 |
|
|
Efficacy of Partnering on the Woodrow Wilson Bridge Project: Empirical Evidence of
Collaborative Problem-Solving Benefits
Lee L. Anderson, Jr., P.E, M.ASCE, and Brian D. Polkinghorn, Ph.D.1
About
This article examines the large body of partnering rating data collected during eight-plus
years of partnered construction contracts on the Woodrow Wilson Bridge replacement project.
The authors report that partnering efficacy, as measured by “collaboration,” was not
predetermined by bid results or regional/national distinctions between construction
contractors. The data show that good partnering is strongly associated with team satisfaction
with budget and schedule results. The correlation among these indicators is not “proof” of the
effectiveness of partnering. However, had the analysis shown negative correlations where
positive correlations were expected, enthusiasm for construction partnering would certainly
have suffered a blow.
Subject Headings
Partnerships
Construction management
Contract management
Project delivery
Human factors
Communication
Conflicts
Dispute Resolution
Introduction
Construction partnering is not new
– it has been a popular strategy since
the 1980’s for minimizing and managing
conflict on construction projects.
Usually initiated by owners, partnering
requires an investment of both money
and staff time. Hence, especially
in times of limited resources, owners
want to know whether partnering “works.”
The quest for affirmation of partnering
efficacy is nearly as old as the practice
of partnering. One informative early
study (Lowe, 1994) used questionnaires
to investigate partnering use in Navy
construction offices. Lowe reported
that, “subjective results are all
but unanimously in favor of partnering,”
but that “the objective analysis is
inconclusive.” Owners and contractors
always have been anxious for some
quantitative demonstration that their
investment in partnering is producing
a positive return. However, partnering
outcomes are not easily measured.
Commonly, participants report on the
issue of how “satisfied” they are
with the partnering process, although
with what they are satisfied often
remains a mystery. More importantly,
the links between the process and
the eventual outcome – a more critical
topic in our estimation – are rarely
examined empirically. Perhaps the
participants “felt like a team,” but
did partnering improve schedule performance?
Did partnering save money?
The answers to questions like these
are hard to measure for at least two
reasons. First, only rarely will a
construction contractor tell the owner
what profit or loss the firm experienced
on a contract. That information is
business-sensitive. If told that the
contractor made a lot of money, the
owner might feel that he paid too
much for the work. If the contractor
lost money, the contractor would not
want his or her competitors to know.
So, information about dollar results
is not publicized.
The effect of partnering is also hard
to analyze because the outcomes of
an individual construction project
are so heavily affected by external
factors. Regardless of whether a job
was partnered or not, bad weather
might delay completion. Unexpected
material cost inflation may skew budget
results. Differing site conditions
might generate legitimate change orders.
There is no way to have a perfectly
controlled experiment – to work exactly
the same job with and without partnering
– because each job will encounter
a different set of external conditions,
invalidating the hoped-for “pure experiment.”
We are therefore left to less precise,
quasiexperimental models of inquiry.
In the nature of construction, all
contracts, partnered or otherwise,
will be subjected to forces and conditions
outside our personal control which
can lead to scheduling delays and
cost challenges. If not properly managed,
these forces and conditions can also
create a secondary source of conflict
among partners.
Owners with large construction programs
comprising many individual contracts
have had some success in measuring
partnering outcomes by looking at
aggregated data across their programs.
The Arizona Department of Transportation
found the welcome news that, after
instituting partnering across the
board, contract claims fell from 60
in 1991 to less than one per year.
After 2003, Arizona DOT experienced
no contract claims, and they abolished
their claims department (Young, 2009).
The Oregon Department of Transportation
performed a programwide analysis in
2002, showing many advantages of partnering
based on anecdotal reports (Rogge,
et.al, 2002); it was based on a survey
of 132 state employees and 42 contractors
who had previous experience with partnering.
The Maryland State Highway Administration
commissioned a study of its partnering
process and program; part of this
study took data from various isolated
internal databases and linked it to
a separate SHA partnering database.
The study found that an increase in
the use of partnering correlated with
a decrease in claims. (Polkinghorn,
et. al, 2006). While this study found
it difficult to control for the impact
of external forces (extraneous or
intervening variables) due to the
nature of the data available, the
results pointed to a genuine impact
of partnering on reducing post-construction
claims.
Results like this, when properly examined
using statistical analysis, serve
many useful purposes. They allow researchers
to examine whether or not certain
variables impact partnered construction
projects, whether extraneous variables
impacted the results, and to some
degree suggest what components of
partnering may have the greatest impact
on critical measures. Empirical research
also provides clarity and may help
steer future use of the partnering
process under specific project, site,
and local political conditions.
Many studies that focus on the experience
of individuals within the partnering
process, typically measured in terms
of process elements and associated
degrees of satisfaction, go no further
in exploring statistical relationships
pertaining to core variables impacting
the project. While these findings
are important for those involved in
managing collaborative decision-making
processes, they may not encompass
a holistic or systemic approach to
the study of partnering on the site
or with community stakeholders. Likewise,
many partnering studies that examine
the process rely on findings that
are limited to the internal workings
of the process and are supported by
descriptive testimonials and/or prescriptive
recommendations. The study at hand
largely escapes these limitations.
This article presents a quantitative
analysis of partnering outcomes resulting
from the Woodrow Wilson Bridge (WWB)
program. The WWB program replaced
the Interstate 95 bridge crossing
the Potomac River at the Maryland/Virginia
state line near outside Washington,
D.C., and reconstructed four major
interchanges, two on each side of
the river. Program costs totaled $2.5
billion (about $1.5 billion in construction
alone). The program comprised 19 separate,
major construction contracts awarded
by two owners, the Maryland State
Highway Administration and the Virginia
Department of Transportation. Most
of the construction was performed
during the years 2001-2009. One large
contract (VB-2/3/6) is still underway
at this writing, and a final smaller
contract is still to be awarded. With
these minor exceptions, for the purposes
of this article the program may be
considered complete.
The WWB program presents a unique
opportunity to study partnering outcomes.
All of the major contracts were partnered,
so admittedly there was no “control”
on the program’s application of partnering.
