The antecedents and effect of ambidexterity in the management of complex engineering projects

David Leitner BE (USYD), MPM (USYD), MIEAust, CPEng, IntPE, Leighton Contractors, NSW, Australia

Abstract

Projects, as the common unit of organizing complex tasks, often feature unique challenges that require both radical innovation and incremental improvements to processes and solutions in order to deliver the project within constraints. Both exploration and exploitation are likely to be needed for complex engineering projects to succeed. This study, through an in-depth case study of a complex infrastructure project facing unique challenges and tight budget and schedule, investigates the antecedents of ambidexterity and the effect of ambidexterity on project performance.

The main contributions of this research include: finding ambidexterity at the project team level is a significant contributor to project performance; the effects of temporal separation and project context on project performance are mediated by the project team's degree of ambidexterity; and, contrary to the prediction from organizational literature, structural separation has been found not to have contributed to the case project team's ambidexterity.

We believe that among the three antecedents for ambidexterity at the project level as predicted by organizational literature, structural separation is rendered ineffective by the natural cycling between exploration and exploitation in the project environment and by the need to integrate the two under time pressure and resource constraints. The development of an effective project context can facilitate the temporal cycling between exploration and exploitation by creating awareness that overcomes the inertia in transitioning between the two by setting performance expectations and appropriate social context. Implications for theory development and practice are discussed.

Keywords: Innovation; Exploration; Ambidexterity; Temporal Separation; Structural Separation; Contextual Ambidexterity; Exploitation

Introduction

Organizations adapt to environment through exploiting existing capabilities (e.g., continuously improving processes for existing products) and exploring new possibilities (e.g., trying new ways of designing products or developing new products). Exploration refers to the acts of searching and discovering innovative solutions, whereas exploitation refers to the refining and gradual improvement of existing capabilities (O'Reilly & Tushman, 2007). The ability of an organization to both explore and exploit is referred to as organizational ambidexterity. Studies have shown that ambidexterity leads to sustained competitive advantage (Holmqvist, 2004; Crossan, Lane, & White, 1999). Most studies on ambidexterity have been conducted at the organizational level.

Projects, as the common unit of organizing complex tasks, often feature unique challenges that require both radical innovation and incremental improvements to processes and solutions in order to deliver the project within constraints. A project is a “temporary endeavour undertaken to create a unique product or service,” where uniqueness is defined as being “different in some way from all similar products or services” (Project Management Institute, 1996). The uniqueness of a project typically requires customization—exploring and tailoring solutions for unique problems. On the other hand, efficiency considerations demands using repeat processes and off-the-shelf solutions for timely and low cost completion, which typically relies on gradual and continuous improvement over time— exploitation. Both exploration and exploitation are likely to be needed for complex engineering projects to succeed. The challenge for achieving ambidexterity in a project is that exploration and exploitation require very different organizational support (Andriopoulos & Lewis, 2009) and that project solutions need to be delivered within specific constraints. Therefore, ambidexterity is a critical capability for projects facing unique challenges and tight constraints. Nevertheless, few studies examined how ambidexterity can be achieved at the project level and its effect on project performance.

This study, through an in-depth case study of a complex infrastructure project facing unique challenges and tight budget and schedule, investigates the antecedents of ambidexterity and the effect of ambidexterity on project performance. The main contributions of this research include: finding ambidexterity at the project team level is a significant contributor to project performance; the effects of temporal separation and project context on project performance are mediated by project team's degree of ambidexterity; and, contrary to the prediction from organizational literature, structural separation has been found not to have contributed to the case study project team's ambidexterity.

In this paper, literature is reviewed and hypotheses developed. Then, research design is elaborated and analysis results are presented. Finally, the implications are discussed and conclusions drawn.

Literature Review and Hypotheses

Organizations adapt to environmental uncertainties through exploration and exploitation (March, 1991). Exploration includes the search for and experimentation with new approaches, processes, or procedures with an aim to find new solutions or develop new products or services (O'Reilly & Tushman, 2007). Exploitation refers to efforts to refine and gradually improve existing capabilities and/or procedures with an objective to capitalize on existing capabilities as much as possible (O'Reilly & Tushman, 2007).

Whereas exploratory learning has a potentially limitless array of outcomes, exploration often leads to dead ends or failure but can also lead to new innovation and ground-breaking change. Exploration involves undertaking high risk ventures and therefore the success rate is low. An over reliance on exploration could result in short-term cash-flow problems. In contrast, exploitation learning narrows the organization's focus to its core competencies, leading to short-term success but, at the same time, may result in the organization falling into so-called “success trap,” where the organization fails to explore opportunities effectively and as a result cannot adapt as the competitive environment changes (Gupta, Smith, & Shalley, 2006). According to Benner and Tushman (2003), organizations focused on exploitation will generally fail to effectively explore new opportunities or vice versa because exploration and exploitation require very different organizational support, respectively, which makes the simultaneous pursuit of both exploration and exploitation extremely challenging. Another challenge with balancing exploitation and exploration is that organizations tend to find comfort with routines developed through exploitation compared with high risks associated with exploration. Consequently, there exists a bias in organizations towards exploitation at the cost of exploration (Andriopoulos & Lewis, 2009). It is also difficult for organizations to cycle between periods of exploration and exploitation as after long periods of exploitation, organizational members’ skill sets for exploration become diminished and as a result the team members tend to adopt very narrow definitions of organization problems (Gupta et al., 2006).

