Academy of Management The Academy of Management Journal

Comparing Alternative Conceptualizations of Functional Diversity in Management Teams:
Process and Performance Effects

Author(s): J. Stuart Bunderson and Kathleen M. Sutcliffe
Source: The Academy of Management Journal, Vol. 45, No. 5 (Oct., 2002), pp. 875-893
Published by: Academy of Management
Stable URL: http://www.jstor.org/stable/3069319
Accessed: 08-12-2017 02:08 UTC
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms
Academy of Management is collaborating with JSTOR to digitize, preserve and extend access
to The Academy of Management Journal
This content downloaded from 137.111.13.200 on Fri, 08 Dec 2017 02:08:56 UTC
All use subject to http://about.jstor.org/terms
? Academy of Management Journal
2002, Vol. 45, No. 5, 875-893.
COMPARING ALTERNATIVE CONCEP1 UALIZATIONS OF

FUNCTIONAL DIVERSITY IN MANAGEMENT TEAMS:
PROCESS AND PERFORMANCE EFFECTS
J. STUART BUNDERSON
Washington University in St. Louis
KATHI.EEN M. SUTCLIFFE
University of Michigan
Functional diversity in teams has been conceptualized in a variety of ways without

careful attention to how different conceptualizations might lead to different results. We
examined the process and performance effects of dominant function diversity (the
diversity of functional experts on a team) and intrapersonal functional diversity (the
aggregate functional breadth of team members). In a sample of business unit manage-
ment teams, dominant function diversity had a negative, and intrapersonal functional
diversity, a positive effect on information sharing and unit performance. These find-
ings suggest that different forms of functional diversity can have very different impli-
cations for team process and performance and that intrapersonal functional diversity
matters for team effectiveness.
Empirical research on functional diversity in concluded that functional diversity is a double-

teams has presented a complex picture. On the one edged sword, in that it has positive implications in
hand, researchers argue that by broadening the some contexts and for some process or performance
range of experience and expertise available to a variables but negative implications in other con-
team, functional diversity can promote team effec- texts and for other process or performance variables
tiveness. Supporting this argument, empirical stud- (Milliken & Martins, 1996: 403; Tsui, Egan, & Xin,
ies have shown that functionally diverse teams can 1995; Williams & O'Reilly, 1998).
be more innovative (Bantel & Jackson, 1989), can This article introduces the possibility that the
develop clearer strategies (Bantel, 1993), can re- positive or negative effects of functional diversity
spond more aggressively to competitive threats may not just be a function of the dependent vari-
(Hambrick, Cho, & Chen, 1996), and can be quicker able or context examined but may also be a func-
to implement certain types of organizational tion of the way in which functional diversity is
change (Williams, Hoffman, & Lamont, 1995) than conceptualized and measured. In much of the ex-
functionally homogeneous teams. On the other isting work on functional diversity in teams, re-
hand, researchers have argued that because func- searchers have conceptualized functional diversity
tional diversity is associated with differences of as the distribution of team members across a range
opinion and perspective, functional differences can of relevant functional categories, overlooking the
inhibit team process and/or effectiveness. Empiri- extent to which these individuals are narrow func-
cal research also seems to support this argument, tional specialists or broad generalists with experi-
indicating that functional diversity can increase ence in a range of functional areas. We explore the
conflict (Knight et al., 1999; Pelled, Eisenhardt, & thesis that, within the same context, the former
Xin, 1999), slow competitive response (Hambrick conceptualization of functional diversity can have
et al., 1996), and even compromise performance negative, and the latter conceptualization, positive
(Murray, 1989; Simons, Pelled, & Smith, 1999). implications for team process and performance.
Given this pattern of results, team researchers have Teams composed of specialists from different func-
tional areas may be unable to exploit their diverse
expertise because of cross-functional communica-
tion and coordination problems. In contrast, teams
We thank Joe Banas, Bill Bottom, Kurt Dirks, Greg
Northcraft, Melissa Thomas-Hunt, and our anonymous composed of individuals with a breadth of func-
AMJ reviewers for helpful comments on this article and tional experiences may be better able to overcome
Mike Ferdinandi and David Hatch for help with data communication barriers (because team members
collection. can relate to one another's functions) while still
875
TABLE 1
Alternative Conceptualizations of Functional Diversity in the Emp
Functional Diversity Variable:

Study Dominant Function Diversity Core Proposition Data
(la) Dominant Function Diversity: Ten Studies

Bantel and Jackson (1989) Blau index from survey data Dominant function diversity -* n = TMTs of
(obtained from HR executive). innovation adoptions function d
(especially
Bantel (1993) Blau index from survey data Dominant function diversity -> n = TMTs of
(obtained from HR executive). strategic clarity function d
Blau index using archival data.

Carpenter and Frederickson (2001) Dominant function diversity - firm n = 207 TMT
globalization (moderated by TMTs wer
environmental uncertainty) were less
was high
low.
Blau index using archival data.

Hambrick, Cho, and Chen (1996) Dominant function diversity - n = All com
competitive action/response industry
function-
respond i
comprom
was also r
Knight, Pearce, Smith, Olian, Blau index using

Sims, survey data.
Smith, and Dominant function diversity -* n = TMTs o
Flood (1999) strategic consensus dominant
.05) strat
conflict (
moderate
function
Michel and Hambrick (1992) Blau index using archival data. Diversification posture - n = TMTs o
performance (moderated by did not m
dominant function diversity) diversific
Murray (1989) Blau and Shannon indexes using Dominant function diversity -* n = TMTs o
archival data. performance (efficiency/short-term function
and adaptability/long-term) term perf
Dominant
to long-te
Pelled, Eisenhardt, and Xin (1999) Shannon index from survey data. Dominant function diversity -* task n = 45 new
conflict -> performance teams fro
diversity
conflict (
task rout
was not r
Smith, Smith, Olian, Sims, O'Bannon, and Scully Blau index from survey data. Dominant function diversity - n = TMTs o
(1994) TMT process (social integration, function
informal communication, or perfor
communication frequency) ->
TMT performance
Williams, Hoffman, and Lamont (1995) Blau index from archival data. Dominant function diversity -> n = TMTs of 7
M-form implementation time TMTs with h
implemented
(p < .10).
(lb) Functional Background Diversity: Two Studies