However, the program was characterized
by many factors that permit insightful
contract-to-contract comparisons to
be made. One factor is simply the
breadth and depth of an exceedingly
large quantity of data, described
in detail below, that was captured
over eight consecutive years. Another
is the constancy of leadership on
the owner side: one person in Maryland
and one person in Virginia were in
charge of their respective agency’s
efforts for the vast majority of the
period of observation. These individuals
set the tone for their subordinates,
so there was less variability in owner-side
priorities across the 19 WWB contracts
than would have existed, say, if the
contracts had been administered by
separate districts in a state highway
department. The individual contracts
ranged in dollar value at award from
about $3 million to $236 million.
The WWB program was a remarkably successful
endeavor. As a public infrastructure
megaproject completed on time, with
only three-percent budget growth over
eight years – resulting in a beautiful
bridge and major improvements to traffic
flow – it was heralded by the public
and lavishly recognized with industry
awards. Notably, one of the contracts
received a 2009 Marvin Black Award
for Partnering from the Associated
General Contractors of America. All
of the 19 contracts were successful,
although the participants would readily
agree that some contracts proceeded
better than others did. Some were
more or less profitable, more or less
timely, or proceeded with a greater
or lesser sense of teamwork and cooperation.
The collective outcome is remarkable,
given the scope of the program and
the countless issues and problems
that could have easily derailed the
entire project. The aggregate success
of the program could well be attributed
to the consistent application of insights
articulated in the standard conflict
resolution literature (Anderson &
Polkinghorn, 2008). None of these
insights resonates quite so strongly
as the Maryland WWB Project Director’s
mantra (lesson #10 in the cited work)
that, “We all succeed together.”
This internal evaluation of the vast
partnering rating data collected on
the WWB program shows – in quite satisfying
ways – that partnering “works.” For
example, the analysis here examines
the relative effectiveness of partnering
on the 19 contracts as a function
of bid results. [Contracting is, in
essence, a give-and-take process,
in which people bring things to the
table to exchange. Likewise, negotiators
create value and incentives, and they
claim value and rewards. The construction
industry is no different, except that
the end-results of the give-and-take
process are a tangible structure or
facility on the job site.] So, had
the analysis shown that when contractors
left a large amount of money “on the
table” at bid time, the team reported
lower partnering satisfaction than
in situations where the winning bids
were just slightly below the next
competitors’ bids, one might surmise
that the partnering effort had been
wasted. If only contractors with full
wallets make good partners, why bother
with partnering? Happily, this is
not the result, and there must be
a more meaningful explanation that
encompasses parts of the current findings.
The analysis here also examines whether
regional contractors made better “partners”
than national contractors, finding
only a modest correlation with geography.
The authors also look at how reported
effectiveness of issue resolution
correlated with the softer human-performance
factors involved in partnering, and
how self-reported results for schedule
and budget correlated with partnering
effectiveness. Showing objective schedule
and budget gains attributable to partnering
would be the “gold standard” – the
object of nearly every other previous
analysis of partnering outcomes. However,
because of the externalities explained
above, the self-reported schedule
and budget outcomes are probably as
valid as any. Success at issue resolution
is important as a postulated causal
link between better schedule/budget
performance and the human performance
factors.
With the large body of WWB program
partnering data available, the authors
are able to examine three other hypotheses
as well. First, this article explores
whether effectiveness at managing
community relations correlated with
good schedule performance (finding
a moderate association). Second, the
paper examines how the partnering
rating indicator for safety matched
actual safety experience – to some
extent, simply as a means to examine
whether the partnering raters were
“paying attention.” [The answer: maybe.]
Finally, the article delves into a
partnering commonplace: that having
construction company senior managers
attend monthly partnering follow-up
meetings is a secret to partnering
success. Interestingly, the data show
a negative correlation for this hypothesis.
Methodology
This study draws on the entire body
of partnering ratings submitted by
participants in each of the 19 WWB
contracts. The rating process was
described in detail in another article
(Anderson, et.al, 2006). Briefly,
the participants would visit the WWB
program website (www.wilsonbridge.com)
before each partnering follow-up meeting,
finding there a dedicated rating form
accessible by a username and password.
To rate each indicator, the participant
would select a score (from 1 to 4
or “I don’t know”) with a mouse click,
type in any comments he or she wished
to offer, and submit the form. A “4”
was consistently the best score, and
a “1” was consistently the lowest
score. The WWB program partnering
coordinator or assistant would receive
an e-mail generated by the internet
server for each rating and tally the
results on a spreadsheet for presentation.
Ratings were collected in the days
immediately preceding each partnering
follow-up meeting for discussion at
the current meeting. In essence, gathering
rating scores and comments served
as a means to flag current issues
as well as to provide an early warning
of potential problems.
This rating process was done for 354
follow-up meetings. The meetings were
attended by over 8800 people [counting
the same people many times, of course].
Average attendance was about 24 at
each meeting. Not every person provided
a rating for every meeting, and team
participants were welcome to rate
even though they could not attend
a particular meeting. The total body
of rating data represents an estimated
6,000 individual rating form submissions.
In comparison to most construction
projects, the number of meetings,
participants involved and subsequent
rating forms is incredibly large.
Also, the fact that these data were
collected over an eight-year period
provides a means of observing partnering
performance within the project over
an extended period of time.
The rating criteria and the narrative
meanings of each score were customized
for each of the 19 contracts. The
rating categories were chosen to support
analysis of achievement of the particular
objectives which each contract team
had agreed to include in its charter
at the partnering kick-off meeting.
This means there were a variety of
measurements taken on different contracts,
and that complicates comparison of
data across contracts. Nevertheless,
many of the rating categories and
score meanings were common among the
contracts. [For Maryland construction
contracts, certain rating elements
were prescribed to support the State
Highway Administration’s business
plan.] Thirteen categories were scored
by at least eight of the 19 contracts,
and six were scored by all 19. Table
1 refers.
For purposes of this study, the authors
create a collective indicator after
the fact to capture the softer, human-performance
side of the partnering effort. Called
“collaboration,” this indicator is
the sum of the average Teamwork, Communication,
and Cooperation scores for each contract
at each meeting. That is to say, the
“collaboration” score attempts to
smooth out anomalies in teamwork or
communication or cooperation scores
by presenting a single indicator as
a representative of how well the individual
contract team was succeeding in working
together. It also allows for comparison
of human factor problem-solving across
various contracts. This also then
allows the researchers to examine
contracts in regard to measures such
as schedule and budget. The authors
hypothesize that contracts presenting
high mean scores for “collaboration”
should present outcomes consistent
with a high level of human factor
cooperation.
