Inter-firm collaboration and project performance
the effects of timing and complexity
Associate Professor of Management
Division of Business and Management
Norwich University, Vermont
Proceedings of the PMI Research Conference 11-14 July 2004 – London, UK
The effective use of knowledge is one of the keys to a successful project (Snider & Nissen, 2003). Galbraith (1977) proposes that a firm needs to design an organizational structure that matches its information processing requirements in order for it to be successful. Ring and Van de Ven (1994) argue that interpersonal relationships are critical in determining the outcomes of inter-organizational ventures. Integrating these and other research streams, this paper explores the factors affecting the transfer of knowledge and the resulting project outcomes. Specifically, I examine the characteristics of the project, the timing of the relations, and the nature of the relationship between the project participants.
The focus of this paper is the commercial construction industry and its management triad of owner, designer, and contractor. I propose that the more complex the characteristics of a project, the greater the need to have early and extensive relations between the project participants. I also propose that a collaborative relationship is beneficial regardless of project complexity.
In the first part of this paper I develop the theoretical rational for my hypotheses. Particular emphasis is placed on research in new product development (NPD) and construction. I follow with a description of the research and the results. The research design follows Yin’s (1989) model for embedded, multiple case studies (Type 4). I conclude with a discussion of the implications and suggestions for future research. The results raise some provocative questions related to the efficacy of planning. The ability to manage change on an ongoing basis rather than the ability to plan appears to be the key to project management success.
New Product Development
NPD literature offers significant insight into the generic project and construction specific issues addressed in this paper. NPD is a project-oriented activity, and the construction of a new entity can be viewed as a unique type of new product. The emphasis on the inputs of the designer, producer, and customer as well as their interaction is particularly relevant to this research with its emphasis on the management triad of designer, contractor, and owner. The challenges associated with optimizing the inputs from all participants in order to create the new knowledge that results in a successful new product are particularly pertinent.
Although this research stream has been primarily intra-organizationally oriented (as opposed to construction’s inter-organizational emphasis) the environment has an inter-organizational flavor resulting from functional and thought world conflicts. In addition, the customer is almost always an external force.
Researchers have committed significant efforts to understanding the factors affecting NPD projects due to the importance in achieving a sustainable competitive advantage (Adler, 1995). Brown and Eisenhardt (1995) review and organize this burgeoning literature, with an eye towards developing a comprehensive model of the NPD process. They identify three streams of research: the rational plan, the communications web, and disciplined problem solving. These streams provide a framework for most NPD studies and inform my research.
The Rational Plan
The rational plan perspective focuses on discovering the factors that are associated with the financial success of NPD processes. Brown and Eisenhardt (1995) synthesize this literature and identify eight factors affecting the performance of the product development process:
1) Team composition
2) Team organization
3) Team process
4) Senior management support
5) Product effectiveness
This perspective argues that ‘...careful planning of a superior product for an attractive market…the execution of that plan by a competent and well coordinated-cross functional team, and…the blessing of senior management’ (Brown and Eisenhardt, 1995, p. 348) will result in successful product development efforts.
The Communications Web
Proponents of the communications web perspective argue that the interaction between team members and outsiders affects the performance of the product development process. Their studies emphasize the importance of gatekeepers as sources of external information to a cohesive product development team. Ancona and Caldwell (1992) and Dougherty (1992) follow in this tradition and explore the composition of the team and its relation to performance. Dougherty (1992) notes that different thought worlds contribute to the introduction of new ideas, but hinders the ability to communicate among cross-functional team members. Brown and Eisenhardt (1995) synthesize four factors that affect performance:
1) The project leader
2) Team composition
3) Team internal communications
4) Team external communications
These studies suggest that a major challenge facing NPD efforts is achieving a balance between the positive and negative effects of diversity.
Disciplined Problem Solving
Disciplined problem solving grew out of studies of the Japanese product development process. This process emphasizes the balance between independent teams, heavyweight project and top managers, and project vision. The importance of cross-functional teams is emphasized as well as the necessity for involvement with a strong supplier network (Clark, 1989). This increase in the scope of the team to involve external as well as internal members may partially explain the emphasis on the heavyweight project manager and project vision (Clark & Fujimoto, 1990) as necessary components of a successful process. These devices keep the negative effects of diversity in check, enabling the team to stay focused on what they are trying to achieve. Brown and Eisenhardt (1995) identify six factors in this stream:
2) Team composition
3) Team organization
4) Team process
5) Project leaders
6) Team managers
All three streams of literature recognize the importance and difficulties associated with knowledge transfer and cross-functional teams. Nonaka, (1994) working at a more fundamental level, asserts a theory of organizational knowledge creation that can inform our thinking on the processes that are accomplished by cross-functional teams. Drawing on Polyani’s (1958) work in the philosophy of knowledge, he identifies knowledge as existing in one of two states, explicit or tacit. Explicit knowledge is that which is codifiable, that is, transmittable in formal systematic languages. Tacit knowledge, on the other hand, resists codification and is rooted in specific contexts and specific actions and is difficult to communicate.
