INTERNATIONAL JOURNAL OF

PROJECT
MANAGEMENT
International Journal of Project Management 22 (2004) 523–532
www.elsevier.com/locate/ijproman

An exploration of project management office features and

their relationship to project performance
a,* b,1
Christine Xiaoyi Dai , William G. Wells
a
Management Science Department, School of Business and Public Management, The George Washington University,
2115 G Street, NW, Washington, DC 20052, USA
b
8900 Carmel Valley Road, Carmel, CA 93923, USA
Received 17 October 2003; received in revised form 2 March 2004; accepted 2 April 2004

Abstract

The advantages of project management have been well documented, but project failure rates still remain high. This suggests
continued exploration of new process models and organization structures to nurture strong project performance. One important
candidate for improvement in this ongoing journey is the project management office (PMO). This paper is based on a two-year
empirical study that investigated the establishment and use of PMOs and the environmental conditions in which they operated. It
also identified and assessed an array of PMO functions and services and their influence on reported project performance. The core
results were generally favorable toward the utilization of such features, with project standards and methods showing the highest
correlation to performance in each of the two distinct populations.

2004 Elsevier Ltd and IPMA. All rights reserved.

Keywords: Project office; Systems; Success and strategy; Processes; Procedures

1. Introduction ous successes and failures due to a lack of effective

Projects have become important instruments for
change and development in organizations [1]. Gareis [2]
and Lundin [3] suggest that the broader utilization
of projects requires a new orientation in project man-
agement (PM) and a new model for more effective op-
erations in project-driven organizations. Munns and
Bjeirmi [4] showed that more effective PM offers great
potential for improving overall organizational perfor-
mance by enhancing the prospects for project perfor-
mance and minimizing the likelihood of failure. In spite
of the advantages of using the project approach, how-
ever, Jessen [5] suggests that there is also a significant
problem. Because of the one-time nature of projects, an
organization may often derive little benefit from previ-
* tion to personnel recruitment/selection and training. A
Corresponding author. Tel.: (202)-994-7375.
E-mail addresses: cdai@gwu.edu (C.X. Dai), wwells1923@aol.com
(W.G. Wells).
1
Tel.: +831-624-2869; fax: +831-622-0852.
knowledge transfer. The study reported in this paper
examined the question of what measures organizations
have taken to enhance transferability of lessons learned
from previous projects, and how these measures have
influenced project outcomes.
There are many dimensions for evaluating project
performance [6,7], while it appears to be easier to de-
velop consensus on determining project failure. The
documentary record is replete with reports of high rates
of project failures across all industries, government
agencies and national boundaries [8–10].
One approach to studying project performance has
been through the investigation of critical success factors
(CSFs) as predictors of performance. For example,
Pintos [11] identified 10 CSFs, ranging from project
mission, top management support, project schedule/
plan, client consultation, technical tasks, communica-
natural next-step would be to determine how organiza-
tions could systematically foster CSFs on an ongoing
basis.

0263-7863/$30.00
2004 Elsevier Ltd and IPMA. All rights reserved.

