Application 2 – Annotated Bibliography

Application 2 – Annotated Bibliography

198 IEEE TRANSACTIONS ON ENGINEERING MANAGEMENT, VOL. 57, NO. 2, MAY 2010

An Alternative Taxonomy of Project Management Structures: Linking Project Management Structures

and Project Success Thomas G. Lechler and Dov Dvir

Abstract—This paper addresses the question of how projects are organized and how these management structures impact project success. Despite its widely accepted managerial importance, empir- ical studies could not provide significant evidence of a relationship between implemented management structures and project success. A major problem in finding meaningful empirical evidence is the conceptualization of the structure measure, which is derived from a typologist’s perspective. In this study, we follow the taxonomists’ perspective and empirically develop an alternative taxonomy of project management structures. We empirically compare both ap- proaches, by using two different samples, collected in the United States and Germany, including together over 600 projects. Our empirical findings show that the validity of the widely accepted project organization typology is in question. The use of cluster anal- yses reveals an alternative taxonomy that encompasses five struc- tural types, differentiated by the entities managing them: project coordinator, supervised project coordinator, autonomous project manager, supervised project manager, and autonomous functional project manager. The results strongly support the widely accepted proposition of a relationship between project organization and project success. The emerging taxonomy of project organization configurations enriches the theoretical and conceptual discussions of organizing projects and unravels the multiple aspects involved in organizing the execution of projects.

Index Terms—Cluster analysis, project organization classifica- tion, project organization structures, project performance.

I. INTRODUCTION

It is widely agreed that the choice of management struc- tures used to implement innovative, temporary, cross-functional and complex project endeavors has important implications for project success [2], [11], [67], [69], [74]. The discussion of alternative project management structures dates back to Galbraith’s [17] conceptual introduction of the matrix organi- zation and its differentiation from functional and product orga- nizations. He systematically compared the advantages and dis- advantages of alternative matrix organization structures. Based on Galbraith’s typology some authors favored matrix project organization structures for their flexibility, their economical use of resources, and the clear differentiation between project au- thority and functional authority [18]. Others criticized matrix

Manuscript received May 19, 2006; revised June 5, 2007 and September 27, 2007. Current version published April 21, 2010. Review of this manuscript was arranged by Department Editor Jeffrey K. Pinto.

T. G. Lechler is with Howe School, Stevens Institute of Technology, Hoboken, NJ 07030 USA (e-mail: tlechler@stevens.edu).

D. Dvir is with Guilford Glazer School of Business and Management, Ben-Gurion University of the Negev, Beer Sheva 84105, Israel.

Digital Object Identifier 10.1109/TEM.2010.2044441

project organization structures due to their complexity [12] and lack of unity of command [76]. On balance, these conceptual discussions lack agreement, thus providing little conclusive the- oretical direction concerning the relationship of specific project structures to project success.

These conceptual disagreements are also reflected in the em- pirical research. With the exception of [38] and [39] empirical studies [44], [49], [66], [70] have not generally revealed signifi- cant associations between project organization types and project success. However, an alternative stream of empirical research suggests the importance of project management structures for project success. Several studies concordantly identified signifi- cant and strong associations between project managers’ (PM’s) decision authority and project success [5], [49], [51], [60], [61].

The differences between these two empirical research streams accentuate the inconclusive discussion of project management structures and their association with project success. The prob- lem seems to be related to the unidimensional operationalization of the project management structure variable used in large-scale empirical studies and the assumption of its linear relationship to project success. The limitations of typologies primarily at- tributed to the unidimensionality of the classification scheme, and their inability to cope with real-life, complex, multidimen- sional organizations, have been previously identified (cf., [43] and [71]). We propose that these problems are evident in the widely used unidimensional approach to analyze the impact of project management structures on project success.

In this study, we address the unexplored antagonism between the different research streams. We also attempt to revive the discussions concerned with structural alternatives in integrating project implementation within the context of larger organiza- tions. Following the taxonomist’s perspective [43], we identified five attributes to describe project management structures. We call our alternative approach the “multidimensional approach.” Using data collected from more than 600 projects in the U.S. and Germany, cluster analyses are employed to empirically derive an alternative typology of project management structures. Fi- nally, the typology derived from the multidimensional approach is empirically compared to the commonly used typology derived from the unidimensional approach.

II. THEORETICAL BACKGROUND AND HYPOTHESES

A. Unidimensional Approach

The predominant approach defining alternative forms of project management structures follows Galbraith’s [17]

0018-9391/$26.00 © 2010 IEEE

LECHLER AND DVIR: ALTERNATIVE TAXONOMY OF PROJECT MANAGEMENT STRUCTURES 199

conceptual differentiation between three basic structural types: 1) functional organization; 2) matrix organization; and 3) prod- uct (project) organization. These structural types are derived from a continuum that differentiates the relative influences of the functional and product perspectives on project-related deci- sions. On one extreme, in the functional organization, product managers do not have project decision authority, whereas on the other extreme, within the product organization, they have full decision authority. The matrix organization is positioned be- tween these two extremes. It is superimposed on the functional organization, resulting in a dual authority shared between the permanent functional organization and the temporary project organization.

Several authors extended and specified Galbraith’s model and adopted it to analyze project management structures [44], [49]. The most commonly used typology differentiates between five project management structure types: functional organiza- tion, weak matrix, balanced matrix, strong matrix, and pure project organization [38]. The Project Management Institute (PMI) adopted this typology with a slight modification in its project management standard handbook—The Project Manage- ment Body of Knowledge [53].

Following this unidimensional perspective, a nominal mea- surement scale was developed to empirically analyze the use of project management structures and their association with project success [18], [37], [38] [44], [49]. However, with the exception of Gobeli and Larson [18], [38], empirical stud- ies [44], [49], [66], [70] have not identified statistically signif- icant associations between project management structures and project success. In contrast to these results, some empirical stud- ies have revealed statistically significant associations between the PM’s level of authority and project success [5], [7], [29], [44], [49], [51], [60], [61].

Furthermore, two studies analyzed the impact of project man- agement structures on project success using the unidimensional approach and a separate measure for PM authority [44], [49]. In both studies, significant positive correlations between the level of PM authority and project performance were found. But both studies identified either weak [44] or insignificant [49] correla- tions between project management structures, specified by the unidimensional approach, and project success.

One area of concern lies in the simplifying assumption of a linear relationship between the authority of the PM and func- tional management. Stuckenbruck [68] indirectly questions this implied assumption by pointing out that there are many reasons why it is almost impossible to have a true balance of power be- tween functional management and project management, and that the balance of power is not constant. This suggests that structural types cannot be fully described by proportionally varying the decision authority between these two perspectives. This differ- entiation does not consider possible interdependencies between both constituents, e.g., functional managers could be assigned to implement projects. The problems in linking empirically project management structures with project success, and the controver- sial empirical results to other approaches suggest some validity problems with the generally accepted unidimensional approach as reflected in our first hypothesis.

Hypothesis 1: The unidimensional approach does not appro- priately represent the variety of project management structure configurations.

B. Multidimensional Approach

The measurement problems associated with project man- agement structure configurations seem to follow the general discourse in the organization science literature [71]. Pugh et al. [56], reflecting on Weber’s [75] typology have concluded that “. . .the single bureaucratic type is no longer useful. . .” (p. 115). Carper and Snizek [8] stipulate that the definition of organizational typologies should be based on multiple attributes in order to be meaningful in a practical sense. Furthermore, the attribution of organizations to conceptually derived typolo- gies is often not clear-cut, and therefore, may be problematic for empirical use [43]. The principal problem with the existing project management typology seems to be its unidimensional nature.

