World Bank projects' critical success factors and their interactions

an empirical investigation

Lavagnon A. Ika, Université du Québec en Outaouais, Gatineau, Québec, Canada

Amadou Diallo, Ph.D. & Denis Thuillier, Ph.D., Université du Québec à Montréal, Montréal, Québec, Canada


Projects are still relevant despite the shift from a project to program approach in international development. However, the abundant project management literature on the critical success factors (CSFs) falls short in paying adequate attention to these specific projects. Although seminal research analyzes the perceptions of National Project Coordinators (NPCs), the perspective of the World Bank Task Managers or Task Team Leaders (TTLs), who are the project supervisors, has not been examined thus far. This paper highlights self-perceptions of TTLs about CSFs and their influences on project success dimensions. Based on a sample of 178 World Bank projects, the interactions between CSFs (monitoring, coordination, design, and training) and project success dimensions (project management success and “deliverable success”) are analyzed. First, results of structural equations modeling show the interactions between the CSFs and their respective influences on project management success and deliverable success, with TTLs ranking the project design, monitoring, and coordination as the most prominent CSF. Second, while project management–related CSFs are positively correlated with one another and have a positive effect on project management success, they fail to show a significant influence on deliverable success. Third, project management success itself does not influence deliverable success. Finally, non–project management but project-related variables such as project duration, budget, and TTL experience do not significantly affect the CSFs.

Keywords: Project success; critical success factors; international development projects; World Bank; structural equations modeling.


There has been early debate regarding the use of a project versus a program approach in international development (ID). Some authors have warned against the rush to condemn projects and to abandon them (e.g., Honadle & Rosengard, 1983). Despite the shift, which began in the mid-1990s, from the long-time prominent project approach to a program approach, the demise of the former, which was predicted more than 40 years ago by Phillip Combs (Honadle & Rosengard, 1983) has yet to happen. In fact, project management is still important under the program approach; for example, projects are still relevant in countries with weak institutional capacity (World Bank, 1998; European Commission, 2007), and specific goals and objectives are met through implementation of projects (Tacconi & Tisdell, 1992, p. 268; Lavergne & Alba, 2003, p. 6). But international development projects are still plagued by deep-rooted problems that would explain their failure rate (Rondinelli, 1976, 1983; Morgan, 1983; Gow & Morss, 1988; de Solages, 1992; Youker, 1992, 1999; Noël, 1997).

Both the fields of project management and ID date back to the 1950s and the 1960s, yet they have grown in parallel. ID has contributed to the wealth of knowledge, with project feasibility studies and evaluations. As well, project management can do more to foster ID. A better management of ID projects is welcomed (Ika, 2005). In that regard, a conceptual framework of project success, success criteria, and success factors is needed in ID (Diallo & Thuillier, 2004, 2005; Ika, 2005; Steinfort & Walker, 2007; Khang & Moe, 2008; Ika, Diallo, & Thuillier, 2010). As a matter of fact, a great deal has been written about critical success factors (CSFs) in the project management literature. As with many academic fields, this literature still shows limitations, three of which are worth mentioning.

First, “Although the multidimensional approach for assessing project success is a common understanding today, most of the project management literature does not differentiate between the impacts of success factors on the various success dimensions”1 (Dvir & Lechler, 2004, p. 3). Second, although there is empirical evidence in the project management literature that CSFs are interrelated, there is a lack of formal studies that analyze the relationships among them (Esteves, Casanovas, & Pastor, 2003; Lechler & Gemünden, 2000). Last but not least, little of this research pays adequate attention either to industry-specific CSF or to nontraditional project management areas (Carden & Egan, 2008).

In particular, very little has been written on ID projects, despite the size of their industry sector (US$100 billion per year), their proliferation, and their questionable outcomes (Themistocleous & Wearne, 2000; White & Fortune, 2002; Crawford & Bryce, 2003; Organization for Economic Co-operation and Development [OECD], 2005; Roodman, 2006; Steinfort & Walker, 2007). Very few contributions address the perceptions of National Project Coordinators (NPCs), the “true” project managers in that specific industry sector2 (Diallo & Thuillier, 2004, 2005; Ika, Diallo, & Thuillier, 2010). The perspective of the NPC counterparts located at the headquarters of the aid agency, in particular, the World Bank Task Managers or Task Team Leaders (TTLs) (who supervise project implementation3 and make sure the guidelines of the World Bank are strictly respected by NPCs and their project implementation unit) has not been examined thus far (Ika, Diallo, & Thuillier, 2009). More specifically, the question of the World Bank projects' CSF is hardly addressed by previous research, and we believe that a careful empirical investigation is required to better understand the World Bank projects' CSFs, the interactions among them, and their influences on project success dimensions.

The first objective was to study CSFs empirically and analyze how a set of World Bank projects' CSFs are interrelated. The second objective was to investigate their influences on two different dimensions of project success (project management success and deliverable success). The third objective was to understand how non–project management but project-related variables affect CSFs.

The next two sections of this paper discuss the theoretical and empirical project management and international development project management (IDPM) literature on the CSFs. In the fourth section,4 we derive the conceptual framework and specify the hypotheses based on that review. In the fifth section, the methodological concerns are discussed, and the conceptual framework is tested with a structural equation modeling. The results are reported in the sixth section. In the seventh section, the theoretical and practical implications of the results are discussed and some suggestions for further research are presented.

Project Success, Success Criteria and Dimensions, and Critical Success Factors in Project Management

Literature on project success has not led to a consensus on a definition or measure of project success. There is a consensus that project success is both project efficiency and effectiveness, that project success is a matter of perspective, that project success is multidimensional, and therefore, that there are project success criteria/dimensions and CSFs (e.g., Shenhar, Levy, & Dvir, 1997; Pinto & Slevin, 1988; Baccarini, 1999; Jugdev & Müller, 2005; Hyväri, 2006; Ika, Diallo, & Thuillier, 2009). While success criteria are characteristics or principles used to assess project success, CSFs are conditions, facts, or circumstances that bring about project success. Many lists of success criteria and CSFs do exist in project management literature and we cannot account for all of them here (for a detailed discussion of project success, see Jugdev & Müller, 2005; Ika, 2009). It is common to distinguish between dimensions such as project management success (project management success or the triangle of time, cost, and quality) or product or deliverable success (e.g., realization of strategic objectives, de Wit, 1988; Baccarini, 1999). Shenhar, Levy, and Dvir (1997) have used four dimensions to assess project success: 1) meeting design goals, 2) benefits to customers, 3) commercial potential, and 4) future potential. Examples of CSFs include: project-mission, top-management support, monitoring, design, context or environment, competencies and experience of the project manager and the project team, training, project size, project duration, project budget, and organization type (Pinto & Slevin, 1988; Belassi & Tukel, 1996; White & Fortune, 2002; Hyväri, 2006). Their relative importance also varies both across research studies, project types, and project phases (Pinto & Slevin, 1988; Hyväri, 2006). Although it is impossible for CSFs to be suitable for all projects, the project management literature on project success, criteria, and CSFs is insightful in IDPM (Diallo & Thuillier, 2004, 2005; Steinfort & Walker, 2007; Khang & Moe, 2008).

“Critical Success Factors” in IDPM

Except for the seminal empirical studies by Diallo and Thuillier (2005) and subsequent work by Khang and Moe (2008), none of the research on ID projects specifically addresses CSFs. Exceptions are those studies that focus on economic or institutional determinants of ID project success (country's growth rate, level of development, governance, inflation, sector, project size, etc.) (Mubila, Lufumpa, & Kayizzi-Mugerwa, 2000). As reported in their paper, however, they neglect project quality, implementation, and management.

Also, Kwak (2002) has outlined 10 internal and external, visible, and invisible CSFs in IDPM. They are: political, legal, cultural, technical, managerial/organizational, economic, environmental, social, corruption, and physical factors; however, this list of 10 CSFs is very general (see Table 1).

