Professional Documents
Culture Documents
Analysis of The Higher Order Partial Correlation PDF
Analysis of The Higher Order Partial Correlation PDF
Abstract: A fast-track project delivery strategy is designed to leverage the ability to execute the design, procurement, and construction
phases simultaneously to substantially reduce the overall project duration. The successful execution of the design phase in fast-track
projects is especially challenging because the concurrent execution of design and construction interferes with the inherently iterative
nature of the design process. In this research, it was hypothesized that a correlation exists between the use of some of the Construction
Industry Institute (CII) best practices and the performance of design phase in fast-track projects. Analysis of the performance of the
design phase in 31 fast-track projects revealed a statistically significant correlation between performance of the design phase and front-
end planning, alignment, constructability, and change management. The results of quantitative analysis are supported by qualitative
analysis of data, opinion of experts from industry, and academia. The projects analyzed in the study were executed by CII member
companies with experience and knowledge of using the best practices effectively. DOI: 10.1061/(ASCE)CO.1943-7862.0000487.
© 2012 American Society of Civil Engineers.
CE Database subject headings: Fast-track construction; Project management; Constructability; Alignment; Best Management Practice.
Author keywords: Fast track; Preproject planning; Constructability; Alignment.
(Hamilton and Gibson 1996; Dumont et al. 1997). The overlap be- minimizing the risks associated with overlapping design activities.
tween design, construction, and procurement processes introduces This framework involves identifying the evolution and sensitivity
an additional element of uncertainty, which results in the formation in design activities. The evolution of a design activity is on the basis
of additional deleterious iterative cycles caused by errors resulting of the level of design optimization, constraint satisfaction, external
from a lack of time to ensure accuracy of design and changes re- information exchange, and standardization. The sensitivity of an
sulting from inadequate front-end planning, which are detrimental activity is determined by evaluating activity constraints, input
to performance of the project (Lee et al. 2003). variables, and the level of design integration. The optimal strategy
In fast-track projects, designers have to make decisions on the involves identifying opportunities to overlap design activities with
basis of incomplete information, and the design is implemented at fast evolution, with downstream activities with less sensitivity.
the site rapidly. It is difficult to optimize the design because of the Elvin (2003), in a practical approach, outlined practices for
lack of time and accurate information. The designers are forced to managing design in fast-track project environment on the basis
over design and ensure that adequate capacity is available. For ex- of interviews with experts. These practices for successful manage-
ample, foundations and structural systems are constructed in earlier ment of design, using conceptualization of the design process as a
phases of the project, but necessary information about loading is
flow of information, include building a shared language between
not always available. In such cases, design cannot be optimized,
participants, enhancing the iterations and feedback processes, en-
and foundations and structural system are typically designed for
suring early downstream input in the design process, using a struc-
a higher load-bearing capacity. On the other hand, lack of informa-
tured team building program, creating flexible project organization,
tion can also result in undersized systems that have to subsequently
colocating the design and construction teams, and synchronizing
be reworked and fixed at additional cost. The lack of adequate de-
the workflow planning.
sign information can lead to rework in fast-track projects.
In a recent survey conducted by the the FMI Corporation and the
Construction Management Association of America (CMAA),
nearly 50% of owners reported that construction documents pre- Study Motivation
pared by the design team that are presented to the construction team
at the beginning of a project are lacking in significant information; The performance of design in a fast-track industrial project depends
15% said that construction documents are, “insufficient with major on the collaboration and participation of multiple stakeholders in
information needed;” and 5% say these documents are, “inadequate the flow of information throughout the pre project planning, design,
with major information needed” (FMI Corporation and CMAA procurement, and construction phases. Consequently, it is impor-
2004). The project team is often advised to anticipate, accept, and tant to study the entire life cycle of a project to identify the practices
budget for changes and rework for a fast-track project to be suc- that need special attention to ensure success of design in an indus-
cessful (Williams 1995). trial project executed using the fast-track project delivery strategy.