Several in-depth case studies of major projects support the multidimensional nature of project management structures [6], [47], [48]. The PMBoK [53] acknowledges three structural at- tributes: 1) the PM’s authority, varying from “little or none” in functional project organizations to “high to almost total” in projectized organizations; 2) the PM’s role, usually part-time in functional organizations, tending to full-time when moving toward a projectized organization; and 3) the PM’s responsibil- ity, in functional organizations usually a coordinator or a leader, sometimes called a “project officer” in a balanced matrix or- ganization and evolving to a “PM” or “program manager” in a projectized organization. These three structural attributes are used to describe the five structural types of the unidimensional approach but they are not used to derive alternative structure types. Other, more recent discussions on alternative project or- ganization structures extend the unidimensional perspective by focusing mainly on the process level [69] or on the organiza- tional level [23].

The linearity assumption fundamental to the unidimensional approach also presents a problem. Galbraith conceptually differ- entiates between alternative organizational structures assuming a linear continuum of project authority from the functional or- ganization to the product organization. This also implies that all different structural attributes, like the PM’s responsibility, au- thority, etc., are behaving in the same linear fashion. The prob- lem with this linearity assumption is evident in the controversial empirical results identified earlier. For example, structural types exhibiting a “low” influence of the project perspective on the project implementation process are not necessarily associated with failure. This “linearity” assumption does not consider po- tential compensating effects on authority resulting from other structural elements.

To overcome the limitations of the unidimensional approach, we suggest a multidimensional approach in order to empiri- cally classify project management structure configurations. An advantage of the multidimensional approach is that possible nonlinear interactions between structural attributes are also taken into account [43]. We follow Galbraith and derive several

200 IEEE TRANSACTIONS ON ENGINEERING MANAGEMENT, VOL. 57, NO. 2, MAY 2010

structural attributes from the perspective of the PM’s formal project position in order to identify basic project management structure types.

1) Managerial Project Authority: A number of scholars have attempted to empirically isolate underlying dimensions of organizational structures [40], [55], [58], [73]. They consider the structure of authority as one of the principal variables describing the level of centralization in organizations. The level of author- ity describes the PM’s locus of influence on project decisions. Some authors extended this perspective and explicitly include the authority over resources [23], [44]. Clark and Fujimoto [9] suggested a typology of product team organizations that differ- entiates between the functional organization that lacks a PM and that relies on the functional organizations to coordinate project tasks, and three other structure types, all utilizing the role of a PM as the focal point of project coordination. The “heavi- ness” of project mangers is largely determined by their level of authority over project team members and over relevant project related areas. An extreme form of the heavyweight PM system is the autonomous project team, where the team is set as a sep- arate organization for the duration of the project [16]. PMs bear a significant responsibility for the well-being of project team members, e.g., depending on the position PMs are likely to have significant control or influence over motivational factors [15]. The PM’s authority over personnel decisions (combined with the authority over technical, financial, and resource decisions) is one of the determining factors of the product (or project) matrix structure type [29], [37], [39]. This influence is partic- ularly important in the context of temporarily assigned project team members. In this case, a mutual incentive responsibility between the project and the functional managers is important, and in the case of full-time project team members, the PMs should be fully equipped with personnel authority. The associ- ation of personnel authority, defined by influence over salaries and promotions, on project success was empirically analyzed by Katz and Allen [29]. Their results suggest that a sole influence of functional managers has a significant negative association with project success, but structures allowing for a mutual influence or even a sole influence of the PM are positively associated with project performance.

Consequently, the authority delegated to the PM concerning project related decisions and the authority over personnel in- centives seem to be important attributes of project management structures that are also directly related to project success.

2) Managerial Project Responsibility: Project authority and responsibility are often seen as similar issues. For example, Larson and Gobeli [37], [39] claim that in the balanced matrix organization responsibilities and authorities for each project are shared between functional managers and PMs. Scott [62], in his review of a half-century of organizational sociology, claims that the archetypical unitary hierarchy has long been the defining characteristic of organizations. A unitary hierarchy is defined by the responsibility that every manager in the hierarchical ladder has over specific subjects. However, most scholarly discussions indicate that project authority and responsibility are two differ- ent attributes. Furthermore, it is recognized that the implemen- tation of project management often involves dramatic changes in both authority and responsibility within the entire organiza-

tion [35]. The PM’s level of responsibility is also an important attribute of heavyweight project teams [10]. We therefore con- sider the project responsibility of PMs as a separate attribute in order to accurately describe a specific, implemented project management structure.

In general, PMs are not independent of the functional or- ganization. Depending on their position, PMs could also bear significant responsibilities within the functional organization. Galbraith’s discussion implies a dichotomy between the role of the PM and the role of the functional manager. In their discus- sion, Clark and Wheelwright [10] explicitly request that PMs of heavyweight project teams should be recruited from senior management positions within the functional organization to as- sure a high level of expertise, experience, and authority nec- essary to successfully implement a project. Furthermore, it is not unusual that PMs are recruited from functional positions that they continue to hold during the project implementation. This dual role also has implications for a PM’s influence on the project implementation. The PMI description of project struc- tures acknowledges the fact that PMs often only work part-time on the project. This aspect is rarely addressed in the literature, which generally treats the role of PMs as agents independent of functional responsibilities. However, in order to measure the structural integration of a project within the context of a larger organization, it is important to understand whether the PM bears functional responsibilities in addition to project responsibilities.

3) Steering Committee: Projects are usually undertaken in a multiproject environment, i.e., in a larger organization ex- ecuting many projects at the same time. When several units must compete for scarce resources, the gains to companywide coordination of resource allocation are likely to be large. Steer- ing committees are adopted to coordinate and monitor multiple projects, to set priorities and allocate resources. They serve as a supervising entity usually consisting of senior managers nom- inated by top management. Steering committees should have the backing of top management and be headed by senior man- agers [13], [42]. Steering committees help integrating the project organization into the functional organization and are an im- portant structural component to assure and to coordinate the involvement of senior managers in the process of project im- plementation. In particular, high management involvement via steering committees could compensate for low PM authority and could act as a balancing power between the project and the functional organization.

They are frequently installed to manage the IS function [14], [30], [50], and were identified as a critical success factor in the implementation of information systems projects [54]. Gupta and Raghunathan [22] empirically confirmed the value of steering committees in avoiding system redundancy, and noted that steer- ing committees have a substantial influence on the achievement of planning goals.

Steering committees are mainly discussed as a structural ele- ment used to implement IT strategies and projects. Their func- tion relative to the management of projects in general is not well understood, although they structurally integrate projects within the functional organization. In this regard, the presence of steering committees poses a challenge to the proposed lin- ear relationship between the PM’s authority and the different

LECHLER AND DVIR: ALTERNATIVE TAXONOMY OF PROJECT MANAGEMENT STRUCTURES 201

structural alternatives used to organize projects. Thus, we see steering committees and their configuration as an important at- tribute of project management structures.

4) Components of the Multidimensional Approach: Our sug- gested multidimensional approach extends the definition of project management structures by integrating three structural components.

PM authorities: This includes the following. 1) The authority delegated to the PM in project matters. 2) The authority delegated to the PM in project-related per-

sonnel issues. PM responsibilities: This includes the following. 1) The responsibility assigned to the PM for project activities

and project outcomes. 2) The hierarchical position and responsibilities of the PM in

functional matters. Top management involvement (via a steering committee):

This includes the following. 1) The level of involvement (supervision, control and coor-

dination) of top management in project-related decisions. We recognize that it is not possible to conceptually derive a

typology of possible project management structures from these suggested five attributes since their specificities and their inter- relationships are not known. Instead, we follow the taxonomy perspective and will empirically identify project management configurations by simultaneously considering these five struc- tural attributes.

Hypothesis 2: The multidimensional (multiattributal) ap- proach will define sets of project structure configurations that differ from the unidimensional approach.

