See discussions, stats, and author profiles for this publication at: [Link]

net/publication/11547039

Team cohesion and team success in sport

Article in Journal of Sports Sciences · March 2002

DOI: 10.1080/026404102317200828 · Source: PubMed

CITATIONS READS

422 31,028

3 authors, including:

Albert Carron Mark Eys

Western University Wilfrid Laurier University
82 PUBLICATIONS 7,834 CITATIONS 171 PUBLICATIONS 6,222 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Mark Eys on 07 April 2015.

The user has requested enhancement of the downloaded file.

This article was downloaded by: [Canadian Research Knowledge Network]
On: 3 June 2011
Access details: Access Details: [subscription number 932223628]
Publisher Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-
41 Mortimer Street, London W1T 3JH, UK

Journal of Sports Sciences

Publication details, including instructions for authors and subscription information:
[Link]

Team cohesion and team success in sport

Albert V. Carron; Steven R. Bray; Mark A. Eys

Online publication date: 09 December 2010

To cite this Article Carron, Albert V. , Bray, Steven R. and Eys, Mark A.(2002) 'Team cohesion and team success in sport',
Journal of Sports Sciences, 20: 2, 119 — 126
To link to this Article: DOI: 10.1080/026404102317200828
URL: [Link]

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: [Link]

This article may be used for research, teaching and private study purposes. Any substantial or
systematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply or
distribution in any form to anyone is expressly forbidden.

The publisher does not give any warranty express or implied or make any representation that the contents
will be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses
should be independently verified with primary sources. The publisher shall not be liable for any loss,
actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly
or indirectly in connection with or arising out of the use of this material.
 Journal of Sports Sciences, 2002, 20, 119± 126

Team cohesion and team success in sport

ALBERT V. CARRON,1* STEVEN R. BRAY2 and MARK A. EYS1
1
School of Kinesiology, University of Western Ontario, London, Ontario N6A 3K7 and 2Department of Kinesiology,
University of Lethbridge, Lethbridge, Alberta T1K 3M4, Canada

Accepted 10 September 2001

The main aim of this study was to examine the relationship between task cohesiveness and team success in elite
teams using composite team estimates of cohesion. A secondary aim was to determine statistically the consist-
ency (i.e. `groupness’ ) present in team members’ perceptions of cohesion. Elite university basketball teams
(n = 18) and club soccer teams (n = 9) were assessed for cohesiveness and winning percentages. Measures were
Downloaded By: [Canadian Research Knowledge Network] At: 19:41 3 June 2011

recorded towards the end of each team’ s competitive season. Our results indicate that cohesiveness is a shared
perception, thereby providing statistical support for the use of composite team scores. Further analyses indicated
a strong relationship between cohesion and success (r = 0.55± 0.67). Further research using multi-level statistical
techniques is recommended.

Keywords: group and individual analysis, task cohesion, task performance.

Introduction associated with reduced team success in co-acting

sports (e.g. bowling).
The de® nition of cohesion ± `a dynamic process that Powerful advantages of a meta-analysis are the
is re¯ ected in the tendency of a group to stick together opportunities to summarize statistically a large body of
and remain united in the pursuit of its instrumental research and to examine the role of potential moderator
objectives and/or for the satisfaction of member aþ ective variables if suý cient data are available. Thus, in 1994,
needs’ (Carron et al., 1998, p. 213) ± implicitly conveys when Mullen and Copper carried out their meta-
the generally held assumption about team cohesion analysis of 49 studies from various subdisciplines in
and team success; that is, greater team cohesiveness psychology (e.g. industrial, sport, military, social), their
is assumed to be related to greater team success. conclusions appeared to provide unequivocal answers
Historically, this assumption has been challenged. for sport psychology. They reported that the relation-
For example, in 1969, Lenk provided data from two ship between cohesion and team success is positive,
champion German rowing eights to `refute the strict albeit small, and the task interaction requirement (i.e.
general validity of a thesis that seems to have been taken interacting vs co-acting sports) does not serve as a
for granted . . . [namely that] only small groups, which moderator variable. They also reported that `real groups
are low in con¯ ict, or highly integrated can produce exhibit signi® cantly stronger [cohesion± success] eþ ects
especially high performances’ (p. 393). Lenk’ s study, as than arti® cial groups, and sport teams exhibit even
well as other studies undertaken inside and outside of stronger eþ ects than nonsport real groups’ (Mullen and
sport psychology over a 20 year period, led others (e.g. Copper, 1994, p. 224). In fact, Mullen and Copper
Landers and L schen, 1974; Carron and Chelladurai, reported that the strongest relationship between co-
1981) to propose that type of sport is a moderator hesion and group success is present in sport teams,
variable in the relationship between cohesion and followed by military groups and then non-military
success. That is, they hypothesized that cohesion is groups.
associated with enhanced team success in interacting Although Mullen and Copper’ s (1994) meta-analysis
sports (e.g. volleyball), but either has no eþ ect or is did provide some valuable insights, we would argue that
it did not provide de® nitive answers about the relation-
* Author to whom all correspondence should be addressed. e-mail: ship between cohesion and success in sport teams. The
bcarron@[Link] fundamental reason for our assertion is that most of the

