Abstract
Efficient capital project execution is a key business objective in the process industries, but existing project management research gives little direction about how team factors influence three important capital project outcomes: cost, schedule, and operability. From an extensive cross-disciplinary review of the team and project management literatures, we constructed and tested a theoretically based, five-dimensional model of organizational context, team design, team leadership, team processes, and project outcome factors. We examined the model by means of an empirical study of 56 newly completed capital projects executed by 15 Fortune 500 companies in the process industries. The results indicate the value of disaggregating project outcomes. Different bundles of team factors were found to drive project cost, schedule, and operability. Team efficacy, cross-functional teams, autonomous project team structure, and virtual office usage were the strongest predictors of project cost. Continuity of leadership, cross-functional teams, and project manager incentives were the strongest predictors of project construction speed. In contrast, clear project goals and an office design to facilitate effective communication were the main predictors of plant operability. Implications of these findings for academics and project practitioners are discussed.
Introduction
Project management research is still in its infancy (Cooke-Davies, 2002). Successful project management depends on identifying key determinants of project success. Early research identified that people management drives project success more than technical issues do (Larson & Gobeli, 1989; Pinto & Slevin, 1988). Despite this finding, there exists only a small body of research that examines the so-called soft project management, the people side of project management (Kloppenborg & Opfer, 2002). Academics (Pinto, 2002) and professional associations such as the Project Management Institute (2002) have identified the need for more in-depth examination of the relationship between team factors and project success. Increasingly, companies are also recognizing that the identification of project team success factors is crucial to achieving world-class competitiveness. This particular study grew out of an industry call for team-focused research in capital projects in the process industries (Business Round Table, 1997).
Recent academic literature has engaged in a fruitful debate (Dvir, Lipovetsky, Shenhar, & Tishler, 1998) over the nature of project success. Traditionally, project success was assessed on the triple constraint of cost, time, and performance (Kloppenborg & Opfer, 2002). A project was considered successful if it was completed within its budget estimate, within its initial scheduled time frame, and performed as it was designed to. However, this triple constraint of hard outcomes is now considered too simplistic, since it ignores important soft outcomes, such as the satisfaction of the client or intended user, and employee development and satisfaction (Hackman, 1987). Other identified dimensions of project success are longer-term business success, and learning that prepares the organization for the future (Shenhar, Levy, & Dvir, 1997). While acknowledging that project success is much broader than the triple constraint, our study has focused on project cost, schedule, and operability for reasons of methodological parsimony, and the appropriateness of these criteria to the process industries. In this particular sector, cost and schedule control, and plant operability are regarded as important measures of capital project success.
The process industries add value to materials by mixing, separating, or forming chemical reactions. These industries typically involve considerable capital investment. Despite their significant financial contribution to national economies, and their numerical strength as a group, the process industries have largely been ignored by project management researchers (Fransoo & Van Donk, 2003; Zobel & Wearne, 2000). Capital projects in the process industries involve the construction of physical plant facilities and materials processing equipment, either to produce a new product for expected profit (Rowings & Behling, 1993), or alternatively to maintain or develop operating-level capabilities. The past decade has witnessed a changing view about the role of capital project systems in a company's business. Improving its capital project system is one strategy a company can employ to gain a competitive edge in both local and global markets. Even small improvements in project management can translate into substantial economic benefits for businesses, shareholders, and the economy (Business Round Table, 1997).
Despite its importance to business, team management in capital projects in general (not just in the process industries) has been largely ignored by researchers, leaving project managers uncertain as to the important human levers of capital project success. Most project team studies have examined development projects conducted in the research and development (R&D) and new product development (NPD) contexts. Pinto and Covin's (1989) comparison of the management of capital projects and R&D projects raised initial concerns about the prevailing view that one size fits all. The proposition that critical success practices may not be similar for all project types (Dvir et al., 1998; Pinto, 2002) should be further tested in this new setting.
Our extensive study of the extant project team literature indicated four common shortcomings. The first limitation relates to the type of research methodology commonly employed. Except for a small number of studies (Belout & Gauvreau, 2004; Pinto & Slevin, 1987), there is limited quantitative research on the impact of project team management practices on project performance. Most of the literature is based on case studies or expert practitioner opinion (Kloppenborg & Opfer, 2002). The current research project set out to redress this problem by applying a systematic empirical research design to the gathering and analysis of data related to the management of human issues in projects. Second, with several notable exceptions (Pinto & Slevin, 1987), studies have tended to adopt a piecemeal approach to team-related success factors. There is a clear need to develop and test a more comprehensive project team model that prioritizes the relative importance of team-related critical success factors.