Glick, Miller, and Huber (1993) Based on complete work histories. Functional background diversity n = TMTs of 79 busine
Euclidean distances between all -- diversity of beliefs about functional backgroun
TMT members. Survey data. effectiveness -* beliefs about effectiven
communication communication
Sutcliffe (1994) Based on complete work histories. Functional background diversity n = 67 TMTs. TMTs with h
Euclidean distance measure. -> accuracy of environmental diversity were less ac
Survey data. perceptions environment.
(lc) Functional Assignment Diversity: Five Studies
Ancona and Caldwell (1992) Shannon index from survey and Functional assignment diversity n = 45 new product teams
archival data. -- communication with other Functional assignment div
groups -* self- and manager extern
ratings of effectiveness ratings of innov
Keck (1997) Blau index from archival data. Functional assignment diversity n = TMTs from 56 cement an
-> performance over time over 65 and 12 yea
(moderated by environmental assignment-divers
stability) turbulent environments,
assignme
environments
Keck and Tushman (1993) Blau index from archival data. TMT context -* functional n = TMTs of 104 firms in th
assignment diversity years. Periods of env
lower func
performan
assignment
Lant, Milliken, and Batra (1992) Blau index from archival data. Functional assignment diversity n = TMTs from 40 furnit
strategic reorientation firms. High-functio
were more
in both indus
Simons, Pelled, and Smith (1999) Blau index from survey data. Functional assignment diversity n = TMTs from 57 manuf
x debate -> firm financial assignment d
performance (debate as the sales and profits. F
moderator) alone related negatively
a "TMT" is "top management team"; "MNC" is "multinational oorporation."
878 Academy of Management Journal October
realizing the performance benefits of diverse func- distributions of functional affiliations or experi-
tional experiences. ences (e.g., Chaganti & Sambharya, 1987).
The purpose of this study was to develop and The results of our literature review are reported
empirically examine this thesis. Drawing on theo- in Table 1. We identified 17 empirical studies that
ries of motivation and social categorization as well included functional diversity as a key variable. Fif-
as research on groups and teams, we developed a teen of these studies examined functional diversity
model in which information sharing mediates the as an antecedent of team processes (such as con-
relationship between these distinct forms of func- sensus, conflict, social integration, and communi-
tional diversity and unit performance. This model cation) and/or team outcomes (such as innovation,
was tested using data from a sample of 44 business strategic clarity, financial performance, speed of
unit management teams in a Fortune 100 consumer response, and accuracy of perceptions). One study
products company. Results underscore the need for examined functional diversity as a moderator
greater care in the definition and measurement of (Michel & Hambrick, 1992), and one study exam-
functional diversity and highlight the significance ined selected antecedents of functional diversity
of intrapersonal functional diversity for team pro- (Keck & Tushman, 1993).
cess and performance. Analysis of these 17 studies reveals three differ-
ent conceptualizations of functional diversity: (1)
diversity in the different functional areas within
THEORY AND HYPOTHESES which team members have spent the greater part of
their careers (dominant function diversity), (2) di-
A review of published research examining func-
versity in the complete functional backgrounds of
tional diversity in teams reveals two important pat-
team members (functional background diversity),
terns. First, although different studies of functional
and (3) diversity in team member functional assign-
diversity presumably examine the same basic con-
ments (functional assignment diversity). In the fol-
struct, they adopt very different conceptualizations
lowing paragraphs, we summarize the core logic
of that construct. It is therefore important, as Pfeffer
and research findings associated with each of these
suggested, "to be sensitive to the use of multiple
three conceptualizations of functional diversity.
measures of demographic characteristics and to de-
Dominantfunction diversity. Ten studies (59%)
veloping theoretical predictions that are related to
examine the extent to which team members differ
the specific measures being developed" (1983:
in the functional areas within which they have
310). Second, although a few researchers have rec-
spent the greater part of their careers. We refer to
ognized the potential significance of examining the
this form of functional diversity as dominant func-
functional breadth of individual team members
tion diversity. This approach to conceptualizing
(see Burke & Steensma, 1998), there have been no
functional diversity rests on the assumption that
attempts to empirically examine this aspect of
each member brings a specific functional perspec-
functional diversity in teams.
tive to a team, a perspective gained through expe-
rience that is typically weighted toward a particu-
lar function. The key question then becomes
Distinguishing between Different
whether the different functional perspectives (that
Conceptualizations of Functional Diversity
is, different dominant functions) among team mem-
We reviewed published research on functionalbers cover some relevant range of functional cate-
diversity in management teams and other multi-gories or are restricted to a subset of those catego-
functional teams in order to determine how func- ries. The extent to which the dominant functions of
tional diversity has been conceptualized and mea- a team's members are evenly distributed across all
sured and to understand the results that have been of the relevant functions is viewed as an indication
reported. Relevant articles were identified through of the team's breadth and balance of knowledge and
keyword searches of on-line databases and reviews expertise related to running all aspects of an organ-
of academic journals in management and organiza- ization. Empirical research using this measure of
tion. Search results were cross-referenced with functional diversity suggests that dominant func-
published reviews of the literature on team diver- tion diversity is associated with innovation (Bantel
sity (e.g., Milliken & Martins, 1996; Tsui et al., & Jackson, 1989), strategic clarity (Bantel, 1993),
1995; Williams & O'Reilly, 1998) as well as refer-slower competitive response (Hambrick et al.,
ence lists in published articles. Since our focus was 1996), lower consensus (Knight et al., 1999), higher
on team functional diversity, we did not include conflict (Knight et al., 1999; Pelled et al., 1999),
lower near-term but higher long-term performance
studies at the individual or interpersonal levels or
team-level studies that considered means but not (Hambrick et al., 1996; Murray, 1989), a less global
2002 Bunderson and Sutcliffe 879
strategy' (Carpenter & Frederickson, 2001), and distance between pairs of team members in the
quicker implementation of the M-form structure amount of time spent in each function.2
(Williams et al., 1995). There is a logical relationship between func-
Dominant function diversity is operationally de- tional background diversity and dominant function
fined by determining (or asking respondents to de- diversity. Specifically, teams composed of individ-
uals whose dominant functions cover all of the
termine) the functional areas within which team
members have spent the greater part of their careers functional bases (dominant function diversity i
or that best represent their dominant functional high) are more likely to be composed of individuals
career tracks. Once each team member has been with different functional backgrounds (functional
background diversity is high). The two measures
categorized into a functional area, some version of
clearly provide unique information, however, since
Blau's (1977) or Shannon's (1948) heterogeneity
it is possible to have a team whose members hav
index is used to compute dominant function diver-
generally similar functional backgrounds but very
sity. This approach does not address or assign any
different dominant functions. This suggests that th
weight to experiences that lie outside an individu- choice between dominant function diversity and
al's dominant functional career track.
functional background diversity is an important
Functional background diversity. In two (12%) research design decision that may influence the
of the studies summarized in Table 1, functional
results that are obtained and the way in which
diversity is conceptualized as the degree of differ- those results should be interpreted. Furthermore
ence in the complete functional backgrounds of we suspect that in most cases, researchers wil
team members. We refer to this form of functional want to select one or the other conceptualization
diversity as functional background diversity. Like rather than employ both in order to avoid problem
dominant function diversity, functional back- of construct overlap and multicollinearity.
ground diversity focuses on the different functional Functional assignment diversity. In five (29%)
experiences of team members. But whereas domi- of the studies summarized in Table 1, functional
nant function diversity focuses on the distribution diversity is conceptualized as diversity in the func-
of dominant functions across some range of functional assignments of team members. We refer to
tional categories, functional background diversity this form of functional diversity as functional as-
focuses simply on the extent to which team mem- signment diversity. The key issue with this concep-
tualization is not whether team members have ex-
bers differ in their functional backgrounds. The
underlying assumption is that different functional perience in different functional areas but whether
backgrounds imply nonoverlapping knowledge their current functional assignments cover some
and expertise, which suggests that team members relevant range of functional categories. By looking
have a broader pool of resources from which to at diversity in current assignments rather than
functional backgrounds, researchers hope to under-
draw in making decisions and taking action. Key stand how the breadth and mix of functional ac-
findings from these two studies are that functional
countabilities on a team relate to team processes
background diversity is positively associated with
and outcomes. Empirical research on functional
a diversity of beliefs and perceptions (Glick, Miller,
assignment diversity suggests that it is positively
& Huber, 1993; Sutcliffe, 1994) and with the ten- related to external communication (Ancona & Cald-
dency to communicate more frequently (Glick et well, 1992), performance in turbulent environ-
al., 1993).
ments (Keck, 1997), sustained performance (Keck &
Functional background diversity is operationally Tushman, 1993), likelihood of strategic reorienta-
defined by reviewing the work histories of each team tion (Lant, Milliken, & Batra, 1992), and (when
member (obtained from surveys or archives) in order accompanied by open debate and dialogue), firm
to determine the amount of time spent in each of the profitability (Simons et al., 1999).
different functional areas. Functional background di- Functional assignment diversity is operationally
versity is then computed as the average Euclidean defined by assigning each team member to one
functional area on the basis of job title and/or re-
sponsibilities. Some version of either the Blau (1977)
1 Although Carpenter and Frederickson (2001) re-