5) Project Structure–Project Success Relation: Most discus- sions of project management structures emphasize that the se- lection of an adequate organizational structure is an important decision for successful project implementation [11], [67]. This argument is supported by those empirical studies analyzing the impact of the previously identified structural variables on project success, e.g., PM decision authority [4], [49], [51], [60], [61] and PM incentive authority [29]. We propose that the relation- ship between structure and performance will be evident if the project configuration typology accurately reflects the complex- ity of actual project structures and, consequently, that the uni- dimensional approach will not be able to explain differences in project performance.

Hypothesis 3: The multidimensional approach will identify different project management structure configurations; some of them will demonstrate significantly better performance than oth- ers.

Hypothesis 4: The unidimensional approach will not differ- entiate project management structures as determining project performance.

III. METHOD

A. Sample and Data

The units of analysis in this study are projects executed within the context of larger organizations. Two different samples were used to identify project management structure types and to as-

sess their impact on project success. One sample of project data was collected in Germany in 1996 and the second was collected between 2001 and 2004 in the U.S. In both cases, a survey was conducted using similar questionnaires designed to measure the impact of project management variables on project success.

In Germany, the questionnaires were sent to the members of the German Project Management Society (GPM, Gesellschaft für Projektmanagement). Each respondent was asked to com- plete two questionnaires, one for a successful project and the other for a failed project. Both projects had to have been com- pleted. This concept of pairwise comparison was first introduced by Rothwell et al. [59] and has the advantage of reducing the personal bias of the key informants. In Germany, a response rate of 43% was achieved, resulting in a total sample size of 448 projects (257 successful and 191 unsuccessful). The lower num- ber of unsuccessful projects can be explained by the experiences of some respondents who reported that they have never been part of failed projects. The respondents had different roles within the projects: 46% were PMs, 27% were core team members (with technical or business related responsibilities), and 27% were ex- ternal consultants responsible for specific project tasks and with intimate knowledge of the project. The projects were imple- mented in different industries, including automobiles, machine tools, software, pharmaceutical, and construction. More than 50% of the projects were related to new product development or software development efforts. The sample provides a fairly rep- resentative cross-sectional distribution of projects carried out in German industry in 1996.

The U.S. sample was gathered with the assistance of project team members and/or PMs. They were asked to select a sin- gle successful or failed project that was recently completed within their organizations, or that was close to completion, with a budget of at least $500 000 and duration of at least six months. These individuals were then given three identical questionnaires, which they were asked to distribute to the PM, a core project team member, and the senior manager responsible for the fund- ing of the project. The questionnaires were completed inde- pendently by the different participants. This sample consisted of 160 projects (112 successful and 48 unsuccessful). In total, data were collected from 160 PMs, 65 senior managers, and 195 project team members from a variety of U.S. companies in financial services (19%), pharmaceuticals (16%), manufac- turing (14%), telecommunications (11%), insurance (4%), and IT (5%). The largest group of projects were new product devel- opment projects (42%) followed by IT system implementation projects (20%), Software- or IT development projects (16%), construction projects (8%), R&D projects (8%), and organiza- tional change projects (5%). The median budget of the projects was $1,200,000.

B. Measurements

The questionnaires used in this study include 199 single items and some quantitative measures of project-specific characteris- tics. Out of these, 67 items were directly taken from Pinto’s [51] questionnaire, with permission of the author. The remaining items were developed for the purposes of this study. Each item

202 IEEE TRANSACTIONS ON ENGINEERING MANAGEMENT, VOL. 57, NO. 2, MAY 2010

was assessed on seven-point rating scales with a range from “strongly agree” to “strongly disagree.” The original question- naire was developed in the German language and the adopted Pinto’s items were translated from English to German. The German questionnaire was pretested and modified after in-depth interviews and responses by a group of experienced PMs. In order to collect data on the U.S. projects the German question- naire was translated into English. To demonstrate consistency and accuracy in translations the documents were back-translated using two experts who discussed the translations and together corrected any inconsistencies. Prior to final data collection in the U.S. the instrument was tested in a pilot of 20 projects.

Following our conceptual discussion we use a set of five vari- ables measuring project management structure attributes and four variables measuring project success. All constructs consist- ing of multiple items were tested for composite validity using Cronbach’s alpha and factor analysis. Some of the initial scales had to be modified to achieve satisfactory composite validity (see the Appendix).

1) Project Management Structure Variables: The scales used to measure the project management structure attributes could be portrayed as follows. The PM project authority scale describes the level of authority delegated to the PM and cov- ers different important decisions typically made or influenced by PMs. The PM functional responsibilities scale measures the position of the PM within the functional organization. The PM personnel authority scale measures whether the PM had the authority to reward project team members. The PM project re- sponsibility scale is used to differentiate the project responsi- bility of the PM from the project responsibility of other con- stituents within the organization. The project influence of steer- ing committees was measured by the level of top management involvement in project decisions.

2) Unidimensional Project Management Structure Scale: We used the scale developed by Larson and Gobeli [38]. This scale is a blend of the scales suggested by Might and Fischer [44] and Murphy et al. [49]. It differentiates between five project management structure types and describes the spectrum from a functional organization to a pure project organization.

3) Project Success Variables: Pinto and Mantel [52] iden- tified three distinct aspects of project performance: 1) the im- plementation process; 2) the perceived value of the project; and 3) client satisfaction with the delivered project outcome. Shen- har et al. [65] have suggested four different criteria to assess project success: 1) meeting design goals; 2) benefits to cus- tomers; 3) commercial success; and 4) future potential. In this study, we use four different project success measures: efficiency, effectiveness, customer satisfaction, and business results (see the Appendix).

C. Statistical Analyses

1) Clustering Procedures: A two-step cluster analysis, as suggested by Ketchen and Shook [32], was performed to iden- tify discrete configurations of project management structures. This two-stage procedure has been followed to improve the quality of solutions when using cluster analyses methods [57].

Many clustering techniques are available and they can gen- erally be divided into hierarchical and nonhierarchical (or it- erative) algorithms. Each category of algorithms has its own strengths and weaknesses in the way that it groups the ob- servations [32], [57]. Complementary clustering methods from both groups (Ward’s and the k-means algorithms) were used to determine the optimal cluster structure. Each cluster anal- ysis step was performed twice, with raw scores and standard- ized (z-scores) values of the model variables. All analyses steps were performed separately for the U.S. sample and the German sample.

In the first step, Ward’s minimum variance method was used to determine the appropriate number of clusters and to pro- duce a starting point for the second cluster analysis step. Ward’s technique is representative of the hierarchical agglomerative algorithms and maximizes intercluster differences. It tends to produce relatively equal cluster sizes and is relatively insensi- tive to outliers [26]. Cluster algorithms do not offer a criterion determining the ideal or “correct” number of clusters, and there- fore, this decision is dependent on the researchers’ judgment as to which solution will be used [32]. We addressed this critical issue by visual inspection of tree plots [1] and by conceptual reasoning. The dendrograms for each solution were inspected by both investigators independently before deciding for the appro- priate number of clusters. Several alternative cluster solutions were discussed and evaluated from a conceptual perspective. The reliability of the final cluster solution was examined by randomly splitting the two samples into two subgroups. Cluster analyses with Ward’s technique were then performed on the re- sulting subsamples to check whether the clustering algorithms were capitalizing on random variation. This technique involves the estimation of the degree of replicability of a cluster solu- tion across a series of datasets. It is a check for the internal consistency of a solution.

