MULTIPLE TEAM MEMBERSHIP: A THEORETICAL MODEL OF ITS EFFECTS ON PRODUCTIVITY AND LEARNING FOR INDIVIDUALS AND TEAMS

Page created by Tiffany Robbins
 
CONTINUE READING
姝 Academy of Management Review
2011, Vol. 36, No. 3, 461–478.

                           MULTIPLE TEAM MEMBERSHIP: A THEORETICAL
                           MODEL OF ITS EFFECTS ON PRODUCTIVITY AND
                             LEARNING FOR INDIVIDUALS AND TEAMS
                                                            MICHAEL BOYER O’LEARY
                                                             Georgetown University

                                                                  MARK MORTENSEN
                                                                      INSEAD

                                                          ANITA WILLIAMS WOOLLEY
                                                          Carnegie Mellon University

                          Organizations use multiple team membership to enhance individual and team pro-
                          ductivity and learning, but this structure creates competing pressures on attention
                          and information, which make it difficult to increase both productivity and learning.
                          Our model describes how the number and variety of multiple team memberships drive
                          different mechanisms, yielding distinct effects. We show how carefully balancing the
                          number and variety of team memberships can enhance both productivity and learn-
                          ing.

   Surveys estimate that 65 to 95 percent of                                          especially in teams (Bunderson & Sutcliffe, 2003;
knowledge workers across a wide range of in-                                          Singer & Edmondson, 2008).
dustries and occupations in the United States                                            Given this tension between productivity and
and Europe are members of more than one proj-                                         learning, understanding when and how both
ect team at a time (which we refer to as multiple                                     can be optimized in the context of multiple team
team membership). In some companies it is com-                                        memberships would be useful. In this article we
mon for people to be members of five, ten, or                                         propose a model of multiple team membership
twelve or more teams at a time (Martin & Bal,                                         and its effects on learning and productivity. The
2006; Zika-Viktorsson, Sundstrom, & Engwall,                                          model is guided by attention and social network
2006). As noted by Milgrom and Roberts (1992),                                        theories, which are particularly useful because
firms adopt this approach to organizing work to                                       people have increasingly unlimited access to
leverage their resources more effectively and to                                      information (through new technologies and rap-
promote knowledge transfer—that is, to en-                                            idly widening networks) but limited abilities to
hance both productivity and learning. However,                                        attend to and process that information (Borgatti,
classic work on the productivity dilemma (Aber-                                       Mehra, Brass, & Labianca, 2009; Gallagher, 2009;
nathy, 1976) suggests that the routines put in                                        Jackson, 2009; Ocasio, 1997). Thus, two critical
place to enhance productivity often hinder the                                        commodities in the current, highly networked
practices that foster learning (Adler et al., 2009;                                   economy—attention and information (Hansen &
Benner & Tushman, 2003), with learning and per-                                       Haas, 2001; Hudson, Christensen, Kellogg, &
formance often working at cross-purposes—                                             Erickson, 2002)—are precisely the resources that
                                                                                      are most central to managing multiple team
                                                                                      memberships. The model we propose explores
   We extend our thanks to Linda Argote, Alberto Espinosa,
Richard Hackman and members of the Harvard Groups-
                                                                                      how multiple team membership draws heavily
Group, Sophie Leroy, Michelle Marks, Randall Peterson, Mi-                            on these resources and addresses two ques-
chael Pratt, Ray Reagans, Denise Rousseau, and Ruth Wage-                             tions: (1) How are productivity and learning af-
man; Cynthia Pickering, Eleanor Wynn, and the Intel Corp.;                            fected by multiple team membership? (2) How
Beth Ahern, Lisa Bender, Kerry Buckley, Jim Farris, Fred                              can both outcomes be enhanced simultane-
Zapp, and the MITRE Corp.; and our anonymous reviewers
for their insightful and constructive comments on earlier
                                                                                      ously?
versions of the manuscript. We especially acknowledge Jeff                               We argue that two elements of multiple team
LePine for his support throughout the revision process.                               membership—the number of teams individual
                                                                                461
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 express written permission. Users may print, download, or email articles for individual use only.
462                                   Academy of Management Review                                   July

workers are members of and the variety of those       RESEARCH ON MULTIPLE TEAM MEMBERSHIP
teams—influence productivity and learning,
                                                         As noted above, being on multiple teams is
and we propose a model that specifies these
                                                      increasingly common and occurs in a wide
influences at the individual and team levels of       range of contexts. It appears to be especially
analysis. As our model shows, the more teams          common in highly competitive settings charac-
people are on, holding the variety of those teams     terized by pressure for both productivity and
constant, the more productivity but the less          learning, such as information technology, soft-
learning there will be. However, these produc-        ware development, new product development,
tivity gains do not continue unabated; they           and consulting (Milgrom & Roberts, 1992; Wheel-
eventually plateau and turn negative as the           wright & Clark, 1992). However, as noted by
number of team memberships continues to in-           other researchers, most of the existing research
crease. Conversely, the more variety there is in      on teams “has focused on intact teams without
the teams, holding the number of team member-         accounting for the possibility of multi-teaming”
ships constant, the more learning but the less        (Chudoba & Watson-Manheim, 2007: 67). Despite
productivity there will be. As with productivity,     some scholars’ acknowledgment that multiple
these learning gains eventually plateau (but do       team membership “is quite prominent these
not turn negative, since people and teams do not      days,” research on it has been “scant” enough
experience “negative learning”). Furthermore,         that a recent review described it as one of six
the number of teams individual workers are            areas in teams research that warrants attention
members of can have a focusing effect on indi-        and noted that “very little is known about its
vidual attention, leading individuals to seek out     implications for teams and individuals alike”
more efficient work practices and leading teams       (Mathieu, Maynard, Rapp, & Gilson, 2008: 442).
to focus on key priorities in their work. This        The need to understand the implications of mul-
focusing of attention can enhance productivity        tiple team membership is heightened in knowl-
while reducing learning. However, the variety of      edge-intensive environments, where attention is
teams individual workers are members of can           an especially scarce resource (March & Simon,
increase the diversity of information individuals     1958); where individuals, teams, and organiza-
and teams encounter, stimulating learning, but        tions strive to allocate their focus and attention
at the expense of productivity.                       in ways that maximize productivity and learn-
   With our model we make three key contribu-         ing; and where those efforts are “much ne-
tions. First, our model shows how multiple team       glected issue[s]” in research (Schmidt & Dolis,
                                                      2009: 690).
membership can enhance both productivity and
                                                         Although only a handful of studies have di-
learning, but only if the countervailing effects of
                                                      rectly addressed multiple team membership,
number and variety are carefully balanced. Sec-
                                                      other research has addressed related constructs
ond, we push beyond simple “multitasking is
                                                      and processes at the individual and team levels.
bad” arguments (Rosen, 2008) by separating
                                                      At the individual level, research on multitasking
multiple team membership into its key compo-
                                                      deals with the microprocesses and cognitive im-
nent parts of number and variety and explicat-        plications of an individual’s switching between
ing their distinct theoretical relationships with     two tasks or of “true” multitasking, where peo-
productivity and learning. Third, we move be-         ple work on two or more tasks simultaneously
yond the individual level of analysis to expli-       (e.g., driving a car while talking on a cell phone
cate the mechanisms that drive its effects for        or using a Blackberry while participating in a
teams. Some elements of individual productivity       team meeting; Leroy, 2009). These two versions
and learning aggregate to affect team produc-         of multitasking have also been described as
tivity and learning, but the mechanisms by            time swapping and time sharing (Waller, 1997).
which multiple team membership drives pro-            Multitasking has been studied in relation to spe-
ductivity and learning at each level are distinct.    cific behaviors like communication (Reinsch,
In the sections that follow we briefly review the     Turner, & Tinsley, 2008), as well as traits like
related research. Then we define key terms,           polychronicity (e.g., Bluedorn, 2002), which influ-
present our model, and discuss scholarly and          ences individuals’ preference for and success at
managerial implications.                              multitasking or interruption handling (e.g., Jett &
2011                                  O’Leary, Mortensen, and Woolley                                 463

