Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
THE ROLE OF CONTEXT IN WORK TEAM DIVERSITYRESEARCH: A META-ANALYTIC REVIEW
APARNA JOSHIHYUNTAK ROH
University of Illinois at Urbana-Champaign
Integrating macro and micro theoretical perspectives, we conducted a meta-analysisexamining the role of contextual factors in team diversity research. Using data from8,757 teams in 39 studies conducted in organizational settings, we examined whethercontextual factors at multiple levels, including industry, occupation, and team, influ-enced the performance outcomes of relations-oriented and task-oriented diversity. Thedirect effects were very small yet significant, and after we accounted for industry,occupation, and team-level contextual moderators, they doubled or tripled in size.Further, occupation- and industry-level moderators explained significant variance ineffect sizes across studies.
Research in the area of work team diversity hasgrown exponentially in the last four decades. How-ever, several comprehensive reviews have notedthat the findings in this area do not provide a clearconsensus regarding the performance effects ofwork team diversity (Harrison & Klein, 2007; Jack-son, Joshi, & Erhardt, 2003; Milliken & Martins,1996; Van Knippenberg & Schippers, 2007; Wil-liams & O’Reilly, 1998). In some studies, research-ers have reported that team diversity is positivelyassociated with performance (e.g., Ely, 2004; Vander Vegt, Van de Vliert, & Huang, 2005). In anotherset of studies, team diversity has been found tonegatively predict performance (e.g., Jehn, North-craft, & Neale, 1999; Leonard, Levine, & Joshi,2004). A majority of these studies, however, havereported a nonsignificant, direct relationship be-tween team diversity and performance. Further-more, even within studies, the effects of gender,race, age, and tenure diversity on team performancehave varied (e.g., Kirkman, Tesluk, & Rosen, 2004;Kochan et al., 2003).
Current theoretical perspectives framing diver-sity research, such as social identity theory, socialcategorization theory, and the attraction-selection-attrition framework, appear to be insufficient forresolving these mixed findings. In general, current
applications of these theoretical perspectives haveoffered broad generalizations for why differenceswithin work groups may manifest in specific atti-tudinal outcomes, such as conflict or cohesion, orin behavioral outcomes, such as turnover or absen-teeism (see Jackson et al., 2003). In addition toasking why diversity manifests in specific out-comes, a careful examination of situational settingswould also ask when, where, and how diversity dy-namics unfold in workplaces; these contextual con-siderations, not often captured (see Johns, 2006), arepertinent for reconciling the mixed findings from pastresearch. Researchers have attempted to reconcilethese findings within prevalent theoretical traditionsby considering the influence of organization-leveland team-level factors on diversity outcomes (e.g.,Kirkman et al., 2004; Kochan et al., 2003). However,these explanations are often offered post hoc; contex-tual factors have less often been incorporated in hy-pothesis development or in study design. We proposethat a unified and comprehensive contextual frame-work for diversity research has the potential to re-solve these inconsistent findings and can contributeto further theoretical and empirical developments inthe field.
This article highlights contextual issues in teamdiversity research. Our study takes a substantivelydifferent approach from past meta-analytic reviewson this topic (Bowers, Pharmer, & Salas, 2000; Hor-witz & Horwitz, 2007; Webber & Donahue, 2001).Rather than test whether diversity attributes have apositive or a negative effect on team performance,we accounted for aspects of context at multiplelevels and examined whether these contextual fac-tors shaped the team diversity-performance rela-tionship. In proposing and testing the contextual
A version of this paper was presented at the 2008 AnnualMeeting of the Academy of Management and won the Do-rothy Harlow Distinguished Paper Award. We are indebtedto Joseph Martocchio for advice and feedback on drafts ofthis work. We are also extremely grateful to Jason Colquittand three anonymous reviewers for invaluable feedbackthroughout the review process. We thank Niti Pandey andErik Young for research assistance.
� Academy of Management Journal2009, Vol. 52, No. 3, 599–627.
599
Copyright of the Academy of Management, all rights reserved. Contents may not be copied, emailed, posted to a listserv, or otherwise transmitted without the copyright holder’s expresswritten permission. Users may print, download or email articles for individual use only.
framework presented in this article, we attempt tomake several contributions to work team diversityresearch. First, we develop and test a theoreticallydriven framework for work team diversity contextacross multiple levels of analysis. Several research-ers have acknowledged that contextual consider-ations are critical in diversity research (e.g., Jack-son et al., 2003; Martins, Milliken, Wiesenfeld, &Salgado, 2003). However, a theoretically drivenmultilevel framework explicating contextual deter-minants of work team diversity outcomes has notbeen forthcoming. Our study addresses this gap.Some scholars have also drawn attention to struc-tural and institutional factors that give meaning todemographic differences in organizations (DiTo-maso, Post, & Parks-Yancy, 2007; Ragins & Sund-storm, 1990). Yet the prevalent theoretical perspec-tives we referred to in our introduction do notaccount for these institutional and structural fac-tors. Therefore, a second contribution we attempt isan integration of work team diversity research withmacro theoretical perspectives that account forthese factors and have received relatively scant at-tention in the past. Furthermore, team task charac-teristics can also shape the salience of diversityattributes within teams (Van Knippenberg & Schip-pers, 2007); we also consider whether these charac-teristics are a relevant team-level context influenc-ing the diversity-performance link. Third, weintegrate 15 years of field research on the perfor-mance outcomes of work team diversity. Our re-view distinguishes between the effects of task-ori-ented (e.g., function, education, and tenure) andrelations-oriented (e.g., gender, race/ethnicity, andage) aspects of diversity in relation to performance.Finally, this study answers a call in the broaderdomain of management research for a more contextbased understanding of workplace phenomena(Bamberger, 2008; Johns, 2006; Rousseau & Fried,2001). Context theorizing in management researchcan enhance its “market orientation,” which notonly makes study findings more accessible to man-agers but is also important for future theory build-ing (Bamberger, 2008; Dubin, 1976).
In the following section, we discuss in detail howwe conceptualize diversity context and outline var-ious aspects of this context. Next, we develop hy-potheses that delineate the moderating effects ofthese contextual factors on team diversity in rela-tion to team performance. Finally, we present thefindings from a meta-analytic review and considerthe implications of these contextual considerationsfor future theoretical and empirical developmentsin diversity research.
KEY CONCEPTS ANDTHEORETICAL BACKGROUND
We define diversity as an aggregate team-levelconstruct that represents differences among mem-bers of an interdependent work group with respectto a specific personal attribute (Jackson et al.,2003). In line with past research, we distinguishbetween task-oriented and relations-oriented as-pects of diversity (Jackson, May, & Whitney, 1995).Relations-oriented diversity attributes such as gen-der, race/ethnicity, and age are cognitively accessi-ble, pervasive, and immutable and are associatedwith social categorization processes (Fiske, 1998;Van Knippenberg, De Dreu, & Homan, 2004). Thesesocial categorization–based processes, which man-ifest in intergroup bias and negative attitudestoward dissimilar others in a group may have neg-ative performance consequences. In contrast, task-oriented diversity attributes, such as education,function, and tenure, are associated with skill-basedand informational differences among work groupmembers (Jackson et al., 1995). These aspects of di-versity are assumed to constitute a team’s cognitiveresource base and are associated with elaboration-based processes, defined as the exchange of in-formation and perspectives among group mem-bers, individual-level information processing, gainingfeedback, and integrating information and per-spectives. These elaboration-based processes ex-plain the positive performance outcomes of workgroup diversity (see Van Knippenberg et al., 2004).In the subsequent sections, we discuss how va-rious aspects of diversity context can influence thecategorization-based processes associated with re-lations-oriented diversity and the elaboration-based processes associated with task-oriented di-versity, and implications for team performance.Although more sophisticated conceptualizations ofteam diversity have been developed (e.g., Harrison& Klein, 2007; Lau & Murnighan, 1998), we draw ona more simplified yet established typology of diver-sity attributes to develop our key propositions re-garding the role of contextual factors in diversityresearch.1
Team performance is defined as the extent to
1 Some researchers have also conceptualized team di-versity by differentiating between surface- and deep-level diversity (e.g., Harrison, Price, & Bell, 1998). Otherresearchers have proposed an approach based on “fault-lines” (e.g., Lau & Murnighan, 1998). Recently, Harrisonand Klein (2007) posed “separation,” “variety,” and “dis-parity” as diversity dimensions to consider. We discussthese alternative approaches later, in the Discussionsection.
600 JuneAcademy of Management Journal
which a team accomplishes its goal or mission (De-vine & Phillips, 2001). We focus on performance asan outcome of diversity because this outcome hasreceived the most research attention and representsan area in which the mixed findings have been themost prevalent (Jackson et al., 2003). Results pertain-ing to performance outcomes of diversity may needthe most integration, and understanding the contex-tual factors shaping performance outcomes may havethe most value for resolving past findings.
Context has been defined as the situational set-ting in which workplace phenomena occur (Cap-pelli & Sherer, 1991). In recent theoretical advance-ments, it has been recognized that various aspectsof context may serve as “situational opportunitiesfor and countervailing constraints against organiza-tional behavior [and] be represented as a tensionsystem or force field comprising such opportunitiesand constraints” (Johns, 2006: 387). Drawing onthis perspective, we propose that context can setspecific constraints and opportunities that eitherenhance or minimize the direct effects of workteam diversity on performance. Opening the dis-course on team diversity to contextual influences ischallenging because it requires one to specify team,organizational, and extraorganizational factors thatmay comprise a “tension system” shaping diversityeffects. Therefore, we undertook an extensive re-view of team diversity research conducted in or-ganizational settings from 1992 through 2008 toidentify aspects of diversity context that past re-search has explicitly acknowledged as either mod-erator or as control variables. Table 1 represents thefindings of this review.
Our review indicated that approximately 60 per-cent of the direct effects reported in past researchwere nonsignificant for various diversity attributes.Among the remainder, 20 percent of the effectsreported were significantly positive, and 20 percentwere significantly negative. Researchers have con-sidered contextual variables primarily at the teamlevel to explicate these mixed effects of team diver-sity on performance. Among the studies we reviewed,over 70 percent accounted for team-level contextualfactors such as task interdependence, complexity, cli-mate, and other team-level perceptual variables (e.g.,Jehn et al., 1999; Pelled, Eisenhardt, & Xin, 1999;Schippers, Den Hartog, Koopman, & Wienk, 2003;Van der Vegt & Bunderson, 2005). Approximately 20percent of the studies represented in Table 1 exam-ined contextual moderators at the organizationallevel, including organizational demography, diver-sity training participation, and organizational culture(e.g., Ely, 2004; Jackson & Joshi, 2004; Jehn &Bezrukova, 2004). Less than 10 percent of the studiesreviewed examined extraorganizational factors.
Among the studies that did incorporate these vari-ables, the focus was on national culture, customerbase demography, market competition, and rate oftechnological change (e.g., Ancona & Caldwell, 1992;Lovelace, Shapiro, & Weingart, 2001; Leonard et al.,2004; Reagans & Zuckerman, 2001; Van der Vegt etal., 2005). On the basis of this review, we aimed totake stock of contextual effects that are conceptuallydistinct and have been considered in past researchand also to incorporate additional contextual vari-ables that have received less attention in the past.
Theoretical and practical considerations gov-erned our efforts to identify key contextual factors.Sociopsychological theoretical perspectives sug-gest that the demography of the job or occupationin which diverse work groups are embedded canshape categorization-based processes in thesegroups (DiTomaso, Post, & Parks-Yancy, 2007; Lar-key, 1996; Reskin, McBrier, & Kmec, 1999).2 Unlikein a balanced setting, in an occupational contextdominated by a single demographic group, negativestereotypes about underrepresented groups can in-fluence categorization-based outcomes withinwork groups (Hilton & Von Hippel, 1996; Larkey,1996). We extend these perspectives to focus onoccupational demography as a contextual factorthat can enhance or minimize the influence ofteam diversity on performance outcomes.
We also consider industry as an embedding con-text for diversity-based outcomes. Strategic man-agement research suggests that interindustry varia-tion in levels of technological change, regulatorypressure, customer demands, and market competi-tion are greater than intraindustry variations (Bour-geois, 1985; Porter, 1980). Furthermore, these in-dustry-level contingencies can serve as situationalenhancers or minimizers of diversity effects on per-formance (Hambrick, Cho, & Chen, 1996; Richard,Murthi, & Ismail, 2007). Within the strategic man-agement domain, considerable research on topmanagement teams and research on the firm diver-sity–performance link has incorporated industry-level context (e.g., service versus manufacturing) asa key moderator (e.g., Haleblian & Finkelstein,1993; Hambrick et al., 1996; Keck, 1997; Richard etal., 2007). We extend these perspectives to considerindustry setting as a relevant aspect of a team’s
2 We acknowledge that the proximal work context rep-resented by organization-level demography, culture, andclimate has received some attention in past research andthat these are also important contextual variables to con-sider. However, of the studies included in this review, onlythree provided information regarding organization-level de-mography, culture, or climate, and therefore these variablescould not be included in the meta-analysis.
2009 601Joshi and Roh
TA
BL
E1
Su
mm
ary
ofT
eam
Div
ersi
tyR
esea
rch
,19
92–2
008a
Stu
die
sD
iver
sity
Att
ribu
tes
Con
sid
ered
Sam
ple
and
Fin
din
gs
Con
text
Var
iabl
esC
onsi
der
edat
Dif
fere
nt
Lev
els
Tea
mO
rgan
izat
ion
Ind
ust
ry/O
ccu
pat
ion
Oth
er
Ain
oya
(200
4)b
Rac
e,te
nu
re59
wor
kte
ams
from
vari
ous
ind
ust
ries
.N
od
irec
tre
lati
onsh
ips,
and
lim
ited
sup
por
tfo
rm
oder
atin
gef
fect
s.
Tas
kin
terd
epen
den
ce,
con
flic
tm
anag
emen
tp
ract
ice
An
con
a&
Cal
dw
ell
(199
2)F
un
ctio
n,
ten
ure
45n
ewp
rod
uct
dev
elop
men
tte
ams
inh
igh
-tec
hco
mp
anie
s.P
arti
alsu
pp
ort
for
dir
ect
rela
tion
ship
s.
Res
ourc
eav
aila
bili
tyd
Com
pan
yca
pab
ilit
y(p
rod
uct
dev
elop
men
tex
per
ien
ce)d
Mar
ket
com
pet
itio
nd
Bal
kun
di,
Kil
du
ff,
Bar
snes
s,&
Mic
hae
l(2
007)
Eth
nic
ity,
gen
der
,ag
e19
pro
du
ctio
nan
dm
ain
ten
ance
team
sat
aw
ood
pro
du
cts
com
pan
y.N
od
irec
tre
lati
onsh
ips.
Bau
gh&
Gra
en(1
997)
Gen
der
,ra
ce31
pro
fess
ion
alp
roje
ctte
ams
ina
gove
rnm
ent
agen
cy.
No
sign
ific
ant
mai
nef
fect
s.
Bu
nd
erso
n(2
003)
Sp
ecif
icst
atu
scu
es(c
omp
any
and
ind
ust
ryte
nu
re,
edu
cati
on),
dif
fuse
stat
us
cues
(gen
der
,et
hn
icit
y)c
35p
rod
uct
ion
team
sin
ah
igh
-tec
hco
mp
any.
No
dir
ect
rela
tion
ship
sbu
tsu
pp
ort
for
mod
erat
ing
effe
cts.
Tea
mp
roce
ss(i
nfl
uen
cece
ntr
alit
y),
team
ten
ure
,le
ader
role
d
Cad
y&
Val
enti
ne
(199
9)R
ace,
gen
der
,ag
e,fu
nct
ion
50p
robl
emso
lvin
gte
ams
ina
hig
hte
chco
mp
any.
Par
tial
sup
por
tfo
rd
irec
tre
lati
onsh
ips.
