You are on page 1of 16

CONSTRUCTION MANAGEMENT AND ECONOMICS

2021, VOL. 39, NO. 11, 879–893


https://doi.org/10.1080/01446193.2021.1993289

Exploring the impact of project size on design-bid-build and design-build


project delivery performance in highways
Phuong H. D. Nguyena, Dai Q. Tranb and Sai P. K. Bypanenic
a
Department of Civil and Environmental Engineering, University of Alberta, Edmonton, Canada; bDepartment of Civil, Environmental,
and Architectural Engineering, University of Kansas, Lawrence, KS, USA; cO’Connor Construction Management Inc, Rockville, MD, USA

ABSTRACT ARTICLE HISTORY


The highway construction industry has been increasingly using design-build (D-B) project deliv- Received 17 November 2020
ery because of its documented benefits. Only limited studies, if any, have compared D-B and Accepted 11 October 2021
traditional design-bid-build (D-B-B) project performance based on the size of projects. The exist-
KEYWORDS
ing gap was addressed by investigating the cost growth, schedule growth, and construction
Highway project delivery
intensity of highway projects delivered under D-B-B and D-B with the project sizes ranging from methods; cost performance;
$2 million to greater than $100 million. These projects were collected from six state depart- schedule performance;
ments of transportations in the U.S. with mature D-B programs. Several statistics tests including construction intensity;
t-test, Welch’s test, and Mann–Whitney U-test were used to analyze 3888 completed highway project size
construction projects. The results show that D-B produced a lower cost growth than D-B-B in
the project size of over $10 million while D-B-B projects had lower cost growth than D-B proj-
ects in the project size from $2 million to $10 million. D-B projects also produced less schedule
growth than D-B-B projects across all groups of project size. D-B-B projects had a faster work
pace than D-B projects in the project size from $2 million to $10 million. The findings contribute
to the project delivery body of knowledge by reporting statistically empirical comparisons
between D-B-B and D-B project performance controlled by the project size. The findings may
help highway agencies better understand how D-B performs under the different sizes of projects
in terms of cost growth, schedule growth, and construction intensity.

Introduction D-B is fundamentally different from the traditional


design-bid-build (D-B-B) method in that contractual
In the highway construction industry, traffic growth,
relationship between the main parties involved in a
deteriorating infrastructure, and increasing population
project. Under the D-B contracting method, a single
have created substantial pressures to move critical entity is responsible for both the design and construc-
projects quickly from the planning and design phases tion of a project. The integration of design and con-
to the construction phase. To respond to those pres- struction services under a single contract is able to
sures, a design-build (D-B) delivery method has been produce earlier cost and schedule certainties and allows
utilized in many highway projects in the world for the construction process to begin before the com-
(Gudiene_ et al. 2013, Mahamid 2013, Park et al. 2015). pletion of the design process (Chen et al. 2016). These
In the United States, many state departments of trans- differences have created varied cost and schedule per-
portation (DOTs) have increasingly implemented D-B formance between D-B-B and D-B projects. Statistically
over the past 20 years. According to the Design-Build comparing project performances between D-B and
D-B-B is a challenging task since there are many param-
Institute of America (DBIA)’s report, as of 2018, only
eters involved, such as project characteristics, internal
six state DOTs (North Dakota, Iowa, Wisconsin, Illinois,
and external conditions, and other relevant administra-
Kansas, and Oklahoma) have not been fully authorized tion issues (Sullivan et al. 2017). A large number of
to practically utilize D-B for transportation projects studies have compared cost and schedule performance
(DBIA 2018). It is noted that some of these six state between D-B and D-B-B projects. For example, the pre-
DOTs, such as Kansas and North Dakota, have experi- vious studies (Ibbs et al. 2003, FHWA 2006, Hale et al.
mented D-B for their highway construction projects. 2009, Shrestha et al. 2011, Goftar et al. 2014,

CONTACT Phuong H. D. Nguyen phuonghdnguyen@ualberta.ca Department of Civil and Environmental Engineering, University of Alberta, 116 St.
and 85 Ave., Edmonton, Alberta T6G 2R3, Canada
ß 2021 Informa UK Limited, trading as Taylor & Francis Group
880 P. H. D. NGUYEN ET AL.

Shrestha and Fernane 2016) showed that D-B outper- growth, schedule growth, and construction intensity)
forms D-B-B in terms of cost and schedule performance between D-B-B and D-B highway projects across differ-
ranging from 5% to 10% and from 4% to 16%, respect- ent groups of the project’s size from $2 million to
ively. In South Korea, D-B construction projects over $100 million. The results are expected to provide
produced a lower cost growth of 5% and a shorter transportation authorities with practical evidence
construction duration by twelve days per floor com- regarding the cost and schedule performance of D-B-B
pared to D-B-B construction projects (Park et al. 2015). and D-B highway construction projects.
On the other hand, several studies (Shrestha et al. The remainder of the article is structured as follows.
2011, Gudiene_ et al. 2013, Minchin et al. 2013) found The literature review section explores recent perform-
that D-B-B construction projects achieved better per- ance comparisons between D-B-B and D-B highway
formance with 25% lower in cost growth and 15% projects and indicates research gaps. The research
lower in schedule growth than D-B construction proj- method section presents research questions and
ects. However, those findings from previous studies hypotheses, data collection, and methods of analysis.
were based on small sample sizes, expert’s judgments, The results and discussion discuss key findings in
and limited case studies (Park and Kwak 2017). detail. These are followed by the conclusion and
Therefore, the common perception that D-B outper- contribution of the study.
forms D-B-B in construction projects needs to be empir-
ically tested with larger sample sizes considering
Literature review
various project attributes, such as project type, dur-
ation, project complexity, and project size. This section summarises (1) impacts of project size on
Project size is one of the important factors that project performance as well as delivery method selec-
affect the success of construction projects regardless tion and (2) current practices in comparing the per-
of project delivery methods (Gudiene_ et al. 2013). For formance of D-B-B and D-B highway construction
instance, Yong and Mustaffa (2012) identified the four projects. Based on the key findings from the literature,
most common factors that affect the success of the research hypotheses development was then discussed
construction project including project size, manpower in detail.
allocation, project complexity, and urgency in meeting
project deadlines. The increase of project size is highly
Impacts of project size on project performance
correlated to project complexity, the number of par-
and delivery method selection
ties involved, and various delivery risk factors (Luo
et al. 2017). A large project typically results in longer Regardless of applied delivery methods, project size is
project duration, intricate construction contracts, one of the critical factors that affect the performance
uncertain cost estimates, and a more comprehensive of construction projects in terms of cost, schedule,
organizational structure (Russell et al. 2014, Qureshi and owner satisfaction (FHWA 2006, Yong and
and Kang 2015). Therefore, different sizes of projects, Mustaffa 2012). In fact, an increase in construction
from small projects to mega projects, lead to varia- project sizes drives the number of uncertainties in pro-
tions of project performance in terms of cost, sched- ject outcomes, the complexness of construction con-
ule, and quality (Love et al. 2015). However, there are tracts, and the comprehensiveness of organizational
limited studies, if any, that have investigated the cost structures (Qureshi and Kang 2015, Luo et al. 2017,
and schedule performance of D-B-B and D-B highway Sullivan et al. 2017). Larger projects are exposed to a
construction projects across different project sizes. variety of factors that affect the success of a construc-
There is also a lack of statistical confirmations regard- tion project, including planning feasibility, design and
ing the significant difference and superiority between construction complexity, delivery risk factors, resource
these two common delivery methods in highway con- availability (i.e. labour, materials, and equipment),
struction in terms of project sizes. To fulfil these exist- environmental and economic impacts, and legal regu-
ing gaps, the current study aimed at empirically lations (Gudiene_ et al. 2013). The difference in project
assessing whether or not there is a difference between sizes should be included in comparing the perform-
D-B-B and D-B project performance in terms of cost ance of construction projects using different delivery
growth, schedule growth, and construction intensity methods (e.g. D-B-B and D-B) (Love et al. 2015, Park
across different project sizes. Specifically, statistical et al. 2015).
inference analyses were utilized to investigate the dif- Project size affects the cost and schedule perform-
ferences in cost and schedule performance (e.g. cost ance comparisons between the traditional D-B-B and
CONSTRUCTION MANAGEMENT AND ECONOMICS 881