Journal of Sports Sciences ISSN 0264-0414 print/ISSN 1466-447X online Ó 2002 Taylor & Francis Ltd
[Link]
 120 Carron et al.

sport studies included in Mullen and Copper’ s behavior ± adherence for example ± then the individual’ s
meta-analysis (e.g. Carron and Ball, 1978; Landers cognition about cohesion is the critical consideration . . .
et al., 1982; Williams and Hacker, 1982) used the Sport Conversely, if the focus was on a group behavior ± group
Cohesiveness Questionnaire (SCQ; Martens et al., performance for example ± then the average level of co-
1972). The SCQ is no longer used for group dynamics hesion in the group would be the appropriate unit of
analysis.
research in sport, mainly because it was not developed
from a strong conceptual base and its psychometric
properties have never been fully established (Gill, Thus, using the individual as the unit of analysis does
1977; Carron et al., 1998). answer the question: Are individual athlete perceptions
Carron et al. (1985) developed the Group Environ- of cohesion related to personal or team productivity or
ment Questionnaire (GEQ), which is based on a con- success? However, such a strategy does not oþ er insight
ceptual model in which cohesion is considered to be a into the relationship between composite team cohesion
result of four primary constructs: Individual Attractions and team success.
to the Group-Task, which re¯ ects a member’ s feelings There have also been studies in which the Group
about his or her personal involvement with the group’ s Environment Questionnaire has been used as the
task; Individual Attractions to the Group-Social, a operational de® nition for cohesion and the group or
Downloaded By: [Canadian Research Knowledge Network] At: 19:41 3 June 2011

member’ s feelings about his or her personal social inter- team has been used as the unit of analysis (i.e. a com-
actions with the group; Group Integration-Task, a posite measure has been used to represent team co-
member’ s perceptions of the similarity and uni® cation hesion and team winning percentage has been used to
of the group as a whole around its tasks and objectives; represent success). Nonetheless, that body of research
and Group Integration-Social, a member’ s perception also oþ ers limited insight into the relationship between
of the similarity and uni® cation of the group as a social team cohesion and team success for two reasons. First,
unit. in some studies (e.g. Shanghi and Carron, 1987; Slater
Slater and Sewell (1994, p. 424) suggested that `the and Sewell, 1994), the sample sizes used were relatively
GEQ holds great potential for furthering the establish- small (i.e. less than eight teams). Thus, questions
ment of a more complete picture of team cohesion associated with statistical power arise. Secondly, in
in sport’ . However, they also noted that, `thus far, other studies (e.g. Widmeyer et al., 1990; Grieve et al.,
the GEQ appears to have been used in relatively 2000), many teams were examined, but these teams
few published studies of the cohesion± performance were created or developed for the study ± they were
relationship’ (p. 424). not intact (real) sport teams. As Mullen and Copper
Subsequent to the observations of Slater and Sewell (1994) showed, `the cohesiveness± performance eþ ect is
(1994), various authors have examined the relation- even more robust in the real world among real groups’
ship between cohesion and success in sport using the (p. 224). Thus, the strength of the relationship between
Group Environment Questionnaire as an operational cohesion and success is probably underestimated in
de® nition for cohesiveness. Unfortunately, however, laboratory studies.
that body of research suþ ers from shortcomings in The main aim of this study was to examine the rela-
terms of answering the fundamental question: Is team tionship between team perceptions of task cohesiveness
cohesion positively associated with team success? One and team success in elite basketball and soccer teams.
shortcoming is that the perceptions of cohesion of Team success was operationally de® ned as the team’ s
individual athletes (on few winning and losing teams) win± loss percentage. Team perceptions of cohesion
have often been the focus of analysis (e.g. Matheson were operationally de® ned using the two task cohesion
et al., 1997; Kozub and Button, 2000). In short, measures from the Group Environment Questionnaire ±
statistical power for the analyses was derived by testing Individual Attractions to the Group-Task and Group
relatively large samples of individuals, not teams. Using Integration-Task. Widmeyer et al. (1993) have sug-
individual athletes as the unit of analysis is not in and gested that the task-related dimensions of cohesion
of itself inappropriate. This point was emphasized by should be most directly related to team success. How-
Carron and Spink (1995, pp. 91± 92): ever, because the Group Integration-Task dimension
taps members’ beliefs about the team’ s integrated pur-
An ongoing debate in the group dynamics literature, suit of its task-relevant goals and objectives, Widmeyer
which can be traced to the 1920s (Allport, 1924), is
et al. suggested that `the cohesion dimension most
whether the individual or the group should be the unit
of analysis. The answer, of course, is that there is no
closely linked [conceptually] to performance outcome is
de® nitive answer ± it depends on the nature of the Group Integration-Task’ (p. 686). Consistent with this
question. Cohesion is a cognition that exists in the minds theorizing, we hypothesized that the Group Integration-
of individual group members. If the question of interest is Task and Individual Attractions to the Group-Task
centered on the relationship of cohesiveness to individual dimensions would be positively related to team success,
 Cohesion and team success 121