Cam
pio
n,
Med
sker
,&
Hig
gs(1
993)
Exp
erie
nce
80ad
min
istr
ativ
ew
ork
team
sin
afi
nan
cial
serv
ice
com
pan
y.L
imit
edsu
pp
ort
for
mai
nef
fect
s.
Tra
inin
g,m
anag
eria
lsu
pp
ort,
com
mu
nic
atio
nan
dco
oper
atio
nbe
twee
ngr
oup
sd
Cam
pio
n,
Pap
per
,&
Med
sker
(199
6)E
xper
ien
ce60
pro
fess
ion
alte
ams
ina
fin
anci
alse
rvic
eco
mp
any.
Lim
ited
sup
por
tfo
rm
ain
effe
cts.
Tra
inin
g,m
anag
eria
lsu
pp
ort,
com
mu
nic
atio
nan
dco
oper
atio
nbe
twee
ngr
oup
sd
Ch
atm
an&
Fly
nn
(200
1)R
ace,
gen
der
,ci
tize
nsh
ip16
1m
anag
ers
from
10bu
sin
ess
un
its
ofa
fin
anci
alse
rvic
efi
rm.
Par
tial
sup
por
tfo
rd
irec
tre
lati
onsh
ips.
Ch
oi(2
007)
Age
,ge
nd
er,
ten
ure
,fu
nct
ion
188
wor
kte
ams
ina
larg
eel
ectr
onic
sco
mp
any
inK
orea
.P
arti
alsu
pp
ort
for
dir
ect
rela
tion
ship
s.
Ch
oi,
Pri
ce,
&V
inok
ur
(200
3)A
ge,
gen
der
,ra
ce,
edu
cati
on16
9tr
ain
ing
grou
ps
from
3or
gan
iza-
tion
s.P
arti
alsu
pp
ort
for
mai
nef
fect
s.
Col
quit
t,N
oe,
&Ja
ckso
n(2
002)
Eth
nic
ity,
gen
der
,ag
e88
pro
du
ctio
nte
ams.
No
dir
ect
rela
tion
ship
and
lim
ited
sup
por
tfo
rm
oder
atin
gef
fect
.
Pro
ced
ura
lju
stic
ecl
imat
e
Dra
ch-Z
ahav
y&
Som
ech
(200
2)T
ask-
rela
ted
(fu
nct
ion
,ed
uca
tion
,te
amte
nu
re),
rela
tion
s-or
ien
ted
(gen
der
,ag
e)
48ad
min
istr
ativ
ete
ams
inm
idd
le/
hig
hsc
hoo
lsin
Isra
el.
Par
tial
sup
por
tsfo
rd
irec
tre
lati
onsh
ips.
Tea
mp
roce
ss(f
requ
ency
ofm
eeti
ngs
)d
Ely
(200
4)A
ge,
gen
der
,ra
ce,
ten
ure
486
reta
ilba
nk
bran
ches
.P
arti
alsu
pp
ort
for
mai
nan
dm
oder
atin
gef
fect
s.
Tea
mp
roce
ss(c
oop
erat
ive
team
pro
cess
)
Div
ersi
tytr
ain
ing
pro
gram
Her
ron
(199
3)b
Gen
der
,et
hn
icit
y93
wor
kte
ams
ina
larg
eae
rosp
ace
com
pan
y.M
ixed
sup
por
tfo
rm
ain
effe
ctre
lati
onsh
ips.
Jack
son
&Jo
shi
(200
4)G
end
er,
eth
nic
ity,
team
ten
ure
365
sale
ste
ams
ina
larg
eF
ortu
ne
500
com
pan
y.N
od
irec
tre
lati
onsh
ips
but
par
tial
sup
por
tfo
rm
oder
atin
gef
fect
s.
Mu
ltid
imen
sion
alte
amd
iver
sity
(in
tera
ctio
ns
amon
gat
trib
ute
s)
Wor
ku
nit
dem
ogra
ph
y,m
anag
erd
emog
rap
hy
Jarr
ell
(200
2)b
Fu
nct
ion
,ed
uca
tion
leve
l11
0in
terd
isci
pli
nar
yh
ealt
hca
rete
ams.
Tas
kin
terd
epen
den
ce,
task
com
ple
xity
Sig
nif
ican
tsu
pp
ort
for
mai
nan
dm
oder
atin
gef
fect
s.
Jeh
n&
Bez
ruko
va(2
004)
Gen
der
,ra
ce,
age,
team
ten
ure
,fu
nct
ion
,ed
uca
tion
alle
vel
1,52
8te
ams
ina
larg
ein
form
atio
np
roce
ssin
gfi
rm.
Par
tial
sup
por
tsfo
rm
ain
and
mod
erat
ing
effe
cts.
Org
aniz
atio
ncu
ltu
re,
busi
nes
sst
rate
gy,
HR
pra
ctic
eor
ien
tati
on
Jeh
n,
Nor
thcr
aft,
&N
eale
(199
9)S
ocia
lca
tego
ry(a
ge,
gen
der
),in
form
atio
n(e
du
cati
on,
fun
ctio
n)
92w
ork
team
sin
the
hou
seh
old
good
sm
ovin
gin
du
stry
.S
up
por
tsfo
rm
ain
and
mod
erat
ing
effe
cts.
Tas
kin
terd
epen
den
ce,
task
typ
e(c
omp
lexi
ty)
Josh
i(2
002)
bA
ge,
race
,ge
nd
er,
ten
ure
,ed
uca
tion
772
serv
ice
team
sin
am
anu
fact
uri
ng
com
pan
y.O
rgan
izat
ion
ald
emog
rap
hy
Par
tial
sup
por
tfo
rm
ain
and
mod
erat
ing
rela
tion
ship
s.
Kea
rney
&G
eber
t(2
009)
Age
,ed
uca
tion
,n
atio
nal
ity
62R
&D
team
sin
aG
erm
anp
har
mac
euti
cal
com
pan
y.N
od
irec
tre
lati
onsh
ipbu
tsi
gnif
ican
tm
oder
atin
gef
fect
s.
Tra
nsf
orm
atio
nal
lead
ersh
ip,
team
lon
gevi
ty,d
task
inte
rdep
end
ence
d
Kea
rney
,G
eber
t,&
Voe
lpel
(200
9)A
ge,
edu
cati
on,
gen
der
,c
nat
ion
alit
y,c
ten
ure
c83
team
sfr
omG
erm
anor
gan
iza-
tion
s.S
up
por
tfo
rsi
gnif
ican
tm
oder
atin
gef
fect
s.
Tea
mm
embe
rs’
nee
dfo
rco
gnit
ion
,ta
skin
terd
epen
den
ce,d
task
com
ple
xity
,d
team
lon
gevi
tyd
Kel
ler
(200
1)F
un
ctio
n93
R&
Dte
ams.
Sig
nif
ican
tm
ain
effe
cts.
Kir
kman
,T
eslu
k,&
Ros
en(2
004)
Age
,ge
nd
er,
race
,co
mp
any
ten
ure
,te
amte
nu
re11
1w
ork
team
sfr
omfo
ur
dif
fere
nt
orga
niz
atio
ns.
Few
mai
nef
fect
san
dp
arti
alsu
pp
ort
for
mod
erat
ing
effe
cts.
Lea
der
-mem
ber
dem
ogra
ph
icm
atch
Org
aniz
atio
nal
wor
ken
viro
nm
entd
Con
tin
ued
onfo
llow
ing
pag
e
TA
BL
E1
Con
tin
ued
Stu
die
sD
iver
sity
Att
ribu
tes
Con
sid
ered
Sam
ple
and
Fin
din
gs
Con
text
Var
iabl
esC
onsi
der
edat
Dif
fere
nt
Lev
els
Tea
mO
rgan
izat
ion
Ind
ust
ry/O
ccu
pat
ion
Oth
er
Leo
nar
d,
Lev
ine,
&Jo
shi
(200
4)R
ace/
eth
nic
ity,
gen
der
,ag
e70
0re
tail
stor
es.
Lim
ited
sup
por
tfo
rm
ain
effe
cts
and
no
sup
por
tfo
rm
oder
atin
gef
fect
s.
Com
mu
nit
yd
emog
rap
hy
(cu
stom
erd
emog
rap
hic
mat
ch)
Lov
elac
e,S
hap
iro,
&W
ein
gart
(200
1)F
un
ctio
n43
new
pro
du
ctd
evel
opm
ent
team
sin
ah
igh
tech
ind
ust
ry.
No
dir
ect
rela
tion
ship
but
sign
ific
ant
mod
erat
ing
effe
cts.
Tea
mp
roce
ss(c
omm
un
icat
ion
styl
e),
lead
eref
fect
iven
ess
Tec
hn
olog
ical
chan
ged
Mar
stel
ler
(200
3)b
Age
,ge
nd
er,
edu
cati
on,
ten
ure
40h
ealt
hca
rete
ams.
Par
tial
sup
por
tfo
rm
ain
effe
cts.
Org
aniz
atio
nty
ped
May
o,P
asto
r,&
Mei
nd
l(1
996)
Gen
der
,ra
ce,
age,
ten
ure
68w
ork
team
sfr
omva
riou
sin
du
stri
es.
Lea
der
dem
ogra
ph
yd
Sig
nif
ican
tm
ain
effe
cts.
Pel
led
(199
6)R
ace,
gen
der
,co
mp
any
ten
ure
42p
rod
uct
ion
team
sin
man
ufa
ctu
rin
gfa
cili
ties
.N
od
irec
tre
lati
onsh
ips.
Pel
led
,E
isen
har
dt,
&X
in(1
999)
Gen
der
,ra
ce,
age,
com
pan
yte
nu
re,
fun
ctio
n45
team
sin
elec
tron
icd
ivis
ion
s.N
od
irec
tre
lati
onsh
ips
but
sign
ific
ant
mod
erat
ing
effe
cts.
Tas
kro
uti
nen
ess,
grou
plo
nge
vity
Org
aniz
atio
nal
wor
ken
viro
nm
entd
Por
tero
-Bro
wn
(199
9)b
Age
,ge
nd
er,
eth
nic
ity,
fun
ctio
n,
com
pan
yte
nu
re37
R&
Dte
ams.
Par
tial
sup
por
tfo
rm
ain
and
mod
erat
ing
effe
cts.
Tea
mid
enti
fica
tion
,co
llec
tive
nor
m,
team
pro
cess
Pu
ck,
Ryg
l,&
Kit
tler
(200
6)E
thn
icit
y20
pro
cess
imp
rove
men
tte
ams
ina
Ger
man
spor
tsw
ear
com
pan
y.L
imit
edsu
pp
ort
for
the
dir
ect
rela
tion
ship
.
Rea
gan
s&
Zu
cker
man
(200
1)C
omp
any
ten
ure
c22
4co
rpor
ate
R&
Dte
ams.
No
dir
ect
rela
tion
ship
.T
eam
task
typ
edM
arke
tco
mp
etit
ion
d
Rea
gan
s,Z
uck
erm
an,
&M
cEvi
ly(2
004)
Ed
uca
tion
,ge
nd
er,
ten
ure
1,51
8p
roje
ctte
ams
ina
pro
fess
ion
alR
&D
firm
.N
od
irec
tre
lati
onsh
ips.
Ric
har
d(2
000)
Rac
e63
ban
ks.
No
dir
ect
rela
tion
ship
but
sign
ific
ant
mod
erat
ing
effe
cts.
Bu
sin
ess
stra
tegy
(gro
wth
vs.
dow
nsi
zin
g)
Ric
har
d,
Bar
net
t,D
wye
r,&
Ch
adw
ick
(200
4)
Rac
e,ge
nd
er15
3m
anag
emen
tte
ams
ina
ban
kin
gin
du
stry
.N
od
irec
tre
lati
onsh
ips
but
sign
ific
ant
mod
erat
ing
effe
cts.
Bu
sin
ess
stra
tegy
(in
nov
atio
n/r
isk
taki
ng)
Sac
co&
Sch
mit
t(2
005)
Rac
e,ge
nd
er,
age
2,37
3qu
ick
serv
ice
rest
aura
nts
.P
arti
alsu
pp
orts
for
mai
nef
fect
sbu
tn
om
oder
atin
gre
lati
onsh
ip.
Com
mu
nit
yd
emog
rap
hic
mat
ch
Sch
ipp
ers,
Den
Har
tog,
Koo
pm
an,
&W
ien
k(2
003)
Age
,ge
nd
er,
edu
cati
onle
vel,
team
ten
ure
54w
ork
team
sin
the
Net
her
lan
ds.
Few
mai
nef
fect
sbu
tsi
gnif
ican
tm
oder
atin
gef
fect
s.
Ou
tcom
ein
terd
epen
den
ce,
grou
plo
nge
vity
Som
ech
(200
6)F
un
ctio
n13
6p
rim
ary
hea
lth
care
team
sin
Isra
el.
Lea
der
ship
styl
e
No
dir
ect
rela
tion
ship
san
dp
arti
alsu
pp
ort
for
mod
erat
ing
effe
cts.
Tyr
an&
Gib
son
(200
8)G
end
er,
eth
nic
ity,
ten
ure
,co
llec
tivi
sm57
ban
kbr
anch
team
s.P
arti
alsu
pp
ort
for
mai
nef
fect
rela
tion
ship
s.
Tea
mou
tcom
ety
pe
Van
Der
Veg
t&
Bu
nd
erso
n(2
005)
Exp
erti
se(f
un
ctio
n)
57R
&D
team
sin
oil
and
gas
ind
ust
ryin
the
Net
her
lan
ds.
No
dir
ect
rela
tion
ship
but
sign
ific
ant
mod
erat
ing
effe
cts.
Col
lect
ive
team
iden
tifi
cati
on,
team
pro
cess
(tas
kan
dre
lati
onsh
ipco
nfl
ict,
com
mu
nic
atio
n)d
Van
Der
Veg
t,V
anD
eV
lier
t,&
Hu
ang
(200
5)
Age
,ge
nd
er,
ten
ure
,fu
nct
ion
270
man
ufa
ctu
rin
gu
nit
sof
am
ult
inat
ion
alco
mp
any.
Par
tial
sup
por
tfo
rm
ain
rela
tion
ship
san
dsi
gnif
ican
tsu
pp
ort
for
mod
erat
ing
effe
cts.
Job
typ
e,jo
bch
alle
nge
dN
atio
nal
pow
erd
ista
nce
(cu
ltu
re)
Yeh
&C
hou
(200
5)F
un
ctio
n,
pos
itio
n88
cros
s-fu
nct
ion
alte
ams.
Su
pp
ort
for
mai
nef
fect
sbu
tn
osu
pp
ort
for
the
mod
erat
ing
effe
ct.
Tea
mle
arn
ing
beh
avio
rs
aN
�43
,co
mp
risi
ng
fiel
dst
ud
ies
con
du
cted
du
rin
gth
e19
92–2
008
per
iod
.b
Un
pu
blis
hed
.c
Div
ersi
tyat
trib
ute
sco
nsi
der
edas
con
trol
vari
able
sin
the
rese
arch
des
ign
.d
Con
text
ual
fact
ors
incl
ud
edas
con
trol
vari
able
s.
diversity context. On a practical note, informationregarding occupational and industry setting wasavailable from most study descriptions. Addition-ally, publicly accessible databases provide a wealthof objective information regarding the occupationaland industry contexts in which teams are embed-ded. Therefore, we focus on industry setting andoccupational demography as contextual factors thatprovide a comprehensive understanding of a team’smacrolevel context and could potentially accountfor the mixed findings reported.