There is a wealth of research on the CSFs of public policy implementation in Western countries. CSFs are indeed context dependent (Belassi & Tukel, 1996; Hyväri, 2006), but the CSFs derived from policy implementation projects and programs funded by ID agencies could apply to ID projects in general (see Table 1 for a summary of the CSFs resulting from those few studies). In an ex post evaluation of a technical assistance financial reforms project in Bangladesh, funded by the government and the Department for International Development (DFID), Khan, Thornton, and Frazer (2003) have identified nine reasons5 for its success. Similarly, in an evaluation of a technical assistance project (modernizing public financial management information systems) in Bosnia and Herzegovina, funded by the U.S. Agency for ID (USAID), Vickland and Nieuwenhuijs (2005) have outlined six CSFs. Struyk (2007) has proposed nine success factors in the literature, and looked across the implementation of 18 pilot program projects in Russia, to identify associations between the factors and project success. Once again, these studies (most of them being case studies) fall short of empirically analyzing CSFs, their interactions, and their influences on project success dimensions.

Diallo and Thuillier (2005) have explored the perceptions of the African NPC regarding the relationship between both communication and trust factors and project success and tested their influence on project success dimensions. They confirm that trust and communication between players are proxy variables. They advise that trust between the TTL and the NPC is the foremost important CSF and that team cohesion is the second-most important. Their research results also suggest that trust between the NPC and its national supervisor (often a high-ranking civil servant or the minister himself) seems not to play a prominent role.

Ika, Diallo, and Thuillier (2010) have highlighted the relationship between project management efforts (the investment of the NPC and his or her team in project management tools and techniques) and project success in the aid industry sector. Their results suggest that project success is not significantly affected by the level of project planning efforts; however, a significant correlation does exist between the use of monitoring and evaluation tools and project “profile,” an early indicator of the project's long-term impact.6 Nevertheless, these studies only take into account the perspective of the African NPCs.

The most comparable research study is the one by Khang and Moe (2008). Their 53-item questionnaire was answered by key stakeholders—that is, project managers, team members, funding and implementing agencies, target beneficiaries, and the general public. More specifically, they suggest a conceptual framework of 19 success criteria and 18 CSFs (see Table 1). However, the originality of their contribution lies more in the definition of the criteria and CSFs according to the project life cycle. The authors confirm the ID community consensus that most problems emerge in the project implementation phase but they fail to find significant links between the CSFs and success of each phase of the project life cycle. However, they have shown that the success of each phase has a carry-over effect to that of the subsequent phase, and that consultation CSFs prove to be the most influential on project management success and more important than the following competency CSFs.

All of these research studies are insightful but they do not analyze the influences of CSFs on project success dimensions, nor do they explore the perspective of a key player in ID: the World Bank.

The Conceptual Framework of the Study

The definition and operationalization of the variables and their interrelationships are discussed in this section. The hypothesized relationships between the model variables are represented in Figure 1. The core model is derived from our literature review, and therefore the test of the framework is confirmatory.

Project success has been shown to be multidimensional in ID (Diallo & Thuillier, 2004). Using three project success dimensions, Diallo and Thuillier have found that project management success is by far the most important dimension, followed by project “profile,”7 and that project impact is rated in last position by the African NPCs. In this study, we measure project success with two success dimensions: project management success and deliverable success (Del success in Figure 1).8 Ika, Diallo, and Thuillier (2010) have shown that in the perception of the African NPCs, there is a significant correlation between project management success and deliverable success (which is assessed by project “profile” and “impact”). Thus, we propose:

Hypothesis 1: Project management success significantly influences project deliverable success.

Hypothesized relations between the CSF variables and project success.9

Figure 1: Hypothesized relations between the CSF variables and project success.9

Table 1: Summary of the research on critical success factors for international development projects

Kwak (2002) Khan, Thorton, and Frazer (2003) Vickland and Nieuwenhuijs (2005) Struyk (2007) Khang and Moe (2008) Diallo and
 Thuillier (2005)
Political (inconsistency, instability, war, revolution, import restriction) Flexible project planning Integrated solutions vs. “Best of Breed” Degree and consistency of local leadership Clear understanding of project environment Trust
Legal (changes in laws, currency conversion, lack of appropriate regulatory systems, role of local courts in arbitration) Implementation approach Big Bang vs. incremental implementation Policy characteristics Competencies of designers, planners, and team members Communication
Cultural (differing socio-cultural backgrounds and thought process of actors) Awareness and sense of urgency for change Strong project management Availability of resources Effective consultations with stakeholders  
Technical (use of technology and standards incompatible with project) Publication of success stories Extensive training Number of implementing actors Adequate resources  
Managerial/organizational (bad project management, lack of appropriate processes and resources) Creation of a powerful group of “champions” of change Use of the appropriate individuals from each functional area Attitude of implementing personnel Continuing support of stakeholders  
Economic (changes economic conditions, competition, regulatory changes) Networking and teambuilding Senior manager's understanding of project Alignment of clients Commitment to goals and objectives  
Environmental (pollution, noise, air, water, visual, unsustainable use of natural resources) Anchoring changes in the organization's culture Top-down implementation approach Learning opportunity among implementers and between projects Compatible rules and procedures for project management  
Social (ethnic hostility, religious fragmentation, security of stakeholders, resistance of beneficiaries to new social values) Project management structure   Past experience of implementers Clear policies by donors and recipients to support sustainability  
Corruption (political participation in investment decision making, lack of regulatory institutions) Selecting the right project team   Local environment Adequate local capacities  
Physical (uncontrollable circumstances---natural disasters, wars, coups, acts of terrorism)       Strong local ownership of the project  

The Critical Success Factor Variables

Monitoring, coordination, design, training, and institutional environment are the CSF variables. In fact, monitoring CSF is commented in IDPM (Rakodi, 1982; Morgan, 1983; Rondinelli, 1983; Honadle & Rosengard, 1983; Cracknell, 1988; Binnendijk, 2000; Canadian International Development Agency [CIDA], 2001, p. 20; Crawford & Bryce, 2003; Easterly, 2003, 2007; World Bank, 2004; Ika, Diallo, & Thuillier, 2010). It has been proved that NPCs, with their privileged coordination role, are key players in project success (Analoui, 1989; Diallo & Thuillier, 2004, 2005; Khang & Moe, 2008). Also, the IDPM practitioners would agree that project design is a CSF (e.g., Smith, 1988; Tacconi & Tisdell, 1992; Hulme, 1995; CIDA, 2001, p. 19; Khang & Moe, 2008). A project can fail in spite of the quality of its design and implementation simply due to a poor institutional environment (e.g., Bremer, 1984; Brinkerhoff, 1994). The training factor has also been singled out in the literature (e.g., Taylor, 1995; Jacobs & McLaughlin, 1996; Kealey, Protheroe, MacDonald, & Vulpe, 2005, p. 289).

Furthermore, empirical studies show that the CSFs are generally highly correlated in IDPM (see Diallo & Thuillier, 2005; Ika, Diallo, and Thuillier, 2010). CSFs may be of different interrelated types: related to the project, to the project manager and team, to the organization, and to the external environment (Belassi & Tukel, 1996). Thus, we can assume that a higher level or second-order CSF latent or construct variable—say, CSF related to the management of the project (hereafter labeled PM CSF)10—does exist and is accountable for the preceding five first-order factors altogether—that is, the monitoring, coordination, design, training, and institutional environment CSFs (e.g., Byrne, 2001; Roussel, Durrieu, Campoy, & El Akremi, 2002). In fact, the first four CSFs are related to the project supervisor, the project coordinator, and the project team members (i.e., to project management), and the last one is somewhat outside the control of the management (i.e., to the external environment), even though the management has to cope with it. It is hypothesized, therefore, that the PM CSF factor positively affects project management success as well as deliverable success.

Hypothesis 2a: The CSFs are correlated and there exists a higher-level CSF related to the management of the project (PM CSF).

Hypothesis 2b: Project management success is positively affected by PM CSF.

Hypothesis 2c: Project deliverable success is significantly affected by PM CSF.

Non–Project Management But Project-Related Variables

Experience shows that in IDPM, strategic or large projects are supervised by senior project managers or supervisors. The empirical literature has shown some evidence of the influence of non–project management but project-related variables such as project size on project success (Mubila, Lufumpa, & Kayizzi-Mugerwa, 2000). Yet their influence on CSFs is not analyzed, which is surprising, because CSFs should reflect the project-related variables influence. It can be concluded that the interactions of the CSF with other non–project management but project-related factors and their influences on the various dimensions of project success have not been studied in-depth. As a consequence, non–project management but project-related variables, such as project duration, project budget, and the TTL's experience at the World Bank, were assumed to influence CSF and as a result, project success dimensions.

