BONUSTECHNICALARTICLE

Evaluating Project
Performance Using
Baseline Schedules
BY DR. NOUR EL IMANE BOUHOU; ANTOINE BECHAALANI; KELLY PETTERSEN; AND DR. MARCELO AZAMBUJA

ABSTRACT indicators were evaluated including, but not capture all project conditions that might
Most construction companies use limited to, unique activity ID and names, affect project performance. Construction
deterministic planning models and historical percentage of missing logic relationships, projects may be impacted by flawed baseline
data to develop a baseline schedule. However, percentage of high activity duration, schedules that produce an unreasonable
deterministic methods may not capture all excessive use of constraints, and percentage forecasted completion date. Accordingly, the
project conditions that might affect project of lags and leads. Ultimately, this article development of a reasonable baseline schedule
performance. Construction projects may be provides guidance that can aid schedulers is a challenge for construction professionals.
impacted by flawed baseline schedules that in developing reasonable baseline schedules Some software programs, authors, and
produce unreasonable forecasted completion and focus on critical indicators to anticipate agencies use certain metrics to check the
dates. Accordingly, the development of a project performance. This article was first integrity and validity of the project schedule.
reasonable baseline schedule is a challenge presented as PS.2961 at the 2018 AACE The title “Health Check” is relatively
for construction professionals. The purpose International Conference & Expo. common for such checks and some of these
of this article is to identify and evaluate metrics are [6, p. 77]: missing predecessors,
a variety of project schedule indicators INTRODUCTION missing successors, redundant relationships,
to determine whether a correlation exists Most construction companies use inaccurate lags, dangling activities, number
between project performance and the project deterministic planning models and of constraints, missing status, out of
baseline schedule. Based on actual data from historical data to develop a baseline schedule. sequence activities, and excessive float. The
more than 40 projects, over 15 schedule However, deterministic methods may not most famous check is the DCMA 14 which

SOURCE DECEMBER 2019 21

 was developed by the U.S. Defense Contract then classified based on their delay quantum review revealed 18 quantitative Industry
Management Agency (DCMA). The DCMA into separate groups. Metrics with recommended thresholds
14 contains a 14-point assessment that is part The Industry Metrics for each project appropriate for analyzing a baseline schedule.
of quality standards for schedules and has were analyzed using a similar methodology Industry Metrics that are typically used
been adopted by several organizations and to Choi and Gonzales [2]. The process was in the evaluation of schedule updates or
software programs to assist in ensuring well- as follows: those that do not have a recommended
built plans. Metrics have been developed to threshold within the researched literature
monitor schedule and cost performance and • Collect baseline schedules for selected were excluded. Table 1 provides a list of the
to forecast schedule slippage based on actual projects. Industry Metrics, a brief description of the
project progress information [1, 4]. However, • Metric values were calculated using the Industry Metric, and the selected threshold.
there are still opportunities to investigate XER toolkit software and Excel functions. In addition to the 18 Industry Metrics,
if a reasonable forecast can be made using • All metric values were compiled into a projects were also evaluated for planned
baseline schedule data. spreadsheet (.xlsx) format. and actual project cost and duration. While
Choi and Gonzales [2] conducted • Each metric value was reviewed and project duration information was captured
a comprehensive literature review and compared to the recommended Industry for most projects based on a review of the
identified over 100 quantitative metrics Metric thresholds identified during the project schedules, the planned and actual
to correlate a project’s performance to its literature review process. cost information was not available for all
baseline schedule. Their study analyzed a • Any metric value that met the Industry projects. As a result, the cost data could not
list of 27 industry-recognized metrics to Metric recommendation was labeled as be used in the analysis as it relates to the
determine if schedules were fatally flawed or “Pass,” whereas any metric value that was baseline schedules but should be considered
represented a reasonable forecast. However, not within the Industry Metric threshold as an area for future research.
the number of schedules analyzed was was labeled as “Fail.”
very limited and additional case study • The number of “Pass” and “Fail” metrics PROJECTS COLLECTION AND
projects were required to develop effective for each project schedule were tabulated CLASSIFICATION
quantitative schedule quality metrics and and analyzed. Over 180 projects in total were collected.
thresholds [2, p. 14]. In addition, according The project information was reviewed,
to Choi and Gonzales, research is needed on METRICS SELECTION AND organized and classified. The selected
setting the proper threshold for each metric. COMPILATION projects represent various types of industries
Based on their literature review, industry Industry Metrics appearing within multiple including commercial, education, industrial,
metrics appear to be somewhat subjective, sources and publications were identified, and residential. Some projects were excluded
rule-of-thumb, or based on the experience of selected and categorized. The literature from the analysis due to:
the organizations/authors [2, p. 14]. These
research needs clearly indicate a significant
gap in the body of knowledge and served as
motivation for this study.
This article identifies and evaluates a
variety of project schedule indicators to
determine whether a correlation exists
between project performance and the project
baseline schedule. Based on actual data from
more than 40 projects, over 15 schedule
indicators were evaluated. This study also
provides insights of the correlations between
the schedule indicators and the project
schedule performance (i.e., project delays).
Ultimately, the purpose is to aid schedulers
in developing reasonable baseline schedules
and focusing on critical indicators to
anticipate project performance.

