group emotional intelligence. The behavioural approach to study group
emotional competence is introduced in section 4, while it also illustrates the GEC’s
measurement model. In section 5, specific measurement issues are identified and
critically assessed. Lastly new avenues for future research are addressed in section 6.
Since Ancient Greece and Aristotle’s (384 BC – 322 BC) formal logic, reason was
believed to be superior to emotion, for people could agree on rational arguments but
would easily engage in conflict and disputes through emotional display (Mayer,
Roberts & Barsade, 2008). Emotion and reason were understood as two separate and
opposite processes, wherein the former was seen to interfere with the latter, to the
jeopardy of rationality and effectiveness. In fact, this millenary credo still remains as
of today, present in organizations that frequently pressure their employees to keep
their affective experiences, emotions and feelings off-limits (Seo & Barrett, 2007).
Nevertheless, there also were times when “emotions and feelings were truer than
reason” (Mayer et al., 2008); the romanticism art movement during the eighteenth
century Europe being one among them. By the mid-twentieth century the term
emotional intelligence begins to appear. At first, in an incidental fashion in literary
accounts (Van Ghent, 1953, 1961), then, in occasional scientific references in
psychiatry (Leuner, 1966) and in other studies for personal and social improvement
(Beasley, 1987; Payne, 1986). Later, the 1980s would witness a bourgeoning in
research on the existence of multiple intelligences (Gardner, 1983; Sternberg, 1985)
being paralleled with a renewed research interest on emotions and how these should
interact with cognition (Lazarus, 1982; Izard, 1985; LeDoux, 1989). As such,
although traditional views had described emotion as “a disorganized response, largely
visceral, resulting from the lack of an effective adjustment” (Shaffer et al., 1940:
505), causing the whole loss of brain control (Young, 1936), modern theories had
5
began to reveal that emotion could both interact and direct cognitive activities
(Mandler, 1975; Simon, 1982). Interestingly, research in artificial intelligence also
questioned whether there was value in introducing emotion in computer systems, so
as to enhance and direct their processing (Sloman & Crocher, 1981).
It was then, in the context of an enthusiastic inquiry on both emotion and intelligence,
that Salovey & Mayer (1990) uncovered the concept of emotional intelligence. They
came first to detect there was a void in research: while scattered articles in various
psychology subfields addressed how emotion could be used to prioritize tasks and
solve problems, research in intelligence had been examining the existence of multiple
intelligences, notably in areas such as social behaviour (Gardner, 1983) and, on
occasion, emotions. In connecting these threads, the authors came to terms to devise
their original conceptualization of EI, which they defined as the ability to: “monitor
one’s own and other’s feelings and emotions, to discriminate among them and to use
this information to guide one’s thinking and actions” (Salovey & Mayer, 1990: 189).
Although Goleman’s (1995) book propelled scientific research in EI, it also raised
such public acclaim that the term began to span over a wide variety of interests and
applications, among social researchers, educators and managers. Indeed, after nearly
two decades of research, EI is being referred to and used in remarkably different
ways. By 2009, there is still substantial controversy regarding both the exact
definition of EI and its proper measurement. The definitions are so varied, and the
field is growing so rapidly, that researchers are constantly refining their own previous
definitions.
6
In result, research in EI has diversified into three main approaches: (i) The ability
based model of EI, following the original approach to EI, as in Mayer & Salovey
(1997) and Mayer, Salovey & Caruso (2004), (ii) The emotional competencies model
of EI as a behavioural approach to EI, following Goleman’s (1995, 1998, 2001) and
Goleman, Boyatzis & McKee (2002), and (iii) The trait EI model, based on a mix of
personality traits, attitudes and affective dispositions, following Bar-On (1997, 2000)
and Petrides & Furnham (2000, 2007).
Among these three approaches, we share the views of the two first models of EI. This
is due to our research interests being focused on the understanding of emotionally
intelligent behaviour in team contexts, which are governed by underlying emotional
competences open to development through training and experience. As such, we do
not subscribe to the views advanced in the trait EI model, whereby emotional
intelligence is seen as a quality associated with certain personality traits and not with
others. Moreover, this approach views EI as a relatively innate trait, which does not
conform to the developmental nature of intelligence (Carroll, 1993). Therefore, we
decide to expose further the ability-based model of EI and the behavioural approach
to EI based on emotional competences, which not only are closer to our views but
also, and most importantly, are far more adequate to answer the two research
questions that underlie the present study.
Under the ability based framework that followed the original approach in Salovey &
Mayer (1990), continuous revisions have led to the definition of EI within a four
branch model of abilities (Mayer & Salovey, 1997), including “the abilities to [1]
accurately perceive emotions, [2] to access and generate emotions so as to assist
thought, [3] to understand emotions and emotional knowledge, and [4] to reflectively
regulate emotions so as to promote emotional and intellectual growth” (Mayer,
Salovey and Caruso, 2004: 197).
The distinctive aspect of the ability based model lies in its attempt to establish EI as a
formal type of intelligence, i.e. “the ability to grasp and reason correctly with
abstractions (concepts) and solve problems” (Schmidt and Hunter, 2000: 3). As such,
7
In result, the construct of EI has been designed to meet the three main criteria for
establishing EI as a formal type of intelligence (Mayer, Caruso & Salovey, 1999).