doi:10.1016/j.ijproman.2004.04.001
524 C.X. Dai, W.G. Wells / International Journal of Project Management 22 (2004) 523–532
Another conceptual guideline was provided by Might
and Fischer [12], who examined how project organiza-
tional structure interrelated with outcomes. They ob-
served structures ranging from functional design at one
extreme to a dedicated project team at the other, with a
matrix format somewhere in between. Their results
suggest that while neither extreme showed any notable
degree of association with performance, the intermedi-
ate forms did possess some positive relationship to
performance. Their study pointed to the potential of a
broader inquiry into alternatives at the organizational
level of analysis.
2. Background
A project office (PO, also be called a program office) is
an organizational entity established to manage a specific
project or a related series of projects, usually headed by
a project or program manager [13]. A project manage-
ment office (PMO, also called a center of excellence or
center of expertise) is an organizational entity estab-
lished to assist project managers, teams and various
management levels on strategic matters and functional
entities throughout the organization in implementing
PM principles, practices, methodologies, tools and
techniques [13]. A PMO carries a much broader mission
and was the focus of this study.
Dinsmore [14], Fleming and Koppelman [15] and
Knutson [16] call for the establishment of such an office
to improve PM effectiveness, particularly by enabling
the acquisition of knowledge from earlier failures and
successes and by providing a range of support and fa-
cilitative services not only for projects but also for var-
ious management levels and support units. A study of
PM best practices in large functional organizations [17]
reinforced the notion that there is value in utilizing
PMOs. Block and Frame [18] suggest that an ad hoc
approach to PM leads to inefficiencies and can even be
dangerous, while establishment of a PMO can foster
consistency and nurture PM professionalism. They
propose the following characteristics to help improve an
organization’s PM effectiveness:
• Project support to offload administrative burdens
such as reporting and software operations from pro-
ject managers.
• Consulting and mentoring, whereby professional PM
expertise such as proposal development and project
planning is provided.
• Development and enforcement of standards and
methods to leverage best practices and to ensure
members of the organization are all ‘‘speaking the
same PM language.’’
• Training to enhance individual skills and to encour-
age professional certification.
• Assistance in staffing projects with appropriate pro-
ject managers.
• Playing a high-tech project support role by enabling
virtual project offices across geographical and organi-
zational distance.
Bates [19] further adds that PMOs should also as-
sume tasks such as providing project risk assessment,
performing post-project evaluation services and ulti-
mately leading the organizational transition to an ef-
fective project environment.
Little systematic empirical research had been done to
test the growing body of anecdotal evidence, however.
The major objective of this study was to enhance the
strength of the empirical research base that comple-
ments these findings, examining the particular question
of what correlations might exist between the presence of
PMO features and project performance.
To clarify the differences and often interchangeable
use among the names of PMO, PO and other possible
forms (e.g., SPO – systems program office), this study
used the notion of PMO presence which focused on the
functions and services an organization performs and
provides, rather than which name was used. This ap-
proach permitted the inclusion of ‘‘intermediate’’orga-
nizations which had no formally established PMO entity
but did have resources providing PMO functions and
services to project managers and teams.
3. PMO presence
Although a standard set of PMO presence features
has yet to be agreed upon in theory or practice, the lit-
erature review led to the identification of the categories
enumerated below.
3.1. Developing and maintaining PM standards and
methods
A PMO can develop and maintain a set of standards
and methods, becoming a steward of documented PM
expertise within the organization. These standard pro-
cedures should be detailed enough to provide guidance
but not so excessively detailed as to inhibit creativity.
The following list includes representative areas, each of
which was reflected in the survey instrument: proposal
development, change management, risk assessment,
documentation standards and project closeout.
3.2. Developing and maintaining project historical ar-
chives
The PMO can provide a centralized archive to sys-
tematically collect and store project knowledge such as
lessons learned and templates. Representative areas in-
clude records of project performance such as status re-
 C.X. Dai, W.G. Wells / International Journal of Project Management 22 (2004) 523–532
ports, variance analysis and changes to the baseline
plan, risk lists and other risk management documents,
information on prior successful and unsuccessful pro-
jects and a database of lessons learned.
3.3. Providing project administrative support
As project numbers and scale grow large, the asso-
ciated administrative requirements also expand. Ad-
ministrative work often is not reflected directly in
project deliverables and thus can represent a distraction
to the core project team. Representative work areas in
this category include maintenance of a project binder or
web site, assistance in generating standardized reports,
provision of a ‘‘war room’’ for reviews and meetings and
standardization and assistance with PM software.
3.4. Providing human resource/staffing assistance
As more organizations carry out their activities
through projects, the demand for qualified project man-
agers has grown. Assistance can be provided in identify-
ing the proper person to manage a project and the proper
skill requirements for the project team, in gathering data
to conduct performance evaluations, in recruiting project
staff outside the organization and in granting awards or
other types of extraordinary recognition.
3.5. Providing PM consulting and mentoring
As organizations become more sophisticated in PM,
the need to move from an ad hoc to a more strategic PM
approach increases. A PMO can contribute by providing
the following areas of consulting and mentoring: assis-
tance in employing PM methodologies and responding
to risk events, mentoring on unique measures that must
sometimes be taken to foster project success (and shar-
ing those same ideas with upper management) and
group sharing sessions for project managers.
3.6. Providing or arranging PM training
As organizations devote more resources to conducting
business on a project basis, the need for PM training
grows. A PMO can take a leadership role in working with a
human resource department in the areas of skill set iden-
tification, training on PM and related software, financial
support to conduct training and one-on-one coaching.
4. Research methodology
4.1. Measurement instrument
The six functions and services above were embodied
in a questionnaire research instrument to characterize
525
the PMO presence associated with a project outcome.
Selected PM and PMO experts were consulted to help
fine-tune the instrument. Each questionnaire item used a
7-point Likert scale to indicate the level of agreement
with a statement regarding the associated PMO feature.
In treating Not Applicable (N/A) responses, a list-wise
deletion method was used. The respondents were asked
to form their responses based on a recently completed
project on which they had participated. The presence of
each feature was then calculated as the mean of the as-
sociated item scores.
Following Pinto and Slevin [6], multiple measures
were used to form the dependent variable, reported
project performance. These included dimensions of the
triple constraint as well as other indicators of whether
the delivered asset was judged to contribute to broader
organizational performance.
The survey instrument was loaded into a web appli-
cation hosted at The George Washington University.
4.2. Samples for data gathering
The Project Management Institute (PMI ) year 2000
membership list served as one population. One thousand
members in North America were randomly selected
from a pool of 35,880 members who had not chosen to
exclude their names from the list. A total of 234 (23.4%)
responses were received. This will be referred to as the
‘‘random’’population sample.
A second population consisted of a targeted group of
organizations that had been identified in advance as
having some version of a PMO. Some candidates were
drawn from those on the Official Attendee List of the
PMI Annual Symposium 2000 who identified them-
selves as being a member of a PMO or a similar entity.
Others were contacts affiliated with the Master of Sci-
ence in Project Management Degree Program at George
Washington University. The target organizations were
not restricted to North America, but only a small
number were from elsewhere. Ninety-six PMO repre-
sentatives participated, with 52 being partnered with a
project manager from the same organizations. The cri-
terion for selecting the partner was that the individual
had recently completed managing a project for the or-
ganization. PMO representatives were instructed not to
be concerned about the completed project’s outcome in
selecting a potential partner to participate. This will be
referred to as the ‘‘targeted’’population. The choice of
the two distinct populations led to an enriched set of
findings and comparisons.
4.3. Data conditioning
Confirmatory factor analysis was used to assess the
construct validity of the factors comprising PMO
presence. Three of the 30 questionnaire items were
526 C.X. Dai, W.G. Wells / International Journal of Project Management 22 (2004) 523–532