The second cluster analysis step is needed to test for the validity of the identified clusters. Hierarchical methods do not allow reallocation of objects that may have been “incorrectly” grouped at an early clustering stage [1]. Therefore, k-means clustering as a nonhierarchical method was used to partition project management structure types. The k-means technique was used because it is an iterative algorithm, shifting objects from one cluster to another until it reaches convergence cri- teria which minimize within-cluster distances and maximize between-cluster distances [27], [63]. The cluster means of the five-cluster solutions obtained in the first analyses steps were used as starting points to perform five-cluster k-means cluster analyses. Although Ward’s method had suggested a five-cluster solution, four- to six-cluster solutions were examined and com- pared with the five-cluster solution. In this step, we also esti- mated the solutions on both samples independently and then compared them in the following steps. The five-cluster solu- tions repeatedly demonstrated the greatest similarities between the different methods and across the two samples. In a final test, the five-cluster solution was tested with a hold-one-out discrim- inant analysis. This analysis showed that in the U.S. sample 94.1%, and in the German sample 92.5%, of the project man- agement structures classified by the cluster analyses remained in

LECHLER AND DVIR: ALTERNATIVE TAXONOMY OF PROJECT MANAGEMENT STRUCTURES 203

the same clusters when reclassified by the discriminant analysis. These findings also supported the five-cluster solutions.

2) MANOVA Procedures: Performance differences between the five clusters were statistically tested by multivariate analysis of variance (MANOVA), using the different performance mea- sures as dependent variables. Bonferroni tests were conducted to identify significant pairwise differences between the project performances levels of the identified project organization struc- ture types [45]. These tests are also seen as another validity test for the cluster solutions [1].

3) CFA Procedures: Finally, a configural frequency analy- sis (CFA) was conducted to analyze the similarities between the unidimensional and our multidimensional approach by com- paring the observed and expected values [72]. This analysis uses assumptions similar to a chi-square test by comparing the distribution of expected and observed response frequencies. The results provide a significance level for the difference between the observed and expected values. When these differences are sig- nificant, the CFA indicates two classifications: types (in which the observed value is significantly higher than the predicted value) and antitypes (in which the observed value is significantly lower than the predicted value). The analysis was conducted using the “HCFA 3.2: a program for Hierarchical Configural Frequency Analysis implemented in R” [21].

In the U.S. sample, we aggregated the data gathered from the 420 stakeholders (PMs, team members, and senior managers) by taking the average of the project scores. This step avoids single- informant bias but the aggregation at the project level had to be tested. We calculated the within unit agreement rWG(j), and the eta-squares [19], [24], [25], [33]. The aggregation tests revealed that the average rWG(j) values for all scales were in the range of 0.95 and 0.84, above the generally acceptable level of 0.70 [19], thus demonstrating within-group agreement. The corresponding eta-square values for the individual variables aggregated to the project level were between 0.89 and 0.77, far exceeding Georgopoulos’ [20, p. 40] minimum criterion of 0.20.

IV. RESULTS

A. Descriptive Statistics

Descriptive statistics and correlations of the model variables are provided in Table I. The correlation patterns across the two samples are quite similar. Only the variable PM functional re- sponsibility exhibits consistently higher correlations with other model variable in the U.S. sample. The similarities between the two samples outweigh the differences and support the com- parison of clustering and performance results in the following steps.

The relationships between the structural variables and the project success variables are presented in Table II. The some- what larger means of the performance variables in the U.S. sample reflect the lower percentage of reported unsuccessful projects (22%) as compared with the German sample (43%). As observed in other empirical studies [44], [49], the PM’s project authority exhibits a significant correlation with all four project success measures in both samples. The relation between the functional responsibility and project success measures is

TABLE I SUMMARY STATISTICS AND CORRELATION COEFFICIENTS

TABLE II CORRELATIONS BETWEEN PROJECT SUCCESS AND ORGANIZATIONAL

VARIABLES

the only noteworthy difference between the U.S. and German samples. This difference may indicate cultural differences in assigning PMs. All other structural attributes and their relations to project success showed high similarities between the U.S. sample and the German sample.

The correlations between the five-point scale of the unidimen- sional approach and the four success measures are also similar across both samples. The only difference is the weak but signif- icant correlation (0.15) between the unidimensional approach and efficiency in the U.S. sample. This result is consistent with the results of Might and Fischer [44].

B. Cluster Analyses

The two clustering procedures (Ward and k-means) resulted in similar configurations in both samples using the original values and the standardized values of the model variables. The simi- larity between the solutions obtained with alternate clustering

204 IEEE TRANSACTIONS ON ENGINEERING MANAGEMENT, VOL. 57, NO. 2, MAY 2010

TABLE III MEANS OF THE FIVE-CLUSTER SOLUTION

algorithms indicates that the cluster solutions are reliable [28]. Two cluster solutions (five-cluster and six-cluster) were further analyzed.

The six-cluster solution contained unevenly sized groups in both samples and two out of the six clusters could be interpreted as subgroups without major differences. In addition, the six- cluster solutions did not show sufficient similarities between the two samples that would permit consistent interpretations. In particular, one cluster differed between both samples, while the other five clusters were highly similar.

The five-cluster solutions provided more balanced cluster sizes and coherent groups within both samples. Also, the five- cluster solutions are similar across both samples, allowing con- sistent interpretations for all clusters. The similarities of the results between the two samples provide sufficient confidence in accepting the five-cluster solution, presented in Table III.

Each cluster represents a specific configuration that is based on the relative levels of the five structural attributes: project authority, personnel authority, project responsibility, functional responsibility, and top management involvement (via a steering committee). They describe the specific position of the PM and are determined by mean values of these variables. We call these structures depending on the attributes of the project managers managing the projects within each structure:

1) project coordinator (cluster 1); 2) supervised project coordinator (cluster 2); 3) autonomous PM (cluster 3); 4) supervised functional PM (cluster 4); 5) autonomous functional PM (cluster 5). The projects in cluster 1 differ from all other projects. The

means of all structure variables (except PM functional respon- sibility) are lower than in all other clusters. The low values for the project responsibility and authority attributes indicate that there is no single agent assigned to implement these projects. It appears that the responsibility for the implementation of these projects is distributed between the participating functional units,

and that the PM’s role in this structure is to coordinate these efforts. This role is also indicated by the low values of the functional responsibilities these PMs have and that they are re- cruited from positions with very low functional responsibilities. Accordingly, we call this type of project management structure the “project coordinator.”

Projects in cluster 2 display relatively low mean values for all but the steering committee variable. The authority and re- sponsibility structure is similar to the structure of cluster 1 with one exception; these projects are coordinated and supported by a steering committee consisting of senior managers. Although the project personnel and functional responsibility variables in the U.S. sample achieve higher mean values and the project respon- sibility achieves a lower mean value than in the German sample the characterization of this structure type is in both samples very similar. Except for the importance of the steering committee the similarities to cluster 1 are substantial indicating that PMs of this structure type also play a coordinating role in project implemen- tation. Therefore, we characterize these projects as managed by a “supervised project coordinator.”

Compared to clusters 1 and 2, projects in cluster 3 exhibit considerable higher levels of project authority and responsibil- ity and also a higher level of personnel authority. Projects in this cluster have relatively low mean values for the role of steering committees. The main difference between the two samples oc- curs in the variable project personnel authority and functional responsibility. The means of both variables are higher in the U.S. sample but they do not affect the main characteristics of this cluster, and are therefore, not relevant for its interpreta- tion. In this configuration, PMs are in a position to implement their projects relatively independent from the functional organi- zation. We therefore characterize the management structure of these projects as being implemented by an “Autonomous PM.”

In clusters 4 and 5, the means calculated for the level of func- tional responsibilities are high indicating that these projects are implemented by functional managers who have simultane- ously to accomplish tasks within the functional organization. Although the variable measuring personnel authority achieved higher means in both clusters, in the U.S. sample these differ- ences do not change their basic interpretation. The difference between the project management structures of the two clus- ters lies in the role of steering committees. PMs in cluster 4 are supported by steering committees. We define this structural con- figuration as projects directed by a “supervised functional PM.”