Second, the correlational criterion implies that intelligence correlates with (yet is
different from) other types of intelligence. In this way, cognitive intelligence is
understood as the specialization of general intelligence in the area of cognition in
ways that reflect experience and learning. Similarly, the authors conceptualize
emotional intelligence as the specialization of general intelligence in the domain of
perceiving, understanding and managing emotions. As such, this conceptualization
implies that cognitive intelligence and emotional intelligence are separate though
related sets of abilities, both subsumed under general intelligence. Accordingly,
empirical findings suggest that over 80% of emotional intelligence is unrelated to
other types of intelligence, while the other 20% is correlated to cognitive and verbal
intelligence (Côté & Miners, 2006).
Third, by the developmental criterion, any form of intelligence must have the
potential to improve over time. For this matter, it has been showed that experience
and learning are important to develop emotional intelligence: on one hand the familiar
environment has been proven important in children’s development of emotional
abilities, but on the other hand, adults can also be trained to recognize emotions and
use better strategies to manage emotions in themselves and in others.
Notwithstanding, the ability-based model of EI has been criticized for lacking both
face and predictive validity to explain performance at the workplace (Becker, 2003;
Landy, 2005; Locke, 2005). Such criticism may stem from the fact that in this
approach, EI is assessed through an ability test, the Mayer Salovey Caruso Emotional
8
Intelligence Test (MSCEIT; see Mayer et al., 2003), which examines EI just as IQ
does with cognitive intelligence. However, its central limitation lies in the fact that
there is no exact algebra in emotional intelligence to provide the right answers.
Instead, the MSCEIT must be scored by consensus, usually by means of an expert
committee. Moreover, even if the MSCEIT rightly identifies EI abilities, it does not
show that these abilities are manifested, in interaction with others, through consistent
behaviour. In result, extensive empirical research on the effect of ability based EI on
job performance has so far accumulated mixed and inconclusive results (Côté &
Miners, 2006).
The emotional competencies approach to EI, takes distance from the ability-based
model in so far as it focuses on behaviour. The research strategy in doing so is that it
captures EI where it is most visible, credible and potentially performing at work.
Indeed, research in this area is providing supportive evidence to the idea that
emotional competencies “account for a substantial and important amount of the
variance in predicting or understanding performance” (Boyatzis, 2009).
scores affect group performance (Côté, 2007). Researchers that follow this
methodology to obtain a measure of group emotional intelligence take the implicit
assumption that team members with a higher average of EI scores make a higher EI
score team. This assumption is present both in group EI research grounded on the
ability based model of EI (Feyerhem & Rice, 2002; Day & Carroll, 2004; Offermann
et al., 2004, Fry et al., 2006) and research that uses specific behavioural measures of
EI adapted to group work contexts (Jordan et al, 2002; Jordan & Troth, 2004).
However, group theorists have long been reporting that team members’
characteristics, such as abilities, competences and personality traits, do not necessarily
pass on to the group level, as in a linear combination, to affect team performance.
More than 40 years ago, McGrath (1964) introduced the widely accepted input-
process-outcome (IPO) framework for studying team effectiveness. Figure 1 presents
an adapted version of this framework that shows how a number of inputs influence
team processes, which in turn affect team performance. Inputs are the antecedent
factors that enable and restrain team members’ interactions. They include from team
member characteristics (e.g. competences, personality) to team-level factors (e.g., task
structure, leader influence) and organizational factors (e.g., organizational culture, HR
policies). These factors all combine to design team processes, which describe team
members’ interactions for task fulfilment. Processes are central because they explain
how inputs are transformed into team outcomes. Outcomes include targeted goals
such as performance (e.g. measures on quantity and quality productivity) and by-
products of team members’ interactions, such as affective outcomes (e.g., satisfaction,
commitment).
11
Over the years, the IPO model has been subject to several modifications and
extensions. For instance, Cohen & Bailey (1997) added complexity to the model by
addressing the environmental factors surrounding organizations as drivers of team and
compositional issues. A good example of the environmental effects on teams can be
seen by the latest trend in adopting self-managed teams, in order to lower costs and
improve decision-making (Lawler, 1998), upon the rising competitive challenges.
Other scholars also make note of the inherent multilevel nature of teams, in that
individuals are nested in teams, which in turn are nested in organizations influenced
by environmental factors (Mathieu et al., 2008). In line with these advances in group
theory, Ilgen et al. (2005) proposed a new model whereby the nesting arrangement of
team inputs imply that the environmental and organizational factors affect the nature
of the team, from the leadership styles and patterns of behaviour to task design and
team processes. This model, coined as the input-mediator-outcome-input (IMOI)
model, so as to differentiate it from the standard IPO framework, is depicted in Figure
2, wherein the multilevel influences among team inputs are shown. The bold inward
and dashed outward arrows illustrate that the outer layers (e.g., organizational culture)
influence inner layers (e.g., team leadership style) substantially more than the reverse.
12
In addition, the IMOI model features time, since it plays a critical role in team
functioning that was otherwise not properly captured in classical unidirectional
IPO frameworks.
Interestingly, apart from team processes, there are other mediating factors that
intervene and transform the influence of team inputs on outcomes. Accordingly,
Marks, Mathieu & Zaccaro (2001) had previously took note that team processes
involve members’ actions whereas other mediating mechanisms are better understood
as cognitive, motivational and affective states. Referred to as emergent states, they
include psychological safety and collective affect, among others.