George, 2003). Multitasking research is an im-         2003), and research on “project overload” (as the
portant foundation for our own model but has           name itself suggests) has focused solely on the
been conducted solely at the individual level,         negative implications of individuals’ overcom-
has dealt primarily with people’s cognitive ca-        mitment (Zika-Viktorsson et al., 2006). This re-
pacities, and generally has addressed switch-          search rarely acknowledges or models simulta-
ing frequency and switching costs in terms of          neously positive and negative effects operating
extremely small time periods— generally less           through different processes.
than a minute (Altmann & Gray, 2008).                    Although the research summarized above is
   Thus, existing research sheds little light on       valuable in various ways, it does not resolve the
the team-level implications of individuals’            critical tensions associated with multiple team
working on two or more teams, and it does not          membership, learning, and productivity. Resolv-
address the effects of individuals’ switching be-      ing these tensions is theoretically interesting
tween or among teams. Like individual multi-           because existing research does not explain how
tasking, multiple team membership leads to             or why multiple team membership can yield
some task switching (by definition), but it can        both positive and negative effects, nor does it
also include a much broader set of switches            address why organizations adopt it despite
between team contexts (rather than between             much advice to the contrary (e.g., Wheelwright &
simple tasks). Those team contexts often include       Clark, 1992). We focus on the relationships with
different tasks, roles, routines, technologies, lo-    productivity and learning because they are the
cations, and so forth, which make switching be-        key outcomes that are in dynamic tension for
tween them both more effortful (in terms of time       organizations using multiple team membership
and attention) and more potentially valuable in        as an approach to structuring work. Further-
terms of learning.                                     more, each has its strong proponents in the lit-
   At the team level, the research most relevant       erature on individual and organizational perfor-
to multiple team membership concerns bound-            mance, some positing that “productivity is
ary spanning. This work (e.g., Joshi, Pandey, &        arguably the most important measure of team
Han, 2009) shows how having members span               success” (Thompson, 2008: 36-37) and others
boundaries can affect teams’ emergent states           claiming that “the ability to learn and adapt is
and processes. Importantly, however, this re-          critical to the performance and long-term suc-
search is framed in terms of individuals’ cross-       cess of organizations” (Argote & Miron-Spektor,
boundary roles and activities regarding a single       in press). Thus, we focus on productivity and
team—paying relatively little attention to the         learning because they are (1) central to many of
multiple team contexts people may span. In ad-         the other outcomes sought by individuals,
dition to the boundary spanning research, a few        teams, and organizations; (2) critical compo-
studies address how groups divide their atten-         nents of a holistic view of performance (Hack-
tion across multiple tasks (Waller, 1997), but         man & Katz, 2010); (3) most vulnerable to the
they do so within the bounds of a single focal         fragmentation of time and attention (Ocasio,
team. They do not address the competing de-            1997); and (4) applicable across levels of analy-
mands on team members’ time that are gener-            sis. Multiple team membership may well affect
ated by their multiple team memberships.               other variables as well (e.g., identity and career
   In summary, while providing valuable in-            progression), and we hope such effects will be
sights into processes and situations related to        explored in future research.
multiple team membership, existing research
typically addresses only individual multitask-
                                                                        MODEL DEVELOPMENT
ing, does not consider the distinct impact of be-
ing on more than one team, and examines only             Before presenting the model, it is important
productivity or learning—not both. In addition,        that we define several of the key concepts on
previous research on the effects of fragmented         which it is based (team, membership, and time
attention has typically addressed either the pos-      period), as well as the two dimensions of multi-
itive or negative effects of that fragmentation,       ple team membership (number and variety).
but not both. For example, research on interrup-       Teams are bounded sets of individuals who
tions has treated them as either harmful (e.g.,        work interdependently toward a shared out-
Perlow, 1999) or beneficial (e.g., Zellmer-Bruhn,      come (Hackman, 2002). Individuals are members
464                                    Academy of Management Review                                     July