subsequently loaded with factors that matched more have the same degree of access to such documents as did
closely. Cronbach a scores were then used to assess the targeted respondents who were PMO representa-
measure reliabilities. Six a values substantially exceeded tives/managers. A wide range of mission statement
the minimum threshold [20] of 0.70, ranging from 0.81 purposes was revealed. Some focused more broadly on
to 0.92. An independent samples t-test between early organizational goals, while others concentrated more
and late respondents (those who did not respond until specifically on improving PM skills within the organi-
after the second mailing) was conducted [21], with re- zation. While there were many variations in the wording
sults indicating no response bias. of the mission statements, several themes appeared
prominently, including:
• Advocate and support the implementation of best-in-
5. Research findings and discussion
class PM practices, processes and principles across
the organization.
The random and targeted samples presented similar
• Structure and promote an environment in which pro-
work patterns. Respondents had, on average and across
cesses, methods and tools for system development,
many industry and governmental areas, 20 years of
change management and PM are optimally employed
work experience with about 12 years in PM. For the
and continuously improved in the business for
random group, 72% had served as a project manager;
achieving strategic goals.
the proportion for the targeted group was 92%.
• Standardize PM skills and disciplines organization-
wide, while creating an integrated delivery process
5.1. Profile of the establishment and use of PMOs
so reliable, effective and responsive that our custom-
ers identify us as giving them a competitive
Of 234 responses in the random sample, 113 indicated
advantage.
having a PMO or an entity similar to a PMO. Of these,
A point of inquiry in the survey dealt with what or-
the overwhelming majority of PMOs were established in
ganization policy documents had been issued on the
the mid-1990s to 2000 (Fig. 1).
establishment and use of PMOs, with the following
The majority of respondents reported multiple moti-
representative elements:
vations for initiating a PMO. The most frequently re-
• PMO charter
ported factors are listed in Table 1. Both the random
• PM policy/strategy
and the targeted samples indicate a significant number
• PM methodology guidelines
of information technology firms or information systems
• Various standard operating procedures
departments were prominent in the movement toward
• Business justification document
PMO establishment.
• Policies on key areas (e.g., training, project tracking,
An overwhelming proportion of PMOs was approved
planning and configuration management)
at a top/upper management level (Table 2), and the
• Project metrics and standards
number that reported to senior management possessed a
• Reporting mechanisms
large margin over those that did not.
• PMO website or corporate website
A significantly higher rate of mission statement use
• IT governance policy
was reported by the targeted respondents compared to
• Best practices database
the random respondents, 72% compared with 23%.
• Capital project cost reduction initiatives
Many respondents of the random survey did not seem to
• Quality assurance policy, risk management imple-
30 mentation
A high proportion (over 93%) of full-time PMO
25 staffing was reported in both surveys (Table 3). Re-
Number of PMOs Established