Over past two decades, the Construction Industry Institute (CII)
has, through rigorous research, identified best practices that can
improve the performance of capital projects. These best practices
Managing Design Process in Fast-Track Projects
cover various aspects of project execution, including front-end
The design process can be conceptually thought of as the flow of planning, design optimization, procurement and materials manage-
information from various stakeholders to the designers, and vice ment, construction, facility start-up and operations, human resour-
versa, during the preproject planning, design, procurement, and ces management, business and project processes, project controls,
construction phases. The design of a large industrial project in- contracts and risk management, safety, and more. Although these
volves a complex process of information flow across engineering practices were identified and developed with a focus on their effect
disciplines and across organizational boundaries and is inherently on overall project performance, this research was initiated to iden-
iterative. The feedback from various stakeholders is received and tify the practices that play an important role in the success of the
addressed in a compressed schedule. This results in many prob- design phase through rigorous quantitative analysis.
lems, including increased communication of nonessential informa- The primary objective of the study was to identify the industry
tion to stakeholders, information overload, delay in information standard best practices, which, when used appropriately, can im-
transmission, translation difficulties between disciplines, lack of prove the performance of the design phase in fast-track industrial
time for reflection and exploration of design, and rapid transmis- projects. The CII best practices identified for analysis included
sion of errors (Elvin 2003). practices that dealt with planning before design (front end
The significant overlap between the design and construction planning), flow of information from downstream stakeholders to
also requires that design and construction personnel cooperate early design (constructability, design for maintainability, design for
and more closely in the project. In a fractured and litigious industry, safety), collaboration between project stakeholders (partnering,
this poses a significant challenge. In the FMI Corporation/CMAA alignment), measuring the performance of the design phase (design
turnover during the project, offshore engineering, and more. To Most respondents to the questionnaire survey reported signifi-
correctly identify the effect of the use of best practices on the cant experience in executing projects using a fast-track execution
performance of the design phase, the effect of these external factors strategy. Some of the respondents commented that fast-track project
on the performance of design has to be isolated and eliminated. execution has become a standard operating practice in their organi-
The hypothesis was structured as follows: zation; 53% of respondents reported that their organization
Null hypothesis Ho∶ ρ ¼ 0 ð1Þ executes 75–100% of projects using a fast-track strategy, and 30%
of respondents reported that their organization executes 50%–75%
No correlation exists between the use of a best practice and of their projects using fast track strategy. The rest of the respond-
performance of the design phase after controlling the external ents reported that their organization executed less than 50% of proj-
factors. ects using a fast-track strategy. The projects analyzed in this study
included 13 grassroots construction projects; 13 expansion proj-
Alternate hypothesis Ha∶ ρ ≠ 0 ð2Þ ects; three retrofit, conversion-type projects; and one project
involving rehabilitation of an industrial structure damaged by
A correlation exists between the use of a best practice and
Hurricane Katrina. Specifically, these projects spanned a wide
performance of the design phase after controlling the external
spectrum of industrial construction, including oil exploration/
factors.
production, power generation, pulp and paper, continuous process,
The p-value is compared with a predetermined significance
gas distribution, coke production, industrial hygiene, automotive
level α to determine whether the null hypothesis should be rejected
manufacturing, and other industrial projects (e.g., natural gas
in favor of the alternative hypothesis. It is common in research of
processing, pipeline, central utility plant, water/wastewater, electri-
this type to set the significance level at 0.10 (Hanna et al.
cal distribution, semiconductor manufacturing, microelectronics
1997, 2002).
manufacturing, metals refining, consumer products manufacturing,
laboratory buildings). After the initial questionnaire surveys were
Data Collection received, the research team contacted some of the respondents for
clarifications and additional information. The data-collection pro-
The data-collection effort in this research study involved the cre- cess took more than 1 year to complete.
ation of a questionnaire survey through a very elaborate develop- The questionnaire survey was designed to collect information
ment process to identify the content, structure, and goals of the about a fast-track project executed by the respondent’s organiza-
data-collection effort. The questionnaire survey was developed tion. It comprised 104 questions and four sections: project charac-
over a period of 6 months, and it went through multiple revisions teristics and performance, detailed questions about specific design
and reviews by independent subject experts from the construction practices and their effect, detailed information about the use of
industry and an expert statistician from academia. The question- CII best practices in the project, and open-ended questions
naire survey was beta tested by a representative of the construction designed to elicit information about the challenges faced in man-
industry before it was sent out. Care was taken to ensure that the aging the design phase in the fast-track project environment. A sig-
collected data were from practices actually implemented in the nificant amount of data on design practices and performance was
project. In a separate section in the questionnaire survey, respond- collected in this research. In the interest of brevity, only the data
ents were explicitly asked for their opinion. To ensure that all relevant to the quantitative analysis presented in this paper are dis-
respondents had the same understanding of key terms and best cussed. Full details of the survey can be found in Salem and
practices used in the survey, a detailed glossary was included with Miller (2007).