Hypothesis 3a: Project duration affects CSF.

Hypothesis 3b: Project budget affects CSF.

Hypothesis 3c: TTL experience at the position affects CSF.

As the core model is derived from the literature review, the test of this part of the framework is confirmatory. However, the last three hypotheses are exploratory statements, since only a few studies analyzed non–project management but project-related variables influences on CSFs. There are indications that the non–project management but project-related variables might influence CSFs, but in IDPM the analysis of their effect on CSFs has not been done in detail yet.


Research Design and Data Collection

Data collection results in a sample of 178 World Bank projects. A detailed questionnaire, which was designed to measure the influence of CSFs, was distributed to the World Bank TTLs (project supervisors). Due to their geographical dispersion and mobility, each respondent was asked to fill out a website questionnaire (Dillman, 2000; Porter & Whitcomb, 2007) gathering data on a project that is completed or nearly completed. The perception versus reality debate, particularly in project success research, is one of the utmost importance (Kleinschmidt & Cooper, 1995). This is acknowledged in the following assertion of Likert and Likert (1976, p.165, as cited in Linberg, 1999):

“People act on the basis of what they perceive the situation to be, whether the perceptions are accurate or grossly inaccurate. Since behavior is based on perceptions, the existence of each of them is a fact to be considered. Similarly, the frustrations, attitudes, loyalties, and hostilities felt by each member and the information and misinformation possessed by each particular course of action under consideration.”

This exposes authors to a methodological dilemma: They must choose between studies of self-perceptions (Uhl-Bien & Graen, 1998; Linberg, 1999; Nah, Zuckweiler, & Lau, 2003; Schmid & Adams, 2008; Diallo & Thuillier, 2004, 2005; Ika, Diallo, & Thuillier, 2009, 2010 for the particular cases of ID projects) and the perceptions of others, which are also biased, although not with the same bias (e.g., Fowler & Walsh, 1998; Keil, Tiwana, & Bush, 2002; Gareth & Martin, 2003; and Khang & Moe, 2008 for the particular cases of ID projects). Perceptions are indeed, by their very nature, ontological, biased, and idiosyncratic (Liu & Walker, 1998).

In this research the first alternative is considered, but some important precautions have been taken to reduce the self-perception bias. The overall assessment of project success must be made on a separate page of the Web questionnaire. Only the respondents who skim through the entire questionnaire before answering will know about the subsequent success items that are available on different pages.

In a 3-week period, three electronic reminders (follow-ups)11 were sent to increase the response rate and proven tactics, such as an accompanying personalized letter, source credibility, and the day of the week the letter, for example, Wednesday, were used (Heerwegh, 2005; Porter & Whitcomb, 2007).

The questionnaire included 48 single items (7 for success criteria and 41 for success factors) and some quantitative information about each project. The success criteria items come from different sources: country relevance and beneficiary relevance (OECD, 2002; Khang & Moe, 2008), efficiency/time, efficiency/cost, effectiveness/objectives, and impact (Diallo & Thuillier, 2005) and sustainability (OECD, 2002). The CSFs come either from research work or the help of some experienced project supervisors. Monitoring (Pinto & Slevin, 1988; CIDA, 1999), coordination (Diallo & Thuillier, 2004, 2005; Khang & Moe, 2008), design (Pinto & Slevin, 1988; CIDA, 1999), training (Pinto & Slevin, 1988), and institutional environment (Bremer, 1984; Brinkeroff, 1994).

The final variables in the questionnaire relevant for this study are listed in Table 2 and Table 3: project management success and deliverable success (Table 2); monitoring, coordination, design, training, and institutional environment CSFs (Table 3); and three non–project management but project-related interval (continuous) variables (project duration, budget, and the TTL's work experience at the World Bank, see Table 2). Each item was assessed on a seven-point Likert scale from strongly disagree to strongly agree. Due to their exploratory nature, all the three non–project management but project-related variables were measured with single items. All other constructs were measured with multiple items and were therefore tested with Cronbach's alpha for scale reliability and with confirmatory factor analysis for unidimensionality. All scales achieve a Cronbach's alpha > 0.7 and principal component factor analysis communalities of > 0.55. The observed variables were not significantly skewed or highly kurtotic but they show an often negative skewness and an often positive kurtosis, suggesting a moderate departure from normality. The reverse and logarithmic transformations did not increase normality. Using SPSS descriptive, some outliers such as | Z | > 2 have been detected; however, as they seem apparently connected to the rest of the cases, they were considered a legitimate part of the sample (Tabachnick & Fidell, 2001, p. 71). Using SPSS regression and Mahalanobis distance, no multivariate outliers have been detected (p < 0.001).

The assumptions of multivariate normality, linearity, multicolinearity, and singularity were also evaluated through Analysis of Moment Structures (AMOS). Since the latter program converges, we assume that the covariance matrix was nonsingular. For this study and the final fitted model, there are 178 participants and 20 observed success variables, so the ratio of cases to observed variables is about 9:1. There are approximately 3% to 7% missing values for success criteria items, and 10% to 15% for CSFs. The fact that the AMOS program generates no “Heywood case” (i.e., rogue estimates such as nonsignificant, negative, or null error variances or regression weights above 1) confirms that outliers, normality, linearity, and missing values do not pose any significant problems (Dillon, Kumar, & Mullani, 1987; Anderson & Gerbing, 1988; Roussel, Durrieu, Campoy, & El Akremi, 2002, p. 89).

Response Rate and Sample Characteristics

The data collection effort achieved an overall theoretical response rate of 12.5%,12 resulting in a final sample size of 178 projects. We strongly believe the real response rate to be near 30%, because: not all TTLs are project supervisors; approximately 50% of participating TTLs sent us an e-mail; it was requested that respondents be familiar with the full life cycles of projects;, TTLs are often traveling for work and therefore would not always have a chance to respond; TTLs tend to be bombarded with questionnaires (sometimes two to three per month); confidentiality posed some problems; and e-mail invitations may have been filtered or blocked for security reasons or even treated as spam (see Faught, Whitten, & Green, 2004; Kaplowitz, Hadlock, & Levine, 2004; Porter & Whitcomb, 2007; Manfreda, Bosnjak, Berzelak, Haas, & Vehovar, 2008;). There seems to be no coverage bias, and the homogeneity of respondents enhances the validity and reliability of the measured results.

The sample is fairly balanced concerning the different project sectors and provides a fairly representative cross-sectoral distribution of projects funded by the World Bank (see Table 4). Approximately 25% of the projects in the sample are from the social development, reform, and governance sectors. The agriculture, water, electricity, and sanitization sectors account for another 25%. Roughly 20% of the projects are from the education, health, nutrition, and population sectors. The transportation and environment sectors account for another 20%. All projects in the sample are either completed or under ongoing implementation, with a 4-year mean and 2-year standard deviation. Their average cost is around US$88 million for all donors and US$75 million for the World Bank. Three fourths of the respondents in the sample are men. There are five permanent employees for every one contractual worker. Approximately 40% of the respondents are economists, and approximately 30% are engineers. Other academic backgrounds, including business administration, humanities, and social sciences, account for another 30%. The average age of the respondents is 49 years, and they have held their position for an average of 7 years. Table 5 summarizes the descriptive characteristics of the TTL.

Table 2: Operationalization of the project success constructs and the non–project management but project-related variables.

Construct Scale Cronbach's α       Measures
Success criteria  
Project management success
α = 0.73
Project was completed on time (efficiency/time)
Project met its objectives (effectiveness/objectives)
Project budget was well-managed (efficiency/cost)
Deliverable success α = 0.88 Project objectives are suited to the country's needs (relevance country)
Project objectives are suited to the beneficiaries' needs (relevance beneficiaries)
Project built institutional capacity within the country (impact)
Project results will continue after the project ends (sustainability)
Non–project management but
project-related variables
Project duration   End date minus start date
Project budget   Amount of initial financing
TTL's work experience Length of time TTL has held his or her position

(Items in italics have been dropped in the final fitted model.)