The recognition that project success is multi-dimensional raises the question whether different input factors may have different effects on different project outcomes (Denison, Hart, & Kahn, 1996). This brings us to the third shortcoming in the project literature: the common methodological practice of aggregating separate measures of project success criteria into a single, overarching measure of project success. This reductionist practice implies that success factors improve all project outcomes, masking the possibility that different success factors may drive different project outcomes. In order to develop a more fine-grained model, we chose to disaggregate project success by testing the model for individuated measures of our target project outcomes: cost, schedule, and operability.
A fourth limitation in the literature is the rarity of objective measures of project outcomes. This is related to the difficulty in comparing projects, which, by definition, are unique. Usually studies use key informant subjective perceptions of project outcomes; a practice that leads to same-source bias, and to conflicting findings depending on informant group membership (Hoegl & Gemuenden, 2002). In our study we broke with tradition. We measured project success criteria using three proven objective metrics developed by Merrow (1997) that enabled us to directly compare projects of different types, scopes, and sizes executed in different process industry sectors.
In summary, this current study set out to address the outlined deficits in the project management literature. From an extensive review of the team literature outlined below, we developed a comprehensive model of theoretically grounded project team variables and explored its explanatory power for three key project outcomes in capital projects executed in the process industries.
Literature Review
In 1998, project management in the United States (US) was an $850 billion industry, with a predicted growth rate of 20% per annum (Bounds, 1998). With this amount of investment capital at stake, cost control in capital projects is vital. However, expensive failures abound. More than fifteen percent of authorized projects run fifty percent or more over budget (Procter & Gamble, 2002). Minimizing time to market is also a key business objective. Slow project execution may cause a product to be late to market, turning a promising investment opportunity into an expensive failure. Plant technical performance is another project success criterion. Production installations in the process industries are extremely costly, so a high utilization with minimal maintenance shutdowns is necessary to maximize throughput and business returns. Despite the importance of these traditional project success criteria to gaining a competitive edge, research on minimizing costs and improving cycle time and plant operability in capital projects is scant.
In recognition that project success criteria are multi-faceted, project research has often separately measured multiple dimensions. Unfortunately, most of these studies have subsequently adopted a data reduction practice of outcome aggregation that has hindered advancement in project research. This practice implies that success factors improve all outcomes equally, masking the possibility that different success factors may actually drive different outcomes. A comparative study of self-managing (SMWTs) and traditionally managed service teams (Cohen, Ledford, & Spreitzer, 1996) demonstrated that different outcomes can have different drivers. Outcomes such as absenteeism, quality of work life, and team ratings of performance were each influenced by different team input factors.
Another difficulty in the literature involves the methodological problem of how to effectively measure project performance. Due to cultural and philosophical differences, each organization has its own way to measure and express the performance of a project team and its outcomes (Cleland, 1999). This hinders large sample research, since researchers are unable to objectively measure project outcomes in a form that enables direct comparison across projects, across companies, across sectors, and across industries (Cohen & Bailey, 1997; Yeatts & Hyten, 1998). Studies using objective measures of work outputs generally have been restricted to the case study method (see Clark & Wheelwright, 1992; Pinto & Kharbanda, 1996). Most large studies tend to rely upon key informants’ subjective ratings of project outcomes (e.g. Pinto & Slevin, 1987; Thamhain & Wilemon, 1987; Campion, Medsker, & Higgs, 1993). However, research has shown that perceived performance ratings are not well correlated, i.e., results vary according to who is asked: a project manager's ratings of performance outcomes differs from team ratings (Campion et al., 1993; Hoegl & Gemuenden, 2002). Additionally, subjective measures do not correlate well with objective measures of outcomes (Gladstein, 1984). These disturbing findings cause potential interpretation problems for a large number of team studies that have relied on opinion-based performance ratings.
In our study we set out to address this problem. We selected three proven objective metric indices of project cost, schedule, and technical performance used by over seventy Fortune 500 companies to measure project success (Merrow, 1997). This enabled us to directly compare projects of different types, scopes, and sizes, executed in different industry sectors of the process industries. These metrics originated from extensive initial research by the RAND Corporation, and have been further developed by Independent Project Analysis (IPA), a global project consultancy. A more detailed description of the metrics is outlined in the methodology section.