the questionnaire survey. In addition, respondents were encouraged The primary objective of this research study was to identify best
to get more information about CII best practices from resources practices for managing the design phase in fast-track projects.
available from the CII website. The questionnaire survey was Respondents were provided with a list of best practices for man-
evaluated and approved by the Institutional Research Bureau at aging projects. To further identify the extent use of each of the best
the University of Cincinnati. practice in the project, respondents were provided with the essential
The questionnaire survey was distributed using postal mail elements of each best practice and asked if each element of the best
and electronic mail to 99 CII member companies. The CII mem- practice was implemented or not. A sample of the data collected for
bership includes contractors, design firms that are involved in one of the best practices (viz preproject planning) is included in
design, companies involved in construction of industrial facili- Appendix I.
ties, and owners. For the planned higher-order partial correlation In addition, examples of sample questions from the question-
analysis, a power of 0.80 and α ¼ 0:10, the required sample size naire survey are included in Appendix II.
was 29 fast-track industrial projects. The research team was able To test the hypotheses, it was also important to identify the ef-
to collect information about 31 fast-track industrial projects in a fect of factors that were a part of the project environment but ex-
period of 6 months. Respondents to the survey included owners, ternal to the use of the best practices. Respondents of the
ensuring the success of design in fast-track projects. tween the criterion variable y and the predictor variable x1 when
the effect of variable of x2 on Y has been removed can be calculated
by the following formula:
Data Normalization
r yx1 ðryx2 × rx1 x2 Þ
The metrics used for identification of the performance of the design r yx1 :x2 ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi ð3Þ
ð1 r 2yx2 Þ ð1 r 2x1 x2 Þ
phase in fast-track projects were design cost performance and de-
sign schedule performance. The metrics for measuring the perfor-
mance of the design phase were normalized to a scale in which a where ryx1 :x2 = coefficient of partial correlation between y and x1
score of 100 was established as the baseline value. If the design in after controlling for x2 ; r yx1 = coefficient of correlation between
y and x1 ; r yx2 = coefficient of correlation between y and x2 ; and
the project was completed on budget, a value of 100 was assigned.
r x1 x2 = coefficient of correlation between x1 and x2 (Blalock 1961)
If the design was completed 5% over budget, a value of 105 was
When the analysis has one criterion variable and two predictor
assigned. Similarly, if the design was completed 5% under budget, variables, the effect of one predictor variable can be eliminated
a value of 95 was assigned. The design schedule performance was when calculating the correlation. This type of correlation is known
similarly normalized to a baseline value of 100. as first-order partial correlation. In higher-order partial correlation
analysis, it is possible to eliminate the effect of more than two var-
Extent of Use of Best Practices iables (Tacq 1997).
Pearson’s product moment correlation coefficient is one of the
One of the challenging aspects of identifying the correlation be- most commonly used correlation coefficients. The assumptions
tween performance of the design phase and use of best practice made for calculating Pearson’s correlation coefficient include linear
was to quantify the extent of use of best practices in a project. relationship between the variables, continuous random variables,
To measure the degree of implementation of a particular best prac- and bivariate normal distribution. The collected data do not meet
tice in the project, respondents were provided with a list of ele- these requirements. Only one criterion variable (i.e., cost or sched-
ments of every best practice. The extent of implementation of ule performance metric) is continuous. The other variables (i.e.,
the best practice was then assessed as the number of elements best practice implementation index, external effect index) are dis-
of each best practice used in the project using the best practice im- tributed discretely and do not follow bivariate normal distribution.
plementation index. For example, front-end planning was broken Pearson’s partial correlation is sensitive to violation of the multi-
down in seven components. If only three of the seven elements variate normality assumption. When the relationship is not linear,
were implemented, a value of 3∕7 (0.43) was assigned to the best Pearson’s partial correlation can be misleading (Korn 1984). On
practice implementation index for the project. If all seven elements account of these reasons, the Spearman’s rank correlation coeffi-
were implemented, a value of 7∕7 (1.0) was assigned to the best cient, which is nonparametric, is used for correlation analysis.