alternative contracting methods, such as D-B (Park Declaration 1: D-B outperforms D-B-B in high-
et al. 2015, Qiang et al. 2015). FHWA (2006) found that way projects
the selection of D-B delivery for highway construction Since the U.S. agencies started to use D-B in their proj-
projects is highly correlated to the size of the project. ects, the majority of research and white papers have
As a result, FHWA (2006) recommended that D-B concluded that D-B projects tend to achieve better
should be used in medium to large highway construc- project performance than D-B-B projects in terms of
tion projects (over $10 million). Different project sizes cost, schedule, and quality. As one of the first studies
lead to diverse cost deviations in D-B-B and D-B proj- in this area, Ellis et al. (1991) evaluated 11 D-B projects
ects because of the increase in resource requirements in the Florida DOT Pilot D-B program and found that
of project controls, team communications, and man- D-B produced approximately 11% cost savings and
agement experience (Mahamid 2013). Project size also 36% faster delivery than D-B-B. With the updated
differentiates D-B-B and D-B in terms of impacts on database, Ellis et al. (2007) found that D-B-B projects
project durations (Russell et al. 2014). Based on these had higher cost growth than D-B projects (9.4% and
salient findings from the literature, it is reasonable to 4.5%, respectively) with a dataset of 66 D-B and 1847
conclude that project size is one of the decisive fac- D-B-B projects. In terms of schedule performance,
tors that contribute to the selection of an appropriate using D-B-B produced a higher schedule growth com-
delivery method for highway construction projects. pared to D-B projects (16.5% and 7.1%, respectively).
Supporting the conclusion that D-B projects outper-
Performance comparisons between D-B-B and D-B form D-B-B projects, Warne (2005) investigated 21 D-B
highway construction projects and 39 D-B-B projects across the U.S. with a total cost
ranging from $83 million to $1.3 billion. The research
There are rich literature to compare project perform-
found that D-B projects had better cost performance
ance between D-B-B and D-B projects across different
than D-B-B projects (4% less in cost growth). The
project types, scope of work, and procurement meth-
researcher also found that 76% of the D-B projects
ods (e.g. low-bid and best-value). Previous studies
were completed ahead of schedule. Shrestha et al.
have investigated the cost and schedule performance
(2007) conducted a quantitative comparison of 4 D-B
of D-B-B and D-B in many studies associated with dif-
and 11 D-B-B projects and found that the cost growth
ferent types of projects, including general buildings
of D-B projects was significantly lower than that of the
(Molenaar and Songer 1998, Konchar and Sanvido
D-B-B projects (5.5% with D-B and 4.1% with D-B-B).
1998); military buildings (Hale et al. 2009); university
buildings (Shrestha and Fernane 2016); industrial proj- The researchers also found that the scheduled growth
ects (Songer and Molenaar 1997, Konchar and Sanvido of D-B projects was lower than that of D-B-B projects
1998); and mechanical projects (Riley et al. 2005). (7.6% with D-B and 12.9% with D-B-B). On the other
Construction intensity between D-B-B and D-B high- hand, the number of change orders of D-B-B projects
way construction projects has been investigated by was found lower than that of D-B projects (5.3% with
Tran et al. (2017) and Antoine et al. (2019). D-B and 3.9% with D-B-B). It is noticed that no statis-
Comparisons of change orders between D-B-B and D-B tical significance was mentioned in these previous
construction projects have been studied by Alleman studies. Reports from the Federal Highway
et al. (2020). The comprehensive summary and discus- Administration (FHWA) show that D-B performed
sions of non-highway D-B-B and D-B project perform- superior to D-B-B in terms of both cost growth (3.6%
ance can be found in Shrestha and Fernane (2016), and 7.4%, respectively) and schedule growth (4.2%
Goftar et al. (2014), Park and Kwak (2017), and Sullivan and 4.8%, respectively) with an analysis of 11 pairs of
et al. (2017). The current study mainly concentrates on D-B and D-B-B projects (FHWA 2006). By analyzing 129
empirical comparisons of performance indicators (e.g. D-B-B and 110 D-B projects, FHWA (2018) stated that
cost, schedule, and intensity) between D-B-B and D-B D-B projects had fewer cost overruns than D-B-B proj-
projects in the highway sector. According to the cur- ects (3.4% and 4.1%, respectively). FHWA (2018) also
rent highway project delivery literature, although the compared the scheduled growth by analyzing 63
performance of D-B projects has been regularly con- D-B-B and 70 D-B projects. The result showed that D-B
cluded to be superior to the performance of D-B-B projects had less schedule growth than D-B-B projects
projects (Declaration 1), some studies have countered (2% and 10%, respectively) (FHWA 2018). These results
this conclusion at some points (Declaration 2) as the were not statistically significant at the 95%
following discussions. confidence level.
882 P. H. D. NGUYEN ET AL.