Both the IPO and IMOI frameworks of team effectiveness unveil the crucial
importance of team member interactions in either enhancing or restraining the
individual members’ abilities to contribute to the team’s performance. In result,
because of the unique circumstances and interactive processes of working in a team,
the knowledge, skills and abilities (KSAs) needed for effective performance differ
from those needed by individuals working alone (see Stevens & Campion, 1994).
Hence, the individual abilities involved in emotional intelligence should not be
expected to simply add up to a group’s emotional intelligence. In fact, because
emotions in a group are a by-product of team members’ interactions, we should next
inspect how emotions arise in group contexts.
In effect, Barsade (2002) has found that emotions spread throughout group members
in a rather unconscious and contagious fashion, therefore confirming that “people do
not live in emotional islands, but, rather, that group members experience mood at
work, these moods ripple out and, in the process, influence not only other group
members’ emotions but their group dynamics and individual cognitions, attitudes, and
behaviours as well” (Barsade, 2002: 670). Emotional contagion and the subsequent
dynamics it enacts within a group conforms with previous claims about emotion in a
group context being able to raise a powerful force that overwhelms individual
differences in emotion and creates a collective group character (LeBon, 1977). In
result, there can be a group-level construct of emotion that is greater than the sum of
its individual parts (Barsade & Gibson, 1998).
In the light of these findings, the quest for an emotional intelligence application to
group contexts led Druskat & Wolff (1999, 2001) to pioneer the conceptualization of
a group-level construct of emotional intelligence. Their working theory proposes to
explain how awareness and management of emotion in groups can be used to make
teams work better and achieve higher performance. Building on the EI theories based
on emotional competencies (Goleman, 1998, 2001), those authors introduced the
model of Group Emotional Competence (GEC), which we present in the next section.
Group emotional competence is defined in Druskat & Wolff (1999, 2008) as the
ability of a group to generate a shared set of behavioural norms that guide and
encourage members’ awareness and regulation of emotions in a way that facilitates
the development of group social capital. In turn, group social capital, as defined by
the value added from the group’s structure and the quality of its social relationships
(see Nahapiet & Ghoshal, 1998) – e.g., trust (Mayer et al., 1995), group identity
(Ashforth & Mael, 1989), group efficacy (Lindsley et al., 1995) and networks (Burt,
1997) – is proposed to improve team processes, resulting in process gains that boost
team effectiveness.
15
The model of GEC is, in fact, nested into a broader model – the socio-emotional
model of team effectiveness (Druskat & Wolff, 1999, 2001) – in which two parallel
processes develop over time as team members work together: (1) an emotional
process described by the emergence of GEC norms through which team members
build a group’s social capital and, (2) a process by which social capital fosters team
member engagement in effective processes (e.g., collaboration, open communication)
which improve group performance.
To be sure, the “intelligence” in GEC, unlike in individual EI, is sought from the
patterns of emotionally intelligent behaviour that emerge as group members work
together. Thus, Druskat & Wolff (1999, 2001) propose the existence of emotionally
competent group norms (ECGNs), i.e., formal or informal patterns of behaviour
sustained through time and directed at the awareness and management of emotion in
the group. The authors explicitly refer to awareness and management of emotion since
these are the critical dimensions used in most theories of EI (Salovey & Mayer, 1990;
Goleman, 2001).
Specifically, they identified nine emotionally competent group norms that give an
indication of a group’s emotional intelligence and predict whether a group of
individuals will function as a high performing team. Furthermore, the authors argue
that while in individual EI awareness and regulation of emotion is directed to either
oneself or the others, in groups however, awareness and regulation of emotions could
occur at multiple levels of interaction. Accordingly, these scholars propose that GEC
must evoke emotionally competent norms directed at the awareness and regulation of
group emotion at multiple levels: at the individual-level, for norms directed at
individual team members; at the group-level, for norms directed at the group as a
whole; and at the cross-boundary level, when norms emerge towards others outside
the group (e.g., other teams in the organization, external supervisors, clients).
The nine emotionally competent norms that form a group’s emotional competence are
shown in Table 1.
16
Source: Koman & Wolff, 2008; Stubbs, 2005; Druskat &Wolff, 1999; Hamme, 2003.
Upon the reading of Table 1, one question may naturally arise: How do norms,
especially as emotionally demanding as the ones above, arise in a team? In answering
this question, empirical evidence shows support for a four-stage process that guides
the emergence of norms in team contexts (Feldman, 1984; Bettenhausen &
Murnighan, 1985).
In the first stage, as a team is formed, members base their behaviour and expectations
on their past experience in similar situations. Therefore if EI behaviour is to emerge
as team norms, some members, or at least an influential one, need to possess the
abilities related to EI and a belief that emotionally intelligent behaviour will benefit
the team.
17
The second stage begins as team members initiate interaction. Here, a series of
observations, actions and reflections enhances members’ sense making of situations.
However, according to social comparison theory (Festinger, 1984), ambiguities and
unmet expectations lead team members to seek one another to compare behaviours
and align expectations in regard of the acceptability of their behaviours and beliefs in
the group.
