Project governance is receiving increasing attention from both academics and practitioners (Central Computer and Telecommunications Agency, 1999; Turner & Keegan, 2001). Inadequate governing could lead to disastrous project outcomes. The Central Artery/Tunnel (CA/T) Project in Boston, also know as the Big Dig project, and the Fast Ferry Project in Canada are two cases in example. The Big Dig project is the largest federally funded public works project in recent United States (US) history. With an initial cost estimate of US$2.4 billion in 1984, the cost ballooned to US$13.6 Billion by 2000 and US$14.6 billion in 2002. A federal task force conducted a review on project cost and oversight in early 2000 and identified, among others factors, a lack of governing as a main reason for cost overruns (Murphy 2002; Federal Highway Administration, 2000). The Fast Ferry project was announced by the provincial government of British Columbia, Canada in 1994 to construct three fast car ferries for the British Columbia Ferry Corporation. The total cost for the project had risen to an estimated C$463 million from the original estimate of C$210 million. An audit conducted by the Office of the Auditor General of British Columbia (1999) has concluded that poor project governing is the main reason for cost overruns and delays.
One of the most commonly used governance mechanisms for senior manager's active involvement in a project is project sponsor. Typically, the project sponsor is someone who is high in an organization's hierarchy, the owner of the business case for the project, and responsible for making sure the conduct of the project is in the best interests of the organization and stakeholders (Kerzner, 2000, p. 237-249). However, since senior management's time and attention is limited, senior management is unable to be involved in all aspects of operation (March & Simon, 1958). The literature typically recommends management support as a universal approach without addressing the contingent effect of management support (Sabherwal & King, 1992). Thus, the normative question is: “Under what conditions should management proactively involve themselves in project management decision process?"
In addition to direct intervention by senior managers, project organizations can govern projects through project reviews and steering committees; organizations can also govern projects by establishing project offices. Existing studies on the effects of project review and project office are mostly conceptual. There is little empirical evidence on their effects. Similarly, studies on steering committees do not provide empirical evidence regarding the effect these have on organizational performance. This study investigates the effects of project sponsor, project review, steering committee, and project office on project delivery capability (PDC) and how these effects are influenced by the project context, the performance risk.
The data includes a survey of senior managers and project managers in Australian construction and IT services companies. The reasons for selecting these two industries are two-fold. First, both manage their work mainly in the form of projects and are comparable in management practices. Second, evidence indicates that project performance across the two industries is very different (Johnson, Boucher, Connors, & Robinson, 2001; Schut & Moonon 2003; Lemon, Liebowitz, Burn, & Hackney, 2002; Sauer, Liu, & Johnston, 2001; Walker & Sidwell, 1998), which enabled us to examine the differential effects of the same governing mechanisms under different context.
This study mainly draws from the information-processing view of organizations (Galbraith, 1977; Tushman & Nadler, 1978) and from management control literature (Simons, 1990). In the following sections, the hypotheses are developed, the research method described, the data analyzed, the findings elaborated, and the implications discussed.
Theory and hypotheses
In this study, the concept of project governance is adapted from corporate governance literature (Daily, Dalton, & Cannella, 2003) and Projects In Controlled Environments II (PRINCE II) (Central Computer and
Telecommunications Agency, 1999), as the determinate of the business objectives of projects, the determination of the uses of organizational resources for the conduct of projects, the monitoring and control of projects for attaining business objectives, and the resolution of conflicts among project stakeholders. Commonly used organizational mechanisms for implementing project governance include project review, project office, project sponsor, and project steering committee. Since the main purpose of project governance is to control projects and eventually achieve business objectives, this study examines the effect of the four project governance mechanisms on project delivery capability (PDC) and the level of control exerted by organizations. PDC is an organization's capability to deliver projects according to the client's expectations regarding time, cost, and quality. PDC differs from the term project performance in that PDC considers an organization's consistency when delivering various projects over an extended period of time; the latter typically refers to the one-off performance in delivering a project.
Organizational control is conceptualized as influences exerted by an organization through control mechanisms that help align the actions of its employees with the interests of the organization (Tannenbaum, 1968). Organizational control theory identifies three types or modes of control—behavior, output, and input—that can be directed at achieving the behavior necessary to secure desired performance. Behavior controls seek to secure a specified type of behavior in the belief that the behavior delivers the right results. Behavior controls are procedures that lay down a sequence of operations that must be followed. Output control specifies the schedule of outputs desired. The assumption is that people, who know what their targets are, adopt appropriate behaviors to achieve them. Input control regulates the knowledge, skills, abilities, values, and motives of the organizational members (Snell, 1992). For example, the selection and development of project managers can be an important input control mechanism for project organizations. Below, we develop hypotheses relating to the effects of the four project governance mechanisms on PDC and organizational control.