spondents in the targeted group also reported various

20
types of support the PMO provided to managers and
project teams (Table 4). Of special significance was that
developing and maintaining PM standards and methods
15
was being attempted in all 96 PMO entities, providing
evidence that PMOs are being used as facilitative units
10
rather than as another line of directive management.
This finding may help defuse concerns that a PMO
5
might contribute another layer of bureaucracy that
would interfere with a project manager’s authority.
0 With regard to PMO funding, 56 out of 113 organi-
76

78

80

82

84

86

88

90

92

94

96

98

00

zations in the random sample provided information, as

Fig. 1. Growth in PMO establishment over time (random sample). did 69 out of 96 organizations in the targeted sample.
 C.X. Dai, W.G. Wells / International Journal of Project Management 22 (2004) 523–532 527

Table 1
Frequently reported motivations for PMO establishment
Motivation Random (113) Targeted (96)
Improving all elements of project management – including performance outcomes, lessons learned, 31.86% 61.46%
and support for project managers – to reducing number of ‘‘troubled projects’’
Achieving a common project management approach – including methodologies, standards and 22.12% 26.04%
accountabilit
Achieving more efficient use of human and other resources in a multiple project environment 14.16% 13.54%
Ensuring more consistent project management training, competence, and performance 0% 10.42%
Improving quality and customer satisfaction 11.5% 10.42%
Incorporating project management with strategic goals and developing competitive advantages 8.85% 8.33%
Directed by an external client, CEO or other senior executive (e.g., CIO) due to positive attitude/ 7.08% 0%
experience with project management
Change to a matrix organization and the complexity of a multiple project environment 3.54% 0%

Table 2
Environmental factors related to the establishment of PMOs
Random (113) Targeted (96)
PMO approval level Top/upper management 86.73% 83.33%
Middle/department management 6.19% 16.67%
PMO reporting level Top/upper management 70.8% 61.46%
Middle/department management 22.12% 38.54%
Yes 22.55% 71.88%
Have a mission statement No 46.9% 26.04%
Yes, but unable to provide detail 19.47% 2.08%
Yes 56.64% 78.13%
Have policy documents No 11.5% 18.75%
In the process of making them Do not know 3.13%

Table 3 The reported data indicated that the proportion of

Summary of PMO staffing type budget devoted to a PMO was estimated at less than 2%
Random (113) Targetted (96) of the organization’s operational budget for 59% in the
Type I only 39.82% 44.79% random sample and 62% in the targeted sample.
Type II only 0.88% 4.17%
Type III only 0.88% 1.04%
Type I and type II 14.16% 16.67% 5.2. Profile of reported projects
Type I and type III 21.24% 9.38%
Type II and type III 0% 1.04% Data on reported industry are shown in Table 5.
Type I, type II and type III 17.7% 22.92%
Do not know 5.31% 0%
Results show that while only 26% of the random re-
Total 100% 100% spondents and 31% of the targeted respondents reported
Type I: Some people perform PMO functions and services as a full- their organizations as being part of the software devel-
time job responsibility; Type II: Some people perform PMO functions opment or computer/data processing/IT industries, about
and services as part of their job description; Type III: Some contrac- 53% and 62%, random and targeted, respectively, of the
tors are used to provide PMO functions and services. reported projects were associated with these areas (see

