.

THE EFFECT OF TEAM BUILDING

ON PERFORMANCE
An Integration

EDUARDO SALAS
Naval Air Warfare CenterTraining SystemsDivision
DREW ROZELL
BRIAN MULLEN
SyracuseUniversity
JAMES E. DRISKELL
Florida Maxima Corporation

In this article, meta-analytic integration of researchexamining the effects of team building

on performance is reported. Overall, therewas no significant effect of teambuilding on per-
formance. Howevel; the effects of teambuilding varied as afunction of thetype ofoperation-
alization ofperformance: Onobjective measuresofperformance, therewas a nonsignificant
tendencyfor teambuilding to decreaseperformance,whereason subjectivemeasuresofper-
formance, there was a significant, albeit small, tendencyfor team building to increaseper-
formance. Examination of the specific components of team building revealedthat interven-
tions emphasizing role clarification were more likely to increase performance, whereas
interventions that emphasizedgoal setting,problem solving, or interpersonal relations were
no more likely to renderan increaseor decreasein performance. Finally, the effects of team
building decreasedas afunction of the size of the team. The discussion considersimplica-
tions of these effects of team building on performance.

Harry, you weren't kidding when you said this Florida trip would be
no vacation! Building a group like us into a high-perfonning team
takes really hard work! But I wouldn't have missed it for worlds!

-J ack, a participant in a team-building intervention

(Shandler & Egan, 1996)

Team building, also called team development or group develop-

ment, is an extremely popular and common intervention. In fact,
..SMALL GROUP RESEARCH, Vol. 30 No.3, June 1999 309-329
@ 1999 Sage Publications,Inc.

309
 310 SMALL GROUP RESEARCH I June 1999

Porras and Berg (1978) observed that team building was one of the
most frequently used organization development interventions.
Enthusiastic reports can be found for the application of team build-
ing to such diverse settings as hotel management organizations
(e.g., Beckhard, 1966), microwave cooking manufacturers
(George, 1977), metropolitan rapid transit authorities (Golem-
biewski & Kiepper, 1976), medical rehabilitation teams (Halstead
etal., 1986), and nursing teams(Robinson-Kurpius & Keim, 1994).
A recent special issue of a scholarlyjournal was devotedto applica-
tions of team building to sportsteams (Hardy & Crace, 1997). Also,
as conveyed in the epigram above, team-building interventions are
evidently believed to lead to a substantive increase in team per-
formance. Shandler and Egan (1996) claim that by applying princi-
ples of team building, "any group can transform itself. ..into a
high-performing team" (p. x).
In spite of this popularity of the concept of team building, how-
ever, several reviewers (e.g., Buller, 1986; Woodman & Sherwood,
1980) have observed that there is no conclusive evidence that team
building renders an increase in team performance. Druckman and
Bjork (1994) noted that the enthusiasmfor theseapproachesamong
practitioners "is not matched by strong empirical support for their
effect on team performance" (p. 125). Similar to this, Smither,
Houston, and McIntire (1996) concluded that "Research findings
on the effectiveness of team building provide a complex mix of
results that make drawing firm conclusions difficult" (p. 324). :
More than 20 years after Beer (1976) attempted to formalize the:
notion of team building, some of the most fundamental questions :
about the effects of team building remain: Does team building
enhance performance? Why? Under what conditions? This article
reports the results of a meta-analytic integration of previous
research examining the effects of team building on performance.
Several issues need to be addressed in examining the effect
of teambuildingonperformance.
First,thesignificance
and ..
AUTHORS' NOTE: The views expressedin this article arethoseof the authorsand do not re-
flect the official position of their organizations. Correspondenceconcerning this article
..should be addressedto Eduardo Salas,Naval Air Warfare CenterTraining SystemsDivision, ..
Code 4961,12350 ResearchParkway,Orlando, FL 32826-3275.
 Salasetal.! TEAM BUILDING 311

magnitude-indeed, the very existence-of an effect of team build-

ing on performance needs to be established. Second, differences
between objective and subjective indices of performance have been
delineated in other domains, but there is no a priori determination
of the extent to which team building would differentially affect
these two different ways of operationalizing performance. Third,
the degreeto which a given team-building intervention engagesdif-
ferent components of team-building might influence the effective-
ness of the team-building intervention. Fourth, the effect of team
size needs to be specified, particularly in light of recentfmdings in
cognate areasindicating that group size exerts a considerable effect
on other group phenomena. Finally, the effect of the duration of
team-building interventions on their effectiveness is of consider-
able practical significance. Each of these considerations is
addressed in turn.

THE EXISTENCE OF A TEAM-BUILDING EFFECT

Team building has at its core the central notion that enlisting the
participation of a group in planning and implementing change will .'
be more effective than simply imposing change on the group from ,:
outside. The process by which the members of the team become
able to effectively participate in the targeted change requires that
the team acquire new skills and perceptions. The various defmi-
tions of team building that have been proposed in the research lit-
erature resonate to these elements of participation and acquisition
of new skills and perceptions. For example, Buller (1986) defined
team building as a planned intervention facilitated by a third-party
consultant that develops the problem-solving capacity and solves
major problems of an intact work group. Woodman and Sherwood
(1980) proposed that team building was designedto enhanceorgan-
izational effectivenessby improving team operation through devel-
oping problem-solving procedures and skills and increasing role
clarity (cf. Beer, 1976, 1980; DeMeuse & Liebowitz, 1981; Dyer,
1977).
The effects of team building are often described in extremely
.sanguinary terms. For example, the epigram at the beginning of this ..
 312 SMALL GROUP RESEARCH I June 1999