of a team when they share the responsibility           etc.) characterizing the teams that individuals
and reward (or penalty) for the outcomes of the        are members of and that a focal team overlaps
team’s work and recognize each other as mem-           with. We draw on Harrison and Klein’s work to
bers of the team. Finally, the time period during      conceptualize variety of team memberships as a
which we consider team membership is context           form of diversity capturing the “composition of
specific. In contexts where teams are relatively       differences in kind, source, or category of rele-
short-lived (e.g., computer emergency response         vant knowledge or experience among unit mem-
teams or hospital emergency room teams), twen-         bers” (2007: 1203). This type of variety is typically
ty-four to forty-eight hours is a period during        measured using entropy indices like Blau’s
which multiple team membership and its effects         (1977) or Teachman’s (1980), which gauge how
could be assessed meaningfully. In contexts            widely spread an entity is (i.e., evenness) across
where teams are longer-lived (e.g., software de-       how many different categories (i.e., richness),
velopment), the relevant period might be weeks         and then standardized and cumulated across
or months. Thus, any empirical study of multiple       the diversity variables relevant in any given
team membership must closely consider the              context. For example, Cummings (2004) aver-
general context (Johns, 2006) and temporal struc-      aged Teachman’s indices of team members’
tures (Ancona, Goodman, Lawrence, & Tush-              geographic locations, functional assignments,
man, 2001) of the research setting.                    reporting managers, and business units to mea-
   In addition, although people can occasionally       sure variety’s relationship with knowledge shar-
extend their work hours, we assume that the            ing and team performance. For our purposes, an
total time available for people’s work is finite.      even spread across the richest number of infor-
Accordingly, the time individuals dedicate to          mation sources and diversity variables yields
any one team must be reduced when they be-             maximum variety for each individual’s portfolio
come members of multiple teams. Furthermore,           of teams or for any given focal team.
for every additional team that someone joins, he          As we describe in detail later, variety or “the
or she must shift his or her attention and activity    number and spread of ‘batches’ of information
at least once from Team 1 to Team 2 (and some          content, experience, or unique network ties
individuals will choose or feel compelled to shift     available across unit members . . . broaden[s]
their focus back and forth more frequently).           the cognitive and behavioral repertoire of the
   As multiple team membership has become              unit” (Harrison & Klein, 2007: 1204). The “hetero-
more common over recent decades, two types of          geneity of new ideas, processes, and routines”
theories have emerged as especially valuable           that is valuable for individuals and teams
for understanding the dynamics of twenty-first-        comes from “other concurrent and past teams”
century work. Attention-based theories high-           (Zaheer & Soda, 2009: 3). There are many vari-
light the increasing number of demands that            ables across which one might measure variety
compete for people’s attention (Hansen & Haas,         of teams; in practical terms, examples of the
2001; March & Simon, 1958; Ocasio, 1997).              variables affecting the relationships among va-
Social network theories highlight how network-         riety, productivity, and learning include (but are
enabled exposure to a wider variety of informa-        not limited to) members’ roles, network ties,
tion affects learning and productivity (Reagans        functional experience, and industry background
& Zuckerman, 2001). Thus, in our model we focus        and teams’ tasks, norms, locations, and technol-
on the number and variety of teams as two re-          ogies in use.
lated but distinct dimensions, which we predict           Variety of multiple team memberships is par-
drive the effects of multiple team membership          tially structural and partially the result of indi-
on productivity and learning. The first dimen-         vidual and managerial actions. Structurally,
sion is the number of teams an individual is           some organizations may just be more complex
concurrently a member of, which would be cap-          and diverse, with the potential for higher variety
tured at the team level as the number of unique,       of team memberships as a result. However, com-
nonoverlapping “other” teams the focal team’s          plex organizations can increase or decrease this
members are also involved with.                        kind of variety, based on how they assign peo-
   The second dimension of multiple team mem-          ple to teams. Individuals can—to varying de-
bership—variety of team memberships—refers             grees—increase or decrease their number of
to the diversity (in tasks, technologies, locations,   team memberships. Similarly, any given team
2011                                  O’Leary, Mortensen, and Woolley                                  465

or team leader can seek out members who are               Before doing so, it is important to note several
able to devote more (or less) of their time to the     assumptions that we make in our model. First,
team, thus decreasing the number of multiple           we hold constant those constructs that could
memberships at the team level.                         potentially affect productivity and learning but
  Conceptually, the number and variety of team         that are not explicitly included in our model. For
memberships parsimoniously capture the chal-           each proposition regarding number of member-
lenges and opportunities for productivity and          ships, we hold variety of memberships constant
learning posed by a multiteam environment. For         (and vice versa for each proposition regarding
example, at the individual level, being a mem-         variety of memberships). Second, a complete
ber of more teams motivates individuals to find        model of multiple team membership’s anteced-
more efficient work practices, but it can also         ents and consequences is not our intent. Rather,
reduce the time and attention necessary for            we hope our model captures the key elements of
learning to occur effectively. Similarly, a focal      multiple team membership’s effects on produc-
team whose members are on a wider variety of           tivity and learning and stimulates additional
other teams exposes the focal team to more             theoretical and empirical work regarding these
unique information, which can stimulate learn-         and other aspects of people’s experience work-
ing, but it also poses coordination challenges for     ing in multiple teams. In the sections that follow
the team, which can reduce productivity. We            we present four core propositions at the individ-
develop these relationships and our overall            ual and team levels. We summarize these prop-
model in more detail below.                            ositions in Figure 1.

                                           FIGURE 1
   Relationships Between Multiple Team Membership Variety and Number and Productivity and
                          Learning at the Individual and Team Levels
466                                    Academy of Management Review                                   July