Our literature search also revealed an absence of any comprehensive research on soft project management and capital project outcomes. Research in capital projects has focused on hard project management. Studies have identified many hard technical and procedural drivers of project success, such as contracting strategies and contract types (Pedwell, Hartman, & Jergeas, 1996), level of project front-end definition, and standardized project delivery processes (Nemes & Lukas, 1996). Due to the lack of capital project team research, we adopted a cross-disciplinary approach and looked to research in other project types and other team types for guidance. Mindful of contingency theorists’ call to research different project contexts (Pinto & Covin, 1989), we looked for already identified team variables to test in this new, under-researched setting.
A second consideration caused us to look beyond even the project literature. “Most research on the management of projects is relatively young and still suffers from a scanty theoretical basis and a lack of concepts,” write Shenhar & Dvir (1996, p. 608). In view of this assertion, we also drew on the well-established theory of the general work team literature to further develop project management theory.
Proposed Conceptual Model
Our extensive multi-disciplinary search of both the generic work team and project literatures formed the basis for the proposed model of project team performance. Previous research has developed many models of how work team performance relates to work outcomes (Cohen et al. 1996; Gladstein, 1984; Hackman, 1987; Yeatts & Hyten, 1998). However, fewer models (see Hoegl & Gemuenden, 2002; Pinto, Pinto, & Prescott, 1993; Thamhain & Wilemon, 1987) have been developed for project teams. Project teams share a similar general definition with all work teams in that they are a group of interdependent individuals who share responsibility for specific outcomes for their organization (Sundstrom, 1999). However, project teams differ from other teams in that they perform a one-off, non-routine task, they are temporary rather than permanent, they are discontinuous and not on-going, they are usually heterogeneous and cross-functional (not uni-disciplinary), and they are highly educated specialists rather than generalists. Until Cohen & Bailey's (1997) major literature review, teams were often treated as generic, with no distinction made between team types and contexts.
From our detailed examination of the interdisciplinary literature we identified thirteen success factors (independent variables) that may influence the successful completion of projects in terms of cost, schedule, and operability (dependent variables). We then developed a holistic model that integrated this comprehensive range of team and organizational inputs. The model was theoretically derived and drew on many perspectives, including the social psychological (Hackman, 1987), human resource (Shea & Guzzo, 1987), cognitive (Bandura, 2000), and ecological (Sundstrom, De Meuse, & Futrell, 1990) approaches. The broad inclusiveness of our model enabled the simultaneous examination of the relative impact of a number of important organizational and team predictors in a way we believe has never before been done in the project literature.
Organizational Context
Project teams are interdependent within the larger organization in which they are embedded. Although early work team research focussed almost exclusively on group dynamics, researchers have now begun to recognize the critical role of organizational context for integrated team performance (Cohen & Bailey, 1997). Defined as the organization's impact on teams, organizational context is conceptualized as external to the team, but internal to the organization (Hackman, 1987). Hackman asserts that it is more effective to create the ideal conditions to enable teams to work efficiently, than to attempt to manage internal team processes and behaviours. Two organizational success factors are common to both the project and the general work team literature: clear project goals, and senior management support of the project.
Exhibit 1: Proposed model of team-related factors that impact project cost, schedule, and operability
Exhibit 1 illustrates our framework for team-related capital project success. The thirteen identified team success factors can be theoretically classified into four dimensions: organizational context, team design, project leadership, and team processes. Our proposed model includes the direct effects on project outcomes of these four key team and organizational dimensions. We now explore these dimensions in detail.
Clear Goals
Clear goals are important to orientating teams towards common objectives (Kirkman & Rosen, 1999). Shared, clearly enunciated strategic goals that are aligned to the organizational mission improve project performance (Pinto & Slevin, 1987). Superordinate goals foster cooperation since they can only be achieved through collaboration (Campion et al., 1993). Senior management's setting of clear project goals in collaboration with the project team has been shown to reduce cycle time in NPD projects (Kessler & Chakrabarti, 1999; Lynn, Skov, & Abel, 1999; Zirger & Hartley, 1996).
