Content uploaded by Rebecca Grossman
Author content
All content in this area was uploaded by Rebecca Grossman on Jul 21, 2015
Content may be subject to copyright.
Objective: The aim of this study was to review literature
relevant to cohesion measurement, explore developing mea-
surement approaches, and provide theoretical and practical
recommendations for optimizing cohesion measurement.
Background: Cohesion is essential for team effective-
ness and performance, leading researchers to focus atten-
tion on understanding how to enhance it. However, cohe-
sion is inconsistently defined and measured, making it difficult
to compare findings across studies and limiting the ability to
advance science and practice.
Method: We reviewed empirical research through which
we uncovered specific information about cohesion’s concep-
tualization, measurement, and relationships with performance,
culminating in a set of current trends from which we provide
suggestions and possible solutions to guide future efforts and
help the field converge toward greater consistency.
Results: Cohesion demonstrates more significant relation-
ships with performance when conceptualized using social and
task (but not other) dimensions and when analyses are per-
formed at the team level. Cohesion is inherently temporal, yet
researchers rarely measure cohesion at multiple points during
the life of a team. Finally, cohesion matters in large, dynamic col-
lectives, complicating measurement. However, innovative and
unobtrusive methodologies are being used, which we highlight.
Conclusion: Practitioners and researchers are encouraged
to define cohesion with task and social subdimensions and to
measure with behavioral and attitudinal operationalizations. Indi-
vidual and team-oriented items are recommended, though team-
level analyses are most effective. Innovative/unobtrusive methods
should be further researched to enable cohesion measurement
longitudinally and in large, dynamic collectives.
Application: By applying our findings and conclusions,
researchers and practitioners will be more likely to find
consistent, reliable, and significant cohesion-to-performance
relationships.
Keywords: organizational behavior/design, organizational
psychology, macroergonomics and the environment, group
processes, social processes, team dynamics, teams and groups,
team collaboration, team communication
IntroductIon
Teams are critical for success in today’s orga-
nizations (Kozlowski & Ilgen, 2006), regardless
of whether teams operate in an office setting
(Simons & Peterson, 2000) or in isolated/con-
fined/extreme environments (Bishop, 2004).
Teams are advantageous to individuals in many
ways. They pool diverse knowledge and skills,
allowing for convergent and divergent thinking,
the building blocks of creativity and knowledge
generation (Hoegl & Parboteeah, 2007). They
also provide a source of backup and assistance
for overworked or underskilled team mem-
bers, and can be a source of positive affect and
increased morale (Salas, Sims, & Burke, 2005).
To be sure, teams offer many benefits, but in
large part these benefits will be realized only
in cohesive teams. Cohesion—the shared bond/
attraction that drives team members to stay
together and to want to work together (Casey-
Campbell & Martens, 2009)—is essential for
teams (e.g., Beal, Cohen, Burke, & McLendon,
2003; Chiocchio & Essiembre, 2009). Indi-
viduals who feel no sense of cohesion with their
team (whether due to distrust, dislike, disinter-
est, or a host of other reasons) are less motivated
and less likely to participate in the “teaming”
behaviors that enable the many positive effects
of teams.
Given the importance of cohesion to team
and organizational performance, accurate mea-
surement of this construct is essential; however,
several issues complicate effective measure-
ment. First, cohesion is an umbrella term used in
many domains, leading to myriad definitions
(e.g., Carron, Widmeyer, & Brawley, 1985;
Shaw, 1981) and complicating the research pro-
cess. Second, because cohesion has team and
individual components, operationalizing, mea-
suring, and analyzing cohesion at different lev-
els is often difficult. Third, cohesion is thought
to emerge over time (Bartone & Adler, 1999),
meaning that researchers should incorporate
578267HFSXXX10.1177/0018720815578267<italic>Human Factors</italic>Measuring Team Cohesion
Address correspondence to Eduardo Salas, Institute for
Simulation and Training, University of Central Florida,
3100 Technology Parkway, Suite 132, Orlando, FL 32826,
USA; e-mail: esalas@ist.ucf.edu.
Author(s) Note: This work is not subject to U.S. copyright
restrictions.
Measuring Team Cohesion: Observations from the Science
Eduardo Salas, University of Central Florida, Orlando, Florida, Rebecca Grossman,
Hofstra University, Hempstead, New York, Ashley M. Hughes, University of Central Florida,
Orlando, Florida, and Chris W. Coultas, Leadership Worth Following, LLC, Irving, Texas
HUMAN FACTORS
Vol. 57, No. 3, May 2015, pp. 365 –374
DOI: 10.1177/0018720815578267
At the Forefront of HF/E
at AXINN LIBRARY*E on June 8, 2015hfs.sagepub.comDownloaded from
366 May 2015 - Human Factors
temporal elements into cohesion research,
though longitudinal studies inherently introduce
logistical difficulties. Finally, there is a growing
interest in measuring cohesion in complex,
dynamic collectives (e.g., Thayer, Gregory,
Grossman, & Burke, 2014), specifically in space
exploration (see Salas et al., in press). Measur-
ing cohesion in these settings presents practical
and logistical challenges that must be addressed.