Similar to cluster 4 PMs in cluster 5 have a high functional responsibility but in contrast to PMs in cluster 4, they are not directly supported by a steering committee; rather they are more autonomous to implement their projects. We character- ize projects in cluster 5 as being managed by an “autonomous functional PM.”

In summary, the differences between the structural variables across the five clusters support our second hypothesis indicating the need for a multidimensional approach to describe project management structures. The empirical results clearly indicate that only the combination of all structural variables allows a differentiation among the identified five structure types. In par- ticular, the combination of the support by steering committees

LECHLER AND DVIR: ALTERNATIVE TAXONOMY OF PROJECT MANAGEMENT STRUCTURES 205

TABLE IV MEDIANS OF PROJECT CHARACTERISTICS

TABLE V (a ) COMPARISON OF SUCCESS MEANS ACROSS FIVE-CLUSTER SOLUTION

(GERMAN SAMPLE)

and the PM’s position in the functional organization support hypothesis 1 and demonstrate that the unidimensional approach does not appropriately represent the variety of project manage- ment structures.

In a further step, we tested the possibility that specific clus- ters depend on situational variables. None of the variables tested could be related to specific clusters. The medians of the project duration and the project team size demonstrate that the occur- rence of specific project management structures is relatively independent of specific contextual settings (see Table IV).

C. MANOVA

Hypothesis 3 follows the general expectation that the choice of project management structures impacts project success. Dif- ferent ANOVA procedures, including Duncan’s multiple range test and multiple comparison tests, were used to analyze whether performance differences exist among the five clusters. The re- sults of all employed tests are quite similar. The results of Bonferroni tests are presented in Tables V(a) and V(b).

The results indicate significant performance differences be- tween several management structures across all four project suc-

cess measures. The performance differences are quite consistent across both samples. Cluster 1 (project coordinator) consistently exhibits the poorest performance results for all four success measures across both samples. A more detailed analysis reveals some differences between the U.S. and the German samples. The mean differences between the U.S. sample and the German sample are related to the higher number of project failures in the German sample but they remain consistent across the five clusters. In the German sample, the highest performance levels are achieved by the project management structures of clusters 2 and 3, but, compared to clusters 4 and 5, the differences are not significant. In the U.S. sample, the supervised functional PM (cluster 4) achieves the highest performance levels across all four success variables, while the supervised project coordinator (cluster 2) achieves the second lowest success ratings. From the standpoint of efficiency, cluster 4 exhibits significantly better performance than cluster 2, and clusters 3 and 4 exhibit bet- ter performance than cluster 2 in terms of business results. In summary, the performance analyses partially support hypothesis 3. Cluster 1 (project coordinator) exhibits significantly inferior performance compared to the other structures across both sam- ples and all four project performance variables.

A similar performance analysis of the five structure types de- fined by the unidimensional approach did not show any signifi- cant performance differences. This result was consistent across both samples and supports hypothesis 4.

D. Comparison of the Unidimensional and the Multidimensional Approach

In the last step of our data analyses, we tested whether the structure types of the unidimensional approach are associated with the structure types of the multidimensional approach.

Table VI represents the final step in testing Hypothesis 1 by directly comparing both typologies. Relationships between the unidimensional and the multidimensional approach were examined with varying assumptions about the distributions and the relations between the two typologies. None of the statistical tests used (Kendall’s Tau, Pearson chi-square) indicated any significant commonalities. The results support our hypothesis since there is only a slight but insignificant resemblance between the two approaches. The configural frequency analyses (CFAs) could not identify any cell that shows significant differences between expected and observed frequencies. In summary, the results support Hypothesis 1 indicating that the unidimensional approach lacks the validity to measure implemented project organization structures.

V. DISCUSSION AND IMPLICATIONS

One of the key questions addressed in this study is to assess why the empirical results of two separate research streams do not generally converge. We suggest an alternative approach to derive a project management structure taxonomy and compare it with the prevalent typology. Our empirical results explain why, in most cases, significant correlations between project manage- ment structures and performance could not be found. As the types resulting from the cluster analyses demonstrate, it is not

206 IEEE TRANSACTIONS ON ENGINEERING MANAGEMENT, VOL. 57, NO. 2, MAY 2010

TABLE VI CROSS-TABULATION OF THE TWO ALTERNATIVE PROJECT

ORGANIZATION TYPOLOGIES

sufficient to differentiate project management structures pri- marily using the project authority of the PM. The integration of project implementation into the functional organization is more complex and cannot be explained by a single attribute, but rather by a combination of several structural attributes. The emerging project management structure taxonomy derived from our suggested multidimensional approach is substantially dif- ferent from the unidimensional approach (see Table VI).

We see two major reasons for this incongruence. First, Galbraith’s conceptual discussion of matrix organization struc- tures differentiates between functional and product organization in order to define their specific roles in the project implementa- tion. This differentiation is important to contrast the advantages and disadvantages of structural alternatives. However, that dis- cussion does not directly address the possibility that PMs could be recruited from the group of functional managers and that they could manage projects in addition to their functional re- sponsibilities. In this case, functional managers do not lose their personnel and positional decision authority. The conceptual sep- aration of functional and project entities could only be observed in two clusters, namely those reflecting the autonomous PM and the supervised project coordinator. Steering committees are an- other organizational concept that overcomes the separation be- tween the functional and the project perspective. For example,

the involvement of senior management via steering committees could compensate for low decision authority and project respon- sibility of PMs, as demonstrated by clusters 2 and 4. These two structural alternatives do not question the conceptual differenti- ation between functional and project organization, but they do have empirical consequences. The underlying assumption that the influence of the functional management vs. project man- agement is a linear function across the assumed five structural alternatives cannot be sustained when the model is augmented by alternative structural components of power distribution.

The second reason lies in the empirical treatment of the com- plex phenomenon of project management structures. A direct translation of Galbraith’s normative perspective, as represented by the unidimensional approach, is not able to capture the spe- cific mechanics used to coordinate between the permanent func- tional organization and the temporary project organization. It is therefore not a surprise that the unidimensional approach can- not empirically differentiate specific authority and responsibility levels and functional responsibilities of PMs.

The multidimensional approach presented in this paper resolves the antagonism present between different research streams and is consistent with published empirical results [4], [49], [51], [60], [61], indicating that the PM’s authority is asso- ciated with project success. However, the different clusters also show that low PM’s project authority could be compensated by the involvement of higher management through steering com- mittees. The PM’s position in the permanent functional organi- zation could be another source for authority, as clusters 4 and 5 suggest. The unidimensional approach does not differentiate between these two sources of project authority. It is therefore not surprising that most empirical studies employing this approach did not find significant associations with project success.

The success relations between the identified five types of the multidimensional approach are consistent with the liter- ature. The strong position of the PM of cluster 3 resembles with heavyweight team structure suggested by [10]. Their “au- tonomous team” structure type could be related with the in- dependent functional PM. The results clearly suggest avoiding the “project coordinator” structure. This form represents the worst case since there is no individual fully responsible for the project’s implementation and the project is basically a free wheel that runs by its own inertia. This type might be comparable to the “lightweight team structure” suggested by [10].

A. Conceptual Implications

Our study suggests that the description of alternative project management structures is indeed consistent with the classic dis- course in the organization science literature. The specific charac- teristics of structural types identified with the multidimensional approach support Carper and Snizek’s [8] request for the use of multiple attributes to derive more realistic definitions of project management structure typologies. Further discussions of project management structures should consider multiple attributes to prescribe the specific mechanisms that coordinate the influence of the project management and the functional organization on project-related decisions.