Senior Management Support
Senior management support is also said to play a key role in project team success by aligning the project support systems with the project team (Pinto & Slevin, 1987; Ranney & Deck, 1995). However, findings on the effects of senior management support are mixed. Many highly functioning teams have still failed due to insufficient resources, or to senior management interference (Sundstrom, 1999). Other studies have found that increased senior management support either has no cycle accelerating effect (Zirger & Hartley, 1996), or that some forms (e.g., process control) lengthen project time cycle (Bonner, Ruekert, & Walker, 2002).
For the dimension of organizational context, we hypothesized that clear goals would be significantly and positively related to project cost, schedule, and operability. In view of the conflicting literature, we formulated a non-directional hypothesis for senior management support, predicting that it would be significantly related to project cost, schedule, and operability.
Team Design
Cross-functional Teams
Until the mid 1990s, team design had been relatively neglected by the project team literature (Denison et al., 1996). One consistent finding concerning project team composition is that cross-functionally integrated teams with members drawn from across all necessary functional departments, such as engineering, R&D, marketing, and operations (Zirger & Hartley, 1996), contribute significantly to shortened cycle times (Griffin, 1997) and drive project profitability (Cooper, 1995). Cross-functional teams have been found to limit average project cost growth, schedule slip, and total project cycle time in capital projects (Nemes & Lukas, 1996).
Autonomous Project Team Structure
There is increasing evidence that for large or complex projects, an autonomous, full-time, and empowered core project team delivers projects on time and within budget (Bommer, De La Porte, & Higgens, 2002). The presence of a heavyweight (i.e. a fully empowered and dedicated project leader) (Clark & Wheelwright, 1992) is considered a key factor in NPD cycle reduction (Cooper & Kleinschmidt, 1994). Research results, however, are conflicting. Some studies have found that heavyweight leadership actually lengthened cycle time (Griffin, 2002).
Experience
Team member experience has been found to impact project outcomes. Firms can improve project performance by selecting more knowledgeable and experienced team members (Pinto & Slevin, 1987).
Continuity of Team Membership
Stable or ongoing team membership (Akgun & Lynn, 2002) has received little coverage in the project literature, but it is considered important since team turnover negatively affects team learning (Liang, Moreland, & Argote, 1995) and slows innovation speed (Kessler & Chakrabarti, 1999) due to team knowledge loss (Argote, 1993).
Office Design to Facilitate Effective Communication
Effective communication is one of the most frequently studied project team success factors. We decided to study the physical design of the project office to determine whether an office intentionally designed to facilitate communication is related to improved project outcomes.
Collocation
Collocation has been long considered an essential component for superior project functioning, by promoting team identity and by providing ready access to informal and task oriented communication, group cohesion, increased group responsibility, and super-ordinate goal achievement (Pinto et al., 1993). Communication increases with increased proximity. Increased distance between team members can delay decision-making and exacerbate any existing team personality difficulties (Allen, 1986). However, more recent studies have found that collocation has no effect on cycle time (Griffin, 2002; Zirger & Hartley, 1996).
Virtual Office Usage
Studies of geographically dispersed teams show that actual physical proximity is not essential provided the team can come together with ready access to each other in a virtual sense via technology and communication systems (Mankin, Cohen, & Bikson, 1997). Virtual NPD teams have been shown to make better decisions than face-to-face teams (Schmidt, Montoya-Weiss, & Massey, 2001). Therefore, the use of virtual technology is included in our model for its potential in assisting in cross-functional integration and communication.
In summary, for the dimension of team design, we predicted that the following variables would be significantly and positively related to project cost, schedule, and operability:
- Cross-functionally integrated teams formed early at project inception
- An autonomous project team structure
- Team member level of experience
- Stable team membership
- Project office design to enhance effective communication
- Virtual technology usage
For collocation, we formulated a non-directional hypothesis that collocation would be significantly related to project cost, schedule, and operability outcomes.
Team Leadership
Project Manager Continuity
Despite increasing interest in leadership in the general management literature, team leadership is still under-researched in project management. The NPD literature has shown that project manager continuity is important to accelerating development speed in a stable environment (Akgun & Lynn, 2002).
Project Manager Incentives
Another under-researched area is the use of incentives to motivate project managers. Consistent with management literature (Jaworski, 1993), it is argued (Bonner et al., 2002) that the tying of extrinsic incentives to project outcome dimensions and return on investment (ROI) goals may improve project performance.