All the projects in this study are ID projects funded by the World Bank. They are similar in terms of finality, objectives, processes, and guidelines. Roughly 17% of the projects were considered more or less a failure by their supervisors, which is possibly below the real failure rate for ID projects at the World Bank.13 Although this may suggest a nonrespondent bias, tests for potential links between project success and other key variables, as well as time to respond to the questionnaire, do not demonstrate any significance.

Table 4: Distribution of project sectors.

Project Sector Frequency (N) Frequency (%)
Education 18 12.2
Energy 4 2.7
Environment 17 11.6
Mines 1 0.7
Transport 14 9.5
Agriculture 21 14.3
Urban development 4 2.7
Water, electricity,
and sanitization
15 10.2
Social development 11 7.5
Reform and governance 27 18.4
Health, nutrition, and population 14 9.5
Communication and
1 0.7
Total 147 100%

Data Analysis

A principal components factor analysis was first applied to test the multidimensionality of the 41 CSFs scale. Then, the non–project management but project-related variables of our model were identified using an exploratory correlation analysis (Table 6). Of an initial list of three interval (continuous) variables—duration, budget, and experience—only the latter two have been found to significantly affect the CSFs.

Table 5: Distribution of project TTLs.

Variables Frequencies or means
Gender (N = 152) Male: 113 (75%)
Female: 39 (25%)
Age (N = 144)      49 years
TTL experience (N =146)      7.2 years
Professional status (N = 152) Permanent: 128 (85%)
Contractual: 24 (15%)
Level of education (N = 148) Master: 88 (60%)
    PhD: 60 (40%)
Academic background (N = 145)  
    Social sciences and humanities 41%
    Business administration/commerce 10%
    Education sciences 11%
    Law 3%
    Health sciences 2%
    Engineering and natural sciences 4%
  29 %

The model was finally tested using structural equation modeling (SEM). SEM is a statistical method for measuring simultaneous hypothesized causal relationships between multiple latent and observed variables14 (Bollen, 1989; Byrne, 2001; Tabachnick & Fidell, 2001; Roussel, Durrieu, Campoy, & El Akremi, 2002; Hair, Black, Babin, Anderson, & Tatham, 2006; Shah & Goldstein, 2006). For the model estimation, AMOS 17 and the full information maximum likelihood (FIML) were used (see Anderson, 1957; Arbuckle, 1996; Marsh, 1998; Byrne, 2001; Roussel, Durrieu, Campoy, & El Akremi, 2002, p. 76 & 87; Shah & Goldstein, 2006; Hair, Black, Babin, Anderson, & Tatham, 2006). Assessing the fit of a model is one of the more complicated aspects of SEM (Bagozzi & Yi, 1988; Byrne, 2001; Tabachnick & Fidell, 2001; Roussel, Durrieu, Campoy, & El Akremi, 2002; Hair, Black, Babin, Anderson, & Tatham, 2006; Shah & Goldstein, 2006). The most popular index has been the chi-square statistic (Bollen, 1989; Byrne, 2001, p. 79), but it is dependent on the sample size and distribution. In light of these limitations and the ongoing debate over superiority or even appropriateness of one index over another, an ever-increasing number of fit indices have been developed (Bagozzi & Yi, 1988; Bollen, 1989; Tabachnick & Fidell, 2001; Hair, Black, Babin, Anderson, & Tatham, 2006; Shah & Goldstein, 2006). Furthermore, there are no consistent cut-off criteria for any single index that can be used in all instances (Marsh, Balla, & Donald, 1988). For absolute measures of fit we report the chi-square statistic (CMIN; and its statistical significance, p > 0.05) and the root mean square error of approximation (RMSEA; Steiger & Lind, 1980 as cited in Bollen, 1989; Browne & Cudeck, 1993). Values of RMSEA as high as 0.08 and, if possible, less than 0.05 are acceptable; values greater than 0.10 are indicative of poor-fitting models (Browne & Cudeck, 1993). Hu & Bentler (1999) have found that, in small samples, the RMSEA tends to overreject good models. For incremental fit indices, we report the normed fit index (NFI, Bentler & Bonett, 1980) and the comparative fit index (CFI, Bentler, 1990). Models with a CFI below 0.85 should be rejected (Bentler & Bonett, 1980). Values close to 0.90 are indicative of a good fit (Bentler, 1990). For parsimony fit indices, we report the normed chi-square statistic (CMIN/DF, Wheaton, Muthen, Alwin, & Summers (1977). Values less than 1 may indicate overfit, and higher values (above 3 to 5) may indicate an under-parametized model (Jöreskog, 1969; Roussel, Durrieu, Campoy, & El Akremi, 2002). Finally, in order to compare the initial and the final model, the Akaike information criterion, AIC (Akaike, 1987), and the expected cross-validation index, ECVI (Browne & Cudeck, 1989), have been used, smaller values of which represent a better fit of the models (Hu & Bentler, 1995; Browne & Cudeck, 1989).


Principal Components Factor Analysis

The factor analysis generates five CSFs (monitoring, coordination, design, training, and institutional environment). Results are assessed in Table 3 with the CSF construct Cronbach's alpha and the percentage of variance each CSF accounts for. Then, the analysis of the bivariate relations using the correlation matrix (Table 6, p. 24) reveals that average correlations among the CSFs are positive but not very low, with some of them reaching 0.50 or 0.60.15 This might suggest the existence of a second-order latent CSF construct or variable that is accountable for the five CSFs, which is theoretically possible, since CSF is a multidimensional, abstract, and complex construct (Byrne, 2001, pp. 120-141; Roussel, Durrieu, Campoy, & El Akremi, 2002, pp. 163-184). Using the Belassi and Tukel (1996) framework of groups of CSFs, we labeled this second-order CSF construct: CSF related to project management (hereafter labeled PM CSF). Table 6 also provides the correlations between the three non–project management but project-related variables and the five CSFs. Project duration does not show any significant correlation with any of the CSFs. Project budget is significantly correlated with both the design and the environment CSFs, and the TTL's experience is significantly correlated with the coordination CSF's experience. All three correlations are positive.

Table 6: Correlation matrix.

CSF Variables 1 2 3 4 5 6 7 8
1. Monitoring 1.00              
2. Coordination .55** 1.00 1.00**          
3. Design .62** .50** .43** 1.00 1.00      
4. Training .54** .38** .49 ** .28** n.s. 1.00    
5. Environment .42** .49** n.s n.s. .03** n.s. 1.00  
6. Duration n.s. n.s. .04** n.s. n.s. n.s. n.s. 1.00
7. Cost n.s. n.s. n.s. n.s        
8. Experience n.s. .03**            

**p < 0.05

Structural Equation Modeling (SEM)

In this step of the data analysis, the interactions of the model variables, including the second-order CSFs (PM CSF) were estimated simultaneously. Similar to the works of Dvir and Lechler (2004), we started the SEM/AMOS analysis with the confirmatory part of the model only; for example, the non–project management but project-related variables were not included in the estimation. Based on the results of the correlation analysis between the non–project management but project-related and the CSF variables, we introduced the non-project management but project-related variables into the model. According to the pattern of those correlations (Table 6), the paths from the experience variable were added to the confirmatory model. Paths from the budget variable were added to both the design and environment CSFs. The initial hypothesized model is a complex one, with 32 observed variables and 105 estimated parameters (7 for success criteria, 23 for CSF, and 2 for project size).

Admittedly, there is some degree of misfit in the initial model despite the acceptable value of the RMSEA and the CMIN/DF. In addition, the project management success, the environment CSF, and the coordination CSF showed poor composite reliability (0.55–0.65 < 0.7) (Jöreskog, 1971). We then decided to drop the following ten unstable items: the first project management success item (time), the last deliverable success item (sustainability), the fourth coordination item (NPC remained the same throughout the project), the third and fourth design items (risk identification; innovative design), the last training CSF item (technical training), and all the four institutional environment CSF items (see items in italics in Table 3).