report is extracted from a primer on team building, entitled

VROOM! Turbo-Charged TeamBuilding (Shandler & Egan, 1996).
Therein, the president of a high-tech research and development
firm uses a team-building intervention to rescue his foundering
company, ensurehis promotion in the corporate hierarchy, and even
savehis failing marriage. It would seemthat any intervention capa-
" ble of such impressive effects would have generated an imposing
body of evidence supporting its effectiveness.
Unfortunately, a clear summary of the effects of team building
, on performance apparently cannot be formulated from narrative
readings of previous research. Indeed, several narrative literature
reviews have been conducted in this domain within the past two
decades, and the judgments rendered by these previous review ',
efforts have been inconclusive. Woodman and Sherwood (1980)
reviewed 30 papers and concluded that there were generally posi-
tive results for most of the studies but that there was no clear-cut :
evidence to suggestthat team building can improve performance.
DeMeuse and Liebowitz (1981) reviewed 36 previous reports and
similarly concluded that there were generally positive descriptions -:'
of the team-building interventions but that there was a dearth of evi-
~, dence for beneficial effects of team building on performance. Bul-
ler (1986) reviewed 9 studies, Sundstrom, DeMeuse, and Futrell:
(1990) reviewed 13 studies, and Tannenbaum, Beard, and Salas
(1992) reviewed 17studies, and (like each of the previous reviews) ..'
each of these narrative reviews concluded that team building was
described in encouragingways by most of the studies reviewed, but
there was a general lack of definitive, compelling evidence for the
beneficial effects of team building on performance.
It is interesting to consider the nature of the body of studies con-
sidered in many previous narrative reviews of the effects of team
building. First, there was a remarkable lack of convergence among
these reviews regarding which studies to include in a review of
team-building interventions. Of the 69 reports examined in at least
one of these narrative reviews, on average only 4 studies (or 5.8%)
were included by both of any two reviews. The point is that more
than 90% of the literature summarized in each of these narrative. .

~
 Salas et aI.1 TEAM BUILDING 313

reviews of team building was referred to in none of the other narra-

tive reviews. This stunning lack of convergence on precisely what
constitutes an includable study of team building is in part a reflec-
tion of the complexity of the notion of team building (to be eluci-
dated further).
It is useful to consider a few other intriguing facets of the studies
considered in previous narrative reviews of the effects of team
building. The most common type of report examined in theseprevi-
ous reviews was the case study (20 out of 69 studies or 29%),
describing the application of a team-building intervention in some
organizational setting with no empirical examination of the effec-
tiveness of the intervention (e.g., Beckhard, 1966; George, 1977).
These case studies, although interesting and illuminating in an
anecdotal way, can hardly be said to provide definitive, conclusive
evidence in support of team building. The second most common
type of report examined in these previous reviews reported the
effects of team building on some affective, nonperformance out-
come measure(15 out of 69 studies or 22%). This tendencyfor stud-
ies of team-building interventions to focus on affective responsesto
the intervention while excluding genuine performance results of
the intervention was recognized in some of the previous narrative
reviews (notably Buller, 1986; DeMeuse & Liebowitz, 1981;
Woodman & Sherwood, 1980). Only 6 of the 69 studies considered !
in the previous narrative reviews provided a precise statistical test .
of the effects of a team-building intervention on performance. :
In sum, we cannot detemline whether there is a beneficial effect
of team building on performance based on previous narrative
perusals of this research domain. Part of the problem lies in the
ambiguity of what precisely is team building and what studies
should be included in an effort to integrate the effect of team build-
ing on performance. Part of the problem is that the team-building
literature includes a variety of case studies, nonempirical reports,
and tests of various types of outcome measures.Therefore, the first
goal of the presenteffort was to establishthe empirical significance
and the strength of the effects of team building on performance.
314 SMALL GROUP RESEARCH I June 1999

COMPONENTSOF TEAM BffiLDING

Beer (1976, 1980), Dyer (1977), and Buller (1986) have dis-
cussedfour current models of teambuilding: Goal setting, interper-
sonal relations, problem solving, and role clarification. Goal set-
ting emphasizes the setting of objectives and the development of
individual and team goals. Team members exposed to a goal-
setting team-building intervention are supposed to become
involved in action planning to identify ways to achieve those goals.
Interpersonal relations emphasizesan increase in team work skills,
such as mutual supportiveness, communication, and sharing of
feelings. Team members exposed to an interpersonal relations
team-building intervention are supposed to develop trust in one
another and confidence in the team. Problem solving emphasizes
the identification of major problems in the team. Team members
exposed to a problem-solving team-building intervention are sup-
posed to become involved in action planning for the solution of .
those problems and for implementing and evaluating those solu-
tions. Role clarification emphasizes increased communication
among team members regarding their respective roles within the
team. Team members exposedto a role-clarification team-building
intervention are supposedto achieve better understanding of their
and others' respective roles and duties within the team. .
,
These models might be best thought of as components of any f
given team-building intervention. That is, any team-building inter- I
vention might engage any or all of these various components in
varying degrees. For example, Wegenast's (1983) team-building
intervention in child protective service workers involved a rela-
tively high emphasis on problem solving, but relatively low empha-
sis on interpersonal relations. Alternatively, Paul and Gross's
(1981) team-building intervention in the communications and elec-
trical division of city governmentinvolved a relatively high empha-
sis on role clarification but relatively low emphasis on goal setting.
This diversity in team-building interventions represents one of
the major challenges to previous efforts to make sense of the re-
searchliterature on team building. In the present effort, we viewed
this diversity among team-building interventions a~ a. !Jnique
 SalasetaI.! TEAM BUILDING 315

opportunity to gauge the relative impact of the four components of

team building delineated above (goal setting, interpersonal rela-
tions, problem solving, and role clarification). To date, there is no
evidence regarding which of thesecomponents of team building are
more critical to the putative effects of team building on perform-
ance. Therefore, another important goal of the present effort is to
determine the relative contributions of these components of team
building to the team building-performance effect.