Table 4
Summary of PMO functions and services from 96 targeted organizations
Targeted (96)
PM standards and methods 100%
Project historical archives 64.58%
Project administrative support 70.83%
Human resource and staffing assistance 28.13%
PM training 70.83%
PM consulting and mentoring 79.17%
Other (frequency) maintaining communication web site: (1) portfolio management (3), configuration management (1), managing projects (3),
monitoring and audit (4), project review (2), quality management assurance (2), sales support (1), financial tracking and execution (1), capital
project financial governance (1), budgeting (1), pm certification program (3), enterprise project management system (1), facilitation, organization
development, strategy (1), project planner (1), assessment review (1), organization change agency (1)
528 C.X. Dai, W.G. Wells / International Journal of Project Management 22 (2004) 523–532

Table 5
Frequency distribution on industry
Type Absolute frequency Relative frequency Absolute frequency Relative frequency
(random) (random) (targeted) (targeted)
Manufacturing 29 12.39% 5 9.62%
Construction 11 4.7% 4 7.69%
Engineering 11 4.7% 0 0%
Business management 11 4.7% 3 5.77%
Software development 26 11.11% 6 11.54%
Telecommunications 31 13.25% 8 15.38%
Computers/data processing /information 35 14.96% 10 19.23%
technology
Transportation 7 2.99% 5 9.62%
Retail 3 1.28% 1 1.92%
Health care 4 1.71% 2 3.85%
Entertainment 2 0.85% 0 0%
Energy 5 2.14% 0 0%
Aerospace 3 1.28% 0 0%
Financial/insurance 25 10.68% 6 11.54%
Pharmaceutical 4 1.71% 1 1.92%
Government 13 5.56% 0 0%
Education 3 1.28% 0 0%
Utility 7 2.99% 1 1.92%
Miscellaneous 4 1.71% 0 1.71%

Grand total 234 100% 52 100%

Table 6
Frequency distribution on end product
Type Absolute frequency Relative frequency Absolute frequency Relative frequency
(random) (random) (targeted) (targeted)
Physical facility 32 13.68% 5 9.62%
Hardware, equipment, or appliance 36 15.38% 7 13.46%
development
Food, drug, or soft goods development 5 2.14% 1 1.92%
New or improved process or software 123 52.56% 32 61.54%
development
Service or test 21 8.94% 5 9.62%
Study or report 8 3.42% 0 0%
Department reorganization or moving 9 3.85% 2 3.85%
to a new facility
Other 0 0% 0 0%
Total 234 100% 52 100%

Table 6). This result reflects the growing importance of
electronic project deliverables in comparison with
physical products.
With regard to project size, 63% and 69% of reported
projects, random and targeted, respectively, fell into one
of the two categories: $100,000–$1 million or $1 million–
$10 million.
5.3. Results of statistical modeling
A comparative summary of simple linear regressions
for the random and the targeted samples is presented in
Table 7. Abbreviations used for the six PMO presence
categories and the dependent variable are as follows:
• ST – PM Standards and Methods
• AR – Project Historical Archives
• AD – Project Administration Support
• HR – Human Resources and Staffing Assistance
• TR – PM Training
• CN – PM Consulting and Mentoring
• DV – Reported Project Performance
All correlations of the DV with the PMO presence
variables were positive, with significant slope coefficient
estimates.
The standards and methods variable (ST) explained
the most variation (R2 ) in reported project performance
from both populations. The Fortune 500 PM Bench-
marking Study [17] reported a similar finding regarding
 C.X. Dai, W.G. Wells / International Journal of Project Management 22 (2004) 523–532 529

Table 7
Summary of simple regression against indicated variables
Variable Estimates Random (n ¼ 234) Targeted (n ¼ 52)
ST b0 intercept 4.319 (17.51)

4.943 (15.77)


b1 slope coefficient 0.269 (5.71)

0.208 (3.68)


R2 0.1243 0.2135
AR b0 intercept 5.317 (30.62)

5.573 (33.39)


b1 slope coefficient 0.095 (2.18)

