Analysis of Critical Success Factors Influence on Critical Delays for Water Infrastructure Construction Projects in the Abu Dhabi emirate Using PLS-SEM Method

The objective of this study is to investigate the significance impact of critical success factors on critical delays in the field of water infrastructure construction projects (WICPs) in the Abu Dhabi emirate in particular. Investigation was conducted utilizing quantitative approach by means of questionnaire survey to examine the understanding of professionals engaged in water infrastructure construction towards several critical success factors influencing critical delays. A total of 323 completed responses from owners, consultants and contractors representatives were gathered against 450 distributed questionnaires. The gathered questionnaires were analysed using an advanced multivariate statistical method of Partial Least Square Structural Equation Modelling (PLS-SEM). Data analysis was conducted in two major phases. The first phase involved a preliminary analysis of the data, to ensure that the data adequately meet the basic assumptions in using SEM. The second phase applied the two stages of SEM. The first stage included the establishment of measurement models for the latent constructs in the research. After confirming the uni-dimensionality, reliability and validity of the constructs in the first stage, the second stage developed to test the research hypotheses through developing the structural models. The results indicated that Project Management Process (PMP), Project Manager’s Competency (PMC), Project Team’s member Competency (PTC), Project Organizational Planning (POP), Project Resources’ Utilization (PRU) and Project Organizational Commitment (POC) had significant positive effects on Critical Delay Factor Evaluation (CDFS). From the results of moderation analysis revealed that Project Benchmark Characteristics (PBC) is positively moderate the effects of Project Management Process (PMP), Project Manager’s Competency (PMC) and Project Team’s member Competency (PTC) and Project Organizational Planning (POP) on Critical Delay Factor Evaluation (CDFS).


Introduction
The construction industry in general and including water infrastructure sector is large, complex, volatile, risky, and requires tremendous capital outlays and tight money (Tumi, Omran & Pakir, 2009).It provides a bigger challenge to maintain its scheduled time, budgetary cost, and appropriate quality (Elawi, Algahtany & Kashiwagic, 2016).A prime critique coming up against the construction sector including water infrastructure construction projects is the increasing rate of occurred delays in construction project delivery (Tumi et al., 2009).From the available review, several studies have spotlight on identifying causes of project delays or critical success factors; however, none of the previous conducted studies have investigated relation among the critical success factors and critical delays in construction industry, in general, nor in water infrastructure construction projects in particular.Hence, this study adopted Structural Equation Modeling (SEM) to assess the influence of critical success factors on critical delays.The analysis selected PLS approach to Structural Equation Modeling (SEM) as this approach is more recommended and advised as most appropriate method for testing the causal relation (Hair, ringle & Sarstedt, 2011).In addition, according to Ng, Tang and Palaneeswaran (2010), Structural Equation Modeling shows better functionality than other multivariate techniques including multiple regression, path analysis, and factor analysis.