The resulting final model is still a complex one, with 22 observed variables and 73 estimated parameters (5 for success criteria, 15 for CSF, and 2 for project size). Table 7 compares the fit statistics of both the initial and final models. The final model presented in Figure 2 does not differ from the confirmatory model; therefore, similarly to Dvir and Lechler (2004), we do not present the confirmatory model.

Table 7: Fit statistics of the two models.

Initial model 2.07 0.85 0.76 0.078 1150.62 6.501
Final model 2.10 0.90 0.84 0.079 569.64 3.218

Table 3: Factor analysis of the critical success factors (CSFs) and their Cronbach's alpha and their common variance share.

Items Principal Components (CSF)
  1 2 3 4 5
1.    Monitoring CSF (α = 0.90; var: 18%)
Project team respected financial accounting policies 0.789        
Project team controlled contracting processes 0.756        
Resource utilization was appropriate 0.753        
Project team anticipated project challenges 0.730        
Project team responded quickly to problems 0.700        
2.    Coordination CSF (α = 0.83; var: 13%)
NPC16 showed leadership   0.795      
NPC had the appropriate interpersonal skills   0.767      
NPC had the required knowledge for the project   0.735      
NPC remained the same throughout the project   0.657      
Good communication between NPC and agency   0.648      
3.    Design CSF (α = 0.86; var: 12%)
Project was well designed     0.774    
Objectives based on understanding of local context     0.765    
Risk identification was done well     0.711    
Design was innovative     0.686    
Project stakeholders agreed on strategic issues     0.591    
4.    Training CSF (α = 0.84; var: 11%)
Project team received appropriate project management training       0.790  
Project team received appropriate training in contracting       0.781  
The design included training       0.747  
Project team received appropriate technical training       0.724  
5.    Institutional environment(α = 0.75; var: 10%)
Project did not require political activity in the country         0.729
Institutional frameworks were favorable         0.664
Other donors wanted the project to succeed         0.658
Favorable political, economic, social, and cultural conditions         0.628
N = 178. Kaiser –Meyer-Olkin (KMO) = 0.913; Orthogonal rotation: VARIMAX; 64% of the common variance; variables that were dropped in the final fitted models are in italics but Cronbach's alphas are for the initial CSF items.
Results of the structural equation model. Fit statistics: CMIN=423. 643; DF=202, p &lt; 0.000; CMIN/DF = 2.10; CFI=0.90; RMSEA=0.079; NFI=0.84

Figure 2: Results of the structural equation model. Fit statistics: CMIN = 423. 643; DF = 202, p < 0.000; CMIN/DF = 2.10; CFI = 0.90; RMSEA = 0.079; NFI = 0.84.

Except for the chi-square index, all test criteria are met in assessing the model fit. Since all other tests achieve the required fit criteria, the final structural equation model should be accepted. Composite reliability above 0.7 (Jöreskog, 1971), average variance extracted above 0.50 (Fornell & Larker, 1981), and the fact that all of the success items were significant (t = 1.96), with all success items' standard regression weights above 0.6, converge to show construct validity (Bagozzi & Yi, 1988; Shah & Goldstein, 2006). Table 8 presents the results of the composite reliability and the average extracted variance.

Table 8: Construct reliability and average extracted variance.

Constructs Jöreskog's rhô Fornell and Larker's Average Variance Extracted
Project management success 0.69 0.53
Del success 0.83 0.62
PM CSF 0.77 0.47
Monitoring 0.87 0.57
Coordination 0.85 0.58
Design 0.77 0.53
Training 0.81 0.59

The results for the hypotheses are shown in Table 9.

Table 8: Hypotheses testing results.

Hypotheses Result
H1        Project management success significantly influences project deliverable success Not supported
H2a   CSFs are correlated and there exists a second-order CSF Supported
H2b   Project management success is positively affected by PM CSF Supported
H2c   Deliverable success is significantly affected by PM CSF Not supported
H3a   Project duration affects CSF Not supported
H3b   Project budget affects CSF Not supported
H3c   TTL experience at the position affects CSF Not supported

Hypothesis 1 (H1) is not supported, since the path between project management success and deliverable success is not significant. The positive correlations among the CSFs and the fact that some correlations are about 0.60 fully support H2a, and therefore the existence of a second-order latent CSF related to project management (PM CSF). The strong positive influence of PM CSF on project management success fully supports H2b. As the path coefficient between PM CSF and deliverable success is not significant, H2c is not supported. The exploratory hypotheses H3a, H3b, and H3c, proposing effects of non–project management but project-related variables on the four CSFs are not supported, as their path coefficients are not significant in the SEM/AMOS model, which takes simultaneously into account the interactions of all variables.

Discussion and Conclusion

The first objective of this study was to provide an in-depth empirical investigation of CSFs and analyze how a set of World Bank projects' CSFs are interrelated. The second objective was to investigate their influences on two different dimensions of project success (project management success and deliverable success). The third and last objective of this research was to estimate the influence of non–project management but project-related variables on CSF.

The most important results of this study are the interactions among the CSFs and their influences on project success dimensions. Only by investigating the CSFs separately, using structural equation modeling (SEM) instead of multiple regression analysis (see Ika, Diallo, & Thuillier, 2009), can we gain insight into the complex relationships among them and explore a CSF phenomenon that would be otherwise unobservable. There exists, in fact, a second-order latent CSF variable that we have called CSF related to project management (PM CSF). The strong, significant, and positive standardized regression weights (t) between PM CSF and each of the four CSFs, in monitoring (t = 0.83), coordination (0.69), design (0.85), and training (0.63) are worth mentioning.

However, the task managers or task team leaders (TTLs) do not weight equally their perceptions on the CSFs. The hierarchy of the CSFs in their perspective confirms the high importance of project design (squared multiple correlation, R2 = 0.73), project monitoring (0.68), and project coordination (0.48). This is consistent with previous results using the multiple regression analysis (see Ika, Diallo, & Thuillier, 2009). In line with the orthodoxy of IDPM, the emphasis on results-based management and its accountability-for-results principle, and the strong procedures or guidelines orientation in IDPM, this research result is consistent with practice and theory (see Binnendijk, 2000; Diallo & Thuillier, 2004, 2005; Khang & Moe, 2008; Ika, Diallo, & Thuillier, 2009, 2010).

The clearly differing influences of the CSFs on the two project success variables indicate the importance of differentiating between these two success dimensions. While the CSFs altogether, and particularly the PM CSF affects significantly, positively, and strongly project management success (t = 0.89; R2 = 0.80), the PM CSF doesn't show any significant influence on deliverable success. If the former result makes sense and reflects the nature of IDPM, which is mainly focused on project-specific objectives and cost, the latter is counterintuitive, since the overall goal of ID is to achieve impact of project outputs. This fourth result seems more surprising if we consider that the path between PM CSF and deliverable success falls just short of being nonsignificant (p = 0.054) but it was just significant in the confirmatory model (p = 0.043), which is not shown here. While this fourth result stands in stark contrast to the findings of Diallo and Thuillier (2004) concerning the TTLs, it is consistent with their findings regarding the national project coordinators (NPCs); that is, the “true” project managers. These authors have found that, although the NPCs seem less sensitive to project-induced results (i.e., project impact) (possibly because the latter requires long-term indicators that are difficult to measure during project implementation), TTLs seem to be sensitive to project impact and do not pay significant attention to project management success. For that reason, we call for caution in the interpretation of this fourth research result. In fact, the squared multiple correlation of deliverable success (R2 = 0.46) indicates that there are influences other than project management–related CSFs causing the deliverable success.

The fifth research result is also insightful. Project management success doesn't significantly affect deliverable success. This suggests that a project may be a project management success but a development failure or that a project could still yield deliverables that conform to stakeholder expectations despite a poor project management. Yet a minimum project management performance has to be achieved in the short term. Otherwise, no deliverable would be completed. Deliverable success is in fact a long-term success dimension. This research result seems consistent with practice as well as theory, as a project may be a failure, project management–wise, and still later become a deliverable success and vice versa (Ika, 2009). But it is intriguing to the extent that one may expect TTLs to deliver, through strong project management by NPCs, development results: deliverable success. Is this research result due to the hardly difficult to define and assess long-term impact criterion? Or is it due to the specificity of our sample? This is difficult to ascertain.