practice implementation index for the project. Thus, the value of The p-value for Spearman’s partial correlations is computed
by treating the t statistic as coming from a t distribution with
this metric can range between 0–1. The assumptions for the quan-
(n k 2) degrees of freedom
tification process were:
1. All elements of each best practice are of equal importance pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
r× nk2
in achieving the intended enhancement of project performance; t¼ pffiffiffiffiffiffiffiffiffiffiffiffiffi ð4Þ
and 1 r2
2. Each element of the best practice was implemented
where n = total number of variables; r = partial correlation coef-
completely.
ficient; and k = number of variables being controlled.
In this study, the statistical analysis was conducted using a
Effect of External Factors SAS software package. The SAS PROC CORR procedure was
Performance of the design phase is affected positively and nega- used to calculate the partial correlation coefficients and to conduct
tively by a number of external factors that are a part of the project hypothesis testing. To derive Spearman’s partial rank-order corre-
lation, PROC CORR applies the Cholesky decomposition algo-
environment but are external to the use of the best practices. The
rithm to Spearman’s rank-order correlation matrix (SAS Institute
effect of these factors must be considered and accounted for while
2010).
establishing correlation between performance of the design phase
and use of best practices. The effect of each external element was
evaluated by the respondent on a Likert scale ranging from 1–5: Results of Hypothesis Testing
1 = “Highly Negative,” 2 = “Negative,” 3 = “As Planned,” 4 = Nine best practices and proposed best practices, including align-
“Positive,” and 5 = “Highly Positive.” ment, change management, constructability, design effectiveness,
all participants are aligned with the project goals, as defined in projects. Constructability analysis assumes increased importance
the project scope definition. Alignment between various stakehold- in fast-track projects that involve addition/expansion of an existing
ers can be improved by ensuring that stakeholders are appropriately facility. The importance of constructability is increased because
represented on the project team; project leadership is defined, ef- construction has to be carried out in congested areas very close
fective; and accountable; priority between cost, schedule, and the to the operational facility. This puts a significant constraint on
required project features is clear; communication within the team the techniques that can be used in the construction process. In such
and with stakeholders is open and effective and the preproject plan- cases, design of the facility is driven by the requirements of
ning process includes sufficient funding, schedule, and scope to construction, and constructability reviews assume increased
meet objectives. importance.
Did your organization use this industry Please indicate if your organization used any elements Please discuss the reasons for not adopting this
best practice in the project? of the corresponding best practice. (Y = Yes, N = No) best practice in the project. Would the design have
benefited by adopting this best practice? How?
Preproject planning: Preproject planning includes putting together the project team, selecting technology, selecting a project site, developing the project scope,
and developing project alternatives. Preproject planning is often perceived as synonymous with front-end loading, front-end planning, feasibility analysis, and
conceptual planning.
__YES __NO Preproject planning team was established. Y or N
Preproject planning effort was adequately funded. Y or N
Preproject planning plan was developed and used. Y or N
Downloaded from ascelibrary.org by Universidad Militar Nueva Granada on 08/16/20. Copyright ASCE. For personal use only; all rights reserved.
Appendix II. Example of Survey Questions Elvin, G. (2003). “Proven practices in design-build and fast track.” ASCE
Conf. on Architectural Engineering, ASCE, Reston, VA.
Was the design in this project completed ahead of schedule/on FMI Corporation and Construction Management Association of America
schedule/behind schedule? (CMAA). (2004). FMI/CMAA survey of owners, FMI, Raleigh, NC.
Griffith, A., and Gibson, G. (2001). “Alignment during preproject plan-
_On Schedule ning.” J. Manage. Eng., 17(2), 69–76.
_Ahead of schedule By how many weeks? _Weeks Griffith, A., Gibson, G., Hamilton, M., Tortora, A., and Wilson, C. (1999).
_Behind the schedule By how many weeks? _Weeks “Project success index for capital facility construction projects.”
J. Perform. Constr. Facil., 13(1), 39–45.
Hamilton, M., and Gibson, G. (1996). “Benchmarking preproject-planning
Acknowledgments effort.” J. Manage. Eng., 12(2), 25–33.