Literature suggests that by separating exploration and exploitation activities into different organizational units, exploration and exploitation could be managed effectively using different incentive organizational instruments. For example, an organization's R&D unit might be focused squarely on exploration, adopt an organic organisational structure and rely on horizontal coordination. In contrast, an organization's production unit might be focused squarely on exploitation, operate in a mechanistic organisational structure, be hierarchical and have centralized decision-making authority. Ambidexterity requires the separated business units to effectively integrate exploration and exploitation (Gupta et al., 2006).

An alternative to structural separation is to temporally separate activities of exploration and exploitation within a single unit (Puranam, Singh, & Zollo, 2006). Punctuated Equilibrium refers to the process of seeking a balance between exploration and exploitation by means of temporal separation (Gibson & Birkinshaw, 2004). Both structural and temporal separation approaches create conditions that allow exploration and exploitation to be effectively managed within a single business unit and ultimately achieve ambidexterity if integrated well. However, from a learning perspective, structural separation is argued to be a more effective approach to organizational learning than temporal separation due to difficulties encountered when switching between conflicting modes of learning (Gupta et al., 2006). On the other hand, structural separation is not always practical. For example, in a small organizational unit, it is costly to split into and manage two very small units. In the cases of projects, it is also often desirable to have team follow through from concept to execution to prevent loss of specific project knowledge. In these cases, temporal separation may be a favourable alternative.

For organizations to be ambidextrous, they must be successful in being both consistent and inconsistent in their actions. To achieve this, the organization must achieve a balance between the need to be small and large, centralized and decentralized and have a short-term and a long-term focus all simultaneously (Benner & Tushman, 2003). Cegarra-Navarro, Sanchez-Vidal, & Cegarra-Leiva (2011) argued that individuals within an ambidextrous organization should be capable of unlearning outdated information or processes and re-learn updated knowledge. Similarly, an individual involved in the exploration must re-learn to adapt to the demands of existing routines and process when moving from exploration to exploitation and vice versa. The concept of contextual ambidexterity focuses on creating an environment that empowers individuals to make a choice between exploration and exploitation activities in their daily work.

Contextual ambidexterity utilizes behavioural and social means of integrating exploration and exploitation. Examples include team-building practices, fostering shared values, and aid in team coordination (Gibson & Birkinshaw, 2004). Contextual ambidexterity focuses on creating an organizational environment that supports, encourages, and develops organisational ambidexterity by developing organizational conditions that drive team performance as well as promote team member creative freedom. Gibson and Birkinshaw (2004) conceptualized contextual ambidexterity as the product of alignment and adaptability. Alignment refers to the capability to seek coherence among organizational activities, whereas adaptability refers to the organization's capability to reconfigure to meet changing demands. Contextual ambidexterity is achieved through empowering individuals to decide on the time spent on exploration activities or exploitation activities.

Organizational context is defined by Ghosal and Bartlett (1994) as having four attributes— stretch, discipline, support, and trust—that interact to define an organizational context. Birkinshaw and Gibson (2004) combined the four attributes of organizational context defined by Ghosal and Bartlett into two dimensions: performance management and social support. The former is concerned with motivating people to deliver best results while holding them accountable. The latter refers to providing people with the “security and latitude they need to perform” (Birkinshaw & Gibson, 2004). According to Birkinshaw and Gibson, both dimensions are equally important in developing a “high performance organizational context” which enables “truly ambidextrous organization.”

At the project level, similar challenges for achieving ambidexterity exist. One of the key differences is projects are bounded by constraints (e.g., scope, time and cost), are progressed in phases and are implemented to address unique needs. At the project initiation phase, it is typically necessary to explore alternative solutions and processes to identify a satisfactory solution that will address client's needs within the project constraints. As the project progresses, the focus shifts to exploitation as the project solutions become familiar and the capability is developed to learn from initial experiences and streamline processes (Brady & Davies, 2004; March, 1991). For projects facing low levels of uncertainty, complexity, and constraints, the need for exploration will be less than those facing high uncertainty, complexity, and tight constraints.

For a typical engineering project, the life cycle can be broken into four phases: concept, design, construction, and post construction. In a study of the Sutong Bridge Project in China, the longest suspension cable-stay bridge in the world, Wang, Liu, and Sheng (2009) found that at the early project phases, the focus was on exploring alternative solutions and as the project moves into the construction phase, the focus quickly shifts to exploitation. Parallel to the findings from organizational level, the ability to both explore and exploit during the time span of a project have positive influence on project performance. Formally, Andriopoulos and Lewis (2009) posit that in project context differentiation seems to entail separation by “iterating between project constraints and freedom” through temporal separation of exploration and exploitation. They propose that these iterations enable “creative expression and commercial success” by enabling an effective channel though which the project team can find an effective balance between the paradoxical forces of both exploration and exploitation.

Hypothesis 1: Project team ambidexterity has a positive effect on project performance.

In complex projects temporal swings between exploration and exploitation are not simply random events in time but rather driven by natural fluctuations of the project programme. In large projects each item of the project work breakdown structure can itself be considered as a small project, requiring the development of a unique solution to a new and unique problem. The temporal separation of exploration and exploitation in project phases enables ambidextrous management of the project. Following Purnanam et al. (2006), temporal separation of exploration and exploitation at the project team level leads to ambidexterity.

Hypothesis 2: The effect of temporal separation on project performance is mediated by project team ambidexterity.

Similarly, the structural separation at the project level, such as separating into teams with distinct roles and responsibilities on exploration and exploitation, should facilitate ambidexterity at the project team level (Raisch & Birkinshaw, 2009).