The third question that this research addresses is how the non–project management but project-related variables influence CSFs. This part of the study is more exploratory. In contrast to our initial hypotheses, all of the non–project management but project-related variables considered simultaneously have a nonsignificant influence on the CSFs. This sixth result stands in contrast to the correlation analysis (Table 6) that shows significant relationships with two non–project management but project-related variables. First, considering the project duration and budget variables, this research result suggests that a kind of “one-size-fits-all” approach is still dominant in IDPM (Analoui, 1989; Ika, Diallo, & Thuillier, 2010). Second, the nonsignificant influence of the TTL's experience with the coordination CSF may be due to the fact that the TTLs are not involved in day-to-day project operations, although they are updated on each step of the project and may not grant a “no objection” to the NPCs, the “true” project managers.

The fact that project management success item efficiency/time and project duration were dropped indicates that, in IDPM, the project objectives and the cost of achieving them are far more important than the time constraint to achieve them. In fact, ID is more long-term oriented. Yet, the longer the project, the higher the risk of undertaking a project for which the needs are no more relevant. This might induce the redesign of the project (objectives-changes), which in turn may affect project management success as well as deliverable success (Gittinger, 1982; Youker, 1989; Ika, Diallo, & Thuillier, 2010).

As the environment CSF has been dropped because of its instability, the research results suggest that some of the other external environment variables, and the other non–project management but project-related variables such as life cycle and urgency, and the project organization variables such as agency support (e.g., Belassi & Tukel, 1996; Ika, Diallo, & Thuillier, 2009), need to be considered in analyzing the influence of the CSFs on the project's deliverable success.

Implications and Outlook

In contrast to dominant project management literature, this study sheds a different light on the influence of the CSFs on project success dimensions. It supports the view that, while CSFs are not everything in understanding the project success phenomenon, interactions among them help to understand their respective and collective influences. By highlighting the interactions among the CSF variables, this study contributes to theory building in three aspects. First, it explores the CSFs in an ID sector that has grown in parallel with project management and particularly the perspectives of key players, the World Bank TTLs (project supervisors). Second, it reveals the interaction structure between the CSFs themselves and with the two dimensions of project success (project management success and deliverable success). Third, it explores the non–project management but project-related variables and their influence on CSFs.

CSFs related to the management of the project (here the project supervisor, the project coordinator, and the project team) affect project success (see Munns & Bjeirmi, 1996). The major lesson is that the most important CSFs are project design, monitoring, and coordination. This stresses the importance of the collaboration between the project supervisor (TTL) and the project coordinator (NPC), which is critical in the context of the new aid management orthodoxy with its pillars or banners such as “local management,” “‘program approach,” “alignment of donors,” and “harmonization”' (Maddock, 1992; Hubbard, 2005; Lavergne & Alba, 2003; European Commission, 2007; Ika, Diallo, & Thuillier, 2009;). It is therefore in the hands of the TTL and the NPC to use the positive effects of those project management–related CSFs in order to deliver project management success. Successful management of ID projects is often “a long voyage of discovery in the most varied domains, from technology to politics” (Hirschman, 1967, p. 35). In this Hirschman journey of discovery, confronted with the “notorious critical implementation problems” (Gow & Morss, 1988), TTLs and NPCs should not commence the voyage “empty headed and empty handed” (Rondinelli, 1983, p. 325). They need tools to implement processes that secure fair levels for the four well-known tested CSFs (Ika, Diallo, & Thuillier, 2009).

Although our basic model is mainly confirmatory, it has some limitations. One limitation is linked to the static treatment of data. The measurements are all ex-post and therefore do not render possible the analysis of the influences of CSFs over the project life cycle (Dvir & Lechler, 2004; Khang & Moe, 2008). The second limitation is that the questionnaire focuses only on the self-perceptions of World Bank TTLs.

Our study opens up opportunities for further research. The fact that project management success does not impact deliverable success indicates an important direction: the search for variables that could explain deliverable success notably impact. A more accurate definition and examination of the environment CSF and other CSFs is required. Such a definition may enable further investigation into the interactions between non–project management but project-related variables and the CSFs. In fact, like all empirical quantitative studies, this research had to trade off questionnaire length and time required to complete it against measurement thoroughness. Further research focused on the project life cycle is needed to deepen the understanding of the interactions between contextual variables, including non–project management but project-related variables, and the CSFs as well as their influences on project success. Finally, as project success is a matter of perspective, the perceptions of the project supervisors and NPCs of other multilateral agencies such as the United Nations Development Program, the European Union, and the Multilateral Development Banks should be explored.


Analoui, F. (1989). Project managers' role: Towards a “descriptive” approach. Project Appraisal, 4(1), 36–42.

Akaike, H. (1987). Factor analysis and AIC. Psychometrika, 52, 317–332.

Arbuckle, J. L. (1996). Full information estimation in the presence of incomplete data. In G.A. Marcoulides & R.E. Schumaker (Eds.), Advanced structural equation modeling: Issues and techniques (pp. 243–277). Hillsdale, New Jersey: Lawrence Erlbaum Associates.

Anderson, T. W. (1957). Maximum likelihood estimates for a multivariate normal distribution when some observations are missing. Journal of the American Statistical Association, 59, 2–5.

Anderson, J. C., & Gerbing, D. W. (1988). Structural equation modeling in practice: A review and recommended two-step approach. Psychological Bulletin, 103, 411–423.

Au, A. K. M., & Tse, A. C. B. (2001). Marketing ethics and behavioural predispositions of Chinese managers of SMEs in Hong Kong. Journal of Small Business Management, 39(3), 272–278.

Baccarini, D. (1999). The logical framework method for defining project success. Project Management Journal, 30(4), 25–32.

Baggozi, R. P., & Yi, Y. (1988). On the evaluation of structural equation models. Academy of Marketing Science, 16(1), 074–094.

Belassi, W., & Tukel, O. I. (1996). A new framework for determining critical success/failure factors in projects. International Journal of Project Management, 14(3), 141–151.

Bellizi, A. J., & Hasty, R. W. (2002). Supervising unethical sales force behaviour: Do men and women managers discipline men and women subordinates uniformly? Journal of Business Ethics, 40(2), 155–166.

Bentler, P. M. (1990). Comparative fit indexes in structural models. Psychological Bulletin, 107, 238–246.

Bentler, P. M., & Bonett, D. G. (1980). Significance tests and goodness of fit in the analysis of covariance structures. Psychological Bulletin, 88, 588–606.

Binnendijk, A. (2000). Results based management in the development of co-operation agencies: A review of experience. Background Report. Organisation for Economic Co-operation and Development: Development Assistance Committee Working Party on Aid Evaluation (DAC-EV).

Bollen, K. A. (1989). Structural equations with latent variables. New York: Wiley.

Bremer, J. A. (1984). Building institutional capacity for policy analysis: An alternative approach to sustainability. Public Administration and Development, 4(1), 1–13.

Brinkerhoff, D. W. (1994). Institutional development in World Bank projects: Analytical approaches and intervention designs. Public Administration and Development, 14(1), 135–151.

Browne, M. W., & Cudeck, R. (1993). Alternative ways of assessing model fit. In K. A. Bollen & J. S. Long (Eds.), Testing structural equation models (pp. 445–455). Newbury Park, CA: Sage.

Browne, M. W., & Cudeck, R. (1989). Single sample cross-validation indices for covariance structures. Multivariate Behavioral Research, 24(4), 445–455.

Byrne, B. M. (2001). Structural equation modeling with AMOS: Basic concepts, applications, and programming. New Jersey: Lawrence Erlbaum Associates.

Canadian International Development Agency (CIDA). (January, 1999). Results based management: An introductory guide to the concepts and principles. Gatineau, Canada: Performance Review Branch.

Canadian International Development Agency (CIDA). (March, 2001). A results approach to developing the implementation plan; A guide for CIDA partners and implementing agencies. Gatineau, Canada: Performance Review Branch.

Carden, L., & Egan, T. (2008). Does our literature support sectors newer to project management? The search for quality publications relevant to nontraditional industries? International Journal of Project Management, 39(3), 96–27.

Cracknell, B. E. (1988). Evaluating development assistance: A review of the literature. Public Administration and Development, 8(1), 75–83.