For the dimension of team leadership, we hypothesized that project manager continuity and project manager incentives would be positively and significantly related to project cost, schedule, and operability.
Team Processes
Compared with the intense focus on internal processes in the general work team literature, scant attention has been paid to internal interaction within project teams (Cohen & Bailey, 1997). We therefore included two variables from the general team literature in our model.
Problem-solving
Effective team problem-solving ability is one internal team process shown to improve project outcomes and to accelerate speed in NPD projects (Schmidt et al., 2001).
Team Efficacy/potency
Teams that share a high belief about their collective efficacy are more committed and prepared to work hard for the team, and achieve higher productivity and task effectiveness (Campion et al., 1996). Derived from cognitive theory (Bandura, 2000), team efficacy or potency refers to a team's collective conviction that together they can successfully achieving the team's tasks (Guzzo, Yost, Campbell, & Shea, 1993).
For the dimension of team process, we hypothesized that high levels of team problem solving and strong team efficacy (potency) would be positively and significantly related to project cost, schedule, and operability.
Project Outcomes
Unlike much of the project management literature, we were able to isolate our dependent variables (project cost, project construction cycle time, and plant operability) as separate, objectively measured outcomes. We chose to measure our dependent variables quantitatively using a set of proven benchmarking tools, or indices, developed by RAND and IPA (Merrow, 1997). We measured each project's outcomes relative to the industry average outcomes for projects of that particular type. Further details of our project success criteria are given below in our methodology section. The major advantage of these indices is that they allowed normative comparisons of project outcomes. This made it possible to make a meaningful objective comparison of the cost, schedule, and operability of projects with very different scopes, on an even basis.
Research Methodology
Sample
The sample was drawn from 56 completed capital projects, ranging from small (US$270,000) to very large (US$203.25 million), executed by 15 large Fortune 500 process industry companies in the US, Canada, France, Germany, Australia, Singapore, and Korea. The level of technological innovation varied from no new technology or minor increments (87.5%) to revolutionary, i.e., an entirely new process, (12.5%). Data were collected from 252 team members, with a mean team size of 7, mean project tenure of 1.9 years, mean company tenure of 14.27 years, and average of 13 years of project experience. The average team age was between 36-45 years. 92% were male, 90% held permanent tenure, and 65% held an undergraduate degree or higher. 76.9% were employees of the owner organization.
Measures
Independent Variables: Project Inputs
Team inputs were assessed using a triangulation of multiple sources of data, as recommended by Kirkman, Tesluk, & Rosen (2001), and included both subjective and objective types of information. A set of proven, reliable, and valid measures of each identified critical team factor was compiled from the literature to form a twenty minute written questionnaire. Pinto & Slevin's (1987) proven Likert scales were used to measure several constructs. The clear goals variable was measured with the five-item project mission scale and included statements such as “the goals of the project were in line with the general goals of the organization” and “the basic goals of the project were made clear to the project team.” The five-item top management support scale measured senior management's support of the team with items such as “senior management granted us the necessary authority and supported our decisions concerning the project.” Team problem solving was measured with the five item troubleshooting scale, which included items such as “immediate action was taken when problems came to the project team's attention.” Guzzo et al.‘s (1993) team potency scale measured team efficacy. The items assessed the extent to which team members agreed or disagreed that their team had confidence in itself, believed it could be extremely good at producing high-quality work, expected to be known as a high-performing team, felt it could solve any problem, believed it could be very productive and could get a lot done when it worked hard, believed that no job was too tough, and expected to have influence.
The questionnaire was initially screened by a ten-member, transnational panel made up of accomplished capital project managers, capital project analysts, and psychologists. After modification, the instrument was piloted on five project teams and further refined. The resulting written questionnaire was completed individually by team members to yield subjective individual responses to measures of the four dimensions of team input variables.
The aggregation of individual results to produce a single team score is a commonly used technique (Campion et al., 1993; Cohen & Bailey, 1997) that rests on the assumption that such an aggregation represents the whole team/group perspective. Recent researchers (Brannick, Salas, & Prince, 1997; Yeatts & Hyten, 1998) have questioned the validity of such an assumption and recommend the whole group response technique.