In light of these four issues, we discuss possible
paths to solving these problems by reviewing
empirical research and leveraging theory.
Although the first two issues have been pres-
ent in the literature for quite some time, the dis-
parate approaches to analyzing cohesion have
not yet been reconciled; we thus use the findings
from our review to provide initial insights about
which approaches should be adopted by both
researchers and practitioners going forward.
Additionally, because the second two issues are
more novel, developing themes in the literature,
we place a greater emphasis on discussing them,
allowing for a stronger contribution to the litera-
ture. Finally, we do not presume to solve all
problems associated with cohesion measure-
ment; rather, our review serves to move the state
of cohesion research forward.
Method
We conducted a literature search to uncover
trends in the cohesion literature using the terms
cohesi* and team within peer-reviewed arti-
cles in EBSCOhost databases (i.e., PsycINFO
[1887–2013], Business Source Premier [1905–
2013]). We included articles if (a) they were
empirical, (b) cohesion was included in the title,
and (c) they explored the relationship between
cohesion and performance. Though cohesion
demonstrates relationships to other constructs,
we limit our review to articles exploring the
cohesion–performance link for parsimony and
because performance is an outcome of particular
importance. We supplemented the search find-
ings with studies from the two most recent cohe-
sion meta-analyses (Beal et al., 2003; Chiocchio
& Essiembre, 2009), excluding unpublished
studies. This process yielded 210 articles, 70
with information sufficient for coding (note that
not all articles are referenced in this work due to
space constraints; a list of references is available
in the online supplementary materials). Of these
70, various team types were used (n = 20 sports
teams, n = 22 student samples, n = 20 adult
working samples, n = 6 military samples, and
n = 2 from other domains).
We qualitatively coded articles to extract infor-
mation pertaining to cohesion’s conceptualiza-
tion, data collection/analysis methodologies, and
whether or not studies established relationships
between cohesion and performance. We calcu-
lated the frequency with which specific measure-
ment practices occurred as well as the ratio of sig-
nificant to nonsignificant cohesion–performance
findings. This calculation was done with the intent
that frequency indicates common measurement
practice whereas significance suggests efficacious
practice. Significance percentages refer to the
ratio of measured relationships between cohesion
and performance that were statistically significant
(p < .05). For example, if cohesion was conceptu-
alized 750 times unidimensionally compared to
250 multidimensionally, that result would indicate
common practice is to conceptualize cohesion
unidimensionally. However, if 50% of unidi-
mensional conceptualizations were significant
whereas 80% were significant when defining
cohesion multidimensionally, this result would
suggest that multidimensional conceptualizations
are more efficacious. To further clarify our cod-
ing/analysis strategy, see Table 1.
defInItIonal Issues
Clearly, cohesion measurement is inherently
complex. Some scholars define cohesion uni-
dimensionally (e.g., members’ attraction to the
group or resistance to leaving; Seashore, 1954),
whereas others view it as a multidimensional
construct (e.g., sum of forces acting on members
to remain in a group; Festinger, 1950; for a list
of the five most common dimensions and their
definitions, see Table 2). Our review revealed
that the majority of measured relationships
(42%) defined cohesion multidimensionally,
though some (16%) defined it unidimension-
ally (for almost 50% of measured relationships,
a definition was not provided). The idea that
cohesion is conceptualized in a number of ways
is certainly not new. As noted earlier, cohe-
sion has been a topic of interest for decades,
but this long and varied history has resulted in
a vast, often ambiguous literature that offers
little insight about which approaches are most
at AXINN LIBRARY*E on June 8, 2015hfs.sagepub.comDownloaded from
Measuring TeaM Cohesion
367
effective. Indeed, divergent definitions, dimen-
sions, and operationalizations have yielded an
array (over 35!) of cohesion measures, not only
obfuscating potential cross-study comparisons
that would yield stronger research conclusions,
but also leaving practitioners, who need to mea-
sure cohesion in order to assess and enhance it,
at a loss for insight about exactly how to do so.
Specifically, due to logistical constraints in
applied settings, it is often necessary to limit atten-
tion to only those dimensions of cohesion that are
most likely to relate to performance improve-
ments; to maximize utility, such dimensions must
be (a) identified, so that team interventions can be
designed to enhance them, and (b) effectively
measured, so that they can be assessed both before
and after interventions are implemented in order
to evaluate their effectiveness and to ensure that
interventions are linked to cohesion dimensions of
interest and not others deemed less critical. Thus,
our goal, in part, is to identify patterns in the litera-
ture that indicate which approaches to defining
and measuring cohesion are most effective in
terms of their ability to detect relationships with
performance.