In stage three, members begin challenging the rising expectations and behavioural
patterns while attempting to voice alternative preferences (Bettenhausen &
Murnighan, 1985). In this stage, if EI norms are emerging in the team, they are likely
to be challenged. However, if they are to endure, they must be supported by the
majority of group members. Otherwise, if group EI norms have not emerged, the
group should be intervened to raise their support. Druskat & Wolff (2001, 2008)
report on five drivers that can persuade a team to adopt an emotionally intelligent
behaviour: external team leaders (e.g., superior managers); informal team leaders,
courageous and persuasive members of the group; training and organizational culture.
Among all these drivers, an organizational culture that supports an EI conduct is
likely the most effective and least costly, especially as compared to training.
Finally in stage four, members begin to behave in alignment to the group beliefs and
expectations and abandon the initial strategies and expectations with which they had
entered the team (Bettenhausen & Murnighan, 1985).
Nonetheless, emotionally intelligent group norms may not always emerge as orderly
as this four-stage process suggests. Although for some people EI behaviour is first
nature, for others it may impose a considerable emotional effort. Especially, if teams
face strong opposition by team members to adopt EI behaviour, it is likely that the
problem will not be solely solved with EI training. There may be a problem with
either intrinsic or extrinsic motivations, which in the latter case can be better dealt
with from an organizational perspective. We will come back to this point later when
we discuss research gaps and direction for future research. We now proceed to
consider the GEC operationalization and measurement model.
18
Figure 3 shows the GEC path diagram wherein the central hypothesis is that the GEC
construct affects group performance by means of an indirect effect through social
capital, captured by the product coefficient βGEC*βSC. The positive effect of GEC
norms on group performance has also been found especially beneficial to enhance
innovation outcomes in creative teams, for example in high tech design firms such as
IDEO (Druskat & Wolff, 2001). However, prior to developing GEC norms in a group,
team members must be committed to a minimum level of norms. These are the
fundamental norms directed at sharing goals and information with a sense of
responsibility and organization, without which a group may lack motives to develop
emotional competences.
Moreover, Figure 3 also depicts the measurement model of GEC, whereby each of the
nine ECGNs (e.g., ‘interpersonal understanding’, IU; ‘confronting members who
break norms’, CM; ‘caring behavior’, CB; and so on) are measured through 5 to 8
indicators, each corresponding to a question item in the survey instrument. These
items are assumed to be the reflective indicators4 of each norm’s latent construct.
4 Traditional scale development usually draws on reflective measurement models wherein the
observed indicators are assumed to be caused by a latent construct. As such, reflective
indicators can be thought of as “mirrors” that reflect observable measures of their underlying
construct. On the contrary, in formative measurement models it is assumed that the indicators
19
In turn, the ECGN scales, obtained from the average of their indicators’ scores, are
assumed to be the formative dimensions of the GEC construct. To be precise, the
ECGN scales are the first-order constructs (measured by reflective indicators), which
in turn, taken altogether (i.e., all 9 ECGNs) are the formative or emergent dimensions
of a second-order construct, captured by the GEC composite scale.
Shortly after the nine emotionally competent norms forming the GEC construct were
identified, research efforts to operationalize the GEC model began. Hamme (2003)’s
PhD dissertation developed and tested a survey instrument, designed to measure a
team’s fundamental norms, the nine ECGNs, and social capital so as to test the fit of
the GEC model. Thereafter, though considerable research efforts have accumulated
(Wolff, Druskat, Koman & Messer, 2006; Koman & Wolff, 2008), the GEC model
has not yet been tested in a culture outside the U.S. nor has the full theoretical model
of emotionally competent norms been successfully validated.
For this reason, Druskat has invited Batista-Foguet and our team at the Survey
Research Centre of ESADE to establish a research collaboration within a two-year
research project (2008-10) that tests the GEC model by means of a cross-cultural
comparison with student project teams from the US and Spain. The next subsection
presents the GEC survey research design with reference to this research project.
The study hereby described is designed to: (1) test and improve the psychometric
properties of the survey instrument with US and Spanish samples, (2) test a model of
emotionally competent team leader and team members that predicts the emergence of
emotionally competent team norms, and (3) test the GEC model with data from the
two cultures.
describe different facets of the latent construct. In result, formative indicators are thought to
jointly define the conceptual meaning of the construct (Bisbe, Batista-Foguet & Chenhall,
2007).
20
Because this research project combines cross-cultural samples it delves into the
critical issues of cultural compatibility, which poise potential challenges to validity
and subsequent data analysis (Batista-Foguet, Boyatzis, Guillén & Serlavós, 2008).
This is due to the fact that questionnaires, as opposed to qualitative methods such as
critical incident interviews, are blind to cultural differences. Nonetheless, differences
in meanings and interpretation of the questions will show up in respondents’ answers,
potentially introducing bias and invalidating any subsequent data analysis. As Batista-
Foguet et al. (2008) explain, once using questionnaires, comparisons across countries
demand a frequently overlooked procedure: prior to any data analysis from a cross-
cultural sample, it is critical to evaluate the factorial invariance of the model factor
structure, that is, we must ascertain to what degree the survey items and their
underlying constructs have the same meaning among respondents pooled from
different groups. Failure to do so, implies that the measured differences in the
distribution of the underlying factors between groups may be imputed to differences
in the meanings associated with those factors. In turn, these differences in meanings
may result either from cultural issues or from different interpretations from the
translation of the questionnaire (Batista-Foguet et al., 2008).