CONTRIBUTIONS OF OPERATIONALIZATION OF PERFORMANCE

Two broad sets of operationalizations of performance typify

research in the area of team performance. On one hand, objective
indicators of performance include direct measures of countable
behaviors, like the number of units sold or the tons of ore mined. On
the other hand, subjective indicators of performance include ratings
of the performance or effectiveness of the team. Two separate
meta-analyses have reported relatively modest correlations
between objective and subjective measures of performance (Bom-
mer, Johnson, Rich, Podsakoff, & MacKenzie, 1995, r = .389;
Heneman, 1986, r = .27). This led Heneman (1986) to conclude that
objective and subjective measuresof performance are not substitut-
able, and that "when reviews of the literature are conducted, results
should be grouped by the type of performance criteria" (p. 820).
There is no a priori expectation for the effect of team building on
performance to be greater for one or the other of these two opera-
tionalizations. However, the relatively modest correlations previ- !
ously observed between objective and subjective measuresof per-
formance led us to examine the effect of team building on
performance separately for each type of performance measure.

CONTRIBUTIONS OF TEAM SIZE

The size of social groups has been shown to affect the magnitude
of several other group phenomena, including the participation-
leadership effect (Mullen, Salas, & Driskell, 1989), the ingroup-
316 SMALL GROUP RESEARCH I June 1999

bias effect (Mullen, Brown, & Smith, 1992), the cohesiveness-

performance effect (Mullen & Copper, 1994), the groupthink effect
(Mullen, Anthony, Salas,& Driskell, 1994), and the effect of stress
inoculation training on performance (Saunders,Driskell, Johnston,
& Salas, 1996) (for a discussion, seeMullen, 1991). Larger groups
tend to encourage deindividuation among group members (e.g.,
Mullen, 1987). Moreover, it is well established that as group size
increases,members' liking for the group (e.g., Indik, 1965) and per-
formance (Mullen, 1987) tend to decrease.In larger groups, group
members would be expected to be less engaged in the group, less
open to the team-building intervention, and less likely to perform
with any concern for excellence. Thus, we would expect that the
effect of team building on performance would be weaker in larger
teams and stronger in smaller teams.
In spite of the consistent tendency for the size of the group to
affect several similar group phenomena, there has been no consid- '

eration to date of whether team size moderates the effect of team

building on performance. Therefore, another important goal of the ...
present effort was to determine the extent to which team size affects "
the impact of team building on performance.

CONTRlBUflONS OF DURATION OF
THE TEAM-BUILDING INTERVENTION

The duration of team-building interventions varies considerably

in the extant literature, from 2 days (Smith, 1976)to 8 months (Bul-
ler & Bell, 1986). Common sensemight suggestthat interventions
of longer duration might exert greater effects: If a little of the inter-
vention works, then a lot of the intervention might work evenmore.
However, interventions of longer duration have paradoxically been
found to render weaker effects in suchdiverse areasasmental prac-
tice effects (Driskell, Copper, & Moran, 1994), self-help treatment
groups (Gould & Clum, 1993), and group-work interventions with
depressed older persons (Gorey & Cryns, 1991). However, the
nature of the relationship between the duration of team building
and the effect of team building on performance.r.emainsunclear.
Therefore, another goal of the present effort is to determine the
" "...'1

Salasetal. IlEAM BUn..DING 317

effect of the duration of the team-building intervention on the team

building-perfonnance effect.

A META-ANALYUC INTEGRATION

In an effort to provide a clearer picture of the effects of team

building on perfonnance, a meta-analytic integration (Mullen,
1989; Mullen & Rosenthal, 1985; Rosenthal, 1991)was conducted.
The goals of this effort were (a) to provide a precise summary of the
significance, the strength, and the direction of the effects of team
building on perfonnance; (b) to examine different magnitudes of
this effect of team building on perfonnance as a function of the
operationalization of perfonnance; (c) to examine the possible
moderation of the effect of team building on perfonnance by the
four components of team building (goal setting, interpersonal rela- :
tions, problem solving, and role clarification); (d) to determine
whether the effect of team building on perfonnance would decrease
as a function of team size; and (e) to scrutinize the effects of the
duration of the team-building intervention on the effects of team
building.

PROCEDURE

Using all of the standardliterature searchtechniques,an exhaus-

tive search was conducted for studies testing the effect of team-
building interventions on perfonnance. First, on-line computer
searcheswere conducted, using the keywords team(s), group(s), or
crew(s) and building, development, perfonnance, productivity, or
effects. These computer searcheswere supplemented by ancestry
approach and descendency approach searches, correspondence
with researchersactive in this domain (the invisible college), and
manually examining the past 30 years of social psychology, applied
psychology, and managementjournals (seeMullen, 1989, for a dis-
cussion of literature searchtechniques). Second, we reviewed each
of the team building studies included in the previous narrative

