'^ o/
Tt^"^
rrt/-
—B'
ALFRED
P.
WORKING PAPER
SLOAN SCHOOL OF MANAGEMENT
Project Team Aging:
Thomas Allen
Ralph Katz
J.J. Grady
Neil Slavin
A Failure to Replicate
November 1987
CT/1963-87
MASSACHUSETTS
INSTITUTE OF TECHNOLOGY
50 MEMORIAL DRIVE
CAMBRIDGE, MASSACHUSETTS 02139
Project Team Aging:
Thomas Allen
Ralph Katz
J.J. Grady
:;eil Slavin
A Failure to Replicate
November 1987
WP/1963-87
FFB
5
1988
INTRODUCTION
The phenomenon
years by
to relate
a
"group aging"
of
number
is
one that has been reported over the
Shepard
of separate investigators.
group performance
to the
mean tenure
(1956)
was the
group members.
of
found that performance increased with increasing average tenure up
about
16
Pelz
and Andrews (1976) found
performance reached
its
maximum
yet another study,
years, just as Pelz and
& Alien,
1982)
Smitfi
at a
In
mean tenure
(1970)
showed
Andrews found.
reported, for 50 projects
a
Finally,
in
a
their study,
of four
years (Figure
performance peak
at
four
the present authors (Katz
U.S. chemical company,
clear curvilinear relationship with performance again
at
to
similar curvilinear
a
relationship between mean tenure and performance.
In
He
months, but decayed as group membership remained stable over
longer periods.
1).
first
reaching
a
a
maximum
about the three to four year point for mean project team tenure and
decaying significantly thereafter (Figure 2).
There
is
therefore substantia! evidence for the existence of the reported
curvilinear relationship between mean tenure and project team
performance.
The research has been conducted by
a
number
of
independent investigators, who have examined groups and project teams
number
very different industrial settings
across
a
time.
Katz and Allen (1982), after reporting their observations, go on to
of
at different points
in
attribute this effect to the well-known "not-invented-here" syndrome.
According
to this
explanation,
and supported by data reported by
themselves and by Pelz and Andrews, groups gradually define themselves
55
t
O
f'O
-.1
LJ
D;
UI
45
<
o
Li.
a;
UI
40
J
.35
L.
1?
MEAN tenure: OF GROUP MEMBERS
Figure
1
,
Performance as o
Furiction of
Mean
TerrJi'- of G: OiJp l/lerribers,
(Adapted from Peir
-
UJ
LO
3.£
&
Andrews. tQTf))
12
narrow
into a
have
a
field of specialization
monopoly on knowledge
in
and convince themselves that they
Such increased
their area of specialty.
specialization creates an appearance to the outside world of decreased
relevance, which leads to
decrease
a
communicate with and respond
the team's motivation to
in
As
Shoemaker, 1971; Katz and Allen, 1985).
expose themselves
information (Janis,
less
and
1972;
more narrow focus which
less to critical
Katz,
(Rogers and
to the outside world
1982;
a
long-tenured teams
result,
sources of external contact and
Allen,
1987).
It
turn leads to poorer performance.
in
group success often creates the conditions for eventual
group has been successful there
"winning team" together.
conditions may set
in
this isolation
is
If
is
a
In
sense,
When
failure.
tendency for management
a
to
and
a
keep the
however,
this continues for too long
which can lead to the observed deterioration
in
the
group's expected performance.
PRESENT STUD Y
Although relationships between performance and group age now have been
reported by at least four different studies, we
still
know very
little
about
those groups that to maintain creative performance over sustained periods
of time.
While there may be an "average" tendency
decline with higher levels of group age,
from which
to
project groups
and
still
there
is
performance
for-
to
as yet no empirical basis
suggest how one might organize and manage long-term
in
order to benefit from the continuity of
overcome the negative aspects
of
"NIH,"
i.e.,
a
team approach
of increased
insulation,
L
specialization,
stability,
and homogeneity (Katz and Allen, 1982
1985).
led the
of this
All
sample of
of the
181
authors to test the phenomenon further with
project teams
nine organizations.
in
study were twofold:
a
large
The primary objectives
from this new sample those
first to identify
long-tenured teams that were able to remain effective over an extended
period of time and secondly to try to discover how they were able to
accomplish this.