Project review and project office
Various types of reviews are used to control project quality, progress, and cost. Examples include user requirement review, design review, project kick-start review, project review, and post-mortem review (Block & Frame, 1998; Frame, 1994; Hallows, 1998; Kerzner, 1995). Expect for post-mortems, typical reviews occur at frequent intervals during a project's lifecycle, with the exact schedule depending on the uncertainty of the project. Attention is directed to budget and cost variances, meeting business objectives, problems in the execution of the project, and needs in resources and support (Block & Frame, 1998; Frame, 1994; Kerzner, 1995). The main function of a project review is to process critical project information to inform decisions about any adjustments required in a project plan to keep the project on track. It is thus expected that the use of project review will increase the level of control by the project organization and positively influence the performance in delivering projects.
Hypothesis 1A: PDC is a positive function of project review.
Hypothesis 1B: The level of organizational control is a positive function of the use of project review.
Evidence suggests that the management of construction projects is outcome driven (Liu, Yetton, & Sauer, 2001). PDC in this situation is primarily driven by output control mechanisms, such as setting project targets and milestones and measuring and tracking performance against these targets and milestones. Project reviews identify major deviations from project targets and standards, and feedback is provided to the project team in a timely fashion to address the problem. In this context, the project office's main role in relation to project review is to develop and manage the processes for administering project reviews. In order words, project review is the primary driver of project performance, enabled by the project office when task uncertainty is low. (See below for discussions on task uncertainty.)
The main focus of project review is to provide outcome feedback. According to information processing literature, as task uncertainty increases, the information processing capability needs to be enhanced (Galbraith, 1977; Tushman & Nadler, 1978). In the context of project organizations, increased task uncertainty is often the result of increased cross-organizational and cross-functional interdependencies, requirement in-determinacy, and technical uncertainties. Increased frequency of project review should increase its information processing capability. On the other hand, increased frequency could dramatically increase management overhead and delay the conduct of projects. It follows that project review is more effective when task uncertainty is low than when it is high.
Hypothesis 1C: When task uncertainty is low, project review mediates the effect of project office on PDC.
Increasingly, organizations are using a dedicated project office to serve the broader project management needs of multiple projects and the organization overall (Bernstein, 2000). The main functions of a project office, according to Bernstein (2000) and Kerzner (2000) include:
- Developing and maintaining project management frameworks and standards
- Accumulating knowledge base on project management and benchmarking
- Conducting project reviews
- Assisting the selection projects
- Allocating organizational resources
The project office is designed to facilitate the management of projects and is therefore expected to increase control and to positively impact PDC.
Hypothesis 2A: PDC is a positive function of the existence of a project office.
Hypothesis 2B: The level of organizational control is a positive function of the existence of a project office
When task uncertainty is high, cognitive feedback replaces outcome feedback as the main performance driver of performance (Sengupta & Abdel-Hamid, 1993). Outcome feedback is not sufficient for effective project management. Instead, a project office, which processes task-related information and provides a broader spectrum of feedback than simply the deviation from targets, provides task information (one of three types of information conveyed by cognitive feedback) to the project team. This enables effective execution of project tasks. The project office integrates information on outcome feedback from project reviews, historical information on similar projects, information on project stakeholders, and other task-related information that may inform the project team about their tasks and their decisions on resource allocation and coordination. In this context, project review is only one of a number of information sources that enable the project office to inform a project team about future plan adjustment and task execution. A project office's information processing capability, as measured by the frequency and intensity of information processing, is higher than that of project reviews because it has dedicated and specialized staff. The information processing view posits that the fit between task uncertainty and information processing capability leads to superior performance. Therefore, a project office is expected to have a better fit with the task than project review when task uncertainty is high.
Hypothesis 2C: When task uncertainty is high, the project office mediates the effect of project review on PDC.
Project sponsors
In addition to using project reviews and establishing a project office, appointing a project sponsor is a project governance mechanism that is widely employed by project organizations to exercise interactive control over projects (Briner, Hastings, & Geddes, 1990; Frame, 1994). Project sponsors are not responsible for the execution of projects. Rather, their key responsibility is to ensure the overall success of the projects (Briner et al., 1990). The role of project sponsors for a specific project depends upon the task environment and the project's characteristics. According to Simons (1990), when strategic uncertainty is high, senior managers (project sponsors in this context) should take interactive control of issues of strategic importance. Strategic uncertainty refers to the issues that are very important to business outcomes but highly uncertain. If an issue is very important, but poses little risk to the final outcome (e.g., cost control), the senior manager can delegate such issues to specialists. Similarly, if an issue has very high levels of uncertainty but little significance to final outcome, the senior manager may simply choose to ignore the issue. In this study, performance risk reflects strategic uncertainty.
Hypothesis 3A: When strategic uncertainty is high, PDC is a positive function of project sponsorship.
Hypothesis 3B: When strategic uncertainty is high, the level of organizational control is a positive function of project sponsorship.