The nature of a team’s task can have a significantinfluence on the extent to which team members areinterdependent in terms of goals and task outcomes(see Ilgen, Hollenbeck, Johnson, & Jundt, 2005;LePine, Hanson, Borman, & Motowidlo, 2000). Asthe review above suggests, team-level contextualvariables have received the most attention in pastresearch. Social categorization theory would sug-gest that aspects of a team’s task can minimize thesalience of diversity attributes by reinforcing acommon group identity or by placing demands onthe team’s diverse cognitive resource base (Gaert-ner & Dovidio, 2000; Jehn et al., 1999). Informationprocessing theory would suggest that the nature oftasks would place requirements on a team’s cogni-tive resource base with implications for the sa-lience of diversity attributes within the team (Jehnet al., 1999; Williams & O’Reilly, 1998). Our reviewalso indicated that information regarding aspects ofteams’ tasks, such as team interdependence andteam type (i.e., short-term versus long-term), wasmore consistently available across studies. There-fore, we chose to focus on task interdependenceand team type as aspects of team-level context.Ultimately, we sought to examine the effects ofconceptually distinct contextual variables that to-gether provide a comprehensive picture of contextat the team level. In the subsequent sections, wediscuss in greater detail how these aspects of occu-pational, industry, and team-level context mayshape the performance outcomes of both task-ori-ented and relations-oriented team diversity.
A CONTEXTUAL FRAMEWORK FOR TEAMDIVERSITY RESEARCH
Occupational Demography
In the United States, workers are classified intooccupational categories based on work performed,skills, education, training, and credentials. Someoccupations are found in just one or two industries(e.g., post office clerks); however, many occupa-tions are found in a large number of industries (e.g.,auditors, accountants, software developers, net-
work analysts). Thus, the term “occupation” refersto this collective description of a number of indi-vidual jobs performed, with minor variations, inmany establishments (U. S. Bureau of Labor Statis-tics, 2007). A significant body of sociological re-search shows that occupation-level demographiccomposition can have important implications forgender-based and ethnicity- or race-based diversityin organizations (for a review, see Reskin et al.[1999]). We propose below that occupational de-mography serves as a situational setting that canenhance the effects of relations-oriented diversityand minimize the effects of task-oriented diversityon team performance.
Considerable research on stereotype formationenables us to specify the psychological processesby which an occupational demography can influ-ence diversity dynamics within teams. This re-search suggests that stereotypes, which emerge inresponse to environmental factors—such as differ-ences in social roles (Eagly, 1995) and in power(Fiske, 1993)—and as a means to justify the statusquo (Jost & Banaji, 1994; Sidanius, 1993), result incategorization-based responses toward targetedgroups (Allport, 1954). When a single demographicgroup dominates an occupation, negative stereo-type-based categorization processes against under-represented groups are likely (Fiske, 1993; Reskinet al., 1999). Once categorization-based processesare invoked, additional information regarding tar-geted group members is filtered out, and individu-ating processes are blocked (Allport, 1954; Brewer,1988). In diverse teams, when the environmentprimes negative stereotypes against specific demo-graphic groups, these categorization-based pro-cesses are likely to influence interactions. In workenvironments, where stereotypes are less salient,individuating information regarding demographi-cally dissimilar individuals is likely to be cogni-tively acceptable and result in more positive inter-actions (see Larkey, 1996). Integrating theseperspectives, we propose that in occupational set-tings dominated by a single demographic group, itis likely that stereotypic reactions against under-represented groups will be triggered (Fiske, 1993;Reskin et al., 1999). These reactions result in cate-gorization-based processes that hinder effectivegroup interactions, with negative performance con-sequences (Larkey, 1996). These categorization-based processes are less likely in demographicallybalanced occupational settings.
A rich body of research on the age, race, andgender typing of jobs corroborates our argumentthat the demographic attributes of a job category oroccupation are associated with job stereotypes thatmay trigger social categorization processes based
606 JuneAcademy of Management Journal
on these demographic attributes (Eagly & Steffen,1984; Perry, 1997). For example, consider the male-dominated occupational category of production en-gineers. On the basis of the research reviewedabove, we would surmise that female engineers (anunderrepresented group) are likely to be targets ofnegative stereotypes (e.g., “women are less techni-cally competent”) enhancing the salience of genderas a basis for categorization when men and womenwork together in a production team (Eagly & Stef-fen, 1984; Ely, 1994, 1995). We propose that genderdiversity is likely to have negative performanceeffects in this setting. In a more gender balancedoccupational setting, such as postsecondary sci-ence teaching, on the other hand, this researchsuggests that gender may not emerge as a salientbasis for categorization within a team of teachersworking on organizing a school science fair. In thissetting, gender diversity is less likely to have anegative influence on team performance. A similarargument can be made with regard to the effects ofoccupational race/ethnic composition on the per-formance outcomes of race/ethnicity-based diver-sity in teams (Reskin et al., 1999).
Regarding the effects of occupational age compo-sition, a significant body of research suggests thatnegative stereotypes against older workers arefairly prevalent and can have detrimental conse-quences for these workers (Fiske, Cuddy, Glick, &Xu, 2002; Rosen & Jerdee, 1977; see Shore andGoldberg [2005] for a recent review). Althougholder workers may experience more unfavorablejob outcomes overall, research also indicates thatthe age composition of a job context can influenceperformance ratings and advancement potential forolder employees. This research suggests that olderworkers may face more unfavorable outcomes inoccupations with fewer older workers (Cleveland,Festa, & Montgomery, 1988; Cleveland, Montgom-ery, & Festa, 1984). In view of this research, wewould propose that among, for example, softwareprogrammers (an occupation dominated byyounger workers), older programmers may be tar-gets of negative stereotypes (e.g., viewed as lessskilled or motivated); the implications of age diver-sity–based outcomes would thus be negative whenolder and younger team members collaborate on ajoint software development project. In work set-tings with a higher proportion of older workers (forinstance, among welfare service workers or postoffice clerks), these workers are less likely to facestereotypes and discrimination (Shore & Goldberg,2005). In these settings, categorization-based effectsbased on age are less likely to disrupt groups’functioning.
To date, diversity research has not addressed the
role of occupational demography in shaping perfor-mance outcomes within diverse teams. At the firmlevel, Frink et al. (2003) found that an invertedU-shaped relationship between gender composi-tion and performance was only observed in gender-balanced occupational settings and not in male-dominated settings. The authors noted that thesefindings may reflect the inability of organizationsin male-dominated contexts to capitalize on thebenefits of gender diversity (Frink et al., 2003).Although the researchers had not accounted forthis contextual factor in their theoretical frame-work, hypotheses, or study design, they discussedthe importance of occupational demography in ex-plaining their findings.
Cumulatively, the research perspectives dis-cussed above suggest that in occupational settingsdominated by a single demographic group, diverseteams may face performance losses primarily fortwo reasons. First, these teams may perform subop-timally because the work context enhances stereo-typing and bias against underrepresented demo-graphic groups that triggers social categorizationbased on these attributes within the teams (Di-Tomaso, Post, & Parks-Yancy, 2007; Skaggs & Di-Tomaso, 2004; Van Knippenberg et al., 2004). Sec-ond, in these settings, teams with higherproportions of underrepresented group members(e.g., women or ethnic minorities) may be valuedless and receive poorer performance ratings or ac-cess to resources, which is likely to impact subjec-tive or objective performance outcomes (Baugh &Graen, 1997; Hultin & Szulkin, 1999; Joshi, Liao, &Jackson, 2006). In view of these considerations,we proposed that occupational demography willmoderate the relationship between relations-ori-ented diversity and team performance and testedspecific aspects of that general relationship meta-analytically with the following:
Hypothesis 1a. Occupational gender composi-tion moderates the negative effect of team gen-der diversity on performance. The negative ef-fect of gender diversity on performance isweaker in gender-balanced settings.
Hypothesis 1b. Occupational race/ethnic com-position moderates the negative effect of teamrace/ethnic diversity on performance. The neg-ative effect of race/ethnic diversity on perfor-mance is weaker in racially/ethnically bal-anced settings.
Hypothesis 1c. Occupational age compositionmoderates the negative effect of team age di-versity on performance. The negative effect of
2009 607Joshi and Roh
age diversity on performance is weaker in rel-atively age-balanced settings.
So far we have argued that occupational demog-raphy creates a context that may enhance or mini-mize categorization-based processes in workgroups. Our arguments raise the possibility thatoccupational demography will moderate the rela-tionship between task-oriented diversity and teamperformance as well. We surmise that when a sin-gle demographic group dominates an occupation,relations-oriented diversity may be correlated withtask-oriented diversity. For example, white menmay be more tenured and have a specific educa-tional background. In these settings, individuals’task-relevant contributions may be confoundedwith their demographic attributes (Berger, Fisek,Norman, & Zelditch, 1977) so that any positiveoutcomes of task-oriented diversity on performancemay also be mitigated. Researchers have also notedthat when the negative effects of categorization-based diversity are salient, information processingin groups is disrupted (Van Knippenberg et al.,2004). Some studies have shown that when rela-tions-oriented diversity is associated with greaterconflict, task-oriented diversity is also less likely tohave positive outcomes (Homan & Van Knippen-berg, 2003; Jehn et al., 1999). In view of these per-spectives, to account for the possibility that occu-pational demography may have implications forthe outcomes of task-oriented diversity, we proposethe following broad hypothesis:
Hypothesis 1d. Occupational demographymoderates the positive effect of task-orienteddiversity on team performance. The effects oftask-oriented diversity are stronger in morebalanced settings.
Industry Setting
Industry settings refer to fairly specific businessenvironments in which teams are nested that mayhave important implications for team diversity dy-namics that go over and above the occupationaleffects discussed above (Batt, 2002; Datta, Guthrie,& Wright, 2005). Porter’s (1980) analysis identifiesfactors such as customers, suppliers, and regulatorygroups that vary by industry and pose varying con-tingencies for firms. Contingency theory suggeststhat these industry factors provide firms with op-portunities as well as challenges for utilizing keyorganizational resources to enhance performance(Burns & Stalker, 1961; Drazin & Van de Ven, 1985;Lawrence & Lorsch, 1967; Porter, 1980). Drawingon this perspective, a significant body of strategicmanagement research has identified industry con-
text as a key contingency that influences the rela-tionship between organizational processes/prac-tices and performance outcomes (see Combs, Lieu,Hall, and Ketchen [2006] for a meta-analysis). In-dustry-level context has received some, albeit lim-ited, attention in past diversity research (e.g., An-cona & Caldwell, 1992; Lovelace et al., 2001;Reagans & Zuckerman, 2001). On the basis of stra-tegic management perspectives and in line with ourconceptualization of context, we consideredwhether contingencies associated with three spe-cific contexts—service, manufacturing, and hightechnology—may serve as situational enhancers orminimizers of diversity effects on performance out-comes. Together, these three industrial settings in-corporate a bulk of the research settings consideredin past diversity research.
Service industries, defined as customer-orientedindustries that require front-line customer interac-tion and engagement, include retail trade, hospital-ity, and education (U.S. Census Bureau, 2002).Relative to manufacturing industries, service in-dustries are characterized by more frequent andcloser interactions with customers and a greateremphasis on discretionary behavior that can man-ifest directly in performance outcomes such assales, customer satisfaction, and customer reten-tion (Datta et al., 2005). Researchers have arguedthat increasing demographic attribute–based diver-sity can enhance a firm’s “market competence,”which is a form of competitive advantage in theservice industry (Richard et al., 2007). Consider asan example that a retail store with a diverse groupof store employees is more likely to attract diversecustomers and thereby also more likely to havestrong store sales. On the other hand, a competingstore that is unable to enhance employee diversityto attract customers and increase market share islikely to perform poorly in comparison. On thebasis of these considerations, we would expect thataspects of service settings such as direct customercontact and higher levels of discretionary behaviorswould serve as situational enhancers of relations-oriented diversity effects on performance. In theservice industry, since these aspects of diversitycan be considered as a form of market competence,we propose that any negative effects of gender,race, and age diversity are likely to be reversed.
Manufacturing industries are based on the fabri-cation, processing, or preparation of products fromraw materials and commodities. These industriesare typically highly capital-intensive and includeautomobile manufacturing, chemical manufactur-ing, and paper and wood manufacturing (U. S. Cen-sus Bureau, 2002). In contrast to service industryfirms, manufacturing firms rely more on plant and
608 JuneAcademy of Management Journal
equipment, technology, and raw materials toachieve business goals (Quinn, Anderson, & Finkel-stein, 1996). Because manufacturing relies more onphysical capital and equipment and less on directcustomer-based interactions, we would expect di-versity attributes to be less likely to directly impactperformance in this setting (Richard et al., 2007).Further, Combs and colleagues (Combs, Hall, &Ketchen, 2006) noted that manufacturing indus-tries are more likely than service industries to im-plement total quality management techniques, reg-ular training, and formalized human resource (HR)practices because of the need to monitor employeesand to develop their knowledge, skill, and ability toutilize expensive and sometimes dangerous ma-chinery. Consider, as an example, the automobilemanufacturing industry, which has implemented anumber of total quality management techniques in-volving team-based interventions and undertakenregular training for employees. These attributes ofthe automobile industry may serve to buffer anydirect effects of team diversity. Thus, we surmisethat aspects of manufacturing settings, such as re-liance on machinery and HR practices that involvegreater monitoring of employee behavior, may actas situational minimizers of diversity effects onperformance outcomes.
High-technology industries follow invention andinnovation in business strategy and compete inglobal and short-cycle product-markets (cf. Milkov-ich, 1987). In contrast to service and manufacturingindustries, these industries rely more heavily onintellectual capital and invest significantly more inresearch and development (Organisation for Eco-nomic Cooperation and Development, 2006). In-dustries such as information technology, biomedi-cal technology, telecommunication, and dataservices are included in this category. Researchershave begun to examine this particular setting as adistinct context framing firm practices and out-comes (Collins & Smith, 2006). This research hassuggested that employers in this setting are alsolikely to adopt commitment-based practices aimedat recruiting and retaining highly skilled employ-ees, which can create a climate fostering knowledgeexchange and combination (Collins & Smith, 2006).These situational contingencies may enhance task-oriented diversity effects. Relative to service andmanufacturing settings, given contingencies suchas rapidly changing technology, reliance on intel-lectual capital, and a need for creativity and inno-vation in a dynamic environment, we would expecttask-oriented attributes that form a team’s cognitiveresource base to significantly impact performanceoutcomes in this setting.
Although a significant body of research on top
management teams has examined industrial envi-ronment as a relevant influence on the outcomes ofdiversity (e.g., Hambrick et al., 1996; Hambrick &Finkelstein, 1987; Keck, 1997), research on lower-level teams has rarely taken this contextual factorinto account. In general, top management team re-search suggests that industry attributes, such as rateof technological change, munificence, and environ-mental uncertainty, can enhance the salience ofdiversity in top management teams in relation tofirm performance (see Carpenter, Geletkanycz, andSanders [2004], and see Hambrick and Finkelstein[1987] for a review). Richard and colleagues (2007)examined whether industry (service versus manu-facturing) moderated the relationship between ra-cial diversity and firm performance in a sample ofover 800 large U.S. companies; in support of thepropositions, results indicated that the relationshipbetween racial diversity and firm performance wasstronger in service than in manufacturing firms(Richard et al., 2007).
On the basis of the theoretical and empiricalperspective detailed above, we would expect thethree distinct industrial settings to impose varyingcontingencies on diversity outcomes. In service set-tings, customer-based contingencies may enhancepositive performance outcomes of relations-orienteddiversity. In manufacturing settings, a reliance onphysical equipment and standardized practices tomonitor employee behavior may minimize any directperformance outcomes of diversity; and in high-technology settings, technological and intellectualcapital–based contingencies may enhance positiveeffects of task-oriented diversity. Thus, we proposethat industry setting moderates the relationship be-tween relations- and task-oriented diversity andteam performance. Specifically:
Hypothesis 2a. Relations-oriented diversity islikely to have a positive effect on performance inservice industries. In manufacturing and high-technology settings, relations-oriented diversityis less likely to have a significant effect onperformance.
Hypothesis 2b. The positive effect of task-orienteddiversity on performance is stronger in high-tech-nology industries than in manufacturing and ser-vice settings.
Team-Level Diversity Context
We propose below that specific team character-istics—team interdependence and team type—con-stitute a team-level diversity context that canenhance or minimize the direct effects of relations-
2009 609Joshi and Roh
oriented and task-oriented diversity on teamperformance.