The theoretical literature on collaborative
problem solving, as it impacts the
core characteristics of partnering,
can be summed up in one sentence.
“If the primary stakeholders can interact
in an open setting while focusing
on interest-based dialogue, then these
conditions substantially aid in the
search for and acquisition of mutually
beneficial outcomes.” The link between
principled negotiation and problem
solving in construction partnering
has been highlighted in previous research
on the WWB project (Anderson and Polkinghorn,
2008).
Statistical
Method
The observations and analyses presented
below result from performing a statistical
correlation test between data sets.
This was done using the CORREL function
in Microsoft Excel. Where an array
contains empty cells, those values
are ignored. A score of +1 indicates
a completely positive correlation
between the variables measured. A
score of 0 indicates no relationship,
and a score of -1 indicates a complete
negative correlation. Interpreting
the meaning of the numbers in between
could employ more sophisticated methods
than are appropriate for the purposes
of this paper. Jacob Cohen (1988)
offered a simple scale of small, medium,
and large correlations which the authors
adopt here. He characterized a “small”
correlation as the range -0.3 to -0.1
or +0.1 to +0.3. Cohen’s “medium”
correlation is the range - 0.5 to
-0.3 or +0.3 to +0.5, and his “large”
or “big” correlation is -1.0 to -0.5
or +0.5 to +1.0. The authors often
use the word “strong” below to mean
“large” or “big.”
In the case of the partnering data,
as will be seen below, the data sets
had few outliers. Hence, the correlation
coefficient is considered an appropriate
tool for gaining insight about these
partnering measures. However, as is
often said, “Correlation does not
imply causation.” Statistical correlation
is no guarantee that one set directly
causes the data in the corresponding
7 set. In the analysis below, the
opposite of this common wisdom is
more apt: if we believe partnering
“works” in some way, it would be revealing
(and disheartening to partnering boosters)
if the expected positive correlations
were negative or only weakly positive.
The 19 construction contracts included
in this study run the gamut in size
(as mentioned above), duration (from
about 18 months to over five years),
and complexity. In the charts below,
each contract is identified on the
X-axis label by a two-letter prefix
and a number or short series of numbers.
Contracts labeled “BR” were those
for the bridge superstructure; the
“VA” contracts were for the U.S. Route
1 interchange in Virginia; the “VB”
contracts were for the Telegraph Road
interchange in Virginia; the “MA”
contracts were for the I-295 interchange
in Maryland; the “MB” contracts were
for the Maryland Route 210 interchange;
and the “MM” contract was a larger
environmental mitigation contract
in Maryland.
Observations
and Analysis: Partnering versus Bid
Results
The first question to ask is whether
partnering outcomes are “fixed” on
bid day. It would be reasonable to
suspect that a construction contractor
who wins a contract with a bid that
is far below his nearest competitor
might enter the contractual relationship
with much regret and trepidation.
Those feelings could color the relationship
and make it well nigh impossible to
enjoy a fruitful partnership between
owner and contractor. Indeed, if his
budget were stretched too thin, one
might expect the contractor to be
looking for change orders at every
opportunity to recover his opportunity
to make a profit. In such a situation,
would partnering matter? Or, would
the contractor sink into the psychological
phenomenon called “buyer’s remorse,”
where one realizes he has entered
into an agreement that is quite unfavorable
to his interests?
The authors look at this question
in two ways. First, we plot the collaboration
score defined above for each contract
against the percentage by which the
winning bid fell below the next lowest
bid. Winning margins varied from a
miniscule 1.05% to a hefty 19.90%.
One contract, VB-4, had only one bidder.
A negative correlation would be expected,
with higher percentage left “on the
table” corresponding to a lower collaboration
score. Indeed, the correlation is
negative, but at a small -0.372, it
is far from strong. Figure 1 plots
the data.
Looking at the same question in another
way, the authors represent the bid
results as a percentage of the next
group of bids (up to three) above
the low (winning) bid. A higher percentage
meant that the winning bidder was
closer to the collective pricing wisdom
of up to three other competitors,
not just the next higher bidder. These
percentages varied from just 77 percent
to over 97 percent. The correlation
coefficient is again negative, as
expected, but scarcely different from
the correlation against the winning
margin at a small 0.382. Figure 2
shows the results. The one anomalous
data point (not figured in the correlation
coefficient) represents there having
been only one bidder for contract
VB-4.
Regional versus
National Firms
It would be reasonable to forecast
a priori that a regional construction
company would perform better as a
partner for a state highway agency
than would a national construction
company. A regional firm, one whose
primary business is conducted within
a few states localized within an easy
drive, might very well have worked
with a particular state highway agency
before. Personal relationships at
the top might have been cultivated
over many years of working together.
The regional firm would have every
incentive to please its regional clients,
especially as the selection for major
construction services moves away from
automatically choosing the lowest
bidder to choosing the firm which
can convincingly offer the best value
to the owner. One would expect the
regional firm to be a stellar partner
with the most to gain as well as the
most to lose.
Conversely, one might expect that
a large construction firm which is
bidding jobs coast-tocoast might never
have worked with a particular agency
before, would have no existing relationships
to nurture, and could reasonably not
expect to work for the same owner
again for many years. One could expect
the national firm to be a good “partner”
only when it suited his or her advantage.
Mitigating this to some extent, one
might assume that the lack of personal
contact would be substituted for by
national reputation. [Before moving
on, it is worth noting that the psychology
here is complex: when people have
a personal relationship with others,
as might be the case with the regional
firm, they have more at stake as they
are not part of an aggregate or large
and ambiguous entity but a person
who can be easily identified and held
accountable. Likewise, social psychological
research tends to suggest that when
individuals feel they are part of
a larger group and have a sense of
being anonymous, they misbehave more
so then when they are singled out
and identified.]