Steering committee
A project steering committee is a high-level team of representatives from key stakeholder groups involved in a project who are entrusted with the task of linking project strategy with business strategy (Karimi, Bhattacherjee, Gupta, & Somers, 2000; Nolan, 1982). The key stakeholder groups typically include clients, project team, sponsors, suppliers, and consultants. Despite the fact that steering committees have been widely used to manage large and complex projects, most of the research on steering committees focus on the roles and effects of corporate-level steering committees (e.g., IT steering committee) (Karimi et al., 2000) rather than project steering committees.
As the highest decision-making mechanism for a project, project steering committees set the strategic direction of the project and coordinate resources necessary for achieving strategic goals. It serves to reduce the performance risk that is generated by a lack of cross-boundary coordination and direction (Nidumolu, 1996). It is used only for large and complex projects because the performance risk for these projects is typically high. Therefore, steering committees have positive effects on PDC and the level of control when performance risk is high.
Hypothesis 4A: PDC is a positive function of project steering committee when performance risk is high.
Hypothesis 4B: The level of organizational control is a positive function of a project steering committee when performance risk is high.
Research design
Unit of analysis
When testing the hypotheses, it is necessary to select an appropriate organizational level as the unit of analysis. Experience had shown us that—at least in some construction groups—well-recognized differences exist in management control systems and information control capabilities across business units within the same group. Whereas, within the business unit across projects, the difference in management control systems are small. We therefore chose the business unit as our unit of analysis.
Sample selection and data collection
We requested a mailing list of construction companies and IT services companies from an information service provider. The construction companies selected for this study were general contractors, heavy construction contractors, and special trade contractors within the definition of Standard Industry Classification (SIC)1 code 1521-1799 and with annual turnovers exceeding A$402 million. The IT companies selected were IT services companies within the definition of SIC 7371-7379 and with a turnover of A$10 million or over. In total, there were 232 in the construction category and 224 contacts in the IT services categories. [SIC, or standard industry classification, is a coding system that designates a four-digit codes to specific industries. The system was developed in the United States and used in many other countries, including Australia. The exchange rate between the Australian dollar and the American dollar between late 1998 and early 1999, when survey was conducted, was one Australian dollar to 0.6-0.7 US dollar.]
A fax was sent to the chief executive officer (CEO) or managing director at each of the 456 companies on the list, asking them to participate in our study by providing lists of project managers and senior managers in their business units overseeing multiple projects. Twenty IT services companies provided the names and contacts of 52 senior managers. Twenty-two construction companies provided the names and contacts of 67 construction senior managers. Questionnaires were sent directly to the 52 senior managers in IT services companies and the 67 senior managers in construction companies. The senior managers surveyed were also asked to provide a list of project managers were under their supervision for the purpose of participating in the second phase of the study.
The number of responses to our invitation fax was 20/224 = 8.9% in the IT sample, and 22/232 = 9.5% in the construction sample. The low response rates, while similar to other studies, signals a potential non-response bias, which threatens the findings’ validity (Schwab, 1999). To evaluate this threat, we questioned whether the samples could represent the two targeted groups as a whole. Our analysis showed us that there is no evidence of a non-response bias for either sample and, therefore, no potential validity to the threat due to a potential non-response bias. Due to page limits, the analysis is not included in this paper; please contact the authors for detailed analysis.
Instrument design and validation
Dependent variables
The dependent variable is the PDC of an organization. Traditionally, PDC is measured against achievement of the triple constraints: time, cost, and quality (Kerzner, 2001, p. 5). This view of project delivery is now under scrutiny for not including considerations about business case and business benefits. Focusing exclusively on project delivery is likely to achieve sub-optimal results at the cost of the whole life cycle performance. For example, the design and construction of a building have implications on the maintenance of the building after the project has been delivered. Similarly, the design and development of an IT system is likely to affect the operation and maintenance of the system after the project has been delivered. Although the trend in project delivery is to move towards a more collaborative approach or outsource the delivery and maintenance to a single entity, the main responsibilities for realizing business benefits is likely to lie with the client and the main responsibility for project delivery lie with project delivery companies. The expectation of the project company in realizing business benefits is likely to differ from project to project. The clients know best the contributions made by the project delivery team to their organizations.
Because this study focuses on IT services and construction companies whose main responsibilities are delivering projects, we adopted the traditional view of PDC with an eye towards extending that view in two distinct ways. First, PDC is measured against the expectations of the business unit and the performance of their competitors. Second, the client's satisfaction benefits are included in all three dimensions of PDC.