.
318 SMALL GROUP RESEARCH I June 1999

reviews by Woodman and Sherwood (1980), DeMeuse and Lie-

bowitz (1981), Buller (1986), Sundstrom, DeMeuse, and Futrell
(1990), and Tannenbaum, Beard, and Salas (1992). Any studies
that were available as of March 1998 were eligible for inclusion in
this integration.
Studies were included if they met the following criteria: Team
members in the studies had to be adolescentsor adults not sampled
from abnormal populations. Studies had to report (or allow the
reconstruction of) a precise test of the effect of a team-building
intervention either on an objective measure of performance (e.g.,
productivity data) or a subjective measureof performance (e.g., rat-
ing of team effectiveness). Case studies were not included (e.g.,
"Teamwork Through Conflict," 1971), nor were T-group (e.g., Argy-
ris, 1962) or managerial grid (e.g., Blake & Mouton, 1966) inter-
ventions. The effect of these criteria for inclusion was to focus on
the effects of team building on performance in studies that were
optimally homogeneous in methodological terms. Hypothesis tests
were coded as having a positive direction of effect if team building
increased performance, and they were coded as having a negative
direction of effect if team building decreasedperformance.
These selection criteria rendered a total of 11 studies (Buller &
Bell, 1986; Eden, 1986; Friedlander, 1967; Howard, 1979; Kim-
berly&Nielsen, 1975; Paul & Gross, 1981; Smith, 1976; Wakeley&
Shaw, 1965; Wegenast, 1983; Wexler, 1990; Woodman, 1978).
These 11 articles yielded 16 separatetests of the effects of team
building on performance, representingthe responsesof 2,806 team
members in 307 teams.! Attesting to the thoroughness of the litera-
ture search, the included reports spanneda period of more than 30
years,and 4 of the 11reports (36%) were unpublished dissertations.
In addition to the requisite statistical information, eachhypothe-
sis test was coded for sevenpredictors. The nature of the operation-
alization of performance (objective vs. subjective), the size of the
team, and the duration of the team-building intervention were all
directly coded from each report by two judges with perfect reliabil-
ity. In addition, each team-building intervention procedure was
also scored for the extent to which t~~t intervention would engage
 Salas et al.1 TEAM BUR-DING 319

each of the components of teambuilding. To be specific, two judges

independently rated each team-building intervention on a scale
from 0 (low) to 100 (high) for goal setting, interpersonal relations,
problem solving, and role clarification (as defined above).
Acceptably high levels of interjudge reliability were obtained for
each of these components: Goal setting (interjudge reliability r =
.693, Spearman-Brown Effective Reliability R = .819), interper-
sonal relations (r = .822, R = .902), problem solving (r = .945, R =
.972), androleclarification(r= .580,R= .734). The extent to which
each intervention engaged each component of team building was
set equal to the average of the judges' ratings for eachcomponent.
The hypothesis tests included in this meta-analysis, along with the
corresponding statistical information and predictors for each
hypothesis test, are presented in Table 1.

RESULTS

General effect. Overall, there was a nonsignificant, z = 0.132, p =

.45, negligible, 4isher= 0.007, r = .007, effect of team-building inter-
ventions on performance. Thus, the weight of available evidence i
fails to substantiate the effectiveness of team-building interven- :
tions to increase performance.2 r.
i
J

Objective versus subjective measures ofperformance. For the k .[

= 8 hypothesis tests that used objective measuresof performance,
the effect of team building on performance was a nonsignificant, z = !.
-0.511, p = .70, negative, 4isher= -0.040, r = -.040. For the k = 8
hypothesis tests that used subjective measures of performance,
there was a significant, z = 2.589,p = .004, albeit still weak, 4isher=
0.140, r = .139, positive effect of team building on performance.
These two effects were significantly different from one another, z =
3.398, p = .0003. Thus, although subjective reports indicate that
team building appears to increase performance, objective indica-
tors of performance reveal no genuine effect of team building on
performance.
 ~
~
~~~~~~~~~~~~~~~~
ooo~oo~~~o~ooo~~
-N-MN---~~~~NN~~
u"
~ ~OOOOOOOOOOOOOOO u
~ NOO~OOOOO~O~~~OO a
~-~N~~~~~~~~~~~~ E
d .g
,~ ~
~ ~~oooooooo~o~~ ~ Q,
e= N ~ N ~ 0 0 0 0 0 0 0 0 N 0 ~ ~.~ "E
M~M~~~~~M~~NM~NN -~
'5 ~
u u
~
~ ~~~~~~~~~~~~~~~~
", M ~
0 ~~~ ~00 N 0 ~ N ~ ~ ~
~
~
~.,
~ ~~N~NMNNN~NNN-~~ ~
UII "'"-
~t.s
§ -;;- biI-e ~
.:: 6' 0 0 0 M ~ ~ N N M N ~ N 0 0 ~ ~ .53 ~ 2
~~ ~~~ --~~~-- >~
§;: NNN o~~
Q~ tl)tij-:5
~52'" 2
"
,!:f MNM-~~OONM~~~~MM
~'O: ~ -o~
.g ,~
~ M NN ~'O
~.c53
II §'~
\) ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ';n
~ »>,...,...»» ,..., > .~~
~ ,... ,... ,... , '" ~
","""uu""""""uuuuu",u = >
..,
~
u u u ~ ~
~~~'~'~~~~~.""'.~'~'~'~~.""'."'~
u u u u ~ ~ ~ ~ ~ u ~ 0 '" 'l::
~
"
oo
.c o
'~.~'~.c.c'~.~.~.~.c.c.c.c.c'~.c
.c .c oooo
=' =' .c .c .c .c =' =' =' =' =' o.c =' ~
l::
~
= bI)
~
tI)tI) tI)tI)tI)tI)tI) tI) ~~=
~Q,~
~ "a ~ ~.
0 1+11++11++++++1+ S~8
Q ~-ea
~ ~o-~ 88 o~ N~~~
NMONN e.='E
2=0
Z -~-~~-~~~~~ ='0",,"
-~-,N""--~N-""""---
~ ~~_N~~M- N~~~~ -=
II~&
O~-~~O~~~N~-~~~~
~","E .~
-~ NO O~-~~-N-MO bI) ~
.~ OM8NN~-0~~~~NN-N =-:5
~ I'II.'II"'."I"~~,S
~ ~ ~ ~ tI)~~
I ~~~N~ -CN ~~ 8 °~ "a.:::,...
~~ ~M _~M~~M~ -~ 00:9
~ ...~~~~ 1I~r-:~ II~N II~ .-:'0: C 'S
~ .:: 0-0 'OO ~-.t NN-- ",8.c
~
,C
.~
~
1IIIIIIIggllllll~II~ "-" II ~lIlIlIlItI)5~ C=,tij
-.~ ~ ~ ~ ~ --~ M M --~ 0 0 ~ ~ .= ~
N,..,..-~~,..,..M~,..,..~~~~... ...
.S ~~~~ti;.'r:~~'i:.~~'i:.~~~~ ~~~
'0~ ~...'c'"
r., ...
'0 ~ ",
~ =~
.a ~ o~~
u
= -~
'-'
= ~ ="'u
.l::
0 Q,~ '0
'"~ ~
~ ~
~ --
~ '-' U u=
'" ,...
:= ~ ~ ~ ~ g =
w -~ ,... -'" ;.., ~ bI)
.a '-' ~ z '-' ~ ~ c .;n
00 -~
U -~'O '"",'cM tI)~ ~
~ 115
='~
~ '-' ~ a e G '0 ~ ~ ~ ~ = ,~
..'0 ~ ~ ->, C ~ a '-' ~
~ = ~'O'-'~ ~ a 8 ~'B
~ ~ -a '0 -;: ] ~ ~ ~ ~ 5 ,. ~ Q,
~ .c- ~ ~:o~,8 ~-:5u 5.£ '00 ~~ ~
=- '"' -~ ~ ...5 -'= ~ bI) >< f-- L. .c=
-< :!"3 ~'C o:~ ~ = ~ ~ ~ ~ 0 r- E--
E-i ~ ~ 11j~~~ ~tI);;oo;;oo;;oo '?" Zd.c
320
---Salas .SaIas
et all TEAMBUlLDIN~
et aI./TEAM BUll-DING 321
~,
~,"
..,;.",