The Curvilinear Effects of the Number of               one or more team projects. In contexts such as
Multiple Team Memberships on Productivity              law and consulting, this is the percentage of
                                                       “billable hours” that individuals have in their
   At the individual and team levels, productiv-
                                                       schedules. In general, organizations strive to
ity refers to the ability to create products or
                                                       keep employees actively engaged in project
services that meet the expectations of key stake-
                                                       work, minimizing everyone’s “beach time” and
holders in a given time period with a given set
                                                       maximizing everyone’s utilization. Utilization is
of human and other resources (Adler et al., 2009).
                                                       measured at the individual level (e.g., what per-
The key distinction between productivity at the
                                                       centage of time the TV crew’s camerawoman is
individual and team levels is the range of re-
                                                       actively working on a story) and, unlike turn-
sources that must be coordinated at the team           around, exists at the team level only as an ag-
level (including disparate information, sched-         gregation of individual team members’ utiliza-
ules, and social dynamics), introducing addi-          tion. While multiple team membership may
tional complexity, which goes beyond simple            affect people’s workloads, it does not exert its
aggregation effects (Steiner, 1972). At both levels    influence solely by increases in workload. The-
productivity is a complex, multidimensional            oretically, one could be on more teams yet have
phenomenon (including various manifestations           less work (or vice versa).
of quality, quantity, time, value, etc.) and is con-      Although turnaround and utilization are distinct
sidered one of the key criteria for work effective-    dimensions of productivity, they are not fully in-
ness (Adler et al., 2009; Thompson, 2008). We          dependent. Individuals can be very productive in
consider two of the most common dimensions of          terms of turnaround on a given project without
productivity—turnaround and utilization—as             being fully utilized (i.e., while still having time
they relate to multiple team membership.               available to work on other team projects); how-
   Turnaround captures the amount of time used         ever, increases in utilization will also eventually
to produce a given quantity of goods or servic-        affect turnaround (Kc & Terwiesch, 2009). As we
es—the elapsed time from the receipt of a task to      describe in detail below, the number of multiple
its completion. For example, if it took a TV news      team memberships affects turnaround via the de-
crew one week to produce a five-minute feature         velopment of better individual and team work
story after it was assigned, the team’s turn-          practices, while it affects utilization through the
around would be one week for that piece of             better allocation of individuals’ time.
work. Turnaround incorporates both the actual             Individual productivity. At the individual
process time (when the crew was actively work-         level, we propose that the number of multiple
ing on the story) and the queue time (when the         team memberships improves productivity in
feature story got set aside for a day because the      terms of both utilization and turnaround time by
crew had to a cover a story about a breaking           facilitating load balancing across team projects
news event). If the crew members find ways to          and by focusing people on key priorities and
produce the story more efficiently, they can           efficient individual work practices within each
lower their process time and, consequently,            team. However, as multiple team membership
their turnaround. Reducing the influence of            increases above a moderate level, queue times
other demands can reduce queue time and, con-          for individual projects increase as individuals
sequently, turnaround. Another tactic would be         struggle with competing demands, offsetting
to reduce quality to reduce turnaround time.           some of the efficiencies initially achieved and,
However, for the purposes of our model, we hold        eventually, decreasing productivity.
quality constant. Turnaround time can be mea-             By enabling people to allocate their time and
sured at the individual or team level, represent-      attention in ways that reduce downtime, being
ing the distinct combination of inputs at each         on more teams increases the utilization aspect
level. The more process losses a team experi-          of individual productivity. Workloads are inevi-
ences, the longer its turnaround time will be          tably uneven, with teams’ demands on individ-
(Steiner, 1972).                                       ual members’ time varying significantly over
   Utilization captures the extent to which re-        the course of those teams’ life cycles (Evans,
sources are being used as opposed to sitting           Kunda, & Barley, 2004; Westenholz, 2006). As a
idle. For our purposes, it refers to the percentage    result, individuals frequently face unproductive
of time an individual is actively engaged with         downtime because they are waiting for hand-
2011                                  O’Leary, Mortensen, and Woolley                                467

offs from someone else or because there is a lull      shown, if people expect that they will have to
in a project (Yakura, 2001). Being on more teams       divide their time and attention between multi-
concurrently gives individuals more opportuni-         ple teams, they will “develop a mode of working
ties to offset the ebbs in one team’s work with        faster . . . to compensate for the time they know
the flows of another team’s work. The more             they will lose” (2008: 110).
teams individuals are on, the less likely those           For example, Kc and Terwiesch (2009) found
individuals are to have gaps in their schedules        that splitting staff across more projects in-
and the more fully “utilized” they will be. Mil-       creased productivity, with the busier staff work-
grom and Roberts (1992: 409) note that being on        ing more efficiently. Jett and George (2003) also
only one team is “directly inefficient” because of     noted that being on multiple teams triggered the
the “uneven time-pattern to the work.” In prac-        kind of interruptions that can create a stimulat-
tice, many firms strive for maximum utilization        ing rhythm in individuals’ work practices, help-
of their employees’ time (Adler, Nguyen, &             ing their long-run performance. Thus, driven by
Schwerer, 1996), and the ability to assign time in     more team memberships, new, more focused
small increments to multiple teams supports            work practices enhance individuals’ productiv-
that utilization-maximizing goal.                      ity. As with load balancing, we expect that work
   However, this positive effect on utilization        practice efficiencies eventually plateau and are
does not continue unabated. Other things being         offset by increased queue time when the num-
equal, the positive load-balancing effect of mul-      ber of multiple team memberships is high.
tiple team memberships on utilization tapers off          As individuals take on larger numbers of
over time. This is consistent with Wheelwright         teams, each additional team exacerbates the
and Clark’s (1992) findings of increased produc-       division of people’s attention and slows their
tivity when engineers began adding projects,           reengagement with any one team’s work (Hopp
but decreasing productivity as they continued to       & Van Oyen, 2004; Huey & Wickens, 1993). Thus,
do so. Similarly, in a medical context where           even though multiple team membership en-
projects were patients, Kc and Terwiesch (2009)        hances utilization in a way that is “very attrac-
found that when hospitals split staff across           tive” for management, and even though it can
more projects, they achieved higher utilization,       lead to the development of more efficient work
but while such high utilization was maintained         practices, beyond a moderate level (possibly at
for long periods of time, productivity increases       only three teams) it introduces bottlenecks and
were not sustainable and eventually dropped            slows turnaround (Slomp & Molleman, 2002).
off.                                                      Team productivity. At the team level, we pro-
   Being on an increasing number of teams cre-         pose that the number of team memberships in-
ates time pressure, which can lead people to           creases productivity by prompting teams to
develop work practices that reduce process time        adopt more efficient collective work practices,
and (as long as queue time is held to a mini-          while it simultaneously decreases a team’s op-
mum) improve overall turnaround (Kc & Ter-             portunities to work collectively. In this sense,
wiesch, 2009; Svenson & Maule, 1993; Waller,           the mechanism by which multiple team mem-
Conte, Gibson, & Carpenter, 2001). When people         berships increase team productivity is analo-
add second, third, fourth, or more teams to their      gous to the one that increases individual pro-
daily or weekly activities, it “demands that in-       ductivity, but it is manifested in terms of team
dividuals enact specific efforts to coordinate,        members’ collective consciousness about the
manage, and track those collaborations” (Gon-          time pressures facing them (Waller et al., 2001)
zalez & Mark, 2005: 144) and leads them to think       and teams’ collective efforts to develop more
more carefully about how they use the fractions        efficient team practices.
of their time that are available to each team             Without at least some mild stress on the sys-
(Karau & Kelly, 1992). Among other things, the         tem, people tend to budget more generously
time pressure created by more team member-             than the task actually demands (Brooks, 1995),
ships leads people to prioritize, sequence, and        and they are less likely to find more efficient
“time box” or compartmentalize their available         methods of conducting their work (Huey & Wick-
hours more actively (Hudson et al., 2002; Tobis &      ens, 1993). Knowing that they have smaller frac-
Tobis, 2002) and to be more focused when they          tions of each other’s time, and knowing that the
are working. As Mark, Gudith, and Klocke have          coordination of that time will be challenging,
468                                   Academy of Management Review                                 July