With this goal in mind, our review provides
suggestions and possible paths for solutions
based on the efficacy of measurement strategies
TABLE 1: Descriptions and Examples of Measurement Practices/Concepts Coded
Code Description Examples
Conceptualization/definition How was cohesion defined?
Did the definition capture a
single dimension or multiple
dimensions of cohesion?
• Unidimensional: Individual
members’ attraction to the
group task
• Multidimensional: The total
field of forces that act on
members to remain in the group
Dimensionality of
measurement
How was cohesion measured?
Did the items capture a
single dimension or multiple
dimensions of cohesion? Which
dimension(s) did the items
capture?
• Task cohesion: “Our team is
united in trying to reach its
goals for performance.”
• Social cohesion: “Our team
would like to spend time
together outside of work.”
Operationalization/focus of
measurement
Did the cohesion measure
capture attitudes, behaviors,
or a mix of both?
• Attitudes: “The members of our
team felt proud to be a part of
the team.”
• Behaviors: “Members of our
team do not stick together
outside of work.”
Level of measurement Did the cohesion measure
capture the individual level,
the team level, or a mix of
both?
• Individual level: “Some of my
best friends are on this team.”
• Team level: “People work well
together as a team.”
Level of analysis How was cohesion analyzed:
at the individual level or the
team level?
• Individual level: A mean across
all participants was calculated
and utilized in analysis.
• Team level: Team member
responses were aggregated
to the team level (e.g., mean);
then a mean across all teams
was calculated and utilized in
analysis.
at AXINN LIBRARY*E on June 8, 2015hfs.sagepub.comDownloaded from
368 May 2015 - Human Factors
present within common cohesion measurement
practices. First, we confirm that researchers
should adopt a multidimensional definition of
cohesion (of the articles that clearly defined
cohesion, multidimensional conceptualizations
found significant cohesion–performance rela-
tionships more frequently [69%] compared to
unidimensional conceptualizations [57%]). This
finding is largely consistent with prior work on
the tripartite view of cohesion, whereby task
cohesion, social cohesion, and group pride have
each shown significant links with performance
(Beal et al., 2003; Mullen & Copper, 1994). Our
review, however, did not reveal a consistent
group pride–performance link (it was also less
frequently studied). Accordingly, we advocate
leveraging group pride dimensions when feasi-
ble, but prioritizing social and task dimensions,
particularly when adapting to logistical con-
straints. Second, cohesion seems to be neither
purely attitudinal (e.g., “Members of this team
like each other”) nor purely behavioral (e.g.,
“Members of this team spend time with each
other off the job”): We found that measures
including both operationalizations more consis-
tently uncovered the cohesion–performance link
(see Table 3). Thus, we suggest that the most
effective cohesion measures are those that assess
the social and task dimensions while spanning
attitudinal and behavioral foci.
MultIlevel Issues
Another critical issue to consider when mea-
suring and understanding cohesion is the role
of multilevel assessment. Particularly, it has
long been unclear whether cohesion should
be defined primarily as an individual, team,
or multilevel construct (Casey-Campbell &
Martens, 2009). Indeed, although cohesion was
frequently defined as a team variable (37% of
measured relationships), authors of some stud-
ies conceptualized cohesion as a multilevel vari-
able (14%) and, further complicating the issue,
40% failed to clarify the level of conceptual-
ization (fortunately, there was agreement that
cohesion should not be considered as strictly an
individual-level variable [in only 1% of stud-
ies was cohesion examined exclusively at the
individual level]; see Table 4). As noted, this
ambiguity presents an issue for both researchers
and practitioners interested in cohesion: Before
we can understand how to enhance cohesion or
diagnose and improve it in practice, we need
to be able to measure it effectively. Our find-
ings provide some insight about how to do so.