but the former would show a stronger relationship basketball and nine soccer teams. The participants
between cohesion and success. had considerable competitive experience at both the
A secondary aim of the study was to determine the developmental and, in most cases, elite intercollegiate
consistency among team members’ perceptions of co- standards. For example, several athletes had experience
hesion before amalgamating responses to create a on provincial and national junior teams. The athletes
composite `team cohesion’ score. Analysis issues have ranged in age from 15 to 30 years (20.9 ± 3.3 years;
attracted considerable attention in the general literature mean ± s) and had been members of their respective
on group dynamics (e.g. Kenny and La Voie, 1985; teams for 1± 10 years (2.4 ± 1.5 years). A heterogeneous
Rousseau and House, 1994; Moritz and Watson, 1998), sample from two sports was used to increase the power
as well as in that on cohesion speci® cally (e.g. Carron for the analysis and to increase the generalizability of the
et al., 1998; Carron and Brawley, 2000; Paskevich et al., ® ndings.
2001). Moritz and Watson (1998), for example, sug-
gested that a failure to consider individuals and groups
Measures
in group research is associated with three potential
biases: (a) possible over-generalization, whereby it is Cohesion. The Group Environment Questionnaire
assumed that relationships at one level are similar to (GEQ; Carron et al., 1985) was used to assess cohesion.
those for a seemingly similar concept at another level; This is a self-report questionnaire that contains 18
Downloaded By: [Canadian Research Knowledge Network] At: 19:41 3 June 2011

(b) possible underestimation of group in¯ uence when items. Four aspects of cohesion are assessed: Individual
individuals are the unit of analysis as well as possible Attractions to the Group-Task (4 items), Individual
underestimation of individual in¯ uence when groups Attractions to the Group-Social (5 items), Group
are the unit of analysis; and (c) possible rei® cation of Integration-Task (5 items) and Group Integration-
group structure. Social (4 items). Responses are provided on a 9-point
In the present study, we selected the team as the Likert scale anchored at the extremes by `strongly dis-
unit of analysis for two reasons. First, the dependent agree’ (1) and `strongly agree’ (9). Thus, higher scores
variable ± team success ± is the product of an integrated re¯ ect stronger perceptions of cohesiveness.
aggregation of all team members’ eþ orts. Also, team Only the task cohesion dimensions were included.
success is a measure that has no individual variability; it The decision to focus on the relationship between task
can only be analysed for the team. Secondly, although cohesion and team success only, and not on that
measures of task cohesion are obtained from individual between social cohesion and team success, was based on
members, these perceptions develop within an inter- two related considerations: participant burden and con-
dependent environment. As a consequence, measures of ceptual rationale. The data used in the present study
cohesion obtained from members within a group may were obtained in research in which many variables were
be similarly biased (Zaccaro et al., 1995). For example, incorporated in the questionnaire. We were concerned
Group Integration-Task refers to team members’ shared that, if the questionnaire was too long, the athletes
beliefs about the group’ s approach to its performance. would either refuse to complete it or would complete it
Thus, while all members of a team may not be entirely in a cursory fashion. Thus, we decided to include only
like-minded in these beliefs, their perceptions of Group the task cohesion scales of the GEQ, since it has been
Integration-Task should show a high degree of consist- suggested that they have a stronger link with team
ency and non-independence re¯ ective of this sharing success (i.e. Widmeyer et al., 1993).
among members. Consistent with this theorizing, we The internal consistency of each scale was computed
hypothesized that the most appropriate unit of analysis for the data obtained in the present study. Both the
for the research question would be the team. In short, Group Integration-Task (a = 0.75) and the Individual
we believed that acceptable consistency about Group Attractions to the Group-Task (a = 0.68) scales
Integration-Task and Individual Attractions to the possessed acceptable reliability.
Group-Task cohesion would be present among the
members of each team, thereby lending statistical sup- Team success. Team success was operationally de® ned
port for the aggregation of individual responses to the as each team’ s total winning percentage for the games
team. played in their regular competitive schedule; play-oþ
games were excluded. A percentage score was cal-
culated by dividing the number of points obtained by
Methods the maximum number of points possible. Basketball
designates two points for a win and no points for a loss;
Participants ties are not possible in basketball. Soccer in Canada
The participants were 294 Canadian (154 females, designates two points for a win, one point for a tie and
140 males) intercollegiate and club athletes from 18 no points for a loss.
 122 Carron et al.