team members develop ways to accomplish              and distinct dimension is the variety of those
more in less time. These practices may include       teams. As we describe below, this variety has a
more focused, structured meetings, in which          negative effect on productivity. In settings
team members consciously spend more time on          where the number and variety of teams are pos-
task and less time on social, relational, or other   itively correlated, the negative effect of member-
interactions. The pressure on team members’          ship variety may offset some or all of the pro-
schedules from being on multiple teams “can          ductivity gains that arise from the initial
trigger certain activities by teams to reassess      increases in the number of memberships.
[their] existing structures and enact new struc-        Individual productivity. At the individual
tures” (Fuller & Dennis, 2004: 2). There is even-    level, higher variety in the teams one is a mem-
tually a quality/quantity trade-off, but teams       ber of means a greater amount of information
working under tighter time constraints do tend       must be managed, necessitating that more of
to produce at a faster rate (Kelly & McGrath,        one’s time and effort must be spent adjusting to
1985; Waller, Zellmer-Bruhn, & Giambatista,          different team contexts and their associated
2002).                                               people, tasks, technologies, roles, locations, and
   Although an increasing number of team mem-        so forth (DeMarco, 2002; Huey & Wickens, 1993;
berships can decrease processing time through        Milgrom & Roberts, 1992). These switching costs
more efficient work practices, it can also in-       reduce individual productivity by increasing
crease queue time, as competing demands on           turnaround. The more different the “working
members’ time from other teams reduce the time       spheres” associated with each team, the more
available for synchronous work in any focal          the switches between those teams disrupt rou-
team. Because they have to divide their time         tines and hurt productivity (Mark, Gonzalez, &
across multiple teams, a focal team’s members        Harris, 2005). Holding the number of member-
have less than 100 percent of their time to work     ships constant, the variety of memberships will
on each team, and the blocks of time they do         be negatively related to individual productivity
have available are less likely to be aligned. This   because of increased information load, leading
temporal misalignment means that more work           to greater processing time and, consequently,
must be done asynchronously, coordinated, and        turnaround. When one switches between three
then reintegrated, which increases the team’s        relatively similar teams, the diversity of infor-
queue time (Postrel, 2009; Wittenbaum,               mation to be managed is reduced, and switch-
Vaughan, & Stasser, 1998). Longer queue times        ing has far less of an effect on productivity than
eventually offset the gains in processing time       switching between three relatively different
arising from better team work practices, and,        teams (Hopp & Van Oyen, 2004; Rubinstein,
thus, we expect the work-practice benefits of        Meyer, & Evans, 2001). When variety is greater,
having many members with multiple team com-          job scope and complexity are greater and are
mitments to yield diminishing returns for            accompanied by high levels of strain, leading to
teams—increasing at a decreasing rate toward         reduced productivity (LePine, Podsakoff, &
an asymptote.                                        LePine, 2005). As we discuss later, individuals
                                                     can manage these switching costs more or less
      Proposition 1: The relationship be-
                                                     effectively depending on their work practices.
      tween the number of teams individu-
                                                        Team productivity. At the team level, the va-
      als are members of and productivity at
                                                     riety members experience as they switch be-
      the individual and team levels is cur-
                                                     tween teams results in lower productivity for the
      vilinear; the positive relationship in-
                                                     focal team. This is due to the increased coordi-
      creases at a decreasing rate and even-
                                                     nation costs among members, leading to longer
      tually turns negative.
                                                     turnaround times. Variety in team membership
                                                     increases the complexity of the information
                                                     teams’ members must manage (Cronin & Wein-
Negative Effects of Variety of Team
                                                     gart, 2007) and the likelihood that members’
Memberships on Individual and Team
                                                     schedules will be difficult to align. In addition,
Productivity
                                                     as the variety of team memberships increases,
  Moving beyond the effects of the number of         teams must devote more time to managing the
teams an individual is a member of, a separate       associated variance in perspectives, mental
2011                                  O’Leary, Mortensen, and Woolley                                  469