Twelve different firms or joint ventures
performed the 19 contracts on the
WWB program. Classifying these entities
as “regional” or “national” in their
outlook was sometimes a matter of
judgment, because a joint venture
often represented a combination of
a local firm with an out of-town
partner. Table 2 shows the classification
that the authors adopt; generally
a joint venture lead by a strong local
partner is classified as “regional”
even if they had brought aboard a
national partner.
The analysis is performed as follows.
The authors assign a value of “1”
for a regional contractor and “0”
for a national contractor. The resulting
array for all 19 contracts is then
correlated with the collaboration
score (just as defined above). The
computed correlation coefficient is
0.419 – a “medium” strength correlation
by the Cohen characterization above.
With all of the seemingly logical
reasons why a regional contractor
should be a better partner than a
national contractor, it is instructive
to see only a medium-strength correlation
coefficient come from the analysis.
And, while its collaboration score
ranked just number 9 of the 19 contracts,
the BR-3B contract – led by a national
firm (Granite Construction Company,
Inc.) – was the one selected to receive
the 2009 Marvin Black Award for Partnering
from the Associated General Contractors
of America.
From the data and the anecdotal evidence
cited, the authors infer that the
“national” or “regional” status of
the prime contractor does not predispose
construction partnering to be less
or more successful. While the authors
were able to construct a “collaboration”
variable (see page 5 above) to examine
the soft side of the data, the same
has not been accomplished in regard
to status of the firms of joint ventures.
Therefore, a number of alternative
reasons or contributing variables,
not directly measured here, are tantalizing
as a subject for future research.
One variable that might temper this
finding could be the status of the
project. WWB was arguably a high profile
project. Both regional and national
contractors could appreciate the visibility
and reputation that their efforts
on the WWB project would earn in the
public eye and with national decision-makers.
Another related avenue of exploration
could be the operational strategy
of a firm, or a joint venture understanding
among partners, in regards to partnering.
Likewise, the character of a firm’s
management staff would be a potential
source to explore, as some construction
firms, regardless of regional or national
character, are adept at partnering,
understand its principles and know
how collaboration can impact outcomes.
Conversely, other firms take a more
cynical or mechanistic view of partnering
and see it as “hoop to jump through”
or a “box to check” and not a way
of doing business. Even more troubling
are firms who are sometimes accused
by others as viewing partnering as
a means of extracting · American Bridge/Edward
Kraemer & Sons National · Corman Construction
Regional · CK Constructors (Corman
– led JV) Regional · R.R. Dawson Bridge
Company National · John Driggs Company
Regional · Glover Construction Regional
· Potomac Constructors, LLC National
· Shirley Contracting Company, LLC
Regional · Skanska USA Civil Southeast
National · Tidewater/Kiewit/Clark
Joint Venture National · Virginia
Approach Constructors (Granite – led
JV) National · G.A. & F.C. Wagman
Construction, Inc. Regional Table
2. National vs. Regional Contractors
concessions not stated in the contract.
Delving into variables of this nature
would be highly subjective and perhaps
impossible to examine convincingly.
Collaboration
versus Issue Resolution
One rating criterion reflected on
all 19 of the WWB construction partnering
rating forms was issue resolution.
Any experienced construction practitioner
would regard the ability to have issues
resolved in a timely, interest-based
manner as a key element of a project’s
success. Sometimes a construction
contractor will encounter a conflict
on the blueprints: one drawing shows
something one way and another shows
it the opposite. Or, the blueprints
and the specifications may differ.
Or, an unexpected obstacle is discovered
when soil excavation begins. The list
of possible issues arising on a construction
project is seemingly endless. Ideally,
many issues are resolved right on
the spot, in consultation between
the superintendent and chief inspector,
or between the contractor’s project
manager and the owner’s resident engineer.
Issues requiring technical analysis
may be returned to the engineer-of-record
as a “Request for Information” or
“RFI” to get an engineering answer.
In any case, timely resolution is
important. Workers may be unable to
divert to another task, or critical
materials and equipment may sit unused
or idle, until the issue is resolved.
As might be expected, the WWB partnering
analysis shows a very strong positive
correlation between the collaboration
indicator and the issue resolution
score: 0.947. Figure 3 charts the
data.
This outcome confirms what one would
expect: a strong association between
good partnering (as represented by
the composite collaboration score)
and the all-important issue resolution
performance. This virtual truism works
whichever way it is postulated: because
collaboration is better, issues are
resolved more effectively. Or in reverse,
because issues are resolved more effectively,
the collaboration is regarded as better.
They mirror one another. The most
plausible reason is the ability to
establish the conditions that reinforce
effective communications: collaborative
problem-solving that supports the
notion of teamwork. Underlying these
composite scores are the notions of
honesty, trust, and the parties being
held accountable to one another.
Collaboration
versus Budget
As explained above, it may be asking
too much to demonstrate that partnering
has a salutary effect on whether the
contractor makes a profit or not,
and whether the final price tag stays
within the owner’s budget. Many external
factors can affect the budget outcome
which are beyond the ability of an
effective partnering program to influence.
On the WWB program, eight of the 19
contracts tracked a “budget” indicator.
A typical “budget” criterion rating
scheme (as used for this example on
the $236-million contract VB-2/3/6)
read as follows:
“Owner and Contractor are both meeting
budget expectations:
Score 1: Overspent-cannot recover
Score 2: Overspending-must compensate
Score 3: Near target
Score 4: Doing better than expected”
The adjectival descriptions assigned
to each numerical rating score were
thus aimed at a “3” being the accepted
score if the project budget was on
track for both owner and contractor.
Generally, neither side had good visibility
over the other’s budget situation,
so owner raters would score how the
owner was doing, and contractor raters
would base their ratings on contractor
financial projections. Generally,
stakeholders not familiar with the
budget situation of either owner or
contractor would simply mark “I don’t
know.”
Even with partners not knowing all
the budgetary details, Figure 4 shows
a strong positive correlation of 0.842
between the collaboration indicator
described above and the average budget
scores assigned on the contracts that
chose to track this criterion.
The authors, while not surprised to
see this result here and above on
issue resolution, want to maintain
a critical view alá Popper’s “falsification”
process of working to prove your own
hypothesis wrong. In other words,
Popper says a good researcher doesn’t
simply accept the conclusion, especially
if it is positive, but seeks to refute
it in order to find a more encompassing
explanation of the results. In this
case, this strong positive correlation
does not necessarily “prove” – but
maybe “tends to suggest” – that well-partnered
contracts produced better budget outcomes.