LECHLER AND DVIR: ALTERNATIVE TAXONOMY OF PROJECT MANAGEMENT STRUCTURES 207

Our results also show that different structural types can- not simply be derived from a linear continuum representing the authority distribution between the functional organization and project management. Moving from an integrated functional project organization toward an autonomous projectized struc- ture [23] does not monotonically increase project performance and success. Each structure type has its specific set of charac- teristics, making it impossible to order them on a continuum along only one dimension, and to suggest a corresponding re- lation with project success. Some structures compensate for a “weak” PM position with a structural integration of senior man- agement via steering committees or the delegation of project responsibilities to functional managers. Consequently, a struc- ture with relatively low PM authority could achieve success levels that are similar to those of a relatively autonomous PM. The significant correlations between the authority of the PM and project success are consistent with this conclusion.

B. Practical Implications

The results of our suggested multidimensional approach indi- cate that the choice of project organizational structure is an im- portant decision that is directly related to project performance. Therefore, attention should be given to this decision before entering the planning phase. Whenever a project is important enough to require top management attention, the “supervised” approach seems to improve performance. However, the main contributor to organizational performance seems to be the level of authority and responsibility delegated to PMs. High perfor- mance levels are achieved when the PM has personnel authority in addition to project authority and responsibility. The high lev- els of functional responsibilities suggest that the PMs should be recruited from functional management positions.

The choice of project management structures will also de- pend on the overall structure of the organizational entity and the specific requirements of the project. If the project could be effectively managed within one functional unit, a functional manager would be the right structural choice. When a project involves several functional units, a “supervised project coordi- nator” or a “supervised functional PM” management structure may be preferred. One guiding principle relevant to all struc- tures is that the PM should be delegated project responsibility and supported with sufficient authority to assure proper project implementation. The one structure that should be avoided under all conditions is the “Project coordinator” type.

VI. CONCLUSION AND LIMITATIONS

This study makes three major contributions advancing our understanding of how projects are organized for success. First, it provides insights into why earlier studies, linking project man- agement structure to project success using the unidimensional approach, have resulted in equivocal results: these studies have relied on an unnecessarily sparse unidimensional differentiation of structures. By introducing the multidimensional approach, that simultaneously considers several structural attributes, this study provides empirical evidence that the choice of a specific project management structure can affect project success.

Second, the results show that there are a number of alternative management structures that can facilitate project success, and that there is one to be avoided. Project structures with lower level PMs lacking appropriate responsibility and authority or senior level support clearly lead to failure. This leads to the conclusion that senior management needs to consider how to integrate PMs more successfully into the organization’s power and authority structures to be successful.

Finally, following a taxonomy approach, we empirically de- rived a typology of five distinct project management structures exhibiting different performance levels over a large sample of projects collected within two different cultures. The results show that the links between project management structure and suc- cess hold across project type, project size, industry and two different cultures. This suggests that the choice of project man- agement structures (PM authority, responsibility, and steering committees) is relatively independent of the context in which it is implemented. Consequently, the findings should influence the further definition of project management standards and training materials.

However, our study raises as many questions as it answers, providing a promising starting point for further discussions and analyses of project management phenomena. Following the three concepts proposed by Walton [73] to define organizational structures, our approach used five structural variables address- ing aspects of centralization and, to some extent, specialization, while formulization was not considered at all. Further studies should extend our measurement model including attributes de- scribing process structure and process integration of projects (cf., [69], [74]), e.g., Winch [74] extends the perspective of project organization structures and integrates the client perspec- tive. The extension of our model with more structural attributes would also further increase the level of content validity [73].

The guiding idea of this study, whether the project organi- zation structure is a real decision requires further study. The design of our study focuses on the single project level and not on the firm level. It is therefore not clear if the identified struc- tural configurations coexist within a single organization or if organizations generally tend toward one specific type. It would be interesting to see how the identified structure types are dis- tributed within a specific organization. Another question in this vein is how the “macro” and the “micro” structures interact, e.g., how do the five attributes of project management struc- tures vary within the project-based organization? The “macro” structures describe the power distribution beyond the PM’s po- sition (e.g., [23]) but they do not consider variations on the “micro” level.

If several structural types are prevalent within a single or- ganization or organizational unit, the factors influencing the structural choice represents a first step to an empirically based contingency approach, and could link the literature on project types [11], [64], [67], with the structural choice. Clark and Wheelwright [10] support this notion by suggesting that “heavy- weight” project teams seem particularly promising in today’s fast-paced competitive environment. The influence of situa- tional variables on the choice of structure is also supported by Hobday [23] and Winch [74]. Hobday suggests that the

208 IEEE TRANSACTIONS ON ENGINEERING MANAGEMENT, VOL. 57, NO. 2, MAY 2010

project-based organization appears to be much better suited to complex systems, which tend to focus on providing unique products, and where direct user involvement can be very signif- icant. Winch [74], discussing mass production and large-scale manufacturing industries, concludes that the choice of project management structures could be influenced by the type of in- dustry; however, both functional divisions and various types of matrix organizations appear ill-adapted to the requirements of advanced production methods. Although these influences might exist, it is not clear whether the choice of a project management structure is executed as an explicit management decision or whether it is implicitly derived from common practice. Still, the “functional” structures are generally used [69], as also suggested by the frequencies of clusters 4 and 5 in both of our samples.

A limitation of our study is that we could not control for structural changes over the life cycle of a project. Some case studies discuss these dynamics and analyze their impact on project success [2], [23]. It is a common practice to change PMs over the project life cycle, and as a consequence, changes in the project management structure seem to be likely, e.g., at project start, a PM is recruited from a functional unit and later replaced by a dedicated PM or vice versa.

The results of this study are a step forward in understand- ing how the implementation of temporary organizational units (projects) is structurally integrated within the context of larger organizations. The five clusters derived herein are robust despite the fact that the data were collected across different industries and different countries.

APPENDIX

OPERATIONALIZATION OF THE USED CONSTRUCTS

All items are measured on seven-point rating scales, Cron- bach’s alpha in brackets.

Success Measures

Scale: Efficiency (alpha: 0.78) The project had come in on schedule. (Pinto) The project had come in on budget. (Pinto)

Scale: Effectiveness (alpha: 0.81) The project met all technical specifications. (Pinto) The results of this project represent an improvement in client performance. (Pinto) The project is used by its intended clients. (Pinto) Important clients, directly affected by the project, make use of it. (Pinto) Clients using this project will experience more effective decision making and/or improved performance. (Own) The project has a positive impact on those who make use of it. (Pinto)

Scale: Customer satisfaction (alpha: 0.90) The clients were satisfied with the process by which this project was completed. (Own) The clients are satisfied with the results of the project. (Pinto)

Scale: Business results (alpha: 0.80) The project was an economic success for the organization that completed it. (Own) All things considered, the project is a success. (Pinto)

Project Organization Measures

Scale: PM project authority (alpha: 0.85) The PM was involved in specifying the project goals. The PM had exclusive authority over the project team re- garding technical aspects. The PM had sufficient authority to make all the necessary decisions to achieve the project goals. The PM had the authority to change objectives in order to achieve the project goal.

Scale: PM personnel authority (alpha: 0.86) The PM had considerable incentive authority to the project team members. The PM had significant influence in assessing the perfor- mance of the project team members. The PM had significant influence in deciding on the kinds of rewards given to the project team members (e.g., pro- motions, bonus, etc.).

Scale: PM functional responsibility (alpha: 0.94) The PM was a high ranking member of the organization. In addition to the project implementation, the PM still had many functional responsibilities.

Scale: PM project responsibility The PM was fully responsible for the project.

Scale: Steering committee top management involvement (al- pha: 0.83)

All major project decisions were supervised by a project control/steering committee. The project control/steering committee consisted only of upper management representatives.

REFERENCES

[1] M. S. Aldenderfer and R. K. Blashfield, Cluster Analysis. Beverly Hills, CA: Sage, 1984.

[2] E. Alsène, “Internal changes and project management structures within enterprises,” Int. J. Proj. Manage., vol. 17, no. 6, pp. 367–376, 1998.