models, and capabilities of team members               that the variety of multiple team memberships
(Massey, Montoya-Weiss, & Hung, 2003; Zaheer           generates more varied inputs and creates suf-
& Soda, 2009), which decreases productivity.           ficient interpersonal connections to stimulate
                                                       learning for individuals and teams.
       Proposition 2: The variety of multiple
                                                          Individual learning. At the individual level, as
       team memberships is negatively re-
                                                       variety of memberships increases, people have
       lated to individual and team produc-
                                                       access to more diverse inputs and, thus, more
       tivity.
                                                       opportunities to learn (Mark et al., 2005). Variety
                                                       is a critical component of individual learning
                                                       (Schilling, Vidal, Ployhart, & Marangoni, 2003;
Positive Effects of Multiple Team Membership
                                                       Wiersma, 2007). Deliberate sequential variation
Variety on Individual and Team Learning
                                                       in employees’ contexts is a traditional element
   Although the variety of multiple team mem-          of job rotation and other personnel movement,
berships will have negative effects on produc-         which scholars across several disciplines have
tivity (Proposition 2), we argue that such variety     shown enhances individual learning (e.g., Or-
can enhance learning, primarily through the in-        tega, 2001). Unlike traditional job rotation, mul-
creased variety of ideas and information from          tiple team membership allows for concurrent
which to draw insights. Holding the number of          and serendipitous variation in the information
multiple team memberships constant, exposure           one has ready access to (Hudson et al., 2002), as
to a wider variety of inputs can reduce the pos-       well as the opportunity for more immediate ap-
sibility of tunnel vision and raise the probability    plication and integration of that new knowl-
that better ideas and approaches will be discov-       edge. Firms that assign employees to only one
ered. Although multiple team membership–               team at a time “might then be at a competitive
driven variety enhances learning initially, it         disadvantage in the labor market” because of
does not do so ad infinitum; eventually, the           the lost opportunities for learning that multiple
learning benefits plateau. We propose that this        team membership affords (Milgrom & Roberts,
effect of membership variety on learning occurs        1992: 409). Contextual variety also can stimu-
at the individual and team levels, but we note         late learning processes themselves, espe-
that learning at these levels is a distinct process    cially when people’s multiple teams expose
(Argote & Miron-Spektor, in press), which differs      them to more “cool” (Grabher, 2002) or motivat-
in three key ways.                                     ing work (Hackman, Pearce, & Wolfe, 1978),
   First, learning at each level differs with re-      involving knowledge or skills that are per-
spect to content. Team learning is, by its very        ceived to be more valued or desirable. Holding
nature, not solely about an increase in domain         the number of teams constant, we argue that
knowledge (which would reflect aggregated              multiple team membership–driven individual
individual learning); it is also about improv-         learning occurs when being on multiple teams
ing processes and team “repertoires” (Wilson,          increases the variety of experiences individu-
Goodman, & Cronin, 2007). Second, learning at          als are exposed to.
each level differs with respect to scope. In              While a moderate amount of variety can aid
contrast to learning at the individual level, in       learning, several studies on analogical learning
which one person samples from a set of teams           have shown that it is difficult for individuals to
and experiences and then integrates that               transfer analogous solutions when contexts are
knowledge, at the team level, many individu-           too different (e.g., Novick, 1988; Reeves & Weis-
als sample from many different teams. Third,           berg, 1994). As a result, we expect that high
although research on learning at different lev-        levels of dissimilarity across teams will eventu-
els is converging (Argote, in press), individual       ally hinder learning. Thus, although we expect
and team learning differ in terms of mecha-            the relationship between variety of teams and
nisms, where individuals can learn from the            learning to be positive, it is likely to reach a
simple exposure to or transfer of new informa-         “saturation point” (Dahlin, Weingart, & Hinds,
tion, whereas teams require shared experi-             2005; Kenis & Knoke, 2002). Beyond that point, the
ence and the development of a new set of               diversity of inputs is so great and members’
behaviors or repertoires in order to learn (Ar-        information so varied that the combination is
gote & Todorova, 2007). Therefore, we argue            unlikely to trigger any additional learning in the
470                                   Academy of Management Review                                 July

team. In short, learning appears greatest when       and team levels, the primary mechanisms
there is both some difference and some overlap       through which learning is undermined are (1)
in members’ skills and experiences (Heimerl &        the reduction in the time available to attend to
Kolisch, 2010).                                      and integrate new information effectively and
   Team learning. At the team level, the diver-      (2) the effect that a greater number of teams has
sity of inputs in teams resulting from member-       in focusing individual and team attention on
ship variety enhances team cognition (Hinsz,         only critical, immediate tasks.
Tindale, & Vollrath, 1997) and, ultimately,             Individual learning. As discussed above,
team learning (Subramaniam & Youndt, 2005;           more team memberships lead individuals to pri-
Wong, 2008). Along with productivity, team           oritize key tasks and to seek out efficient meth-
learning is one of the “key criteria” for team       ods of task completion. Furthermore, while more
effectiveness (Thompson, 2008: 36). This link        team memberships might lead individuals to
between diverse inputs and team learning has         work more efficiently, they also deprive individ-
been found by many scholars in multiple con-         uals of the time needed to seek out and integrate
texts (Bunderson & Boumgarden, 2010). In-            new information. Time pressure that is too high
creasing membership variety means that               limits people’s exploratory thoughts, behaviors,
members of the focal team are experiencing           and ability to encode and retrieve knowledge
work in more different teams and can bring           (Amabile & Mueller, 2008; Bailey, 1989), which is
knowledge acquired in those other teams to           detrimental for learning (Jett & George, 2003;
the focal team. As Ruff notes, membership va-        Perlow, 1999). In contrast, individuals working
riety promotes learning because “each team           on a smaller number of teams have more time
member maintains a broad set of knowledge            per team. This results in less time pressure and
and methods,” and concurrent “work in very           allows individuals to leverage brief breaks be-
different projects encourages the discovery of       tween projects for subconscious learning
‘latent’ opportunities and promotes the ex-          (Zhong, Dijksterhuis, & Galinsky, 2008). More
change of knowledge” across different teams          slack time allows individuals to experiment
(2006: 292). However, as with the individual-        more actively with new approaches, to appreci-
level effects on learning, we expect a similar       ate new nuances more mindfully, and to gener-
threshold point above which increasing the           alize experiences from team to team (DeMarco,
variety of teams a given team overlaps with          2002).
(through its shared members) only makes in-             Team learning. As the average number of
formation gained from those teams less appli-        teams per member increases, each team mem-
cable to the focal team’s context and only           ber has less time to dedicate to the focal team,
makes it more difficult to sustain meaningful        making it more difficult for the team to integrate
connections with those teams (Kenis & Knoke,         knowledge and develop shared repertoires (Wil-
2002).                                               son et al., 2007). When teams have high schedule
                                                     constraints (e.g., their members’ schedules lack
      Proposition 3: Holding the number of
                                                     contiguous blocks of time), coordinating their
      team memberships constant, the vari-
                                                     efforts is more difficult (McGrath, 1991) and team
      ety of teams individuals are members
                                                     members typically have less slack time for the
      of is positively related to learning at
                                                     activities that foster collective learning (Haas,
      the individual and team levels, with
                                                     2006). Shared information processing activities
      learning increasing at a decreasing
                                                     are critical to team learning, and, as shown in
      rate.
                                                     analogous situations where membership actu-
                                                     ally changes, team members’ frequent comings
                                                     and goings hinder team learning (Van der Vegt,
Negative Effects of the Number of Team
                                                     Bunderson, & Kuipers, 2010). In contrast, when
Memberships on Individual and Team
                                                     teams spend more time together, they become
Learning
                                                     more familiar with one another and are better
  While increasing the variety of multiple team      able to generalize team-encoded roles and rou-
memberships can improve learning, increasing         tines across tasks (Lewis, Lange, & Gillis, 2005;
the number of concurrent team memberships            Staats, Gino, & Pisano, 2010), which is an impor-
can undermine it. At both the team individual        tant hallmark of team learning (Wilson et al.,
2011                                   O’Leary, Mortensen, and Woolley                                471