METHODOLOGY
A literature review and analysis were
performed to select and compile schedule
metrics (Industry Metrics) relevant to the
evaluation of a baseline schedule. The
referenced information was imported into
a table and rows were added when a new
metric was found. In addition, a range of
projects were collected. These projects were
TABLE 1 Industry Metrics Descriptions and Thresholds

22 DECEMBER 2019 SOURCE

 • Inadequate information and project type, The planned durations of the 46 selected SCHEDULE QUALITY METRICS
including but not limited to damage projects range from 107 to 1,439 calendar ANALYSIS RESULTS
analysis, cost estimation or standard of days. The schedules were developed between Projects were sorted by schedule
care projects. 2006 and 2015, with the projects being performance group and categorized into
• Insufficient baseline information where completed between 2009 and 2017. three groups. Nine projects were included
the project did not have a baseline or the The 46 projects were then classified into in Group 1, 11 in Group 2 and 26
baseline schedule was in PDF format. three groups based on the average percentage projects in Group 3. The projects analyzed
Baseline schedules in PDF format could of delay relative to the planned project were also categorized by project types:
duration. This was calculated by taking the
not be analyzed due to the inability to view industrial, commercial, infrastructure,
logic relationships, resource assignments,proportion of the total project delay over the government, residential, education,
total planned project duration. The three
project calendars and other schedule details. healthcare, hospitality. The relationship
• Insufficient project schedule information groups were determined as follows: between project delays and project types
where the project was not completed. was examined to understand if there is an
Projects that were not completed could • Group 1 was any project that experienced observed trend in project delays based on
not be analyzed due to the inability to less than or equal to 10% schedule delay. the project types. Figure 1 describes the
quantify the final delay or gain incurred • Group 2 was any project that distribution of percent project delays by
on the project. experienced greater than 10% schedule different project types. (Figures 1 and 2
delay, but less than or equal to 30% were created using the software RStudio,
In total, 46 projects were selected based schedule delay. Version 3.2.2).
on the criteria that a baseline schedule was • Group 3 was any project that experienced Figure 1 indicates that, on average
available in P6 or MS Project format and greater than 30% project schedule delay. for projects studied residential projects
had completion dates. The selected 46 incurred the highest project delay
projects correspond to the construction The 9 projects in Group 1 had an percentage (roughly 70%). Results also
industry sectors, as shown in Table 2. average of 3.81% schedule delay. The show that infrastructure projects had the
11 projects in Group 2 had an lowest delay amounts, compared to other
average of 20.17% schedule delay. project types, with an average project delay
The 26 projects in Group 3 had percentage of roughly 11%.
an average of 68.13% schedule Schedule performance metrics were
delay. The minimum schedule then analyzed for the individual projects.
delay experienced by all projects Projects that meet the industry thresholds
was 0%, or an on-time project are coded as “PASS” and those that do not
delivery, and the maximum are identified as “FAIL.” Table 3 is a screen
schedule delay experienced by all capture (only the first 8 metrics are shown)
TABLE 2 Number of Projects per Construction projects was 160.75%. Of the 46 of the industry threshold analysis for all
Industry Sector projects analyzed, two projects projects included in this analysis.
were delivered on-time and five Table 4 summarizes the schedule
projects experienced a schedule quality and metrics analysis results for
delay greater than 100%. the three groups.