Despite cohesion being more frequently defined
as a team (as opposed to a multilevel) variable,
multilevel measures performed better (74% of
measured relationships were significant) than
those with strictly team (65% of measured rela-
tionships were significant) or individual (54%
TABLE 2: List of Cohesion Subdimensions and Their Definitions
Subdimension Definition Citation
Task An attraction or bonding between group
members that is based on a shared commitment
to achieving the group’s goals and objectives
Carron, Widmeyer, & Brawley
(1985); Festinger (1950)
Social A closeness and attraction within the group that is
based on social relationships within the group
Carron et al. (1985); Seashore
(1954)
Belongingness The degree to which members of a group are
attracted to each other
Shaw (1981)
Group pride The extent to which group members exhibit
liking for the status or the ideologies that the
group supports or represents, or the shared
importance of being a member of the group
Beal, Cohen, Burke, &
McLendon (2003)
Morale Individuals’ high degree of loyalty to fellow
group members and their willingness to endure
frustration for the group
Cartwright & Zander (1960)
at AXINN LIBRARY*E on June 8, 2015hfs.sagepub.comDownloaded from
Measuring TeaM Cohesion
369
of measured relationships were significant) oper-
ationalizations. Additionally, analytic strategies
seem to favor team-level cohesion; aggregat-
ing cohesion to the team level more frequently
yielded significant findings (76% of measured
relationships) than dyadic (50%) or individual-
level analyses (53%). We therefore urge future
researchers and practitioners to adopt a multi-
level view of cohesion—from research design,
to measure development, to statistical analyses.
That is, cohesion should be measured at both the
individual and the team level. This approach will
enable greater flexibility while also allowing for
analyses to be run at the team level, where cohe-
sion seems to operate the strongest.
teMporal Issues
Cohesion is a relational “emergent state”
(Marks, Mathieu, & Zaccaro, 2001), meaning
it emerges over time as teammates interact.
Despite this inherently temporal component,
in only a few studies reviewed was cohesion
measured two or more times, likely due to
the logistical constraints placed on researchers
that often prevent longitudinal measurement.
Nonetheless, limited understanding of cohesion
TABLE 3: Frequency and Significance of Measured Relationships for Each Dimension and Foci of
Cohesion Measures
Measure Characteristic
% Frequency of Measured
Relationships
% Signicance of Measured
Relationships
Dimension (n= 116)
Task 33 76
Social 53 69
Belongingness 10 58
Group Pride 2 50
Morale 2 0
Focus (n= 217)
Attitudinal 45 53
Behavioral 33 60
Mixed 22 74
Note. n refers to the number of times the specific cohesion–performance relationship was examined across all 70
studies included in the review.
TABLE 4: Frequency and Significance of Measured Relationships for Each Level of Measurement and
Analysis
Measure Characteristic
% Frequency of Measured
Relationships
% Signicance of Measured
Relationships
Measure/item level (n= 236)
Individual 48 54
Team 37 65
Mixed 14 74
Level of analysis (n= 202)
Individual 59 53
Dyadic 2 50
Team 39 76
Note. n refers to the number of times the specific cohesion–performance relationship was examined across all 70
studies included in the review.
at AXINN LIBRARY*E on June 8, 2015hfs.sagepub.comDownloaded from
370 May 2015 - Human Factors
emergence complicates its measurement and our
understanding of its temporal nature. Empirical
evidence on cohesion has demonstrated that
cohesion within teams varies as a function of
time (Carless & De Paola, 2000). This idea is
also a matter of common sense; when two teams
are characterized by the same quantitative indi-
ces of cohesion (means, dispersion, skewness,
etc.) but with wildly different tenures (e.g., 1
hr vs. 10 years), should these teams be consid-
ered to have equivalent cohesion? Should these
teams even be asked the same cohesion-relevant
questions? We do not presume that simply
emphasizing the importance of time in cohesion
measurement will result in more longitudinal
studies. Accordingly, we offer a few thoughts
for time-conscious and practical measurement.
Team development theories (e.g., Kozlowski,
Gully, Nason, & Smith, 1999) and research on
emergence over time (e.g., Coultas, Driskell,
Burke, & Salas, 2014) offer insights into longi-
tudinal cohesion measurement. Kozlowski
et al.’s (1999) process model of team develop-
ment posits that teams develop through phases
consisting of (a) team formation (i.e., members
familiarizing with each other at surface levels),
(b) task compilation (i.e., members identifying
with/mastering tasks), and (c) role compilation
(i.e., members learning/negotiating details of
intrateam relations). These phases have clear
implications for cohesion emergence. First,
group-level agreement is less likely during team
formation (Mullen & Copper, 1994), meaning
that group-level cohesion–performance relation-
ships will also be unlikely. Additionally, social
cohesion indices are less reliable than task cohe-
sion during early team life (Siebold, 2006).
To address this issue, we encourage research-
ers to begin developing the construct of “swift
cohesion” (Coultas et al., 2014). Research on
“swift trust” (Meyerson, Weick, & Kramer, 1996)
and “swift psychological safety” (Dufresne, 2007)
suggests that when constructs rapidly emerge,
they may be conceptually different from their
more gradual counterparts (see also transactive
memory systems; Kanawattanachai & Yoo,
2007). If swift cohesion emerges under different
conditions or has different effects than “tradi-
tional” cohesion, expanding on this concept may
enable more reliable indices of cohesion at the
early phases of team development (Quintane,
Pattison, Robins, & Mol, 2013). Furthermore,
empirical evidence suggests that cohesion mea-
sured later in the team life cycle, as opposed to
during its formation or early phases of develop-
ment, demonstrates stronger links to improved
performance (Bradley, Baur, Banford, &
Postlethwaite, 2013; Siebold, 2006).