Using different projects scopes, Tran et al. (2017) Research hypotheses development
compared the cost and schedule performance of 139
The common perception in the construction industry is
pairs of D-B-B and D-B highway projects, including
that D-B projects can produce less cost growth and
new construction, reconstruction, resurfacing, restor-
schedule growth compared to D-B-B projects.
ation, rehabilitation (3R) projects, intelligent transpor-
Additionally, D-B projects are expected to have a more
tation system, and other types. A two-sample t-test
intense working pace than D-B-B projects. However, pre-
analysis was conducted along with descriptive ana-
vious studies found that in certain situations involving
lysis. They found that D-B projects statistically design errors, utility conflicts, or right-of-way issues, D-B
achieved lower cost growth and schedule growth in highway construction projects produce higher cost
3R projects and other project work types. In new con- growth and schedule growth compared to D-B-B projects
struction and reconstruction projects, the use of D-B (Shrestha et al. 2011, Gudiene_ et al. 2013, Minchin et al.
produced a higher construction intensity compared to 2013, Park and Kwak 2017). Therefore, the difference and
D-B-B. superiority in performance between D-B-B and D-B high-
way projects need more justifications based on multiple
Declaration 2: D-B-B outperforms D-B in high- factors, such as project types, project sizes, project com-
way projects plexity, project duration, and delivery risk factors.
Although the majority of previous studies found that The increase of project size is highly correlated to
D-B outperforms D-B-B in terms of cost and schedule various factors, including project complexity, project
performance in highway construction projects, some durations, organizational structures, and project risk
studies showed that D-B-B performed better than D-B. profiles, that affect project performance (Luo et al.
Under some circumstances where there are design 2017). Larger highway projects typically have a longer
errors, high rates of change orders, and miscommuni- project duration, sophisticated construction contract,
cation between the contractor and the design team, uncertain cost estimate, and higher number of parties
D-B projects suffer cost and schedule performance. involved that lead to variations of project performance
For example, Shrestha et al. (2011) found that the total (Qureshi and Kang 2015). Thus, conducting compari-
cost growth of D-B projects achieved 1.5% higher sons between D-B-B and D-B project performance in
than that of D-B-B project (7.8% with D-B and 6.3% terms of project size can help explain the contradicting
with D-B-B) by analyzing 6 D-B and 16 D-B-B projects. results in the current project delivery literature. Further,
The study also indicated that the total schedule previous studies showed a lack of powerful statistical
growth of D-B projects was 15.4% higher than that of inferences regarding the influence of project sizes on
D-B-B projects (20.5% with D-B and 5.1% with D-B-B). D-B-B and D-B performance. Goftar et al. (2014) found
Minchin et al. (2013) analyzed 60 transportation proj- that alternative project delivery methods tend to be
ects collected from the Florida DOT database and more frequently utilized in large construction projects
found that D-B-B projects performed significantly bet- than in small construction projects. Sullivan et al. (2017)
ter in cost performance. Specifically, the cost growths indicated that project type and size critically affect pro-
of D-B and D-B-B projects were 45.3% and 20.4%, ject performance. Differences in project size also indi-
respectively. The results contradicted many previous cate a variation in project complexity and risks, which
studies in the construction literature. The authors’ jus- are often considered as the main factors in increasing
tification was that a large number of the highway cost growth (Nguyen et al. 2020).
projects in their study were completed about 15 years To address these existing gaps, the primary object-
ago while D-B was still developing while D-B-B had ive of the current study was to empirically investigate
been implemented for a long time (Minchin the performance of D-B-B and D-B highway construc-
et al. 2013). tion projects across different project sizes ranging
Table 1 summarizes key findings of comparisons from $2 million to over $100 million. Based on the
between D-B-B and D-B highway projects in terms of findings in the literature presented in the Declaration
common project performance metrics, including cost 1: D-B Outperforms D-B-B in Highway Projects, three
growth, total cost and unit cost, schedule growth, and research hypotheses (Hi) of cost growth, schedule
delivery time. Accordingly, the majority of the studies growth, and construction intensity were formed to
show that D-B projects outperform D-B-B projects in test the performance of D-B-B and D-B highway con-
terms of cost and schedule performance while a few struction projects as follows:
studies disagree with that. It is noted that project sizes H1: D-B highway projects produce less cost growth than
are largely varied in the previous studies. D-B-B highway projects across different project sizes.
CONSTRUCTION MANAGEMENT AND ECONOMICS 883

Table 1. Summary of highway D-B-B and D-B project performance comparisons.


Metric Studies Sample size (n) Major findings and statistical results
Cost growth Warne (2005) D-B:21 D-B outperformed D-B-B by 4%. No statistical results
D-B-B:39 were reported.
FHWA (2006) D-B:11 D-B-B outperformed D-B by 3.8%. No statistical results
D-B-B:11 were reported.
Ellis et al. (2007) D-B: 66 D-B-B outperformed D-B by 4.9%. No statistical results
D-B-B: 1,847 were reported.
Shrestha et al. (2007) D-B:4 D-B outperformed D-B-B by 9.6%. F-test was used to compare
D-B-B:11 means of two samples with p ¼ 0.03. Note that F-test is often
used to test variances of two samples.
Shrestha et al. (2011) D-B:6 D-B-B outperformed D-B by 1.5%. ANOVA was used for analysis,
D-B-B:16 but no significant results were found, p ¼ 0.751.
Minchin et al. (2013) D-B:30 D-B-B outperformed D-B by 24.9%. Nonparametric statistical tests
D-B-B:30 were used but no significant results were found, p ¼ 0.209.
FHWA (2018) D-B:129 D-B outperformed D-B-B by 0.7%. No statistical results
D-B-B:110 were reported.
Total cost/ Ellis et al. (1991) Ernzen and D-B: 11 D-B produced approximately 11% cost savings and 36% faster
unit cost Schexnayder (2000) D-B-B: 11 delivery
D-B: 1 D-B outperformed D-B-B by 15%. No statistical results
D-B-B: 1 were reported.
FHWA (2006) D-B:11 D-B outperformed D-B-B by 3%. No statistical results
D-B-B:11 were reported.
Schedule growthFHWA (2006) D-B:11 D-B outperformed D-B-B by 9%. No statistical results
D-B-B:11 were reported.
Ellis et al. (2007) D-B: 66 D-B outperformed D-B-B by 9.4%. No statistical results
D-B-B: 1,847 were reported.
Shrestha et al. (2007) D-B:4 D-B outperformed D-B-B by 5.3%. F-test was used to compare
D-B-B:11 means of two samples with p ¼ 0.51.
Shrestha et al. (2011) D-B:6 D-B-B outperformed D-B by 15.4 %. ANOVA was used for
D-B-B:16 analysis, but no significant results were found, p ¼ 0.17.
Minchin et al. (2013) D-B:30 D-B outperformed D-B-B by 2.8%. Nonparametric statistical tests
D-B-B:30 were used but no significant results were found, p ¼ 0.229.
FHWA (2018) D-B:70 D-B outperformed D-B-B by 8%. No statistical results
D-B-B:63 were reported.
Delivery time Warne (2005) D-B:21 100% interviewees agreed that D-B was faster than D-B-B. No
D-B-B:39 statistical results were reported.
FHWA (2006) D-B:11 D-B outperformed D-B-B by 14%. No statistical results
D-B-B:11 were reported.

H2: D-B highway projects produce less schedule were formed based on the literature review. Then,
growth than D-B-B highway projects across different three project performance metrics, including cost
project sizes. growth, schedule growth, and construction intensity,
H3: D-B highway projects have higher construction were defined. The data collection and data analysis
intensity than D-B-B highway projects across different including multiple statistical tests were then performed
project sizes. to examine the hypotheses. Finally, hypothesis testing
The established hypotheses align with the common results were discussed to draw a conclusion and claim
perception in the construction industry that D-B projects contributions of the current study. The following sec-
produce more cost and schedule certainties compared tions discuss these steps in more detail.
to D-B-B projects. Hypotheses H1 and H2 are supported
by Shrestha et al. (2007), Tran et al. (2017), FHWA (2018), Project performance metrics
and Nguyen et al. (2020) while Hypothesis H3 is in line Many project performance metrics have been used to
with Tran et al. (2017), Antoine et al. (2019), and evaluate highway construction projects, such as cost
Alleman et al. (2020). The results are expected to sup- growth, total cost, unit cost, the total cost of change
port transportation authorities at both national and orders (design and construction), schedule growth, deliv-
international levels in understanding the difference in ery speed, and quality. Three common performance met-
cost and schedule performance between D-B-B and D-B rics in highway construction, including cost growth,
construction projects in different project sizes. schedule growth, and construction intensity, were utilized
in the current study.
Research method
Cost growth
Figure 1 shows an overview of the research process for Cost growth is the percentage change in cost between
the current study. It is noted that research hypotheses the contract awarded amount and the final cost.
884 P. H. D. NGUYEN ET AL.

DATA COLLECTION DATA ANALYSIS

DESCRIPTIVE
DELIVERY METHODS STATISTICS
Design-Bid-Build (D-B-B) Project Size
Design-Build (D-B) • $2M - $10M
• $10M - $20M
• $20M - $50M
• $50M - $100M
DATA SAMPLING • Over $100M
LITERATURE REVIEW Highway Projects from RESULTS AND
Performance comparisons Six State DOTs DISCUSSION
between D-B-B and D-B
• Florida • Ohio Normality Variance
• Indiana • Oregon Test Test
• North Carolina • Utah
INFERENTIAL
STATISTICS
Hypothesis Testing
PERFORMANCE METRICS
• t-test
Cost Growth
• Welch's t-test
Schedule Growth
• Mann–Whitney U-test
Construction Intensity

Figure 1. Research process.