However, our critical view leads us
to acknowledge the possibility that
the reverse phenomenon may have been
operative instead: on contracts where
the budget was being met, perhaps
the partnering team members found
themselves under less stress and hence
behaved as better team members, communicated
better, and cooperated more with their
counterparts. Still, had a weak or
negative correlation been manifest
from the WWB data, that result would
have been quite disheartening for
partnering advocates.
As an aside, notice the relatively
lower average “budget” score for the
award-winning BR-3B contract (it was
a 2.93). It is instructive to know
from one author’s personal reminiscence
that the parties recognized the budget
difficulties on that contract, but
in their discussions they acknowledged
that the collaboration which was occurring
kept the budget difficulties from
becoming considerably worse. The teachable
idea from this example is rather simple:
if partners know there will likely
be problems ahead, they should continue
to communicate and solve problems
as a team as clearly and often as
possible. That is, they should do
so if there is trust. Without trust,
one could be wiser to adopt the opposite
tactic: namely, to withdraw and run
for cover while placing blame on others.
In fact, among all these findings
so far, one has to consider the possibility
of a self-fulfilling prophesy coming
into play. If we are predisposed to
work well with others (high collaboration),
then we are predisposed to give others
the benefit of the doubt when something
goes wrong and to chalk up any problems
to external conditions rather than
to personal characteristics. Likewise,
if the working relationship is not
good, the opposite often occurs. We
do not give the partner the benefit
of the doubt; we chalk up the problem
to an internal flaw in their character,
or we attribute their actions to some
evil intent.
Collaboration
versus Schedule
All 19 of the WWB contracts had a
partnering rating indicator to track
how well each contract was doing at
conforming to its established schedule.
Typically, the adjectival descriptions
for each score of 1 to 4 were highly
customized. On contracts with incentive
milestones (where the contractor could
earn a bonus payment for meeting certain
target dates), the “schedule” rating
scheme focused on the likelihood of
his meeting the incentive dates. Other
contracts did not have incentive milestones,
so the adjectival scoring was somewhat
different. It is also relevant to
observe, from one author’s reminiscence,
that some raters gave greater weight
to the process of submitting and maintaining
the required critical path method
schedule under this criterion, while
other raters focused exclusively on
the results.
Notwithstanding these caveats, it
is at least somewhat informative to
see a strong positive correlation
between the average collaboration
scores and the average “schedule”
scores. Figure 5 charts the result,
displaying a correlation coefficient
of 0.682.
Many of the same things said above
about “budget” could also be said
about “schedule.” The substantial
positive correlation does not “prove”
that well-partnered contracts produce
better schedule outcomes. On contracts
where the schedule was being met,
perhaps the partnering team members
found themselves under less stress
and hence behaved as better team members,
communicated better, and cooperated
more with their counterparts. However,
partnering advocates can take heart
that a weak or negative schedule correlation
is not manifest from the WWB data.
Community
Relationships versus Schedule
In addition to addressing the basic
questions about the efficacy of partnering
discussed above, the voluminous body
of WWB partnering data allows the
authors to explore some pertinent
sidebar issues. One interesting issue
is how the relationship between the
partnering team and the community
at large – “outside the fence” – may
have been associated or not associated
with schedule performance. It would
be reasonable to surmise that an antagonistic
relationship between the project team
and the community would lead to schedule
delays. Opponents might have found
ways to raise continuing objections
to the project which would have impeded
its progress. Conversely, delays in
project completion might be associated
with a decline in public goodwill
towards even a project which the public
welcomed.
There are countless methods by which
the public can impact a construction
project. These range from attending
and taking over public meetings, engaging
state and federal agencies in an attempt
to enforce particular regulations
or laws that meet their agenda; demanding
meetings with project managers; writing
letters to the editor, giving interviews
on the radio and public access television,
organizing petition drives, calling
elected officials and demanding that
they investigate constituent complaints,
or filing lawsuits and requesting
injunctions – to name just a few.
There are also countless reasons why
members of the public might object
to construction projects. Some might
be: environmental justice concerns
over disproportionate impact to minority
communities, destruction of historical
sites, protection of sensitive aquatic
or forest ecosystems, impact to threatened
or endangered species, aesthetic disputes,
and increases in “noxiants” such as
noise and air pollution. Members of
the public can also engage in disputes
on construction projects because of
more abstract political and ideological
agendas. Disputes of this type, while
they may be unfounded, also take time
and resources to address. With NIMBY
(“Not in My Back Yard) groups ranging
from all socio-economic, ethnic, and
racial groups to ideologically driven
deep ecologist/environmentalists such
as BANANAS (“Build Absolutely Nothing
at All Near Anything”), the stakes
climb even higher. Together the methods
by which the public can engage in
disputing along with the untold issues
and/or (ideological) reasons
to do so make for a dynamic and sometimes
unstable mixture that can easily disrupt
or destabilize a construction project.
The results of such disputing can
show up in scheduling delays, cost
overruns, radical reworking of the
project itself or, in some cases,
the demise of a construction project
altogether.
Realizing the immense nature of the
Woodrow Wilson Bridge Project, it
was important for the program to deal
effectively with community relations.
Nine of the WWB contracts included
a specific community relations or
public relations indicator within
their partnering rating process. Just
as there were differences in the adjectival
definitions for “schedule,” the adjectival
definitions for each numerical score
under community relations had differences.
Furthermore, different raters easily
may have had different interpretations
about the substance of this particular
criterion. Hence, any conclusion from
the correlation between community
relations and schedule ought to
be treated with particular skepticism.
Nevertheless, Figure 6 charts the
average scores for community relations
and schedule, showing a just-barely-large
positive correlation of 0.524.
The correlation of community relations
with schedule suggests that some of
the mechanisms discussed above were
operational on the WWB project. It
is the authors’ mutual sense that
the fact that the WWB program generally
stayed on schedule had a salutary
effect on community relations. This
was a more important effect, based
on the authors’ anecdotal recollections,
than situations where community issues
might have posed obstacles to schedule
progress. However, so many other factors
influenced schedule that it seems
unlikely that the positive correlation
shown here can demonstrate that community
relations were a major driver in schedule
outcomes.