[3] T. J. Allen, D. M. S. Lee, and M. L. Tushman, “R&D performance as a function of internal communication, project management, and the nature of work,” IEEE Trans. Eng. Manag., vol. EM-27, no. 2, pp. 2–12, Feb. 1980.

[4] T. J. Allen, “Organizational structure, information, technology, and R&D productivity,” IEEE Trans. Eng. Manag., vol. 33, no. 11, pp. 212–117, Nov. 1986.

[5] T. J. Allen, R. Grady, and J. J. Slevin, “Project team aging and perfor- mance: The roles of project and functional manager,” R&D Manage., vol. 18, no. 4, pp. 295–308, 1988.

[6] R. D. Archibald, “Organizing the project office and project team: Duties of project participants,” in Project Management Handbook, D. Cleland and W. R. King, Eds. New York: Wiley, 1988, pp. 85–110.

[7] R. Balachandra and J. A. Raelin, “When to kill that R&D project,” Res. Manage., vol. 27, pp. 30–33, 1984.

[8] W. B. Carper and W. E. Snizek, “Nature and types of organizational taxonomies: An overview,” Acad. Manage. J., vol. 5, no. 1, pp. 66–75, 1980.

[9] K. B. Clark and T. Fujimoto, Product Development Performance. Boston, MA: Harvard Business School Press, 1991.

LECHLER AND DVIR: ALTERNATIVE TAXONOMY OF PROJECT MANAGEMENT STRUCTURES 209

[10] K. B. Clark and S. C. Wheelwright, Revolutionizing Product Development. New York: Free Press, 1992.

[11] D. Cleland, “The cultural ambience of project management: Another look,” Proj. Manage. J., vol. 19, no. 3, pp. 49–56, 1988.

[12] S. M. Davis and P. R. Lawrence, Matrix. Reading, MA: Addison- Wesley, 1977.

[13] W. J. Doll, “Avenues for top management involvement in successful MIS development,” MIS Quart., vol. 9, no. 1, pp. 17–35, 1985.

[14] D. H. Drury, “An evaluation of data processing steering committees,” MIS Quart., vol. 8, no. 4, pp. 257–265, 1984.

[15] S. C. Dunn, “Motivation by project and functional managers in matrix organizations,” Eng. Manage. J., vol. 13, no. 2, pp. 3–9, 2001.

[16] J. D. Ellison, K. B. Clark, T. Fujimoto, and Y. Hyun, Product Development Performance in the Auto Industry: 1990s Update. Cambridge, MA: MIT, 2002.

[17] J. R. Galbraith, “Matrix organization designs—How to combine functional and project forms,” Bus. Horiz., vol. 17, no. 1, pp. 29–40, Feb. 1971.

[18] D. H. Gobeli and E. W. Larson, “Matrix management: More than a fad,” Eng. Manage. Int., vol. 4, no. 1, pp. 71–76, Oct. 1986.

[19] J. George, “Personality, affect, and behavior in groups,” J. Appl. Psychol., vol. 75, pp. 107–116, 1990.

[20] B. S. Georgopoulos, Organizational Structure, Problem Solving and Ef- fectiveness: A Comparative Study of Hospital Emergency Services. San Francisco, CA: Jossey-Bass, 1986.

[21] S. T. Gries, HCFA 3.2—A Program for Hierarchical Configural Frequency Analysis for R for Windows. (2004). [Online]. Available: http://people. freenet.de/Stefan_Th_Gries

[22] Y. P. Gupta and T. S. Raghunathan, “Impact of information systems (IS) steering committees on IS planning,” Decis. Sci., vol. 20, no. 4, pp. 17–35, 1989.

[23] M. Hobday, “The project-based organization: An ideal form for managing complex products and systems,” Res. Policy, vol. 29, no. 2, pp. 871–893, 2000.

[24] L. R. James, R. G. Demaree, and G. Wolf, “Estimating within-group interrater reliability with and without response bias,” J. Appl. Psychol., vol. 69, pp. 85–98, 1984.

[25] L. R. James, R. G. Demaree, and G. Wolf, “Rwg: An assessment of within- group interrater agreement,” J. Appl. Psychol., vol. 78, pp. 306–309, 1993.

[26] J. D. Jobson, Applied Multivariate Data Analysis, Volume II: Categorical and Multivariate Methods. New York: Springer-Verlag, 1992.

[27] R. A. Johnson and D. W. Wichern, Applied Multivariate Statistical Anal- ysis, 6th ed. Englewood Cliffs, NJ: Prentice-Hall, 2007.

[28] T. D. Klastorin, “Merging Groups to maximize object partition compari- son,” J. Psychom., vol. 46, pp. 425–433, 1980.

[29] R. Katz and T. J. Allen, “Project performance and the locus of influence in the R&D matrix,” Acad. Manage. J., vol. 28, no. 1, pp. 67–87, 1985.

[30] J. Karimi, A. Bhattacherjee, Y. P. Gupta, and T. M. Somers, “The ef- fects of MIS steering committees on information technology manage- ment sophistication,” J. Manage. Inf. Syst., vol. 17, no. 2, pp. 207–230, 2000.

[31] D. J. Ketchen, Jr., J. B. Thomas, and C. C. Snow, “Organizational con- figurations and performance: A comparison of theoretical approaches,” Acad. Manage. J., vol. 6, pp. 1278–1313, 1993.

[32] D. J. KetchenJr. and C. L. Shook, “The application of cluster analysis in strategic management research: An analysis and critique,” Strateg. Manage. J., vol. 17, pp. 441–458, 1996.

[33] K. J. Klein, A. B. Conn, and J. S. Sorra, “Implementing computerized tech- nology: An organizational analysis,” J. Appl. Psychol., vol. 86, pp. 811– 824, 2001.

[34] H. Knoepfel, C. Gray, and S. Dworatschek, “Projektorganisationsformen: Internationale studie ueber ihre Verwendung und ihren Erfolg,” Proj. Manage., vol. 1, pp. 3–14, 1992.

[35] H. F. Kolodny, “Evolution to a matrix organization,” Acad. Manage. Rev., vol. 4, pp. 543–553, 1979.

[36] E. W. Larson and D. H. Gobeli, “Project management structures: Is there a common language?” Proj. Manage. J., vol. 16, no. 2, pp. 40–44, 1985.

[37] E. W. Larson and D. H. Gobeli, “Matrix management: Contradictions and insights,” Calif. Manage. Rev., vol. 29, no. 4, pp. 126–138, 1987.

[38] E. W. Larson and D. H. Gobeli, “Organizing for product development projects,” J. Prod. Innov. Manage., vol. 5, no. 3, pp. 180–190, 1988.

[39] E. W. Larson and D. H. Gobeli, “Significance of project management structure on development success,” IEEE Trans. Eng. Manage., vol. 36, no. 2, pp. 119–125, 1989.

[40] P. R. Lawrence and J. W. Lorsch, Organization and Environment. Home- wood, IL: Richard D. Irwin, Inc., 1969.

[41] S. Lipovetsky, A. Tishler, D. Dvir, and A. Shenhar, “The relative im- portance of project success dimensions,” R&D Manage., vol. 27, no. 2, pp. 97–106, 1997.

[42] F. W. McFarlan, “Portfolio approach to information systems,” Harv. Bus. Rev., vol. 59, no. 5, pp. 142–150, 1981.

[43] A. S. Meyer, A. S. Tsui, and C. R. Hinings, “Configurations approaches to organizational analysis,” Acad. Manage. J., vol. 36, no. 6, pp. 1175–1195, 1993.

[44] R. J. Might and W. A. Fischer, “The role of structural factors in determining project management success,” IEEE Trans. Eng. Manag., vol. EM-32, no. 2, pp. 71–77, May 1985.