Team interdependence. Teams display varyinglevels of interdependence based on their tasks,goals, and outcomes. Task interdependence is de-fined as the extent to which team members rely oneach other to complete their task (Shea & Guzzo,1987). In highly task interdependent teams, teammembers engage in both sequential and reciprocalexchanges to accomplish the team tasks; in less taskinterdependent teams, team members’ independentcontributions are aggregated to accomplish theteam tasks (Saavedra, Earley, & Van Dyne, 1993;Thompson, 1967; Van de Ven, Delbecq, & Koenig,1976). Teams may also vary on goal and outcomeinterdependence. Goal interdependence refers tothe extent to which a team as a whole has a collec-tive goal. Outcome interdependence refers to theextent to which team members are interdependentin terms of rewards and feedback. These aspects ofinterdependence tend to be highly correlated andtherefore researchers have suggested combiningthem into an overall team interdependence con-struct (Gully, Incalcaterra, Joshi, & Beaubien, 2002).
Social categorization theory would predict thathigher outcome and goal interdependence is likelyto unite team members to work toward a commongoal and motivate them to cast aside differences(Gaertner & Dovidio, 2000). Task interdependencemay facilitate intergroup contact conducive to re-ducing categorization-based processes in teams(Pettigrew, 1998). Team interdependence has re-ceived considerable attention as a moderating in-fluence in team diversity research. For example,Schippers and colleagues (2003) found that out-come interdependence reinforces common groupgoals that can counteract the negative effects ofdiversity, so that highly outcome-interdependentteams with high levels of diversity display moretask-related discussions and communication thanhighly diverse teams with low outcome interdepen-dence. Jehn and colleagues also reported that de-mographic diversity was positively associated withsatisfaction and commitment when task interde-pendence was high (Jehn et al., 1999). Other re-search has also supported this pattern of findings(e.g., Van der Vegt & Janssen, 2003). We wouldexpect that high interdependence creates a contextfor elaboration-based processes within a team andthus that the outcomes of task- and relations-oriented diversity may be more positive or lessnegative when the level of interdependence is high.Hence, we propose:
Hypothesis 3. Team interdependence moder-ates the relationship between task- and rela-
tions-oriented diversity and team performance.The positive effect of task-oriented diversity onteam performance is stronger among highlyinterdependent teams. The negative effect ofrelations-oriented diversity on team perfor-mance is weaker among highly interdependentteams.
Team type. The durability of a team’s member-ship—that is, whether the team has been assembledto accomplish a short-term goal or whether it isinstead a stable and permanent unit in an organi-zation—is likely to have consequences for interper-sonal interactions among diverse team members.Task-related contingencies are likely to differ inshort- and long-term teams. In short-term teams,greater urgency may surround goals and missions.On the other hand, in long-term teams, task require-ments may be more stable, and distribution of tasksand roles may also be more clearly defined (DeDreu & Weingart, 2003). Another dimension alongwhich these teams are likely to differ is the longev-ity of team membership. The members of short-term teams likely have shorter tenure, than themembers of long-term teams.3 Thus, temporal dy-namics are also likely to vary in these two types ofteams.
Some studies that have examined temporal influ-ences on team diversity outcomes have shown thatthe length of time team members spend togethermay diminish the salience of visible aspects of di-versity and enhance the salience of attitudinal orvalue-based aspects of diversity (Harrison et al.,1998; Harrison, Price, Gavin, & Florey, 2002).Schippers and colleagues (2003) found that over alonger duration, highly diverse teams were lesslikely to display elaboration-based processes. In theshort run, however, diverse teams engaged in moretask-relevant debates and discussions that had pos-itive consequences for team performance. The au-thors noted that in long-standing diverse teams,team members may attribute conflicts to relationaldifferences, and the motivation and willingness toresolve differences through greater communicationmay erode over time. In the short term, on the otherhand, the members of highly diverse teams aremore likely to communicate across differences toaccomplish the teams’ tasks (Schippers et al.,2003). Corroborating this finding, Watson, Johnson,and Merrit (1998) also found that demographic di-
3 Although team tenure or longevity could also be im-portant variables to consider, our review indicated thatthese variables were sparsely reported in extant researchand we were, therefore, unable to include these in thepresent study.
610 JuneAcademy of Management Journal
versity was negatively associated with outcomesover time. We propose, on the basis of this empir-ical research, that in short-term teams, urgency sur-rounding the accomplishment of the tasks may re-quire team members to overlook differences andaim at utilizing diversity based on task-relevantdimensions. Since short-term teams are also asso-ciated with shorter team tenure, team membersmay be less likely to attribute task-based differ-ences to deeper attitudinal or personality-based dif-ferences (Harrison et al., 1998; Schippers et al.,2003). In long-term teams, divisions based on di-versity attributes may become more entrenchedand self-reinforcing, so that conflicts and differ-ences based on relations-oriented attributes have amore debilitating impact on team performance(Schippers et al., 2003; Watson et al., 1998). In viewof these considerations, we meta-analytically testedthe following:
Hypothesis 4. Team type moderates the rela-tionship between task- and relations-orienteddiversity and team performance. The positiveeffect of task-oriented diversity is stronger inshort-term teams than in long-term teams. Thenegative effect of relations-oriented diversity isstronger in long-term teams than in short-termteams.
METHODS
Literature Search
We employed multiple search techniques toidentify prior empirical research that examined therelationship between work team diversity and per-formance. First, we searched the computerized da-tabases PsycINFO, ABI/Inform, and SocIndex usingkeywords such as “team/group diversity,” “team/group composition,” “team/group performance,”and “team/group effectiveness” as well as searchterms associated with specific team diversity at-tributes (e.g., “gender,” “race/ethnicity,” “age,”“tenure,” “education,” and “functional backgrounddiversity”). Second, the electronic search was sup-plemented by a manual search of 19 major journals,including Academy of Management Journal, Ad-ministrative Science Quarterly, Journal of AppliedPsychology, Personnel Psychology, OrganizationScience, and Journal of Organizational Behavior,and others considered the most highly cited jour-nals in the field of management (see Gomez-Mejia &Balkin, 1992). Third, we also consulted the refer-ence lists from previous reviews on this topic, in-cluding both narrative (Harrison & Klein, 2007;Jackson et al., 2003; Milliken & Martins, 1996; VanKnippenberg & Schippers, 2007; Williams &
O’Reilly, 1998) and quantitative (Bowers et al.,2000; Horwitz & Horwitz, 2007; Webber &Donahue, 2001) reviews. Finally, in an effort toidentify relevant unpublished studies, we searchedProQuest Digital Dissertations and conference pro-ceedings for the annual meetings of the Academy ofManagement and the Society of Industrial and Or-ganizational Psychology for the previous five years.Researchers in related areas were also contacted toobtain current and unpublished studies that mightfit our criteria for inclusion.
Since this study concerned detecting the moder-ating effects of several contextual variables embed-ded at multiple levels, we focused only on studiesthat had been conducted in field settings wherethese variables were likely to influence diversityoutcomes. We did not include studies that relied onstudent samples, were conducted in laboratory set-tings, or involved simulated tasks or tasks in artifi-cial environments. Although these types of studiesare of great value in theory development, for thepurposes of this research we were interested inidentifying contextual moderators influencing di-versity outcomes in naturally occurring, intactwork teams in business settings. Research suggeststhat these types of teams differ substantively fromlaboratory teams (e.g., McGrath, 1984). From aninitial set of 95 field studies conducted in organi-zational settings, we applied the following addi-tional criteria to select articles for our meta-analy-sis. First, although diversity research has oftenbeen conducted at multiple levels of analysis, werestricted the present meta-analysis to the teamlevel. For a study to be included in the meta-anal-ysis, both diversity and performance had to be mea-sured at the team level. Second, team diversityvariables had to include either relations-orienteddiversity attributes (race/ethnicity, gender, or age)or task-oriented diversity attributes (education,functional background, or organizational tenure)pertinent to our theoretical arguments. Third, wealso excluded studies that examined top manage-ment teams (n � 47) because outcomes for suchteams were often measured at the firm level (e.g., asorganizational financial performance), and topmanagement teams were generally considered asoperating under different dynamics than lower-level general work teams in organizations (e.g.,Hambrick & Mason, 1984; Webber & Donahue,2001). Furthermore, team effectiveness models alsosuggest that performance is a proximal outcome ofteam composition among general work teams; thisassumption may not hold among top managementteams (Hackman, 1987). Finally, a study had toreport sample sizes and an appropriate statistic(e.g., mean and standard deviation, chi-square, t, F)
2009 611Joshi and Roh
that allowed the computation of a correlation coef-ficient with the formula provided by Hunter andSchmidt (1990: 272). Based on these criteria, thefinal data set included 8,757 teams from 39 studiesconducted between 1992 and 2009, yielding a totalof 117 effect sizes. The data included in the finalanalyses represent approximately twice the num-ber of effect sizes and three times the number ofteams included in past published meta-analyses onthis topic (e.g., Webber & Donahue, 2001).
Coding of Studies
Our initial coding scheme was based on opera-tionalization of team characteristics (team interde-pendence and team type) and research setting(occupational and industry setting). Using thisscheme, each author and two additional raters in-dependently coded a random selection of five arti-cles; initial interrater agreement ranged between 75and 94 percent. To resolve disagreements, we wentback to the studies and reached consensus throughdiscussion. From this discussion, we developed adetailed set of decision rules and used them to codean additional seven articles. Interrater agreementwas almost 100 percent for these articles. The au-thors then coded the remaining articles using thedecision rules we had developed.
Measures
Team diversity. Drawing on past research (e.g.,Jackson et al., 1995; Webber & Donahue, 2001), weclassified diversity attributes into two categories:relations-oriented (race/ethnicity, gender, and age)and task-oriented (organizational tenure, educa-tion, and functional background). In the studiesincluded in the analyses, categorical diversity at-tributes were measured using Blau’s (1977) indexor Teachman’s (1980) entropy measure; Allison’s(1978) coefficient of variation was used for contin-uous variables such as age and tenure. Becausemeasures based on “faultlines” (hypothetical linessplitting a group into attribute-based subgroups[Lau & Murnighan, 1998: 328; 2005: 645]) wereused less frequently and in some cases combineddemographic and task-oriented attributes, thesewere not included in the present analysis. In gen-eral, the operationalization of relations- and task-oriented diversity in this study is theoreticallydriven and consistent with past research. Our ini-tial review also indicated that the diversity at-tributes included in this meta-analysis were themost commonly studied variables in past research.However, since the measurement of diversity at-tributes in studies involved in the sample predates
recent reconceptualizations (e.g., Harrison & Klein,2007), we were unable to incorporate these newerdirections in this study.
Team performance. Measures of team perfor-mance included financial and operational mea-sures (e.g., sales, productivity), product quality/quantity, team innovation, supervisor ratings ofteam performance/effectiveness, and self-ratingsby team members. Most studies provided objectiveteam performance measures or ratings by supervi-sors. For subjective performance measures, reli-abilities of measurement instruments were re-corded whenever reported. In those cases in whichno reliabilities were reported, we took the averagereliability of the same variable from all other stud-ies. The average reliabilities were .82 for subjectiveperformance measures. To maintain the statisticalindependence of the data set, when multiple mea-sures for team performance were available, we in-cluded only the most objective external measure;when no other source was available, we relied onteam members’ own assessments of their perfor-mance. In a study measuring multiple dimensionsof team performance (e.g., its quality and quantity),we calculated a composite effect by averaging eachcorrelation coefficient.
Occupational demography. We relied on the ar-chival data from the Labor Force Statistics of theCurrent Population Survey (U.S. Bureau of LaborStatistics, 2006) to obtain definitions and statisticsof occupational categories. Initially we coded theresearch settings (based on sample descriptions) ofstudies included in the meta-analysis using Bureauof Labor Statistics (BLS) definitions; then we ob-tained the occupational gender, race/ethnicity, andage composition data from the BLS data set andassigned this demographic composition to the oc-cupations included in the studies. For example, if astudy’s respondents were described as members ofproduction teams from multiple electronics manu-facturing firms, we obtained data on electrical,electronics, and electro-mechanical assemblers’gender, ethnicity, and age from the BLS websiteand used them to measure occupational demogra-phy (n � 16). For teams that included multipleoccupations (e.g., cross-functional product devel-opment teams consisting of engineers and produc-tion supervisors), we found demographic data foreach occupational category involved in a team andthen calculated a composite value by averaging thedemographic information on all participating occu-pations from the BLS data set (n � 8). In aggregatingdata over multiple occupational categories, wewere careful not to include studies that incorpo-rated occupations that were differently skewed de-mographically. For example, when a research sam-
612 JuneAcademy of Management Journal
ple involved multiple occupations in singleteams—combining, say, the white-male-domi-nated occupation of network analyst and the rel-atively balanced occupation of insurance under-writer or HR professional (Campion, Papper, &Medsker, 1996)—we performed no aggregationand excluded the study from this set of analyses.Moreover, when studies were unclear about thesamples used, had non-U.S. samples for whichdata were not available, or had samples withmultiple types of teams, we also excluded themfrom the analyses.
Among the studies included in these analyses,the occupational percent female ranged from 20.6to 73.4 percent, and the occupational percent eth-nic minority (i.e., nonwhite) ranged from 15.7 to37.6 percent. If either of these percentages wasbelow the overall mean for all U.S. occupations(i.e., 46.3 percent for female composition and 29percent for minority composition), the occupa-tional setting was characterized as “majority male”or “majority white” for the purposes of categoricalanalyses. Occupational settings with above-averagepercents female or minority were categorized as“balanced” settings. In our sample, we did not en-counter enough study settings that could be cate-gorized as majority female or majority ethnic mi-nority; therefore, we were unable to test the effectsof diversity in these settings. Occupational agecomposition data were based on the informationprovided by the BLS, which offers data on four agecategories by occupation. We considered an occu-pation as “majority younger worker” if the propor-tion of older (over 55 years) workers in that occu-pation was lower than the average acrossoccupations (i.e., 18%) and as “balanced” if theproportion of older workers was above the averageacross occupations.
Industry setting. Industry was also coded when-ever it was reported and whenever the researchsetting or sample was described. Following strate-gic management research (e.g., Collins & Smith,2006; Datta et al., 2005) and drawing from the de-tailed industry descriptions based on the NorthAmerican Industry Classification System (NAICS;U. S. Census Bureau, 2002), we classified indus-tries examined in previous research into threebroad categories: service, manufacturing, andhigh technology. Service industries in the sampleincluded wholesale/retail trade, finance/insur-ance, health care, educational service, moving/transportation, and government service; manu-facturing industries included automobilemanufacturing, paper and wood manufacturing,textile manufacturing, oil and gas, chemicalproduct manufacturing, and general equipment
manufacturing. High technology included semi-conductor/electronics, information processing,telecommunication, and professional R&D ser-vices. Studies that included samples from multi-ple industry settings were excluded from ouranalyses (n � 5).
Team interdependence. Drawing on conceptualdefinitions of interdependence presented in previ-ous research (e.g., Campion, Medsker, & Higgs,1993; Saavedra et al., 1993; Shea & Guzzo, 1987),we determined team interdependence using threeseparate ratings for task, goal, and outcome inter-dependence. Each dimension was rated on a scaleranging from 1 to 3 (low, moderate, and high) andaveraged to provide a composite score of overallteam interdependence. Low-interdependence teams in-cluded, for example, production or sales teamswith sequentially related activities or individual-based goals and rewards; cross-functional R&Dteams whose members frequently exchanged ideasand shared common goals were deemed highlyinterdependent.
Team type. We also coded team type based onthe length of time a team was expected to exist(Schippers et al., 2003). Two broad categories,short-term and long-term, were used. Cross-func-tional project teams existing for a limited periodwere considered short-term teams, for example.Permanent work teams and general work teams ex-isting for longer than two years were generally con-sidered long-term.