With this understanding, knowledge creation can be seen as efforts to combine and convert various forms of existing knowledge to new knowledge. The NPD effort—characterized by planning, communications, and problem solving—is the process of converting the knowledge embedded in team members into a new form. The creation of new knowledge results from new combinations of experiences. Explicit in Nonaka’s (1994) theory of knowledge creation is the need for teams of individuals, with diverse experiences, as the sources of knowledge creation.
Efforts geared towards involving the customer in the knowledge creation process are viewed as critical. The rational plan, as explained by Rothwell (1977), argues that creating a product that responds to a customer’s needs is best achieved through early involvement with the customer. Brown and Eisenhardt (1995) in their synthesis of the communications web literature, argue that one of the major external communication sources should be the customer. Von Hippel’s (1988) work on the sources of innovation demonstrates the importance of maintaining communication with the customer in designing successful products.
At an operational level, Rosenthal and Tatikonda (1992) describe design tools and practices that can aid in the information processing requirements involved in NPD efforts. These tools can help overlap the phases of product development, allowing reciprocal learning between production and design. These design tools and practices recognize the cross-functional nature of NPD efforts and the resulting thought world differences. These also attempt to convert knowledge into an explicit form that all members of the team can understand.
The difficulties in fashioning an effective NPD effort are evident in the above discussion. Diversity is critical, yet diversity hampers internal communication. External knowledge sources are critical, but difficult to integrate into the process. The tools and practices that attempt to resolve these issues often solve one while increasing another.
Construction Project Management
The NPD literature suggests that project relations can be fraught with pitfalls. The divisions that exist can be significant. The nature of the relationship and the timing and tools employed will have a significant affect on the flow of knowledge and the resulting project outcomes. In the construction industry, with its history of inter-organizational NPD projects, these issues are magnified.
Nam and Tatum (1992) identified ‘disintegration’ (goal conflict and the resulting divergent behaviors) as a significant contributor to the high failure rate observed in the design-construct-manage (DCM) process. Most efforts to address this situation have put forward that much of this conflict is based on the inter-organizational aspect of construction. These are different firms with different goals united by a contract that defines the relationship’s formal organizational structure (Gordon, 1994). This organizational structure could either hinder or facilitate the sharing of knowledge across the discrete members of the project (integration). Nauoum (1994) offers evidence that contracts which facilitate an integrative organizational structure are most appropriate on projects of higher complexity.
One of the main methods for sharing knowledge is the process of planning, and while planning is generally recognized as a positive activity, it does have costs. The significant time and resources expended on many planning activities suggest that maximizing project performance must include the appropriate allocation of planning resources. (Faniran, Love, & Li, 1999). Planning is an activity that should occur throughout the construction process, however, planning that occurs at the beginning of the project is generally viewed as having the dominant impact (Cohenca-Zall, Laufer, Shapira, & Howell, 1994).
Planning is often viewed as an objective exercise with the goal of maximizing outcomes with minimal inputs. Objectivity, however, is a difficult goal to achieve. Polanyi’s (1958) work argues that much knowledge is of a subjective nature, based on beliefs that are rooted in an individual’s value system. The result of this personalized knowledge is clearly seen in the contentious interactions between the various firms engaged in the DCM process. Conflict often seems to be the only constant in a project. Puddicombe (1997) argues that these conflicts are often based on a bias (the firms might argue knowledge) that members of the DCM team bring to the project from past experiences. The expectations of certain individual behavior will extend beyond the individual, become embedded in the firm, and impact the conduct of project activities. However, many of the areas thought to be sources of conflict are in fact areas where there is unanimity. The actual conduct of inter-firm relationships will have a significant impact on project performance (Larson, 1995) by either reinforcing or negating preconceived notions of the motivation of other members of the DCM team.