0.147 (3.32)


R2 0.0207 0.1810
AD b0 intercept 5.216 (27.76)

5.73 (26.29)


b1 slope coefficient 0.105 (2.55)

0.08 (1.73)

R2 0.0280 0.0574
HR b0 intercept 5.113 (25.41)

5.526 (26.55)


b1 slope coefficient 0.129 (2.88)

0.133 (2.90)


R2 0.0348 0.1467
TR b0 intercept 4.935 (21.24)

5.193 (15.81)


b1 slope coefficient 0.159 (3.28)

0.171 (2.73)


R2 0.0444 0.1299
CN b0 intercept 5.082 (23.74)

5.58 (17.93)


b1 slope coefficient 0.130 (2.85)

0.096 (1.61)
R2 0.0341 0.0494
YI ¼ b0 þ bI XI þ eI ; Entries are parameter estimates (t-statistics); e ¼ error term. Models reported here are the original pre-diagnostic versions.
*
p < 0:10 [two-tailed test for variables with hypothesized directions].
**
p < 0:05 [two-tailed test for variables with hypothesized directions].

the importance of standardized methodologies as a ‘‘key
success factor and core best practice’’. It seems that as a
practical matter, organizations that take steps – either
via a PMO or otherwise – to standardize their PM
practices across the organization are more likely to have
stronger project performance.
The variable with the next highest proportion of
variation explained for the targeted population was AR
(18%), but for the random population it was TR (4.4%).
It seems that the targeted population placed much more
emphasis on maintaining and using lessons learned from
previous projects than the random population. This
view regarding the value of organizational learning was
compatible with that of Jessen [5] and the Fortune 500
Benchmarking Study [17].
A rather surprising finding from the targeted popu-
lation was that PM consulting and mentoring (CN) had
the lowest R2 (<5%) and failed to generate a statistically
significant slope coefficient estimate – notwithstanding
that this service was acclaimed among PM authorities
and PMO practitioners. Perhaps consulting and men-
toring are often provided for projects that are already at
risk for other reasons, thus diluting their perceived
efficacy.
Similar weak performance resulted for project his-
torical archives (AR) in the random population. This
was particularly surprising given the stronger perfor-
mance of this variable for the targeted population and
high status accorded this feature by the Fortune 500
Benchmarking Study [17]. This may reflect that either
archives have not yet been established in many organi-
zations or the established archives need to improve their
service and support to project managers and teams.
In general, data from the targeted population yielded
higher R2 values, indicating that correlations between
PMO functions and services and project performance
are stronger here than in the random population. One of
the reasons could be that those project managers from
the targeted group who were selected by their PMO
representatives/managers were inherently more aware of
a PMO presence, which highlighted the PMO contri-
bution to project performance. On the other hand, it
seems that in many organizations there are factors in
addition to PMO presence that are influencing project
outcomes.
The diagnostic tests for several variables in the ran-
dom population called for weighted least square trans-
formation to satisfy regression assumptions, but the
resulting model estimates ended up being very close to
the original estimates. Details of all regression models,
including diagnostic tests on the residuals, are described
in [22].
The targeted population data yielded a multiple linear
regression model on ST and AR with an F-statistic of
8.455 (Table 8) and an adjusted R2 of 22.62%, a small
but significant improvement over the simple regression
on ST alone. The addition of the second predictor AR
was significant at a 90% confidence level, given that ST
was already in the model. The targeted population
consistently emphasized saving and applying lessons
530 C.X. Dai, W.G. Wells / International Journal of Project Management 22 (2004) 523–532