ported an overall negative relationship between domi- 2 Since this approach does not take differences in ten-
nant function diversity and firm globalization, this rela- ure into account, an alternative would be to compute
tionship became positive under conditions of low Euclidean distances based on the proportion of an indi-
environmental uncertainty. vidual's work history spent in each functional area.
or the Shannon (1948) heterogeneity index is then making biases such as escalation of commitment
used to compute functional assignment diversity. and overconfidence. These propositions have not
In sum, empirical work on functional diversity in been directly tested, although they are intuitively
management teams has adopted three different con- appealing and indirectly supported by some evi-
ceptualizations of functional diversity: dominant dence. For example, Rulke (1996) found that teams
function diversity, functional background diver- of MBA students formed using a functional gener-
sity, and functional assignment diversity. These alist selection strategy performed better at a man-
three conceptualizations focus on different expres- agement simulation exercise than did teams formed
sions of functional diversity and are grounded in using a functional specialist selection strategy.
different assumptions about which form of func- In sum, the significance of intrapersonal func-
tional diversity matters most. tional diversity for management team process and
performance is a largely unexplored phenomenon in
organizational research. Consequently, we do not
Intrapersonal Functional Diversity
know how this form of functional diversity compares
A review of the literature summarized in Table 1 with more traditional forms (summarized earlier) in
suggests that existing empirical research has over- its effect on team process and performance.
looked a fourth, potentially very important form of
functional diversity in teams-the diversity repre-
A Research Model and Hypotheses
sented in the functional backgrounds of individual
team members. This conceptualization of func- The above review raises an important question
tional diversity-which we refer to as intrapersonal about the study of functional diversity in manage-
functional diversity-focuses on the extent to ment and cross-functional teams: How do the vari-
which the individuals on a team are narrow func- ous conceptualizations of functional diversity
tional specialists with experience in a limited range found in the literature differ in their implications
of functions, or broad generalists whose work ex- for team process and performance? This question
periences span a range of functional domains. We cannot be answered using existing empirical re-
suggest that this fourth conceptualization of func- search, since variance in results across studies may
tional diversity can have significant implicationsbe a consequence of different process or perfor-
for team process and performance. mance variables or different research settings rather
Although there have been no attempts to empir- than (or in addition to) different conceptualizations
ically examine the significance of intrapersonal of functional diversity. To begin to address the
functional diversity for teams, there have been aabove question, one would need to compare the
few scattered attempts to examine its significance effects of at least two different forms of functional
for individual managers. For example, Campion, diversity on the same process and performance
Cheraskin, and Stevens (1994) found that experi- variables in the same research setting. Further-
ence in a range of functional domains was posi- more, this comparison should be grounded in
tively associated with salary level, promotion op- sound theoretical arguments that make clear why
portunities, overall positive affect, and perceptions effects can be expected to differ across conceptual-
of skill acquisition. And whereas Walsh (1988) andizations of functional diversity. In line with this
Beyer and colleagues (Beyer, Chattopadhyay, research agenda, the present section outlines a spe-
George, Glick, ogilvie, & Pugliese, 1997) found that cific model and set of hypotheses comparing two of
intrapersonal functional diversity was not signifi- the four forms of functional diversity described
cantly associated with the way managers perceived above.
an experimental problem situation, Hitt and Tyler Research on functional diversity in teams has
(1991) found that executives with broad functional typically adopted some form of the basic input-
backgrounds evaluated strategic acquisitions differ- process-outcome model of group effectiveness (e.g.,
ently than did executives with narrower functional Guzzo & Shea, 1992; Hackman & Morris, 1975). In
backgrounds. the simple form of this model, group characteristics
In a recent conceptual paper, Burke and and context factors (such as functional diversity
Steensma (1998) suggested that intrapersonal func- and task characteristics) influence patterns of be-
tional diversity is important for management teams havior and interaction within a group (such as con-
and not just for individual managers. They argued flict, communication, and cohesion) that, in turn,
that management teams composed of people with affect the outcomes achieved by the group (such as
wide-ranging functional backgrounds will have competitive response, innovation, and perfor-
broader "dominant logics" (Prahalad & Bettis, mance). But although all of the studies cited above
1986) and will be less susceptible to decision- and summarized in Table 1 are grounded in this
model, only a few represent attempts to "open the focused on a variety of issues related to team fun
black box" (Lawrence, 1997) of group process and tioning, including the effect of functional diversi
empirically examine the mediated relationships on team process and performance. Data from thes
that are assumed to exist (e.g., Pelled et al., 1999). interviews were used to inform and enrich the de-
The model proposed here was designed to exam- velopment of our theory and hypotheses, and se-
ine the role of information sharing in mediating the lected quotes appear in the following discussion to
relationship between different forms of functional illustrate or elaborate specific points.
diversity and performance outcomes. Information Intrapersonal functional diversity and infor-
sharing involves conscious and deliberate attempts mation sharing. Researchers interested in under-
on the part of team members to exchange work- standing the relationship between team diversity
related information, keep one another apprised of (of whatever type) and team process or perfor-
activities, and inform one another of key develop- mance often invoke social categorization theory to
ments. Since information related to both internal support their arguments or explain their findings
and external issues is unequally distributed among(Williams & O'Reilly, 1998). Social categorization
team members (given the division of labor and ac- theory suggests that individuals seek to maintain
countability), their ability and willingness to con-high self-esteem by defining themselves in ways
tinually share pertinent information is critical if the that lead to favorable social comparisons. Individ-
team is to make decisions and take actions that uals place themselves and others into social groups
appropriately address the range of relevant factors (defined in terms of age, background, status, and so
(see Argote, Gruenfeld, & Naquin, 1999). It is there- forth) and then attribute positive characteristics to
fore not surprising that research has tended to sup- their own groups (their in-group) and negative
port the significance of information sharing for characteristics to other groups (out-groups). These
team effectiveness (Ancona & Caldwell, 1992; stereotypes and biases can severely restrict commu-
Eisenhardt, 1989; Gladstein, 1984). nication, undermine cohesion, and impede collab-
Our approach was to examine the different ef- orative problem solving (Tajfel, 1981; Turner,
fects of dominant function diversity and intraper- 1982).
sonal functional diversity on information sharing Since management and cross-functional teams
and unit performance. We focused on these two are composed of individuals who are responsible
forms of functional diversity for several reasons. for different functional areas, stereotyping and in-
First, dominant function diversity is by far the mostgroup/out-group biases resulting from functional
common conceptualization of team functional di- background differences are a potential problem. If
versity in the literature and therefore constitutes a team members develop negative stereotypes about
useful standard against which to compare alterna- individuals in other functions, these stereotypes
tive conceptualizations. Second, intrapersonal can restrict communication and undermine group
functional diversity provides a nice theoretical cohesion. In contrast, if an individual sees himself
counterpoint to dominant function diversity since or herself as similar to other team members in terms
dominant function diversity concerns a team's of functional background, positive attributions will
breadth of experience across functional categories, likely be made and the lines of communication
whereas intrapersonal functional diversity con- opened (Williams & O'Reilly, 1998).
cerns the breadth of the functional experiences of We suggest that teams composed of individuals
the individuals on a team. Third, intrapersonal with broad functional backgrounds will be less sus-
functional diversity has been generally overlooked ceptible to functionally grounded biases and ste-
in the empirical literature, suggesting an opportu- reotypes because these individuals will identify
nity to broaden understanding of functional diver- with several functional areas. An individual who
sity in teams. And fourth, a careful comparison of has spent time in a variety of functional areas
all four forms of functional diversity is beyond the should be less strongly identified with a singl
scope of a single study. function and less likely to view individuals i
In developing our hypotheses, we reference data other functions in stereotyped and biased ways. In
obtained from 43 interviews conducted with man- other words, there are simply fewer function-bas
agement team members from four different busi- out-groups in a team composed of functional gen
ness units in a Fortune 100 consumer products eralists since team members can claim membership
company. This same company provided the setting in a variety of functional domains. These argu
for our empirical analysis, although these four ments suggest that teams composed of functionall
teams were not included in that sample. Interviews broad individuals will be better at sharing informa
lasted about an hour and were tape-recorded and tion than will teams composed of functional
transcribed. Interviews were semistructured and specialists.
The following quote from our interview data gets ence a given team member's responses to both the
at functional stereotypes and the way in which expectancy and the instrumentality questions. In a
functional breadth can break down these stereo- team composed of functionally broad individuals, a
types: given team member will be more likely to believe
that other team members can understand activities
It's just people don't understand what the other and developments in his or her functional area
person's experience base is or what's important to because he or she knows that other team members
their job. You know, the typical manufacturing guy
have had experience in that functional area (ex-
always wants [accurate forecasts] or the typical sales
guy all he knows how to do is go after volume, pectancy is increased). Furthermore, an individual
volume. It's less of a problem if you have cross- with experience in a range of functions will be
fertilization between departments and a better un- more likely to recognize the relevance and impor-
derstanding of what those people are up against. tance of information that he or she possesses for
individuals in other functional areas and for the
Although social categorization provides one ex- team as a whole (instrumentality is increased).
planation for how functional diversity can affect Consequently, since each member of a team of
information sharing, the sharing of information in functionally broad individuals should be more
teams can also be seen as a basic question of moti- strongly motivated to share information with oth-
vation-that is, are team members motivated to ers, levels of information sharing within such a
share information about activities and develop- team should be high.
ments in their area with others in their team? The above arguments suggest the following hy-
Whereas social categorization theory provides anpothesis about the relationship between intraper-
explanation for why an individual might not wantsonal functional diversity and information sharing
to share information with other team members (that
in management teams:
is, negative stereotypes), a motivation approach
Hypothesis 1. The intrapersonal functional di-
(specifically, expectancy theory [Vroom, 1964])
versity of a team will be positively associated
provides an explanation for why an individual
would want to share information with other team with information sharing within the team.
members. By viewing the problem of information Given our earlier argument, that information
sharing in teams in terms of expectancy theory, we sharing will be positively related to team effective-