FIGURE 1 (Above) Distribution of Percent Project

Delay by Project Types. Percent Delay is Equal to the
Total Project Delay Over the Project Contract Duration.

TABLE 3 (Right) Screen Caption of the Results of

Some Industry Threshold Analysis. Detailed Schedule
Quality Metrics Analysis Results are Described in
Appendix 1.

SOURCE DECEMBER 2019 23

 Table 4 shows that on average, among
all 18 schedule quality measurements,
61% of the projects in Group 1 passed the
schedule quality thresholds, 52% in Group 2
passed and 51% in Group 3 passed. Results
also show that the percentage of projects
with unique activity names decreases for
projects with higher delay, indicating
that the formatting of activity names and
representation of how the project scope is
translated into the schedule might impact
how the delay is captured in the schedules.
This finding could also be applicable to
activity codes, Work Breakdown Structure
(WBS) and reference. Table 4 also indicates
that the projects that include milestones
missing predecessors or successors incurred
higher delays. Finally, the data shows that
the incurred delays increase for projects
where the percent missing predecessors and
successors is higher than 5%. These results TABLE 4 Summary of Schedule Quality Analysis Results. Detailed Schedule Quality
seem to be in alignment with some of the Metrics Analysis Results are Described in Appendix 1.
DCMA 14-Point Assessment particularly the
missing logic ties. Furthermore, additional
indicators not referenced in the DCMA
14-Point Assessment seem to be correlated
with delays, including the percentages of
unique activity names, unique activity codes
and Work Breakdown Structure (WBS).
The relationship between the schedule
performance (i.e., project delay) and the
baseline schedule quality was further
analyzed using the quantitative correlation
coefficients. The authors evaluated the
“Pearson,” “Spearman” and “Kendall”
correlation coefficients and the statistical
significance of these correlations [3, 5].
In a statistical context, a correlation is a
bivariate analysis measuring the strength of
association between two variables and the
direction of the relation. The correlation
coefficient varies between -1 and +1. A
correlation coefficient closer to 0 means that
the relationship between the two variables is
weak. The “Pearson” coefficient measures the TABLE 5 Correlation Coefficients between Schedule Performance and Baseline Quality
relationship degree between linearly related Metrics
variables. The “Spearman” and “Kendall”
coefficients are non-parametric tests that Table 5 shows that the number smaller the delay percentage. However,
assess the degree of association and strength of activities with unique activity IDs, results indicate that the percent of activities
of dependence between two variables, milestones missing predecessors or successors, on the critical path is directly correlated to
respectively. Table 5 describes the results of activities on the critical path, unresourced the percent project delays, which means that
the correlation analysis between the schedule tasks, and percentage of activities with the higher the percentage of activities on the
performance and baseline quality metrics. lags are correlated to the project schedule critical path, the higher the project delays.
The “p-value” provides an indication on performance (i.e., amount of project delays). Some of the above discrete correlation
the statistical significance of the correlation For example, Table 4 shows that statistically results described in Table 5 are also
results. The correlation coefficients that the percentage of lags in a schedule is illustrated in Figure 2, to show the
were interpreted as statistically significant inversely correlated to the percent project magnitude, statistical significance and
have a p-value that is lower than 0.2 (also delays, which indicates that the higher the direction of the correlation between specific
highlighted in gray in Table 5). percentage of lags used in a schedule, the schedule metrics and performance.