Table 8 functions and services, but no formal entity had been

Regression model with independent variables ST and AR established. The corresponding research hypothesis was:
Parameter Estimate T for H0 : Pr > jT j SE of
Parameter ¼ 0 estimate Ha : The mean scores of reported project performance differ be-
tween organizations having a PMO, having no PMO and hav-
Intercept 4.9864316 16.16 0.0001 0.308567 ing something in-between.
ST 0.147581 2.23 0.0303 0.066149
AR 0.0853084 1.68 0.0997 0.050845 In addition, we examined whether the use of PMO
Model – reported project performance ¼ 4.986 + 0.148
ST + functions and services differed significantly among or-
0.085
AR. F value ¼ 8.455; Prob > F ¼ 0.0007; R2 ¼ 0:2566; Adj ganizations. It was intended to determine whether the
R2 ¼ 0:2262. establishment of a PMO entity contributed to the use of
PM standards and methods, historical archives, ad-
ministrative support, human resources/staffing assis-
learned from previous projects vis- a-vis the random
tance, training, and mentoring and consultancy. The
population.
second research hypothesis was:
The simple regression against ST was the best linear
model for the random population. No variable offered Ha : The mean scores for the PMO functions and services
any statistically significant predictive power once the ST variables (ST, AR, AD, HR, TR and CN) differ between orga-
variable was included. nizations having a PMO, having no PMO, and having some-
thing in-between.

5.4. Project performance in organizations with PMOs
versus those without PMOs
Data were also gathered from the random population
on whether project performance varied between orga-
nizations that had established PMO entities versus those
that did not. The organizations compared were: (1)
those having a fully established PMO, (2) those having
no PMO and (3) those having something in-between,
where somebody was tasked with working on PMO
Sheffe’s a posteriori test was conducted to investigate
variable status, with groupings presented in Table 9.
When the populations share identical codes (such as
three As), the means among three groups are not sta-
tistically different; but when they have different codes
(such as A, B and AB), the means of the three groups are
statistically different.
For example, for the PM training variable TR, be-
tween populations of PMOs and No-PMOs, there is
statistical evidence (at a 95% confidence level) to con-