Meta-Analytic Techniques
We used Hedge and Olkin’s (1985) meta-analyticprocedures to analyze the data. Zero-order correla-tions between work team diversity and perfor-mance were taken or calculated from each studyand corrected for measurement error. FollowingHunter and Schmidt’s (1990) formula, we also cor-rected the correlations for unreliability using theartifact distributions for subjective team perfor-mance measures. We calculated weighted meancorrelations by adopting the inverse varianceweights and applying Fisher’s Z transformationprocedures (Hedge & Olkin, 1985; Lipsey & Wilson,2001). Ninety-five percent confidence intervalswere calculated around the sample-weighted corre-lation as a measure of accuracy of the effect size(Whitener, 1990). The failsafe k was also calculatedto identify the number of “file drawer” (unknown)studies of the same relationship with a true corre-lation of zero needed to widen the reported confi-dence interval enough to include zero (Orwin,1983; Rosenthal, 1979).
2009 613Joshi and Roh
Heterogeneity of effect sizes. To determinewhether effect sizes were consistent over the stud-ies reviewed, we tested the homogeneous distri-bution of the effect sizes by calculating the Q-statistic (Hedge & Olkin, 1985). A significant Qindicates the likelihood of moderators that ex-plain variability in correlations over studies (Lip-sey & Wilson, 2001). We also examined the po-tential impact of outliers by calculating thesample-adjusted meta-analytic deviancy statisticsuggested by Huffcut and Arthur (1995). Threeoutliers were identified by use of this statistic,but a detailed review of the potential outliersrevealed no problematic correlations, so thesewere not eliminated from the analyses.
Moderator analysis. We conducted detailedmoderator analyses to determine whether multi-level contextual variables were related to the het-erogeneity of effect sizes (Hedge & Olkin, 1985;Lipsey & Wilson, 2001). The logic of the categoricalmodel moderator test is analogous to analysis ofvariance (ANOVA). Calculating the categoricalmodels results in (1) the between-group goodness-of-fit statistic QB, which has an approximate chi-square distribution with p � 1 degrees of freedom,where p is the number of groups, and (2) the with-in-group goodness-of-fit statistic Qw, which has anapproximate chi-square distribution with m � 1degrees of freedom, where m is the number of effectsizes in the group. That is, QB is analogous to amain effect in an ANOVA, and Qw indicates homo-geneity within each group in an ANOVA. In thepresent analysis, as recommended in previous re-search on this topic (e.g., Webber & Donahue,2001), we used a QB statistic to test whether thecategorical moderator model was statistically sig-nificant and then examined each subgroup withinthe sample by testing the confidence intervals forstatistical significance and by comparing the effectsizes across subgroups whenever possible. For thecontinuous moderators (i.e., occupational percentfemale and occupational percent minority),weighted least squares (WLS) regression was alsoused, as suggested by Hedge & Olkin (1985). Thisprocedure involves weighting each observed effectsize by the inverse of its variance, as with formulaefor weighted means and confidence intervals (Lip-sey & Wilson, 2001). Using such an approachavoids the artificial categorization of continuousmoderating variables. Two indexes assessing theoverall fit of the weighted regression model canbe calculated: a Q attributable to the regressionand a Q error or residual (denoted as QR and QE,respectively, and both distributed as a chi-square). QR is analogous to an F for a regressionmodel and, if significant, indicates that the re-
gression model explains significant variability inthe correlations of the relationship between teamdiversity variables and team performance (Lipsey& Wilson, 2001).4
RESULTS
Main Effects: Work Team Diversityand Performance
Using the meta-analytic techniques describedabove, we tested the main effects of work teamdiversity on performance outcomes as well as themoderating effects of contextual factors embeddedat multiple levels. Table 2 presents the main effectresults. We first examined the correlations betweenall types of diversity and performance and obtaineda near-zero, nonsignificant result (r � �.01, k �117, 95% CI � �.02 to .00). This initial resultcorroborated past meta-analytic findings (e.g., Web-ber & Donahue, 2001). We then conducted separateanalyses for relations- and task-oriented diversityand found a different pattern of results for eachtype of diversity. For relations-oriented diversity,we found a very weak negative but significant re-lationship with performance (r � �.03, k � 69,95% CI � �.05 to �.02). The relationship betweentask-oriented diversity and performance was alsovery weak but positive and significant (r � .04, k �48, 95% CI � .02 to .06). The failsafe k’s in Table 2suggest that, although the effect sizes are verysmall, at least 118 (for relations-oriented diversity)and 73 (for task-oriented diversity) file-drawer nulleffects would need to be reported before those tworelationships would lose statistical support. Wealso conducted additional analyses for each diver-sity attribute and found that functional backgrounddiversity was most positively related to perfor-mance (r � .13, k � 20, 95% CI � .09 to .17) andthat age diversity showed the most negative perfor-
4 We undertook additional OLS regressions whereby weregressed the effect of diversity-performance on the keyindependent variables. Results indicated that the pattern offindings for OLS regressions mirrors the pattern of findingsreported here. We also undertook a stepwise regressionprocedure and found that adding each contextual variablewas associated with significant incremental variance in thediversity-performance effect. Collectively, in these analy-ses, contextual variables accounted for 56 percent of thevariance in relations-oriented effect sizes across studies.Further, the variance inflation factors in all cases were wellbelow 10, which is the rule-of-thumb cutoff for high mul-ticollinearity. The full results are available from the authorsupon request. We thank our action editor, Jason Colquitt,for this suggestion.
614 JuneAcademy of Management Journal
mance effect (r � �.06, k � 21, 95% CI � �.09 to�.04).
Table 2 also shows that considerable heterogene-ity among effect sizes exists, as indicated by theQ-statistic. Both Q values for relations- and task-oriented diversity are highly significant (p’s � .01),indicating that correlations vary across studies andthat potential moderators might exist that can ex-plain these correlations.
Occupational Demography
Hypotheses 1a and 1b predicted that the negativeeffects of gender and race/ethnicity diversity wouldbe weaker in more gender-balanced and ethnicallybalanced settings, respectively. We conducted cat-egorical moderator analyses and contrasted the dif-ference between majority male or majority whiteoccupations and relatively gender-balanced andethnically balanced occupations. Table 3 summa-rizes the results. The categorical model testing themoderating effect of occupational percent femalewas highly significant (QB[1] � 39.19, p � .01). Ashypothesized, gender diversity had a significant,negative effect on team performance in majoritymale occupational settings (r � �.09, k � 12, 95%CI � �.12 to �.05). The effect of gender diversitywas significantly positive in relatively gender bal-anced settings (r � .11, k � 7, 95% CI � .06 to .15).The moderating effect of occupational percentminority was also significant (QB[1] � 48.65, p �.01). The average correlation was significantlynegative in majority white occupations (r � �.07,k � 10, 95% CI � �.10 to �.04) and positive inrelatively balanced occupations (r � .11, k � 6,95% CI � .07 to .14). In addition, because theoccupational gender and race/ethnicity demogra-
phy variables were measured as continuous vari-ables (i.e., occupational percent female and occu-pational percent minority), to further understandthe patterns of these moderating effects, we alsoperformed WLS regressions. In support of bothhypotheses, we found that occupational percentfemale and minority accounted for significantamounts of variance in the correlations betweengender and race/ethnicity diversity and team per-formance (� � .32, p � .01, R2 � .10, for genderdiversity; � � .37, p � .01, R2 � .14, for race/ethnicity diversity). Significant values of QR forgender diversity (QR[1] � 16.78, p � .01) andrace/ethnicity diversity (QR[1] � 22.71, p � .01)also indicate that both regression models werestatistically significant.
Hypothesis 1c predicted that the negative effectsof age diversity would be strengthened in occupa-tions composed of relatively younger workers andweakened in relatively age-balanced occupations.We conducted a categorical analysis and did notfind a strong moderating relationship (QB[1] �0.85, p � .10) (see Table 3). Although occupationscomposed of younger workers displayed slightlymore negative effects than occupations that weremore balanced in terms of age, considering thenonsignificant QB and significantly overlappingconfidence intervals, the two occupational groupswere not statistically different.
Hypothesis 1d proposed that the positive perfor-mance outcome of task-oriented diversity wouldbe stronger in more balanced occupational set-tings. Contrary to this proposition, task-orienteddiversity showed more positive performance ef-fects in majority male and white settings. How-ever, considering the nonsignificant subgroup re-sults for balanced occupational settings across all
TABLE 2Main Effects: The Relationship between Team Diversity and Performancea
Diversity TypeEffect Sizes
(k)Total Teams
(N)WeightedMean r
95%Confidence
IntervalFailsafe
k Q
All diversity 117 29,608 �.01 �.02, .00 635.16**Relations-oriented diversity 69 19,779 �.03 �.05, �.02 118 479.54**
Gender 26 5,473 �.02 �.04, .01Race/ethnicity 22 7,089 �.01 �.04, .01Age 21 7,217 �.06 �.09, �.04 48
Task-oriented diversity 48 9,829 .04 .02, .06 73 155.63**Function 20 3,085 .13 .09, .17 65Education 9 2,863 �.02 �.06, .01Tenure 19 3,881 .03 �.01, .06
a N is the total number of teams counted by effect sizes; failsafe k indicates the number of unpublished studies reporting null resultsneeded to reduce the cumulative effect across studies to the point of nonsignificance (p � .05) and is only reported for statisticallysignificant results (p � .05); Q is the effect size heterogeneity statistic indicating the possibility of moderators.
** p � .01
2009 615Joshi and Roh
diversity attributes and the overlapping confi-dence intervals in general, we were unable inter-pret this finding meaningfully.
Industry Setting
Hypotheses 2a and 2b proposed that industrysetting moderated the relationship between teamdiversity and performance outcomes. Table 4 pre-sents the results. With regard to relations-orienteddiversity, the overall categorical model testing in-dustry impact was highly significant (QB[2] �209.89, p � .01). As predicted in Hypothesis 2a,relations-oriented diversity had a positive effect onperformance in service industries (r � .07, k � 21,95% CI � .05 to .09). Inconsistently with Hypoth-esis 2a, however, in the manufacturing industrysetting, the effect of relations-oriented diversitywas negative (r � �.04, k � 16, 95% CI � �.07 to
�.01) and interestingly, relations-oriented diver-sity displayed the strongest negative performanceeffect in high-technology industry settings (r ��.18, k � 21, 95% CI � �.20 to �.15). This findingis also resistant to unpublished null effects with afailsafe k of 152.
Regarding Hypothesis 2b, we found weak sup-port for the moderating effects of industry settingon the performance outcome of task-oriented diver-sity. The overall categorical model was modestlysignificant (QB[2] � 7.28, p � .05). Task-orienteddiversity was positively related to performance inthe high-technology industry setting, as hypothe-sized (r � .06, k � 23, 95% CI � .04 to .09), but thiseffect was only slightly larger than the overall maineffect of task-oriented diversity (r � .04). We didnot find any significant support for industry mod-erating effects on task-oriented diversity effects inthe manufacturing and service settings.
TABLE 3Contextual Influences: Occupational Demographya, b
Team Diversity � OccupationalDemographyc
Effect Sizes(k)
Total Teams(N)
WeightedMean r
95%Confidence
IntervalFailsafe
k QB
Gender diversity (Hypothesis 1a) 39.19**Majority male settings 12 2,952 �.09 �.12, �.05 20Balanced settings 7 1,832 .11 .06, .15 12
Race/ethnicity diversity (Hypothesis 1b) 48.65**Majority white settings 10 4,071 �.07 �.10, �.04 17Balanced settings 6 2,584 .11 .07, .14 13
Age diversity (Hypothesis 1c) 0.85Majority younger worker settings 6 4,758 �.08 �.10, �.05 13Balanced settings 9 1,869 �.05 �.10, �.00 3
Task-oriented diversity (Hypothesis 1d)Occupational gender demography 9.67**
Majority male settings 22 6,866 .06 .03, .09 39Balanced settings 14 1,823 �.03 �.08, .02
Occupational race/ethnicity demography 0.60Majority white settings 32 8,497 .04 .02, .06 39Balanced settings 4 192 �.02 �.17, .13
Occupational age demography 5.36*Majority younger worker settings 9 5,199 .06 .03, .09 13Balanced settings 27 3,490 .01 �.03, .04
a N is the total number of teams counted by effect sizes; failsafe k indicates the number of unpublished studies reporting null resultsneeded to reduce the cumulative effect across studies to the point of nonsignificance (p � .05) and is only reported for statisticallysignificant results (p � .05); QB is the between-group heterogeneity statistic indicating the statistical significance of the categoricalmoderator model.
b For continuous occupational demography variables (i.e., occupational percent female and occupational percent minority), we alsoconducted the WLS regression analyses and obtained the same pattern of findings as reported in this table of results (in relation toHypothesis 1a and 1b).
c Occupational settings in which the percent female or minority was below the overall mean level (i.e., 46.3 percent for femalecomposition and 29 percent for minority composition) were considered as majority male or majority whites, respectively, in the analyses;we considered an occupation as a majority-younger-workers setting if the proportion of older workers (over 55 years old) in that occupationwas less than the overall average (i.e., 18%).
* p � .05** p � .01
616 JuneAcademy of Management Journal
Team Interdependence and Team Type
Table 5 presents the results regarding the effectsof team-level moderators on the relationship be-tween team diversity and performance. Hypothesis3 proposed that effects of task-oriented diversitywill be stronger and the effects of relations-orienteddiversity will be weaker among highly interdepen-dent teams. The findings with regard to relations-oriented diversity were contrary to this hypothesis.Although the categorical model for relations-ori-ented diversity was significant (QB[2] � 174.21,p � .01), among teams with low interdependencerelations-oriented diversity was positively relatedto performance (r � .08), and among teams withmoderate and high interdependence, relations-ori-ented diversity was negatively related to perfor-mance (r � �.12 and r � �.04, respectively). Thecategorical analysis for task-oriented diversity sup-ported the hypothesized pattern, although the over-all moderating model only showed limited statisti-cal support (QB[2] � 7.14, p � .05). The positiveperformance effect of task-oriented diversity in-creased as team tasks, goals, and outcomes becamemore interdependent; however, these results alsoneed to be interpreted cautiously because the 95percent confidence interval of the low-interdepen-dence subgroup included zero and overlapped withthose for the other two subgroups.
Our final hypothesis addressed whether teamtype—the length of time a team was expected toexist (i.e., long-term versus short-term)—affected
the relationship between team diversity and perfor-mance. Hypothesis 4 proposed that the negativeeffects of relations-oriented diversity would bestrengthened in long-term teams. We found strongsupport for relations-oriented diversity: the cate-gorical moderator model was highly significant(QB[1] � 222.91, p � .01). The performance effect ofrelations-oriented diversity was positive in rela-tively short-term teams (r � .09, k � 23, 95% CI �.07 to .12) but became negative in more stable orlong-term teams (r � �.14; k � 43, 95% CI � �.16to �.12). Both findings are resistant to unpublishednull effects, with failsafe k’s of 89 and 305, respec-tively. For task-oriented diversity, we did not find astatistical support for the hypothesis (QB[1] � 0.63,p � .10). Although the performance effect of task-oriented diversity was more positive in short-termteams (r � .08) than in long-term teams (r � .04),these subgroup results were not statistically differ-ent from each other.