The discussion above suggests that two important factors impacting the exchange of knowledge and subsequent project performance are the nature of the relationship and the characteristics of the project. The relationship between members of the management triad is referred to as Joint Action and has three dimensions that emerge from our previous discussion. In the model below, Work encompasses two dimensions, the amount of the interaction and the timing of the interaction. Manner refers to the actual conduct of the relationship. High values for Joint Action should result in a high exchange of knowledge.
Just as different firms have different information processing requirements, different projects have differing requirements as to the transfer of knowledge. The characteristics of a project, specifically project complexity, are viewed as determining the project's information processing requirements. This research tests the relationship between the complexity of the project and the need for a high exchange of knowledge. Complexity is a contingent variable defining the match between Work and Performance.
Despite the inter-organizational divisions that exist between members of the project team, relationships are critical in determining the success of a project. The formal and informal conduct of inter-firm relationships (Manner) will have a significant impact on project performance.
I propose that the more complex the characteristics of a project, the greater the need to have early and significant Work. I also propose that a collaborative Manner of work would be beneficial, regardless of project complexity.
Hypothesis 1: High and early use of joint work is positively and significantly associated with high
performance in projects of high complexity.
Hypothesis 2: A collaborative manner of work is positively and significantly associated with high
Exhibit 1: The Full Model
The research was conducted in the industrial/commercial segment of the building industry. The research design followed Yin’s (1989) model for embedded, multiple case studies (Type 4). The unit of analysis for the study was a project. I collected data on 108 building projects, treating each project as one mini-case. The embedded aspect of the design results from the multiple levels of project analysis.
The multiple case approach employs a replication logic that Yin (1989) describes as analogous to that employed in multiple experiments. In this study, each project is considered an experiment in which the effect of different patterns of inter-organizational relations on project performance is examined. The breadth of analysis is more tightly focused and the number of cases much higher than would be considered the norm for a multiple case study. As a result, this study exhibits a strong resemblance to a multiple experiment design.
Multiple informants and multiple methods were used to collect the data. I enlisted firms from each sector of the management triad—the designer, the constructor, and the owner—to act as informants. Information was also gathered via archival analysis. The primary and secondary informants would vary between different members of the management triad and the written record. The written record informed the research via archival analysis of contracts, requisitions, correspondence, and job meeting minutes. The management triad supplied information via self-completed questionnaires and structured and open-ended interviews.
The timing and the degree to which there was joint work between the parties were captured by two constructs: pre-construction and construction work. Both were measured on seven point Likert-type scales. Pre-construction was measured as the average of a single measure of the overall degree of early joint work supplied by the secondary and primary informant plus the average of the use of three specific tools (constructability, budgeting, and value engineering) supplied by the primary informant. Construction work was captured by the average of a single measure of the overall degree of joint work supplied by the secondary informant plus the average of two specific tools (constructability and value engineering) supplied by the primary informant. The degree of inter-rater reliability was high and significant (p < .00).
The manner of the work was the average of the informants’ rating of the relationships between each member of the triad (Designer-Constructor, Designer-Owner, Constructor-Owner). This was measured pre-construction and during construction. Multiple informants were employed and all correlations were significant (p < .00), though these varied in their strength. The manner of work construct was tested with a seven point Likert-type scale that ran from No Collaboration - Arms Length – Collaboration.
The project characteristics were captured by three variables:
1) The average of three scales measured across multiple informants that captured relative product, process, and context complexity.
2) The log of the final square-foot cost of the project.
3) A measure based on the description of the overall complexity of the project. The Cronbach’s alpha for this scale (α = .8176) indicated high reliability.
Operational performance was captured by traditional concerns for cost, quality, and schedule. Cost was captured by the percentage deviation from the final budget and the quality for the cost. Quality was captured by a measure of the absolute quality of the project. Schedule was captured as the percentage deviation from the budgeted time to complete. Quality for the cost and absolute quality were single measures averaged across multiple respondents. There was significant inter-rater reliability (p < .02), however, the correlations were not high. This difference was not a significant concern as the quality constructs were predominately measured according to the average of the primary and secondary informants.
Statistical Models and Results
Multivariate and univariate OLS regression models were employed to perform a moderated regression analysis (MRA) (Sharma, Durand, & Gur-Arie, 1981) that tested for both main and interaction effects. The preliminary analysis of the data showed no association between the cost measure’s deviation from final budget and its joint action. This variable was eliminated from the final analysis, leaving a single scale for each performance construct. A variance inflation factor of less than 1.5 for all variables indicated that multicollinearity was not a concern (Neter, Wasserman, & Kutner, 1990).