can identify additional ways in which intraper- ness, Hypothesis 1 suggests (by extension) that in-
sonal functional diversity might enhance team trapersonal functional diversity can influence team
member motivation to share information and there- effectiveness by increasing information sharing.
fore increase information sharing within a team. We suspect, however, that intrapersonal functional
According to expectancy theory, motivation to diversity has additional positive implications for
perform a task results from an individual's re- group process and performance. For example, teams
sponses to three questions: (1) If I exert effort on composed of functionally broad individuals may be
this task, will I achieve the relevant outcome? (ex- better at making well-informed decisions (individual
pectancy), (2) If I achieve this outcome, will it lead and collective) than teams composed of functionally
to a particular reward? (instrumentality), and (3) Do narrow individuals (Burke & Steensma, 1998). Re-
I value the reward? (valence). In the case of infor- search suggests that accumulated prior knowledge is
mation sharing, these questions can be framed as necessary for new knowledge to be assimilated and
follows: (1) If I share information with my fellow used because of the self-reinforcing aspect of memory
team members about activities and developments development (Bower & Hilgard, 1981). That is, the
in my area, will they understand me? (expectancy), more objects, patterns, and concepts that are "stored
(2) If my team members understand activities and in memory," the more readily individuals acquire
developments in my area, will it help us to perform new information about these constructs and the better
better as a team/organization? (instrumentality), they are at using the information in new settings
and (3) Do I value performing better as a team/ (Bower & Hilgard, 1981: 424). Without appropriate or
organization? (valence). If we assume that most sufficient prior knowledge, individuals may have dif-
team members will respond in the affirmative to ficulty making new knowledge fully intelligible
the valence question (given that bonuses, job secu- (Lindsay & Norman, 1977: 517). These research find-
rity, and other incentives are involved), motivation ings imply that teams composed of individuals who
to share information in teams really boils down to have worked in a number of different areas may be
issues of expectancy and instrumentality. better prepared both to make sense of information
We propose that the intrapersonal functional di- and to integrate information related to different func-
versity of a management team will positively influ- tional domains. In addition, teams composed of indi-
viduals with broad capacity (intrapersonal functional tion diversity to exacerbate the problems of stereo-
diversity) may be more capable of recombining exist- typing and in-group/out-group biases described
ing knowledge, skills, and abilities into novel combi- above, as each team member is likely to identify
nations (Burke & Steensma, 1998; Weick, Sutcliffe, & with his or her particular (dominant) functional
Obstfeld, 1999). specialty. When an individual specializes in a par-
The following quotes, taken from our interview ticular functional area, identification with that
data, summarize some of these benefits of intraper- function becomes central to his or her sense of self,
sonal functional diversity for decision making: and favorable comparisons between his or her func-
tion and other functions become increasingly im-
The more each individual manager knows generally
portant. Unfavorable characterizations of out-group
about the business, the more effective they're going
functions, which lead to stereotyping and restricted
to be in making the right decision for their individ-
ual functional responsibility because they're not communication, are the likely result. Although
making a decision in a vacuum. They're actually these problems may not arise for groups composed
trying to think first and foremost about their respon- of individuals who are specialists in the same func-
sibility but also thinking about that in the right tion, they become a factor as the range of functional
context. specialists on a team increases. In other words, as
dominant function diversity increases, problems
The better you know the business, the better you're
with communication and lack of cohesion stem-
going to be able to run your small segment of it
because you've gotta be able to connect the dots. ming from social categorization processes becom
You've gotta be able to put the pieces of the puzzle more likely.
together. You've gotta know how it all mixes in Furthermore, motivation to share information
together. It's nice that you're great at what you do may be compromised in teams whose members
and I think it's nice that you run your part of the have nonoverlapping functional backgrounds. If
business very efficiently but you may not know how team member A perceives that the other members
that connects to the bigger scheme of things. of his or her team are specialists in different func-
tional areas, A will be less likely to believe that
These arguments suggest that although intraper-
those other team members can understand activi-
sonal functional diversity may improve team effec-
ties and developments in A's area (expectancy i
tiveness by improving information sharing, the
low). Furthermore, the fact that A is also a func-
benefits of intrapersonal functional diversity may
tional specialist decreases the likelihood that A
not be limited to that one intervening variable. Inwill see how issues in his or her functional area
other words, we did not expect that information
impact other functions and the team as a whol
sharing would fully mediate the relationship between
(instrumentality is low). So, although A may still go
intrapersonal functional diversity and performance
to other team members for advice about issues and
outcomes. To state this argument formally:
problems that he or she sees as related to their
Hypothesis 2. Information sharing within a team functions, A may be less willing to share informa-
will partially mediate the positive relationship tion related to his or her own function. This implies
between the intrapersonal functional diversity of that information sharing will be lower in function-
the team and (near-term) performance. ally heterogeneous teams.
The following quote from our interview data il-
Dominant function diversity and information lustrates how a narrow functional background can
sharing. As discussed above, the construct of dom-inhibit information sharing:
inant function diversity emphasizes the different
and unique functional specializations of team Because I understand sales, I have a tendency to
members. Dominant function diversity is highergravitate a lot more to the [sales managers] because
when team members have specialized functionalI feel like I can add some value there. I don't have a
experiences that complement one another in a way true appreciation for what X does in the manufac-
that covers all the functional bases. In contrast, turing side of the business because I don't really
understand all the nuances that go into that. And
intrapersonal functional diversity is higher when
because I don't have a good capacity at that, I don't
team members are generalists with experience in a
interact with X as much as I would if I really under-
range of functional areas. We therefore expected stood what he was up against. I think the same thing
that dominant function diversity would have very would apply to Y. I don't really know what he does
different implications for management team pro- on a day-to-day basis or how he goes about devel-
cess and performance than would intrapersonal oping his budget, what process he has in place to
functional diversity. make sure that he's got a declining balance. I don't
For example, we would expect dominant func- know the kind of measures he's looking at. I think
the fear of the unknown causes people to gravitate to this company are organized geographically, and
things that they do understand. each unit's management team is responsible for
These arguments lead to the following hypothe- overseeing the production, sales, marketing, and
sis about the relationship between dominant func- distribution of the company's product line in a
tion diversity and information sharing in manage- defined market area. Although each team is given
ment teams: considerable autonomy in how they carry out these
activities, ambitious performance targets are devel-
Hypothesis 3. The dominant function diversity
oped for each business unit under the supervision
of a team will be negatively associated withof the corporate office, and each team is evaluated
information sharing within the team. and compensated in strict accordance with these
performance targets. In the words of one manager,
Combined with our earlier argument that infor-
"We have a lot of autonomy. We have to coordinate
mation sharing will be positively related to team
with other business units, of course, but we have a
effectiveness, this hypothesis suggests that domi-
lot of leeway." Gross revenues across all business
nant function diversity will contribute negatively
units averaged $125 million in 1994.
to team effectiveness by decreasing information
Following previous research (see Hambrick,
sharing within the team. But research adopting this
conceptualization of functional diversity has sug- we defined a business unit management
1994),
gested other problems associated with dominant team as a unit's general manager and his or her
function aiversity besides a restricted exchange"direct
of reports." The teams we studied included
management-level individuals in each of the fol-
information. For example, research has suggested
lowing functional areas: finance, marketing, sales
that dominant function diversity can increase con-
anditdistribution, production (or product logistics),
flict (Knight et al., 1999; Pelled et al., 1999), make
more difficult for a team to reach consensus (Knightequipment management, administrative support,
and human resources. Since teams typically have
et al., 1999), and slow competitive response (Ham-
brick et al., 1996). These process losses help to than one sales and distribution manager to
more
cover different regions within their market area, the
explain why dominant function diversity has been
teams in our sample vary in size.
negatively associated with near-term financial per-
formance in past research (Murray, 1989). This sample was particularly appropriate for test-
ing our hypotheses. One of the problems facing
Given these findings, it seems clear that domi-
researchers
nant function diversity can compromise near-term studying management teams is that
performance through means other than simplysampled
de- teams are often not directly comparable on
key
creased information sharing. In other words, we did dimensions. By examining management teams
from
not expect information sharing to fully mediate the a single organization, we held constant the
role
negative relationship between dominant function di- composition of the teams, how performance
versity and unit performance. Stated formally, was measured and reported, and the work per-
formed. We could therefore focus on the hypothe-
Hypothesis 4. Information sharing withinsizeda relationships while minimizing variance on
team will partially mediate the negative rela-
many of the more common industry (competitive-
tionship between the dominantfunction diver-ness, turbulence) and organizational (structure,
sity of the team and (near-term) performance.
culture, reporting idiosyncrasies) confounds.
To summarize, we suggest that intrapersonal Furthermore, since the teams in this sample were
responsible
functional diversity will be positively associated for the execution of day-to-day opera-
tional
with information sharing in teams and that domi- details, ongoing information sharing across
functions
nant function diversity will be negatively associ- was an especially important team process
ated with information sharing. Furthermore, wethat helped to promote mutual adjustment and real-
suggest that information sharing will partially time
me- coordination. The managers we interviewed
strongly
diate the relationship between these two functional emphasized this point:
diversity variables and unit performance.
The [sales people] can't meet their goals and take
care of their customers if my group doesn't support
METHODS them correctly. So we have to have very good dia-
logue and communication. None of these things
Research Site and Sample stand alone, not at all .... So to be successful we
have to interact very well, extremely well.
We collected data from the management team
members of business units in a Fortune 100 con- Communicate. That's the number one thing-com-
sumer products company. The business units in municate, communicate and re-communicate. Ev-
885
2002 Bunderson and Sutcliffe
erybody needs to know what track the other guy is mining the dominant function of each te
on. People get off on a tangent, and it has ripple member (the functional area in which he or she
effects throughout the organization. spent the greater part of his or her career) and t
Clearly a case can be built that the best operating computing the heterogeneity index proposed
teams are the ones who work closely together. Blau (1977):3
9
Data Collection 1- pi 2
i=1
Survey and archival data from a random sample