Contributions of components ofteambuilding. The meanjudged

level of each component of team building was used to examine the
contributions of the various components of team building. The
team building-perfonnance effect did not vary as a function of the
level of goal setting ~ngagedby the team-building intervention, (r=
-.164, z= 1.165,p = .12). In other words, the extent to which a given
team-building intervention engagedthe component of goal setting
neither increased or decreasedthe magnitude of the effect of team
building on perfonnance. Similar to this, the extent to which a given
team-building intervention engaged the component of interper-
sonalrelations (r=-.056, z=0.485,p= .3138) or problem solving
(r = -.049, z = 0.342, p = .37) neither increased or decreasedthe
magnitude of the effect of team building on perfonnance. The only
component of team building to significantly predict the effect of
team building on performance was role clarification (r = .759, z =
5.533, p < .001). In other words, the more a given team-building ;
intervention engaged the component of role clarification, the
stronger the magnitude of the effect of team building on
performance.
It should be noted that there is no moderation of these patterns
for components of team building by the operationalization of per-
formance. That is, for both objective and subjective indices respec-
tively, there was no effect of goal setting (r = -.060, z = 0.182, p = .43
and r=-.110, z=0.972,p = .17), interpersonal(r=-.382,z= 1.258,
p = .10 and r = -.035, Z = 0.325, p = .37), or problem solving
(r= -.305, Z= O.780,p =.22 and r= .088, Z= 1.298,p = .10). How-
ever, for both objective and subjective indices respectively, there
was a significant effect of role clarification (r = .712, Z= 4.565,
p < .001 and r = .752, Z = 4.594, p < .001).

Contributions of team size. Overall, the effects of team building

on performance decreasedas a function of the size of the team, r =
-344, Z = 5.965, p < .001. This inverse effect of team size on the
effect of team building held for both objective measures of per-
formance (r = -.400, Z= 2.383, p = .009) and subjective measuresof
performance (r = -.407, Z = 7.796, p < .001). Thus, whatever

.
322 SMALL GROUP RESEARCH I June 1999

positive effect team building may have is most likely to emerge in

small groups.

Contributions ofduration ofthe team-buildingintervention. Over-

all, there was a marginal tendency for the effects of team building
on performance to decreaseas a function of duration of the team-
building intervention, r = -.197, z = 1.452, P = .07. This inverse
effect of duration on the effect of team building did not obtain for
objective measuresof performance (r = -.056, z = 0.194, p = .42),
but did obtain for subjective measures of performance (r = -.208,
z = 2.562, p = .005). Thus, the negligible effects of team building on
objective indices of performance are not influenced by the duration
of the team-building intervention, but the appearance of benefits
from team building on more subjective measuresseemsto dissipate
for more lengthy team-building interventions.