DISCUSSION
This meta-analytic review took stock of past re-search on the diversity-performance relationshipconducted in organizational settings over the past15 years and examined the sensitivity of this rela-tionship to contextual variables at multiple levels.Our findings revealed that when one considers therelationship between all types of diversity (i.e., col-lapsing relations- and task-oriented distinction)
TABLE 4Contextual Influences: Industry Settinga
Team Diversity � Industry SettingbEffect Sizes
(k)Total Teams
(N)WeightedMean r
95%Confidence
IntervalFailsafe
k QB
Relations-oriented diversity (Hypothesis 2a) 209.89**High-technology industry 21 6,068 �.18 �.20, �.15 152Service industry 21 9,139 .07 .05, .09 58Manufacturing industry 16 3,687 �.04 �.07, �.01 8
Task-oriented diversity (Hypothesis 2b) 7.28*High-technology industry 23 6,475 .06 .04, .09 45Service industry 16 1,702 �.00 �.05, .05Manufacturing industry 3 1,194 .01 �.05, .06
a N is the total number of teams counted by effect sizes; failsafe k indicates the number of unpublished studies reporting null resultsneeded to reduce the cumulative effect across studies to the point of nonsignificance (p � .05) and is only reported for statisticallysignificant results (p � .05); QB is the between-group heterogeneity statistic indicating the statistical significance of the categoricalmoderator model.
b Three industry categories were analyzed: (1) high-technology included electronics/semiconductors, information processing, telecom-munication, and professional R&D service; (2) service included retail trade, finance/insurance, health care, education service, moving/transportation, and government service; (3) manufacturing industries included automobile manufacturing, paper and wood manufacturing,textile manufacturing, oil and gas extraction, and general and chemical product manufacturing.
* p � .05** p � .01
2009 617Joshi and Roh
and performance, the direct effect of diversity onperformance is essentially zero. Relations-orienteddiversity attributes such as gender, race/ethnicity,and age had very small, though significantly nega-tive, effects on team performance. Although func-tional diversity had a more substantial positive ef-fect, other forms of task-oriented diversity (i.e.,education and tenure) also had very small effectson team performance. These findings make it ap-pear that diversity does not really matter for teamperformance. However, we contend that these weakdirect relationships may be obscuring the specificconditions under which diversity can have benefi-cial or detrimental effects on performance out-comes. Our study developed and tested a frame-work identifying boundary conditions under whichthe diversity-performance relationship is likely tobe significant. Specifically, we found that after weaccounted for moderating variables at multiple lev-els, diversity effects doubled or tripled in size. Fur-ther, industry and occupational moderators, whichhave received relatively scant attention in past re-search, explained significant variance in relations-oriented diversity effects across studies. Below we
discuss the theoretical, empirical, and practical im-plications of our findings.
The Role of Occupational Demography andIndustry Setting as Context
We found that in occupations dominated by maleor by white employees, gender and ethnic diversityhad more negative effects on performance out-comes. These findings draw attention to the impor-tance of extraorganizational context in shaping di-versity outcomes, and we call for a more detailedand comprehensive acknowledgement of macro-level context in future research. Apart from stereo-types associated with underrepresented groups in aparticular occupational context, implicit status dif-ferences between demographic groups may also bea mechanism by which contextual factors such asoccupational demography filter into team-level in-teractions (see Ridgeway, 1991, 1997). The domi-nance of a particular demographic group within aparticular occupational setting can signal greateraccess to resources and privilege for this group. Theprivilege accrued by a demographic group in a par-
TABLE 5Contextual Influences: Team Interdependence and Team Typea
Team Diversity � Team ContextEffect Sizes
(k)Total Teams
(N)WeightedMean r
95%Confidence
IntervalFailsafe
k QB
Team interdependence (Hypothesis 3)b
Relations-oriented diversity 174.21**Low interdependence 17 8,051 .08 .06, .10 56Moderate interdependence 38 10,770 �.12 �.14, �.10 238High interdependence 14 958 �.04 �.11, .03
Task-oriented diversity 7.14*Low interdependence 5 557 �.03 �.11, .06Moderate interdependence 24 7,604 .04 .02, .06 23High interdependence 19 1,668 .10 .05, .15 23
Team type (Hypothesis 4)c
Relations-oriented diversity 222.91**Short-term teams 23 7,733 .09 .07, .12 89Long-term teams 43 9,730 �.14 �.16, �.12 305
Task-oriented diversity 0.63Short-term teams 13 684 .08 �.01, .16Long-term teams 34 8,373 .04 .02, .07 45
a N is the total number of teams counted by effect sizes; failsafe k indicates the number of unpublished studies reporting null resultsneeded to reduce the cumulative effect across studies to the point of nonsignificance (p � .05) and is only reported for statisticallysignificant results (p � .05); QB is the between-group heterogeneity statistic indicating the statistical significance of the categoricalmoderator model.
b Low-interdependence teams included, for example, production or sales teams with sequentially related activities and/or individual-based goals or rewards; high-interdependence teams included, for example, cross-functional R&D teams that frequently exchanged ideasand shared common goals.
c Short-term teams involved, for example, project-based teams existing for a limited period of time; permanent work teams or generalwork teams existing for longer than two years were generally considered relatively long-term teams.
* p � .05** p � .01
618 JuneAcademy of Management Journal
ticular context may provide team members belong-ing to this group with an “expertise advantage”;these group members may be considered morecompetent, and the contributions of the members oflow-status demographic groups may be devalued inthese settings (Berger, Ridgeway, Fisek, & Norman,1998; Berger, Ridgeway, & Zelditch, 2002). Indeed,a rich body of sociopsychological research hasshown that, in small groups, interpersonal interac-tions tend to replicate larger societal status differ-entials (Berger et al., 1998, 2002) and may accountfor the suboptimal performance outcomes of gen-der-diverse and ethnically diverse teams in major-ity male and white settings.
We propose that a closer examination of the oc-cupational context in which teams are embeddedcan provide researchers with an understanding ofbroader structural inequality and its implicationsfor team-level diversity outcomes. Further, al-though we were not able to test the role of organi-zational practices in these settings, research sug-gests that these status differentials are furtherlegitimized by organizational practices that favorsome demographic groups over others (Ridgeway,1991, 1997). Thus, another direction for futureresearch could be to examine the role that organ-izational practices play in either enhancing ormitigating status-based processes in a specificoccupational context.
Although statistically inconclusive, our findingwas that task-oriented diversity had weakly posi-tive effects in male/white dominated settings.This pattern is contrary to our theoretical argu-ments. Although we caution against overinterpret-ing this finding, we acknowledge that in these morehomogeneous settings, categorization based on de-mographic characteristics may not have been suf-ficient to mitigate the elaborative potential of task-oriented diversity. We call for more research onpossible mechanisms mediating task-oriented di-versity effects in these occupational settings.
Our findings indicated that the industry settingsin which teams were embedded also had interest-ing implications for team-level diversity-based out-comes. We found that relations-oriented diversityhad positive effects in service industry settings andslightly negative effects in manufacturing settings.In addition to the market competence perspective(Richard et al., 2007) discussed in developing ourhypotheses, some additional considerations shouldalso perhaps be taken into account. Service settings(for example, retail establishments and restaurants)involve front-line customer contact, and the costsof interactions based on negative categorizationsare high in this context. Hence, firms embedded inthese industries may engage in proactive diversity
management efforts to address gender-, ethnicity-,and age-based issues in the workplace.5 Specificforms of training intended at changing behaviorstargeted at demographically dissimilar employeesand customers may be implemented in these set-tings. In contrast, in manufacturing settings, wherewe found negative effects, firms do not face similarpressures and may be less likely to directly addressthese aspects of team diversity.
Surprisingly, we found that in high-technologysettings, relations-oriented diversity had a moresubstantial negative effect. Recently, DiTomaso andcolleagues found that, in 24 firms that fit our defi-nition of high-technology settings, white men re-ceived more training, mentoring, and coaching andalso the most favorable performance evaluations,relative to any other demographic group (DiTo-maso, Post, Smith, Farris, & Cordero, 2007). Thesecharacteristics of high-technology firms may en-hance ethnicity- and gender-based status differen-tials. It would appear that white men occupy ahigh-status position (reflected in these favorableemployment outcomes) in the high-technology sec-tor (DiTomaso, Post, Smith et al., 2007) and wouldenjoy an “expertise advantage” relative to womenor minorities. Therefore, the status-based processesthat we discussed earlier could also account for thenegative effects of relations-oriented diversity inthis industry. Furthermore, like manufacturingfirms, high-technology firms may not face pres-sures to proactively adopt diversity managementpractices. These findings call for a finer-grainedunderstanding of the industry context framing di-versity-based outcomes. For example, future re-search could apply contingency-based perspectivesand examine whether a fit between firms’ diversitymanagement strategies and industry-level contin-gencies explains the outcomes of diversity at theteam level. Longitudinal studies that examine howspecific industry trends influence firms’ responsesto diversity and their implications for team perfor-mance could be an avenue for future inquiry.
We conclude this section by noting that our find-ings challenge the assumption, born from social-categorization theory, that some aspects of diver-sity necessarily have detrimental effects on teamperformance. Although relations-oriented diversityattributes have been typically associated with cat-egorization-based processes (Jackson et al., 1995;
5 Consider as an example the negative publicity re-ceived by Denny’s Restaurants as a result of discrimina-tion charges and their subsequent efforts to train andreward appropriate behaviors toward demographicallydissimilar employees and customers.
2009 619Joshi and Roh
Jehn et al., 1999), we found that the effects of theseattributes on team performance ranged from signif-icantly negative (in the high-technology industryand in male/white dominated occupations) to sig-nificantly positive (in the service industry and ingender-/ethnicity-balanced occupations). The ef-fects of task-oriented diversity, however, were in-variant across these contexts. Current applicationsof social identity theory or social categorizationtheory in diversity research are insufficient for ex-plaining these findings. These theoretical frame-works were originally developed to explain the out-comes of differences in minimal group settings andmay not capture the effects of demographic at-tributes among naturally occurring work groups inthese contexts (see also DiTomaso, Post, Smith,Farris, & Cordero, 2007; Linnehan & Konrad, 1999).Therefore, these findings suggest that a priority forfuture research is to investigate how the demogra-phy of a work context and the characteristics of theindustry settings in which teams are embedded caninfluence the performance outcomes of gender orethnic diversity. We call for more context-focuseddiversity research, which would entail targetingparticular settings in which either the negative orthe positive consequences of team diversity aremore likely. Such efforts could facilitate the devel-opment of new theoretical approaches that incor-porate the status-based or strategic managementperspectives we outlined above.
Team-Level Contextual Influences
Our review of past literature suggested that therehas been a growing emphasis on the role of moder-ators in diversity research and that, predominantly,past research has focused on team-level moderatorssuch as team tenure, task interdependence, andtask complexity (see Van Knippenberg & Schippers[2007] for a review). In view of this research, weproposed that contextual factors such as team in-terdependence and team type would moderate theeffects of diversity on performance in such a waythat elaboration-based processes would be morelikely to be observed in more interdependent orshorter-lived teams (Bowers et al., 2000; Jehn et al.,1999). Contrary to our predictions, we found thatrelations-oriented diversity had the most negativeeffects in moderately interdependent teams. Thesefindings suggest that the interactive effects of teamdiversity and interdependence may be more com-plex than acknowledged in the past. Moderatelyinterdependent tasks may impose constraints thatinterfere with the elaborative potential of diversityattributes but are not so demanding that team mem-bers must overcome categorization-based processes
and collaborate with each other. Although the ef-fect sizes were smaller, a lower level of interdepen-dence was associated with positive performanceoutcomes for relations-oriented diversity. In teamswith low interdependence, the distinct nature ofeach individual team member’s task and a lowerneed to exchange information may create a situa-tion in which team members are not frustrated withdissimilar group members and are able to recognizethe contributions of other members to the teamgoals (Pelled et al., 1999).
In support of our propositions, findings indi-cated that in short-term teams (such as temporaryproject teams), relations-oriented diversity had apositive effect on performance. In these teams, theexpectation of a finite amount of time to completetasks may force group members to exploit the fullelaborative potential of diversity attributes. In long-term teams, team members may feel less compelledto identify and utilize diverse perspectives andmay also have more opportunity to bring up con-flicts that may be detrimental to group functioning.We note that we had limited ability to test temporalinfluences on diversity effects more directly in thismeta-analysis. Measures of team tenure or longev-ity may yield a different pattern of findings (Harri-son et al., 2001; Schippers et al., 2003).
Overall, it is interesting to note that these team-level moderators, which have received the mostattention in past research, had weaker moderatingeffects on diversity outcomes than industry andoccupational variables. Moreover, these variableshad a more significant moderating effect on rela-tions-oriented diversity effects than on task-ori-ented diversity effects. In general, we found thatmoderately interdependent teams and long-termteams could face challenges with increasing diver-sity; less interdependent teams and short-termteams would benefit from greater demographic di-versity among team members. These findings sug-gest that more research on the psychological pro-cesses underlying these moderating effects isneeded in these specific team contexts. For exam-ple, understanding whether role stress or role over-load associated with greater team interdependencehas implications for the manner in which teammembers deal with demographic differences wouldbe useful. These team-level moderators may alsoneed to be considered in conjunction with othermacro-level contextual variables. For example, itwould be interesting to understand whether task-related contingencies are more likely to induceconflict in demographically diverse groups whenthe macrolevel context enhances categorization-based processes.
620 JuneAcademy of Management Journal
Caveats, Limitations, and Future Directions
The meta-analytic approach we applied in thisstudy has some limitations. One concern may bethat the sample is small and may be associated withsecond-order sampling error, which is particularlyrelevant for moderator analysis (Hunter & Schmidt,1990). However, the failsafe k’s we report are fairlylarge and suggest that, even at the rate at whichdiversity research is being conducted, at least adecade more of research may be required to over-turn the significant findings reported here.
We note that we were also limited in our abilityto code contextual variables, as the studies’ de-scriptions of research settings and team tasks wereoften scanty. For example, despite the increasingattention to temporal effects on team diversity out-comes, the lack of information regarding team ten-ure or team duration was especially troubling. Wepropose that future studies provide more detaileddescriptions of research settings and acknowledgethe role that these factors play in explaining re-search findings. Because of theoretical consider-ations and data availability, we considered only asubset of the possible moderating influences onteam diversity outcomes. Missing from this analy-sis is the role that organizational context and otherextraorganizational factors can play in shaping di-versity outcomes. Aspects of organizational contextsuch as managerial demography, climate, culture,and leadership also merit closer scrutiny in futureresearch. Other extraorganizational factors, such associetal and political events (e.g., immigrationtrends, passage of key legislation), may also be im-portant to consider in the future but were outsidethe scope of this study. Further, we note that fortesting the effects of occupational demography, wecould only include U.S.-based samples. These find-ings may vary in other cultural contexts.
Finally, we used a simplified relations-orientedversus task-oriented typology while consideringcontextual effects on diversity outcomes; thisdichotomy corresponded to the categorization-versus elaboration-based processes underlying di-versity effects. Other typologies, such as surface-versus deep-level diversity, may also be pertinentto this analysis and should be considered in thefuture. Contextual effects on the outcomes of deep-level diversity variables, such as personality, cog-nitive ability, values, and attitudes, were also out-side the scope of the present study and may be ofinterest in future research. We were also unable toincorporate the faultlines approach that has been agrowing focus in diversity research. Of the studieswe initially reviewed, 5 percent applied a fault-lines-based operationalization of diversity. We also
note that, although conceptually interesting, fieldresearch directly applying faultlines-based mea-sures is not abundant and has yielded inconsistentresults (e.g., Gibson & Vermeulen, 2003; Greer &Jehn, 2008; Lau & Murnighan, 2005; Li & Hambrick,2005). We propose that future research also exam-ine how contextual moderators may enhance thestrength and number of faultlines in teams to re-solve these mixed findings.
Harrison and Klein (2007) offered several guide-lines for the measurement of diversity in terms ofthree dimensions: separation, variety, and dispar-ity. These authors proposed that the measurementof various diversity attributes should be based onthe specific dimension of diversity considered. Inthis meta-analytic review, because we relied onmeasurement of diversity that predated this 2007article, we were unable to adopt these guidelines.However, complementing Harrison and Klein’s(2007) arguments, we note that the malleability ofrelations-oriented diversity to occupational, indus-try, and team-level contextual variables under-scores the importance of clarifying the specificdimension associated with these demographic at-tributes in the future. A comprehensive contextualanalysis may also help researchers determinewhether a specific demographic attribute is likelyto manifest as “separation,” “variety,” or “dispar-ity” in teams. For example, diversity could be con-ceptualized as variety in short-term teams, in ser-vice settings, and in demographically balancedsettings but as disparity in long-term teams, inmale/white dominated occupations, and in high-technology settings.