24 DECEMBER 2019 SOURCE

 quantum in order to analyze the 2. Choi, J. and Gonzales, A. (2014)
correlation between Industry Evaluation of Baseline Schedule
Metrics identified through the Metrics for Successful Project
literature review and project Schedule Performance. 2014 AACE
delays. 18 Industry Metrics International Transactions. AACE®
were ultimately selected and International. Pages 1-16.
compared across 46 project 3. Chok, Nian Shong (2010) Pearson’s
baseline schedules. Versus Spearman’s and Kendall’s
The results show that some Correlation Coefficients for Continuous
Industry Metrics can be associated Data. Master’s Thesis, University of
with overall project schedule Pittsburgh.
performance. The number of 4. Grau, D. and Abbaszadegan, A.
milestones missing predecessors or (2015) Impact of Real-Time Project
successors, activities on the critical Control on Capital Project Cost and
path, unresourced tasks, and Schedule Performance. Organization,
FIGURE 2 Correlation Matrix Between Percent Project percentage of activities with lags Technology and Management in
Delay and Most Significant Schedule Quality Metrics are correlated to the percentage Construction: An International
of project incurred delays. This Journal, 7 (2), Pages 1289-1294.
Positive correlations are displayed in finding further validates the set 5. Hauke, J. and Kossowski, T. (2011)
Figure 2 in blue and negative correlations of health checks and guidelines described Comparison of Values of Pearson’s and
in red color. Color intensity and the size of in the DCMA 14-Point Assessment. Other Spearman’s Correlation Coefficients on
the circle are proportional to the correlation indicators such as the number of activities the Same Sets of Data. Quaestiones
coefficients. Figure 2 shows that the project with unique activity IDs, unique activity Geograophicae 30(2).
type, number of unique activity ID’s, percent codes and Work Breakdown Structure (WBS) 6. Mubarak, S. (2015) Construction
of activities on the critical path, unique are correlated to the percentage of project Project Scheduling and Control.
activity name and percent activities with incurred delays but are not included in the John Wiley & Sons, Inc. 3rd Edition.
a high float are the metrics that have the DCMA 14-Point Assessment. In addition,
most significant correlation with the percent the study identified that correlations exist ABOUT THE AUTHORS
project delay. For example, the uniqueness between the Industry Metrics themselves, Dr. Nour El Imane
of activity IDs in the schedule is inversely regardless of the delay incurred. When looking Bouhou, is with the
correlated to project delays, indicating at the sample data categorically, based on Spire Consulting
that the projects with a higher amount of construction industry type, the results indicate Group. She can
unique activity IDs had lower project delays. that residential projects incur the highest delay be contacted by
However, percent of activities on the project percentages whereas infrastructure projects sending email to:
critical path is directly correlated to delays, exhibit lower delay amounts. nourb@spirecg.com
which means that projects with a higher Further research is recommended to
percent of activities being on the critical analyze additional factors that can be Antoine Bechaalani,
path incurred more delays. addressed in a baseline schedule and impact is with the Spire
In addition, Figure 2 depicts relationships project schedule performance. Some of these Consulting Group.
between the various Industry Metrics. For factors may include the amount of weather He can be contacted
instance, baseline schedules that contain days planned and incurred, the number and by sending email to:
activities with a significant number of timing of change orders, design submittals, antoineb@
constraints tend to also have many activities the number of subcontractors involved in the spirecg.com
that are lacking predecessor or successor project, the construction delivery method that
relationships. In addition, the number of was selected, and many others. In addition, Kelly Pettersen, is
lags has a strong, direct correlation with the additional analysis could be performed that with the Spire
number of milestones missing predecessor include a larger sample of case study projects Consulting Group.
or successor relationships. This means with a more complete data set. A project’s She can be contacted
that projects with a higher number of lags contract value and actual cost is an example by sending email to:
typically also had more milestones lacking of data that the authors were not able to fully kellyp@spirecg.com
predecessors or successors. capture at this time but could aid in future
understanding of project performance. Dr. Marcelo Azambuja,
CONCLUSION is with the Spire
In this study, the authors performed a REFERENCES Consulting Group.
literature review to select existing schedule 1. Arcuri, F. (2007) A Development of He can be contacted
metrics related to the evaluation of the Performance Metrics for Forecasting by sending email to:
quality of a project baseline schedule. Schedule Slippage. Master Thesis, marceloa@
Projects included in this study were selected Virginia Polytechnic Institute and State spirecg.com
and were classified based on their delay University, 171 pages.

SOURCE DECEMBER 2019 25