In
essence, what are some of the particular managerial
and organizational factors that significantly differentiate high-performing
long-tenured teams from low-performing ones, and what are the
group together and creative
implications of these findings for keeping a
over time?
Sample
The
selection of the nine participating organizations could not be
random, but they were chosen
industries.
the U.S.
Two
to
represent several distinct sectors and
of the organizations are
Department
Space Administration;
of
Defense the other
private industry:
one
in
government laboratories, one
in
in
the National Aeronautics and
three are not-for-profit firms doing most of their
business with government agencies.
in
made
two
in
The four remaining organizations are
aerospace, one
the packaged goods industry.
in
the electronics industry and
Project Performance
Since objective measures of performance that are comparable across
different technologies have yet to be developed,
measure similar
to that of
many
and Katz and Tushman
(1967)
a
subjective
including Lawrence and Lorsch
studies,
each organization, we measured
In
(1981).
we used
project performance by interviewing managers
who were
at least
one
hierarchical level above the project and functional managers, asking them
to indicate on a 5-point Likert-type scale
performing above, below, or
at
whether
a
project team was
the level expected of them, given the
particular technical activities on which they were working.
Managers
evaluated only those projects with which they were personally familiar and
knowledgeable.
Evaluations were made independently and submitted
As
confidentially to the investigators.
we did
in
the previously mentioned studies,
not prescribe the criteria to be used
by the managers, believing
that they should employ those criteria that they believed to be most
relevant.
included,
Discussions with the evaluators indicate that the criteria used
but were not limited
to,
schedule, budget, and cost performance;
innovativeness; adaptability; and the ability to cooperate with other parts
of the organization.
On the average, between four and
evaluated each project.
The evaluations showed very strong
consensus within each organization (Spearman- Brown
from
a
five
low of 0.74 to a high of 0.93).
It
is
managers
internal
reliabilities
range
therefore safe to average the
ratings of individual managers to yield reliable project performance scores.
In
addition to the internal evaluations,
5
an expert panel of independent,
outside
R&D
professionals exhaustively evaluated
project base (N=8).
The ordering
a
of their project
small subset of
our
performance evaluations
agreed perfectly with the ordering of our own aggregated measures of
Such agreement between two separate sources strongly
performance.
supports the validity of our project performance measures.
among
distinction between high and low project performance
project teams,
with
a
mean
performance measures were converted
of zero
to
To
clarify the
all
181
standardized scores
(the original sample mean was 3.32).
Team Tenure
Over 2,000 individuals
questionnaire.
in
the nine organizations completed
The instrument
fairly lengthy
dealt with a wide range of issues
which have been reported elsewhere (Katz L Allen,
1986).
a
1985;
Among the questions was one which asked each
many
of
Allen & Katz,
individual to
indicate the exact duration of his association with the project team which
was identified on the front page
(or
of
of the questionnaire.
Mean team tenure
group age) was calculated by iweraging the individual project tenures
all
project members.
There was
also a long series of questions dealing
with various relationships specific to the project assignment, to the
organization
In
s
management,
to the individual's
work orientation, and
each organization, short meetings were scheduled with
all
so on.
members
of
the technical staff to explain the general purposes of the study, to solicit
their voluntary cooperation and to distribute the questionnaire to each
6
Respondents varied
engineer individually.
mean
of 43
age from
in
21
to 65 with a
and standard deviation of 9.6 years.
RESULTS
When
project performance scores for the 181 projects are standardized and
plotted against the mean tenure of project team members, there
indication whatever of the curvilinear relationship reported
mean
of zero so that high
range.
tenure
of
There are
in
the earlier
performance scores
a
large
number
is
line in
the entire sample had
a
In
fact,
standard deviation units.
symmetrical throughout the tenure
of low
performing projects with mean
excess of four years, but there
high performers as well.
to an
and low performance deviations are
shown above and below the middle dotted
The dispersion
no
The performance scores have been standardized
studies (Figure 3).
overall
in
is
is
a
roughly equal number of
the highest performing project team
in
mean tenure greater than eight years, and two
project teams with mean tenures greater than
15
years have performance
scores that are more than one standard deviation above the mean.