Here, PDC is measured by the perceptions of senior managers who are looking at performance from the point of accomplishing the triple constraints, i.e., cost, time, and quality, in accordance with respondent satisfaction, relative performance with competitors, and the satisfaction of the client. (See Appendix 1 for the instruments). Table 1 outlines good reliability for PDC regarding time, cost, and quality.
| Construction | IT | |||
| PDC—Quality | 0.85 | 0.90 | ||
| PDC—Cost | 0.71 | 0.75 | ||
| PDC—Time | 0.86 | 0.70 | ||
| Behavior control | 0.72 | 0.79 | ||
| Output control | 0.73 | 0.69 | ||
| Input control | 0.47 | 0.74 | ||
| Task programmability | 0.72 | 0.72 |
Table 1: Cronbach alphas for the dependent variables
The existing instruments for measuring control modes are quite diverse. Outcome control has been operationally defined with an emphasis on project targets (Cardinal, 2001; Kirsch, Sambamurthy, Ko, & Purvis, 2002; Snell, 1992), reward-outcome link (Cardinal, 2001; Kirsch, 1996), and pre-specified targets (Kirsch, 1996). Each of these three perspectives has been measured in this study. Specifically, output control is measured here in regards to specifying output targets, reporting performance results to coordinate activities, directing attention to measurable results, and making project managers accountable (see Appendix 1). These instruments have been adapted to the requirements of project management at a business unit level.
Behavior control has been operationally defined as pre-specified behavior (Kirsch, 1996; Kirsch et al., 2002), reward-behavior link (Cardinal, 2001; Kirsch, 1996), and centralization (Cardinal, 2001; Snell, 1992). Each of these three aspects are measured in this study with the wording adapted to project management at business unit level.
Input control has been operationally defined as staff selection and skill development (Cardinal, 2001; Snell, 1992). Our measurement of input control is also influenced by Sauer et al.‘s (2001) study that identified the importance of project manager development as a form of input control. In this regard, input control is measured on the extent of mentoring, job rotation, development of project manager responsibility in line with experience, and selection of project managers based on past performance (Graham & Englund, 1997; Kerzner, 1995).
Table 1 reports that the Cronbach alphas for the independent variables are near or above the conventional cut-off point of 0.7, except for input control in the construction sample (Nunnally, 1978). The reliability of input control is below the 0.7 cut-off point in the construction sample (α = 0.47). Data on this variable in the construction sample is reported here for completeness. The instruments are presented in Appendix 1.
Independent variables
Table 2 reports that the Cronbach alphas for the independent variables are near or above the conventional cut-off point of 0.7.
| Construction | IT | |||
| Steering committee roles | 0.87 | 0.74 | ||
| Project review | 0.66 | 0.70 | ||
| Project office | 0.74 | 0.86 | ||
| Project sponsor roles | 0.74 | 0.79 |
Table 2: Cronbach alphas for the constructs
Project office is measured by two questions concerning the degree to which the business unit or its parent has a dedicated office for supporting projects, and the degree that other functional units are geared toward servicing the projects (Bernstein, 2000; Block & Frame, 1998). Project review is measured by four questions covering the extent to which formal reviews are for providing advice to project teams, controlling and monitoring the project, and conveying information about the project's future. It also shows the extent to which the organization uses close supervision to control and coordinate activities and the extent to which other organizational units are made accountable to support the project (Busby, 1999).
Project sponsorship is measured on two dimensions, the use of project sponsors and the roles and purposes of project sponsorship. The former is measured by the percentage of projects undertaken by the business unit in the last 3 years that had project sponsors. The latter is measured by using as the definition the combination of four questions covering the various roles of the sponsor, such as the seat of formal decision making power and the ability to command resources, oversee corporate control, managing stakeholder interests, and assessing project outcomes (See Appendix 1).
Similarly, the project's steering committee is measured by its use of the steering committees and the roles and purposes of the steering committees. The former is indicated by the percentage of projects undertaken by the business unit in the past 3 years that had steering committees. The latter is measured on the degree to which the steering committees were given authority for allocating resources over the projects, advising on projects, controlling projects, facilitating stakeholder involvement, and accepting responsibility for project outcomes (Karimi et al., 2000; Nolan, 1982).
Task uncertainties in the construction and IT industries
One of the key assumptions that propelled this study is that the level of task uncertainties for construction projects (average across projects) is lower than that for IT projects. Below we compare the levels of task uncertainties in the two industries, using as our guide the three dimensions defined by Lawrence and Lorsch (1967).
First, on the dimension of clarity of information, the clarity of task requirement informs the way projects are conducted. In the construction industry, the requirement for this first dimension has not yet been identified as a major factor causing performance problems. In contrast, the requirements for an IT project are very difficult to clearly specify ex ante and change throughout the project lifecycle due to the dynamic nature of business environment, technological change, and embedded organizational information systems (Johnson 1995; Johnson et al., 2001; Lyytinen & Robey, 1999).
Second, the time span for obtaining definitive feedback from construction project tasks is in general shorter than that for IT project tasks. Construction projects are clearly observable. Construction managers can obtain a clear picture of the progress of a construction project by just walking through the construction site or even observing through an on-site video monitor. In contrast, the outcomes of IT tasks are neither visible nor visualizable, which makes some of the most powerful conceptual tools such as visualization tools (e.g., structural maps or Gantt charts) less useful for IS projects than for construction projects (Brooks, 1987). It is also often difficult to obtain accurate feedback for typical IT projects due to their dynamic nature (Sengupta & Abdel-Hamid, 1993).