Second, because sustainable/reliable task cohe-
sion typically emerges first (see preceding para-
graph), measures of task cohesion should be more
reliable at the group level early on, whereas social
cohesion indices may take longer to be effective
predictors of performance. Accordingly, we sug-
gest that assessing the team’s developmental phase
will enable researchers to determine which ele-
ments of cohesion are most likely to be reliable,
especially if measuring every aspect of cohesion
or measuring it longitudinally is infeasible.
logIstIcal and practIcal Issues
That cohesion is often assessed in the lab,
or in small-scale teams, does not mean that
cohesion is unimportant elsewhere. For exam-
ple, there is interest in assessing cohesion in
large organizations (e.g., the military) and fast-
paced, dynamic teams (e.g., surgical teams);
in these settings, self-report measures may be
cumbersome and/or practically impossible to
administer. Unfortunately, simply asking fewer
questions is often not a solution. In part, this
idea is due to the fact that longer, more reliable
measures are typically more effective. Indeed,
we found that longer measures were likelier to
predict (for measures with five or fewer items,
57% of measured relationships were significant;
six to 10 items, 74%; ≥11 items, 86%); a similar
pattern was found with more reliable measures
(when alpha ≥.70, 78% of measured relation-
ships were significant; alpha < .70, 41%).
Obviously, concise, reliable, and content-valid
measures are ideal; but in some complex set-
tings, it may be infeasible or even impossible
to administer self-report measures of cohesion.
Given these logistical constraints, unobtrusive/
indirect measures of cohesion are essential.
Indeed, we identified several articles that lever-
aged innovative techniques that may facilitate
the collection and analysis of cohesion data in
complex team settings (see Table 5).
at AXINN LIBRARY*E on June 8, 2015hfs.sagepub.comDownloaded from
Measuring TeaM Cohesion
371
Big data (i.e., data sets that proliferate with
automated updated information), sociometric
radio frequency identification (RFID) badges,
and certain physiological metrics (e.g., lexical
analysis, eye gaze, electroencephalogram read-
ings) have shown promise for capturing team-
work processes and states (e.g., Gonzales, Han-
cock, & Pennebaker, 2010). We encourage
researchers to leverage these techniques (see
Table 5) more frequently—especially when
studying cohesion in large, fast-paced, dynamic,
or high-risk team settings. Before we accom-
plish the ultimate goal of having reliable and
valid unobtrusive measures of cohesion readily
available, several things must happen. First,
valid unobtrusive indicators of cohesion must be
developed. To do so, researchers must supplement
unobtrusive collection and analytic methodolo-
gies with traditional approaches to assess con-
struct validity (see Bobko, 2001; Hughes et al.,
2014). The advantage of developing multiple
collection methodologies is it enables research-
ers to select more effective methods, adaptive to
different contexts. For example, in fast-paced
and/or high-risk, dynamic teams (e.g., surgery,
long-duration space flight), sociometric RFID
badges may be especially useful, because they
do not draw resources away from task perfor-
mance and are not subject to recall bias. Current
methods of leveraging RFID to assess team per-
formance and teamwork are nascent (Parlak,
Sarcevic, Marsic, & Burd, 2012); however,
advances in validating RFID metrics with team-
work frameworks is furthering the validity of
their use (Rosen, Dietz, Yang, Priebe, & Prono-
vost, in press). An alternative method for devel-
opment and validation of unobtrusive metrics is
to develop indicators based on a framework that
is grounded in the science of assessment to help
ensure construct validity (Hughes et al., 2014;
Rosen et al., in press). External observations—
though not necessarily innovative—may be a
TABLE 5: Indirect Methods for Analyzing and Measuring Cohesion
Methodology Description Citation
Big data Automated data collection of cohesion
indices via e-mail, text messaging, social
network exchanges, search records, etc.;
indices may be derived from interaction
patterns or the nature/content of
interactions
Brzozowski (2009)
Sociometric badges Sociometric electronic tag that tracks team
member locations; can be used to infer
cohesion based on temporal proximity
and interaction duration and frequency
Olguín-Olguín &
Pentland (2010)
Physiological metrics Physiological indices of teamwork including
cohesion, such as eye gaze, bodily
gestures, and brainwave data, may be
analyzed with algorithms adapted from
linguistic style matching
Gonzales, Hancock, &
Pennebaker (2010)
Content analysis Extracting trends and indicators of cohesion
indices from existing or unobtrusively
collected intrateam textual or verbal
exchanges
Hung & Gatica-Perez
(2010)
External observer Raters with substantial knowledge of the
team’s functioning can estimate a team’s
cohesion, circumventing the need to
measure perceptions of every team
member
Chang, Jia, Takeuchi, &
Cai (2014)
at AXINN LIBRARY*E on June 8, 2015hfs.sagepub.comDownloaded from
372 May 2015 - Human Factors
helpful, unobtrusive way to assess cohesion,
provided external leaders have sufficient expo-
sure to the team and are able to provide unbiased
or trained ratings. Big data may be especially
relevant in large, complex systems (e.g., military
brigades, multiteam systems), where a wealth of
intracollective interactions are available. The
strategies provided and discussed help to pave a
way in an emerging field of real-time assess-
ment of dynamic team emergence, which has
been cited as “possibly the most difficult [mea-
surement] principle to implement” (Rosen et al.,
2011, p. 119).