Even when the data are
method, as they were
in
statistically
smoothed using Tukey's (1977) 3RSSH
our earlier study (Katz & Allen, 1982), no clear
pattern emerges (Figure 4).
A decay
in
performance
Many
inevitable consequence of "group aging"!
is
therefore not an
project teams
manage
maintain their performance levels with high levels of team tenure.
important question now becomes, what
7
is
it
to
The
that distinguishes the higher
4
fef-
m
K
t-
performing long tenured teams?
more closely
Project
in
But
first,
let
us examine the data
a
comparison with the earlier studies.
Type and Comparison
with Earlier Results
Recent studies suggest that not
way they should be managed,
R&D
all
project groups are alike
the way they communicate, or
in
the
in
in
the
kinds of tasks they pursue (Allen,
et.
projects can be categorized along
continuum ranging from research
development
When
little
a
al.,
1979;
1980).
In
particular,
to
and Tushman, 1981).
to technical service (Katz
projects are separated into the three types and the data are
reanalyzed using the same statistical smoothing process as before, some
patterns emerge that at least partially explain some of the curvilinear
findings from the earlier studies.
differences
the patterns shown
in
There are very
in
Figures
5,
6
real
and
and clear
7.
In
the case of
research projects, performance decays after about four years, only to
point closer to the me?n
return to
a
"older".
Similarly,
in
in
the case of teams that are
still
the cases of development and technical service
teams, there are mean tenure points at which performance drops to
fairly low level only to recover to an
tenured teams.
average or better
Whether these patterns
really indicate
level for longer-
something that
happening with "middle-aged" teams or whether they are just noise
impossible to determine from the present data.
the drops
in
performance occur
at different
8
a
is
is
Moreover, the fact that
mean tenure points for the
R&D
©
o
<
a.
o
LU
X
I—
o
o
UJ
-:>
O
li:
^^
different project types makes this even more difficult to determine.
different patterns,
however, may help to explain our inability to simply
reproduce previous findings.
None
of the earlier studies
teams whose mean tenure exceeded eight years.
whatever
causing the decay
is
These
teams of Figures 5 through
in
It
examined project
could well be that
performance among the "middle-aged"
caused the observed decay
7 also
in
the earlier
studies.
There are distinct drops and recoveries
Figures
5,
6
and
These occur
7.
each of the three types of
R&D
in
performance that can be seen
in
at different points for projects within
activity:
between four and five years for
research teams, at two and four years for product development teams, and
at
three and ten years for technical service teams.
pattern to some degree
in
in
One can even
our earlier data (Figure 2).
performance that occurs between four and
six years,
be some recovery by the two longest-tenured projects.
may be simply spurious or an
Nevertheless,
in
a
it
is
see this
Following the drop
there appears to
Now
all
of this
artifact of the smoothing technique.
worth speculating whether there could be certain points
team's development that are quite critical.
And
it
may be that
at
these points some set of relationships develop, both internal and external
to the team,
that ultimately have
a
strong effect on the project group's
overall performance.
The
fact that long-tenured project teams exhibit such a substantial
variance
in
performance, some apparently falling victim to the "NIH"
9
syndrome and becoming
others maintaining high
less effective while
performance, provides an opportunity to see what kinds of relationships
underlie the variance
project team
members
in
a
Fortunately, the survey asked
performance.
number
of questions
about themselves, their
and most importantly, about their
relationships with the organization,
relationships with both their project and functional management.
Comparing High and Low Performing Long Term Project Teams
There are
a
number
of interesting
ways
in
which the high performing
long-tenured project teams differ from their less successful counterparts.
This
is
particularly true of the
way
in
which the teams relate to both
their project and their functional management.
Each project team member was asked to evaluate on
a
seven-point scale
both the project manager and the relevant department head or functional
manager along
a
number
of leadership
management and technical support
dimensions related to the
of the project.
The questions were
independently asked for both project and functional managers, affording
the opportunity to test the degree to which the responses might be similar
or different for each of these managerial roles.
Within each team,
responses were aggregated separately to obtain overall team measures of
project and functional managers that could then be related to project team
performance.