Third, the uncertainty of cause-and-effect relationships is generally lower for construction projects than for IT projects. As compared to IT projects, construction projects demonstrate greater similarity between projects, lower levels of technological change (IBIS, 2000), and a longer history of engineering (Sommerville, 2000; Straub, 1964). All of these factors enable the construction industry to obtain a clear picture of cause-and-effect relationship driving this industry's projects. In contrast is the IT industry, where the complexity of software systems, the embedded processes and structures guiding organizations, and the continuously changing task environment make it difficult, if not impossible, to foreshadow causal relationships.
For an organization that organizes its work mainly by projects, task uncertainty is ultimately reflected in its ability to predict project outcomes. This study uses performance risk as a surrogate measure for task uncertainty in managing project tasks at the business unit level. Performance risk is defined as the likelihood of a deviation from expected project outcomes (Nidumolu, 1996). Specifically, outcomes are measured on the deviations from defined cost, time, and specification targets (See Appendix 1 for the question items). Performance risk in the construction and IT services industries can be estimated from the past performance of projects in the same industries. Since it is unlikely that most of the senior managers surveyed will have specific information on the deviation of expected targets for all projects, we decided to ask project managers from the IT services and construction companies surveyed about performance risk. The senior managers who agreed to participate in our survey were asked to provide a list of project managers under their supervision. We then send survey questionnaires to these project managers, asking them about the deviations from project targets in regards to the last project they completed. These deviations were used as estimates for determining performance risk on the corresponding business units. In case respondents supplied multiple responses from the same business unit for different projects, the mean of the deviations were used as estimates. (See Appendix 1 for the questions.)
The literature shows that the performance of IT projects has been dismal (Johnson, 1995; Johnson et al., 2001). In contrast, the performance of construction projects has consistently improved during the past few decades (Walker & Sidwell, 1998).
Assumption 1: Performance risk is higher in the construction industry than in the IT services industry.
Results
Characteristics of the samples
In total, we received 57 completed questionnaires from the construction sample and 40 for the IT services sample, representing a senior manager response rate of 57/67 = 0.85 and 39/52 = 0.75, respectively.
The 57 responses from the construction sample represent 55 business units. There are two business units with two responses each. Multiple responses for the two business units were aggregated using the mean of the responses. In addition, one response was excluded because only 5% of the employees of that business unit were engaged in project tasks. The resulting sample size is 54 for the construction sample.
The 39 responses in the IT sample represents 37 business units. One response was excluded because it was an internal business unit. There are two business units with two responses each. Because the two respondents for each of the two business units have very different tenures with the business units (15 years versus 2 years in one case and 18 year versus 1 year in the other), the response from the respondent with shorter tenure with the business unit was excluded from the sample. The size of the IT services sample is 36.
| IT | Construction | |||||
| Minimum | Maximum | Mean | Minimum | Maximum | Mean | |
| Operating budget of business unit (A$ millions) | 2.00 | 650.00 | 51.46 | 1.00 | 1800.00 | 153.96 |
| Full-time equivalent employees in business unit | 5.00 | 1500.00 | 225.37 | 4.00 | 5000.00 | 341.61 |
| Proportion of employees in projects | 10.00 | 100.00 | 61.53 | 20.00 | 100.00 | 74.53 |
| Percentage of revenue from projects | 5.00 | 100.00 | 72.16 | 10.00 | 100.00 | 85.18 |
| Size of project (A$ millions) | .50 | 800.00 | 29.03 | 1.10 | 700.00 | 62.43 |
| Person years involved in the project | 2.00 | 900.00 | 58.14 | 1.10 | 900.00 | 207.70 |
Table 3: Sample characteristics
Table 3 presents some key attributes of the business units included in the samples. The budget size and number of employees are very different in the two industries. Despite the differences, the ratios on the proportions of employees on projects and revenue from projects in the two samples are greater than 50%, supporting our assumption that the two industries are project-centered.
The project managers questionnaire was sent directly to 52 IT project managers whose names have been provided by the participating organizations. The questionnaire was also sent to some participating organizations to distribute to their project managers. No information exists concerning the number of project managers who were contacted by the initial group of project managers. In total, we received 62 responses from IT project managers representing the participating organizations. Our questionnaire was sent directly to 54 construction project managers; we received 41 responses, resulting in a 76% response rate.
Table 3 reports that the project size and number of people involved are quite different. In total, 11 small projects were excluded from the IT sample, as was one response from a project manager of an internal IT department. The sample size in the IT project manager sample is 50.
Assumption 1: Comparison of performance risk across the two industries.