Critical Delays Review
Numerous studies have questioned several dissimilar factors that lead to delay in various types of infrastructure construction projects.Generally; delay in construction projects is regarded as one of the most repeated difficulties in the construction field and it has an unfavorable impact on construction project success against time, cost, quality, and safety and no kind of construction projects has got out of the astounding ghost of time overruns (J.Sweis, Rateb Sweis, Abu Rumman, Abu Hussein & Dahiyat, 2013).The causes and impacts of delay factors in construction industry not only vary from project to project but also from geographical location to another due several reasons including and not limited to the environmental, the topographical and the technological constraints (Sweis, 2013;Shebob, Dawood & Shah, 2012).Shebob et al. (2012) mentioned in his study that in addition to country and projects variances in term of delay there is a certain projects are only a few days late while some projects are delayed by over a month or a year.Kazaz, Ulubeyli and Tuncbilekli (2012) examined various causes of time delay in the context of Turkish construction industry and the levels of their significance, design and material changes, delay of payments and cash flow problems are the most predominant delay factors in Turkish construction industry.Motaleb and Kishk (2013) examined problems causing delays on construction projects in the United Arab Emirates; they investigated the causes and effects behind the delays that pertain to the delivery of construction projects in the United Arab Emirates, they identified and ranked the most key factors as follows: Change orders, Inadequate capabilities of client delegate, Delay in decision making by client delegate, Poor experience of client in construction, Insufficient management and supervision, Lack of experience of project team, Inflation/prices fluctuation, Poor time estimating, Construction materials delivery related problems, Improper project planning / scheduling, Imprecise cost estimating, High bank interest rate, Client's poor financial statement, Extravagant restriction to client ,Improper construction methods.Elawi et al. (2016) investigated the reasons of the time delay in infrastructure projects in the Mecca province in Saudi Arabia.Their study concluded that factors contributed for the majority of time overruns were; land acquisition, contractor' lack of expertise, change order, and obtaining approvals and permits against underground utilities.Obodoh and Obodoh (2016) studied the major causes and effects of cost overrun and time delays in the infrastructure construction projects in Nigeria, the study revealed that, insufficient number of equipment, Imprecise time assessment, payment difficulties, change orders, poor cost estimate, inadequate site supervision and management, lack of modern equipment, shortage of construction materials, poor skills of project team, inaccurate project planning and scheduling and contractors' financial difficulties were the main caus es of delay in Nigeria's construction projects.Durdyev, Omarov and Ismail (2017) studied causes of construction delay in infrastructure construction projects of residential nature in Cambodia, the study showed that shortage of materials on site, unrealistic project scheduling, late construction material delivery, shortage of competent labor, change orders, complexity of project, labor absenteeism, delay in payment by the owner against invoiced completed works, poor site management, delay by subcontractor, accidents due to inadequate site safety are ranked and evaluated by the representatives of two main stakeholders of contractors and consultants as the major causes of project delays in Cambodia.

Critical Success Factors Review
Around the world, many researchers have been inspired to investigate project critical success factors.Toor and Ogunlana (2009) attempted to extract the understanding of construction experts on critical success factors (CSFs) pertained to construction projects of large-scale size in Thailand.Their study revealed that most of the high-rated CSFs are related to project planning and control, personnel, and involvement of client.However, the top ten CSFs according to the study were ranked as follows: effective project planning and control, sufficient resources, clear and detailed written contract, clearly defined goals and priorities of all stakeholders, competent project manager, adequate communication among related parties, competent team members, Knowing what client really wants, responsiveness of client and awarding bids to the right designers/contractors.Tabish and Jha (2012) investigated important factors for success of construction projects pertains to public sector in India, the success factors resulted from this study were categorized into generic and specific natures, findings for generic type were: owners requirements need thorough understanding and precise definition, a high level of trust among the project bodies participants, on time and helpful decision from higher management, availability of all required resources as planned during all execution phases of project, top management's support, and consistent monitoring and feedback by higher management, while success factors of particular character were: thorough understanding of project manager and contractor on their part scope, comprehensive and thorough investigation of the project site in the pretender stage, regular and periodic monitoring and feedback by the owner representative, avoid bureaucratic interference, absent of social and political interferences , well identified and clear threaded scope of work, quality control and quality assurance activities, and adequate communication among all project participants.Mustaffa and Yong (2013) evaluated the severity-identified factors on construction project success distributed to clients, consultants and contractors.Their study identified fifteen (15) factors to be accepted as a critical to the success of construction projects and suggested a strong consistency in perception between respondents in recognizing the significance of human-related factors such as competence, commitment, communication and cooperation towards the success of a construction project.Thi and Swierczek (2010) have also studied causes of CSFs in Vietnam construction projects and their study revealed that manager competencies, member competencies and external stability have important positive relationships to the success criteria.Gudienė, Banaitis, Podvezko & Banaitienėet (2014) conducted an empirical study in Lithuania to evaluate critical success factors for construction projects, based on the study results, ten factors including project manager competence, project management team members' competence, project manager coordinating skills, client clear and precise goals/objectives, project value, project management team members' relevant past experience, project manager organizing skills, project manager effective and timely conflict resolution, client ability to make timely decision, and project manager experience were determined as the most significant success factors for Lithuanian construction projects.
Several researchers have pointed out various findings about critical success factor in construction projects such as Gunduz and Yahya (2015) conducting a study aimed to determine the critical success factors in the construction industry in Middle East region and in the United Arab Emirates market specially.These factors were evaluated for their influence and contribution to the real performance of the project from the perspective of three criteria: schedule, cost, and quality.Mukhtar, Amirudin, Sofield & Mohamad (2016) investigated success factors in public housing projects in Nigeria and serves as a guide reference to housing policy makers.The study identified seven CSFs for public housing projects in Nigeria, these factors are; availability of competent personnel, effective project management, proper design and appropriate location, powerful financing system for housing, and sufficient political support.
Based on an analysis of the literature that has been outlined earlier, it has become apparent that there is a plenty of factors with the potential to influence the project success.However, according to Altarawneh, Thiruchelvam & Samadi (2017), due to their frequent use in previous studies and because much researches were concluded their studies results by them in some way, the six most significant success factors in determining project success identified by various number of researchers and their attributes have been chosen for further investigation in this study are listed in Table 1.Contractor is committed to zero variation orders