Overall, the GEC survey at hand is set to provide empirical support to one of the main
hypothesis to be tested: that group emotional competence, as captured by the survey
instrument, positively affects group performance (specially distinguishing average
from high performing teams). The survey itself is designed to measure the level of
group emotional competence and social capital among teams, which can then be
associated with group performance measures.
Subjects
The GEC survey will be administered to 490 team members in 98 teams (245 team
members in 49 teams per academic year 2008-09 and 2009-10). The teams will be
randomly selected from student teams at the undergraduate and MBA courses of two
business schools: 54 teams from the Whittemore School of Business and Economics
at the University of New Hampshire, in the US, and 44 teams from ESADE Business
School at the Universitat Ramon Llull, in Spain.
21
Unit of analysis
The GEC unit of analysis is set at the group-level. As such, although each team
member answers the survey individually, every item in the survey is meant to collect
each team member’s perceptions of behaviour in his team. Therefore, the GEC survey
enables the aggregation of the team members’ answers to create a composite mean
score of GEC for each team. Hence, the GEC model directly captures group-level
constructs, as opposed to other research that measures group emotional intelligence as
an average composite score of individual EI abilities or behaviours.
Variables
The GEC survey intends to collect 57 reflective indicators that measure the nine
emotionally competent group norms (i.e., the nine ECGNs in Table 1 above). The
survey instrument also includes 20 indicators to measure fundamental team norms,
group social capital (i.e., psychological safety, group identity, group efficacy and
networks), and innovative outcomes.
Provided that the GEC survey is being administered in Spanish to ESADE MBAs and
undergraduates, there was a need to mitigate potential factor invariance arising from
differences between English meanings and their Spanish translation. For this reason,
the original English survey was subject to double-blinded translation, i.e., back and
forth translation (ENESEN) by two official and independent translators. The
resulting translation was, indeed, remarkable: only 2 survey items out of 77 needed
additional attention to make sure the meanings across the two languages exactly
matched.
Respondents rate each survey item on an A/D 7-point Likert scale with seven fixed
reference points: completely disagree; mostly disagree; somewhat disagree; neutral;
somewhat agree; mostly agree; and completely agree. Nevertheless, Revilla, Saris &
Krosnick (forthcoming) assert that A/D scales should provide better results with a
smaller 5-point, instead of a 7-point scale. Accordingly, the A/D scale is generally
understood as too similar to a true/false or a yes/no scale, such that two respondents
may find it difficult to discriminate their answer much further once they agree or
disagree. As such, they may tend to randomize their answers within the “agree” and
“disagree” subscales.
Overall, the respondent is being asked to correctly perform four cognitive steps for
each survey item. This is most of the times too much to expect.
Especially if respondents have a propensity to minimize their efforts, they may tend to
answer a series of items in the same way. Researchers have identified that this
phenomenon, known as response set, increases the correlation between the answers in
batteries of requests. Therefore, the battery format is said to exert a method effect by
“artificially” increasing the correlation between indicators. If not accounted for, this
method effect may bias the test of fit of the model. Particularly, if response set is
stronger among similar items, it may enhance the correlations among reflective
indicators of an underlying construct. In result, the reliabilities of each construct
would become artificially inflated, and so would the test of fit of the model.
Although empirical research on the GEC model is still scarce, the analysis of the
findings reported in Wolff, Druskat, Koman & Messer (2006) and Koman & Wolff
(2008) provides enough material to recognize some of the recurrent issues in testing
the GEC model.
Regarding the measurement model of the GEC construct, we are concerned with both
the validity and reliability of the measures conveyed by the survey instrument.
Construct validity refers to the extent to which a variable accurately captures the
underlying concept it is supposed to measure. That is, validity regards what should be
measured, whereas reliability relates to the quality and accuracy of the measurement
tool. For instance, reliability can be thought of as: (i) the degree of stability of
24
repeated observations over time or under different circumstances (e.g., test-retest); (ii)
inter-judges agreement, or (iii) the internal consistency of the measures within the
same underlying construct (Batista-Foguet & Gallart, 2000; Bisbe, Batista-Foguet &
Chenhall, 2007). Within the present discussion we refer to reliability as the internal
consistency of the survey items, based on the average correlations, as measured by the
Chronbach’s alpha5 (Batista-Foguet, Coenders & Alonso, 2004).
Koman & Wolff (2008) test the GEC model using a sample of 349 military aircrew
and maintenance team members within 81 teams. To ascertain the reliabilities of the
ECGN constructs, they report the Chronbach’s alpha (α). Accordingly, all ECGNs
scales had acceptable reliabilities with their alphas close or above .70 (see Bollen,
1989). For example, among the nine emotionally competent norms, ‘interpersonal
understanding’ (IU) obtained the highest reliability of .83. This means that, while
assuming that the GEC measurement model is valid, and specifically that the IU items
compose a valid indicator, 83% of the variance of the IU construct is explained by its
underlying true factor. Conversely, the norm for ‘confronting members who break
norms’ obtained the smallest reliability with α = .67. However, due to the difficulty of
meeting Chronbach’s alpha requirements for application5, these reliabilities are
possibly misestimated.