Assumption 1 is supported. All the three performance risk dimensions show the IT sample wit significantly higher levels of risk than the construction sample. There is no significant difference (p < 0.22) in quality performance risk across the two samples.
| 0-con 1-IT | N | Mean | Std. deviation | Std. error mean | ||
| Performance risk | Time | Construction | 40 | -.44 | 21.72 | 3.43 |
| IT | 48 | -10.13 | 22.30 | 3.22 | ||
| Cost | Construction | 38 | -3.88 | 10.09 | 1.64 | |
| IT | 46 | -23.83 | 60.28 | 8.89 | ||
| Quality | Construction | 34 | 94.79 | 16.09 | 2.76 | |
| IT | 46 | 98.33 | 3.16 | .47 | ||
| Levene's test for equality of variances | t-test for equality of means | ||||||
| F | Sig. | t | df | Sig. (2-tailed) | |||
| Perform.risk | Time | Equal variances assumed | .53 | .47 | 2.05 | 86 | .04 |
| Equal variances not assumed | 2.06 | 83.89 | .04 | ||||
| Cost | Equal variances assumed | 13.06 | 0.00 | 2.02 | 82 | .05 | |
| Equal variances not assumed | 2.21 | 48.04 | .03 | ||||
| Quality | Equal variances assumed | 6.75 | 0.01 | -1.46 | 78 | .15 | |
| Equal variances not assumed | -1.26 | 34.89 | .22 | ||||
Table 4: Mean comparison of performance risk
The main effects of project review and project office (H1A, H1B, H2A and H2B)
In Hypothesis 1A, project review has a main effect on PDC; in Hypothesis 2A, project office shows a main effect on PDC. Table 5 reports that project review has strong effects on PDC in the construction sample; it shows, however, only marginal effects in the IT sample. Table 5 also shows that project office has significant and similar effects on PDC across both samples.
As a result, hypotheses 1B and 2B are supported. Table 5 reports that project review has significant effects on the level of controls in both samples. Similarly, project office also has significant effect on the level of control in both samples.
| Project office | Project review | |||||
| β (R2) | p (Data points) | β (R2) | p (Data points) | |||
| Hypothesis 1A, 1B--Low task uncertainty (Construction) | ||||||
| Quality | 0.36 (0.13) | 0.01 (n=48) | 0.52 (0.27) | 0.00 (n=41) | ||
| Time | 0.35 (0.12) | 0.02 (n=48) | 0.46 (0.21) | 0.00 (n=41) | ||
| Cost | 0.35 (0.12) | 0.02 (n=48) | 0.43 (0.19) | 0.00 (n=41) | ||
| Output control | 0.26 (0.07) | 0.08 (n=48) | 0.47 (0.22) | 0.00 (41) | ||
| Behavior control | 0.42 (0.18) | 0.00 (n=48) | 0.22 (0.05) | n.s. (n=41) | ||
| Input control | 0.11 (0.01) | n.s. (n=47) | 0.30 (0.09) | 0.05 (n=40) | ||
| Hypotheses 2A, 2B --High task uncertainty (IT) | ||||||
| Quality | 0.36 (0.13) | 0.05 (n=30) | 0.26 (0.07) | n.s. (n=34) | ||
| Time | 0.32 (0.10) | 0.08 (n=30) | 0.31 (0.10) | 0.07 (n=33) | ||
| Cost | 0.42 (0.18) | 0.02 (n=30) | 0.10 (0.01) | n.s. (n=34) | ||
| Output control | 0.48 (0.23) | 0.01 (n=30) | 0.61 (0.37) | 0.00 (n=34) | ||
| Behavior control | 0.43 (0.19) | 0.02 (n=30) | 0.51 (0.26) | 0.00 (n=34) | ||
| Input control | 0.21 (0.05) | n.s. (n=30) | 0.49 (0.24) | 0.00 (n=34) | ||
Table 5: The main effects of Project Review and Project Office
The effects of project sponsorship and steering committee
Our study shows that our hypotheses 3A and 3B are supported. Table 6 reports that in the IT sample, the use of a sponsor has significant effects on PDC (quality), output control, and input control. Similarly, sponsor roles also have significant effects on PDC (quality) and output control. In the construction sample, use of sponsor has a significant effect on PDC (cost), but no effects on the level of control. Sponsor roles have no effect on either the PDC or the level of control.