Moderator Factors Review
In addition to the critical success factors that have been identified in the literature, the impact of two other moderate factors has been investigated, Project Benchmark Characteristics (PBC) and Project External Environments (PEE), which are believed to affect the relationship between the critical success factors and project critical delays (Park, 2009;Tan & Ghazali, 2011;Yang, Huang & Wu, 2011;LI, Arditi & Wang, 2012;Gudienė et al., 2013;Yong & Mustaffa, 2013;Gudienė et al., 2014).In the available literature, project Benchmark characteristics and project external environments have long been disregarded as being critical success factors; however, many construction projects witnessed status of failure due to problems within projects (Thi and Swierczek, 2010).
This external environmental factor contains several items, which are external to the project but have an influence on the construction project performance, either positively or negatively (Thi and Swierczek, 2010).A number of external environmental factors, such as economic, political, legal, social and those factors linked to new technologies or even factors related to nature, may influence construction project performance (Hwang, Zhao & Ng, 2013).However, according to Jin et al. (2012), some of these externals influence the construction project at all phases of the project life cycle, such as weather conditions or the social environment.According to some researchers, these factors sometimes, have a considerable impact that they resulted in project termination at the construction stage (Jin et al., 2012;Yong and Mustaffa, 2013;Zhao et al., 2013;Gudienė et al., 2014).
According to several researchers, project size, value, uniqueness of project activities, the density of project and project urgency were specified as major critical success factors within the project (Ng, Wong & Wong, 2012;Gudienė et al., 2013;Zhao et al., 2013;Gudienė et al., 2014;Yang et al., 2015;Shehu et al., 2014;LI et al., 2012;Yong & Mustaffa, 2013;Tan & Ghazali, 2011).In addition to that, Gudienė et al (2014) pointed out that several large construction projects that contain more than 100 activities exceed their contractual deadlines.Also, several researchers highlighted that; the project manager's performance in the work can be significantly affected by the uniqueness of the construction activities (LI et al., 2012;Gudienė et al., 2013;Yong & Mustaffa, 2013;Gudienė et al., 2014).They believed that, it is easier for project managers to plan, schedule and monitor construction project activities if a project has tasks that are more standard rather than complex activities.
According to them, Project density also affects the overall performance.That is, will influence the allocation of project resources, including man-hours and machineries.In a way, due to imposed resource constraints, project managers are often constrained to implement overtime procedures, which lead to exceed the allocated budget, or they are strained to delay activities running for the same manpower resources, which cause delays in project completion.Some researchers related urgency to project success (Gudienė et al., 2014).On the other hand, project performance criteria for some cases are not met due to the urgency impact (Yang et al., 2011;LI et al., 2012).From the presented literature review and many previous several researches, two moderator factors were identified in several studies and listed with their attributes in Table 2.

Research Hypothesis
Following the conduct of thorough and intensive literature review, codes and description of the research hypotheses are represented in Table 3.