Implications for Diversity Management
A context-based approach to workplace diversityresearch can potentially provide practical insightsthat might enhance the effectiveness of diversitymanagement practices. In some work contexts, it isimportant to recognize that diversity at the workgroup level may be problematic. In these settings,diversity interventions aimed directly at targetingbehaviors driven by stereotyping and bias againstunderrepresented groups and ensuring representa-tion at higher levels may be necessary to reversenegative diversity-based outcomes. Furthermore, or-ganizations can directly address the dominance of ademographic majority in a particular labor marketby proactively implementing partnerships with ed-ucational institutions to increase gender- and race-based diversity in the applicant pool. These obser-vations may be particularly important in the case ofthe high-technology settings included in our anal-ysis. At a time when this sector of the economy is
2009 621Joshi and Roh
growing rapidly, there is also growing concernamong policy makers regarding the declining ratioof women and minorities in this industry (Informa-tion Technology Association of America, 2003).The negative performance effects of diversity thatwe found in high technology may also reflect an“unfriendly” atmosphere for women and minori-ties in this setting. Partnerships with high schoolsand college campuses to increase gender and ethnicdiversity in mathematics, science, and engineeringcould enhance the diversity of the applicant pooland provide a more balanced representation of var-ious demographic groups in firms, thereby mitigat-ing the problematic consequences of demographicdiversity at the work group level. We note that afew prominent firms in this industry are alreadymaking these proactive efforts, and our findingsunderscore the value of these initiatives.
With respect to our findings regarding the effectsof team-level moderators, we propose that diversitymanagement practices need to specifically addressthe level of interdependence or longevity that teammembers will encounter. Teams performing moreinterdependent tasks over the long term may needongoing training interventions or team coachingthat facilitates group decision making and con-flict resolution. A more tailored team-focusedapproach to designing diversity managementpractices may enhance the relevance and effec-tiveness of these practices in teams. In general, a“contextual diagnosis” will allow firms to de-velop diversity management practices that aretailored to reduce categorization-based processesand enhance elaboration-based processes at theteam level. Overall, long-term teams that performinterdependent tasks in less demographicallybalanced occupational settings may be the mostvulnerable to categorization-based processes andmay need more direct and focused interventionsthan other types of teams.
Conclusion
Past reviews have labeled differing theoreticalperspectives and contradictory findings in diver-sity research as the “double-edged sword of diver-sity” (e.g., Webber & Donahue, 2001). In this article,we propose a new agenda for diversity research—one that moves beyond a debate regarding the po-tential benefits or costs of diversity and highlightsthe inherent context dependence of diversity ef-fects in organizations. We note that inadequatelyreporting and acknowledging context not only ob-scures the important consequences of diversity inorganizations, but also hampers efforts to synthe-size and integrate the cumulative evidence from the
past; handicaps future theory building; and limitsresearchers’ ability to distill the practical implica-tions of findings. This study presents a roadmap forcontext-focused research that we hope encouragesresearchers to more carefully account for contextin future studies, facilitates greater theoreticalintegration of the macro and micro levels of anal-ysis, and paves the way for new theoretical andmethodological developments.
REFERENCES
*Ainoya, N. 2004. Demographic diversity, team pro-cess, and team performance: Assessing moderatoreffects of cognitive conflict management practicesand task interdependence. Unpublished doctoraldissertation, University of Southern California.
Allison, P. D. 1978. Measures of inequality. AmericanSociological Review, 43: 865–880.
Allport, G. 1954. The nature of prejudice. Cambridge,MA: Addison-Wesley.
*Ancona, D. G., & Caldwell, D. F. 1992. Demography anddesign: Predictors of new product team performance.Organization Science, 3: 321–341.
*Balkundi, P., Kilduff, M., Barsness, Z. L., & Michael,J. D. 2007. Demographic antecedents and perfor-mance consequences of structural holes in workteams. Journal of Organizational Behavior, 28:241–260.
Bamberger, P. 2008. Beyond contextualization: Usingcontext theories to narrow the micro-macro gap inmanagement research. Academy of ManagementJournal, 51: 839–846.
Batt, R. 2002. Managing customer service: Human re-source practices, quit rates, and sales growth. Acad-emy of Management Journal, 45: 587–597.
*Baugh, S. G., & Graen, G. B. 1997. Effects of team genderand racial composition on perceptions of team per-formance in cross-functional teams. Group and Or-ganization Management, 22: 366–383.
Berger, J., Fisek, M. H., Norman, R. Z., & Zelditch, M.1977. Status characteristics in social interactions:An expectations states approach. New York:Elsevier.
Berger, J., Ridgeway, C., Fisek, M. H., & Norman, R. Z.1998. The legitimation and delegitimation of powerand prestige orders. American Sociological Review,63: 379–405.
Berger, J., Ridgeway, C., & Zelditch, M. 2002. The con-struction of status and referential structures. Socio-logical Theory, 20: 157–179.
Blau, P. M. 1977. Inequality and heterogeneity: A prim-itive theory of social structure. New York: FreePress.
Bourgeois, L. J. 1985. Strategic goals, perceived uncer-
622 JuneAcademy of Management Journal
tainty, and economic performance in volatile envi-ronments. Academy of Management Journal, 28:548–573.
Bowers, C. A., Pharmer, J. A., & Salas, E. 2000. Whenmember homogeneity is needed in work teams: Ameta-analysis. Small Group Research, 31: 305–327.
Brewer, M. B. 1988. A dual process model of impressionformation. In T. K. Srull & R. S. Wyer (Eds.), Ad-vances in social cognition, vol. 1: 1–36. Hillsdale,NJ: Erlbaum.
*Bunderson, J. S. 2003. Recognizing and utilizing exper-tise in work groups: A status characteristics perspec-tive. Administrative Science Quarterly, 48: 557–591.
Burns, T., & Stalker, G. M. 1961. The management ofinnovation. London: Tavistock.
*Cady, S. H., & Valentine, J. 1999. Team innovation andperceptions of consideration: What difference doesdiversity make? Small Group Research, 30: 730–750.
*Campion, M. A., Medsker, G. J., & Higgs, A. C. 1993.Relations between work group characteristics andeffectiveness: Implications for designing effectivework groups. Personnel Psychology, 46: 823–850.
*Campion, M. A., Papper, E. M., & Medsker, G. J. 1996.Relations between work team characteristics and ef-fectiveness: A replication and extension. PersonnelPsychology, 49: 429–452.
Cappelli, P., & Sherer, P. 1991. The missing role of con-text in OB: The need for a meso-level approach. InL. L. Cummings & B. M. Staw (Eds.), Research inorganizational behavior, vol.13: 55–110. Green-wich, CT: JAI Press.
Carpenter, M. A., Geletkanycz, M. A., & Sanders, W. G.2004. Upper echelons research revisited: Anteced-ents, elements, and consequences of top manage-ment team composition. Journal of Management,30: 749–778.
Chatman, J. A., & Flynn, F. 2001. The influence of demo-graphic heterogeneity on the emergence and conse-quences of cooperative norms in work teams. Acad-emy of Management Journal, 44: 956–974.
*Choi, J. N. 2007. Group composition and employeecreative behavior in a Korean electronics company:Distinct effects of relational demography and groupdiversity. Journal of Occupational and Organiza-tional Psychology, 80: 213–234.
Choi, J. N., Price, R. H., & Vinokur, A. D. 2003.Self-efficacy changes in groups: Effects of diversity,leadership, and group climate. Journal of Organiza-tional Behavior, 24: 357–372.
Cleveland, J. N., Festa, R. M., & Montgomery, L. 1988.Applicant pool composition and job perceptions:Impact on decisions regarding an older applicant.Journal of Vocational Behavior, 32: 112–125.
Cleveland, J. N., Montgomery, L., & Festa, R. M. 1984.
Group composition and job perceptions: Impact ondecision about sustainability of applicants forgroup membership. Unpublished manuscript, Ba-ruch College, City University of New York.
Collins, C. J., & Smith, K. G. 2006. Knowledge exchangeand combination: The role of human resource prac-tices in the performance of high-technology firms.Academy of Management Journal, 49: 544–560.
*Colquitt, J. A., Noe, R. A., & Jackson, C. L. 2002. Justicein teams: Antecedents and consequences of proce-dural justice climate. Personnel Psychology, 55: 83–109.
Combs, J., Liu, Y., Hall, A., & Ketchen, D. 2006. Howmuch do high-performance work practices matter? Ameta-analysis of their effects on organizational per-formance. Personnel Psychology, 59: 501–528.
Datta, D. K., Guthrie, J. P., & Wright, P. M. 2005. Humanresource management and labor productivity: Doesindustry matter? Academy of Management Journal,48: 135–145.
De Dreu, C. K. W., & Weingart, L. R. 2003. Task versusrelationship conflict, team performance, and teammember satisfaction: A meta-analysis. Journal ofApplied Psychology, 88: 741–749.
Devine, D. J., & Phillips, J. L. 2001. Do smarter teams dobetter? A meta-analysis of cognitive ability and teamperformance. Small Group Research, 32: 507–532.
DiTomaso, N., Post, C., & Parks-Yancy, R. 2007. Work-force diversity and inequality: Power, status, andnumbers. Annual Review of Sociology, 33: 473–501.
DiTomaso, N., Post, C., Smith, D. R., Farris, G. F., &Cordero, R. 2007. Effects of structural position onallocation and evaluation decisions for scientistsand engineers in industrial R&D. AdministrativeScience Quarterly, 52: 175–207.
*Drach-Zahavy, A., & Somech, A. 2002. Team heteroge-neity and its relationship with team support andteam effectiveness. Journal of Educational Admin-istration, 40: 44–66.
Drazin, R., & Van de Ven, A. 1985. Alternative forms of fitin contingency theory. Administrative ScienceQuarterly, 30: 514–539.
Dubin, R. 1976. Theory building in applied areas. InM. D. Dunnette (Ed.), Handbook of industrial andorganizational psychology: 17–39. Chicago: RandMcNally.
Eagly, A. H. 1995. The science and politics of comparingwomen and men. American Psychologist, 50: 145–158.
Eagly, A. H., & Steffen, V. J. 1984. Gender stereotypesstem from the distribution of women and men intosocial roles. Journal of Personality and SocialPsychology, 46: 735–754.
Ely, R. J. 1994. The effects of organizational demograph-ics and social identity on relationships among pro-
2009 623Joshi and Roh
fessional women. Administrative Science Quar-terly, 39: 203–238.
Ely, R. J. 1995. The power in demography: Women’ssocial constructions of gender identity at work.Academy of Management Journal, 38: 589–634.
*Ely, R. J. 2004. A field of group diversity, participationin diversity education programs, and performance.Journal of Organizational Behavior, 25: 755–780.
Fiske, S. T. 1993. Controlling other people: The impact ofpower on stereotyping. American Psychologist, 48:621–628.
Fiske, S. T. 1998. Stereotyping, prejudice, and discrimi-nation. In D. T. Gilbert, S. T. Fiske, & G. Lindzey(Eds.), Handbook of social psychology: 357–411.Boston: McGraw-Hill.
Fiske, S. T., Cuddy, A. J. C., Glick, P., & Xu, J. 2002. Amodel of (often mixed) stereotype content: Compe-tence and warmth respectively follow from per-ceived status and competition. Journal of Personal-ity and Social Psychology, 82: 878–902.
Frink, D. D., Robinson, R. K., Reithel, B., Arthur, M. M.,Ferris, G. R., Kaplan, D. M. et al. 2003. Gender de-mography and organizational performance: A twostudy investigation with convergence. Group andOrganization Management, 28: 127–147.
Gaertner, S. L., & Dovidio, J. F. 2000. Reducing inter-group bias: The common intergroup identitymodel. Philadelphia: Psychology Press.
Gibson, C., & Vermeulen, F. 2003. A healthy divide:Subgroups as a stimulus for team learning behavior.Administrative Science Quarterly, 48: 202–239.
Gomez-Mejia, L., & Balkin, D. 1992. Determinants of fac-ulty pay: An agency theory perspective. Academy ofManagement Journal, 35: 921–955.
Greer, L. L., & Jehn, K. A. 2008. Where perception meetsreality: The effects of different types of faultlineperceptions, asymmetries, and realities on inter-subgroup conflict and workgroup functioning.Working paper, Leiden University, The Netherlands.
Gully, S. M., Incalcaterra, K., Joshi, A., & Beaubien, J. M.2002. A meta-analysis of team efficacy, potency andperformance: Interdependence and level of analysisas moderators of observed relationships. Journal ofApplied Psychology, 87: 819–832.
Hackman, J. R. 1987. The design of work teams. In J. W.Lorsch (Ed.), Handbook of organizational behav-ior: 315–342. Englewood Cliffs, NJ: Prentice-Hall.
Haleblian, J., & Finkelstein, S. 1993. Top managementteam size, CEO dominance, and firm performance:The moderating role of environmental turbulenceand discretion. Academy of Management Journal,36: 844–863.
Hambrick, D. C., Cho, T. S., & Chen, M. 1996. The influ-ence of top management team heterogeneity onfirms’ competitive move. Administrative ScienceQuarterly, 41: 659–684.
Hambrick, D. C., & Finkelstein, S. 1987. Managerial dis-cretion: A bridge between polar views on organiza-tions. In L. L. Cummings & B. M. Staw (Eds.), Re-search in organizational behavior, vol. 9: 369–406.Greenwich, CT: JAI Press.
Hambrick, D. C., & Mason, P. A. 1984. Upper echelons:The organization as a reflection of its top managers.Academy of Management Review, 9: 193–206.
Harrison, D. A., & Klein, K. J. 2007. What’s the differ-ence? Diversity constructs as separation, variety, ordisparity in organizations. Academy of Manage-ment Review, 32: 1199–1228.
Harrison, D. A., Price, K. H., & Bell, M. P. 1998. Beyondrelational demography: Time and the effects of sur-face- and deep-level diversity on work group cohe-sion. Academy of Management Journal, 41: 96–107.
Harrison, D. A., Price, K. H., Gavin, J. H., & Florey, A. T.2002. Time, teams, and task performance: Changingeffects of surface- and deep-level diversity on groupfunctioning. Academy of Management Journal, 45:1029–1045.
Hedge, L. V., & Olkin, I. 1985. Statistical methods formeta-analysis. Orlando, FL: Academic Press.
*Herron, M. T. 1993. The effects of the ethnic andgender diversity of work team on the perceptionsof performance outputs. Unpublished doctoral dis-sertation, California School of Professional Psychol-ogy, Los Angeles.
Hilton, J. J., & von Hippel, W. 1996. Stereotypes. In J. T.Spence, J. M. Darley, & D. J. Foss (Eds.), Annualreview of psychology, vol. 47: 237–271. Palo Alto,CA: Annual Reviews.
Homan, A. C., & Van Knippenberg, D. 2003. The benefi-cial effects of cross-categorizing informational anddemographical diversity in groups. Paper pre-sented at the 11th European Congress of Work andOrganizational Psychology, Lisbon.
Horwitz, S. K., & Horwitz, I. B. 2007. The effects of teamdiversity on team outcomes: A meta-analytic reviewof team demography. Journal of Management, 33:967–1015.
Huffcutt, A. I., & Arthur, W., Jr. 1995. Development of anew outlier statistic for meta-analytic data. Journalof Applied Psychology, 80: 327–334.
Hunter, J. E., & Schmidt, F. L. 1990. Methods of meta-analysis: Correcting error and bias in researchfindings. Newbury Park, CA: Sage.
Hultin, M., & Szulkin, R. 1999. Wages and unequal accessto organizational power: An empirical test of genderdiscrimination. Administrative Science Quarterly,44: 453–472.