DISCUSSION

One of the most important results to emerge from the present

analyses is the documentation of the overall nonsignificant and
negligible effect of team building on performance. These results
suggest that the enthusiastic testimonials on behalf of team-build-
ing interventions should be interpreted with caution. The overall
magnitude of the effect of team building on performance of r = .007
indicates that approximately .005% of the variability in a team's
performance might be accounted for by knowing whether the team
had gone through a team-building intervention. In other words,
approximately 99% of the variability in a team's performance is
attributable to factors other than whether the team had gone
through a team-building intervention. Relative to some rather bold
proclamations regarding the beneficial effects of team building
interventions (e.g., Shandler & Egan, 1996), future summaries of
the effects of team building on performance should be a bit more
modest.
When the data were disaggregated into objective performance
indicators versus subjective performance indicators, it was shown

.
 Salasetat. / TEAM BUILDING 323

that objective performance indicators rendered functionally no

effect of team building on performance, whereas subjective per-
formance indicators rendered a significant, albeit still weak, effect
of team building on performance. This difference between results
for objective performance indicators and subjective performance
indicators is not surprising, given the modest convergence previ-
ously observed between these different operationalizations (Bom-
mer et al., 1995; Heneman, 1986). The analyses for these two dif-
ferent performance indicators are even more sobering than the
overall lack of beneficial effects of team building: Not only is there
an overall lack of beneficial effects of teambuilding, but there is not
even much of a measurement-specific benefit of team-building
interventions of which to speak.
The examination of the contributions of the four components of
team building yielded a very intriguing result: The team building-
performance effect did not vary asa function of the level of goal set-;
ting, interpersonal relations, or problem solving engaged by the
team-building intervention. It was only role clarification that
seemedto make a genuine contribution to the effect of team build-
ing on performance, and this pattern emerged for both of the opera-
tionalizations of performance. This significant effect for role clari-
fication is entirely consistent with evidence from other quarters
regarding the importance of role clarification and role ambiguity in
team performance. For example, Abramis's (1994) recent meta-
analysis reported that the general weight of evidence indicates that
role ambiguity exerts a negative impact on performance. The pres-
ent results suggest that future research on team building and per-
formance should further examine the positive effects of the role
clarification component of team building.
The results for team size on the effects of team building are con-
sistent with patterns documented in other group phenomena (e.g.,
Mullen et al., 1989;Mullen et al., 1992; Mullen et aI., 1994; Mullen &
Copper, 1994; Mullen, 1991). For years, researchers studying
group and team performance have searchedfor the magic numbers
that could summarize the effects of the group on the individual. It is
interesting to note that Old (1946) long ago proposed that the opti-
mum size for a group was 0.7 persons. This tongue-in-cheek
 324 SMALL GROUP RESEARCH I June 1999

proposal, although somewhat limiting in terms of implementing

group training, keeps with the results of the presentanalyses:As the
size of the team increases,the effectiveness of team-building inter-
ventions decreases.Optimal benefit from a team-building interven-
tion seems most likely to be obtained with relatively small teams.
The results for the effects of duration of the team-building inter-
vention are more complex. Although there was a marginal ten-
dency overall for longer team-building interventions to render
weaker effects, this pattern was actually due only to the influence of
duration on the effects of team building on subjective indicators of
performance. Thus, although shorter duration team-building inter-
ventions might be of some benefit to subjective performance mea-
sures,there does not seemto be any difference between shorter and :
longer team-building interventions on objective performance
measures.
Of course, there are limitations to our analysis and to the impli- :
cations that can be drawn from it. As noted earlier, the term team
building has been defined somewhat broadly in the literature to
include various types of group interventions that are conceptually
quite dissimilar. As Lipsey and Wilson (1993) noted in a recent
review of the general efficacy of psychological treatments, there
are often gray areas as to what constitutes a certain intervention.
Our approach was to cast a very narrow net, adopting a precise
approach to defining team building and to including studies in the
analysis. This decision resulted in a small, tightly-focused database
of studies, yet one that we feel is highly representative of this
researchdomain. Second, this analysis allowed us to examine sev- ,.
eral theoretically and practically interesting moderators of the
effect of team building on performance, suchas team size and dura-
tion of training. There are other potentially informative modera-
tors, such as the type or task (cf. Tannenbaumet al., 1992), that the
literature did not allow us to examine. Further primary level
researchis needed to examine other conditions under which team
building may be more effective. Finally, there are methodological
limitations of this analysis. There were no studies in this database
..that reported reliability coefficients for the performance measures~
 Salas et al.! TEAM BUILDING 325

used, so adjustments for attenuation were not attempted. The fail-

ure of the source information to report reliabilities for performance
measuresis not unique to this particular meta-analysis, but it repre-
sents a common limitation of the researchliterature.
In summary, the present results indicate that overall there is no
significant effect of team building on performance. Moreover, what j
little benefits team building might exert on performance are likely
to be seen in interventions that emphasize role clarification in
smaller groups (and perhaps, in interventions that are more brief).
Future research might be directed toward further examinations of
the relative contributions of the four components of team building
and to scrutinizing the differential effects of team building on dif-
ferent indicators of performance. Finally, we note that in examining
team building almost 20 years ago, Scherer (1979) posed the ques-
tion, "How can something that feels so good not be worthwhile?"
(p. 335). Subsequentreviews of team-building researchcautioned
that evidence of an effect of team building on performance was
"inconclusive" (Buller, 1986, p. 147), "unsubstantiated" (Wood-
man & Sherwood, 1980, p. 166), "equivocal" (Tannenbaumet al., ..'
.1992, p. 146), and "mixed" (Sundstrom et al., 1990, p. 128). The
-: present analysis of the effect of team building on performance pro-
vides an empirical basis for the caution voiced by thesereviewers.
: ~i:

NOTES'
~
1. The included studies reported varying numbersof hypothesistests,ranging from one
per study (e.g., Eden, 1986) to three per study (e.g., Buller & Bell, 1986). In the meta-
analysis reported, each hypothesis test was treated as an independentobservation. This
assumptionof independenceis patentlyfalse. For example,eachof the threehypothesistests
included in Buller and Bell (1986)was derived from the samesubject population at the same
time. However,without making this assumptionof nonindependence,one would beforced to
selectthe besthypothesis test from a study such asBuller and Bell or to pool the resultsfrom
the reported hypothesistestsinto a single test. In the presentcontext,thesealternatives seem
even more arbitrary and capricious than the presentassumptionof independence.The effects
of this assumptionof independenceare examined later.
2. As indicated in Note 1,the assumptionthat eachof the 16hypothesistestsrepresented
an independentobservationis false. However,it can be seenthat suchan assumptiondoesnot
seem to render a distorted summary of this researchdomain. Consider the results of a
 326 SMALL GROUP RESEARCH I June 1999

supplementary meta-analysisof wholly independenteffects, in which multiple hypothesis

testsobtained from a single studywere combinedinto a single test. This heavy-handedsolu-
tion precludesthe examination of the effects of any of the moderatorsconsideredabove,but
it does eliminate the problem of nonindependence.This produced 11 distinct, wholly inde-
pendenthypothesis tests,one from eachincludable study. The results of this supplemental
meta-analysis revealed the samepatterns reported previously: There was a nonsignificant,
z = 0.440, P = .33, negligible, ZFisher
= 0.032, r = .032, effect of team-building interventions
on performance. These results indicate that the degree of distortion engenderedby the
assumptionof independenceof the original 16 hypothesis testsis (at worst) tolerable.

REFERENCES

Abramis, D. J. (1994). Work role ambiguity, job satisfaction,and job performance: Meta-
analysesand review. Psychological Reports, 75, 1411-1433.
Argyris, C. (1962). Interpersonal competenceand organizational behavior.Homewood,IL:
Irwin.
Beckhard, R. (1966). An organization improvement program in a decentralizedorganiza-
tion. Journal ofApplied Behavioral Science,2, 3-25.
Beer, M. (1976). The technology of organization development. In M. D. Dunnette (Ed.),
Handbook of industrial and organizational psychology (pp. 937-994). Chicago: Rand
McNally.
Beer, M. (1980). Organization change and development: A systemsview. Glenview, IL: .:.:
Scott, Foresman.
i Bigelow, R. C. (1971). Changing classroominteraction through organizationdevelopment.
In R. A. Schmuck& M. B. Miles (Eds.),Organization developmentin schools.Palo Alto,
CA: National Press.
Blake, R. R., & Mouton, J. S. (1966). Someeffects of managerialgrid training on union and
managementattitudes towards supervision. Journal ofApplied Behavioral Science,2,
387-400.
Bommer, W. H., Johnson,J. L., Rich, G. A., Podsakoff,P. M., & MacKenzie, S. B. (1995).
On the interchangeability of objective and subjective measuresof employee perform-
ance: A meta-analysis. PersonnelPsychology,48, 587-605.
Buller, P. F. (1986). The team building-task performance relation: Some conceptual and
methodological refinements. Group and Organization Studies,I I, 147-168.
Buller, P. F., & Bell, C. H. (1986). Effects of teambuilding and goal setting on productivity:
A field experiment. Academy of ManagementJournal, 29, 305-328.
DeMeuse,K. P.,& Liebowitz, S. J. (1981). An empirical analysisof team building research.
Group and Organization Studies,6, 357-378.
Driskell, J. E., Copper,C. M., & Moran, A. (1994): Does mentalpractice enhanceperform-
ance?Journa.z ofApplied PsJ!chology,79,481-492.
Druckman, D., & Bjork, R. A. (Eds.). (1994). Learning, remembering,believing: Enhancing
human performance. Washington,DC: National Academy Press.
Dyer, W. G. (1977). Teambuilding: Issuesand alternatives.Reading,MA: AddisonWesley.
Eden,D. (1986). Teamdevelopment:Quasi-experimentalconfirmation among combatcom-
...panies. Group and Organization Studies,11,133-146.
 Salaset al.! TEAM BUILDING 327

Friedlander,F. (1967). The impact of organizationaltraining laboratories uponeffectiveness