Change orders are one of the factors that increase the of the entire project. The duration is calculated by
cost of the project causing a positive cost growth. the difference between the planned construction-
Possibly overestimated or limiting scopes of work can started date and the bid-contract-end date. The
reduce project costs and result in negative cost actual construction duration is measured as the
growth. With D-B-B, the cost growth value refers to period (in days) from the date that construction work
construction costs only. With D-B, this value is inclu- started to the date of substantial completion of the
sive of construction and design costs by the design- work. With D-B-B, the schedule growth value involves
builder. The final cost is the total cost of installation of activities in the construction phase. With D-B, this
all project components after changes and miscellan- value is inclusive of both construction and design
eous expenses accrued. The contract awarded amount performed by the design-builder. The schedule
is the price quoted by the bid winner and should not growth metric was calculated using Equation (2) as
follows:
be included in construction engineering inspections,
right-of-way (ROW), and other costs unless it is a part Final duration – Contract duration
Schedule growth ð%Þ ¼ %
of the original bid. The cost growth metric was calcu- Contract duration
(2)
lated using Equation (1) as follows:
Final cost – Contract award cost
Cost growth ð%Þ ¼ % Construction intensity
Contract award cost
(1) Construction intensity is an indicating factor of how
quick the construction pace of a project is; in other
words, it presents the amount of dollars that the
Schedule growth contractor needs to employ per day. Accordingly, this
Schedule growth indicates that the project is either performance metric is the ratio between the final cost
completed within the planned construction schedule and the actual duration of a construction project. The
(negative value) or beyond the planned construction construction intensity metric was calculated using
schedule (positive value). This value is the percentage Equation (3), where the final cost and the actual con-
change from the actual construction duration of struction duration are defined above. It is important to
the project to the planned construction duration of note that the actual construction duration for D-B
the project. The planned construction duration is projects in the current study was recorded since the
estimated by the contractor or the design-builder to time that the design-builder received the notice to
execute and complete the physical building activities proceed after the contract was awarded.
CONSTRUCTION MANAGEMENT AND ECONOMICS 885

Table 2. Descriptive statistics of D-B-B and D-B project costs.


D-B-B (Total n ¼ 15,786) D-B (Total n ¼ 596)
State DOT n Mean ($) Min. ($) Max. ($) n Mean ($) Min. ($) Max. ($)
Florida 4592 3,625,385 4000 192,789,217 407 10,616,073 24,447 242,787,000
Indiana 2728 1,382,846 7411 45,922,565 34 5,085,408 133,305 58,527,877
North Carolina 41 15,770,642 553,500 116,470,217 33 49,926,293 2,462,594 192,040,143
Ohio 6829 1,738,477 9650 219,996,000 77 2,957,932 95,000 23,444,848
Oregon 522 3,236,505 55,834 53,219,440 10 45,098,429 619,000 129,900,000
Utah 1074 2,257,206 15,980 99,681,923 34 30,713,016 147,998 182,885,103

Construction intensity ð$=dayÞ analysis. Based on the discussion with experts from
Final cost these state DOTs, five groups by project size were
¼ (3) used. The group sizes were chosen to provide the
Actual construction duration
practical resolution possible. These five groups include
the projects with the contract award amount ranged
Data collection from $2 million to $10 million ($2M to $10M); $10 mil-
lion to $20 million ($10M to $20M); $20 million to $50
To compare cost and schedule performance between
million ($20M to $50M); $50 million to $100 million
D-B-B and D-B highway projects, empirical project
($50M to $100M); and more than $100 million (over
data from six DOTs (Florida, Indiana, North Carolina,
Ohio, Oregon, and Utah) that have extensive experi- $100M). Table 3 provides descriptive statistics of the
ence in using D-B delivery were collected. The primary contract awarded amounts and the sample sizes (n) of
collected information in the request included: (1) pro- D-B-B and D-B projects associated with the five
ject name or identification number; (2) project delivery selected groups.
methods: D-B-B and D-B; (3) project cost and schedule It is noted that D-B is often used for specific proj-
data. The project cost data contained information ects. As a result, the sample size of D-B-B projects is
related to engineers’ estimate, contract awarded substantially larger than that of D-B projects in almost
amount, final cost, construction engineering and all project size groups (Table 3). For example, there
inspection cost, and final design cost. The project are 2917 D-B-B projects versus 167 D-B projects in a
schedule data contained information related to the $2M to $10M range. There are 371 D-B-B projects ver-
date advertised, award date, construction start date sus 42 D-B projects in a $10M to $20M range. There
(notice to proceed), bid contract end date, and final are 212 D-B-B projects versus 55 D-B projects in a
contract end date (substantial completion). The final $20M to $50M range. There are 64 D-B-B projects ver-
dataset included 15,786 D-B-B and 596 D-B highway sus 31 D-B projects in the $50M to 100M range.
projects. Table 2 summarises descriptive statistics of Finally, for the project size over $100M, there are 17D-
D-B-B and D-B projects associated with each B-B and 12 D-B projects included for subse-
state DOT. quent analysis.
Table 2 shows a wide range of project costs. For
example, the contract awarded a cost of 15,786 D-B-B Data analysis
projects ranged from a minimum of $4000 to a max-
imum of $219,996,000. Likewise, the contract awarded To examine the research hypotheses, in addition to
a cost of 596 D-B projects varied from a minimum of descriptive statistics, inferential statistics were con-
$24,477 to a maximum of $242,787,000. By analyzing ducted to test the aforementioned hypotheses. A series
the contract award amount of these projects, it was of statistical testing analyses, including both parametric
recognized that many small projects (less than $2 mil- (t-test and Welch’s t-test) and non-parametric
lion), which are typically not represented the nature of (Mann–Whitney U-test), was conducted with pairwise
D-B concepts, existed in the raw data set. Most of comparisons between D-B-B and D-B projects in terms
these projects often involved a portion of work that of cost growth, schedule growth, and construction
did not present the typical features of D-B-B and D-B intensity. The independent two-sample t-test is the
project delivery. As a result, the authors took a conser- most common statistical testing method to examine
vative approach to remove all projects less than $2 the equality of mean values between two groups.
million in the data set from the aforementioned six The t-test, referred to as a special case of the one-way
DOTs. This process resulted in a total of 3888 projects, analysis of variance (ANOVA) test – a testing method to
including 3576 D-B-B and 312 D-B projects for further compare mean values of two or more data samples, is
886 P. H. D. NGUYEN ET AL.

Table 3. Classification of D-B-B and D-B projects by size.