10
Role of the Project Manager
of the data in
Table
is
I
Perhaps the most striking characteristic
.
the general lack of any strong relationship
between the team's perceptions of project manager's characteristics and
Only three of the 19
behaviors and the project team's performance.
leadership attributes that were measured relate significantly to
performance.
It
role of the project
the organization.
this
is
interesting to note that these three
is
manager
in
We have argued elsewhere (Katz and
the project manager's principal role.
little
project
s
relate to the
coupling the project team to the rest of
especially critical for long-tenured teams.
how
all
direct effect the other project
It
It
is
is
Allen,
1985) that
possible that this role
also interesting to see
manager attributes have on the
performance.
Role of Functional Managers
Once again, most
.
of the leadership
attributes do not correlate very strongly with performance (Table
two significant exceptions are noteworthy, however.
relate to the role of the functional
underlying technologies.
this
is
role as
This
is
manager
in
1985;
Allen,
1986;
The
Both statements
as that of connecting to
accord with our earlier description of
one that manages the technological input
(Katz & Allen,
II).
Allen & Hauptman,
to the organization
1987).
Long-tenured
teams perform better when functional managers provide technical
information; maintain currency regarding professional activities; are
technically and professionally involved with team
high performance standards.
In
members and maintain
short, they function to protect the long-
run technical capabilities and competencies of the project team.
11
But as
Tabic
Project
I
Manager AUiibuIes and Performance of Long- Icnurcd Project learns
Correlation with
Project
manager
project team
performance
disseminates important and relevant information concerning
slale-of-thc art technical advances.
manages meetings very
is
effective in keeping
keeps current and
is
erfeclively.
me informed
my
about
well-informed about ihc
overall performance.
lalcsl professional.
has the ability to recogni7e and mediate conflicts between group-: or individuals
is
an excellent sounding board for new
is
particularly effective al providing original ideas or fresh approaches.
encourages us to participate
in
ideas.
important decisions.
RftD
has important and useful contacts with other
professionals
outside this organi/.alion,
maintains high standards of performance
is
heavily involved
in the technical details
of
my
work.
has been very instrumental in iny professional development
and I have learned a great deal from him.
has considerable influence which is useful in obtaining the
various resources necessary to carry out my work, effectively.
provides excellent and constructive feedback on
materials and reports
has excellent conceptual understanding of
is
effective al providing appreciation
for
«ork
my
my
written
work.
and recognition
well done.
has important and useful contacts with other R.^'I) professionnl";
within this organization.
has a good understanding of ihe applied techniriues and methc\f|s
use in my work.
I
assigns
me
to jobs or tasks
on which
professionally to perforin well.
I
am
challrngcd
0.04
Tabic
Manager
riinclinnal
Altriliulc":
II
and Performance of
I
nnp-Teniired Project
Correlation with
team
performance
Functional manager
project
disseminates important and relevant information concerning
state-of-the-art technical advances
manages meetings very
is
m
effective
is
effectively
me informed
keeping
keeps current and
n..-)7
about
my
overall perforinance.
well-informed about the latest professional
activities.
has the ability to recogni7e and mediate conflicts between groups or
individiials-
is
an excellent sounding board for
IS
particularly effective
a'
new
ideas.
providing oiiginal ideas or fresh approaches.
encourages us to participate
in
important decisions.
RScD
has impoitani and useful contacts with other
Drgani7ation
professionals outside this
maintains high standard'; of performance,
IS
heavily involved
m
the technical details of
has been very instrumental
a
great deal
m my
my
work.
professional development and
have learned
I
from him
has ccinsiderable influence which is useful in obtaining the various resources
necessary to carry nut my work, effectively
provides excellent and constructive fccriback on
my
has excellent conceptual understanding of
IS
effective at providing appreciation
mv
written materials and reports
work.
and recognition
for
work
veil
done
has iinpoitan! and useful contacts with other RA'I) professionals within
this
organi7ation.
has a good understanding of the applied techniques and methods
I
use
work
assigns rne to jobs or tasks on which
perform
well.
I
am
challenged profcssionalh' to
in
my
long-tenured project teams perform better when the functional manager
provides the technological connection, they also need and perform better
when the project manager provides
a
strong organizational connection for
integrating and using these technical abilities.