Collaboration
versus Company Officer Attendance
The conventional wisdom among partnering
practitioners is that it is very important
to have construction company senior
managers attend partnering follow-up
meetings, along with customer agency
senior managers. That way, so the
logic goes, the partnering meetings
can be genuine problem-solving sessions,
and subordinates will be motivated
to behave in the collaborative ways
modeled by their bosses. In addition,
having managers with decision-making
authority take part in face-to-face,
interest-based partnering discussions
should allow for accurate and timely
resolution of issues. Interestingly,
the pesky WWB partnering data suggest
that the opposite is true: less frequent
attendance by construction company
senior managers is correlated with
higher collaboration scores.
This analysis is conducted as follows.
First, the authors identify by name
the individuals who were vice presidents
or higher in their construction companies
and who also attended partnering followup
meetings as documented in the meeting
minutes. Some contracts had only one
vice president (or higher) ever attend
a follow-up meeting, whereas some
of the large joint-ventured contracts
had up to eight. Referring to these
populations of executives, the follow-up
meetings are coded according to whether
any corporate officer had attended.
On four contracts, at least one company
officer attended 100 percent of the
time; on one contract (MM-6), no company
officer ever attended a follow-up
meeting. On average, 66 percent of
the follow-up meetings had one or
more company officers present. Then,
the percentage of meetings where a
company officer attended is compared
against the collaboration score for
the contract. Figure 7 shows the respective
data series, which have a negative
correlation of 0.318.
This negative correlation was surprising
to the authors but not necessarily
to some practitioners with whom it
has been shared. Some explain the
negative correlation by observing,
“Of course that’s the case: the executives
only showed up for a meeting when
the job was in trouble. They were
managing by exception. If the job
was going well—and being partnered
well—there was no need for them to
show up.” If the folks lower in the
organization are doing well, then
it is in the best interest of the
executives to empower them to work
on issues and simply to keep the higher-ups
informed. This also makes good business
sense, as the adage “time is money”
is true, and in this case, “one head
is better than two.”
Inspection of these data suggests
further that including the MM-6 contract
might have skewed the statistical
analysis. Interestingly, that contract
was small in dollar value (about $3
million) and had only four follow-up
meetings. A company officer attended
the initial kick-off meeting but did
not come to any of the four follow-up
meetings. This contract proceeded
very smoothly, and the partnership
between the owner and contractor was
a great example of “doing it right.”
If the MM-6 data are excluded from
the analysis, the correlation between
the respective arrays remains negative
but falls to a weaker 0.126.
The negative correlation between collaboration
and officer attendance data suggests
that it is unnecessary for partnering
advocates continually to be urging
senior managers (on the contractor
side, at least) to attend partnering
follow-up meetings. Perhaps these
managers can just as well decide for
themselves when it would be helpful
for them to attend. The iconoclast
might phrase this implication even
more strongly: senior managers should
be discouraged from attending
partnering follow-ups. Maybe they
just get in the way! The advice should
be, “Leave the problem-solving to
the working-level team members, and
just watch from afar.”
Safety Results
versus Partnering Safety Scores
On the WWB program, all 19 of the
partnered contracts kept track of
a rating indicator regarding safety.
This rating indicator was a companion
to the actual accident experience
and exposure ratings, which were also
recorded concurrently for every contract
on the program. Maintaining a separate
partnering rating indicator for safety
had at least two advantages: it ensured
that safety would be discussed at
each partnering follow-up meeting,
and it gave the raters an opportunity
to add insights about safety which
the raw safety statistics might not
show. Paramount among the latter was
the opportunity to add a comment about
the safety program in general or a
specific incident in particular.
The adjectival descriptions for the
“safety” criterion varied among the
19 contracts. Some partnering teams
chose to keep the meanings of the
scores from 1 through 4 highly quantitative:
they were interested just in measurable
results. Other teams chose to describe
the meanings of the scores in ways
which spoke to the broader effectiveness
of the safety program and the zeal
with which it was pursued. These teams
allowed for the fact that an element
of randomness enters into the accident
experience on a short duration contract,
and they were most interested in seeing
that a strong safety program was in
place to make the workplace a safer
one. Naturally, this variability in
definitions makes an overall analysis
of the partnering safety data quite
equivocal. Nevertheless, it proves
interesting to compute the correlation.
The authors compare the array of partnering
safety scores with the final OSHA
incidence rate (or “case rate”) on
each contract. The OSHA incidence
rate, simply explained, is the number
of workplace accidents (of more consequence
than a first-aid case) that a 100-person
workforce experienced in a year. [The
200,000-hour divisor in the incidence
rate formula is 40 hours x 50 weeks
x 100 workers.] Figure 8 shows the
comparison between the average partnering
safety scores and contract-long OSHA
incidence rates for all 19 contracts.
The computed correlation coefficient
between these series is negative 0.50.
Inspection of the data shows one clear
outlier: the VB-4 contract had an
OSHA case rate of 15.8. VB-4 was a
short-duration contract that, unfortunately,
experienced more than its share of
accidents. The partnering raters recognized
the somewhat random nature of this,
and scored safety a mid-range 3.51
in the partnering surveys. Without
the VB-4 data, the correlation between
the two arrays would be a much stronger
negative 0.712.
What can be learned from these data?
The direction of the correlation is
negative, as should be expected. Higher
numerical accident experience records
are associated with lower partnering
safety scores (where the higher score
is always the better score). To that
extent, at least, the partnering rater
population was answering the surveys
in ways consistent with the accident
experience which was occurring in
the field. Much more than that cannot
be learned, however, since there was
an element of evaluating safety program
strength, rather than accident experience,
in some of the individual contract
rating schemes.
Data Discussion
Planners of the WWB project not only
broke a complex project down into
incredibly detailed and carefully
orchestrated steps, they also set
up mechanisms to make sure unforeseen
and unintended consequences were identified
and managed as early and effectively
as possible. Part of that overarching
planning process included the foresight
to develop and use a sophisticated
data collection and tracking protocol
on the partnering process itself.