conclusIons
Though long considered a key contributor to
team success, cohesion is perhaps more impor-
tant than ever. As organizations continue to seek
competitive advantage, teams are increasingly
looked to in the hopes of facilitating knowl-
edge, morale, and creativity. Researchers have
long lamented the inconsistency plaguing the
cohesion literature (Casey-Campell & Mar-
tens, 2009), much of it revolving around dis-
agreements about cohesion’s conceptualization
and measurement. Practitioners interested in
monitoring and improving cohesion face a vast,
confusing literature with myriad measurement
options. Our review offers potential solutions to
problems that cohesion researchers frequently
face.
First, we reiterate that cohesion is indeed a
multidimensional construct and clarify that task
and social cohesion should be prioritized when
measuring cohesion, but that more research is
needed on group pride. Second, we illustrate that
cohesion is multilevel, though it seems to oper-
ate most consistently at the team level. Research-
ers should focus cohesion research at the team
level; however, they should also be aware there
are individual-level components of cohesion
that may warrant different analytic techniques.
Third, we emphasize that cohesion is inherently
temporal but that researchers rarely model cohe-
sion longitudinally. We offer a few suggestions
for incorporating time into cohesion research
(e.g., developing the swift cohesion construct),
but ultimately, more research is needed. Fourth,
we acknowledge that cohesion is important in
“messy” team settings, too, and we emphasize a
few methods, such as the use of RFID badges
and social network analysis of big data sources,
for circumventing problems inherent to these
settings.
And although significant progress has been
made over the years, more robust, precise, theo-
retical-driven, practical, and innovative mea-
sures are needed—a tall order that will take time
and resources to continue exploring and testing,
access to expert team-participants, and a man-
date that makes assessment of team cohesion a
priority. The science of team effectiveness has
this challenge ahead, but it is well suited to suc-
ceed. Time will tell.
acknowledgMents
This work was supported by NASA Grant
NNX09AK48G awarded to the University of Central
Florida. The authors thank the following researchers
for their contributions and insights on an earlier ver-
sion of this manuscript: Tripp Driskell, Jim Driskell,
John Mathieu.
key poInts
•Cohesion is critical for team effectiveness, but
inconsistencies in how it has been defined and
measured limit the ability to advance science and
practice.
•On the basis of a qualitative review of a subset of
the cohesion literature and our leveraging of the-
ory, we present the following recommendations
for improving the measurement of cohesion:
•A multidimensional approach to defining and
measuring cohesion should be adopted with
priority given to the social and task cohesion
dimensions; group pride should be examined
when logistical constraints allow for it.
•A multilevel approach to measuring cohesion
should be adopted, whereby cohesion is assessed
at both the individual and the team level. This
approach allows for greater flexibility in analy-
sis, but priority should be given to conducting
analyses at the team level of analysis.
•The developing literature on nonobtrusive
measurement approaches (e.g., sociomet-
ric badges) should be adopted in particularly
complex or high-stakes settings (e.g., military
contexts); multiple measurement approaches
should be utilized initially to ensure the con-
struct validity of nonobtrusive techniques.
at AXINN LIBRARY*E on June 8, 2015hfs.sagepub.comDownloaded from
Measuring TeaM Cohesion
373
•The team’s developmental phase should be
considered when making measurement deci-
sions—certain dimensions of cohesion may be
more or less salient depending on the life span
of the team. Additional research is needed to
explore the notion of “swift cohesion” in newly
formed teams, which may represent a slightly
different construct than cohesion as it applies
to teams that are further developed.
suppleMentary MaterIal
The online reference list is available at http://hfs
.sagepub.com/supplemental.
references
Bartone, P. T., & Adler, A. B. (1999). Cohesion over time in a
peacekeeping medical task force. Military Psychology, 11,
85–107.