Project cost ($)
Delivery method Project size ($) Sample size (n) Percent (%) Mean Min. Max.
D-B-B $2M to $10M 2917 81.6 4,060,029 2,002,110 9,997,604
$10M to $20M 371 10.4 13,822,798 10,008,331 19,973,500
$20M to $50M 212 5.9 30,154,280 20,050,209 49,532,515
$50M to $100M 64 1.8 70,250,171 50,340,834 99,681,924
Over $100M 12 0.3 137,728,788 103,744,995 219,996,000
D-B $2M to $10M 167 53.5 4,831,440 2,012,000 9,985,655
$10M to $20M 42 13.5 13,944,459 10,073,110 19,825,827
$20M to $50M 55 17.6 30,812,368 20,066,295 49,005,000
$50M to $100M 31 9.9 65,676,528 51,292,855 99,746,802
Over $100M 17 5.4 151,177,500 101,930,000 242,787,000

limited in testing equality of mean values between Table 4 shows a descriptive summary of the mean
only two groups. values of cost growth, schedule growth, and construc-
The use of t-test depends strictly on the satisfaction tion intensity between D-B-B and D-B projects across
of the required assumptions: (1) normal distribution of five different groups of project size. Accordingly, the
the populations and (2) equality of variances between mean value of cost growth of D-B projects is less than
two groups. If the D-B-B and D-B samples are normally that of D-B-B projects in the project sizes greater than
distributed and have equal variances, the two-sample t- $10 million while D-B-B projects produced less cost
test is selected. If the first assumption is violated (i.e. growth than that of D-B projects in the project sizes
the samples are not normally distributed), the less than $10 million. The mean value of schedule
Mann–Whitney U-test, a non-parametric test equivalent growth of D-B projects is less than that of D-B-B proj-
to t-test, is utilized to examine the difference between ects across all five different groups of project size. The
mean values of two samples. If the second assumption mean value of construction intensity of D-B projects is
is violated (i.e. the samples have unequal variances), a greater D-B-B projects in three groups, $10M to $20M,
variation of t-test, the Welch’s t-test, is utilized to com- $20M to $50M, and over $100M, while D-B-B projects
pare mean values between two samples. Prior to con- were constructed quicker than D-B projects in two
ducting the hypothesis testing process, a confidence groups, $2M to $10M and $50M to $100M.
level of 95% (a ¼ 0.05) was set to calculate the signifi-
cant value (p-value) for the subsequent testing proc-
Selection of statistical hypothesis testing methods
esses. The null hypothesis for the used tests was that
the mean values of performance metrics (i.e. cost To select appropriate statistical tests, the normality
growth, schedule growth, and construction intensity) assumption of D-B-B and D-B samples was tested
between D-B-B and D-B projects are equal within the across the five categories. The Shapiro-Wilk test was
five groups of project size. If the p-value is less than used to examine whether the samples of D-B-B and D-
0.05, the difference in mean values between D-B-B and B projects in each project size group were normally
D-B projects is statistically significant. distributed. If the p-value for these tests is less than
0.05 at 95% confidence, it is concluded that the sam-
ple is not normally distributed. On the other hand, if
Results
the p-value is larger than 0.05, it is concluded that the
Before performing inferential analysis, outliers in each sample is normally distributed. Table 5 shows the
dataset of the used project performance metrics were results of the normality tests indicating that the D-B-B
determined and removed. Specifically, the cost and D-B samples of cost growth, schedule growth,
growth, schedule growth, and construction intensity and construction intensity in the project size between
were calculated using Equations (1)–(3). For each pro- $2 million and $10 million were not normally distrib-
ject size group, the box plots were generated to visu- uted. In addition, the D-B-B samples of schedule
alize and determine the outliers for the D-B-B and D-B growth and construction intensity in the project size
projects. Figures 2–4 display the box plots of cost between $10 million and $20 million were also from a
growth, schedule growth, and construction intensity non-normal distribution.
for each project size, respectively. The figures con- For the samples that were normally distributed, the
firmed that the final dataset did not have any outliers authors used Levene’s test to examine equality of var-
and was ready for subsequent statistical test- iances (homogeneity of variances) between them across
ing processes. five project size groups. If the p-value for those variance
CONSTRUCTION MANAGEMENT AND ECONOMICS 887

Figure 2. Box plot of cost growth of D-B-B and D-B across project sizes.

Figure 3. Box plot of schedule growth of D-B-B and D-B across project sizes.

tests is less than 0.05, it is concluded that the variances across the groups of project size except for Group 5
of these two samples are not equal at 95% confidence, (project sizes over $100 million). On the other hand, the
and vice versa. Table 6 shows that the variances of variance of D-B-B and D-B samples of schedule growth
D-B-B and D-B samples of cost growth were not equal were equal across the five groups of project size.
888 P. H. D. NGUYEN ET AL.

Figure 4. Box plot of construction intensity of D-B-B and D-B across project sizes.

Table 4. Summary of cost growth, schedule growth, and con- Table 6. Levene’s test for equality of variances between D-B-
struction intensity. B and D-B samples.
Sample size (n) Mean (m) Statistical significance (p)
Performance metric Project size D-B-B D-B D-B-B D-B Construction
Cost growth $2M to $10M 2500 141 2.15% 2.79% Group Project size Cost growth Schedule growth intensity
$10M to $20M 300 36 5.86% 3.08% 1 $2M to $10M 0.00 0.60 0.96
$20M to $50M 184 45 6.90% 3.46% 2 $10M to $20M 0.00 0.37 0.47
$50M to $100M 53 29 10.37% 4.21% 3 $20M to $50M 0.00 0.73 0.03
Over $100M 10 16 11.65% 8.99% 4 $50M to $100M 0.00 0.88 0.14
Schedule growth $2M to $10M 2162 143 13.34% 11.48% 5 Over $100M 0.68 0.64 0.35
$10M to $20M 272 36 19.89% 17.53% Statistically significant at 95% confidence interval.
$20M to $50M 153 46 19.14% 16.98%
$50M to $100M 44 22 18.74% 15.19%
Over $100M 7 7 5.63% 4.81%
Construction intensity $2M to $10M 1755 146 $11,608 $9,827
$10M to $20M 269 38 $21,364 $22,736
Cost growth
$20M to $50M 133 47 $31,299 $33,549
$50M to $100M 44 28 $59,198 $57,199 Table 7 shows that D-B highway projects produced less
Over $100M 7 14 $103,552 $115,853 cost growth than D-B-B highway projects across differ-
ent groups of project size, except the project from $2
Table 5. Shapiro–Wilk normality test across five groups of million to $10 million. This result does not support
project size. Hypothesis H1 stating that D-B highway projects pro-
Statistical significance (p) duce less cost growth than D-B-B highway projects
Construction across different project sizes. In addition, for three
Cost growth Schedule growth intensity
groups of project size of $10M to $20M, $20M to $50M,
Group Project size D-B-B D-B D-B-B D-B D-B-B D-B and $50M to $100M, there was a statistically significant
1 $2M to $10M 0.00 0.00 0.00 0.00 0.00 0.00
2 $10M to $20M 0.05 0.10 0.00 0.88 0.00 0.14
difference between cost growth of D-B and D-B-B proj-
3 $20M to $50M 0.80 0.14 0.14 0.10 0.42 0.22 ects at the 95% confidence level. On the other hand,
4 $50M to $100M 0.10 0.37 0.15 0.44 0.08 0.99
5 Over $100M 0.80 0.96 0.98 0.81 0.13 0.41
the outperformance of D-B-B projects over D-B projects
Statistically significant at 95% confidence interval. for Group 1 with the project size between $2 million
and $10 million was not statistically significant.
The variance of D-B-B and D-B samples within construc- Additionally, there was no statistically significant result
tion intensity were unequal only with Group 3 (project found in the mean value difference between D-B-B and
sizes between $20 million and $50 million). D-B projects in the project size over $100 million.
CONSTRUCTION MANAGEMENT AND ECONOMICS 889

Table 7. Difference in cost growth between D-B-B and D-B Projects.