of business units were used for testing the hypoth-where pi equals the percentage of a group whose
eses. Because corporate management limited thedominant functional background is in the ith func-
number of units that we could ask to participate intional area (with nine different functions, in this
the study, we selected a random sample of 45 unitscase). Following Teachman (1980), we normed
(47 percent of the population) using a two-step,dominant function diversity so that its values would
stratified sampling procedure designed to maxi- range from 0 (low diversity) to 1 (high diversity).
mize variance on the dependent variable. We first To measure intrapersonal functional diversity,
grouped the population of business units into high,we first computed an intrapersonal functional di-
medium, and low performers according to their versity score for each team member (see Walsh,
performance averaged over a two-year period pre- 1988) and then computed the average of these
scores within each team (see Burke & Steensma,
ceding the data collection. Next, we randomly se-
lected 15 units from each performance level, ob-1998) as follows:
taining a final sample of 45 teams. n 9
Surveys were distributed to each unit manager E 1 - Pij2 In,

and his or her direct reports along with a letter from i=1\ j=1 I
the corporate vice president of human resources
introducing the study, asking for participation, andwhere pij equals the percentage of manag
promising confidentiality. We collected survey years of experience spent in the jth func
data early in the year for which performance was and n equals the number of team mem
assessed so that there would be a lag of nearly one approach assesses the central tendency
full year between survey administration and mea- tribution of intrapersonal functional
surement of the criterion performance variable.scores within a team. An alternative m
Surveys were returned from 438 of the 521 individ-calculate the percentage of team member
uals surveyed (84%) with valid data for 44 of the 45trapersonal functional diversity scores ab
teams (98%). Responding teams had an average ofcritical level (for instance, 0.50). We exam
11 members (s.d. = 2.4). The average age of teamapproach and found it to yield virtually
members was 40 years, and 86 percent were male.results in all regression models. Like d
At the time of the survey, team members had beenfunction diversity, intrapersonal functio
in their current positions an average of 1.9 yearssity was normed (Teachman, 1980) so t
and had been with the company an average of 9.6would range from 0 (low diversity)
years. diversity).
Performance data and data on team size were Dominant function diversity ranged from 0.27 to
0.92 (x = 0.66, s.d. = 0.17), and intrapersonal func-
collected from corporate records with the coopera-
tion and assistance of corporate personnel. tional diversity ranged from 0.12 to 0.51 (x = 0.28,
s.d. = 0.09), suggesting good variation on these two
variables across teams in this sample.
Measures Information sharing. We measured information
sharing by asking each team member to evaluate
Functional diversity. Team members were asked
the extent to which (1) information used to make
to indicate their years of previous work experience
key decisions was freely shared among the mem-
in each of nine functional areas: sales or marketing,
manufacturing, finance or accounting, personnel/
HR, distribution or warehouse, R&D, equipment
3 Although a majority of management researchers cite
management, administrative support, and general
Blau (1977) as the origin for this measure, it is sometimes
also attributed to Hirschman (1945, 1964), Herfindahl
management. These responses were used to calcu-
late our two measures of functional diversity. (1950),
Dom-or Gibbs and Martin (1962), who developed es-
inant function diversity was computed by deter-sentially the same measure.
bers of the team, (2) team members worked hard to objectives are realized in ex post results (Ezzamel,
keep one another up to date on their activities, and 1992: 25; McGrath, MacMillan, & Venkataraman,
(3) team members were kept "in the loop" about 1995). Second, financial target achievement explic-
key issues affecting the business unit. The response itly takes differences in business unit context into
set for all items ranged from 1, "very strongly dis- account, since the denominator (the profitability
agree," to 7, "very strongly agree." A Cronbach's target) is based on an evaluation of historic and
alpha of .89 supported the internal consistency re- projected market conditions. And third, the
liability of this three-item scale. achievement of explicit financial objectives re-
We averaged responses to these three items quires ongoing information sharing in order to co-
across all of the managers on a team to create a ordinate progress toward the goal, suggesting that
group-level index of information sharing. The va- this was the most appropriate performance mea-
lidity of a group-level measure was examined using sure for our research model.
the intraclass correlation coefficient, ICC (1,k), Financial target achievement for the teams in this
which measures the extent to which responses sample ranged from 70 to 113, with a mean of 98
within a given team are consistent with one another and a standard deviation of 8, a distribution that
but different from the responses of other teams compares favorably with the population distribu-
(Shrout & Fleiss, 1979; Kenny & LaVoie, 1985). An tion on this variable (p, = 98, a = 9.6, min. = 63,
ICC of 0.46, along with one-way analysis of vari- max. = 118).4
ance results suggesting that responses differed sig- Control variables. A review of the literature and
nificantly across teams (p < .001), supported the consideration of our research context suggested the
aggregation of individual scores to create a group- need to control for several variables in our model in
level measure of information sharing. Measures of order to establish the robustness of the hypothe-
information sharing ranged from 3.8 to 5.7 (x = 4.8, sized relationships. These control variables were
s.d. = 0.45).
team size, average years of work experience, age
Performance. As indicated above, these busi-
diversity, tenure diversity, and market growth.
ness unit management teams were held account- Research on group behavior and performance has
able for setting and achieving annual profitability
established quite clearly that group size has impor-
targets, defined in terms of net operating profit
tant implications for group processes and outcomes
before tax (NOPBT). Profitability targets were de-
(Goodman, Ravlin, & Argote, 1986). In the present
veloped under the supervision of corporate man-
study, it was important to control for team size for
agement and were intended to be stretch goals-
at least two reasons. First, in larger teams, sharing
goals not easily achievable without some information with all team members may be more
improvement in revenues and/or reduction in
difficult. And second, since larger teams are asso-
costs. Consequently, proposed targets were ap-
ciated with larger business units, it is important to
proved only after a careful evaluation of historic
control for any possible relationship between size
market conditions and potential market improve-
and performance. Team size was computed simply
ments. Corporate personnel also sought to apply a
consistent set of criteria across all business units so as the number of managers on a team (general man-
ager and direct reports) based on data obtained
that target achievement would represent compara-
ble improvement. Once set, these targets con- from corporate personnel.
Whereas the two functional diversity variables
stituted the key performance bar against which
included in the model got at the mix of functional
management teams were evaluated to determine
bonuses, promotions, and other recognition. expertise on a management team, they did not tell
Given the primacy of these profitability targets us how many years of work experience team mem-
for the teams in this sample, we measured perfor- bers had overall. Since competence and expertise
mance in terms of this variable, specifically, as a are at least partially a function of time, it is impor-
management team's actual profitability relative to tant to also control for the average years of work
its targeted profitability for the survey year (multi-
plied by 100 to simplify the interpretation of re-
sults). This performance measure has a number of
4 Financial target achievement in the survey year was
advantages-beyond its importance in the minds of not significantly correlated with our stratification vari-
participating managers-that made it particularly able-financial target achievement averaged over the
well suited for this sample and research model. prior two years-in the business unit population. This
First, as a number of scholars have noted, financial finding mitigated concern about possible bias owing to
target achievement captures an important aspect of correlations between the dependent variable and the
economic efficiency: the extent to which ex ante sampling variable (Quesenberry & Jewell, 1986).
experience of team members. This variable was positive and significant coefficient for intraper-
computed from survey responses. sonal functional diversity would support Hypoth-
Although functional diversity was the central fo- esis 1, and a negative and significant coefficient
cus of the present study, other forms of diversity for dominant function diversity would support
have also been shown to relate significantly to man- Hypothesis 3.
agement team process and performance. Several In the second equation, model 2, the dependent
studies have suggested, for example, that diversity variable (unit performance) was regressed on the
in team member ages or tenures can negatively independent variables and the controls. And in
affect intragroup communication (O'Reilly et al., model 3, the dependent variable was regressed on
1993; Smith et al., 1994; Zenger & Lawrence, 1989). the independent variables, the mediator, and the
Furthermore, a number of studies have examined controls. We would have evidence for mediation if
the relationship between age or tenure diversity (1) the two functional diversity variables signifi-
and performance outcomes, with some scattered cantly predicted information sharing in model 1,
evidence for significant relationships (Williams & (2) the two functional diversity variables signifi-
O'Reilly, 1998: 96, 103-104). We therefore con- cantly predicted unit performance in model 2, and
trolled for age and tenure diversity in the present (3) information sharing significantly predicted unit
model in order to account for any possible effects. performance in model 3. Given these three condi-
We computed age diversity from self-reported age tions and effects in the predicted directions, the
and tenure diversity from self-reported organiza- final requirement for mediation is that the effect of
tional tenure. Both variables were computed with the two functional diversity variables on unit per-
the coefficient of variation (dividing the standard formance be less in model 3 than in model 2. If we
deviation among team members by the mean for the found that the two functional diversity variables
team, following Allison [1978]). had no effect on unit performance in model 3, we
Finally, since business unit performance is sig- would have evidence for "perfect mediation" (Bar-
nificantly influenced by the market context within on & Kenny, 1986: 1177). Since Hypotheses 2 and 4
which a unit operates, we controlled for market predict partially mediated relationships rather than
context in two different ways. First, our perfor- perfectly mediated relationships, support for Hy-
mance variable accounts for market context by potheses 2 and 4 would be obtained if the coeffi-
assessing performance relative to the level of per- cients for the two functional diversity variables
formance that would be expected given past trends were smaller in model 3 than in model 2 but not
and anticipated market conditions. Second, we in- necessarily insignificant.
cluded a measure of market growth for the geo-
graphic region served by each unit. We obtained
RESULTS
region-specific data on the value of product ship-
ments in the company's core business for the pe- Table 2 presents the means, standard deviation
riod of this study from the U.S. Economic Census, and correlations for all variables included in the
Manufacturing Industry Series (U.S. Department of model. Table 3 presents the mediated regression
Commerce, 1999: Table 6b). Market growth was results for models 1 through 3. In model 1, the two
computed as the value of core business product functional diversity variables and five control vari-
shipments within a business unit's region (U.S. ables accounted for a significant 19 percent (p <
state) at t + 2 divided by the value of product .05) of the variance in information sharing. Of the
shipments in that region at t - 3 (where t is the five control variables included in the model, only
survey year and t - 3 and t + 2 are census years). team size demonstrated a moderately significant
Market growth ranged from 0.87 to 2.56 for the (p < .10) association with information sharing. The
business units in this sample (x = 1.4, s.d. = 0.48). coefficients for dominant function diversity and
intrapersonal functional diversity were significant
Analyses (p < .01) and in the predicted directions, providing
support for Hypotheses 1 and 3.
We tested hypotheses using the mediated regres- In model 2, the five control variables and two
sion approach recommended by Baron and Kenny functional diversity variables accounted for a sig-
(1986). This approach involves examination of nificant 18 percent (p < .05) of the variance in unit
three separate regression equations. In the first performance. Only one of the five control variables
equation, model 1, the mediator (information shar- exhibited a significant relationship with unit per-
ing) was regressed on the independent variables formance-organizational tenure diversity was
(dominant function diversity and intrapersonal negatively related to unit performance at the .10
functional diversity) and the control variables. A level. The coefficients for the two functional diver-
TABLE 2
Means, Standard Deviations, and Correlationsa
Variable Mean s.d. 1 2 3 4 5 6 7 8
1. Team size 10.82 2.42