| Hypotheses 3A, 3B sponsorship | Performance risk | |||||
| High (IT) | Low (construction) | |||||
| β (R2) | p (data points) |
β (R2) | p (data points) |
|||
| Use of project sponsors | ||||||
| Quality | 0.39 (0.16) | 0.02 (n=35) | -0.05 (0.00) | n.s. (n=52) | ||
| Time | 0.10 (0.01) | n.s. (n=34) | 0.21 (0.04) | n.s. (n=52) | ||
| Cost | -0.02 (0.00) | n.s. (n=35) | 0.37 (0.14) | 0.01 (n=52) | ||
| Output control | 0.45 (0.20) | 0.01 (n=35) | 0.04 (0.00) | n.s. (n=52) | ||
| Behavior control | 0.25 (0.06) | n.s. (n=35) | -0.02 (0.00) | n.s. (n=52) | ||
| Input control | 0.45 (0.18) | 0.01 (n=34) | 0.02 (0.00) | n.s. (n=51) | ||
| Sponsor roles | ||||||
| Quality | 0.35 (0.12) | 0.05 (n=31) | 0.08 (0.01) | n.s. (n=40) | ||
| Time | 0.05 (0.00) | n.s. (n=31) | 0.13 (0.02) | n.s. (n=40) | ||
| Cost | -0.07 (0.01) | n.s. (n=31) | -0.03 (0.00) | n.s. (n=40) | ||
| Output control | 0.35 (0.12) | 0.05 (n=31) | 0.08 (0.01) | n.s. (n=40) | ||
| Behavior control | 0.05 (0.00) | n.s. (n=31) | 0.13 (0.02) | n.s. (n=40) | ||
| Input control | 0.12 (0.01) | n.s. (n=31) | 0.14 (0.02) | n.s. (n=39) | ||
Table 6: The effect of project sponsorship on PDC and control
Requirement changes, or scope creep, have consistently been identified as one the major factors causing IT project failures (Boehm, 1991; McConnell, 1996; Schmidt, Lyytinen, Keil, & Cule, 2001). Similarly, other studies show that the quality of the requirements analysis phase impacts on later phases (McConnell, 1996; Zmud, 1980). Errors not identified at the earliest stages of a software project are expensive to fix later (Boehm, 1991, McConnell, 1996). Therefore, managing software requirements and achieving quality for the client are both important and uncertain, demanding close attention from senior management for successful realization (Simons, 1990). The task of achieving quality that satisfies a client's needs has high levels of strategic uncertainty. The intent of senior management's involvement is to protect the system's quality requirement. The impact on time and cost performance is secondary (Yetton, Martin, Sharma, & Johnston, 2000). The significant effect on use of sponsors and sponsor roles on PDC (Quality) in the IT sample supports hypothesis 3A.
In contrast is the construction industry, where the market is volatile rather than the tasks (Eccles, 1981a, 1981b, IBIS, 2000; Langford & Male, 1991). Competition in the construction industry is very much based on cost. Therefore, cost performance appears to be the strategic issue that most demands the attention of senior managers. The significant effect of use of sponsors on PDC (cost) in the construction sample supports hypothesis 3A.
Hypothesis 4A is not supported. Project steering committee has no significant effect on PDC in either sample.
Hypothesis 4B is supported. Table 7 shows that when performance risk is high, the use of a steering committee has a marginal effect on the level of behavior control (β = 0.31, p < = 0.07, R2 = 0.10) and steering committee roles have significant effect on the level of output control (β = 0.35, p < = 0.05, R2 = 0.12).
| Hypotheses 4A, 4B steering committee | Performance risk | ||||
| High (IT) | Low (construction) | ||||
| β | p | β | p | ||
| Use of project steering committee | |||||
| Quality | 0.01 (0.00) | n.s. (n=34) | -0.11 (0.01) | n.s. (n=52) | |
| Time | -0.15 (0.02) | n.s. (n=33) | -0.04 (0.00) | n.s. (n=52) | |
| Cost | -0.17 (0.03) | n.s. (n=34) | 0.21 (0.04) | n.s. (n=52) | |
| Output control | 0.25 (0.06) | n.s. (n=34) | 0.06 (0.00) | n.s. (n=52) | |
| Behavior control | 0.31 (0.10) | 0.07 (n=34) | 0.10 (0.01) | n.s. (n=52) | |
| Input control | -0.03 (0.00) | n.s. (n=33) | 0.01 (0.00) | n.s. (n=51) | |
| Steering committee roles | |||||
| Quality | 0.21 (0.04) | n.s. (n=28) | 0.16 (0.03) | n.s. (n=17) | |
| Time | 0.07 (0.01) | n.s. (n=27) | -0.19 (0.04) | n.s. (n=17) | |
| Cost | -0.14 (0.02) | n.s. (n=28) | 0.17 (0.03) | n.s. (n=17) | |
| Output control | 0.35 (0.12) | 0.05 (n=31) | 0.10 (0.01) | n.s. (n=17) | |
| Behavior control | 0.05 (0.00) | n.s. (n=31) | -0.04 (0.00) | n.s. (n=17) | |
| Input control | 0.24 (0.06) | n.s. (n=27) | 0.01 (0.00) | n.s. (n=17) | |
Table 7: The effect of project steering committee on PDC and control
The joint effect of project review and project office
Hypothesis 1C is partially supported. The regression analysis results explaining the mediating effects of project review in relation to project office on project performance when task uncertainty is low are reported in Tables 5 and 8. The results partially satisfy the test criteria for a mediated relationship (Baron & Kenny, 1986).