Procedure Group-Task and Group Integration-Task were con-

sistent with group eþ ects, lending support to our
Basketball and soccer team coaches were contacted by
hypothesis; for Group Integration-Task, r = 0.21 and
one of the researchers, who explained that he was con-
g2 = 0.27; for Individual Attractions to the Group-Task,
ducting a study examining athletes’ feelings about their
r = 0.17 and g2 = 0.24. Myers (1972) suggested using
team and team success. After permission was obtained,
a liberal criterion (i.e. a < 0.25) to test for group
we met with the athletes and coaches to explain the
phenomena rather than the conventional one (a < 0.05).
study and to request their participation. Individual
The use of a liberal criterion was not necessary, how-
participation was voluntary and informed consent was
ever, as the eþ ects observed were highly signi® cant
obtained from all participants. When individuals were
(P < 0.001). The eta-squared statistics provide com-
under 18 years of age, parental consent was also
plementary evidence of group perceptions, as each
obtained.
exceeded the criterion of 0.20 for aggregation that has
The Group Integration-Task and Individual
been used in previous research (e.g. Jehn and Shah,
Attractions to the Group-Task scales of the Group
1997; Dirks, 2000).
Environment Questionnaire were completed under the
supervision of the researchers approximately 2 weeks
Downloaded By: [Canadian Research Knowledge Network] At: 19:41 3 June 2011

before the end of each team’ s regular competitive Descriptive statistics

season. Questionnaires were administered at a con-
Table 1 provides a summary of the descriptive statistics
venient team meeting or practice that was neither
for the soccer and basketball samples. Before looking for
immediately before nor immediately after a com-
an overall relationship between cohesion and success,
petition, thus avoiding competition-speci® c biases in
preliminary analyses were performed to determine if
the participants’ responses. Although all members of
diþ erences were present in the samples. As indicated
each team completed the questionnaire at a group
previously, both male and female basketball and soccer
meeting, the importance of independent responses was
teams were recruited. Thus, a 2 (basketball, soccer) ´
stressed at the time of administration. The participants
2 (female, male) multivariate analysis of variance
were required to complete the questionnaire on their
(MANOVA) was computed with Individual Attractions
own and without conversing with their team-mates.
to the Group-Task, Group Integration-Task and success
as the dependent variables. Neither sex (Wilks’ l:
F3,21 = 0.46, P > 0.05) nor the interaction of sex and
Results sport (Wilks’ l: F3,21 = 0.32, P > 0.05) was statistically
signi® cant. However, a signi® cant diþ erence was found
Aggregation of data
for sport (Wilks’ l: F3,21 = 3.25, P < 0.04). Post-hoc
Team members’ GEQ scores were aggregated to pro- analyses found no diþ erences in success or Group
vide team scores for Individual Attractions to the Integration-Task (P > 0.05), but Individual Attractions
Group-Task and Group Integration-Task. As indicated to the Group-Task was signi® cantly greater in the soccer
above, this procedure is consistent with the unit of sample (F1,23 = 6.41, P < 0.02). Thus, we decided
analysis identi® ed in the research question ± that is, the to combine the male and female teams within sports
team. To determine whether aggregation was empiric- and to examine the cohesion± success relationship
ally justi® ed, intraclass correlation coeý cients (Kenny independently for basketball and soccer and, sub-
and La Voie, 1985) and eta-squared (Georgopolous, sequently, in the total sample of teams by transforming
1986) were calculated. The intraclass correlation the data to eþ ect sizes and computing an average eþ ect
coeý cient and associated eta-squared (g2) provide size.
evidence of the consistency of variance in responses The statistical techniques used to compute eþ ect
among members of groups nested within a larger sam- sizes were those outlined by Hedges (1981, 1982) and
ple in relation to the scores of non-group members. Hedges and Olkin (1985) and summarized by Thomas
It should be noted that the intraclass correlation coef- and French (1986). That is, the correlations were ® rst
® cient (r) can range from -1 to +1; however, the typical converted to Cohen’ s (1969, 1992) eþ ect size; then,
range of scores falls between zero and one (Kenny and because eþ ect sizes are positively biased in small
La Voie, 1985). If r = 1, then all members of the same samples, a correction factor was used on each eþ ect size
group have identical scores; if r = 0, then the scores before subsequent analyses. Also, each eþ ect size was
of the members belonging to one group are no more then weighted by the reciprocal of its variance before
similar to one another than to those of members of summation. An overall weighted mean estimate and an
other groups. estimate of the variance of each eþ ect size were obtained
The intraclass correlation coeý cients and eta- using the formula provided by Hedges and Olkin
squared statistics for both Individual Attractions to the (1985). The designation eþ ect size is used here to
 Cohesion and team success 123