Crawford, P., & Bryce, P. (2003). Project monitoring and evaluation: A method for enhancing the efficiency and effectiveness of aid project implementation. International Journal of Project Management, 21(1), 363–373.

de Solages, O. (1992). Réussites et déconvenues du développement dans le tiers-monde: Esquisse d'un mal-développement. Paris: L'Harmattan.

Dillon, W., Kumar, A., & Mulani, N. (1987). Offending estimates in covariance structure analysis: Comment on the causes and solutions to Heywood causes. Psychological Bulletin, 101, 126–135.

de Wit, A. (1988). Measurement of project success. Project Management Journal, 6(3), 164–170.

Diallo, A., & Thuillier, D. (2004). The success dimensions of international development projects: The perceptions of African project coordinators. International Journal of Project Management, 22(1), 19–31.

Diallo, A., & Thuillier, D. (2005). The success of international development projects, trust and communication: An African perspective. International Journal of Project Management, 23(1), 237–252.

Dillman, D. A. (2000). Mail and internet surveys: The tailored design method. New York: Wiley.

Dvir, D., & Lechler, T. (2004). Plans are nothing, changing plans is everything: The impact of changes on project success. Research Policy, 33, 1–15.

Easterly, W. (2003). Can foreign aid buy growth? Journal of Economic Perspectives, 17(3), 23–48.

Easterly, W. (2007). Are aid agencies improving? Economic Policy, 22(52) 633–678.

Esteves, J., Casanova, J., & Pastor, J. (2003, 4–6 August). Modeling with partial least squares critical success factors interrelationships in ERP implementations. Paper presented at the ninth Americas Conference on Information Systems, Tampa, Florida, USA.

European Commission. (2007). Support to sector programs. Covering the three financing modalities: Sector budget support, pool funding and EC project procedures. Tools and Methods Series, Guidelines, No. 2.

Faught, K. S., Whitten, D., & Green, K. W. (2004, Spring). Doing survey research on the internet: Yes, timing does matter. Journal of Computer Information Systems, 26–34.

Fornell, C., & Larker, D. F. (1981). Evaluating structural equation models with unobservable variables and measurement error. Journal of Marketing Research, 18, 39–50.

Fowler, A., & Walsh, M. (1998). Conflicting perceptions of success in an information systems project. International Journal of Project Management, 17(1), 1–10.

Gareth, D. G., & Martin, L. A. (2003). The Ashville project: Participants' perceptions of factors contributing to the success of a patient self-management diabetes program. Journal of the American Pharmaceutical Association, 43(2), 185–190.

Gittinger, J. P. (1982). Economic analysis of agricultural projects. Baltimore: The Johns Hopkins University Press.

Gow, D. D., & Morss, E. R. (1988). The notorious nine: Critical problems in project implementation. World Development, 16(12), 1399–1418.

Hair, J. F., Black, W. C., Babin, B. J., Anderson, R. E., & Tatham, R. L. (2006). Multivariate data analysis (6th ed.). Upper Saddle River, NJ: Prentice-Hall.

Heerwegh, D. (2005). Effects of personal salutations in e-mail invitations to participate in a web survey. Public Opinion Quarterly, 69(4), 588–598.

Hirschman, A. O. (1967). Development projects observed. Washington, DC: Brookings Institution.

Honadle, G. H., & Rosengard, J. K. (1983). Putting “projectized” development in perspective. Public Administration and Development, 3, 299–305.

Hu, L. T., & Bentler, P. M. (1995). Evaluating model fit. In R. H. Hoyle (Eds.), Structural equations modeling: Concepts, issues, and applications, Thousand Oaks, CA: Sage.

Hu, L. T., & Bentler, P. M. (1999). Cutoff criteria for fit indices in covariance structure analysis: conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1–55.

Hubbard, M. (2005). Aid management: Beyond the new orthodoxy. Public Administration and Development, 25, 366–371.

Hulme, D. (1995). Projects, politics and professionals: Alternative approaches for project identification and project planning. Agricultural Systems, 47, 211–233.

Hyväri, I. (2006). Success of projects in different organizational conditions. Project Management Journal, 37(4), 31–41.

Ika, L. A., Diallo, A., & Thuillier, D. (2010). Project management in the international industry: The project coordinator's perspective. International Journal of Managing Projects in Business, 3(1), 61–93.

. Ika, L. A., Diallo, A., & Thuillier, D. (2009, 11–13 October). The most critical success factors for World Bank projects: the Task Team Leaders' perspective. Paper accepted in the proceedings of the International Network on Organising by Projects (IRNOP) IX, Berlin.

Ika, L. A. (2009). Project success as a topic in project management journals. Project Management Journal, 40(4), 6–19.

Ika, L. A. (2005). La gestion des projets d'aide au développement: Historique, bilan et perspective. Perspective Africaine, 2, 128–153.

Jacobs, C., & McLaughlin, P. (1996). Making a difference: Results of a pilot investigation into the impact of technical co-operation training on developing countries. Public Administration and Development, 16, 123–129.

Jöreskog, K. G. (1969). A general approach to confirmatory maximum likelihood factor analysis. Psychometrika, 60, 77–93.

Jöreskog, K. G. (1971). Statistical analysis of sets of congeneric tests. Psychometrika, 36, 109–133.

Jugdev, K., & Müller, R. (2005). A retrospective look at our evolving understanding of project success. Project Management Journal, 36(4), 19–31.

Kaplowitz, M. D., Hadlock, T. D., & Levine, R. (2004). A comparison of Web and mail survey response rates. Public Opinion Quarterly, 68(1), 94–101.

Kealey, J., Protheroe, D. R., MacDonald, D., & Vulpe, T. (2005). Re-examining the role of training in contributing to international project success: A literature review and an outline of a new model training program. International Journal of Intercultural Relations, 29, 289–316.

Keil, M., Tiwana, A., & Bush, A. (2002). Reconciling user and project manager perceptions of IT project risk: A Delphi study. Information Systems Journal, 12, 103–119.

Khan, Z. A., Thornton, N., & Frazer, M. (2003). Experience of a financial reforms project in Bangladesh. Public Administration and Development, 20, 33–42.

Khang, D. B., & Moe, T. L. (2008). Success criteria and factors for international development projects: a life-cycle-based framework. Project Management Journal, 39(1), 72–84.

Kleinschmidt, E. J., & Cooper, R. J. (1995). The relative importance of new product success determinants-perception versus reality. R&D Management, 25(3), 281–298.

Kwak, Y. H. (2002, September, 9–13). Critical success factors in international development project management. Paper presented at the CIB 10th International Symposium Construction Innovation & Global Competitiveness, Cincinnati, Ohio.

Lavergne, R., & Alba, A. (2003). Guide d'introduction aux approches-programmes à l'ACDI. Gatineau: ACDI.

Lechler, T., & Gemünden, H. (2000). The influence structure of the success factor of project management: A conceptual framework and empirical evidence. Technical Report.

Likert, R., & Likert, J. G. (1976). New ways of managing conflict. New-York: McGraw-Hill.

Linberg, K. R. (1999). Software developer perceptions about software project failure: A case study. The Journal of Systems and Software, 49, 177–192.

Liu, A. M. M., & Walker, A. (1998). Evaluation of project outcomes. Construction Management and Economics, 6, 209–219.

Maddock, N. (1992). Local management of aid-funded projects. Public Administration and Development, 12, 399–407.

Manfreda, K. L., Bosnjak, M., Berzelak, J., Haas, I., & Vehovar, V. (2008). Web surveys versus other survey modes: A meta-analysis comparing response rates. International Journal of Market Research, 50(1), 79.

Marsh, H. W. (1998). Pairwise deletion for missing data in structural equation models: Non positive definite matrices, parameter estimates, goodness of fit, and adjusted sample sizes. Structural Equation Modeling, 5, 22–36.

Marsh, H. W., Balla, J. R., & McDonald, R. P. (1988). Goodness-of-fit in confirmatory factor analysis: the effect of sample size. Psychological Bulletin, 103(3), 391–410.

Morgan, E. P. (1983). The project orthodoxy in development: Re-evaluating the cutting edge. Public Administration and Development, 3, 329–339.