of ongoing work groups. PersonnelPsychology,20, 289-307.
George,W. W. (1977). Taskteams for rapid growth. Harvard BusinessReview,55, 71-80.
Golembiewski,R. T., & Kiepper,A. (1976). MARTA: Towardan effective opengiant. Public
Administration Review,36, 46-60.
Gorey, K. M., & Cryns, A. G. (1991). Group work as interventive modality with the older
depressedclient: A meta-analytic review. Journal ofGerontological Social Work, 16,
137-157.
Gould, R. A., & Clum, G. A. (1993). A meta-analysisof self-help treatment approaches.
Clinical Psychology Review,13, 169-186.
Halstead,L. S., Rintala, D. H., Kanellos, M., Griffin, B., Higgins, L., Rheinecker,N., White-
side, W., & Healy, J. E. (1986). The innovative rehabilitation team: An experiment in
team building. Archives of Physical Medicine and Rehabilitation, 67, 357-361.
Hardy,C. J., & Crace,R. K. (1997). Foundationsof teambuilding: Introduction to the team
building primer. Journal ofApplied Sport Psychology,9, 1-10.
Heneman,R. L. (1986). The relationship betweensupervisoryratings and results-oriented
measuresof performance: A meta-analysis.PersonnelPsychology,39, 811-826.
Howard,J. (1979). Organizational teambuilding in church-relatedteams.Unpublished dis-
sertation, California School of Professional Psychology,San Diego.
Indik, B. P. (1965). Operational size and memberparticipation: Someempirical testsof alter-
native explanations. Human Relations, 18, 339-350.
Kimberly, J. R., & Nielsen,W. R. (1975). Organizationdevelopmentandchangein organiza-
tional performance. Administrative Science Quarterly, 20,191-206.
Lipsey, M. W., & Wilson, D. B. (1993). The efficacy of psychological, educational, and
behavioral treatment: Confinnation from meta-analysis. American Psychologist, 48,
1181-1209.
Mullen, B. (1987). Self-attentiontheory. In B. Mullen & G. R. Goethals (Eds.), Theories of
group behavior (pp. 125-146). New York: Springer-Verlag.
Mullen, B. (1989). Advanced BASIC Meta-analysis. Hillsdale, NJ: Lawrence Erlbaum.
Mullen, B. (1991). Group composition, salience,and cognitive representations:The phe-
nomenology of being in a group. Journal of Experimental Social Psychology, 27,
297-323.
Mullen, B., Anthony, T., Salas,E., & Driskell, J. E. (1994). Group cohesivenessand quality
of decision making: An integration of tests of the groupthink hypothesis. Small Group
Research,25, 189-204. :
Mullen, B., Brown, R. J., & Smith, C. (1992). Ingroup bias as a function of salience,rele-
vance,and status: An integration. EuropeanJournal ofSocial Psychology,22, 103-122.
Mullen, B., & Copper,C. (1994). The relation betweengroup cohesivenessand perform-
ance: An integration. Psychological Bulletin, 115, 210-227.
Mullen, B., & Rosenthal,R. (1985). BASICMeta-analysis: Proceduresandprograms. Hills-
dale, NJ: Lawrence Erlbaum.
Mullen, B., Salas,E., & Driskell, J. E. (1989). Salience,motivation, and artifact as contribu-
tions to the relation between participation rate and leadership.Journal ofExperimental
Social Psychology, 25, 545-559.
Old, B. S. (1946). On the mathematicsof committees,boards,andpanels.Scientific Monthly,
63,75-78.

..4
 328 SMALL GROUP RESEARCH I June 1999

Paul,C. F., & Gross,A. C. (1981). Increasing productivity and morale in a municipality:
Effects of organizational development. Journal of Applied Behavioral Science, 17,
59-78.
Porras,J. I., & Berg, P.O. (1978). The impact of organization development. Academy of
Management Review,3, 249-266.
Robinson-Kurpius, S.E., & Keirn, J. (1994). Teambuilding for nursesexperiencing burnout
and poor morale. Journalfor Specialistsin Group Work,19, 155-161.
Rosenthal,R. (1991). Meta-analytic proceduresfor social research.Newbury Park, CA:
Sage.
Saunders,T., Driskell, J. E., Johnston,J.,& Salas,E. (1996). The effect of stressinoculation
training on anxiety and performance.Journal of Occupational Health Psychology, 1,
170-186.
Scherer,J. J. (1979). Can team building increaseproductivity? or How can something that
feels so good not be worthwhile? Group and Organizational Studies,4, 335-351.
Shandler, M., & Egan, M. (1996). VROOM! Turbo-charged team building. New York:
American ManagementAssociation.
Smith, P.E. (1976). Managementmodeling training to improve morale and customersatis- , '.-
faction. Personnel Psychology,29, 351-359. : .
Smither, R. D., Houston, J. M., & Mcintire, S. A. (1996). Organizational development:
Strategiesfor changing environments.New York: HarperCollins.
Sundstrom, E., DeMeuse,K. P., & Futrell, D. (1990). Work teams:Applications and effec-
tiveness. American Psychologist,45,120-133.
Tannenbaum,S. I., Beard,R. L., & Salas,E. (1992). Teambuilding and its influence on team
effectiveness: An examination of conceptualand empirical developments.In K. Kelley
(Ed.), Issues,theory,and researchin industrial/organizational psychology.Amsterdam:
Elsevier.
Teamwork through conflict. (1971). Business Week,2168, pp. 44-50.
Wakeley,J. H., & Shaw,M. F. (1965). Managementtraining: An integratedapproach. Train-
ing Directors Journal,19, 2-13.
Wegenast,D. P. (1983). Teambuilding: An application of leadershipskills to casemanage-
ment in child protective services.Unpublished doctoral dissertation,City University of i
New York, New York City.
Wexler, K. J. (1990). Team building: A multidimensional analysis. Unpublished doctoral
dissertation, Hofstra University, Hempstead,New York.
Woodman, R. W. (1978). Effects of team developmentintervention: A field experiment.
Ph.D. dissertation, Purdue University.
Woodman,R. W., & Sherwood,J. J. (1980). The role of team developmentin organizational
effectiveness: A critical review. Psychological Bulletin, 88, 166-186.

Eduardo Salas, Ph.D., is a senior researchpsychologist at the Naval Air Warfare

Center Training SystemsDivision. His research interests include team training,
training effectiveness,and teamperformance.

Drew Rozell, Ph.D., is a personal and businesscoach with Evolution Coaching in

Syracuse,New York.His interestsinclude teambuilding, managementtraining, and
.-.executive success. ...

.,
 Salas et al. IlEAM BUll..DING 329

James E. Driskell, Ph.D., is president of Florida Maxima Corporation and an

adjunct professor of psychology at Rollins College, Winter Park, FL His research
interestsinclude statusprocesses,group dynamics,and performanceunder stress.

Brian Mullen, Ph.D., is a professor in the Department of Psychology at Syracuse

University. His research interests include social cognition perspectiveson (inter)
group phenomenaand meta-analysis.