2. Average years experience 15.84 4.44 -.03
3. Age diversity 0.18 0.05 .23 .27t
4. Organizational tenure diversity 0.84 0.42 .05 -.37* -.26t
5. Market growth 1.38 0.48 .17 .10 .10 .15
6. Dominant function diversity 0.68 0.18 -.10 .10 .20 -.30* -.03
7. Intrapersonal functional diversity 0.28 0.09 -.05 .25 -.21 -.04 .24 .30*
8. Information sharing 4.84 0.45 -.22 .28t .01 -.11 .01 -.19 .30*
9. Unit performance 98.41 8.37 .08 -.07 -.21 -.05 .19 -.28t .28t .40**
an = 44.
tp < .10
* p < .05
** p< .01
TABLE 3
Results of Mediated Regression Analysis of Functional Diversity Variables, Information Sh

Unit Performancea
Model 1: Model 2: Model 3:

Independent Variables Information Sharing Unit Performance Unit Performance
Controls
Team size -.274 .05 .14
Average years experience .12 -.22 -.26*
Age diversity .21 -.08 -.15
Organizational tenure diversity -.10 -.30t -.26t
Market growth -.08 .15 .17
Functional diversity variables
Dominant function diversity -.44** -.44 * -.29*
Intrapersonal functional diversity .45** .40* .25
Mediator
Information sharing .34*
F 2.47* 2.36* 2.80*
R2 .32 .31 .39
Adjusted R2 .19 .18 .25
AR2 .08*
df 7, 36 7, 36 8, 35
a n = 44. Standardized re
tp < .10
* p < .05
**p < .01
sity variables were both significant (p < .01 forcreased from -.44 (which was significant at the .01
dominant function diversity; p < .05 for intraper- level) to -.29 (marginally significant at the .10
sonal functional diversity) and in the predictedlevel). In the case of intrapersonal functional diver-
directions. sity, the coefficient decreased from .40 (p < .05) to
Finally, the addition of information sharing in .25 (p = .16, n.s.). According to Baron and Kenny
model 3 increased the variance explained to a sig- (1986), these results suggest that information shar-
nificant 25 percent (p < .05). The coefficient for ing partially mediates the relationship between
information sharing in model 3 was positive and dominant function diversity and unit performance
significant as predicted. Furthermore, the coeffi- (as specified in Hypothesis 4) but that information
cients for both functional diversity variables de- sharing may fully mediate the relationship between
creased in magnitude and significance. In the case intrapersonal functional diversity and unit perfor-
of dominant function diversity, the coefficient de- mance (in contrast to Hypothesis 2, which predicts
a partially mediated relationship). The fact that the sonal functional diversity and unit performance is
coefficient for intrapersonal functional diversity largely (if not completely) explained by improved
was still fairly large (.25), however, and ap- information sharing. This finding is inconsistent
proached significance (p = .16) suggests that we with our argument that intrapersonal functional
may want to preserve the possibility that intraper- diversity will have positive implications for perfor-
sonal functional diversity exerts an effect on unit mance that go beyond just improved information
performance beyond what can be accounted for by sharing (such as better-informed decisions). Given
improved information sharing. the magnitude of the coefficient for intrapersonal
In order to directly test the approximate signifi- functional diversity in model 3, however, it ap-
cance of the mediated effects hypothesized in Hy- pears that intrapersonal functional diversity does
potheses 2 and 4, we used procedures outlined in account for some residual variance in unit perfor-
Baron and Kenny (1986: 1177) that build on the mance. Further exploration of this relationship
work of Sobel (1982) and Goodman (1960). Specif- therefore provides one promising direction for fu-
ically, we computed the standard error for each ture research.
mediated effect (following Goodman, 1960) and As expected, we found a very different (and gen-
then used this parameter to compute Z-scores for erally opposite) pattern of results for dominant
each mediated effect. Results support our hypoth- function diversity. Specifically, we found that
eses that information sharing mediates the relation- dominant function diversity was negatively associ-
ship between both dominant function diversity and ated with information sharing for the management
intrapersonal functional diversity (p < .001 in both teams in our sample. This finding is consistent
cases). with our argument that the dispersion of team
members across functional areas of expertise in-
creases the likelihood that team members will have
DISCUSSION
very different backgrounds and experiences and
Our objective in this study was to suggest willthat
therefore have difficulty communicating with
and relating to one another. We also found that
different conceptualizations of functional diversity
can have very different implications for teaminformation
pro- sharing partially mediated the rela-
cess and performance. We examined two forms tionship
ofbetween dominant function diversity and
functional diversity-intrapersonal functional di-
unit performance. This result supports our conten-
tion
versity and dominant function diversity-and de- that the negative implications of dominant
function
veloped a model to predict how these two forms of diversity for unit performance are not
fully explained by decreased information sharing
functional diversity would relate to information
sharing and unit performance. Drawing on but may also be a consequence of increased conflict
social
categorization and motivation theories as and welldissensus,
as slower decision making, and an in-
ability to take decisive and coordinated action.
research on groups and teams, our model predicted
that intrapersonal functional diversity would have The findings presented here enrich understand-
positive, and dominant function diversity, negativeing of functional diversity in management teams
effects on both information sharing and unit and have important implications for management
per-
formance. A test of this model in 44 business unit theory and practice. For example, our findings un-
management teams from a Fortune 100 consumer derscore the wisdom of Pfeffer's (1983) recommen-
products company provided strong support for the dation that researchers be sensitive to the differ-
different effects of these two conceptualizations of ences between demographic measures and make
functional diversity on management team process sure that their measures match their theories. In
and performance. this study, different measures of functional diver-
Specifically, we found that intrapersonal func- sity exhibited different relationships to the same
tional diversity was positively associated with in- mediating and outcome variables for the same
formation sharing. This finding is consistent with group of teams. These results suggest that studies
our expectation that teams composed of function- using different measures of functional diversity (or
ally broad individuals would be more strongly mo- any other demographic variable) may not be com-
tivated to exchange information and would be less parable and that researchers should therefore be
susceptible to the stereotypes and in-group/out- careful and explicit in defining and measuring de-
group biases that restrict the open sharing of infor- mographic constructs.
mation. We also found that information sharing Furthermore, although empirical research has
mediates the relationship between intrapersonal demonstrated the benefits of functional breadth for
functional diversity and unit performance and that, individual careers (Campion et al., 1994; McCall,
in fact, the positive relationship between intraper- Lombardo, & Morrison, 1988), this study is the first
to examine the significance of intrapersonal func- rather than differences across contexts. Future re-
tional diversity for teams. The results suggest that search might therefore consider how differences in
this form of functional diversity has significant and contextual factors (for instance, environmental sta-
positive implications for team process and perfor- bility) affect the relationships examined here.
mance. These findings imply that organizations can And finally, the fact that the two forms of func-
benefit considerably by seeking and developing tional diversity examined here-dominant func-
management teams composed of individuals who tion diversity and intrapersonal functional diver-
are functionally broad and not just narrowly spe- sity-had such very different implications for team
cialized in a single functional area. process and performance suggests that similar com-
parisons involving the two other forms of func-
Future Research Directions
tional diversity-functional background diversity
and functional assignment diversity-may also be
A number of avenues for future research follow significant and substantively meaningful. In pursu-
from this study. First, although intrapersonal func- ing this promising research direction, it may be
tional diversity emerged as an important variable in useful to consider not only the comparative effects
this study, the seemingly low intrapersonal func- of the different forms of functional diversity, but
tional diversity scores for these business unit man- also the way in which they interact. Jehn, North-
agement teams (the average was 0.28) suggests that, craft, and Neale (1999) found that the relationship
for the most part, managers in this sample had between certain forms of team diversity and perfor-
fairly narrow ranges of functional experience. This mance can be moderated by other forms of team
finding raises a number of interesting questions diversity. For example, they found that the rela-
about the extent to which intrapersonal functional tionship between informational diversity (diversity
diversity might vary across organizations, indus- in education, functional assignment, and position)
tries, or hierarchical levels and whether these dif- and team performance is moderated by value diver-
ferences might affect the relationships observed sity (diversity in goals and work values). These
here. An examination of intrapersonal functional findings underscore the possibility that the various
diversity in other settings is therefore an important forms of functional diversity we identify here
direction for future research. may interact in their effect on team processes and
Second, this study considered the performance performance.
outcomes of functional diversity using a near-term For example, the theory and results presented in
measure of performance-performance in the year this article raise an interesting paradox-what