Hypothesis 3: The effect of structural separation on project performance is mediated by project team ambidexterity.

Gibson and Birkinshaw (2004) observed the effects of organizational context on team ambidexterity and team performance. Similar to the concept of contextual ambidexterity, project context—social and performance context at the project level that empowers team members to decide time allocation between exploration and exploitation activities could lead to ambidexterity at the project team level.

Hypothesis 4: The effect of project context on project team performance is mediated by project team ambidexterity.

 

Research Design

This section provides an overview of the case project and describes how the data are collected, coded, and analyzed, and reports sample descriptive statistics and construct validity indicators.

Case Study Project Overview

The Newmarket Viaduct Replacement Project is a complex construction project in the busy metropolitan suburb of Newmarket in downtown Auckland city, New Zealand. The project consists of a program of sub-projects, including the construction of a new twin box girder viaduct, the deconstruction of an existing twin box girder viaduct and all associated minor structures and roadworks. Since the Newmarket Viaduct Replacement Project has been referred to in press and in our interviews as a project, for consistency, in this paper, we continue referring to the program as a project. The Newmarket Viaduct Replacement project is complex and high risk due to the fact that the project team must successfully replace the fragile existing viaduct with a modern, stronger and wider twin box girder viaduct without disrupting access or traffic on the viaduct or underneath the viaduct, which crosses several pedestrian accesses, roads, and railways. The deconstruction of a viaduct over a heavily populated metropolitan area is a major technical challenge, so the project team has devised radical, never before attempted techniques to deconstruct the existing viaducts.

At the time of study in September 2011 the project team had successfully constructed and opened one of the new viaduct box girders and deconstructed one of the existing viaduct's box girders with one box girder remaining to be constructed and one existing box girder remaining to be deconstructed. With the deconstruction of the southbound viaduct being a world first in terms of deconstruction technique adopted, the construction team has an opportunity to review the lessons of the southbound viaduct deconstruction and re-visit its methodologies for the northbound viaduct deconstruction. This is an opportunity for exploratory learning where original assumptions adopted for the deconstruction of the southbound viaduct can be challenged and significant changes in the deconstruction sequence and methodologies can be made to improve safety, efficiency, and greatly reduce cost. With the construction team currently focused on the construction of the new northbound viaduct, the construction team also has the opportunity to review construction processes and exploit learning from the southbound viaduct to increase site productivity and decrease construction costs.

The Newmarket Viaduct Replacement Project team consisting of an alliance between the client New Zealand Transport Agency and Leighton Contractors, Fulton Hogan, VSL, URS New Zealand, Beca, Tonkin & Taylor and Boffa Miskell has been contracted to design and construct the replacement viaduct as well as deconstruct the existing viaduct in a carefully planned and unique methodology necessary due to the highly populated area surrounding the viaduct. The methodology and sequencing have been selected to keep State Highway 1, a vital link in the country's motorway network and surrounding infrastructure, fully operational during the four-year viaduct replacement period (NZTA, 2009). The project started in October 2008 and has an expected completion date of December 2012. The total project budget is $NZ210 million.

The Newmarket Viaduct Replacement Project has a projectized organizational structure, where the project manager has a high degree of authority to assign projects, resources, and direct the work of the personnel appointed to the project. The project has a flat organizational structure, where the construction engineering teams are responsible for subdivided sections of the project by the project work breakdown structure (WBS). Key project teams include the permanent works design team, which is structurally separated from the construction team reporting directly to the project manager, the temporary works design team, and the construction engineering teams that report to the construction manager who in turn reports to the project manager. The site supervision team has a separate organizational structure that, although has linkages to the construction engineering team, formally reports directly to the project manager. A simplified organizational structure of the Newmarket Viaduct Replacement Project is illustrated in Figure 1.

Project organization structure

Figure 1: Project organization structure.

Large construction engineering projects, as in the case of the Newmarket Viaduct Reconstruction project, can be considered as a program comprised of several sub-projects; in the case of the Newmarket Viaduct Reconstruction Project, the project can be considered as a program comprised of the new viaduct construction, the deconstruction of the existing viaduct, and a road construction project. A simplified WBS of the project is represented in Figure 2. In the case of the Newmarket Viaduct Reconstruction Project, each construction engineering team is effectively responsible for one of the project level deliverables.

Project work breakdown structure

Figure 2: Project work breakdown structure.

Sources of Data and Data Collection

The target population in this study has been the project population of the project middle management construction staff who is responsible for the management and execution of the project. This sample includes the construction engineering teams, the temporary and permanent works design teams, the site supervision team, and selected members from project controls staff. The study has been carried out by means of a questionnaire aimed at measuring the perceptions of the key staff in terms of their own capabilities and performance with respect to organizational structure, organizational context, ambidexterity, and team performance.

Due to the relatively small team on the project, the data sampling has been an attempt to sample the entire population by means of a team census. By attempting to capture the entire project middle management team in the study, several sampling errors have been minimized and defined by Wild and Seber (2000) as follows:

- Selection bias, where the population being sampled is an unrepresentative subgroup.

- Non-response bias, where part of the population fails to complete the questionnaire.

- Self-selection, where people themselves decide whether or not to be surveyed.

Questionnaire questions for the data analysis have been carefully selected from successful past studies. By using carefully selected questions adapted to meet the context of the project question effects, where questions are slanted or loaded to help obtain the desired results have been minimized.