To address this concern, a further set of questions was developed to elicit a team-level response during a face-to-face team interview. Where no pre-existing scales were found for particular constructs, questionnaire items were devised based on theoretical or empirical sources in the literature. Questions designed for the group interview measured team experience, cross-functional integration, project office design, virtual office usage, team and project manager continuity, autonomous project team structure, and whether the project manager's salary was linked to achieving project objectives/ROI. This procedure also provided an added richness of qualitative data not accessible with written, closed-question surveys. Objective archival data such as organizational charts, project objectives documents, task and role descriptions documents, information and communication channels documents, and interface management procedures were also collected to verify the subjective information supplied by the team.
Dependent Variables: Project Outputs
Our project success criteria are relative measures of each project's actual performance, as compared to the industry average for capital projects of that type executed in the process industries. The industry average figure has been statistically derived from IPA‘s proprietary processing projects database of over 4,000 projects of all types, executed over the past 30 years on 4 different continents, by over 70 of the world's leading processing companies in petroleum and chemical processing, consumer products, pharmaceutical, and pulp and paper industries (Merrow, 1997).
Project Cost
Project cost performance was evaluated using a single measure or index, derived by dividing the actual cost of the particular project by the industry average cost (expressed as 1.00) for a similar project of that size, type, and complexity. Total installed capital cost included all owner and contractor costs for engineering (process design, production engineering, and project management services), as well as any other project-related costs normally capitalized, such as licensing fees and initial catalyst charges. Cost also included expenditure for post-mechanical completion modifications. It did not include expenditure for basic R&D or test facilities.
Project Schedule
Project schedule performance was objectively evaluated using a single index measure. The construction index measured the phase that overlaps the completion of the detailed engineering phase. The phase terminates prior to start-up. For each individual project in the sample, the time taken for the construction phase (from the first foundation work, with site preparation activities excluded, through to mechanical completion) was expressed in relation to the industry average (expressed as 1.00). The main activities of this stage include procuring resources to perform the project work, executing the activities identified in the project plan, monitoring and reporting on project progress, and replanning as needed until the facility is complete for start-up (Merrow, 1997).
Plant Operability
Plant operability is a measure of the constructed facility's operational performance, or rather, its fitness for use. Operability is defined as the average percent nameplate capacity during months 7-to-12 after startup, i.e., the ratio of average production for the second 6 months of operating divided by nameplate (design rate) capacity. It was measured by IPA‘s Operability Index, which uses the “percentage of heat” and “materials balance” equations based on data from commercialization units, feedstock, and the number of process steps using new technology.
Results
As our outcome measures were normalized for project type, complexity, technology, and size, we only examined main effects in the present sample. Descriptive statistics were calculated and Pearson's r bivariate correlations were performed for all variables measured at project level (n = 56). These are presented below in Exhibit 2.
Exhibit 2: Descriptive statistics and correlation matrix
Project Cost
An examination of the correlation matrix revealed that nine of the thirteen team factors were positively correlated with project cost. Significant correlations between team factors and project cost varied in magnitude from .27 (significant at the 0.05 level, 1-tailed) for senior management support, and for the linking of the project manager's salary with achieving objectives and ROI, to .44 for autonomous project team structure and .46 (significant at the 0.01 level, 1-tailed) for team potency. The correlation matrix showed that many of the independent variables were more strongly correlated with each other than they were with the dependent variable cost. We then conducted a hierarchical regression analysis on the cost (dependent) variable, to identify the fewest independent variables necessary to predict that variable, where each independent variable predicts a substantial and independent segment of the variability in the dependent variable (Tabachnick & Fidell, 1996). The final cost efficiency model included four significant independent variables, namely team potency, cross-functional team integration, autonomous project team structure with sole responsibility to the project manager, and virtual office usage. This four-factor model accounted for a substantial 35.6% of the explained variance of project cost in our sample. Exhibit 3 shows the coefficients and significance values for that regression.
Exhibit 3: Regression models for project cost, schedule and operability
Project Schedule
Seven of the thirteen team factors were positively correlated with fast project construction cycle: team experience, cross-functional integration, project office design to maximize effective communication and integration, virtual office usage, project manager's continuity, and project manager salary linked to project objectives and ROI, along with team problem- solving ability. Significant correlations between the team factors and project construction schedule varied in magnitude from .26 (significant at the 0.05 level, 1-tailed) to .41 (significant at the 0.01 level, 1-tailed). Three other factors previously found to affect NPD cycle time (clear project goals, team continuity, and autonomous team structure) were positively, but non-significantly related to construction cycle time in this present sample. Senior management support was negatively (though non-significantly) related to project construction schedule.