Research Model
In order to specify the research hypotheses targeted in Table 3, a research structural model was developed in this study.The research structural model is intended to test 6 hypotheses related to direct effects from (PMP), (PMC), (PTC), (POP), (PRU) and (POC) on Critical Delay Factor Evaluation (CDFS).The study also examined the moderation effects of (PBC) and (PEE) on the relationships of the other constructs.Figure 1 illustrates the hypothesized direct and moderation effects in the research structural model.

Figure 1. Hypothetical model
The factors are known as exogenous latent variables meanwhile the items are known as relative manifest variables.The details of the exogenous latent and relative manifest variables of the adopted model are shown in Table 1 &2.

Research Method
This study followed quantitative research approach including data collection by means of structured questionnaire survey.The survey was conducted between main owners of water projects, qualified consultants and contractors registered in the vender's list of the main owners who are either handle or conduct all released projects for the last ten years.A total number of 450 questionnaires were released between the selected companies (owners, consultant & contractors).As a result, 323 completed questionnaires were returned back by the participants.The collected questionnaires were analysed using SPSS software for evaluate the received questionnaires against the demographic information of the respondents as summarized in Table 4.

Overall CFA Model
As highlighted earlier, structural equation, modelling is a data analytic technique commonly used to examine patterns of relationships among constructs (Cooper & Schindler, 2006).The latent constructs in individual CFA models were all measured by several multi-item scales.The inclusion of all items and relative errors in the measurement and structural models leads to a complex and non-stable model because too many parameters need to be estimated.Thus, to overcome this problem, this research utilised parcels as indicators of latent constructs in individual CFA models.Parcels are aggregations (sums or averages) of several individual items.Using parcels as indicators of latent construct commonly have better reliability as compared with the single items (Coffman & MacCallum, 2005).As the result of using item-parcelling procedure, the latent constructs in individual CFA models of (PMP), (PMC), (PTC), (POP), (PRU), (POC), (PBC) and (PEE) were converted into observed variables so that they could easily construct the overall measurement and structural model and reduce the model complexity.
Confirmatory factor analysis was used to assess the overall measurement model.The model comprises all of the first and second order constructs proposed in this study.Figure 2 depicts the overall CFA model.
Figure 2. Overall CFA Model

Reliability and Convergent Validity
Table 5 represents the result of Cronbach alpha and convergent validity for the Overall CFA model.As shown in Table 5, the results of assessing the standardized factor loadings of the model's items indicated that the initial standardised factor loadings of items were all above 0.6, ranged from 0.824 to 0.914.

Discriminant Validity
Table 6 represents the discriminant validity of the Overall CFA Model.Note: Diagonals represent the square root of the average variance extracted while the other entries represent the square correlations.
The inter-correlations between the 9 sub-constructs in Overall CFA Model ranged from -0.074 to 0.283, which were below the threshold 0.85 as recommended by Kline (2005).Further, as shown in Table 4 20 , the correlations were less than the square root of the average variance extracted by the indicators, demonstrating good discriminant validity between these factors (Kline, 2005).Upon examining goodness to fit of data, convergent validity and discriminant validity of the measurement model, it can be concluded that modified measurement scale to assess the constructs and their relative items in overall measurement model was reliable and valid.

Structural Models
The structural equation model is considered as the second major process of structural equation modeling analysis.Once validation process of the measurement model is confirmed, then representation of the structural model can be established by identifying the relationships between the constructs.The structural model provides details on the links between the variables (Nafisi, A. & Nafisi, S., 2015).It displays the particular details of the relationship among the independent or exogenous and dependent or endogenous variables (Hair, et al., 2006;Ho, 2006).Evaluation of the structural model spotlight firstly on the overall model fit, followed by the size, direction and significance of the hypothesized parameter estimates, as shown by the one-headed arrows in the path diagrams (Hair, et al., 2006).The final part included the confirmation process of the structural model of the study, which was established on the projected relationship among the identified and assessed variables.In the present study, the structural model was supposed to test the research hypothesizes, utilizing PLS method and bootstrapping with 1000 replications.
The next sub-sections discuss the development of structural model to test the research hypotheses described in Table 3.