In testing Hypothesis 1, which states that project complexity moderates the relationship between joint action and project performance, the MRA required that four multivariate regressions be conducted. Quadrants A and B in exhibit 2 contain the results from testing early and construction work separately. Also in exhibit 2, quadrant C contains the results from testing early work, construction work, and project characteristics in a single model. Finally, quadrant D contains the results from testing the full, hypothesized model.
The early use of joint work (quadrant A) demonstrates no multivariate significance. This is reinforced in the univariate analysis where the cost and quality models are non-significant. The deviation from the schedule model is significant for the early use of the joint work model
(p < .02) and the sign is in the correct direction. The construction work model (quadrant B) demonstrated weak multivariate significance (p < .06). In the univariate analysis, the quality
(p < .01) and cost (p < .05) models are both significant, with signs in the appropriate direction. Schedule is non-significant.
The third model (exhibit 2, quadrant C) included the main effects for all variables. The multivariate significance of this model was very high (p < .00). The univariate analyses of the quality and cost models were also highly significant (p < .00), while the schedule model was non-significant. In both the quality and cost models, the main complexity effect was significant and positive. The Early and Construction work variables just miss significance. In addition, the sign on the early use of joint work becomes negative in both cases. The construction use of joint work maintains a positive sign. The overall schedule model was non-significant, but the early joint work remained significant (p < .05) and with the correct sign.
The fourth model (exhibit 2, quadrant D) tested the main and interaction effects. The model maintains high multivariate significance (p < .00). The univariate analyses of the quality and cost models were highly significant (p < .00), while the schedule model was non-significant. The main effects—early and construction joint work—were non-significant in both the quality and cost models, as was the interaction of complexity and early joint work. Complexity and the interaction of complexity and the construction joint work were significant with the sign in the right direction in these models. The adjusted R2 for both models also increased with the full model. In terms of schedule, only early work demonstrated significance.
Exhibit 2: MRA for joint work and performance
The MRA for Hypothesis 1 indicates a fairly complex set of relationships. Complexity, which was hypothesized to be a pure moderator, is found to be a quasi-moderator. It interacted with the predictor variables as well as affecting the dependent variable individually. Construction joint work had a positive effect on cost and quality as an individual construct, but its impact was best understood in relationship to the complexity of the project. The early use of joint work had no effect, either individually or in a moderated context, on cost or quality. However, it is seen to have a positive affect on deviation from the schedule without regard to the project characteristics.
Hypothesis 2 stated that the manner of work was positively related to the outcome regardless of project complexity. This hypothesis was supported for the manner of joint work during the construction process but not for the manner of early work. Quadrants A and B in Exhibit 3 detail the results for early and construction manners of work. The early manner of work is non-significant for both the multivariate and univariate tests. The construction manner of work is significant in the multivariate analysis (p < .02), as are the univariate analyses for quality ( p <.01) and cost (p < .01). The schedule model just misses significance ( p <.08). Combining the two constructs in quadrant C of exhibit 3 increases the univariate R2 but decreases the significance. The multivariate significance is decreased.
Exhibit 3: The manner of work and performance
Quadrant D of exhibit 3 details the results of testing the interaction of manner and complexity. The models for quality and cost are significant, but this is attributable to the addition of the complexity variable. The interaction effect is non-significant in both models. It appears that the manner of construction joint work is significant without regards to project complexity. Exhibits 4 and 5 graphically consolidate the results from Exhibits 2 and 3.
Exhibit 4: Early joint action and performance
Exhibit 5: Construction joint action and performance
This study examined the effect that the timing, the degree, and the manner of the interaction between members of the project team had on project performance. The results suggest a much more complex relationship than was hypothesized. Of particular note is that contrary to my hypothesis, the effectiveness of extensive pre-project planning on complex projects is called into question. Complexity was posited to be a significant force in explaining the effectiveness of group effort. The higher the complexity of the project, the more effective early and copious group effort would be. The research did not support this proposition. Early effort between the parties was associated with superior performance solely in terms of scheduling, but without regard to the complexity of the project. Group effort during the construction phase was associated with superior performance in terms of cost and quality, and this was moderated by complexity.
The importance of timing was also observed in terms of collaboration. Collaboration was not sensitive to complexity, as was hypothesized, but it did have a temporal dimension. The manner of group effort between the parties is non-significant for the early manner of work for all three performance measures. This is not as hypothesized, but it is in keeping with the lack of significance observed with early joint work. The significance of the manner of construction joint work is also consistent with the actual performance of that work. It is highly significant for quality and cost and just misses significance for schedule. Collaboration that takes place during the construction process is most effective.