2007). Thus, because temporal constraints asso-         interpersonal relationships at work (Sluss &
ciated with the number of team memberships              Ashforth, 2007), and the geographic dispersion of
significantly inhibit teams’ ability to have such       work (Thatcher & Zhu, 2006) may all trigger com-
real-time interaction, team learning suffers.           peting identities. Multiple team membership
                                                        creates potentially competing team-level bases
       Proposition 4: The number of teams
                                                        for identification, increases the number of rela-
       individuals are members of is nega-
                                                        tionships people have, and appears to be corre-
       tively related to learning at the indi-
                                                        lated with geographic dispersion. Thus, given
       vidual and team levels.
                                                        how easy it is to trigger intergroup competition
                                                        (Tajfel, 1981), membership in multiple teams
                                                        within the same organization may be enough to
        DISCUSSION AND IMPLICATIONS                     cause identity-related tensions and conflict
                                                        (Fiol, Pratt, & O’Connor, 2009). Because most re-
   Although many (if not most) academics have           search on identification has addressed organi-
personal experience with multiple team mem-             zational targets (Johnson, Morgeson, Ilgen,
bership (working concurrently on multiple               Meyer, & Lloyd, 2006), multiple team member-
teaching, research, and service teams), to the          ship represents an important context (and
best of our knowledge, this article is the first        cause) in which to understand how individuals
attempt to model the mechanisms driving its             identify with multiple, alternative, work-related
effects on individuals and teams. We believe            targets.
this has numerous theoretical and method-                  In addition, organizational and social skills,
ological implications for scholars and practi-          as well as other individual characteristics re-
tioners.                                                lated to multitasking, time allocation, and the
                                                        pursuit of multiple goals (e.g., Hecht & Allen,
                                                        2005; Schmidt & Dolis, 2009; Schmidt, Dolis, &
Scholarly Implications—Theoretical                      Tolli, 2009), are likely to rise in importance in
                                                        settings where individuals must navigate ten-
   We believe a focus on multiple team member-          sions among competing teams and priorities.
ship suggests a number of intriguing and impor-         Individuals and teams are likely to adopt a va-
tant directions for future research. These include      riety of practices in response to the pressures
individual-level research on identity issues and        and opportunities they experience in a multi-
employee skills that are conducive to multiple          team environment. The effectiveness of these
team membership, team-level research on the             practices will be an important topic for future
connections to geographic distribution, cross-          research, as will individuals’ ability to manage
level research on context switching and produc-         multiple commitments and to say no to requests
tivity, and multilevel research on information          that exceed their capacity.
transparency. While we briefly address each of             At the team level, the relationship between
these in turn, we do not intend them as an ex-          geographically dispersed work and multiple
haustive list but, rather, as examples of areas         team membership is another area for future
for future work in this domain and an attempt to        research. Accessing individual expertise is a
stimulate future research.                              key motivator for using multiple team mem-
   At the individual level, a shift to the multi-       bership as a way to structure work, since it
team perspective has strong implications for re-        allows teams to leverage the time of experts
search on identity and multiple identities. Be-         more efficiently by allowing them to use their
ginning with early work by Tajfel (1981), we now        time on an as-needed, less-than-100-percent
have a large body of theory and research on             basis. Similarly, distributed work in organiza-
social identity and categorization within organi-       tions is often motivated by the desire to take
zations (see Hogg & Terry, 2000). There is also a       advantage of specific expertise that is not
burgeoning body of literature on multiple and           physically collocated (e.g., Boh, Ren, Kiesler, &
dual identities (e.g., Foreman & Whetten, 2002;         Bussjaeger, 2007). Cummings and Haas (in
Hillman, Nicholson, & Shropshire, 2008). Poten-         press) began to explore these issues, finding
tially competing spheres of one’s life (e.g., work      that being on multiple teams (and having
and family; Rothbard, Phillips, & Dumas, 2005),         members committed at high levels of time to
472                                   Academy of Management Review                                  July