In what regards discriminant validity, the reflective indicators within each construct
should be significantly more correlated with each other than with all other indicators
of different constructs (Batista-Foguet et al., 2004). Whenever there is a theoretical
model guiding the research design and subsequent data collection, construct validity
should be tested with a confirmatory factor analysis (CFA) based on the correlation
matrix of the indicators. Koman & Wolff (2008) tested the GEC factor structure by
means of a CFA; however, they found that the theoretical factors at the individual,
group and cross-boundary levels did not produce a good-fitting model. As such, the
5
Although Chronbach’s alpha is reported in most empirical papers, its conditions of
application are often implicitly assumed but seldom verified. This is due to Chronbach’s
alpha highly demanding requirements, which assume indicators are at least tau-equivalent and
parallel. This means that they should have equal loadings, λ, on the underlying factor as well
as equal variances, θ. Whenever these conditions are not observed (in the covariance matrix
of the indicators) Chronbach alpha provides an inferior limit of reliability, reason why it is
often used as a conservative estimate (Batista-Foguet et al., 2004).
25
authors decided to run an exploratory factor analysis (EFA) to improve the GEC
model fit. However, the EFA should only be used in the absence of a theoretical
model behind the data. Essentially, it is of limited use testing the factor structure of a
model without imposing any of the theoretical restrictions that have already guided
the survey design and data collection process. For instance, an EFA does not impose
that the correlations between indicators of different competency constructs – the
ECGNs, assumed to be the formative constructs of the overall GEC construct – should
be small enough, if not zero.
Notwithstanding, Koman and Wolff (2008) ran an EFA on the variance covariance
matrix of the indicators and found that although the GEC norms did not load on the
individual, group and cross-boundary levels, they loaded on two factors, which were
later interpreted as the awareness and the regulation of emotion norms. In result, the
norm for ‘caring behaviour’ was found to load highly on the awareness factor (.58),
whereas it had been theorized and defined as a regulation of emotion norm. Running
an EFA to fit a theoretically complex model, such as the GEC model, is bound to
produce odd results.
Finally, the last step in the estimation process involved a structural equation
modelling (SEM) approach to test the theoretical hypotheses. However, the model
being tested by SEM was not the original GEC model, but instead a rather different
model of GEC structured only on awareness and regulation of emotion norms as
obtained from the previous EFA estimation. Not surprisingly, the SEM approach
resulted in a poor fit. As such, the authors modified the model, yet again, to propose
an even more complex model of GEC. They hypothesize that the awareness of
emotions precedes behaviour and thus also precedes the actual regulation of emotions.
Therefore only the latter should affect group performance in a direct way. In other
words, the authors considered that GEC norms directed at the awareness of emotion
do not affect team performance directly; they only affect performance by means of an
indirect effect through the regulation of emotion norms. The SEM estimation of this
model, despite a slight improvement of fit, still accounted for a poor fitting model.
Wolff et al. (2006) obtained somewhat better results in his study of 109 teams in six
companies. They tested however a previous version of the GEC model with only six
ECGN norms. In particular, the structure of this model was considerably different
26
from the actual GEC measurement model: the ECGNs were not compiled into a
broader GEC scale; instead, each EGCN was let free to directly affect social capital
and through it, indirectly affect group performance. Although the Chi-square test
rejected the model (P-value = .000), this test is based on a too demanding and highly
sensitive statistic, as it follows a central chi-square distribution under the assumptions
of multivariate normality and perfect model fit of the data – the null hypothesis of the
test. A more adequate test is the Root Mean Square Error Approximation (RSMEA),
which is based on a non-central Chi-square that instead of perfect model fit, only
hypothesizes a close fit, making for a wiser and more realistic null hypothesis. As
such, the RMSEA measures the error of approximation of the model to the data, as an
indication of the degree of misspecification of the model. An acceptable model should
have a RMSEA of at most .08 (see Batista-Foguet & Gallart, 2000), however the GEC
model obtained a RMSEA of .89, thus accounting for a poor fit to the data. Yet, it
does denote a better fit than the one achieved in Koman & Wolff (2008).
Finally, social capital, being influenced by all norms, had a positive impact on
performance, by a coefficient of .5. Overall, the GEC model explained as much as
25% of the variance in performance.
27
Group emotional competence research is a brand new research avenue that emerged
within the field of emotional intelligence. In less than a decade of existence it has
advanced considerably and gathered the attention of both EI and group researchers.
The topic makes two legitimate and relevant contributions to the field of emotional
intelligence. First, GEC focuses on groups’ emotional intelligence, a choice justified
by the widespread prevalence of group work structures in most organizations.
Moreover, the increasing adoption of self managed teams, whereby team members are
given much greater discretion to make their own decisions, poises ever greater
challenges to both team members and their external leaders. Copping with such
challenges requires the kind of soft skills, closely related to emotional intelligence,
that empower and influence a team to build trust, collaborate and engage
wholeheartedly at work to achieve better performance (for an account on the unique
facets of effective leadership of self-managed teams, see Druskat & Wheeler, 2003).
Second, because teams’ emotions emerge from the interpersonal interactions among
members (which are rooted in their behaviours), and build up in contagious ways
(Barsade, 2002), there seem to be a logical case to study a behavioural approach to
group emotional intelligence. In the limit, irrespective of what a member may feel, the
way he acts is what counts to trigger other’s emotions, and for these to ripple out
through a team’s emotional dynamics. According to Druskat & Wolff (2001, 2008) a
team’s emotional dynamics is seen to influence its ability to build on a group’s social
capital. The social capital, in turn, enacts greater trust, collaboration and engagement,
much need for greater team performance.