Direct Effects of Constructs
In the structural model, the direct causal effects of (PMP), (PMC), (PTC), Project Organizational Planning (POP), (PRU) and (POC) on Critical Delay Factor Evaluation (CDFS) were examined.These effects refer to the 6 hypotheses namely: H1, H2, H3, H4, H5 and H6 respectively.The Smart-PLS model is portrayed in in Figure 3.The coefficient parameters estimates are then examined to test the hypothesized direct effects of the variables, which were addressed in Table 3.The path coefficients and the results of examining hypothesized direct effects are displayed in Table 7.  7, all paths from (PMP), (PMC), (PTC), (POP), (PRU) and (POC) to Critical Delay Factor Evaluation (CDFS) were statistically significant as their p-values were all below the standard significance level of 0.05.Thus, the hypotheses H1, H2, H3, H4, H5 and H6 were supported.

Moderation Effects of Project Benchmark Characteristics (PBC)
The Smart-PLS model with interaction terms to examine the moderation effects of Project Benchmark Characteristics (PBC) is portrayed in Figure 4.The moderation (PBC) on the effects of (PMP), (PMC), (PTC), (POP), (PRU) and (POC) as independent variables on Critical Delay Factor Evaluation (CDFS) as dependent variable (DV) were examined as presented in Table 8.Further, the path coefficient was used to evaluate the contribution of each interaction term on the DVs.As shown in Table 8, the interaction terms of (PBC) with (PMP), (PMC) and (PTC) and (POP) had significant effects on Critical Delay Factor Evaluation (CDFS) as their p-values were all lower than the standard significance level of 0.05.These results demonstrated that (PBC) moderates the effects of (PMP), (PMC), (PTC) and (POP) on Critical Delay Factor Evaluation (CDFS).Therefore, hypotheses H7a, H7b, H7c and H7d were supported.
Conversely, the interaction terms of (PBC) with (PRU) and (POC) had not any significant effects on Critical Delay Factor Evaluation (CDFS) as their p-values exceeded the standard significance level of 0.05.This result demonstrated that (PBC) could not moderate the effects of (PRU) and (POC) on Critical Delay Factor Evaluation (CDFS).Therefore, hypotheses H7e and H7f were rejected.

Moderation Effects of Project External Environments (PEE)
The Smart-PLS model with interaction terms to examine the moderation effects of Project External Environments (PEE) is portrayed in Figure 5.In sum, the model exhibits acceptable fit and high predictive relevance.The moderation effects of (PEE) on the effects of (PMP), (PMC), (PTC), (POP), (PRU) and (POC) as independent variables on Critical Delay Factor Evaluation (CDFS) as dependent variable (DV) were examined as presented in Table 9.Further, the path coefficient was used to evaluate the contribution of each interaction term on the DVs.As shown in Table 9, the interaction terms of (PEE) with (PMC), (POP), (PRU) and (POC) had significant effects on Critical Delay Factor Evaluation (CDFS) as their p-values were all lower than the standard significance level of 0.05.These results demonstrated that (PEE) moderates the effects of (PMC), (POP), (PRU) and (POC) on Critical Delay Factor Evaluation (CDFS).Therefore, hypotheses H8b, H8d, H8e and H8f were supported.
Conversely, the interaction terms of (PEE) with (PMP) and (PTC) had not any significant effects on Critical Delay Factor Evaluation (CDFS) as their p-values exceeded the standard significance level of 0.05.This result demonstrated that (PEE) could not moderate the effects of (PMP) and (PTC) on Critical Delay Factor Evaluation (CDFS).Therefore, hypotheses H8a and H8c were rejected.