Beal, D. J., Cohen, R. R., Burke, M. J., & McLendon, C. L. (2003).
Cohesion and performance in groups: A meta-analytic clarifi-
cation of construct relations. Journal of Applied Psychology,
88, 989–1004.
Bishop, S. L. (2004). Evaluating teams in extreme environments:
From issues to answers. Aviation, Space, and Environmental
Medicine, 75(Suppl. 1), C14–C21.
Bobko, P. (2001). Correlation and regression: Applications for
industrial organizational psychology and management. Thou-
sand Oaks, CA: Sage.
Bradley, B. H., Baur, J. E., Banford, C. G., & Postlethwaite, B. E.
(2013). Team players and collective performance: How agree-
ableness affects team performance over time. Small Group
Research, 44, 680–711.
Brzozowski, M. J. (2009, May). WaterCooler: Exploring an orga-
nization through enterprise social media. In Proceedings of
the ACM 2009 International Conference on Supporting Group
Work (pp. 219–228). New York, NY: ACM.
Carless, S. A., & De Paola, C. (2000). The measurement of cohe-
sion in work teams. Small Group Research, 31, 71–88.
Carron, A. V., Widmeyer, W. N., & Brawley, L. R. (1985). The
development of an instrument to assess cohesion in sport
teams: The Group Environment Questionnaire. Journal of
Sport Psychology, 7, 244–266.
Cartwright, D., & Zander, A. (1960). Group cohesiveness: Intro-
duction. In D. Cartwright & A. Zander (Eds.), Group dynam-
ics: Research and theory (2nd ed., pp. 69–94). New York, NY:
Harper Row.
Casey-Campbell, M., & Martens, M. L. (2009). Sticking it
all together: A critical assessment of the group cohesion–
performance literature. International Journal of Management
Reviews, 11, 223–246.
Chang, S., Jia, L., Takeuchi, R., & Cai, Y. (2014). Do high-
commitment work systems affect creativity? A multilevel com-
binational approach to employee creativity. Journal of Applied
Psychology, 99, 665–680.
Chiocchio, F., & Essiembre, H. (2009). Cohesion and performance:
A meta-analytic review of disparities between project teams,
production teams, and service teams. Small Group Research,
40, 382–420.
Coultas, C. W., Driskell, T., Burke, C. S., & Salas, E. (2014). A
conceptual review of emergent state measurement: Current
problems, future solutions. Small Group Research, 45, 671–
703.
Dufresne, R. (2007, August). Ad hoc team learning: The proposed
effects of swift psychological safety and team leadership. Paper
presented at the annual meeting of the Academy of Manage-
ment, Philadelphia, PA. doi:10.5465/AMBPP.2007.26523723
Festinger, L. (1950). Informal social communication. Psychologi-
cal Review, 57, 271–282.
Gonzales, A. L., Hancock, J. T., & Pennebaker, J. W. (2010). Lan-
guage style matching as a predictor of social dynamics in small
groups. Communication Research, 37, 3–19.
Hoegl, M., & Parboteeah, K. P. (2007). Creativity in innovative
projects: How teamwork matters. Journal of Engineering and
Technology Management, 24, 148–166.
Hughes, A. M., Volante, W. G., Stowers, K., Leyva, K., Oglesby, J.
M., Bisbey, T., . . . Vidulich, M. A. (2014). Cognition and phys-
iological response: Towards a model of validated physiological
measurement. In Proceedings of the Human Factors and Ergo-
nomics Society 58th Annual Meeting (pp. 1009–1013). Santa
Monica, CA: Human Factors and Ergonomics Society.
Hung, H., & Gatica-Perez, D. (2010). Estimating cohesion in small
groups using audio-visual nonverbal behavior. IEEE Transac-
tions on Multimedia, 12, 563–575.
Kanawattanachai, P., & Yoo, Y. (2007). The impact of knowledge
coordination on virtual team performance over time. MIS
Quarterly, 31, 783–808.
Kozlowski, S. W., Gully, S. M., Nason, E. R., & Smith, E. M.
(1999). Developing adaptive teams: A theory of compilation
and performance across levels and time. In D. R. Ilgen & E.
D. Pulakos (Eds.), The changing nature of work performance:
Implications for staffing, personnel actions, and development
(pp. 240–292). San Francisco, CA: Jossey-Bass.
Kozlowski, S. W., & Ilgen, D. R. (2006). Enhancing the effective-
ness of work groups and teams. Psychological Science in the
Public Interest, 7(3), 77–124.
Marks, M. A., Mathieu, J. E., & Zaccaro, S. J. (2001). A temporally
based framework and taxonomy of team processes. Academy
of Management Review, 26, 356–376.