Complexity appears to be important to the timing of group efforts. However, its effect is contrary to that which was hypothesized. The timing of the joint work affects different performance measures and interacts with the project’s characteristics in different ways. The early and high use of joint work shows a significant relationship with maintaining or improving on the budgeted schedule, but show no relationship with performance related to cost or quality. When complexity is introduced to the schedule model, no additional explanatory power is achieved. The schedule is positively affected by the early use of joint work, regardless of the complexity of the project. The early joint work of the triad members appears to result in a planning process that produces a project schedule that more accurately reflects actual project progress. Early buy-in to the schedule on the part of the triad may ensure that sub-contractors also commit to the schedule, as it could become part of their contract. In addition, the perceived validity of the schedule could induce efforts to perform at a level higher.
The construction use of joint work shows a significant relationship with cost and quality performance, but no relationship with schedule. When complexity is introduced to the model, the interaction between complexity and construction becomes significant and the main effects become non-significant, indicating that complexity does moderate the relationship between the construction use of joint work and performance.
The results suggest that in projects of high complexity, high joint work during construction will positively affect project performance. An approach that focuses on the limits of presentation and the ability to foresee and plan for the future appears to be a critical factor in determining the efficacy of joint action.
The degree of presentation appears to offer a powerful rational for deciding where and when resources should be expended. Simple projects that are highly analyzable exhibit a high degree of presentation. Given the ability to foresee the future, a structure that resolved issues a priori is appropriate. Complex projects exhibit a high degree of uncertainty, making analysis difficult, resulting in low presentation. A structure that resolves problems on an ongoing basis is appropriate.
A number of researchers have presented evidence that this view has validity. Kenney (2003), building on the work of Shenhar and Dvir (1996), observes that in projects of high complexity, with high levels of uncertainty, plans need to be open to change. As a project progresses and new knowledge emerges, flexibility becomes a key to success. Additional support is found in Kim, Ritzman, Benton, and Snyder (1992). In a mathematical programming model, these researchers investigated the effects of the linkages between product and process in different environments. In a continuous production environment, they found that as uncertainty increased, the value of integration diminishes. Presentation as a planning concept is also supported by Adler’s (1995) argument, which states that the optimum phase for coordination will vary depending upon the analyzability of the project. Contrary to the received wisdom, and the premise of this study, that early involvement is beneficial on complex projects, Adler (1995) argues that projects that are simple and highly analyzable are those that would benefit the most from pre-project coordination. Based on a number of case studies, he found that the potential benefits from pre-project planning decreased as product/process fit analyzability decreased. The more complex a project, the more difficult it is to pre-plan and impose a realistic structure. He argues for a coordination mechanism that functions when the problems occur. A structure that is flexible, recognizes the limitations of pre-planning, and allows the opportunity for continuing interaction among the parties is the most effective with complex projects.
One of the limiting factors in this research is its industry specificity and the resulting limitation in the types of complexity modeled. Although A Guide to the Project Management Body of Knowledge (PMBOK® Guide) (PMI, 2000) suggests that project management processes can be applied to all projects, industry characteristics do vary. In the architecture, engineering, and construction (AEC) industry, in general, and in this sample specifically, the range of complexity that is addressed is limited. In most cases, complexity revolves around logistical issues. Although the logistical issues may become extremely complex and cause a high degree of uncertainty, innovations in building projects is largely incremental and architectural (Henderson & Clark, 1990). The result is that the effectiveness of joint action may be different from that which I have portrayed. If the project involves modular or radical innovation, early joint action may even be of limited benefit in achieving schedule goals. Meeting scheduled goals in these types of projects may be best served by maintaining flexibility in the assignment of resources so that adjustments can be made as the actual schedule unfolds.
This research—in conjunction with Kenney’s (2003), Kim et al.’s (1992), and Adler’s (1995)—raises serious questions as to the efficacy of expending resources early in an environment where we do not have a clear vision of the future. Future research needs to explore these findings in a variety of settings. Larger scale empirical research across multiple project oriented industries is now required.
The results of this research suggest that a significant shift in our perception of project management skills may be warranted. Acknowledging uncertainty, rather than trying to plan it away, requires a significant change in the way we attempt to accomplish tasks. The lack of the ability to accurately predict the course of future events suggests that a structure that resolves issues as they emerge, and weights the actual conditions higher than planned, is appropriate. Superior performance, as opposed to being predicated upon planning for the unknown, appears to be linked to recognizing what can and cannot be planned for, then adjusting accordingly.