following data collection. Results indicated that about a team composed entirely of specialists from
intrapersonal functional diversity was positively the same function? On one hand, such a team
associated with near-term performance and, like should be able to easily share information (given a
the results of other studies (Murray, 1989; Simons common functional background) but, on the other
et al., 1999), that dominant function diversity was hand, increased information sharing may not trans-
negatively associated with near-term performance. late into better-informed decisions because the in-
There is, however, evidence to suggest that al- formation shared represents a single functional per-
though functional diversity can cause performance spective. This particular scenario is practically
problems in the near term, it may be beneficial in unlikely in most management teams, given that the
the long term (Keck & Tushman, 1993; Murray, different functional positions in a management
1989). An important direction for future research, team are filled by individuals with presumed ex-
therefore, would be to examine the relative effects pertise in different functions. Nevertheless, it sug-
of intrapersonal functional diversity and dominant gests that there may be settings in which the inter-
function diversity on performance over the long action between dominant function diversity and
term. intrapersonal functional diversity becomes partic-
Third, past research has suggested that the posi- ularly important for predicting information sharing
tive or negative effects of functional diversity may and performance outcomes.
be at least partially a function of contextual factors. Furthermore, there may be situations in which
For example, researchers have suggested that func- the interaction between intrapersonal functional
tional diversity may be most strongly related to diversity and functional assignment diversity has
performance under conditions of high rivalry (Mur- implications for team process and performance. In
ray, 1989) or high environmental turbulence (Keck, many teams (including most management teams),
1997). In the present study, industry context was functional accountabilities are clearly assigned so
held constant so that we could focus on differences that people know who is responsible for what and
across conceptualizations of functional diversity who has the last word on what issues. Where func-
2002
Bunderson and Sutcliffe 891
tional accountabilities are not clearly assigned, search: Conceptual, strategic, and statistical consid-
however, informal roles often evolve around per- erations. Journal of Personality and Social
ceived areas of functional expertise. In a team com- Psychology, 51: 1173-1182.
posed largely of functional generalists (that is, a Beyer, J. M., Chattopadhyay, P., George, E., Glick, W. H.,
team with high intrapersonal functional diversity), ogilvie, dt, & Pugliese, D. 1997. The selective percep-
this informal role evolution could become prob- tion of managers revisited. Academy of Manage-
lematic, since several people might legitimately ment Journal, 40: 716-737.
claim expertise in a given area. This suggests that Blau, P. M. 1977. Inequality and heterogeneity: A prim-
where intrapersonal functional diversity is high, itive theory of social structure. New York: Free
functional assignment diversity might also need to Press.
be high in order to avoid coordination and account-
Bower, G. H., & Hilgard, E. R. 1981. Theories of learning.
ability problems.
Englewood Cliffs, NJ: Prentice-Hall.
Burke, L. A., & Steensma, H. K. 1998. Toward a model for
Conclusion relating executive career experiences and firm per-
formance. Journal of Managerial Issues, 10: 86-
Research on management team process and per- 102.
formance has increasingly recognized the impor-
Campion, M. A., Cheraskin, L., & Stevens, M. J. 1994.
tance of diversity in functional backgrounds, affil- Career-related antecedents and outcomes of job rota-
iations, and assignments for team effectiveness. tion. Academy of Management Journal, 37: 1518-
The purpose of this study was to suggest that the 1542.
construct of functional diversity can be conceptu-
Carpenter, M. A., & Frederickson, J. W. 2001. Top man-
alized in different ways and that these different
agement teams, global strategic posture, and the
conceptualizations have very different impli- moderating role of uncertainty. Academy of Man
cations for team process and performance. Our ex- agement Journal, 44: 533-546.
amination of two different forms of functional
diversity-dominant function diversity and intra- Chaganti, R., & Sambharya, R. 1987. Strategic orientatio
and characteristics of upper management. Strategic
personal functional diversity-confirmed this ex-
Management Journal, 8: 393-401.
pectation. These results underscore the importance
of careful articulation, definition, and measure- Eisenhardt, K. M. 1989. Making fast decisions in high
ment of demographic constructs and suggest that velocity environments. Academy of Management
Journal, 32: 543-576.
intrapersonal functional diversity may be more im-
portant for management team effectiveness than the Ezzamel, M. 1992. Business unit and divisional perfor-
simple distribution of team members across func- mance measurement. San Diego: Academic Press.
tional categories. Gibbs, J., & Martin, W. 1962. Urbanization, technology,
and the division of labor: International patterns.
American Sociological Review, 27: 667-677.
REFERENCES
Gladstein, D. 1984. Groups in context: A model of task
group effectiveness. Administrative Science Quar-
Allison, P. D. 1978. Measures of inequality. American
Sociological Review, 43: 865-880. terly, 29: 499-517.
Ancona, D. G., & Caldwell, D. F. 1992. DemographyGlick,
and W. H., Miller, C. C., & Huber, G. P. 1993. The
design: Predictors of new product team performance.impact of upper-echelon diversity on organizational
Organization Science, 3: 321-341. performance. In G. P. Huber & W. H. Glick (Eds.),
Argote, L., Gruenfeld, D., & Naquin, C. 1999. Group learn- Organizational change and redesign: Ideas and
ing in organizations. In M. E. Tuner (Ed.), Groups at insightsfor improvingperformance: 176-214. New
York: Oxford.
work: Advances in theory and research: 369-411.
New York: Erlbaum.
Goodman, L. A. 1960. On the exact variance of products.
Bantel, K. A. 1993. Strategic clarity in banking: Role of Journal of the American Statistical Association,
55: 708-713.
top management-team demography. Psychological
Reports, 73: 1187-1201. Goodman, P. S., Ravlin, E. C., & Argote, L. 1986. Cu
Bantel, K. A., & Jackson, S. E. 1989. Top management and thinking about groups: Setting the stage fo
innovations in banking: Does the demography of the ideas. In P. Goodman (Ed.), Designing effe
top team make a difference? Strategic Management work groups: 1-33. San Francisco: Jossey-Bas
Journal, 10: 107-124.
Guzzo, R. A., & Shea, G. P. 1992. Group performan
Baron, R. M., & Kenny, D. A. 1986. The moderator-medi- intergroup relations in organizations. In M. D. D
ator variable distinction in social psychological re- nette & L. M. Hough (Eds.), Handbook of indust
and organizational psychology (2nd ed.), 3: 270- executives develop on the job. Lexington, MA: Lex-
313. Palo Alto, CA: Consulting Psychologists Press. ington Books.
Hackman, J. R., & Morris, C. G. 1975. Group tasks, group McGrath, R. G., MacMillan, I. C., & Venkatraman, S.
interaction process, and group performance effec- 1995. Defining and developing competence: A stra-
tiveness: A review and proposed integration. In L. tegic process paradigm. Strategic Management
Berkowitz (Ed.), Advances in experimental social Journal, 13: 137-161.
psychology: 45-99. New York: Academic Press. Michel, J. G., & Hambrick, D. C. 1992. Diversification
Hambrick, D. C. 1994. Top management groups: A con- posture and top management team characteristics.
ceptual integration and reconsideration of the Academy of Management Journal, 35: 9-37.
"team" label. In B. M. Staw & L. L. Cummings (Eds.),
Milliken, F., & Martins, L. 1996. Searching for common
Research in organizational behavior, vol. 16: 171-
threads: Understanding the multiple effects of diver-
214. Greenwich, CT: JAI Press.
sity in organizational groups. Academy of Manage-
Hambrick, D. C., Cho, T., & Chen, M. 1996. The influence ment Review, 21: 402-433.
of top management team heterogeneity on firms'
Murray, A. I. 1989. Top management group heterogeneity
competitive moves. Administrative Science Quar-
and firm performance. Strategic Management Jour-
terly, 41: 659-84.
nal, 10: 125-141.
Herfindahl, O. C. 1950. Concentration in the U.S. steel
O'Reilly, C. A., Snyder, R. C., & Boothe, J. N. 1993. Effects
industry. Unpublished doctoral dissertation, Colum-
of executive team demography on organizational
bia University, New York.
change. In G. P. Huber & W. H. Glick (Eds.), Or-
Hirschman, A. 0. 1945. National power and the struc- ganizational change and redesign: Ideas and in-
ture of foreign trade. Berkeley, CA: University of sights for improving performance: 147-175. New
California Press. York: Oxford.
Hirschman, A. 0. 1964. The paternity of an index. Amer- Pelled, L. H., Eisenhardt, K. M., & Xin, K. R. 1999. Ex-
ican Economic Review, 54: 761-762.
ploring the black box: An analysis of work group
Hitt, M. A., & Tyler, B. B. 1991. Strategic decision mod- diversity, conflict, and performance. Administrative
els: Integrating different perspectives. Strategic Science Quarterly, 44: 1-28.
Management Journal, 12: 327-351. Pfeffer, J. 1983. Organizational demography. In L. L.
Jehn, K. A., Northcraft, G. B., & Neale, M. A. 1999. Why Cummings & B. M. Staw (Eds.), Research in organ-
differences make a difference: A field study of diver- izational behavior, vol. 5: 299-357. Greenwich, CT:
sity, conflict, and performance in work groups. Ad- JAI Press.
ministrative Science Quarterly, 44: 741-763. Prahalad, C. K., & Bettis, R. A. 1986. The dominant logic:
Keck, S. L. 1997. Top management team structure: Dif- A new linkage between diversity and performance.
ferential effects by environmental context. Organi- Strategic Management Journal, 7: 485-501.
zation Science, 8: 143-156.
Quesenberry, C. P., & Jewell, N. P. 1986. Regression anal-
Keck, S. L., & Tushman, M. L. 1993. Environmental and ysis based on stratified samples. Biometrika, 73:
organizational context and executive team structure. 605-614.
Academy of Management Journal, 36: 1314-1344.