Group Project Size Mean D-B-B (mD-B-B) Mean D-B (mD-B) Test used Difference (mD-B - mD-B-B) p
1 $2M to $10M 2.15% 2.79% U-test 0.64% 0.63
2 $10M to $20M 5.86% 3.08% Welch’s test 2.78% 0.00
3 $20M to $50M 6.90% 3.46% Welch’s test 3.44% 0.00
4 $50M to $100M 10.37% 4.21% Welch’s test 6.16% 0.00
5 Over $100M 11.65% 8.99% t-test/ANOVA 2.66% 0.35
Statistically significant at 95% confidence interval.

Table 8. Difference in schedule growth between D-B-B and D-B Projects.


Group Project size Mean D-B-B (mD-B-B) Mean D-B (mD-B) Test used Difference (mD-B - mD-B-B) p
1 $2M to $10M 13.34% 11.48% U-test 1.86% 0.11
2 $10M to $20M 19.89% 17.53% U-test 2.36% 0.54
3 $20M to $50M 19.14% 16.98% t-test/ANOVA 2.16% 0.51
4 $50M to $100M 18.74% 15.19% t-test/ANOVA 3.55% 0.35
5 Over $100M 5.63% 4.81% t-test/ANOVA 0.82% 0.85

Table 9. Difference in construction intensity between D-B-B and D-B Projects.


Group Project size Mean D-B-B (mD-B-B) Mean D-B (mD-B) Test used Difference (mD-B - mD-B-B) p
1 $2M to $10M $11,608 $9827 U-test $1,781 0.00
2 $10M to $20M $21,364 $22,736 U-test $1,372 0.37
3 $20M to $50M $31,299 $33,549 Welch’s test $2,250 0.15
4 $50M to $100M $59,198 $57,199 t-test/ANOVA $1,999 0.58
5 Over $100M $103,552 $115,853 t-test/ANOVA $12,301 0.49
Statistically significant at 95% confidence interval.

The results also show that both D-B-B and D-B projects higher construction intensity than D-B-B highway proj-
produced a high-cost growth (11.65% and 8.99%, ects across different project sizes. On the other hand,
respectively) in the project size over $100 million. the mean value of construction intensity of D-B proj-
ects was higher than that of D-B-B projects in three
project size groups: $10M to $20M, $20M to $50M,
Schedule growth
and over $100M; these differences were not statistic-
Table 8 shows that D-B projects produced lower sched- ally significant at the 95% confidence level.
ule growth than D-B-B projects in all five groups of pro-
ject size ranging from $2 million to over $100 million.
Discussion
This result supports Hypothesis H2 stating that D-B high-
way projects produce less schedule growth than D-B-B The study attempted to investigate and compare the
highway projects across different project sizes. However, performance between D-B-B and D-B highway projects
no statistically significant result was found in these in terms of cost growth, schedule growth, and
groups (p > 0.05). It is noticed that, in the project size construction intensity across five different project sizes,
between $10 million and $20 million, both D-B-B and including $2M to $10M; $10M to $20M; $20M to $50M;
D-B projects were recorded with a high schedule $50M to $100M; and over $100M. These project sizes
growth of 19.89% and 17.53%, respectively while, in the only consist of highway construction projects completed
project size over $100 million, the least schedule growth within the past two decades. The data sample does not
from both D-B-B and D-B projects across the five groups include any vertical projects, such as maintenance facili-
was recorded with 5.63% and 4.81%, respectively. ties and office buildings; non-highway construction
projects are excluded for statistical analysis and discus-
sion. It is important to note that although the findings
Construction intensity
of cost growth and schedule growth in the current
Table 9 indicates that the mean value of construction study are supported by the literature, none of the previ-
intensity of D-B projects was less than that of D-B-B ous studies focussed on investigating the project
projects in two project size groups: $2M to $10M and performance between D-B and D-B-B projects based on
$50M to $100M; there was a statistically significant dif- project sizes. As one of the first attempts to conduct the
ference at the 95% confidence level was found in the rigorous analysis of performance differences between D-
group of $2M to $10M. This result does not support B-B and D-B across different project sizes, the findings
Hypothesis H3 stating that D-B highway projects have are expected to help highway agencies better
890 P. H. D. NGUYEN ET AL.

understand how D-B performs under the different sizes possible reasons for the selection of D-B for small pro-
of projects in terms of cost growth, schedule growth, ject sizes. One of the possible reasons for the selection
and construction intensity. of D-B for small project sizes is that DOTs attempt to
reduce owner’s risks (e.g. change orders) that might
burden their budget and project schedule by using D-
Difference in cost performance across
B to bear the uncertainty for the design completeness
project sizes
(Park and Kwak 2017). Additionally, the majority of the
There was a statistically significant difference in cost collected highway projects within this size include
growth between D-B-B and D-B highway projects in new construction or widening road, new construction
the project size from $10 million to $100 million. bridge and tunnel, and 3R road projects, which were
Specifically, D-B projects produced less cost growth rated as the most suitable project types for the use of
than D-B-B projects. This finding is supported by previ- D-B delivery methods by FHWA (2006). Finally, the
ous studies in the highway construction literature DOTs may attempt to use D-B regardless of project
(Warne 2005, Shrestha et al. 2007, Minchin et al. 2013, sizes because their goal is to have the final product
FHWA 2018, Nguyen et al. 2020). For example, Warne that meets end users’ needs with the opportunities to
(2005) observed better performance in D-B projects enhance innovation and potentially reduce project
compared to D-B-B projects with a 4% difference in schedules (DBIA 2018).
cost growth. Likewise, Shrestha et al. (2007) confirmed The cost performance result implies that D-B proj-
that D-B projects had 9.6% of cost growth less than D- ects might not be constantly superior to D-B-B proj-
B-B projects. With the project size greater than $10 ects as stated in the studies of Minchin et al. (2013)
million, the study found that D-B projects produced and Shrestha et al. (2011). The main reasons for the
better cost performance than D-B-B projects (Shrestha lack of consensus in comparing cost growth between
et al. 2007). The significant difference in cost perform- D-B-B and D-B projects involve limited data, opinion-
ance between D-B-B and D-B projects can be based data, and small and non-representative samples
explained by a larger number of change orders that (Park and Kwak 2017). Non-statistically significant
often occur in D-B-B projects compared to D-B proj- results were found either in the literature of project
ects. Alleman et al. (2020) found that using the D-B delivery methods or in the current study. Additionally,
delivery method provides highway agencies with sev- lack of data availability and insufficient statistical ana-
eral benefits over the use of D-B-B delivery. First, D-B lysis methods may lead to inherent biases when com-
projects often go with initial contract surpluses (i.e. paring project performance between D-B-B and D-B
the difference between the engineering estimate delivery methods.
amount and the construction contract award amount),
which then can be used to add value to the final pro-
Difference in schedule performance across
ject cost. Second, plan errors and design omissions are
project sizes
transferred to the design-builder in D-B projects that
releases the owner’s risks for potential cost overruns D-B highway projects produced less schedule growth
related to plans and design changed. Third, using the than D-B-B highway projects across all five groups of
D-B delivery method also reduces the risks of change project sizes from $2 million to over $100 million,
orders related to unforeseen conditions, which fre- which confirms the well-cited time-saving benefit of
quently occur in D-B-B projects. The involvement of the D-B delivery method. Using D-B projects can
the design-builder during design development expedite the project delivery process by minimizing
reduces the likelihood of submitting change orders design efforts of highway agencies and commencing
due to unforeseen conditions or requesting quantity the construction process early to shorten project dura-
change orders (DBIA 2018). tions (Chen et al. 2016). This time acceleration cannot
For the project size between $2 million and $10 occur in D-B-B projects due to its fundamental require-
million, D-B-B projects produced a lower cost growth ment of 100% design completeness. Another possible
than D-B projects; however, no statistically significant reason is due to the small number of change orders
result was found. The better cost performance of often recorded in D-B projects (Alleman et al. 2020).
D-B-B projects compared to D-B projects in the project Minimizing the number of change orders can help
size of $2M to $10M can be explained by the fact that avoid potential delays and unexpected cost overruns.
small projects are typically less complex and lack The other reason might relate to the high construction
opportunities for industry innovation. There are some intensity of D-B projects, which is recognized as one
CONSTRUCTION MANAGEMENT AND ECONOMICS 891