To minimize survey format effects where respondents reply to sensitive questions dishonestly due to lack of confidentiality in the data collection process (Wild & Seber, 2000), an online survey system was deployed, where confidentiality of response was guaranteed and respondents could remain completely anonymous if they wished. Same source bias effects were also minimized by distributing the questionnaire to different levels and roles within the project construction team.

The questionnaire was sent to 52 construction middle management staff from various levels in the project hierarchy and various project roles. All personnel selected were directly responsible for delivering both efficiency on the project and for delivering innovative development at the project team level. A total of 38 valid responses were collected, with most responses collected from office-based staff; a lower response rate was collected from staff members who were more site oriented, such as supervisors and foremen or staff who were no longer engaged full time on the project, as in the case of some of the design engineering staff. The distribution of questionnaires and the response distribution are graphed in Figure 3.

Distribution of invited participants and responses

Figure 3: Distribution of invited participants and responses.

The respondents to the questionnaire were each asked basic questions regarding their experience levels to help develop a profile for the project team and ensure that the respondents’ time on the project and in the industry was sufficient to ensure that quality of data received. Of the 38 respondents, the average duration on the project was two years, with the project under way for five years and the execution phase of the project under way for three years. The average duration in the industry for the respondents was 12 years and the average number of large-scale infrastructure projects (defined as a civil works project of value greater than $NZ50 million) completed was three. Table 1 shows the basic statistics on the respondents to the questionnaire.

Duration on project (years) Number of years experience in the construction industry Number of similar scale projects completed ($NZ50M+)
Minimum 0.2 1.0 0.0
Maximum 5.0 35.0 15.0
Average 2.0 11.7 3.5
Median 2.0 10.5 3.0
Standard Deviation 1.3 7.9 3.2

Table 1: Respondent profile.

To get an overall picture of where the project priorities lie, each respondent was asked to rank six project aspects to give an overview of where the project priorities were at the time of the study. Each respondent was only allowed to assign each number once, forcing the respondents to rank each aspect with a different value. A few of the respondents left optional comments that the rankings indicated were a very “within a very tight banding,” suggesting that all aspects were perceived of high importance to the team. The project aspects were ranked as follows, with 1 being the highest priority and 6 being the lowest project priority.

Table 2 shows the basic statistics of the how each respondent ranked each aspect of the project.

Project efficiency Project innovation Project risk Project safety Project quality Project schedule
Minimum 1.0 4.0 1.0 1.0 1.0 1.0
Maximum 6.0 6.0 6.0 4.0 6.0 6.0
Average 3.6 5.3 3.4 1.5 3.9 3.2
Median 4.0 5.0 3.0 1.0 4.0 3.0
Standard Deviation 1.3 0.7 1.5 0.9 1.6 1.6

Table 2: Project aspect priority rankings.

Construct Measurement and Reliability Assessment

To ensure content validity, this study builds upon and utilizes instruments validated in previous studies as much as possible, including the work of Jansen, Ven Den Bosch, and Volberda (2006), Jansen, Templelaar, Ven Den Bosch, and Volberda (2009) and Gibson and Birkinshaw (2004). A new instrument has been developed to capture two dimensions of temporal separation. First, the extent to which the project team pursue temporally separated periods of exploration followed by periods of exploitation has been measured (Gibson & Birkinshaw, 2004). Second, the degree to which the project team are driven by the management team, project systems, and project environment to temporally separate activities of exploration and exploitation has been measured (Nickerson & Zenger, 2002). The constructs and corresponding questions, source and Cronbach's alpha are tabulated in the Appendix. All questions measuring the constructs are measured on a seven-level Likert scale anchored from 1, strongly disagree, to 4, neutral, and 7, strongly agree. Exploratory factor analysis has been used to check convergent versus discriminant validities. All constructs utilized have eigenvalues greater than 1 with minimum factor loadings of 0.4 (Straub, Boudreau, & Gefen, 2004). Cronbach's alpha is used to gauge construct reliability with a 0.6 cut off for exploratory studies (Straub et al., 2004).

All constructs utilized in this study have returned a Cronbach's alpha, ranging between 0.75 to 0.91 except for Temporal Separation with an alpha of 0.68 marginally below the 0.7 cut off for confirmatory studies but well above the cut off of 0.6 for exploratory studies (Straub et al., 2004). The descriptive statistics of the constructs and the bi-variate correlations among them are reported in Table 3.

There are two composite constructs used in this study: ambidexterity and project context, each having two sub-constructs. A review of methods of construct aggregation indicated that alternative methods for aggregation utilized in previous studies include addition (Jansen et al., 2006; Lubatkin et al., 2006), multiplication (Gibson & Birkinshaw, 2004) and subtraction (He & Wong, 2004). A test on the suitability of aggregation methods carried out by Jansen et al., (2009) by means of regression testing of the alternatives utilized by studies for aggregation of results found that addition of exploration and exploitation constructs represented the actual value of the effect of exploration and exploitation on performance better than alternative methods of aggregation that have been utilized in some other studies. The construct for ambidexterity for this study hence has been conceptualized and developed as the addition of exploration and exploitation constructs as carried out by Lubatkin et al. (2006) and Jansen et al. (2009). The construct for project context has also been taken as the sum of social context and performance context in contrast to the original Gibson and Birkinshaw (2004) study, which conceptualized these effects as multiplicative.