The body of data thus collected now
allows researchers an empirical means
by which to examine process and outcome
relationships. The primary benefit
of partnering is to prevent and or
effectively manage issues. If partnering
is such a valuable function as proponents
state, then they shouldn’t rely solely
on descriptive accounts or prescriptive
recommendations to measure whether
it’s working. The secondary benefit
of diligently tracking the WWB project
over an eight-year period is that
we now have the rare [and lucky!]
opportunity to examine a collaborative
decision-making process empirically
on a much larger scale than is normally
seen in construction or any other
industry.
This may indeed be one of the few
cases where so many data points have
been collected that making basic findings
can be both quite easy and, ironically,
a bit difficult. There is so much
richness and quantity in the data
set that data patterns are readily
apparent and rarely subject to statistical
anomalies. In smaller or “normal”
size cases, i.e., those comprising
fewer partnering meetings (measurement
points), fewer participants (sample
size) or fewer objects of measurement,
the usual worry is that one or two
outliers can easily skew not only
the mean descriptive data but also
the inferential statistics. Small
size is obviously not an issue with
the voluminous aggregated data on
the WWB project, and in the rare instances
when outliers do arise, there are
clear reasons why. Conversely, a huge
data set can be difficult to approach
because of the urge to dig deeper
into this rich source of data to identify
more subtle findings. In other words,
the data set may well hold much more
knowledge than is presented in this
article.
Overall, what do these general findings
mean? Before launching into that discussion,
recall that the partnering process,
as used on the WWB project, is both
an early warning conflict prevention
method as well as an in situ conflict-management
process. The distinction between the
two is a matter of timing. Early identification
of potential problems can, when the
time arises, allow them to be prevented
or managed more effectively. Based
on the results of the data analysis
for collaboration presented above,
we believe that a number of antecedent
conditions and variables were likely
developed early on in order for the
conflict prevention and management
functions to operate as well as they
did.
The first condition or variable to
consider is the context within which
the project occurred. While context
is not directly measured in the WWB
partnering data, except indirectly
under “Community Relationship” on
some contracts, there is a strong
possibility of external forces or
actors impacting the process (for
more information on external control,
see Pfeffer & Salincik, 1978). The
measurement of an external environment
is a tough undertaking, even under
ideal conditions and, as a result,
most of the literature in this area
is of a theoretical nature (cf. Emery,
F. and Trist, E. 1965; also Baburoglu,
O., 1988). Depending on the level
and type of dynamics discussed, the
literature describes a range of external
environments going from placid (where
no external forces or actors cause
problems), to clustered (where allies
clump together to meet a common foe
head on) to disturbed (where the impact
of the environment is less controlled
by the parties, to turbulent (where
the environment is so chaotic that
most responses by combatants are maladaptive
and only lead to more problems). Given
the potential sources of potential
conflict in scheduling, budget, dealing
with the public and other issues,
the Woodrow Wilson Bridge project
arguably had the key ingredients of
being undertaken in either a disturbed
or a turbulent environment. With old
WWB bridge presenting such a locally
loathed bottleneck on the Capitol
Beltway and the Interstate 95 corridor,
the chances were high that that the
project would generate many negative
“above the fold” Washington Post headlines.
But that didn’t happen, which begs
the question why?
This was no ordinary project. Two
states, the District of Columbia and
the Federal government were involved.
This was the only Interstate highway
bridge owned by the Federal government,
and a flagship national bridge. Additionally,
among the 200,000 daily travelers
who use the bridge or live in its
shadow are members of the U.S. House
of Representatives and the U.S. Senate.
Given the ambitious nature of the
project, its high profile status and
constant oversight by many governments
and the media (print and broadcast),
this was not the project on which
to be a partner and have things go
wrong. Another factor was the healthy
relationship with the general-interest
media that the WWB project leadership
consciously cultivated. As a result
of this proactive effort – itself
a type of partnership – the media
took a keen interest in the project
– not so much looking for problems,
but to highlight unusual aspects such
as the unique design, human-interest
stories, and other things that played
well on the History and Discovery
Channels’ documentaries on the WWB
project.
In this case, while there are no data
to support this hypothesis, the high
profile, high publicity nature of
the project may have been an external
force or intervening variable that
helps explain, perhaps in some small
way, why so many indicators came out
so well. With so much attention being
lavished on the project, it could
be that, in one sense, this project
was the largest example in modern
history of the Hawthorne Effect (Landsberger,
1958)? [See Appendix.] Could this
be a contextual layer not found in
the partnering data that impacts the
success of the project? Could this
help explain how, given the countless
sources of potential show-stoppers,
the WWB project came out so well?
We cannot empirically say one way
or another, but we can offer this
as a plausible counter explanation
for the good results, alá Popper’s
falsification process.
The authors have mentioned occasionally
throughout this article that there
is room to consider the opposite time-ordered
relationship between various partnering
measures. However, one thing that
we think is truly an antecedent condition
or variable is the development of
trust within the context of this project.
While it cannot be analyzed quantitatively,
trust may help explain the findings
above, both the expected and the unexpected.
Trust does not spring fully formed
from a partnering workshop but more
likely from consistent, transparent
communication on the job, collaborative
problem-solving in practice, a sense
of interdependence among all partners,
and perhaps even having the unstated
feeling of being part of history.
Trust, as least identity-based trust,
is earned, and it is not a precondition
of participation (Lewicki and Wiethoff,
2000). Without trust, people interact
in a defensive mode. Communication
by memorandum replaces informal discussion,
and agreements are calculated more
on exposure to risk than on wise and
efficient means to ends.
There are some important caveats to
consider in weighing the observations
from the WWB data set. One has to
do with groupthink and/or self-fulfilling
prophesy. It would not be surprising
to find that, by meeting face-to-face
nearly every month, partnering teams
began to take on certain normative
behaviors and started echoing each
other’s sentiments. It could be that
some over-reporting of positive results
occurred (a classic “trip to Abilene”
– see Harvey, 1974). With only one
exception (contract VA-6/7), the rating
results were presented anonymously,
which may have served to reduce expectations
to “go along” with the group. When
individuals feel compelled to go along
with the others, groupthink may set
in. Conversely, when things are really
going well a snowball effect tends
to arise and people come to expect
good things to happen – a nice self-fulfilling
prophesy. Likewise, reaching deeper
into speculative territory, perhaps
some of the high scores could also
represent a Hawthorne Effect, or they
could stem from a collective sense
of being a part of history.