Contrast with Newly Formed Groups
Given that the leadership roles of project and functional managers are
albeit in
critical,
very different ways,
teams, to what degree
other teams
in
performance.
to the
performance
our sample,
i.e.,
long-tenured
The
such teams?
this a characteristic only of
is
of
more newly-formed teams, also vary
in
Does the differentiation of project and functional roles also
explain some of the performance variations
in
these "younger" teams or
are other differences going to be more important?
To examine
this question,
long-tenured teams are compared with more
newly-formed teams (mean tenure of
degree
to
which they differ
in
less
than 18 months) to see the
terms of how the leadership factors
correlate with performance.
When
this
comparison
is
done, the pattern of overall results can best be
seen by simply contrasting the direction of the bars
shown
in
Figures 8 and
9.
In
in
the graphs as
each of these figures, the magnitudes and
directions of the performance correlations for each of the leadership
statements are represented by bars.
12
The correlations
for project
manager
LJ
<
t-
aQ,A
Li_l
Lu
z
<;x
tL
til
LiJ
m
Lt
UJ
o
i-'JLl.
X
u!
X
q:
Li_l
O
•<
1—
^
?;'
attributes for both the newly-formed and long tenured teams are shown on
the
left;
those for the functional managers are shown on the right.
each side,
On
"p-values" are displayed when the correlations between "young"
and "old" teams are significantly different.
What
is
immediately obvious
in
these comparisons
is
newly-formed and long-tenured teams are generally
on the project manager side but often
functional manager side.
is
in
not true for functional managers.
management needs
to be
in
the same direction
different directions on the
other words, project managers should treat
In
both "young" and "old" teams similarly
This
in
that the bars for
order to foster high performance.
The
role of functional
very different; either more or
less
pronounced,
depending upon how long group members have been working and
interacting together.
technology and
its
This
is
particularly true
development.
in
areas relating to
Functional managers need to be much
more concerned with providing technical information, professional
development opportunities, feedbacl< and challenge, and
to insist on
high
performance standards for project team members who have been away from
their specialty departments for extended time periods.
13
DISCUSSION AND CONCLUSIO NS
The
notion that both project and functional
team performance, but
discovery.
just
What
is
in
different ways,
most interesting
in
is
management
neither
affect project
new nor
a
the present results
how important effective functional management may be
is
a
to
surprising
understand
to the
performance of long-tenured teams.
While the project manager's roles and responsibilities for connecting the
project team to the rest of the organization are equally important
regardless of the length of time that the team has been working together,
the importance of functional management increases with team age.
and Allen (1982) have shown that teams tend to
their basis of technical knowledge as they age.
isolate
Pelz
themselves from
and Andrews (1976)
showed that teams prefer narrower, more specialized work,
is
Katz
as they age.
It
the role of the functional managers to prevent these two effects from
developing.
The functional managers must insure
that those of their staff,
who are on longer term project assignments, continue
to be well connected
to the relevant technologies either directly or indirectly
interactions with key technical gatekeepers (Allen,
through on-going
1977).
Long-tenured
team members must also be sufficiently stretched and stimulated to remain
broad and open-minded
in
their approaches to and uses of those
technologies
14
Too often, there
is
a
tendency to allow just the opposite
longer a department member
the less concern there
management.
This
managers devote
who are
a
is
is
is
assigned to
They
project or series of projects,
Functional
matrix organizations.
in
of attention to those
not on project assignment or
assignments.
The
happen.
for that individual on the part of functional
true even
lot
a
to
members
departments
of their
who are between
project
either want to get these people into direct charges
(projects) or they have them performing
charters of their departments.
work which
falls
(IR&D, for example).
As
within the
a
consequence,
those who have been given long term, continuing project responsibility
and, "taken care of," so to speak often receive less attention until
problem emerges.
They are
a
real
not on the functional manager's budget, they
are well acquainted with their assignments, they are well experienced, and
as a result,
seem to need
less direction
functional or even project management.
functional
in
It
their
is
work either from
no wonder then that
management often forgets about them and that they
consequently are allowed to
slip easily
into the "not invented here"
syndrome.