Meyerson, D., Weick, K. E., & Kramer, R. M. (1996). Swift trust
and temporary groups. In R. M. Kramer & T. R. Tyler (Eds.),
Trust in organizations: Frontiers of theory and research
(pp. 166–196). Thousand Oaks, CA: Sage.
Mullen, B., & Copper, C. (1994). The relation between cohesive-
ness and productivity: An integration. Psychological Bulletin,
115 , 210–227.
Olguín-Olguín, D., & Pentland, A. (2010). Sensor-based organisa-
tional design and engineering. International Journal of Organ-
isational Design and Engineering, 1, 69–97.
Parlak, S., Sarcevic, A., Marsic, I., & Burd, R. S. (2012). Intro-
ducing RFID technology in dynamic and time-critical medical
settings: Requirements and challenges. Journal of Biomedical
Informatics, 45, 958–974.
Quintane, E., Pattison, P. E., Robins, G. L., & Mol, J. M. (2013).
Short- and long-term stability in organizational networks:
Temporal structures of project teams. Social Networks, 35,
528–540.
Rosen, M. A., Bedwell, W. L., Wildman, J. L., Fritzsche, B. A.,
Salas, E., & Burke, C. S. (2011). Managing adaptive perfor-
mance in teams: Guiding principles and behavioral markers
for measurement. Human Resource Management Review, 21,
107–122.
at AXINN LIBRARY*E on June 8, 2015hfs.sagepub.comDownloaded from
374 May 2015 - Human Factors
Rosen, M. A., Dietz, A. S., Yang, T., Priebe, C. E., & Pronovost, P.
J. (in press). An integrative framework for sensor-based mea-
surement of teamwork in healthcare. Journal of the American
Medical Informatics Association.
Salas, E., Sims, D. E., & Burke, C. S. (2005). Is there a “Big Five”
in teamwork? Small Group Research, 36, 555–599.
Salas, E., Tannenbaum, S. I., Kozlowski, S. W. J., Miller, C.,
Mathieu, J. E., & Vessey, W. B. (in press). Teams in space
exploration: A new frontier for the science of team effective-
ness. Current Directions in Psychological Science.
Seashore, S. E. (1954). Group cohesiveness in the industrial work
group. Ann Arbor: University of Michigan, Survey Research
Center Institute for Social Research.
Shaw, M. E. (1981). Group dynamics: The psychology of small
group behavior. New York, NY: McGraw-Hill.
Siebold, G. L. (2006). Military group cohesion. Military Life: The
Psychology of Serving in Peace and Combat, 1, 185–201.
Simons, T. L., & Peterson, R. S. (2000). Task conflict and rela-
tionship conflict in top management teams: The pivotal role of
intragroup trust. Journal of Applied Psychology, 85, 102–111.
Thayer, A. L, Gregory, M. E., Grossman, R., & Burke, C. S. (2014,
May). Toward development and validation of an unobtrusive
unit cohesion measure. Paper presented at the 29th Annual
Conference for the Society for Industrial and Organizational
Psychology, Honolulu, HI.
Eduardo Salas will be a professor of psychology at
Rice University as of July 1, 2015. Previously, he
was trustee chair and Pegasus Professor of Psychol-
ogy at the University of Central Florida. His research
focuses on teamwork, team training, simulation-
based training, patient safety, safety culture, learning
methodologies, and performance assessment.
Rebecca Grossman is an assistant professor of indus-
trial/organizational (I/O) psychology at Hofstra Uni-
versity. She earned her PhD in I/O at the University
of Central Florida in 2014, where she conducted
research at the Institute for Simulation and Training.
Her research focuses on understanding team cohe-
sion and related processes and emergent states
within both traditional and complex (e.g., multicul-
tural, virtual, distributed, multiteam systems) set-
tings and on improving such variables through the
use of training and development interventions.
Ashley M. Hughes is a doctoral candidate in the
applied experimental human factors psychology
program at the University of Central Florida
(UCF). She received her MS in modeling and
simulation from UCF in 2013 and performs
research at the Institute for Simulation and Train-
ing, where her research interests are teamwork,
team training, and simulation-based training pri-
marily in medical settings.
Chris W. Coultas graduated from the University of
Central Florida (UCF) in 2014 with a PhD in indus-
trial/organizational psychology. While at UCF, he
worked at the Institute for Simulation and Training,
under Dr. Eduardo Salas, where he conducted
research on teams, training, culture, leadership, lead-
ership development, and coaching. He has published
works in Small Groups Research and Consulting
Psychology Journal as well as book chapters on
training and leadership and has presented at numer-
ous conferences.
Date received: August 4, 2014
Date accepted: February 19, 2015
at AXINN LIBRARY*E on June 8, 2015hfs.sagepub.comDownloaded from