Mean Standard Deviation 2 3 4 5 6 7
1 Exploration 4.78 1.17 .56** .69** .17 .40* .66** .57**
2 Exploitation 4.43 1.15 .82** .27 .57** .52** .53**
3 Performance 4.51 1.10 .13 .55** .60** .57**
4 Structural Separation 4.50 1.20 .26 .14 .11
5 Temporal Separation 4.55 1.10 .40* .34*
6 Performance Context 4.59 1.07 .71**
7 Social Context 3.94 1.22

**. Correlation is significant at the 0.01 level (2-tailed).

*. Correlation is significant at the 0.05 level (2-tailed).

Table 3: Mean, standard deviation, and bi-variate correlations.

 

Analysis

Regression analysis has been adopted to analyse the data and test the hypotheses. The analysis process for mediator hypothesis has been used to test the Hypotheses. Figure 4 describes the testing process of mediated relationships. Mediator effects are effects that explain how external events take on psychological significance; mediator variables explain why certain effects occur (Baron & Kenny, 1986).

Barron and Kenny procedure for testing for mediation

Figure 4: Barron and Kenny procedure for testing for mediation.

Figure 4 shows the first step of detecting a mediated relationship is to test if the independent variable x has a significant direct effect on the dependent variable y, shown as relationship 1. If true, then in step 2, the independent variable x should also have a significant direct effect on the mediator variable m, shown as relationship 2 and the mediator variable m should have a significant direct effect on the dependent variable y, shown as relationship 3. In the final step, the dependent variable is regressed on both the independent variable and the mediator. If relationship 1 becomes much weaker or even non-significant, while relationship 3 remains strong and significant, then the effect of the independent variable x on the dependent variable y is said to be mediated by the mediator (Barron & Kenny, 1986).

Collinearity results in increased estimates of parameter variance, which can result in misleading results. Variance Inflation Factor (VIF) provides a reasonable indication of the effects of multicollinearity on the variance of regression coefficients (O'Brien, 2007). A VIF of 10 or more is a clear indication of multicollinearity. In cases where the sample size is relatively small a VIF of more than 4 has been used as the cut off (O'Brien, 2007). The maximum VIF of 2.356 from the regression analysis in this study indicates no or negligible multicollinearity threat to the findings.

Results

The regression analysis results for each of the hypothesis are presented next.

Hypothesis 1 is supported. Table 4 reports that ambidexterity has significant effect on performance (β=0.85, p≤0.00).

Table 4 also reports the results of testing the mediated effect of temporal separation on project performance by ambidexterity. The results show that temporal separation has direct, significant effects on performance (β=0.55, p≤0.00) and ambidexterity (β=0.55, p≤0.00), respectively. Ambidexterity also has direct, significant effect on performance (β=0.85, p≤0.00). When performance is regressed on both temporal separation and ambidexterity, the effect of ambidexterity remains high and significant (β=0.79, p≤0.00) while the effect of temporal separation becomes non-significant (β=0.12, p≤0.27). Following Baron and Kenny (1986), Hypothesis 2 is supported—the effect of temporal separation on project performance is mediated by ambidexterity.

R Square Dependent Variable Independent Variable Beta T(sig.) VIF
0.30 Performance Temporal Separation 0.55 3.93 (0.00) 1.00
0.30 Ambidexterity Temporal Separation 0.55 3.93 (0.00) 1.00
0.73 Performance Ambidexterity 0.85 9.77 (0.00) 1.00
0.74 Performance Temporal Separation 0.12 1.12 (0.27) 1.43
Ambidexterity 0.79 7.59 (0.00) 1.43

Table 4: Results for Hypotheses 1 and 2.

Table 5 shows the analysis results for Hypothesis 3. The results show that structural separation has non-significant, direct effects on project performance (β=0.13, p≤0.45) and ambidexterity (β=0.25, p≤0.13), respectively. Following Baron and Kenny (1986), Hypothesis 3 is not supported—structural separation does not have significant effects on either ambidexterity or project performance.

R Square Dependent Variable Independent Variable Beta T(sig.) VIF
0.02 Performance Structural separation 0.13 0.76 (0.46) 1.00
0.06 Ambidexterity Structural separation 0.25 1.54 (0.13) 1.00
0.73 Performance Ambidexterity 0.85 9.77 (0.00) 1.00
0.73 Performance Structural separation -0.09 -1.03 (0.31) 1.07
Ambidexterity 0.88 9.73 (0.00) 1.07

Table 5: Results for Hypothesis 3.

Table 6 outlines the results of testing the mediated effect of organizational context on project performance by ambidexterity. The results show that organizational context has significant effects on performance (β=0.63, p≤0.00) and ambidexterity (β=0.70, p≤0.00), respectively. Similarly, ambidexterity also has significant effects on performance (β=0.85, p≤0.00). When performance is regressed on both organizational context and ambidexterity, the effect of ambidexterity remains high and significant (β=0.81, p≤0.00) while the effect of organizational context becomes non-significant (β=0.07, p≤0.60). Following Baron and Kenny (1986), Hypothesis 4 is supported—the effect of organizational context on project performance is mediated by ambidexterity.

R Square Dependent Variable Independent Variable Beta T(sig.) VIF
0.40 Performance Context 0.63 4.86 (0.00) 1.00
0.49 Ambidexterity Context 0.70 5.88 (0.00) 1.00
0.73 Performance Ambidexterity 0.85 9.77 (0.00) 1.00
0.73 Performance Context 0.07 0.53 (0.60) 1.96
Ambidexterity 0.81 6.54 (0.00) 1.96

Table 6: Results for Hypothesis 4.