Table 1. Descriptive statistics (mean ± s)

Individual

Group Attractions to Team

Group Integration-Task the Group-Task performance

Basketball (n = 18) 6.05 ± 0.93 6.11 ± 1.04 46.51 ± 27.32

Soccer (n = 9) 6.33 ± 0.76 7.04 ± 0.69 52.38 ± 25.54

Note: Cohesion was assessed on a 9-point scale with low scores representing less co-
hesiveness. Team performance represents a winning percentage out of 100%.

Table 2. The relationship between cohesion and performance

Cohesion±
performance Eþ ect
Downloaded By: [Canadian Research Knowledge Network] At: 19:41 3 June 2011

Group Cohesion measure relationship size

Basketball (n = 18) Group Integration-Task 0.60** 1.42

Individual Attractions to the Group-Task 0.62** 1.49
Soccer (n = 9) Group Integration-Task 0.55 1.16
Individual Attractions to the Group-Task 0.74* 1.94
All teams (n = 27) Group Integration-Task 0.57** 1.29
Individual Attractions to the Group-Task 0.67** 1.71

* P < 0.05, ** P < 0.01.

represent eþ ect sizes that underwent all of the above Total sample. The summation of eþ ect sizes showed
transformations. that, within the total sample, the relationship between
cohesion and success was very strong. That is, for
Group Integration-Task, an overall eþ ect size of 1.29
Cohesion and team success
was obtained, which represents a correlation coeý cient
Type of sport. Table 2 provides a summary of the corre- of 0.57. For cohesion operationally de® ned as Indi-
lation analysis of the relationship between cohesion and vidual Attractions to the Group-Task, the overall eþ ect
success in the basketball and soccer samples. For co- size was even larger at 1.71, which represents a corre-
hesion as manifested by Individual Attractions to the lation coeý cient of 0.67. Our ® nding that the relation-
Group-Task, a statistically signi® cant relationship was ship between cohesion and success was greater for
observed for both basketball (r = 0.62, P < 0.01) and cohesiveness manifested as Individual Attractions to the
soccer (r = 0.74, P < 0.05). In both instances, the size Group-Task than as Group Integration-Task was con-
of the eþ ect was very large: 1.49 and 1.94 for basket- trary to our hypothesis.
ball and soccer, respectively. The designations `small’ ,
`medium’ and `large’ were applied by Cohen (1992)
to represent eþ ect sizes of 0.20, 0.50 and 0.80, Discussion
respectively.
Analysis of the relationship between cohesion and The main aim of this study was to examine the relation-
success using Group Integration-Task as the operational ship between task cohesiveness and team success in
measure of cohesiveness showed a similar pattern. That elite teams using aggregate (i.e. composite team)
is, for basketball, a signi® cant relationship (r = 0.60, measures of cohesion, with team success represented
P < 0.01) of very large magnitude (eþ ect size = 1.42) by season game winning percentages. A secondary
was obtained. For soccer, the relationship was not statis- aim, addressed before undertaking the main part of
tically signi® cant (r = 0.55, P > 0.05). However, this the study, was to determine statistically the consistency
non-signi® cant correlation was undoubtedly a function (i.e. `groupness’ ) of team members’ perceptions of
of statistical power, given that the eþ ect size was again cohesion.
very large (i.e. eþ ect size = 1.16). One of the most important aspects of our study is
 124 Carron et al.