Hanna, A. S., Camlic, R., Peterson, P. A., and Nordheim, E. V. (2002).
The authors would like to gratefully acknowledge the funding sup- “Quantitative definition of projects impacted by change orders.”
J. Constr. Eng. Manage., 128, 57–64.
port provided by the CII for the study. The authors would also like
Hanna, A. S., Russell, J. S., and Gotzion, T. W. (1997). ‘‘An investigation
to thank the members of the CII research team RT 222 for their
into the impacts of change orders on labor efficiency in the mechanical
support and guidance throughout this study. construction industry: Phase II.’’ Technical Rep. to the Mechanical
Contracting Foundation and the Sheetmetal and Air Conditioning
National Association.
References Hanna, A. S., Russell, J. S., Gotzion, T. W., and Nordheim, E. V. (1999a).
“Impact of change orders on labor efficiency for mechanical construc-
Arditi, A., Elhassan, A., and Toklu, C. (2002). “Constructability analysis in
tion.” J. Constr. Eng. Manage., 125(3), 176–184.
the design firm.” J. Constr. Eng. Manage., 128(2), 117–126.
ASCE Construction Management Committee of the Construction Division. Hanna, A. S., Russell, J. S., Nordheim, E. V., and Bruggink, M. J. (1999b).
(1991). “Constructability and constructability programs: White paper.” “Impact of change orders on labor efficiency for electrical construc-
J. Constr. Eng. Manage., 117(1), 67–89. tion.” J. Constr. Eng. Manage., 125(4), 224–232.
Blalock, H. (1961). Causal inferences in nonexperimental research, UNC Hester, W., Kuprenas, J., and Chang, T. (1991). Construction changes and
Press, Chapel Hill, NC. change orders: Their magnitude and impact, Construction Industry
Bogus, S. M., Diekmann, J. E., and Molenaar, K. R. (2005a). “Evaluating Institute, Austin, TX.
the consequences of overlapping dependent activities.” Proc., Con- Ireland, V. (1985). “The role of managerial actions in the cost, time and
struction Research Congress, ASCE, Reston, VA. quality performance of high-rise commercial building projects.” Constr.
Bogus, S. M., Molenaar, K. R., and Diekmann, J. E. (2005b). “Concurrent Manage. Econ., 3(1), 59–87.
engineering approach to reducing design delivery time.” J. Constr. Eng. Korn, E. L. (1984). “The ranges of limiting values of some partial corre-
Manage., 131(11), 1179–1185. lations under conditional independence.” Am. Stat., 38(1), 61–62.
Bogus, S. M., Molenaar, K. R., and Diekmann, J. E. (2006). “Strategies for Lee, S., Peña-Mora, F., and Park, M. (2003). “Quality and change manage-
overlapping dependent design activities.” Constr. Manage. Econ., ment framework for concurrent design and construction.” Proc.,
24(8), 829–837. Construction Research Congress, ASCE, Reston, VA.
Cho, C., Furman, J., and Gibson, G. (1999). “Development of project de- Prasad, B. (1996). Concurrent engineering fundamentals: Integrated prod-
velopment rating index (PDRI) for building projects.” Research Rep. uct and process organization, Prentice Hall, Upper Saddle River, NJ.
155-11, Construction Industry Institute, Austin, TX. Russell, J. S., Gugel, J. G., and Radtke, M. W. (1993). “Documented con-
Construction Industry Institute (CII). (2002). Implementation of CII best structability savings for petrochemical-facility expansion.” J. Perform.
practices: Summaries and a self assessment guide, CII, Austin, TX. Constr. Facil., 7(1), 27–45.
de la Garza, J. M., Alcantara, P., Jr., Kapoor, M., and Ramesh, P. S. (1994). Russell, J. S., Meier, J, and Moua, B. (1999). A model process for main-
“Value of concurrent engineering for A/E/C industry.” J. Manage. Eng., tainability implementation, Construction Industry Institute, Austin, TX.
10(3), 46–55. Salem, O., and Miller, R. A. (2007). “Best modern practices for design in
Dumont, R. P., Gibson, E. G., and Fish, J. R. (1997). “Scope management fast track projects.” Rep. RT-222, Construction Industry Institute,
using project definition rating index.” J. Manage. Eng., 13(5), 54–60. Austin, TX.