The
net conclusion to this study
is
a
very simple one.
As
a
team ages,
two different but essential leadership roles become increasingly important.
There must be strong,
influential
management
to link
and integrate the
team with the organization's goals, resources, expectations, etc.
project manager
who must make sure
It
is
that the team has the appropriate
attention of the organization and that the team and the organization are
15
the
in
Research has consistently shown that one of the most
synchronization.
important factors inhibiting the speed and development of new products
the lack of continuity
in
the focused commitment on the part of the
organization to the project team's efforts (Schon,
For
a
is
variety of reasons,
1967;
including reorganizations,
Maidique, 1980).
resource reallocations,
promotions and reassignments, etc., the continuity of organizational
attention
is
often compromised.
development efforts
in
For example,
his
in
study of long-term
the design and manufacturing of semiconductor
chips during the 60's and 70's, Vanderslice (1983) showed that there was
simply too much rotation of project managers and their organizational
counterparts to maintain
a
continuity of effort that would foster speedy
and effective performance.
manager, then,
is
to
One
of the
most important roles of the project
make sure that the team continues
appropriate attention of the organization and that
it
is
to
have the
able to interface
effectively within this political context.
But this managerial
role alone
is
not sufficient to overcome the strong
trends toward NIH that often emerge within long-tenured teams.
management
also needs to be effective functional
development
of
NIH.
to
There
prevent the
Functional management has the responsibility to keep
their staffs up-to-date and continuously challenged.
serious and credible effort for
it
is
a
the "technical culture" they try to
establish that will ultimately affect the obsolescence,
and communication networks
This must be
of the technical
1986).
16
work orientations,
workforce (Allen and Katz,
REFERENCES
Investigating the not invented here (NIH)
and T. Allen (1982)
A look at performance, tenure and communication patterns of
12
50 R&D project groups, RS-D Management
(1), 7-19.
Katz,
R.
syndrome:
,
D.
Pelz,
and
Andrews
F.
,
Scientists
(1976)
Organizations
in
Ann Arbor:
.
University of Michigan Press (revised edition).
Shepard, H.A.
(1956)
Creativity
R8-D teams.
in
Research and Engineering
,
10-13.
Smith, CO. (1970)
Relations, 23, 81-92.
Age
of
R&D groups:
Organization and Environment
Lawrence, P. and J. Lorsch (1967)
Harvard Business School.
,
Boston:
An investigation into the managerial
R. and M. Tushman
(1981)
and career paths of gatekeepers and project supervisors in a major
Katz,
roles
R&D
A reconsideration, Human
facility,
R&D Management
,
H,
3,
103-110.
The dual ladder: Motivational solution or
Allen, T.J. and R. Katz (1986)
managerial delusion? R&D Management 16 2, 185-197.
,
,
Project performance and the locus of
Katz, R. and T.J. Allen (1985)
influence in the R&D matrix. Academy of Management Journal 28, 1, 67.
87.
Tukey,
J.
(1977)
Exploratory Data Analysis
.
Reading, MA:
Addison-
Wesley
Allen, T.J., M.L. Tushman and D.M. Lee
(1979)
Technology transfer as
function of position in the spectrum from research through development to
technical services. Academy of Management Journal
22
(4), 694-708.
,
,
Allen, T.J., D.M. Lee and M.L. Tushman
(1980)
R&D performance as a
function of internal communication, project management and the nature of
the work, IEEE Transactions on Engineering Management 27
(1), 2-12.
,
,
T.J.
Managing the Flow of Technology, Cambridge:
(1986)
Press (revised edition).
Allen,
Janis,
I.L.
(1972)
Victims of Group Think
17
,
Boston:
Houghton
MIT
Mifflin.
a
Rogers, E.M. and F.I. Shoemaker (1971) The Communication of
Innovations:
A Cross-Cultural Approach, New York: Free Press.
The influence of communication
(1987)
Allen, T.J. and O. Hauptman
A conceptual model.
technologies on organizational structure:
Communication Research
,
14
.
(5).
18
h53h 09S
Date Due
Lib-26-67
U8WARJES
^ "lO&O
005
s 33
&50