Conclusion
Structural model was developed to examine 6 hypothesized direct effects and 12 hypothesized moderation effects of Benchmark Characteristics (PBC) and Project External Environments (PEE).These were done by conducting the path analysis using SMART-PLS 2.0 and testing the significant of the path coefficients for each hypothesized path.
The results indicated that Project Management Process (PMP), Project Manager's Competency (PMC), Project Team's member Competency (PTC), Project Organizational Planning (POP), Project Resources' Utilization (PRU) and Project Organizational Commitment (POC) had significant positive effects on Critical Delay Factor Evaluation (CDFS).The results also indicated that Project Organizational Planning (POP) is the most significant predictor of Critical Delay Factor Evaluation (CDFS), followed by Project Team's member Competency (PTC) and Project Manager's Competency (PMC).
From the results of moderation analysis, it was found that Project Benchmark Characteristics (PBC) positively moderate the effects of Project Management Process (PMP), Project Manager's Competency (PMC) and Project Team's member Competency (PTC) and Project Organizational Planning (POP) on Critical Delay Factor Evaluation (CDFS).
The results also showed that Project External Environments (PEE) positively moderates the effects of Project Manager's Competency (PMC), Project Organizational Planning (POP) and Project Resources' Utilization (PRU) on Critical Delay Factor Evaluation (CDFS).While the effect of Project Organizational Commitment (POC) on Critical Delay Factor Evaluation (CDFS) was inversely moderated by Project External Environments (PEE).

:
a : Average Variance Extracted = (summation of the square of the factor loadings)/{(summation of the square of the factor loadings) + (summation of the error variances)}.bComposite reliability = (square of the summation of the factor loadings)/{(square of the summation of the factor loadings) + (square of the summation of the error variances)}.

Figure 3 .
Figure 3.PLS Analysis of the Structural Model for Direct EffectsThe value of R2 for Critical Delay Factor Evaluation (CDFS) was 0.172.This indicates, 17 percent of variations in Critical Delay Factor Evaluation (CDFS) are explained by its 6 predictors (i.e, (PMP), (PMC), (PTC), (POP (PRU) and (POC)).Overall findings showed that the R² value satisfies the requirement for the 0.30 cut off value as recommended by Patterson (2013).The values of Q2 for Critical Delay Factor Evaluation (CDFS) was 0.129, far greater than zero, which refers to predictive relevance of the model as suggested by Chin (2010).In sum, the model exhibits acceptable fit and high predictive relevance.

Figure 4 .
Figure 4. PLS Analysis of the Structural Model for Moderation Effects of Project Benchmark Characteristics (PBC) The values of R2 for Critical Delay Factor Evaluation (CDFS) was 0.222, above the threshold of 0.1 as recommended by Patterson 2013.The values of Q2 for Critical Delay Factor Evaluation (CDFS) was 0.166, far greater than zero, which refers to predictive relevance of the model as suggested by Chin 2010.In sum, the model exhibits acceptable fit and high predictive relevance.

Figure 5 .
Figure 5. PLS Analysis of the Structural Model for Moderation Effects of Project External Environments (PEE) The value of R2 for Critical Delay Factor Evaluation (CDFS) was 0.191, above the threshold of 0.1 as recommended by Patterson (2013).The values of Q2 for Critical Delay Factor Evaluation (CDFS) was 0.141, far greater than zero, which refers to predictive relevance of the model as suggested by Chin 2010.In sum, the model exhibits acceptable fit and high predictive relevance.The moderation effects of (PEE) on the effects of (PMP), (PMC), (PTC), (POP), (PRU) and (POC) as independent variables on Critical Delay Factor Evaluation (CDFS) as dependent variable (DV) were examined as presented in Table9.Further, the path coefficient was used to evaluate the contribution of each interaction term on the DVs.

Table 1 .
Critical Success Factors (CSFs) and their attributes

Table 2 .
Moderator factors and their attributes

Table 3 .
Research Hypotheses Codes and Descriptions

Table 4 .
Demographic information of respondents

Table 5 .
Results of Cronbach Alpha and Convergent Validity for Overall CFA Model

Table 6 .
Discriminant validity of Overall CFA Model

Table 7 .
Examining Results of Hypothesized Direct Effects of the Constructs

Table 8 .
Moderation Effects of Project Benchmark Characteristics (PBC)

Table 9 .
Moderation Effects of Project External Environments (PEE)