the focal team) improves focal team perfor-           positive effects on learning. Being more spe-
mance, except when geographic dispersion is           cific about the dimensions of multiple team
high. In that case committing significant time        membership (and the associated outcomes)
still helped, but being on multiple teams hurt        provides a more nuanced view of the phenom-
performance—a finding consistent with                 enon—a view that helps explain why organi-
Lojeski, Reilly, and Dominick (2007).                 zations have adopted it as a way of structuring
   Although we do not specifically address the        work. This more nuanced view may also be
dynamics of multiple team membership at the           useful in helping us understand why there is
organizational level in this article, there are in-   an apparent disconnect between the individ-
teresting challenges for coordination and re-         ual and team experience of multiple team
source sharing across teams that are intercon-        membership and the organizational adoption
nected by membership. Work on multiteam               of it. For example, organizational decisions to
systems has provided numerous insights into           adopt multiple team membership may be
related issues of cross-team coordination. For        driven by a managerial focus on utilization
example, Marks and colleagues (DeChurch &             and flexible resource deployment, without a
Marks, 2006; Marks, DeChurch, Mathieu, Panzer,        clear understanding of the implications for in-
& Alonso, 2005) have examined issues of leader-       dividuals and team. Further research and the-
ship, teamwork, and coordination in environ-          orizing could explore this potential relation-
ments where multiple teams work together to-          ship.
ward a single ultimate goal. To date, however,
that research has largely conceptualized such
                                                      Scholarly Implications—Methodological
teams as independent with respect to member-
ship. Thus, it would help to have a better under-        Enhancing our understanding of multiple
standing of how these processes unfold when           team membership also provides an opportunity
teams are not only interdependent with respect        for methodological innovation. Innovative mul-
to their goals but also with regard to their mem-     tilevel analysis (see Klein & Kozlowski, 2000) is
bership. For example, how can organizations           particularly important because of the noninde-
best coordinate the work of teams when they           pendence of teams in multiple team member-
share members? How can human and technical            ship contexts, and even studies not explicitly
systems support that coordination most effec-         focused on multiple team membership should
tively?                                               consider controlling for the nonindependence of
   We began this article by noting that organi-       teams. The interdependence in team member-
zations adopt multiple team membership as a           ships may also fuel certain phenomena (e.g.,
way to organize work, despite the apparent            contagion or diffusion).
problems associated with switching between               Furthermore, multiple team membership
teams. When multiple team membership is not           poses an interesting challenge for acquiring
managed carefully, these problems can un-             accurate information regarding the amount of
doubtedly be profound. However, we believe            effort individuals put into different projects, as
that a purely negative view of multiple team          well as their performance on them (Meyer, Ol-
membership is shortsighted. As we have ar-            sen, & Torsvik, 1996). Workers in multiteam
gued in our model, its effects are both positive      environments may underreport or overreport
and negative, depending on the dimension of           their hours on different projects for a variety of
multiple team membership involved (variety            reasons. This can result from organization-
or number) and the outcome in question (learn-        based or information systems–based limits to
ing or productivity). The specific aspect of pro-     the number of hours or number of projects em-
ductivity (turnaround or utilization) also mat-       ployees can report, or it can result from indi-
ters. The majority of commentary on multiple          viduals’ attempts to carry over, buffer, or
team membership focuses on the number of              hoard time (Yakura, 2001). In situations where
teams and the turnaround aspect of productiv-         such underreporting or overreporting is likely,
ity, where the effects of number are decidedly        researchers can assess individuals’ time com-
negative. However, the number of team mem-            mitments using multiple methods, such as sur-
berships has positive effects on utilization,         veys or time diaries of individuals and man-
just as the variety of team memberships has           agers, as well official organizational time-
2011                                O’Leary, Mortensen, and Woolley                                  473

tracking systems. Triangulating among these          viduals can minimize the potential delays as-
data sources will provide a more robust under-       sociated with the number of team member-
standing of how people divide their time, as         ships by time-boxing portions of their
well as a better sense of how actual and “of-        schedule (Jalote, Palit, Kurien, & Peetham-
ficial” time use compare. Studying people on         ber, 2004), by not switching midtask (Louko-
multiple teams may also be helpful from a            poulos, Dismukes, & Barshi, 2009), and by not
methodological standpoint because people on          switching to easier work. However, doing so is
multiple teams have a current basis for com-         not simple given people’s tendency to be
parison; they do not have to reach back in their     “switchy” and to shift toward easier work
memories to answer common survey questions           (Payne, Duggan, & Neth, 2007). In addition,
beginning with the phrase “In comparison to          scholars have found that people interrupt
other teams I have been a member of. . . .”
                                                     themselves by switching between the work of
                                                     multiple teams at least as often as they are
Managerial Implications                              forced to switch by external forces (Hudson et
                                                     al., 2002; Mark et al., 2005), and—when faced
   Knowing how multiple team membership af-
fects individual- and team-level learning and        with two challenging tasks and the belief that
productivity, individuals, team leaders, and         they cannot finish both—they tend to focus on
managers can be more mindful of the implica-         the work they can complete more easily
tions. The effects of multiple team member-          (Schmidt & Dolis, 2009). To the extent that in-
ship are not purely structural and are subject       dividuals exercise some volition and can man-
to individual agency or managerial interven-         age their shifts from team to team and avoid
tion. In this sense they are akin to other types     these general tendencies, they can minimize
of opportunities provided by networks—               the productivity-decreasing delays that multi-
opportunities that are both “purposive” (agen-       ple team memberships cause. By effectively
tic) and “positional” (Zaheer & Soda, 2009: 4),      batching or sequencing their work so that they
enabling and constraining (Giddens, 1984). We        make fewer switches between widely varying
believe our model identifies potential lever-        teams, individuals can also ameliorate the
age points for practitioners seeking to maxi-        switching costs of membership variety on pro-
mize the upside of multiple team membership          ductivity.
while minimizing its downside. Some of these            From a managerial perspective, the effects of
leverage points include the timing and selec-        multiple team membership on team productivity
tion of switching between teams, the active          and learning also will be moderated by the ac-
coordination of schedules across teams, and          tive coordination of schedules across teams.
the explicit definition of roles within teams.       When teams’ schedules have nonoverlapping
Individuals’ and managers’ interventions at          deadlines and more temporally contiguous
these and other leverage points can make a           blocks of time devoted to the team’s work such
major difference in how effectively these envi-
                                                     that members are ready to receive hand-offs
ronments operate and can potentially help ex-
                                                     from teammates and proceed with their portions
plain how some firms prosper with staff com-
                                                     of the task without a lag, the teams can reap the
mitted to two to six times more teams than
                                                     benefits of greater team member utilization and
their competitors (Milgrom & Roberts, 1992:
449).                                                efficiencies in processing time, without the off-
   Multiple team membership can lead to si-          setting costs of greater queue times. Further-
multaneous multitasking and overly frequent          more, aligning team member schedules pro-
task switching, but it need not involve high         vides greater opportunities for team members to
levels of either if people can control their         engage in the synchronous interaction neces-
schedules and work habits (Spink, Cole, &            sary to share critical information, reflect on les-
Waller, 2008). The timing and selection of           sons learned, and codify new routines and rep-
switches typically is a combination of individ-      ertoires (Edmondson, Dillon, & Roloff, 2007;
ual, managerial, and contextual factors, with        Wilson et al., 2007). Savvy managers can adopt
individuals almost always able to exert some         scheduling practices (e.g., regular weekly meet-
control over their switches. For example, indi-      ings at fixed times) to enable teams to learn and
You can also read