Table 7 provides further analysis results for both Hypotheses 1 and 3. The results show that temporal separation and organizational context have simultaneous significant effects on performance and ambidexterity. When performance is regressed on organizational context, ambidexterity and temporal separation, the effect of ambidexterity remains high and significant (β=0.75, p≤0.00) while the effects of both temporal separation and organizational context (β=0.06, p≤0.62) become non-significant. Following Baron and Kenny (1986), both Hypotheses 1 and 3 are supported—the effect of organizational context and temporal separation on project performance are mediated by ambidexterity. The added insight here is that even after controlling for each other's effects, the results still support the hypothesis that both organizational context and temporal separation on project performance are mediated by ambidexterity.

R Square Dependent Variable Independent Variable Beta t (sig.) VIF
0.50 Performance Context 0.49 3.76 (0.00) 1.19
Temporal Separation 0.36 2.72 (0.01) 1.19
0.58 Ambidexterity Context 0.57 4.77 (0.00) 1.19
Temporal Separation 0.32 2.67 (0.01) 1.19
0.73 Performance Ambidexterity 0.85 9.77 (0.00) 1.00
0.74 Performance Context 0.06 0.50 (0.62) 1.96
Temporal Separation 0.12 1.10 (0.28) 1.03
Ambidexterity 0.75 5.53 (0.00) 2.36

Table 7: Results for Hypothesis 3.

Discussion

This section summarizes the findings, discusses implications for theory development and for practice, and the limitations and future research directions.

Summary of Findings

Adding to literature, this study finds that ambidexterity at the project team level is a significant contributor to project performance; the results further show that a properly nurtured project context that empowers individuals to make exploit versus explore decisions (Gibson & Birkinshaw, 2004), directly impacts a project team's ambidexterity; temporal separation has been found to be a significant predictor of project team's ambidexterity, whereas positive effect of structural separation on ambidexterity at project level as predicted by the organizational literature has not been supported.

Implications for Theory Development

The first contribution from this study is that ambidexterity at the project team level has a significant impact on project performance. Previous studies have mostly focused on organizational level ambidexterity. This is the first study to have identified a link between project ambidexterity and project performance.

The second major contribution to literature is the finding that temporal separation has a significant effect on project performance, which is fully mediated by ambidexterity. In other words, temporal separation facilitates ambidexterity, which impacts performance. The concept of punctuated equilibrium describes organizations oscillating between periods of exploration and exploitation (Gibson & Birkinshaw, 2004). At the project level, it is necessary to identify the solutions for the project before implementation. Therefore, temporal separation of exploration and exploitation is a natural occurrence in projects. This shifting between exploration and exploitation could be iterative for projects with emerging requirements. As project risks diminish with the progress of the project, the overall focus on exploration activities is likely to decrease while the focus on exploitation activities is likely to increase.

The third contribution to literature is the finding that structural separation does not lead to ambidexterity in the project team context, contrary to the predictions from organizational literature on ambidexterity (Gupta et al., 2006). This difference may be due to the fact that project teams are often collocated and multi-tasked as in the case project. Further, time pressure and the typical focus on project delivery demand horizontal coordination, which breaks down the walls between various parties involved in a project. As a result, there is simply no effective mechanism for structural separation within the project team. In very large programs that last decades such as the Apollo program, where project teams are functionalized, there may be a case for structural separation.

The final contribution is the finding that project context has an impact on project performance, which is fully mediated by ambidexterity. This is the first study to have provided evidence on the effects of project context explicitly on project team ambidexterity and subsequently on project performance.

Implications for Practice

Research on organizational ambidexterity has, to date, been focused mostly on organizational structure as a means of achieving separation between exploration and exploitation with the aim of achieving organisational ambidexterity. Despite structural separation in the case project between project teams and separate responsibilities on exploration and exploitation activities, respectively, it is temporal separation that drives team ambidexterity and thus team performance along with project context. The implication for practitioners is, in a project setting, planning efforts should be focused on temporal separation of exploration and exploitation tasks while avoiding structural separation of teams as a primary means of achieving ambidexterity. It is critically important for project teams to work closely together, often collocated, with team follow through on work packages through different phases to effectively integrate exploration and exploitation through retention of project specific knowledge.

Figure 5 illustrates how individual project teams could cycle through exploration and exploitation activities as they progress from the planning with high levels of exploration to execution with high levels of exploitation. In a program of projects such as in the case of many construction projects several teams will oscillate between exploration and exploitation activities, however, not all at the same time resulting in program continuity, as illustrated.

Temporal separation in the project context

Figure 5: Temporal separation in the project context.

The findings that project context impacts on ambidexterity coincides with temporal cycling between exploration and exploitation. It is unlikely that empowered individuals are actually driving this temporal cycling. Instead, the cycling is driven by the necessity of the natural progression from design to implementation and the internal project cycles of a project as it moves through its life cycle phases and through the execution of project work packages. The effect of context on ambidexterity found in this study is likely due to its effect on directing people's attention to achieving the best outcome for the project. Such awareness facilitates people to transition from exploration mode to exploitation mode and vice versa—a transition often hampered by inertia in the current mode (Gupta et al., 2006).