Hypothesis 2C is supported. The regression analysis results explaining the mediating effects of project office in relation to project review on project performance when task uncertainty is high are summarized in Tables 5 and 8. The three conditions required for a mediating relationship are supported (Baron & Kenny, 1986).
| Task uncertainty | |||||
| Low (construction) | High (IT) | ||||
| Dependent variables | Independent variables |
β | p | β | p |
| Project office | Project review | N/A | N/A | 0.41 (0.17) | 0.03 (n = 29) |
| Project review | Project office | 0.34 (0.12) | 0.03 (n = 40) |
N/A | N/A |
Quality (0.33, n=40) |
Project office | 0.30 | 0.04 | 0.27 (0.15) | n.s. (n = 29) |
| Project review | 0.40 | 0.01 | 0.20 | n.s. | |
Time (0.31, n=40) |
Project office | 0.31 | 0.04 | 0.37 (0.15) | 0.07 |
| Project review | 0.37 | 0.01 | 0.03 | n.s. (n = 29) |
|
Cost (0.16, n=40) |
Project office | 0.08 | n.s. | 0.51 (0.22) | 0.01 (n = 29) |
| Project review | 0.37 | 0.03 | -0.13 | n.s. | |
Table 8: Joint effects of project review and project office
Findings and implications
Consistent with management control theory (Simons, 1990), this study finds that project sponsorship has a significant and positive impact on PDC (quality) in the IT sample and PDC (cost) in the construction sample. Given that the construction industry competes on costs and IT industry competes on problem solving (quality), the findings provide further evidence supporting the contingent nature of management intervention. The finding that project sponsorship positively impacts on controls when performance risk is high rather than lower suggests that management is very concerned about high levels of performance risk and intervenes to enhance project control. The finding also suggests that when performance risk is low, management's attention is not associated with project controls.
Project review is found to have a significant and positive impact on PDC when task uncertainty is low; this impact has only a marginal effect on PDC when task uncertainty is high. Project office has a significant and positive impact on PDC in both samples. Extending the existing literature, the study finds that when task uncertainty is low, the effect of project office on PDC is mediated through project review. In contrast, when task uncertainty is high, the effect of project review on PDC is mediated by project office; the effect of project sponsor is moderated by strategic uncertainty. In the former context, project feedback through project review is the performance driver enabled by project office. In the latter context, the coordination of organizational resources is the performance driver enabled by the feedback from project reviews. Senior management's involvement is critical for the performance of strategic tasks. The results also indicate that both project review and project office have significant positive effects on the level of control in both samples, suggesting that regardless of the level of task uncertainty and performance risk, project review and project office are used to enhance organizational control with varying levels of effects on PDC.
Steering committee is found to have significant effects on the level of output control when performance risk is high; steering committee has no effect on PDC in both samples. This finding is consistent with steering committee's key function of strategy setting (output control) and coordination of resources by adjusting formal coordination processes (behavior control).
Two different sets of implications for practice follow from this study. One set explores how PDC can be improved without changing the context. The other focuses on how benefits could follow from changes in context.
In the typical IT context where strategic and task uncertainties/performance risks are high, senior managers have an immediate opportunity to improve the performance of IT projects when they focus their time resources on issues of strategic importance, namely, protecting the quality of strategic tasks and leaving the issues of cost and time management as the responsibilities of project managers. Senior managers should concern themselves with achieving a product quality that meets the client's needs, simultaneously providing the resources required to satisfy these goals. To do this, senior managers must intervene directly to influence project outcomes (output control) and the selection and development of key project personnel.
These findings also suggest that for IT services companies the outcome feedback from project reviews to project teams is neither adequately nor sufficiently timely to improve their current project performance. In the absence of sound knowledge of project tasks and lack of experienced personnel, support from the project office for a task model that has proven—on similar tasks—to help improve project performance in the short run. Best practice and benchmarking represent such an approach. A dedicated organizational unit informed by project reviews can facilitate cross-functional and cross-organizational coordination of project resources, provide task models and task information (cross-functional and/or cross-organizational) to project teams, and improve processes over time. Project steering committees could control the strategic direction of projects and assist in cross-boundary coordination.
In the typical engineering/construction context, where strategic and task uncertainties/performance risk are low, project review can be relied upon as the main governance mechanism enabled by the establishment of a corporate project office. Project sponsors should be appointed for high-impact projects; these sponsors would be responsible for achieving the bottom line.
The second set of implications that emerged from this study is based on the notion that by reducing the level of strategic uncertainty and performance risk, senior managers can be relieved from monitoring and intervening to protect project outcomes and concentrate more on other issues critical to corporate performance. By reducing strategic uncertainty, PDC becomes more consistent by removing human capability as a key performance driver. One way of reducing strategic uncertainty and performance risk is to improve the risk management system. An effective risk management system would result in the two benefits discussed above, and enable the organization to rely more on the direct feedback from project teams during project review and less on the project office as a performance driver.