We then conducted multiple regression analysis on the construction schedule dependent variable. This analysis identified a parsimonious three-factor model comprised of project manager continuity, cross-functional integration, and performance-contingent project manager incentives. As predicted, human resources were also able to explain a significant proportion of variance in project construction cycle efficiency. The most important predictor in our schedule model was the continuity of project manager variable that accounted for most variance in construction speed. In our sample, a surprisingly high proportion of projects (20%) experienced project manager turnover. This discontinuity of leadership resulted in a mean construction cycle that was 37% longer than projects with no manager turnover. Since project manager continuity has received little study in the project literature, this is an interesting new finding. Another little studied variable that emerged as prominent in our schedule model is performance-contingent project manager incentives (used in 24% of our sample). High outcome interdependence, achieved by linking project manager compensation to achieving project objectives and ROI, appears to motivate the project manager to achieve faster project schedules. As with our project cost model, cross-functional integration was also a key predictor of schedule efficiency, underlining the importance of cross-functional teams to project success. Using the more conservative adjusted R2 figure, this model still accounted for 26.9% of the explained variance of project construction cycle time (see Exhibit 3).
Project Operability
The correlation matrix revealed that only four of the thirteen team factors were positively correlated with plant operability. The variables significantly correlated with operability were clear goals, senior management support, office design to facilitate communication, and problem solving. Multiple regression analysis conducted on the operability dependent variable produced a two-factor model. The fitted model indicates that 31.8% of plant operability can be explained by two team-related practices: design of the project office to facilitate effective communication, and the alignment of the team around clear project goals.
Discussion
In this research, we have explored the direct effects relationships between critical team success factors identified from the project and team literatures, and the triple constraints of project cost, schedule, and operability. Clearly, the so-called soft team factors are able to explain a significant proportion of variance in these three hard project outcomes in capital-intensive projects. Despite our use of the conservative adjusted R-Square measure, each of our regression models indicated strong predictive power that ranged from 26.9% to 35.6%. This is a particularly strong result, given that the use of objective outcome measures is known to produce lower results than subjective measures, and that several technical factors excluded from our models are strongly predictive of capital project outcomes (Nemes & Lukas, 1996).
Our results also clearly demonstrate the value in empirically unpacking project outcomes. When project success criteria are disaggregated, critical success factors vary depending on the success criterion measured. The correlation matrix reveals that not all critical success factors have the same impact on each different success outcome. An examination of the regression models for each outcome adds further weight to the contention that different project outcomes are influenced by different combinations of team factors. When the most parsimonious model is selected for cost, schedule, and operability, the value of disaggregating project outcome dimensions becomes even clearer. Only one predictor (cross-functional integration) is common to two project success criteria (to both low cost and fast construction). As predicted from previous research, cross-functional teams—where all necessary functions are represented and well integrated—are associated with better cost performance and faster construction schedules. Cross-functional integration allows for the timely input from members with different areas of expertise, which greatly enhances the project's chance of success. In our sample, apart from this one common variable, the remaining key predictors of cost, schedule, and operability all differ.
Our findings also yield some interesting insights that advance the project management body of knowledge in regard to the current contingency debate over the generalizability of critical success factors to different project types and contexts. Our results lend weight to the view that not all factors are generic. It is clear that not all team success factors reported in the project literature apply in a critical way to this sample of capital projects executed in the process industries. Team collocation and team continuity exhibited no significant relationship to any of the studied outcomes. The setting of clear project goals is positively, but non-significantly correlated to schedule. Strong senior management support is negatively, but non-significantly related to project construction schedule. This last finding warrants further research. Several NDP studies suggest that micro-management or interference by senior management can slow down project schedules.
Limitations and Benefits of the Study
Some potential limitations should be kept in mind when interpreting these findings. One limitation is the moderately small sample size of 56 teams, although this is common in team research (e.g., Stewart & Barrick (2000) studied 45 teams; Wurst, Hoegl, & Gemuenden, (2001) studied 39 teams). Our modest sample size precluded us from testing a larger, more comprehensive model of team factors, and from testing for moderating or mediating interactions between variables. A positive innovation to our research design is that we tested research findings from other project types and applied them to an under-researched type of project in an under-researched context. Our study has demonstrated that caution should be exercised when generalizing findings to other project team contexts. Finally, our cross-sectional research design can only imply, rather than prove causality. A larger sample and a longitudinal design would be beneficial for future research.