Therefore, it is critical for the project organization to nurture an environment that is conducive to temporal cycling of exploration and exploitation. In fostering such an environment, it is important to strike a balance between performance management and social context—i.e., setting out performance expectations while empowering individuals to freely switch between exploration and exploitation as needed. Project senior managers can play significant roles in developing an effective project context that facilitates the temporal cycling between exploration and exploitation activities. Given the difficulty of shifting from an exploitative mindset to an exploratory mindset (Gupta et al., 2006), it is important that senior managers are aware of the pitfalls of temporal separation and work towards developing a project context that facilitates the transition and the effective integration of exploration and exploitation. Integration, a critical part of achieving ambidexterity (Andriopoulos & Lewis, 2009), can be achieved by using the same team to follow through work packages throughout the project's life cycle.

Senior management must find a balance between setting team performance expectations and developing an effective social context to assist project team members temporally switch between exploration and exploitation activities. For example, projects could put in place a process and set targets for the development of project solutions along with a peer review process for project solutions. Such procedures will assist project team members overcome the learning and psychological difficulties encountered when rotating from project exploitation to project exploration and also ensure feedback for the proposed project solutions.

Validity Threats and Future Research Directions

The marginally low Cronbach's alpha for temporal separation poses a threat to the validity of the construct and findings (Straub et al., 2004). However, considering the suggested cut off for exploratory studies of 0.6 for exploratory studies (Straub et al., 2004) and the exploratory nature of this study, the Cronbach's alpha of 0.68 is acceptable. The threat is further mitigated by the good convergent and descriminant validity from exploratory factor analysis, good Cronbach's alpha for most constructs and the use of instrument validated in previous studies.

This study is conducted in the setting of a specific engineering project. Therefore, factors that vary across projects such as uncertainty, complexity, and project size are not included in the investigation. These factors could well have impacts on the relationships identified. Future cross-sectional studies should control for these variables.

The generalizability of the findings is limited due to the nature of a single case study. Future studies should investigate the relationships in varied project settings. Large-scale survey studies will be possible after we have gained sufficient understanding at the project level on the antecedents of ambidexterity and its subsequent effects on project performance.

Conclusions

This study, through case study of a complex engineering project, makes four contributions to literature. First, ambidexterity at project team level has a significant impact on project performance. Second, temporal separation has a significant effect on project performance and this effect is fully mediated by ambidexterity. Third, structural separation does not lead to project team ambidexterity, contrary to the predictions from organizational literature on ambidexterity (Gupta et al., 2006). Fourth, project context has an impact on project performance and this effect is fully mediated by ambidexterity.

We believe that among the three antecedents for ambidexterity at the project level as predicted by organizational literature, structural separation is rendered ineffective by the natural cycling between exploration and exploitation in the project environment and the need to integrate the two under time pressure and resource constraints. Effective project context could facilitate the temporal cycling between exploration and exploitation by creating awareness that overcome the inertia in transitioning between the two, and by setting performance expectations and appropriate social context.

References

Andriopoulos, C., & Lewis, M. (2009). Exploitation-exploration tensions and organizational ambidexterity: Managing paradoxes of innovation. Organization Science, 20(4), 696–717.

Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51(6), 1173–1182.

Benner, M., & Tushman, M. (2003). Exploitation, exploration, and process management: The productivity dilemma revisited. Academy of Management Review, 28(2), 238–256.

Birkinshaw, J., & Gibson, C. (2004). Building ambidexterity into an organisation. MIT Sloan Management Review, 45(4).

Brady T., & Davies, A. (2004). Building project capabilities: From exploratory to exploitative learning. Organization Studies, 25(9), 1601–1621.

Cegarra-Navarro, J., G., Sanchez-Vidal, M., E., & Cegarra-Leiva, D. (2011). Balancing exploration and exploitation of knowledge through an unlearning context. Management Decision, 49(7), 1099–1119.

Crossan, M., Lane, H., & White, R. (1999). An organizational learning framework: From institution to institution. Academy of Management Review, 24(3), 522–537.

Gibson, C., B., & Birkinshaw, J. (2004). The antecedents, consequences and mediating role of organisational ambidexterity. Academy of Management Journal, 47(2), 209–226.

Ghoshal, S., & Bartlett, C. A. (1994). Linking organizational context and managerial action: The dimensions of quality management. Strategic Management Journal, 15, 91–112.

Gupta, A., Smith, K., & Shalley, C. (2006). The interplay between exploration and exploitation. Academy of Management Journal, 49(4), 693–706.

He, Z. L., & Wong, P. L. (2004). Exploration vs exploitation: An empirical test of the ambidexterity hypothesis. Organisational Science, 15(4), 481–494.

Holmqvist, M. (2004). Experimental learning process of exploitation and exploration: An empirical study of product development. Organizational Science, 15(1), 70–81.

Jansen, J. J. P., Ven Den Bosch, F. A. J., & Volberda, H. W. (2006). Exploratory innovation, exploitative innovation and performance: Effects of organizational antecedents and environmental moderators. Management Science, 52(11), 1661–1674.

Jansen, J. J. P., Tempelaar, M. P., Ven Den Bosch, F. A. J., & Volberda, H. W. (2009). Structural differentiation and ambidexterity: The mediating role of integration mechanisms. Organization Science, 20(4), 797–811.

Lubatkin, M. H., Simsek, Z., Ling, Y., & Veiga, J., F. (2006). Ambidexterity and performance in small-to medium-sized firms: The pivotal role of top management in behavioural integration. Journal of Management, 32(5), 646–672

March, J. (1991). Exploration and exploitation in organizational learning. Organization Science, 2, 71–87.

Nickerson, J., & Zenger, T. (2002). Being efficiently fickle: A dynamic theory of organizational choice. Organization Science, 13(5), 547–566.