For most project managers, this is akin to the tightrope walker foregoing a net. The success of the project becomes dependent on the members of the project team, as opposed to the well-executed project plan. The ability to successfully adapt as a project progresses is related to the nature of the relationship between the project members. Collaborative relationships allow the project team to adapt; the= result is high performance. The ability to manage change rather than the ability to plan would therefore become a key project management competency. The project team becomes the net.
Adler, P. (1995). Interdepartmental interdependence and coordination: The case of the design/manufacturing interface. Organization Science, 6(2), 147-167.
Ancona, D., & Caldwell, D (1992). “Demography and design: Predictors of new product development team performance.” Organization Science, 3,(3),321-341.
Brown S., & Eisenhardt, K. (1995). Product development: Past research, present findings, and future directions. Academy of Management Review, 20(2), 343-378.
Clark, K. (1989). Project scope and project performance: The effects of parts strategy and supplier involvement on product development. Management Science, 35(10), 247-263.
Clark, K., & Fujimoto, T. (1990, November/December). The power of product integrity. Harvard Business Review, 107-118.
Cohenca-Zall, D., Laufer, A., Shapira, A., & Howell, G. (1994). Process of planning during construction. Journal of Construction Engineering and Management, 120(3), 561-578.
Dougherty, D. (1992). Interpretative barriers to successful product innovation in large firms. Organization Science, 3, 179-202.
Faniran, O., Love, P., & Li, H. (1999). Optimal allocation of construction planning resources. Journal of Construction Engineering and Management, 125(5), 311-319.
Galbraith, J. (1977). Organization design. Reading, MA: Addison-Wellesly.
Gordon, C. (1994). Choosing appropriate construction contracting methods. Journal of Construction Engineering Management, 120(1), 196-210.
Henderson, R., & Clark, K. (1990). Architectural innovation: The reconfiguring of existing product technologies and the failure of established firms. Administrative Science Quarterly, 35, 9-30.
Kenney, J. (2003). Effective project management for strategic innovation and change in an organizational context. Project Management Journal, 34(1), 43-53.
Kim, J., Ritzman, L., Benton, W., & Snyder, D. (1992). Linking product planning and process design decisions. Decision Science, 23(1), 44-60.
Larson, E. (1995). Project partnering: Results of study of 280 construction projects. Journal of Management in Engineering, 11(2),30-35.
Nam, C. H., & Tatum, C. B. (1992). Noncontractural methods of integration on construction projects. Journal Of Construction Engineering and Management, 118(2), 385-398.
Nauoum, S. (1994). Critical analysis of time and cost of management and traditional contracts. Journal of Construction Engineering and Management, 120(4), 687-705.
Neter, J., Wasserman, W., & Kutner, M. (1990). Applied linear statistical models. Homewood, IL: Irwin.
Nonaka, I. (1994). A dynamic theory of organizational knowledge creation. Organization Science, 5(1), 14-37.
Polanyi, M. (1958). Personal knowledge. Chicago: University of Chicago Press.
Project Management Institute. (2000). A guide to the project management body of knowledge. Newtown Square, PA: Author.
Puddicombe, M. (1997). Designer and contractors: Impediments to integration. Journal of Construction Engineering and Management, 123(3), 245-252.
Ring, P., & Van de Ven, A. (1994). Developmental processes of cooperative interorganizational relationships. The Academy of Management Review, 19(1), 90-118.
Rosenthal, S., & Tatikonda, M. (1992). Competitive advantage through design tools and practices. In G. Susman (Ed.), Integrating design and manufacturing for competitive advantage (pp. 15-35). New York: Oxford University Press.
Rothwell, R. (1977). The characteristics of successful innovators and technically progressive firms (with some comments on innovation research). R & D Management, 7(3), 191-206.
Sharma, S., Durand, R., & Gur-Arie, O. (1981, August). Identification and analysis of moderator variables. Journal of Marketing Research, 291-300.
Shenhar, A. J., & Dvir, D. (1996). Toward a typology theory of project management. Research Policy, 25(4), 607-632.
Snider, K., & Nissen, M. (2003). Beyond the body of knowledge: A knowledge-flow approach to project management. Project Management Journal, 34(2), 4-11.
Von Hippel, E. (1988). The sources of innovation. New York: Oxford University Press.
Yin, R. (1989). Case study research: Design and methods. Newbury Park, CA: Sage Publications.