that it has provided evidence of a very strong relation- such as team success. Our results show that perceptions
ship between cohesion and success in sport teams. of team task cohesiveness are relatively consistent
Interestingly, in their meta-analysis, Mullen and Copper among members of the same team. That is, team
(1994) reported that a small to moderate relationship members perceive their team’ s task unity similarly,
is present in sport. One reason for the diþ erence in the which provides support for a conceptualization of
magnitude of this relationship in our study compared `cohesion-as-shared-beliefs’ (Carron et al., 1998;
with that in the meta-analysis of Mullen and Copper Carron and Brawley, 2000; Paskevich et al., 2001).
could be the operational de® nition of cohesion used. Paskevich (1995, Study 2), using intercollegiate volley-
We focused on task cohesion only, whereas Mullen and ball teams, was the ® rst to provide statistical support
Copper’s ® ndings were based on studies that used both for group cohesion as shared individual beliefs. Thus,
task and social cohesion. A second reason is that we there is increasing support for the measurement of
operationally de® ned task cohesion using the Group individual perceptions to obtain a valid estimate of a
Environment Questionnaire (Carron et al., 1985). group property such as cohesion. In short, aggregating
Mullen and Copper statistically summarized studies the scores of individual team members to produce a
that overwhelmingly used psychometrically un- single measure of team cohesion does not run the risk
developed measures of cohesion. A third possible reason of combining `apples and oranges’ .
Downloaded By: [Canadian Research Knowledge Network] At: 19:41 3 June 2011

might be related to our use of composite team scores to It is also worth issuing a note of caution and inspir-
represent cohesion. The Sport Cohesiveness Question- ation for future researchers of the aggregation of indi-
naire, a psychometrically undeveloped measure of co- vidual cohesion scores for the relationship between
hesion, was used by Carron and Ball (1978), Landers cohesion and success. Although statistical indicators of
et al. (1982) and Williams and Hacker (1982) to examine consistency (e.g. the intraclass correlation coeý cient)
the relationship between composite team cohesion and may show a high degree of sharing, because of
team success. One of the items in the Sport Cohesive- the nature of common personal experiences it is likely
ness Questionnaire is teamwork, a construct that is very that some variability in perceptions of cohesion
similar to task cohesion. In these three studies, using exists within groups. For example, the intraclass
composite team scores, the relationship between team- correlation coeý cients for Group Integration-Task
work (cohesion) and team success when both constructs and Individual Attraction to the Group-Task in the
were assessed at the end of the season (i.e. a time frame present study showed modest consistency, but were well
similar to that used in our study) was generally high, short of scores of 1.00 that would indicate identical
ranging from r = 0.69 (Landers et al., 1982) to r = 0.88 scores among all members within each team. The
(Carron and Ball, 1978) and r = 0.90 (Williams and question arises as to the potential value of information
Hacker, 1982). oþ ered by individual members in their estimates of
Widmeyer et al. (1993) suggested that, because of the cohesion that is lost due to aggregation. As Hoyle and
conceptual nature of the construct, Group Integration- Crawford (1994, p. 465) pointed out: `A single score
Task is likely to have a stronger association with team that represents the thoughts, feelings, or behavioral
performance than Individual Attractions to the Group- tendencies of the group results in a loss of information
Task. Our results do not support this hypothesis; the about the character of the group’ . In the present study,
relationship between Individual Attractions to the the nature of both the research question and the
Group-Task and team success was slightly stronger dependent variable required group analysis. However,
than that between Group Integration-Task and team future research should try to combine individual
success. Although our ® ndings do not support the and group analysis (e.g. multi-level modelling) to
hypothesis of Widmeyer et al. (1993), it is important to explore more fully the relationship between cohesion
note that very large eþ ect sizes were found for each rela- and success.
tionship. Clearly, both the Group Integration-Task and Our search of the literature identi® ed several issues
Individual Attractions to the Group-Task dimensions of that need to be addressed relating to the relationship
cohesion are strongly associated with team success. between cohesion and success in sport. One of these
However, future research should investigate further the pertains to the role that cohesion might play in the
independent relationships between the task cohesion success of individual sport teams. To date, research has
dimensions and team success. examined the relationship between cohesion and suc-
Cohesion is a construct that can be examined in rela- cess in swimming (Everett et al., 1992), golf (Williams
tion to both individual and group outcomes (Hoyle and and Widmeyer, 1991), track events (Berardinis et al.,
Crawford, 1994). However, the results of research using 1983) and bowling (Landers and L schen, 1974).
the Group Environment Questionnaire indicate the Interestingly, the study of Landers and L schen
strongest relationship appears to be that between group (1974) is one of the few to report a negative relationship
estimates of task cohesion and group-related constructs between cohesion and team success. Thus, the
 Cohesion and team success 125