NZTA. (2009). SH1 Newmarket Connection: Viaduct Replacement Project. Retrieved from http://www.nzta.govt.nz/network/projects/project.html?ID=93.

O'Brien, R. M. (2007). A caution regarding rules of thumb for variance inflation factors. Quality & Quantity, 41, 673–690.

O'Reilly, C., & Tushman, M. (2007). Ambidexterity as a dynamic capability: Resolving the innovator's dilemma. Stanford Graduate School of Business, RP No. 1963.

Project Management Institute. (1996). A guide to the project management body of knowledge (PMBOK®guide). Upper Darby, PA: Project Management Institute.

Puranam, P., Singh, H., & Zollo, M. (2006). Organizing for innovation: Managing the coordination-autonomy dilemma in technology acquisitions. Academy of Management Journal, 49, 263–280.

Raisch, S., Birkinshaw, J., Probst, G., & Tushman, M. (2009). Organizational ambidexterity: Balancing exploitation and exploration for sustained performance. Organization Science, 20(4), 685–695.

Straub, D., Boudreau, M., & Gefen, D. (2004). Validation guidelines for positivist research. Communications of the Association for Information Systems, 13, 380–427.

Wang, X. R., Liu, L., & Sheng, Z. H. (2009). Ambidextrous management of a large, complex engineering project with significant innovations: Case study of the Sutong Bridge Project. Industrial Engineering and Engineering Management, 16th International Conference. Beijing.

Wild, C. J., & Seber, G. A. F. (2000). Chance encounters a first course in data analysis and inference. Hoboken, NJ: John Wiley & Sons.

Appendix: Constructs, Instruments, Sources of Reference and Cronbach's Alphas

In this section, we define each construct, the corresponding Cronbach's alpha, the questions used and the source(s) from which the questions in each construct are based upon, respectively. For each construct herein, the question number has been presented in parentheses.

Exploration, α = 0.77, (1) We develop solutions for the project that are completely new to our parent organisations; (2) We proactively seek to take advantage of opportunities that present themselves to the project; (3) Risk concerns force the project team to find innovative solutions to project risks, adapted from Jansen et al. (2009).

Exploitation, α = 0.81, (4) We are constantly improving the project's efficiency and cost effectiveness; (5) We are constantly improving our team's capabilities to undertake similar future projects; (6) The project team constantly improves product quality, adapted from Jansen et al. (2009).

Performance, α = 0.85, (7) This project team is achieving its full potential; (8) The team members of the construction team are satisfied with their own team performance; (9), This project team overall is doing a good job in terms of delivering results for the project; (10) I have been given the opportunity and encouragement to do the best work I am capable of on this project, adapted from Gibson and Birkinshaw (2004).

Structural Separation, α = 0.75, (11) Innovation and production activities within the middle management (includes engineering, supervision, safety, etc.) teams are separated in terms of organisational structure; (12) The middle management teams are clearly separated within our project in terms of organisational structure; (13) We have separate teams devoted to innovative planning and devoted to improving project efficiency (Innovative planning defined as planning resulting in major strategic changes in the project direction), adapted from Jansen et al. (2009)

Temporal Separation, α = 0.68, (14) All teams carry out innovative planning and developing project efficiency, although not at the same time (Innovative planning defined as planning resulting in major strategic changes in the project direction); (15) The project construction team is driven to carry out periods of innovation followed by periods of seeking project efficiency.

Performance context, α = 0.79, “The managers on this project.........” (16) set challenging and aggressive goals; (17) issue creative challenges rather than narrowly defining goals; (18) encourage me to be more focused on doing the job well rather than personal gain; (19) make a point of stretching people; (20) hold people accountable for their performance, adapted from Gibson and Birkinshaw (2004).

Social Context, α = 0.91, “The managers on this project.........” (21) devote considerable effort to develop peers or subordinates; (22) give everyone sufficient authority to do their jobs well; (23) push/keep decisions down to the lowest appropriate level; (24) give ready access to information that others need; (26) work hard to develop the capabilities needed to execute our overall strategy/vision; (27) base decisions on facts and analysis, not politics; (28) treat failure (in a good effort) as a learning opportunity, not something to be ashamed of; (29) are willing and able to take prudent risks; (30) set realistic goals adapted from Gibson and Birkinshaw (2004).

©2012 Project Management Institute

Advertisement

Advertisement

Related Content

  • PM Network

    Digital Pulse member content locked

    In the healthcare arena, doctors, patients and project sponsors are all clamoring for more digital therapies and connected devices. That makes this trend a project hot spot this year, according to a…

  • PM Network

    Going with the Wind member content locked

    By Fister Gale, Sarah Offshore wind projects have long been plagued by high technology and connectivity costs, making initiatives all but impossible without government subsidies. "But in the last few years, we've seen a…

  • PM Network

    Artificial Exterminators member content locked

    By Ali, Ambreen Billions of lines of code are written each year, with many applications built on open-source code shared widely among developers. Therein lies a problem: A single error can ripple into thousands of…

  • PM Network

    IoT Takeover member content locked

    Artificial intelligence (AI) and robotic projects may be grabbing headlines around the globe, but a recent survey of tech leaders shows that it's internet of things (IoT) projects that are most…

  • PM Network

    Familiar Faces member content locked

    By Bishel, Ashley Widespread use of facial recognition technology isn't just for phones anymore. Retail companies are launching projects to drive revenue, cut costs, eliminate theft and improve the customer…

Advertisement

Publishing or acceptance of an advertisement is neither a guarantee nor endorsement of the advertiser's product or service. View advertising policy.