Rulke, D. L. 1996. Member selection strategy
Kenny, D. A., & LaVoie, L. 1985. Separating individual performance: Cognitive integration vs. social
and group effects. Journal of Personality and Social tion in cross-functional teams. Academy o
Psychology, 48: 339-348. agement Best Paper Proceedings: 424-428
Knight, D., Pearce, C. L., Smith, K. G., Olian, J. D., Sims,Shannon, C. 1948. A mathematical theory of com
H. P., Smith, K. A., & Flood, P. 1999. Top manage- cations. Bell System Technical Journal, 2
ment team diversity, group process, and strategic 423, 623-656.
consensus. Strategic Management Journal, 20: Shrout, P. E., & Fleiss, J. L. 1979. Intraclass correlations:
445-465.
Uses in assessing rater reliability. Psychological
Lant, T. K., Milliken, F. J., & Batra, B. 1992. The role of 86: 377-386.
Bulletin,
managerial learning and interpretation in strategic
Simons, T., Pelled, L. H., & Smith, K. A. 1999. Makin
persistence and reorientation: An empirical explora-
use of difference: Diversity, debate, and decision
tion. Strategic Management Journal, 13: 585-608.
comprehensiveness in top management teams.
Lawrence, B. S. 1997. The black box of organizational
demography. Organization Science, 8: 1-22.
Smith, K. G., Smith, K. A., Olian, J. D., Sims, H. P.,
Lindsay, P. H., & Norman, D. A. 1977. Human informa-
O'Bannon, D. P., & Scully, J. A. 1994. Top manage-
tion processing. Orlando, FL: Academic Press.
ment team demography and process: The role of
McCall, M. W., Lombardo, M. M., & Morrison, A. M. social integration and communication. Administra-
1988. The lessons of experience: How successful tive Science Quarterly, 39: 412-438.
Sobel, M. E. 1982. Asymptotic confidence intervals for Williams, K. Y., & O'Reilly, C. A. 1998. Demography and
indirect effects in structural equation models. In S. diversity in organizations: A review of 40 years of
Leinhart (Ed.), Sociological methodology 1982: research. In B. M. Staw & L. L. Cummings (Eds.),
290-312. San Francisco: Jossey-Bass. Research in organizational behavior, vol. 20: 77-
Sutcliffe, K. M. 1994. What executives notice: Accurate 140. Greenwich, CT: JAI Press.
perceptions in top management teams. Academy of Williams, R. J., Hoffman, J. J., & Lamont, B. T. 1995. The
Management Journal, 37: 1360-1378. influence of top management team characteristics on
Tajfel, H. 1981. Human groups and social categories: M-form implementation time. Journal of Manage-
rial Issues, 7: 466-480.
Studies in social psychology. Cambridge, England:
Cambridge University Press. Zenger, T. R., & Lawrence, B. S. 1989. Organizational
Teachman, J. D. 1980. Analysis of population diversity. demography: The differential effects of age and ten-
ure distributions on technical communication.
Sociological Methods and Research, 8: 341-362.
Tsui, A. S., Egan, T. D., & Xin, K. R. 1995. Diversity in
organizations: Lessons from demography research.
In M. M. Chemers, S. Oskamp, & M. A. Costanzo
(Eds.), Diversity in organizations: New perspec-
tives for a changing workplace: 191-219. Thou-
sand Oaks, CA: Sage. J. Stuart Bunderson (bunderson@olin.wustl.edu) is an
assistant professor of organizational behavior at the Joh
Turner, J. C. 1982. Towards a cognitive redefinition of the
M. Olin School of Business at Washington University in
social group. In H. Tajfel (Ed.), Social identity and
St. Louis. He received his Ph.D. from the University
intergroup relations: 15-40. Cambridge, England:
Minnesota with a concentration in strategic managemen
Cambridge University Press.
and organization. His current research focuses on process
U.S. Department of Commerce. 1999. 1997 economic associated with learning, knowledge management, and th
census, manufacturing, industry series. Washing- integration of distributed experience and expertise.
ton, DC: U.S. Census Bureau.
Kathleen M. Sutcliffe is an associate professor of organiz
Vroom, V. H. 1964. Work and motivation. New York:
tional behavior and human resource management at the
Wiley.
University of Michigan Business School. She received
Walsh, J. P. 1988. Selectivity and selective perception: Ph.D. in management from the University of Texas at Au
An investigation of managers' belief structures and tin. Her research is focused on understanding how organ
information processing. Academy of Management zations and their members come to know and cope with
Journal, 31: 873-896. their environments through the processes of attention, i
Weick, K. E., Sutcliffe, K. M., & Obstfeld, D. 1999. Or- terpretation, and action and how an organization's desig
ganizing for high reliability: Processes of collective contributes to its members' abilities to sense, cope with
mindfulness. In B. M. Staw & L. L. Cummings (Eds.), and effectively respond to changing demands.
Research in organizational behavior, vol. 21: 81-
123. Greenwich, CT: JAI Press.

Academy of Management The Academy of Management Journal

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Academy of Management The Academy of Management Journal

Uploaded by

Copyright:

Available Formats

Comparing Alternative Conceptualizations of Functional Diversity in Management Teams:

Process and Performance Effects

COMPARING ALTERNATIVE CONCEP1 UALIZATIONS OF

Functional diversity in teams has been conceptualized in a variety of ways without

Empirical research on functional diversity in concluded that functional diversity is a double-

Alternative Conceptualizations of Functional Diversity in the Emp

Functional Diversity Variable:

(la) Dominant Function Diversity: Ten Studies

Blau index using archival data.

Blau index using archival data.

Knight, Pearce, Smith, Olian, Blau index using

(lb) Functional Background Diversity: Two Studies

(lc) Functional Assignment Diversity: Five Studies

a "TMT" is "top management team"; "MNC" is "multinational oorporation."

1 Although Carpenter and Frederickson (2001) re-

Survey and archival data from a random sample

Surveys were distributed to each unit manager E 1 - Pij2 In,

Variable Mean s.d. 1 2 3 4 5 6 7 8

1. Team size 10.82 2.42

Results of Mediated Regression Analysis of Functional Diversity Variables, Information Sh

Model 1: Model 2: Model 3:

Academy of Management Journal, 36: 1314-1344.

You might also like