Mubila, M. M., Lufumpa, C. L., & Kayizzi-Mugerwa, S. (2000). A statistical analysis of determinants of project success: Examples from the African Development Bank. Economic Research Papers, 56.

Munns, A. K., & Bjeirmi, B. F. (1996). The role of project management in achieving project success. International Journal of Project Management, 14(2), 81–87.

Nah, F. F-H., Zuckweiler, K. M., & Lau, J. L-S. (2003). ERP implementation: Chief information officers' perceptions of critical success factors. International Journal of Human-Computer Interaction, 16(1), 5–22.

Noël, G. (1997). Le développement international et la gestion de projet. Québec: Presses de l'Université du Québec.

Organization for Economic Co-operation and Development (OECD). (2002). Glossary of key terms in evaluation and results based management. Paris: OECD.

Organization for Economic Co-operation and Development (OECD). (2005). The 2004 development cooperation report. Statistical annex. Paris, France: OECD.

Pinto, J. K., & Slevin, D. P. (1988). Critical success factors across the project life cycle. Project Management Journal, 19(3), 67–74.

Porter, S. R., & Whitcomb, M. E. (2007). Mixed-mode contacts in Web surveys: Paper is not necessarily better. Public Opinion Quarterly, 71(4), 635–648.

Rakodi, C. (1982). The role of monitoring and evaluation in project planning in relation to the upgrading of unauthorized housing areas. Public Administration and Development, 2, 129–146.

Rondinelli, D. A. (1976). Why development projects fail: Problems of project management in developing countries. Project Management Quarterly, 7(7), 10–15.

Rondinelli, D. A. (1983). Projects as instruments of development administration: A qualified defense and suggestions for improvement. Public Administration and Development, 3, 307–327.

Roodman, D. (2006). Aid project proliferation and absorptive capacity. Center for Global Development. Working paper; 75, 1-45.

Roussel, P., Durrieu, F., Campoy, E., & El Akremi (2002). Méthodes d'équations structurelles: Recherche et applications en gestion. Paris: Economica.

Schmid, B., & Adams, J. (2008). Motivation in project management: The project manager's perspective. Project Management Journal, 39(2), 60–71.

Shah, R., & Goldstein, S. M. (2006). Use of structural equation modeling in operations management research: Looking back and forward. Journal of Management, 24, 148–169.

Shenhar, A., Levy, O., & Dvir, D. (1997). Mapping the dimensions of project success. Project Management Journal, 28(2), 5–13.

Smith, P. (1988). Improving the project identification process in agricultural development. Public Administration and Development, 8, 15–26.

Steiger, J. H., & Lind, J. C. (1980, June). Statistically based tests for the number of common factors. Paper presented at the Psychometric Society Annual Meeting, Iowa City, IA.

Steinfort, P., & Walker, D. H. T. (August 8–11, 2007). Critical success factors in project management globally and how they may be applied to aid projects. Paper presented at the PMOZ Achieving Excellence–4th Annual Project Management Australia Conference, Gold Coast, Australia.

Struyk, R. J. (2007). Factors in successful program implementation in Russia during the transition: Pilot programs as a guide. Public Administration and Development, 27, 63–83.

Tabachnick, B. G., & Fidell, L. S. (2001). Using multivariate statistics (4th ed.). Boston: Allyn & Bacon.

Tacconi, L., & Tisdell, C. (1992). Rural development projects in LDCs: Appraisal, participation and sustainability. Public Administration and Development, 12, 267–278.

Taylor, H. (1995). From general training to projectization: Implications for learning processes and the roles of trainers. Public Administration and Development, 15, 481–494.

Themistocleous, G., & Wearne, S. H. (2000). Project management topic coverage in journals. International Journal of Project Management, 18(1), 7–11.

Uhl-Bien, M., & Graen, G. B. (1998). Individual self-management: Analysis of professionals' self-managing activities in functional and cross-functional work teams. Academy of Management Journal, 41(3), 340–350.

Vickland, S., & Nieuwenhujis, I. (2005). Critical success factors for modernizing public financial management information systems in Bosnia and Hergzegovina. Public Administration and Development, 25(2), 95–103.

Wheaton, B. (1987). Assessment of fit in overidentified models with latent variables. Sociological Methods & Research, 16, 118–154.

Wheaton, B., Muthen, B., Alwin, D. F., & Summers, G. (1977). Assessing reliability and stability in panel models. Sociological Methodology, 8(1), 84–136.

White, D., & Fortune, J. (2002). Current practice in project management: An empirical study. International Journal of Project Management, 20(1), 1–11.

World Bank (1998). Assessing aid: What works, what doesn't and why. Oxford: Oxford University Press.

World Bank (2004). Monitoring and evaluation: Some tools, methods and approaches. Operations Evaluation Department (OED) 2004; available at

Youker, R. (1989). Managing the project cycle for time, cost and quality: Lessons from World Bank experience. Project Management, 7(1), 52–57.

Youker, R. (1992). Managing the international project environment. International Journal of Project Management, 10(4), 219–226.

Youker, R. (1999). Managing international development project: Lessons learned. Project Management Journal, 30(2), 6–7.


1 Project success dimension refers to a set of project success criteria.

2 Project manager as a title does not exist at the World Bank. So TTLs include not only TTLs for investment projects but also TTLs for research projects, for implementation completion reports (ICRs), and TTLs who are workshop lecturers, etc.

3 TTLs are not involved in day-to-day IDPM, although they are updated on each step of the project. They may supervise but not manage a project portfolio. Unlike in traditional project management, although TTLs may suggest specific activities to the NPC and discuss any issue pertaining to project implementation, they have no authority to request the NPC to do this or that. In fact, they may only grant a “no objection” to the NPC when it comes to proceeding with specific transactions, such as terms of references, short lists, etc.

4 Introduction is actually section one.

5 First, although these are not factors per se, they are normative reasons that can be instructive if we consider the factors as the root causes of the reasons. Second, we can derive factors from any kind of normative “best practices” or “lessons-learned” if they are stripped of their value-orientation (Struyk, 2007, p. 65).

6 Please note here that “profile” stands for a group of project success criteria that include conformity of the goods and services, national visibility of the project, project reputation with ID agencies, and probability of additional funding for the project (Diallo & Thuillier, 2004).

7 See footnote 6.

8 As one of the main objectives of this study is to highlight the influence of the CSFs on project success dimensions, we do not consider a single project success construct but instead two distinct project success dimensions. Anyway, the structural equation model with a single project success construct shows a poor fit (see section “Data Analysis”).

9 The items may be defined using two measurement models: formative or reflective models. Here we have chosen the reflective model, since we assume that the manifest or observed variables (not shown in Figure 1) are a reflection of the latent constructs CSF, project management success, and Del success. This is different from the formative models that are often used for multiple regression analysis, for example.

10 The first-order five CSFs are assumed to be a reflection of the second-order latent CSF construct (PM CSF); hence, the paths are from PM CSF to the five CSFs.

11 The two first invitations were sent on Mondays, March 3 and 17, 2008, the third, on Wednesday, March 26 (Faught, Whitten, & Green, 2004 suggest Wednesday as the ideal day), and the last, on Tuesday, April 1, 2008.

12 Actually, this rather modest response rate is of the same magnitude as the response rates other researchers obtained under similar mailing conditions (see, for example, Au & Tse, 2001; Bellizi & Hasty, 2002; Diallo & Thuillier, 2004, 2005).

13 This is actually better than the 12% rate that Diallo and Thuillier (2004, 2005) arrived at in their study. Khang and Moe (2008) didn't reveal the perceived failure rate in theirs.

14 As such, SEM has been used to gain insight into the complex interactions between the CSFs and project success dimensions. In fact, the alternative multiple regression analysis requires only one dependent variable at a time and low correlations among the CSFs, which is not often the case in project management research (e.g., Ika, Diallo, & Thuillier, 2009).

15 Once again, this is one of the main reasons why we have turned in this study to SEM instead of multiple regression analysis.

16 NPC (national project coordinator)

This material has been reproduced with the permission of the copyright owner. Unauthorized reproduction of this material is strictly prohibited. For permission to reproduce this material, please contact PMI or any listed author.

© 2010 Project Management Institute



Related Content