This literature review was designed to provide both a theoretical and empirical
framework to advance future research in group emotional competence. The major
impediments to providing supportive evidence that emotionally competent group
norms do make a positive difference for group performance are related to
measurement issues. According to our review of empirical findings (Koman & Wolff,
2008; Wolff et al., 2006), these indicate that the GEC model has considerable threats
to construct validity. However, we argue that the model may obtain a better fit to the
data when the GEC construct is given a simpler structure.
28
Indeed, through our critical review we may suggest two modifications to enhance
parsimony of the GEC model: (1) reducing the number of emotionally competent
group norms, and (2) removing the formal separation between a team’s levels of
interaction (i.e., individual, group and cross-boundary levels). Also, we note that it
may be more revealing to examine the separate effects of the GEC norms on group
performance (Wolff, 2006), rather than averaging them into a GEC composite
(Koman & Wolff, 2008).
Finally, from the last point, we can also acknowledge that truly effective changes in
team’s behaviour to adopt group EI norms may well be framed in a broader
organizational perspective. Indeed even the trend of research in EI, going from the
individual to the group-level emotional intelligence, seems to indicate there is much
sense in extending EI research towards an organizational-level approach. For this
matter, organization emotional intelligence (OEI) would benefit from a research pool
between the literature on organizational culture and organizational values as well as
research in human resource management. In effect, companies that turn to emotional
intelligence consulting, introduce changes in their organizations via human resource
policies (Goleman, 2001). Hence, we believe that the behavioural approach to
emotional intelligence as in emotional competencies - and in particular its extension
to group-level and organizational-level phenomena, which enable closer association
29
between EI competencies and de facto organizational performing units - may well fall
into the precepts of a “scholarship that works” (Druskat, 2005: 952).
References
Ancona, D. G., Goodman, P. S., Lawrence, B. S., & Tushman, M. L. 2001. Time: A
new research lens. Academy of Management Review, 26(4): 645-663.
Ancona, D. G., Okhuysen, G. A., & Perlow, L. A. 2001. Taking time to integrate
temporal research. Academy of Management Review, 26: 512-529.
Ashforth, B. E., & Mael, F. 1989. Social identity theory and the organization.
Academy of Management Review, 14: 20-39.
Bar-On, R. 2000. Emotional and Social Intelligence: Insights from the Emotional
Quotient Inventory (EQ-i). In R. Bar-On & J. D. A. Parker (Eds.), Handbook of
Emotional Intelligence. San Francisco: Jossey Bass, pp. 363-288.
Barsade, S. G., & Gibson, D. E. 1998. Group emotion: a view from top and bottom. In
D. Gruenfeld (Ed.), Composition, vol. 1. Stamford, CT: JAI Press, pp. 81-102.
Batista-Foguet, J. M., Boyatzis, R. E., Guillén, L., & Serlavós, R. 2008. Assessing
emotional intelligence competencies in two global contexts. In R. J. Emmerling, V. K.
Shanwal & M. K. Mandal (Eds.), Emotional Intelligence: Theoretical and cultural
perspectives. San Francisco: Nova Science Publishers, pp. 89-114.
Bollen, K. A. 1989. Structural equations with latent variables. New York: John
Wiley.
Boyatzis, R. E. 2001. How and why individuals are able to develop emotional
intelligence. In C. Cherniss and D. Goleman (Eds.), The emotionally Intelligent
Workplace: How to select for, measure, and improve emotional intelligence in
individuals, groups, and organizations. Jossey-Bass, San Francisco, pp. 234-253.
Caroll, J. B. 1993. Human cognitive abilities: A survey of factor analytic studies. New
York: Cambridge University Press.
Cohen, S. G., & Bailey, D. E. 1997. What makes teams work: Group effectiveness
research from the shop floor to the executive suite. Journal of Management, 23: 239-
290.
31
Côté, S., & Miners, C. T. H. 2006. Emotional intelligence, cognitive intelligence and
job performance. Administrative Science Quarterly, 51: 1-28.
De Dreu, C. K. W., & Weingart, L. R. 2003. Task versus relationship conflict: Team
performance, and team member satisfaction: A meta-analysis. Journal of Applied
Psychology, 88: 741-749.
Druskat, V. U., & Pescasolido A. T. 2006. The impact of the emergent leader
emotionally competent behavior on team trust, communication, engagement, and
effectiveness. In W. J. Zerbe, N. Ashkanazy, & C. Härtel (Eds.), Research on Emotion
in Organizations, Volume 2: Individual and organizational perspectives on emotion
management and display. Oxford, UK: Elsevier JAI, pp. 25-55
Druskat, V. U., & Wheeler, J. V. 2003. Managing from the boundary: The effective
leadership of self-managing work teams. Academy of Management Journal, 46: 435-
457.
Druskat, V. U., & Wolff, S. B. 1999. The link between emotions and team
effectiveness: how teams engage and build effective task processes. Academy of
Management Best Paper Proceeding, Organizational Behavior Division, 1999.
Druskat, V. U., & Wolff, S. B. 2001. Building the emotional intelligence of groups.
Harvard Business Review, 79 (3): 81-90.
32
Feyerhem, A. E., & Rice, C. L. 2002. Emotional intelligence and team performance:
The good, the bad and the ugly. International Journal of Organizational Analysis, 10:
343-362.