Another caveat is that some raters
assuredly put more thought and effort
into assigning ratings than others
did. In performing an analysis of
this type, the authors are effectively
assuming that the number of raters
who simply plugged in numbers so their
bosses would not hector them for not
responding is distributed evenly across
the 19 contracts.
The WWB project is unique; it set
or broke engineering records, produced
an elegant structure, and un-corked
a traffic bottleneck that had bedeviled
the region for decades. Considering
the findings above, one might construct
an unscientific formula for success:
if, in a complex project, the right
people work effectively together,
at the right time and in the right
place, then they have optimized the
chances of reaching the right solutions/outcomes.
While this is highly simplistic, it
may well be to add that in cases like
the WWB the right context or special
circumstances, i.e., being part of
a major or vital project or perhaps
being a part of history, provides
an added emphasis that instills a
sense of pride and source of collaboration
that provides even greater chances
of success.
Summary and
Conclusions
This article examined the large body
of partnering rating data collected
during eight-plus years of partnered
construction contracts on the Woodrow
Wilson Bridge program. The authors
found that partnering efficacy, as
measured by “collaboration,” was not
predetermined by bid results or regional/national
distinctions between construction
contractors. The data showed that
good partnering is strongly associated
with team satisfaction with budget
and schedule results. The authors
assert that the self-reported scores
for budget and schedule are an appropriate
surrogate for “real” budget and schedule
performance, since the latter are
affected by so many external forces
that any attempt to analyze them for
partnering efficacy is unlikely to
be any more persuasive.
The data showed a strong association
between good partnering (“collaboration”)
and effective issue resolution. The
authors believe that this is consistent
with the hypothesis that effective
issue resolution results in positive
budget and schedule outcomes. The
common association among these indicators
is not “proof” of the effectiveness
of partnering. However, had the analysis
showed negative correlation where
these positive correlations were expected,
enthusiasm for construction partnering
would certainly have suffered a blow.
Acknowledgements
The authors wish to acknowledge the
Maryland State Highway Administration
and the Virginia Department of Transportation
for access to the WWB partnering data
and permission to publish this study.
Special thanks go to Mr. Terry Garrett,
the WWB partnering coordinator from
late 2008 and hence, for his assistance
in updating the partnering rating
data through June 2009.
References
- Anderson, L., Douglass, R.,
and Kaub, B. (2006). Anatomy of
a Successful Partnering Program
on a Megaproject.” Leadership
and Management in Engineering,
6(3), 110-116.
- Anderson, L., and Polkinghorn,
B. (2008). “Managing Conflict
in Construction Megaprojects:
Leadership and Third-Party Principles.”
Conflict Resolution Quarterly,
26(2), 167-198.
- Baburoglu, Oguz, N. (1988).
“The Vortical Environment: The
Fifth in the Emery-Trist Levels
of Organizational Environments.”
Human Relations, 41(3),
181-210.
- Cohen, J. (1988). Statistical
power analysis for the behavioral
sciences (2nd edition). Erlbaum,
Hillsdale, N.J.
- Emery, F.E. and Trist, E.L.
(1965). “The Causal Texture of
Organizational Environments.”
Human Relations, 18(1),
21-32.
- Harvey, J. B. (1974). “The Abilene
Paradox and other Meditations
on Management.” Organizational
Dynamics, 3(1), 63 and ff.
- Landsberger, H. A. (1958), Hawthorne
Revisited, Cornell University
Press, Ithaca, N.Y. 1958.
- Lewicki, R., and Wiethoff, C.
(2000). “Trust, Trust Development,
and Trust Repair.” The Handbook
of Conflict Resolution: Theory
and Practice, M. Deutsch and
P. T. Coleman, eds., Jossey-Bass,
San Francisco, Ca., Ch. 4.
- Lowe, Scott W. (1994). “An Examination
of the Effectiveness of Partnering
in Navy Construction Contracts.”
MS thesis, University of Washington,
Seattle, Wa.
- Pfeffer, Jeffrey and Gerald
Salincik (1978) External Control
Theory: A Resource Dependence
Perspective. Harper and Row,
New York.
- Polkinghorn, Brian, Robert La
Chance and Haleigh La Chance.
(2006), “An Analysis of the Maryland
Department of Transportation State
Highway Administration’s Partnering
Program and Process.” Maryland
Department of Transportation,
State Highway Administration.
(Internal report to MDSHA)
- Rogge, D., Griffith, A., and
Hutchins, W. (2002). “Improving
the Effectiveness of Partnering,”
State Planning and Research Report
No. 344, Oregon Department of
Transportation and Federal Highway
Administration, November 2002.
- Young, James (2009). Personal
communication providing link to
on-line resource, “Introduction
to Partnering, ADOT Partnering
Office,”
http://www.azdot.gov/ccpartnerships/partnering/education/intro_to_partnering/index.htm
Appendix
In a review of old studies on worker
productivity from the 1920s and 1930s
at the Hawthorne Works, Henry Landsberger
coined the term “Hawthorne Effect”
to describe the phenomenon whereby
workers who received attention, in
this case the reactivity of researchers
in their presence, demonstrated higher
productivity. In other words, when
people are the subject of others’
supportive observations and attention
they tend to react to this condition
by being more productive. Add this
to the positive reports that resulted,
in terms of news reports and visits
by VIPs to the WWB construction site,
and conditions were ripe for the Hawthorne
Effect to arise on this project.
1 Larry Anderson, P.E.,
is a Senior Program Manager with MBP,
Inc., and coordinated the partnering
on the Woodrow Wilson Bridge project
from 2001 through 2008. E-mail: LAnderson@mbpce.com;
Mail: 10440 Little Patuxent Parkway,
Suite 250, Columbia, MD 21044
Professor Brian Polkinghorn, PhD,
is Executive Director of Salisbury
University’s Center for Conflict Resolution.
E-mail: BDPolkinghorn@salisbury.edu;
Mail: 1101 Camden Avenue, Salisbury,
MD 21801
|
|
|
