I Can Reach My Full Potential with a Little Help from My Friends: Understanding Construction Project Team Dynamics and Performance Through Personality Profiling

Abstract

Over six trillion dollars were spent in the global construction market in 2020 (Global Construction Market Size 2020-2030 2022). This number is both significantly larger than a decade ago, and yet a fraction of what it is projected to be in decades to come. The demand for built structures is ever increasing as the world’s population increases and its existing infrastructure ages. Accordingly, there is a significant interest in maximizing project outcomes in order to deliver the highest possible amount of value - interest that is reflected in the significant volume of research conducted in academia and throughout the construction industry regarding cost and schedule control, productivity, project delivery, and procurement. However, while that extant body of literature is indeed vast, there is a lack of research that specifically investigates the dynamics of project teams on project success in a quantitative and repeatable manner.

This thesis presents the results of a study that performed such a quantitative and repeatable examination of the project team’s qualities, dynamics, and traits on project success. In performing this analysis, two meaningful and actionable deliverables were created that are company-agnostic and can be readily applied in similar organizations. These are: 1) an antecedent team performance predictive tool; and 2) a data-backed roadmap for professional development to improve individual performance and increase project team performance. What differentiates this study from other previous efforts that may have examined or sought to examine team cohesion is its use of the Drake P31 behavioral and personality assessment tool to provide quantitative data from which conclusions could be drawn.

The purpose of the Drake P3 tool is to identify the qualities of an individual’s personality that relate to their behavioral style. It assesses degrees of variance from a midline (or average) person in the following areas: dominance, extroversion, patience, and conformity, as well as perceived energy, stress, conscientiousness, etc. These results, along with other key independent variables identified in the course of this study as potentially related to individual performance (such as the average distance an employee worked from their home, and the teams’ average experience at the company), were used to inform the analysis. To gather real-world project data from which timely and actionable analysis can be drawn, the researcher partnered with a major general contractor in the midwestern United States.

The Drake P3 profiles of over 250 employees were provided, as well as project performance data from 202 projects spanning 22 U.S. states. The projects totaled 1.72 billion dollars in final revenue and 84 million dollars in final profit. The projects provided encompassed the healthcare, industrial, institutional, power generation/transmission, and commercial market sectors; and ranged in revenue between one and 116 million dollars. Project duration ranged from as little as four days to over six years, and contracting methods included four contract types: Lump-Sum (LS), Time and Materials (T&M), Guaranteed Maximum Price (GMP), and Time and Materials with GMP (T&M-GMP). The ample dataset provided more than sufficient information for the statistical analyses performed.

Two methods of analysis were used: 1) multiple factor importance analyses that utilize t-tests to determine the importance of independent variables, and 2) linear regression models that study the effects of the independent variables on profit. The first analysis was a manner of factor importance, whereby each variable was put through an individual mean hypothesis test and split based on its mean; the split data of individual variables were then compared in the context of the overall success variable, which in this case was change in margin (CIM). A significant difference between these groups in terms of CIM implies a potentially meaningful impact of that variable on a team’s performance.

Change in margin is a metric used in this study to characterize the impact of a project team on a project’s performance by comparing the forecast margin when the project was bid to the actual margin realized at project completion. CIM is used because it is independent of the actual financial value and initial profit of the project and is more reflective of the project team’s effort while executing than other performance metrics.

The next factor importance analysis performed was the generalized unbiased interaction detection and estimation (GUIDE) program. GUIDE is statistical software that was developed at the University of Wisconsin by Dr. Wei-Yin Loh. Akin to the method used in the manner of importance model characterized previously, GUIDE splits data based on variables that show a significant difference in a key target variable (in this case, CIM). An advantage of the GUIDE method is that it can consider multiple variables and create a regression tree that further identifies differentiating values that a variable can take, expanding beyond the mean-based analysis previously described.

A parallel analysis effort was performed to facilitate the development of a predictive tool based on the project outcomes in the data. The projects were split into subsets based on revenue, duration, industry sector, and project type, and linear models were fit to each subset in turn, as well as to the 1942 projects in the dataset as a whole.

15 linear regression models were developed to capture the factors or combination of factors that impact project success, with a goal of finding model(s) that maximize the prediction power and reduce mean squared error3. Of these models, the highest adjusted R2 was .95 and the lowest adjusted R2 is -.08, with an average across the models of .30, indicating a moderate relationship between the independent variables and the CIM. In a predictive use, the models predict the CIM of a project within 2% of the actual value 42% of the time, within 5% of the actual 75% of the time, and within 10% of the actual 90% of the time; this suggests a moderate ability to predict project success with the tool.

It was in 1917 that Cyrus McCormick coined the now oft-cited proverb: if you want to go fast, go alone; if you want to go far, go together (McCormick, 1917) - a maxim which is as true today as it was then. With declining construction productivity and an aging workforce in need of new hires, it is evidently critical that project teams be correctly selected given their impact on project success. The results of this study indicate that informed team selection, alongside proper training and development, can push correctly structured teams to follow the best practices of previously successful teams within the organization and realize a more successful project.

1 The Drake P3 Behavioral and Personality Assessment is a product of Drake International 2 202 projects were collected but 8 collected projects were excluded as outliers from the analysis 3 See section 1.4 for definitions of terms