manifestation of the relationship between cohesion and development of an instrument to assess cohesion in sport
success in individual team sports is unclear. Given the teams: the Group Environment Questionnaire. Journal of
inconsistency of results, future studies examining Sport Psychology, 7, 244± 266.
the combined individual and group eþ ects of cohesion Carron, A.V., Brawley, L.R. and Widmeyer, W.N. (1998).
and success should be illuminating. Because multi-level Measurement of cohesion in sport and exercise. In Advances
in Sport and Exercise Psychology Measurement (edited by
statistical modelling requires individual scores for all
J.L. Duda), pp. 213± 226. Morgantown, WV: Fitness
variables, the investigation of the relationship between Information Technology.
cohesion and success in individual sport teams lends Cohen, J. (1969). Statistical Power Analysis for the Behavioral
itself to such an approach. Sciences. New York: Academic Press.
A second potential avenue of research that could Cohen, J. (1992). A power primer. Psychological Bulletin, 112,
prove fruitful is to examine the `why’ of the cohesion± 155± 159.
success relationship. Paskevich (1995), for example, Dirks, K.T. (2000). Trust in leadership and team per-
found some support for the conclusion that collective formance: evidence from NCAA basketball. Journal of
eý cacy is a mediator in the relationship between Applied Psychology, 85, 1004± 1012.
cohesion and team performance outcome. Greater Everett, J.J., Smith, R.E. and Williams, K.D. (1992). Eþ ects
team cohesion contributes to greater collective eý cacy, of team cohesion and identi® ability on social loa® ng in
relay swimming performance. International Journal of Sport
which, in turn, contributes to enhanced team per-
Downloaded By: [Canadian Research Knowledge Network] At: 19:41 3 June 2011

Psychology, 23, 311± 324.

formance. Georgopolous, D.B. (1986). Organizational Structure, Problem
Our ® nding that there is a strong relationship between Solving, and Eþ ectiveness. San Francisco, CA: Jossey-Bass.
cohesion and success in naturally forming sport teams Gill, D.L. (1977). Cohesiveness and performance in sport
leads us to echo the view of Mullen and Copper (1994, groups. In Exercise and Sport Science Reviews, Vol. 5 (edited
pp. 223± 224): by R.S. Hutton), pp. 131± 155. Santa Barbara, CA: Journal
Publishing Aý liates.
the cohesiveness± performance eþ ect does not seem to be Grieve, F.G., Whelan, J.P. and Meyers, A.W. (2000). An
a rare and delicate `hothouse’ variety phenomenon experimental examination of the cohesion± performance
restricted to the controlled con® nes of the research relationship in an interactive team sport. Journal of Applied
laboratory. Rather, the cohesiveness± performance eþ ect Sport Psychology, 12, 219± 235.
is even more robust in the real world among real groups. Hedges, L.V. (1981). Distribution theory for Glass’ s estimator
of eþ ect size and related estimators. Journal of Educational
It also leads us to suggest that coaches and sport Statistics, 6, 107± 128.
Hedges, L.V. (1982). Fitting categorical models to eþ ect sizes
psychologists would do well to develop eþ ective team-
from a series of experiments. Journal of Educational Statistics,
building strategies in an attempt to in¯ uence cohesive-
7, 119± 137.
ness directly. Hedges, L.V. and Olkin, I. (1985). Statistical Methods for Meta-
analysis. New York: Academic Press.
Hoyle, R.H. and Crawford, A.M. (1994). Use of individual-
References level data to investigate group phenomena: issues and
strategies. Small Group Research, 25, 464± 485.
Allport, F.H. (1924). Social Psychology. Boston, MA: Jehn, K. and Shah, P. (1997). Interpersonal relationships and
Houghton Miü in. task performance: an examination of mediating processes in
Berardinis, J.D., Barwind, J., Flaningam, R.R. and Jenkins, V. friendship and acquaintance groups. Journal of Personality
(1983). Enhanced interpersonal relations as predictor and Social Psychology, 72, 775± 790.
of athletic performance. International Journal of Sport Kenny, D.A. and La Voie, L. (1985). Separating individual
Psychology, 14, 243± 251. and group eþ ects. Journal of Personality and Social
Carron, A.V. and Ball, J.R. (1978). Cause± eþ ect charac- Psychology, 48, 339± 348.
teristics of cohesiveness and participation motivation Kozub, S.A. and Button, C.J. (2000). The in¯ uence of
in intercollegiate hockey. International Review of Sport a competitive outcome on perceptions of cohesion in
Sociology, 12, 49± 60. rugby and swimming teams. International Journal of Sport
Carron, A.V. and Brawley, L.R. (2000). Cohesion: conceptual Psychology, 31, 82± 95.
and measurement issues. Small Group Research, 31, Landers, D.M. and L schen, G. (1974). Team performance
89± 106. outcome and cohesiveness of competitive coacting groups.
Carron, A.V. and Chelladurai, P. (1981). Cohesion as a factor International Review of Sport Sociology, 9, 57± 69.
in sport performance. International Review of Sport Sociology, Landers, D.M., Wilkinson, M.O., Hat® eld, B.D. and Barber,
16, 2± 41. H. (1982). Causality and the cohesion± performance
Carron, A.V. and Spink, K.S. (1995). The group size± relationship. Journal of Sport Psychology, 4, 170± 183.
cohesion relationship in minimal groups. Small Group Lenk, H. (1969). Top performance despite internal con¯ ict:
Research, 26, 86± 105. an antithesis to a functionalistic proposition. In Sport,
Carron, A.V., Widmeyer, W.N. and Brawley, L.R. (1985). The Culture, and Society: A Reader on the Sociology of Sport (edited
 126 Carron et al.