Table 9
Scheffe’s a posteriori test for PMOs, No-PMOs and in-betweens for the random sample
Variable Population N Mean SD Scheffe codes
(a ¼ 0:05)
ST PMO 111 5.32 1.56 A
No PMO 45 4.43 1.43 B
In-between 76 4.95 1.26 AB
AR PMO 110 4.06 1.77 A
No PMO 44 2.83 1.68 B
In-between 74 3.37 1.56 AB
AD PMO 110 4.47 1.82 A
No PMO 44 3.85 2.09 A
In-between 74 3.94 1.68 A
HR PMO 112 4.47 1.77 A
No PMO 45 3.78 1.48 A
In-between 75 4.04 1.55 A
TR PMO 113 4.93 1.44 A
No PMO 45 4.05 1.55 B
In-between 76 4.29 1.48 B
CN PMO 112 4.75 1.59 A
No PMO 45 3.85 1.69 B
In-between 75 4.26 1.58 AB
DV PMO 113 5.66 1.22 A
No PMO 45 5.51 1.29 A
In-between 76 5.75 0.92 A
 C.X. Dai, W.G. Wells / International Journal of Project Management 22 (2004) 523–532
clude that the mean scores differ. Likewise, between
populations of PMOs and In-Betweens, there is statis-
tical evidence to conclude that the mean scores differ.
However, between populations of No-PMOs and In-
Betweens, there is no statistical evidence to conclude
that the mean scores differ.
The results in the final row of Table 9 show that re-
ported project performance is higher in organizations
that have a PMO in comparison with organizations that
do not (5.66 vs. 5.51), but not high enough to merit
statistical significance. On the other hand, organizations
that have a PMO have clearly done more than those that
do not have a PMO in promoting PM standards and
methods, historical archives, training, and consulting
and mentoring.
6. Discussion of implications
The results from this research support several con-
clusions in terms of the stated research question. They
also point to a number of practical guidelines for or-
ganizations already operating a PMO and for those
considering a move in this direction. Among the specific
ways in which the results might be practically applied
are the following:
• Many organizations are moving in the direction of es-
tablishing PMOs or have already done so, demon-
strating a high level of management confidence in
the utility of this innovation.
• There is strong evidence that PM standards and
methods are most highly correlated with project per-
formance. This PMO feature should take priority
over the others studied, whether a formal PMO entity
is being established or not.
• The use of project historical archives also showed a
significant correlation with project performance.
• Pioneers in establishing PMOs are providing infor-
mation and advice on essential policies and docu-
ments that should accompany the establishment and
use of a PMO (see [22] for examples of such docu-
ments currently in use).
Several areas where additional research in the future
might be worthwhile include:
• There were indications that PMOs have been adopted
more rapidly in the newer technological areas (e.g.,
IT) than in the older, more mature industries. This
phenomenon requires further inquiry to determine
its validity and whether the source is simply an arti-
fact of the research design or the nature of the survey
populations.
• PMOs in early stages of their use may not be fully
representative of what will finally unfold in later
years. It may be necessary to design a research proto-
col that is based on the examination of PMOs after
they have been in operation for some considerable pe-
531
riod of time and accumulated a significant data base.
Moreover, the use of detailed case studies would
complement the broader survey approach used in this
study.
• A number of disparities emerged in the findings when
comparing the different populations. The major one
was with respect to the value of a PMO in contribut-
ing to project performance as seen by organizations
that have PMOs and those that do not. Additional re-
search may help illuminate further correlates with
these reported performance differences.
Acknowledgements
Appreciation and gratitude for providing financial
and other types of support for this research are extended
to The George Washington University, ESI Interna-
tional, and, of course, to the hundreds of individuals
who responded via the surveys. Thanks also to Jeffrey
Pinto for allowing us to adapt his project performance
questionnaire for use in this research.
References
[1] Balck H. Projects as a form of change. In: Gareis R, editor.
Handbook of management by projects. Vienna: MANZsche
Verlags; 1990. p. 22–8.
[2] Gareis R. Management by projects: the management strategy of
the ‘new’ project-oriented company. In: Gareis R, editor. Hand-
book of management by projects. Vienna: MANZsche Verlags;
1990. p. 35–47.
[3] Lundin R. Incentives for chief executives to manage by projects.
In: Gareis Roland, editor. Handbook of management by projects.
Vienna: MANZsche Verlags; 1990. p. 48–53.
[4] Munns AK, Bjeirmi BF. The role of project management in
achieving project success. Int J Project Manage 1996;14(2):81–7.
[5] Jessen SA. The nature of project leadership. Scandinavian
University Press; 1992.
[6] Pinto J, Slevin D. Project success: definitions and measurement
techniques. Project Manage J 1988;XIX(1):67–72.
[7] Dvir D, Lipovetsky S, Shenhar A, Tishler A. In search of project
classification: a non-universal approach to project success factors.
Res Policy Amsterdam 1998;27(9):915–35.
[8] The Standish Group. The CHAOS Report. International Inc.
copyright 1995. Available from http://www.standishgroup.com/
sample_research/chaos_1994_1.php.
[9] Cringely RX. How to forfeit millions in exchange for nothing.
Forbes ASAP; August 1994. p. 60–4.
[10] Morris PWG, Hough GH. The anatomy of major projects – a
study of the reality of project management. Wiley; 1987.
[11] Pinto JK. Project implementation: a determination of its critical
success factors, moderators, and their relative importance across
the project life cycle. Dissertation, University of Pittsburgh; 1986.
[12] Might RJ, Fischer WA. The role of structural factors in
determining project management success. IEEE Trans Eng Manag
1985;32(2):71–7.
[13] Ward L. Project management term – a working glossary. 2nd ed.
ESI International; 2000.
[14] Dinsmore PC. Wining in business with enterprise project man-
agement. New York: AMACOM; 1999.
532 C.X. Dai, W.G. Wells / International Journal of Project Management 22 (2004) 523–532
[15] Fleming QW, Koppelman JM. Project teams: the role of the
project office. Cost Eng 1998;40(8):33–6.
[16] Knutson J. The project office: an evolutionary implementation
plan, PM Network; September 1998. p. 14–6.
[17] Toney F, Powers R. Best practices of project management
group in large functional organizations: results of the fortune
500 Project Management Benchmarking Study. PMI; 1997. p.
84.
[18] Block TR, Frame JD. The project office – a key to managing
projects effectively. Crisp Publications; 1998.
[19] Bates WS. Improving project management: better project man-
agement begings with a project management office. Indust Eng
1998;30(10):42.
[20] Hair JF, Anderson RE, et al. Multivariate data analysis. 4th ed.
Prentice-Hall; 1995.
[21] Oppenheim AN. Questionnaire design and attitude measurement.
New York: Basic Books; 1966. p. 34.
[22] Dai CX. The role of the project management office in achieving
project success. Dissertation, The George Washington University;
2002.