of the benefits of alternative contracting methods to with results found from Antoine et al. (2019), in which
achieve a faster rate of project delivery (Konchar and D-B projects were found to have higher construction
Sanvido 1998). intensity than that of D-B-B projects. It is important to
Although no statistically significant differences were note that Antoine et al. (2019) utilized a much smaller
found in schedule performance, the results of D-B proj- sample size (n ¼ 20) compared to the current study
ects facilitating the schedule acceleration over D-B-B (n ¼ 1901). This may explain why there are only partly
projects concur with previous studies in the highway consistent findings on the construction intensity
delivery literature (Ellis et al. 2007, Shrestha et al. 2007, between D-B-B and D-B from the current study and
Minchin et al. 2013, FHWA 2018). For instance, Shrestha the results found by Antoine et al. (2019). This out-
et al. (2007) produced the same results, but inferential come may warrant further investigations because facil-
testing results did not show any statistically significant itating a faster rate of project delivery is a benefit of
difference. Similarly, Minchin et al. (2013) using non- alternative contracting methods, including D-B, over
parametric statistical testing of 50 D-B and D-B-B high- the traditional D-B-B delivery method (Konchar and
way projects indicated that D-B projects lessened D-B-B Sanvido 1998, Sullivan et al. 2017). The construction
by 2.8% in the scheduled growth with no statistically intensity factor also utilizes the ratio of an amount of
significant results. One potential reason is due to the investment over the duration, which results in the nor-
lack of a sufficient amount of D-B highway projects malizing effect to better compare the performance of
compared to D-B-B projects, which are widely found in different delivery methods (Antoine et al. 2019).
the current project delivery literature. It is also noted In summary, the current study is the first attempt to
that collecting identical highway projects delivered by empirically compare D-B-B and D-B performance in
D-B-B and D-B is challenging, if not impossible, due to highways based on the project size although there is a
many reasons, such as project locations, internal and wide range of studies examining the performance of D-
external features, and different project characteristics. B-B and D-B projects. The finding from the current study
The inferential comparisons of project performance may provide more insights into how D-B-B and D-B per-
between these two delivery methods should be form with regard to varying project sizes. Specifically,
enhanced with collecting more empirical data of com- the results of the current study help explain the existing
pleted highway construction projects. contradictions in the literature regarding whether the D-
B delivery method provides better project performance
(cost growth, schedule growth, and construction inten-
Difference in construction intensity across
sity) than the D-B-B delivery method in the highway
project sizes
sector. Regarding cost growth, although Shrestha et al.
There was a statistically significant difference at the (2011) and Minchin et al. (2013) concluded that using
95% confidence level in construction intensity D-B-B in highway projects can end up with better cost
between D-B-B and D-B highway projects in the performance compared to the use of D-B, the current
project size from $2 million to $10 million; as a result, study only supports these conclusions with the project
D-B-B had a faster work pace than D-B. A possible rea- size from $2 million to $10 million. Conversely, D-B proj-
son for this finding is that for small projects there ects produced better cost performance than D-B-B proj-
exists more competition of D-B-B than D-B. The other ects in the project size from $10 million to $100 million
reason may relate to the measurement approach of with a statistically significant result at the 95% confi-
construction intensity in small D-B projects for which dence level. Regarding schedule growth, the current
the start date of the contract – including the builder’s study supports the conclusion from Tran et al. (2017),
design time after the award, or the start date of the FHWA (2018), and (Alleman et al. 2020) that D-B proj-
construction process is not as clear as large projects. ects produced less schedule growth than D-B-B projects
Similarly, for project sizes of $50M to $100M, D-B-B regardless of project size. The current study also shows
projects were found to have higher construction inten- that D-B projects had a higher construction intensity
sity than that of D-B projects. However, no statistically than D-B-B projects. Finally, the findings of the current
significant difference was found. study regarding the cost growth, schedule growth, and
For project sizes (from $10 million to $50 million) construction intensity between D-B-B and D-B highway
and over $100M, the findings show that D-B projects construction projects can be used as a reference in add-
were found to have higher construction intensity than ition to pertinent factors, such as project type, project
that of D-B-B projects. No statistically significant differ- goals/objectives, project complexity, duration, project
ence was found. These findings are partly consistent risk profiles, staff experience and availability, regulations,
892 P. H. D. NGUYEN ET AL.

and project cost, during the selection process of project that although the data collected from these six state
delivery methods. DOTs is sufficient for analyzing the cost and schedule
performance of D-B-B and D-B, it is expected that ana-
lyzing more data from other states and countries will
Conclusion
improve the validity of the findings. Second, only
Comparisons between performance of D-B and D-B-B highway project data were collected and analyzed.
projects have been widely documented in highway Interpreting or generalizing the performance results
construction. There still existed an argument regarding requires examining external factors for different char-
the common perception that D-B outperforms D-B-B acteristics of highway construction. Future research
in terms of cost and schedule performance. In add- may extend the findings by analyzing the non-
ition, previous studies are still a shortfall of exploring highway project data such as building and industrial
differences in D-B-B and D-B performance across vari- projects. Third, only D-B and D-B-B delivery methods
ous project sizes. The current study empirically investi- were considered because of limited data available. It is
gated the statistical differences between cost growth, suggested that other main delivery methods including
schedule growth, and construction intensity of D-B-B construction manager/general contractor or public-
and D-B projects across different project sizes ranging private partnership should be included in future
from $2 million to over $100 million. A series of statis- studies. Finally, it is important to note that the per-
tical hypothesis testing approaches were conducted to formance of D-B-B and D-B depends on many factors
analyze cost growth, schedule growth, and construc- such as project types, the scope of work, project com-
tion intensity of 3576 D-B-B and 312 D-B highway plexity, and delivery risks, and their interrelationships.
projects collected from six state DOTs. As mentioned in the literature review, there is a wide
D-B projects tend to have better cost and schedule range of studies focussing on comparing the D-B-B
performance than D-B-B projects in large project sizes and D-B performance based on these aforementioned
from $10 million to over $100 million. The results are in factors. To address the knowledge gap, the current
line with previous studies investigating the differences study’s scope is to investigate the difference in per-
in performance between D-B-B and D-B projects where formance between D-B-B and D-B highway construc-
D-B performs better than D-B-B in large project sizes tion projects based on only project size. Future work
(over $10 million) in terms of cost growth and schedule may integrate the project size with other factors to
growth (Sullivan et al. 2017, Tran et al. 2017). develop more comprehensive decision-making models
Specifically, the results show that in the project size for the selection of project delivery methods in high-
from $10 million to $100 million, D-B projects had a way construction.
statistically significant less cost growth than D-B-B proj-
ects at the 95% of confidence level. However, on aver-
Acknowledgements
age (no statistical significance), D-B-B highway projects
produced less cost growth than D-B highway projects in The authors would like to sincerely thank the members of
the Florida DOT, Indiana DOT, North Carolina DOT, Ohio
the project size ranging from $2 million to $10 million.
DOT, Oregon DOT, and Utah DOT for their participation in
Regarding schedule performance, D-B projects produced garnering all project data used for the study. Without their
less schedule growth than D-B-B projects across all five willingness to participate, the current study would not have
groups of different project sizes from $2 million to over been possible.
$100 million although no statistically significant result
was found. Regarding construction intensity, D-B pro-
Disclosure statement
duced more construction intensity than D-B-B in the
project size between $10 million and $50 million as well No potential conflict of interest was reported by the author(s).
as over $100 million (no statistical significance).
However, D-B-B projects had a faster work pace than D- Data availability statement
B projects in two groups of the project size: $2 million
Data generated or analyzed during the study are available
to $10 million (statistical significance at the 95% confi- from the corresponding author by request.
dence level) and $50 million to $100 million (no statis-
tical significance).
There were several limitations. First, the perform- References
ance comparison’s results were based on analyzing Alleman, D., et al., 2020. Project delivery methods’ change-
data collected from only six state DOTs. It is noted order types and magnitudes experienced in highway
CONSTRUCTION MANAGEMENT AND ECONOMICS 893