[45] R. G. Miller, Jr., Simultaneous Statistical Inference. New York: Springer-Verlag, 1981.

[46] J. D. Mooney and A. C. Reiley, Onward Industry. New York: Harper & Row, 1931.

[47] W. G. Morris, “Managing project interfaces—Key points for project success,” in Project Management Handbook, D. Cleland and W. R. King, Eds. New York: Wiley, 1988, pp. 16–55.

[48] W. G. Morris and G. H. Hough, The Anatomy of Major Projects: A Study of the Reality of Project Management. New York: Wiley, 1988.

[49] D. Murphy, N. Baker, and D. Fisher, “Determinants of project success,” Boston College, National Aeronautics and Space Administration, Boston, MA, Report, National Technical Information Services, N-74-30392, 1974.

[50] R. L. Nolan, “Managing information systems by committee,” Harv. Bus. Rev., vol. 60, no. 4, pp. 72–79, 1982.

[51] J. K. Pinto, “Project implementation: A determination of its critical success factors, moderators and their relative importance across the project life cycle,” Ph.D. dissertation, University of Pittsburgh, Pittsburgh, PA, 1986.

[52] J. K. Pinto and S. J. Mantel, Jr., “The causes of project failure,” IEEE Trans. Eng. Manag., vol. 37, no. 4, pp. 269–276, Nov. 1990.

[53] Project Management Institute, A Guide to the Project Management Body of Knowledge (PMBOK R© Guide), 3rd ed. Newtown Square, PA: Project Management Inst., 2004.

[54] G. Porter and D. Kohanski, “The MIS steering committee,” Manage. Account., vol. 63, no. 2, pp. 10–12, 1981.

[55] D. S. Pugh, D. J. Hickson, C. R. Hinings, and C. Turner, “Dimensions of organizational structure,” Administ. Sci. Quart., vol. 14, pp. 115–126, 1968.

[56] D. S. Pugh, D. J. Hickson, C. R. Hinings, and C. Turner, “An empirical taxonomy of structures of work organizations,” Administ. Sci. Quart., vol. 14, pp. 115–126, 1969.

[57] G. Punj and D. W. Stewart, “Cluster analysis in marketing research: Re- view and suggestions for application,” J. Market. Res., vol. 20, pp. 134– 148, 1983.

[58] B. C. Reimann, “Dimensions of structure in effective organizations: Some empirical evidence,” Acad. Manage. J., vol. 17, no. 4, pp. 155–160, 1974.

[59] R. Rothwell, C. Freeman, A. Horsley, V. Jervis, A. Robertson, and J. Townsend, “SAPPHO updated—Project SAPPHO phase II,” Res. Pol- icy, vol. 3, pp. 258–291, 1974.

[60] R. Rubenstein, A. Chakrabarti, R. O’Keefe, W. Souder, and H. Young, “Factors influencing innovation success at the project level,” Res. Man- age., vol. 5, pp. 15–20, 1976.

[61] I. Rubin and W. Selig, “Experience as a factor in the selection and per- formance of project managers,” IEEE Trans. Eng. Manage., vol. EM-4, no. 3, pp. 131–135, Sep. 1967.

[62] W. R. Scott, “Reflection on a half-century of organizational sociology,” Annu. Rev. Sociol., vol. 30, pp. 1–21, Aug. 2004.

[63] S. G. Sireci, F. Robin, and T. Patelis, “Using cluster analysis to facilitate standard setting,” Appl. Meas. Educ., vol. 12, no. 3, pp. 301–325, 1999.

[64] A. J. Shenhar and D. Dvir, “Toward a typology theory of project manage- ment,” Res. Policy, vol. 25, no. 2, pp. 607–632, 1996.

[65] A. J. Shenhar, D. Dvir, and O. Levy, “Mapping the dimensions of project success,” Proj. Manage. J., vol. 28, no. 2, pp. 5–13, 1997.

[66] A. J. Shenhar, A. Tishler, D. Dvir, S. Lipovetskey, and T. Lechler, “Re- fining the search for project success factors: A multivariate typological approach,” R&D Manage., vol. 32, no. 2, pp. 111–126, 2002.

[67] C. L. Stuckenbruck, “The matrix organization,” Proj. Manage. Quart., vol. 10, pp. 21–33, 1979.

[68] C. L. Stuckenbruck, “Integration: The essential function of project man- agement,” in Project Management Handbook, D. Cleland and W. R. King, Eds. New York: Wiley, 1988, pp. 56–81.

[69] J. Söderlund, “Managing complex development projects: Arenas, knowl- edge processes and time,” R&D Manage., vol. 32, no. 5, pp. 419–430, 2002.

210 IEEE TRANSACTIONS ON ENGINEERING MANAGEMENT, VOL. 57, NO. 2, MAY 2010

[70] A. Tishler, D. Dvir, A. Shenhar, and S. Lipovetsky, “Identifying critical success factors in defense development projects: A multivariate analysis,” Technol. Forecasting Soc. Change, vol. 51, pp. 151–172, 1996.

[71] D. Ulrich and B. McKelvey, “General organizational classification: An empirical test using the United States and Japanese electronics industries,” Org. Sci., vol. 1, pp. 99–118, 1990.

[72] A. von Eye, Introduction to Configurational Frequency Analysis: The Search for Types and Antitypes in Cross-Classifications. Cambridge, U.K.: Cambridge Univ. Press, 1990.

[73] E. J. Walton, “The comparison of measures of organization structure,” Acad. Manage. Rev., vol. 6, no. 1, pp. 155–160, 1981.

[74] G. M. Winch, Managing Production: Engineering Change and Stability. London, U.K.: Oxford Univ. Press, 1994.

[75] M. Weber, The Theory of Social and Economic Organization (Transl.: by A. M. Henderson and Talcott Parsons). New York: Free Press, 1947.

[76] R. Youker, “Organizational alternatives for project managers,” Manage. Rev., vol. 66, pp. 46–53, Nov. 1977.

Thomas G. Lechler received the Diplom- Wirtschaftsingenieur and Ph.D. degrees in manage- ment from the University of Karlsruhe, Karlsruhe, Germany.

He was the cofounder and CEO of the Vivat- ech GmbH, Germany. From 2003 to 2005, he was a NASA Research Fellow in project management. He is currently an Associate Professor at the Howe School, Stevens Institute of Technology, Hoboken, NJ. His re- search interests include value creation through inno- vation with particular emphasis on the management

of projects and the recognition and exploitation of business opportunities. Dr. Lechler is a member of the German Association for Project Management

(GPM) and the Academy of Management.

Dov Dvir received the B.Sc. degree in electrical en- gineering from the Technion-Israel Institute of Tech- nology, Haifa, Israel, the M.Sc. degree in operations research, MBA, and Ph.D. degrees from Tel Aviv University, Ramat Aviv, Israel.

He was with RAFAEL (Armament Development Authority) as a System Engineer and the IDF as a Manager of a technological center. He was the Head of the Management of Technology (MOT) Depart- ment, Holon Center for Technological Education. He headed the Management Department for two terms at

Ben-Gurion University, Beersheba, Israel. His current research interests include in the areas of project management and entrepreneurship. He is a member of the Project Management Institute and is a member of the Board of the Israeli chapter. He is a coauthor of Reinventing Project Management: The Diamond Approach to Successful Growth and Innovation (HBSP, 2007).

Engage Quality Experts

Excellent Client Support

Get it before Your Deadline


You Need a Professional Writer To Work On Your Paper?

Privacy Policy




 


Do you have an Urgent Assignment?

 


 






 ;
BestQualityExperts
X

Disclaimer: The services that we offer are meant to help the buyer by providing a guideline.
The product provided is intended to be used for research or study purposes only.

©2005-2023 Bestqualityexperts.com  All rights reserved