Ilgen, D. R., Hollenbeck, J. R., Johnson, M., & Jundt, D.2005. Teams in organization: From input-process-output models to IMOI models. In S. Fiske, D. L.Schacter, & A. Kasdin (Eds.), Annual review of psy-
624 JuneAcademy of Management Journal
chology, vol. 56: 517–543. Palo Alto, CA: AnnualReviews.
Information Technology Association of America. 2003.Report of the Information Technology Associationof America (ITAA) blue ribbon panel on IT diver-sity. Presented at the National IT Workforce Convo-cation, Arlington, VA.
*Jackson, S. E., & Joshi, A. 2004. Diversity in socialcontext: A multi-attribute, multi-level analysis ofteam diversity and sales performance. Journal ofOrganizational Behavior, 25: 675–702.
Jackson, S. E., Joshi, A., & Erhardt, N. L. 2003. Recentresearch on team and organizational diversity:SWOT analysis and implications. Journal of Man-agement, 29: 801–830.
Jackson, S. E., May, K. E., & Whitney, K. 1995. Under thedynamics of diversity in decision-making teams. InR. A. Guzzo & E. Salas (Eds.), Team effectivenessand decision making in organizations: 204–261.San Francisco: Jossey-Bass.
*Jarrell, K. A. 2002. Job-related diversity and serviceteam outcomes: New insights into the roles of taskstructure and process conflict. Unpublished doc-toral dissertation, Syracuse University.
*Jehn, K. A., & Bezrukova, K. 2004. A field study of groupdiversity, workgroup context, and performance.Journal of Organizational Behavior, 25: 703–729.
*Jehn, K. A., Northcraft, G. B., & Neale, M. A. 1999.Whydifferences make a difference: A field study of diver-sity, conflict, and performance in workgroups.Administrative Science Quarterly, 44: 741–763.
Johns, G. 2006. The essential impact of context on organ-izational behavior. Academy of Management Re-view, 31: 386–408.
*Joshi, A. 2002. How does context matter? Examiningthe process and performance outcomes of workteam heterogeneity. Unpublished doctoral disserta-tion, Rutgers University, New Brunswick, NJ.
Joshi, A., Liao, H., & Jackson, S. E. 2006. Cross-leveleffects of workplace diversity on sales performanceand pay. Academy of Management Journal, 49:459–481.
Jost, J. T., & Banaji, M. R. 1994. The role of stereotypingin system-justification and the production of falseconsciousness. British Journal of Social Psychol-ogy, 33: 1–27.
*Kearney, E., & Gerbert, D. 2009. Managing diversity andenhancing team outcomes: The promise of transfor-mational leadership. Journal of Applied Psychol-ogy: In press.
*Kearney, E., Gerbert, D., & Voelpel, S. 2009. When andhow diversity benefits teams: The importance ofteam members’ need for cognition. Academy ofManagement Journal, 52: 581–598.
Keck, S. 1997. Top management team structure: Differ-
ential effects by environmental context. Organiza-tion Science, 8: 143–156.
*Keller, R. T. 2001. Cross-functional project groups inresearch and new product development: Diversity,communications, job stress and outcomes. Academyof Management Journal, 44: 547–555.
*Kirkman, B. L., Tesluk, P. E., & Rosen, B. 2004. Theimpact of demographic heterogeneity and team lead-er-team member demographic fit on team empower-ment and effectiveness. Group and OrganizationManagement, 29: 334–368.
Kochan, T., Bezrukova, K., Ely, R., Jackson, S., Joshi, A.,Jehn, K. et al. 2003. The effects of diversity on busi-ness performance: Report of the diversity researchnetwork. Human Resource Management, 42: 3–21.
Larkey, L. K. 1996. Toward a theory of communicativeinteractions in culturally diverse workgroups. Acad-emy of Management Review, 21: 463–491.
Lau, D. C., & Murnighan, J. K. 1998. Demographic diver-sity and faultlines: The compositional dynamics oforganizational groups. Academy of ManagementReview, 23: 325–340.
Lau, D. C., & Murnighan, J. K. 2005. Interactions withingroups and subgroups: The effects of demographicfaultlines. Academy of Management Journal, 48:645–659.
Lawrence, P., & Lorsch, J. 1967. Organization and itsenvironment. Cambridge, MA: Harvard UniversityPress.
*Leonard, J. S., Levine, D. I., & Joshi, A. 2004. Do birds ofa feather shop together? The effects on performanceof employees’ similarity with one another and withcustomers. Journal of Organizational Behavior, 25:731–754.
LePine, J. A., Hanson, M. A., Borman, W. C., & Motow-idlo, S. J. 2000. Contextual performance and team-work: Implications for staffing. In G. R. Ferris & K. M.Rowland (Eds.), Research in personnel and humanresources management, vol. 19: 53–90. Greenwich,CT: JAI Press.
Li, J., & Hambrick, D. C. 2005. Factional groups: A newvantage on demographic faultlines, conflict, and dis-integration in work teams. Academy of Manage-ment Journal, 48: 794–813.
Linnehan, F., & Konrad, A. M. 1999. Diluting diversity:Implications for intergroup inequality in organiza-tions. Journal of Management Inquiry, 8: 399–414.
Lipsey, M. W., & Wilson, D. B. 2001. Practical meta-analysis. Thousand Oaks, CA: Sage.
*Lovelace, K., Shapiro, D. L., & Weingart, L. R. 2001.Maximizing cross-functional new product teams’ in-novativeness and constraint adherence: A conflictcommunications perspective. Academy of Manage-ment Journal, 44: 779–793.
*Marsteller, J. A. 2003. The relationship between non-racial diversity in team composition and perfor-
2009 625Joshi and Roh
mance in creativity in a chronic illness carequality improvement intervention. Unpublisheddoctoral dissertation, University of California,Berkeley.
Martins, L. L., Miliken, F. J., Wiesenfeld, B. M., & Sal-gado, S. R. 2003. Racioethnic diversity and groupmember’s experience. Group and OrganizationManagement, 28: 75–106.
Mayo, M., Pastor, J., & Meindl, J. R. 1996. The effects ofgroup heterogeneity on the self-perceived efficacy ofgroup leaders. Leadership Quarterly, 7: 265–284.
McGrath, J. E. 1984. Groups: Interaction and perfor-mance. Englewood Cliffs, NJ: Prentice-Hall.
Milkovich, G. T. 1987. Compensation systems in hightechnology companies. In D. B. Balkin & L. R. Go-mez-Mejia (Eds.), New perspectives on compensa-tion. Englewood Cliffs, NJ: Prentice-Hall.
Milliken, F. J., & Martins, L. L. 1996. Searching for com-mon threads: Understanding the multiple effects ofdiversity in organizational groups. Academy ofManagement Review, 21: 402–433.
Organisation for Economic Cooperation and Develop-ment. 2006. OECD science, technology and indus-try outlook. Organisation for Economic Cooperationand Development.
Orwin, R. G. 1983. A fail-safe N for effect size inmeta-analysis. Journal of Educational Statistics, 8:157–159.
Pelled, L. H. 1996. Relational demography and percep-tions of group conflict and performance: A fieldinvestigation. International Journal of ConflictManagement, 7: 230–246.
*Pelled, L. H., Eisenhardt, K. M., & Xin, K. R. 1999.Exploring the black box: An analysis of work groupdiversity, conflict, and performance. AdministrativeScience Quarterly, 44: 1–28.
Perry, E. L. 1997. A cognitive approach to understandingdiscrimination: A closer look at applicant genderand race. In G. R. Ferris (Ed.), Research in personneland human resource management, vol. 15: 175–240. Greenwich, CT: JAI Press.
Pettigrew, T. F. 1998. Intergroup contact theory. In J. T.Spence, J. M. Darley, & D. J. Foss (Eds.), Annualreview of psychology, vol. 49: 65–85. Palo Alto, CA:Annual Reviews.
Porter, M. E. 1980. Competitive strategy. New York: FreePress.
*Portero-Brown, R. A. 1999. The missing link betweenteam heterogeneity and effectiveness: Individualand contextual influences on social categorysalience. Unpublished doctoral dissertation, Uni-versity of California, Berkeley.
*Puck, J., Rygl, D., & Kittler, M. 2006. Cultural anteced-ents and performance consequences of open commu-nication and knowledge transfer in multiculturalprocess-innovation teams. Journal of Organiza-
tional Transformation and Social Change, 3:223–241.
Quinn, J. B., Anderson, P., & Finkelstein, S. 1996. Leverag-ing intellect. Academy of Management Executive,10(3): 7–27.
Ragins, B., & Sundstorm, E. 1990. Gender and powerorganizations: A longitudinal perspective. Psycho-logical Bulletin, 105: 51–88.
*Raver, J. L., & Gelfand, M. J. 2005. Beyond the individualvictim: Linking sexual harassment, team processes,and team performance. Academy of ManagementJournal, 48: 387–400.
*Reagans, R., & Zuckerman, E. W. 2001. Networks, diver-sity, and productivity: The social capital of corporateR&D teams. Organization Science, 12: 502–517.
Reagans, R., Zuckerman, E. W., & McEvily, B. 2004. Howto make the team: Social networks vs. demographyas criteria for designing effective teams. Administra-tive Science Quarterly, 49: 101–133.
Reskin, B. F., McBrier, D. B., & Kmec, J. 1999. The deter-minants and consequences of workplace sex andrace composition. In J. Hagan & K. Cook (Eds.), An-nual review of sociology, vol. 25: 355–361. PaloAlto, CA: Annual Reviews.
Richard, O. C. 2000. Racial diversity, business strategy,and firm performance: A resource-based view. Acad-emy of Management Journal, 43: 164–177.
Richard, O. C., Barnett, T., Dwyer, S., & Chadwick, K.2004. Cultural diversity in management, firm perfor-mance, and the moderating role of entrepreneurialorientation dimensions. Academy of ManagementJournal, 47: 255–266.
Richard, O. C., Murthi, B. P. S., & Ismail, K. 2007. Theimpact of racial diversity on intermediate and long-term performance: The moderating role of environ-mental context. Strategic Management Journal, 28:1213–1233.
Ridgeway, C. L. 1991. The social construction of statusvalue: Gender and other nominal characteristics. So-cial Forces, 70: 367–386.
Ridgeway, C. L. 1997. Interaction and the conservation ofgender inequality: Considering employment. Amer-ican Sociological Review, 62: 218–235.
Rosen, B., & Jerdee, T. H. 1977. Too old or not too old.Harvard Business Review, 55(6): 97–106.
Rosenthal, R. 1979. The “file drawer problem” and toler-ance of null results. Psychological Bulletin, 86:638–641.
Rousseau, D. M., & Fried, Y. 2001. Location, location,location: Contextualizing organizational research.Journal of Organizational Behavior, 22: 1–13.
Saavedra, R., Earley, P., & Van Dyne, L. 1993. Complexinterdependence in task-performing groups. Journalof Applied Psychology, 78: 61–72.
*Sacco, J. M., & Schmitt, N. A. 2005. A dynamic multi-
626 JuneAcademy of Management Journal
level model of demographic diversity and misfit ef-fects. Journal of Applied Psychology, 90: 203–231.
*Schippers, M. C., Den Hartog, D. N., Koopman, P. L., &Wienk, J. A. 2003. Diversity and team outcomes: Themoderating effects of outcome interdependence andgroup longevity and the mediating effect of reflexiv-ity. Journal of Organizational Behavior, 24: 779–802.
Shea, G. P., & Guzzo, R. A. 1987. Groups as humanresources. In K. M. Rowland & G. R. Ferris (Eds.),Research in personnel and human resource man-agement, vol. 5: 323–356. Greenwich, CT: JAI Press.
Shore, L. M., & Goldberg, C. B. 2005. Age discriminationin the workplace. In R. L. Dipboye & A. Colella(Eds.), Discrimination at work: The psychologicaland organizational bases: 203–225. Mahwah, NJ:Erlbaum.
Sidanius, J. 1993. The psychology of group conflict andthe dynamics of oppression: A social dominanceperspective. In S. Iyengar & W. J. McGuire (Eds.),Explorations in political psychology: 183–219.Durham, NC: Duke University Press.
Skaggs, S., & DiTomaso, N. 2004. Understanding the ef-fects of workforce diversity on employment out-comes: A multidisciplinary and comprehensiveframework. In N. DiTomaso & C. Post (Eds.), Re-search in the sociology of work: 279–306. NewYork: Elsevier.
*Somech, A. 2006. The effects of leadership style andteam process on performance and innovation infunctionally heterogeneous teams. Journal of Man-agement, 32: 132–157.
Teachman, J. D. 1980. Analysis of population diversity.Social Methods and Research, 8: 341–362.
Thompson, J. D. 1967. Organizations in action. NewYork: McGraw-Hill.
*Tyran, K. L., & Gibson, C. B. 2008. Is what you see, whatyou get? The relationship among surface- and deep-level heterogeneity characteristics, group efficacy,and team reputation. Group and Organization Man-agement, 33: 46–76.
U. S. Bureau of Labor Statistics. 2006. Labor force sta-tistics from the Current Population Survey. Wash-ington, DC: U. S. Department of Labor.
U. S. Bureau of Labor Statistics. 2007. Occupationaloutlook handbook, 2006–2007. Washington, DC:U.S. Department of Labor.
U. S. Census Bureau. 2002. North American IndustryClassification System (NAICS)–Revisions for 2002.Washington, DC: U. S. Census Bureau.
*Van der Vegt, G. S., & Bunderson, J. S. 2005. Learningand performance in multidisciplinary teams: Theimportance of collective team identification. Acad-emy of Management Journal, 48: 532–547.
Van der Vegt, G. S., & Janssen, O. 2003. Joint impact of
interdependence and group diversity on innovation.Journal of Management, 29: 29–51.
Van der Vegt, G. S., Van de Vliert, E., & Huang, X. 2005.Location-level links between diversity and innova-tive climate depend on national power distance.Academy of Management Journal, 48: 1171–1182.
Van de Ven, A. H., Delbecq, A. L., & Koenig, R. 1976.Determinants of coordination modes within organi-zations. American Sociological Review, 41: 322–338.
Van Knippenberg, D., De Dreu, C. K. W., & Homan, A. C.2004. Work group diversity and group performance:An integrative model and research agenda. Journalof Applied Psychology, 89: 1008–1022.
Van Knippenberg, D., & Schippers, M. C. 2007. Work-group diversity. In M. I. Posner & M. K. Rothbart(Eds.), Annual review of psychology, vol. 58: 2.1–2.27. Palo Alto, CA: Annual Reviews.
Watson, W. E., Johnson, L., & Merritt, D. 1998. Teamorientation, self-orientation, and diversity in taskgroups: Their connection to team performance overtime. Group and Organization Management, 23:161–282.
Webber, S. S., & Donahue, L. M. 2001. Impact of highlyand less job-related diversity on work group cohe-sion and performance: A meta-analysis. Journal ofManagement, 27: 141–162.
Whitener, E. M. 1990. Confusion of confidence intervalsand credibility intervals in meta-analysis. Journal ofApplied Psychology, 75: 315–321.
Williams, K., & O’Reilly, C. A. 1998. Demography anddiversity in organizations: A review of 40 years ofresearch. In B. M. Staw & L. L. Cummings (Eds.),Research in organizational behavior, vol. 20: 77–140. Greenwich, CT: JAI Press.
*Yeh, Y., & Chou, H. 2005. Team composition and learn-ing behaviors in cross-functional teams. SocialBehavior and Personality, 33: 391–402.
Aparna Joshi ([email protected]) is an assistant pro-fessor of labor and employment relations at the Univer-sity of Illinois, Urbana-Champaign. She received herPh.D. from the School of Management and Labor Rela-tions, Rutgers University. She conducts research in thearea of work team diversity, global and distributedteams, team social capital, and generational issues inthe workplace.
Hyuntak Roh ([email protected]) is a Ph.D. candidatein the School of Labor and Employment Relations at theUniversity of Illinois at Urbana-Champaign. His researchinterests include workforce diversity, group process andperformance, employee turnover, and multilevel contex-tual issues in HR/OB research.
2009 627Joshi and Roh