by J.W. Loy and G.S. Kenyon), pp. 393± 396. Toronto: Shanghi, G.M. and Carron, A.V. (1987). Group cohesion and
Collier-Macmillan. its relationship with performance and satisfaction among
Martens, R., Landers, D.M. and Loy, J.W. (1972). Sport high school basketball players. Canadian Journal of Sport
Cohesiveness Questionnaire. Unpublished manuscript, Sciences, 12, 20.
University of Illinois, Champaign, IL. Slater, M.R. and Sewell, D.F. (1994). An examination of the
Matheson, H., Mathes, S. and Murray, M. (1997). The eþ ect cohesion± performance relationship in university hockey
of winning and losing on female interactive and coactive teams. Journal of Sports Sciences, 12, 423± 431.
team cohesion. Journal of Sport B ehavior, 20, 284± 298. Thomas, J.R. and French, K.E. (1986). The use of meta-
Moritz, S.E. and Watson, C.B. (1998). Levels of analysis analysis in exercise and sport: a tutorial. Research Quarterly
issues in group psychology: using eý cacy as an example of a for Exercise and Sport, 57, 196± 204.
multilevel model. Group Dynamics: Theory, Research, and Widmeyer, W.N., Brawley, W.N. and Carron, A.V. (1990).
Practice, 2, 285± 298. The eþ ects of group size in sport. Journal of Sport and
Mullen, B. and Copper, C. (1994). The relation between Exercise Psychology, 12, 177± 190.
group cohesiveness and performance: an integration. Widmeyer, W.N., Carron, A.V. and Brawley, L.R. (1993).
Psychological Bulletin, 115, 210± 227. Group cohesion in sport and exercise. In Handbook of
Myers, J.L. (1972). Fundamentals in Experimental Design, 2nd Research in Sport Psychology (edited by R.N. Singer,
edn. Boston, MA: Allyn & Bacon. M. Murphey and K.L. Tennant), pp. 672± 692. New York:
Downloaded By: [Canadian Research Knowledge Network] At: 19:41 3 June 2011

Paskevich, D.M. (1995). Conceptual and measurement Macmillan.

factors of collective eý cacy in its relationship to cohesion Williams, J.M. and Hacker, C.M. (1982). Causal relationships
and performance outcome. Unpublished doctoral disserta- among cohesion, satisfaction, and performance in women’ s
tion, University of Waterloo, Waterloo, ON. intercollegiate ® eld hockey teams. Journal of Sport and
Paskevich, D.M., Estabrooks, P.A., Brawley, L.R. and Carron, Exercise Psychology, 4, 324± 337.
A.V. (2001). Group cohesion in sport and exercise. In Williams, J.M. and Widmeyer, W.N. (1991). The cohesion±
Handbook of Sport Psychology, 2nd edn (edited by R.N. performance outcome relationship in a coacting sport.
Singer, H.A. Hausenblas and C.M. Janelle), pp. 472± 494. Journal of Sport and Exercise Psychology, 13, 364± 371.
New York: Wiley. Zaccaro, S.F., Blair, V., Peterson, C. and Zazanis, M.
Rousseau, D.M. and House, R.J. (1994). Meso organizational (1995). Collective eý cacy. In Self-eý cacy, Adaptation,
behavior: avoiding three fundamental biases. In Trends in and Adjustment: Theory, Research, and Application (edited
Organizational Behavior, Vol. 1 (edited by C.L. Cooper and by J.E. Maddux), pp. 305± 328. New York: Plenum
D.M. Rousseau), pp. 13± 30. London: Wiley. Press.

View publication stats