Despite these limitations, the present study also adds theoretical and practical value by addressing some of the methodological problems in project team research. Our rigorous, quantitative methodology using objective performance measures avoided same-source bias. Although objective measures do not produce an explanatory power as strong as perceptual measures do, our results are still quite robust. Another contribution of this study is the use of metric outcome benchmarks, which enabled the direct comparison of projects with each other, and allowed for the quantification of the influence of each success factor on each of our three project outcome components.
In addition, our study extends current team and project knowledge in terms of content. Our findings lend empirical weight to the contention that skilful project team management is related to project success in terms of efficiency and effectiveness. We examined a comprehensive range of team success factors to gauge the relative importance of each. Arguably, our major contribution to the project management body of knowledge has been through demonstrating the value of unpacking project success criteria into distinct, separately measured outcome components, and demonstrating the different strengths of correlation of these outcome components with different combinations of drivers.
Implications for Academics and Practitioners
There are several key implications for academics from this research. Our study empirically extends project team theory by developing a comprehensive model derived from team theory and testing its generalizability to capital projects in the process industries. The current finding of different effects of team factors in this setting provide further evidence for the conjecture (Pinto, 2002) that project team success factors are not necessarily generic. Some of the key team success factors identified for other project types in NPD and R&D settings do not appear to be related to improved project performance for capital construction projects in the process industries.
By unpacking separate project outcomes, the study also provides compelling evidence that team factors may not be equal drivers of all criteria of project success. This new finding suggests that researchers need to review the common practice of aggregating project outcomes into one global measure. That particular methodology appears to be obscuring a clearer, more fine-grained differentiation of project team success factors. In view of our moderate sample size, we recommend further research using a larger sample. This would enable the use of structural equation modelling for simultaneous path analysis of all outcomes, as well as for the exploration of possible moderating and mediating interactions. In addition, we recommend future team research in other project settings, as well as the exploration of a broader comparative range of project success criteria, including the important soft outcomes of client/end-user and team satisfaction.
Our study also has important practical implications for capital project practitioners. We identify that it is possible to leverage project cost, construction speed, and plant operability through effective people management. There are clearly significant project performance gains to be made from managing human resources in capital projects. Our results add weight to the contingency argument that all project types and settings are not governed by the same set of generic practices. These results will alert managers to the dangers of indiscriminately applying research results from projects and settings different to their own.
Our findings also call into question the common assumption that all team practices are equally critical for all project success criteria. Our results highlight the need for project practitioners to recognize that not all team practices will achieve all desired outcomes. In any given project, some outcomes may be more important than others. Different management emphases are required according to the outcome desired. By separately measuring and modeling project cost, schedule, and operability, we have provided more highly differentiated information to allow capital project managers to leverage key team practices to achieve desired project outcome objectives.
A major implication of our findings is that project managers need to clearly prioritize their goals for each project so they can adopt the appropriate bundles of team practices that will facilitate their achievement. Our empirical findings offer valuable insights into what specific team practices might be used to achieve greater efficiencies in capital project cost and schedule control, and to improve plant effectiveness.
If project cost efficiency is an important goal, our model suggests that project managers need to focus on team and office design aspects. Best cost results are achieved by establishing integrated cross-functional teams early in the project life cycle. These teams would ideally have an autonomous project structure, would be empowered to engender team efficacy, and would be supported by virtual technology in their communication activities.
If speedy project construction is a business objective, management may achieve maximum benefit from focussing on team design and leadership variables. Forming an integrated and cross-functional team early in the project life cycle, ensuring project manger continuity, and offering incentives to the project manager for achieving specified project objectives are key factors related to fast construction. To achieve the highest proportion of plant operating capacity, the most effective levers will be designing the project office to facilitate effective communication and ensuring strong team alignment around clear project goals.
Armed with these insights, capital project managers can set about optimizing project performance through effective team management.
Acknowledgements
We wish to thank Ed Merrow, Rob Young, Andrew Griffith, Loretta Merrow and Independent Project Analysis for their support and invaluable assistance with data collection. We also gratefully acknowledge the contribution of managers and project teams at the companies surveyed.