Frye, C. M., Bennett, R., & Caldwell, S. 2006. Team emotional intelligence and team
interpersonal process effectiveness. Mid-American Journal of Business, 21 (1): 49-56.
Gardner, H. 1983. Frames of the mind: The theory of multiple intelligences. New
York: Basic Books.
Goleman, D. 1995. Emotional intelligence: Why it can matter more that IQ. New
York. Bantam Books.
Goleman, D., Boyatzis, R. E., & McKee, A. 2002. Primal Leadership: Realizing the
power of emotional intelligence. Harvard Business School Press, Boston.
Ilgen, D. R., Hollenbeck, J. R., Johnson, M., & Jundt, D. 2005. Teams in
organizations: From input-process-output models to IMOI models. Annual Review of
Psychology, 56: 517-543.
33
Jordan, P. J., Ashkanazy, N. M., Härtel, C. E. J., & Hooper, G. S. 2002. Workgroup
emotional intelligence: scale development and relationship to team process
effectiveness and goal focus. Human Resource Management Review, 12: 195-214.
Jordan, P. J., & Troth, A. C. 2004. Managing emotions during team problem solving:
emotional intelligence and conflict resolution. Human Performance, 17(2): 195-218.
Kelly, J. R., & Barsade, S. G. 2001. Mood and emotions in small groups and work
teams. Organizational Behavior and Human Decision Processes, 86: 99-130.
Koman, E., & Wolff, S. 2007. Emotional intelligence competencies in the team and
team leader. Journal of Management Development, 27 (1): 55-75.
Landy, F. J., & Farr, J. L. 2005. Some historical and scientific issues related to
research in emotional intelligence. Journal of Organization Behavior, 26: 411-424.
LeBon, G. 1977. The Crowd. Middlesex, UK: Penguin; original work published 1895.
Lindsley, D. H., Brass, D. J., & Thomas, J. B. 1995. Efficacy performance spirals: a
multilevel perspective. Academy of Management Review, 20 (3): 645-78.
Marks, M. A., Mathieu, J. E., & Zaccaro, S. J. 2001. A temporally based framework
and taxonomy of team processes. Academy of Management Review, 26(3): 356-376.
Mathieu, J., Maynard, M. T., Rapp, T., & Gilson, L. Team Effectiveness 1997-2007:
A review of recent advancements and a glimpse into the future. Journal of
Management, 34 (3): 410-476.
Mayer, J. D., Caruso, D. R., & Salovey, P. 1999. Emotional Intelligence meets
traditional standards for an intelligence. Intelligence, 27: 267-298.
Mayer, J. D., Roberts, R. D., & Barsade, S. G. 2008. Annual Review of Psychology,
59: 507-536.
Mayer, J. D., Salovey, P., Caruso, D. R., & Sitarenios, G. 2003. Measuring emotional
intelligence with the MSCEIT V2.0. Emotion, 3: 97-105.
Mayer, J. D., Salovey, P., & Caruso, D. R. 2004. Emotional Intelligence: Theory,
findings and implications. Psychological Inquiry, 15 (3): 197-215.
Morgeson, F. P., & Hoffmann, D. A. 1999. The structure and function of collective
constructs: implications for multilevel research and theory development. Academy of
Management Review, 24 (2): 249-65.
Nahapiet, J., & Ghoshal, S. 1998. Social capital, intellectual capital, and the
organizational advantage. Academy of Management Review, 24 (2): 249-65.
Offermann, L. R., Bailey, J. R., Vasilopoulos, N. L., Seal, C., & Sass, M. 2004. The
relative contribution of emotional competence and cognitive ability to individual and
team performance. Human Performance, 17 (2): 219-243.
Revilla, M., Saris, W. E., & Krosnick, J. Forthcoming. Comparison between Liss
panel (web) and ESS data (face to face). Survey Research Methods.
Schaffer, L. F., Gilmer, B., & Schoen, M. 1940. Psychology. New York: Harper &
Brothers, pp. xii-521.
Seo, M.-G., & Barrett, L. 2007. Academy of Management Journal, 50 (4): 923-940.
Sloman, A., & Crocher, M. 1981. Why robots will have emotions. In A. Drinan
(Eds.), Proceedings of the Seventh International Joint Conference on Artificial
Intelligence, Vol. 1. Vancouver.
Sternberg, R. J. 1985. Human Intelligence: the model is the message. Science, 230:
1111-1118.
Stevens, M. J., & Campion, M. A. 1999. Staffing work teams: Development and
validation of a selection test for teamwork settings. Journal of Management, 25: 207-
227.
Stubbs, E. C. 2005. Emotional intelligence competencies in the team and team leader:
a multi-level examination of the impact of emotional intelligence on group
performance. Unpublished dissertation. Case Western Reserve University, Cleveland,
OH.
Van Ghent, D. 1953. The English Novel: Form and function. New York: Harper &
Row.
Wolff, S. B, Druskat, V. U., Koman, E. S., & Messer, T. E. 2006. The link between
group emotional competence and group effectiveness. In V. U. Druskat & F. Sala &
G. Mount (Eds.), Linking emotional intelligence and performance at work: current
research evidence with individuals and groups. Mahway, NJ: LEA.
Young, P. T. 1936. Motivation of Behavior. New York: John Wiley & Sons.