construction. Journal of legal affairs and dispute resolution Molenaar, K.R. and Songer, A.D., 1998. Model for public sec-
in engineering and construction, 12 (2), 04520006. tor design-build project selection. Journal of construction
Antoine, A., Alleman, D., and Molenaar, K., 2019. Examination engineering management, 124 (6), 467–479.
of project duration, project intensity, and timing of cost Nguyen, P., Tran, D., and Lines, B., 2020. Fuzzy set theory
certainty in highway project delivery methods. Journal of approach to classify highway project characteristics for
management in engineering, 35 (1), 04018049. delivery selection. Journal of construction engineering and
Chen, Q., et al., 2016. Time and cost performance of design- management, 146 (5), 04020044.
build projects. Journal of construction engineering and Park, H.S. and Kwak, Y.H., 2017. Design-bid-build (DBB) vs.
management, 142 (2), 1–7. design-build (DB) in the U.S. public transportation proj-
DBIA. 2018. Design-Build Authorization for Transportation 2018 ects: the choice and consequences. International journal of
[online]. Design-Build Institute of America. Available from: project management, 35 (3), 280–295.
https://dbia.org/advocacy/state/ [Accessed 10 May 2020]. Park, H., et al., 2015. Comparing project performance of
Ellis, R.D., et al., 2007. Evaluation of alternative contracting design-build and design-bid-build methods for large-sized
techniques on FDOT construction projects. Tallahassee, FL: public apartment housing projects in Korea. Journal of
Florida Department of Transportation. Asian architecture and building engineering, 14 (2), 323–330.
Ellis, R.E., Herbsman, Z.J., and Kumar, A., 1991. Evaluation of Qiang, M., et al., 2015. Factors governing construction pro-
the FDOT design/build program. Tallahassee, FL: Florida ject delivery selection: a content analysis. International
Department of Transportation. journal of project management, 33 (8), 1780–1794.
Ernzen, J. and Schexnayder, C., 2000. One company’s experience Qureshi, S. and Kang, C., 2015. Analyzing the organizational
with design/build: labor cost risk & profit potential. Journal of factors of project complexity using structural equation
construction engineering management, 126 (1), 10–14. modelling. International journal of project management, 33
FHWA. 2006. Design-build effectiveness study. Final report to (1), 165–176.
Congress as required by TEA-21, U.S. Department of Riley, D.R., Diller, B.E., and Kerr, D., 2005. Effects of delivery
Transportation, Federal Highway Administration. systems on change order size and frequency in mechan-
FHWA. 2018. Alternative contracting method performance in ical construction. Journal of construction engineering and
U.S. highway construction. (DTFH61-13-C-00024). Federal management, 131 (9), 953–962.
Highway Administration. Russell, M., et al., 2014. Causes of time buffer and duration
Goftar, V.N., El Asmar, M., and Bingham, E., 2014. A meta- variation in construction project tasks: comparison of per-
analysis of literature comparing project performance ception to reality. Journal of construction engineering and
between design-build (D-B) and design-bid-build (D-B-B) management, 140 (6), 04014016.
delivery systems. In: Construction in a global network. Shrestha, P., Connor, J., and Gibson, G.J., 2011. Performance
Atlanta, GA: CGN, 1389–1398. comparison of large design-build and design-bid-build
Gudiene, _ N., Banaitis, A., and Banaitiene,
_ N., 2013. Evaluation highway projects. Journal of construction engineering and
of critical success factors for construction projects – an management, 138 (1), 1–13.
empirical study in Lithuania. International journal of stra- Shrestha, P., et al., 2007. Benchmarking of large design-build
tegic property management, 17 (1), 21–31. highway projects: one-to-one comparison and comparison
Hale, D., et al., 2009. Empirical comparison of design/build and with design-bid-build projects. Transportation research
design/bid/build project delivery methods. Journal of con- record, 1994 (1), 17–25.
struction engineering and management, 135 (7), 579–587. Shrestha, P. and Fernane, J., 2016. Performance of design-build
Ibbs, C.W., Kwak, Y., and Odabasi, A., 2003. Project delivery sys- and design-bid-build projects for public universities. Journal
tem and project change: a quantitative analysis. Journal of of construction engineering management, 9364, 1–10.
construction engineering and management, 129 (4), 382–387. Songer, A.D. and Molenaar, K.R., 1997. Project characteristics
Konchar, M. and Sanvido, V., 1998. Comparison of US project for successful public-sector design-build. Journal of con-
delivery systems. Journal of construction engineering man- struction engineering management, 123 (1), 34–40.
agement, 124 (6), 435–444. Sullivan, J., et al., 2017. Two decades of performance compari-
Love, P., et al., 2015. Estimating construction contingency: sons for design-build, construction manager at risk, and
accommodating the potential for cost overruns in road design-bid-build: quantitative analysis of the state of know-
construction projects. Journal of infrastructure systems, 21 ledge on project cost, schedule, and quality. Journal of con-
(2), 04014035. struction engineering and management, 143 (6), 04017009.
Luo, L., et al., 2017. Construction project complexity: research Tran, D.Q., Diraviam, G., and Minchin, E.R., 2017. Performance
trends and implications. Journal of construction engineer- of highway design-bid-build and design-build projects by
ing and management, 143 (7), 04017019. work types. Journal of construction engineering and man-
Mahamid, I., 2013. Effects of project’s physical characteristics agement, 144 (2), 04017112.
on cost deviation in road construction. Journal of King Warne, T.R., 2005. Design-build contracting for highway proj-
Saud University, 25 (1), 81–88. ects: a performance assessment. South Jordan, UT: Tom
Minchin, R., et al., 2013. Comparison of cost and time per- Warne & Associates, LLC.
formance of design-build and design-bid-build delivery Yong, Y. and Mustaffa, N., 2012. Analysis of factors critical to
systems in Florida. Journal of construction engineering and construction project success in Malaysia. Engineering, con-
management, 139 (10), 1–5. struction and architectural management, 19 (5), 543–556.
Copyright of Construction Management & Economics is the property of Routledge and its
content may not be copied or emailed to multiple sites or posted to a listserv without the
copyright holder's express written permission. However, users may print, download, or email
articles for individual use.

You might also like