A Structural Equation Modelling Approach to Improving Progress Payment Systems Through Common Data Environment (CDE) Implementation

Abstract

The construction industry in South Africa faces challenges with the current payment system used to manage progress payments. Contractors often experience delays in progress payments for completed works. These late payments stem from the improper management of progress payment procedures, namely, information, communication, and collaboration, as well as corruption. This study proposes the integration of common data environment (CDE) as it has emerged central in managing information, improving communication and collaboration in a transparent manner. However, the implementation of CDE is facing challenges in the industry. Therefore, the study aimed at developing a model based on the implementation of CDE to uphold efficiency in the management of payment systems for progress payments. A systematic review was conducted to examine the enabling factors, characteristics of CDE in managing progress payment challenges, and benefits of integrating a payment system in a CDE platform. Furthermore, the study utilised questionnaire surveys to purposively collect data from construction professionals who implemented CDE in their projects. From 201 valid questionnaire responses, a structural equation model was developed; testing for the reliability, validity, model fit, and hypotheses was conducted using AMOS and ADANCO. The findings revealed that enabling factors such as quality technology and quality assurance team are the strongest enablers, followed by training and policy. The findings further predict that CDE integration will improve the management of the payment system by 0.589. The study provides theoretical and practical guidance for researchers, policy makers, and construction professionals seeking to strengthen CDE-based payment system frameworks in South Africa. Furthermore, it is recommended to adopt the method of questionnaire surveys and SEM to validate variables and establish their influence on one another to improve generalisation.

1. Introduction

Since 1999, South Africa’s public sector has been working to improve the management and funding of infrastructure development through a public–private partnership approach [1]. A public–private partnership is a procurement system that utilises private service providers, managerial expertise, and innovative commercial skills to develop and manage public infrastructure and services [2]. According to Gcaba and Yalezo [3], the construction industry in South Africa commonly adopts design-bid-build and design-build methods in this procurement system. Common in these methods is the employment of a contractor who executes the project [4].

The contractor is employed to undertake the management and construction of the project [5]. The appointment is finalised by signing a contract outlining the client’s payment schedules during the construction phase at regular intervals to the contractor called progress payments [6]. The construction phase payment claims from clients have been a long-standing issue, often being complicated, late, or non-paid, causing frustration for contractors [7]. According to Mahamid [8], late payments pose a significant threat to project funding as contractors often rely on the project’s cash flow due to financial constraints. Furthermore, these late payments are a major problem that affects the construction industry from one project to another [9]. This late payment challenges stems from different circumstances.

Despite the planned current payment system for progress payments, the procedures have been poorly managed, creating an opportunity for public officials to manipulate contractor payments within government departments for their personal gain [10]. Additionally, the current practices of processing, managing, and administering payment information have been unsuccessful, even though the relevant personnel are employed [11]. Furthermore, contractors frequently encounter conflicts with other parties due to disagreements arising from lack of collaboration during the execution of these payments leading to disapproved claims [12]. According to Akinsiku and Ajayi [13], this delay in progress payments starts when there is no agreement on the variation of works between client, consultant, and contractor, coupled with improper documenting in claims. Consequently, late payments are primarily attributed to inadequate information management, poor communication, lack of collaboration, and corruption activities [14,15,16]. Thus, this study suggests integrating a common data environment for effective transparent information sharing and collaboration, as it provides immutable data storage, fostering a trustworthy environment for payment processing [17,18].

The Common Data Environment (CDE) is a shared digital project space with defined access areas, status definitions, and robust workflows for data collection, evaluation, sharing, and approval [19]. According to Matthei and Klemt-Albert [20], the central structuring of project information in a CDE platform ensures that the system provides updated data transparently. Furthermore, the CDE is crucial for establishing a centralised information system, ensuring proper access and data collection and their maintenance for all relevant parties [21]. Moreover, the CDE can significantly enhance progress payment management by simplifying the bill of quantity generation, managing cash flow, and enabling automated payments [22]. Additionally, the CDE is a tool for managing contracts that evaluates requests, modifications, variations, claims, progress payments, tenders, bids, and offers [21]. Moreover, this system avoids the difficulties associated with a traditional approach in processing payments by making development a step-by-step procedure for clients and contractors [23]. There are several studies which have illustrated CDE-based frameworks for payment system in the construction industry.

For example, Sigalov et al. [24] designed a collaborative system using BIMcontracts to automate progress payments. The system is developed in a way such that all the data needed can be stored in the model and transferred to smart contracts to update payments at intervals. This creates security regarding transactions and ensures transparency about payments. The study used literature review and observation (case study) methods. Nguyen et al. [25] illustrated the application of BIM QTO for generating bills of quantities and establishing progress payments during the design phase, with updates throughout the project. The model facilitates automated cost estimation by extracting quantity data from design inputs, ensuring accurate information for project updates. This study conducted a literature review and employed the Delphi technique. Sarkar et al. [26] designed an integration of BIM-IoT and Blockchain to transfer information from site to the system and automate contactors’ payments through smart contracts. Project data can be collected in a logical manner in real time with BIM using IoT. Then, the data can be extracted from BIM to blockchain to create a transparent payment system. This study conducted a literature review and observations (case study). Nevertheless, the utilisation of CDE technologies faces challenges in the industry and thus its implementation is not effective [27]. This is because these frameworks do not include enabling factors for a successful implementation. According to the contingency theory, for a system to function effectively in information processing, context and structure need to be reasonably compatible, including the driving external and internal factors with the mediator [28,29]. Therefore, there was a need to establish strategies to drive the implementation of CDE-based payment systems. Thus, this study aimed at developing a common data environment centred model which employs enabling measures to manage current challenges and ensure an effective payment system. This was done by achieving the following objectives:

• To establish enabling factors to improve the implementation of a CDE.
• To evaluate the characteristics of a CDE in managing progress payment challenges.
• To identify the benefits of integrating a CDE in construction payment system.
• To develop a model for enhancing the management of payment systems through the implementation of a CDE.

1.1. Challenges Faced During CDE Implementation

The challenges develop from the lack of technology training and the inability to value the functionality of the CDE [30]. Kassem et al. [31] concurred that when participants are not trained on how to operate the CDE, they do not succeed in implementing the platform. Furthermore, participants express concerns regarding the implementation of the CDE process due to the absence of established guidelines or standards [32]. Poirier and Soyez [30] asserted that the lack of policy impedes the use of CDE since participants are uninformed of what it is, their rights and obligations, its functions, and usage guidelines. Additionally, the utilisation of CDE is hindered by changing from the normal way of operating to adopting technology without appointing managers to maintain the platform and ensure utilisation [33]. According to Migilinskas et al. [32], participants face difficulties in utilising the platform because of a lack of top management mediation. Moreover, the platform struggles to operate due to faults in the technology because it is not properly installed [34]. Furthermore, the model was not properly developed, did not fulfil its anticipated capacity, and failed to accomplish the mission [35]. These challenges revealed that the CDE does not reach implementation maturity because of the lack of training, policy, top management mediation, and the poor quality of technology used. The following section evaluates how these challenges can be managed.

1.2. Enabling Factors for Improving CDE Implementation

The concepts behind the model are built from the literature. These concepts are endorsed by contingency theory, which states that in order to have a well performing system, external and internal variables should be considered as suitable inputs and supported by mediators as third variable [28,29]. Additionally, the theory provides explanations of how external elements like technology, culture, and environment influence the structure and operations [36]. Loukis et al. [37] asserted that maximum performance results from adopting proper values of some structural factors that match the situation. According to Shao et al. [38], in contingency theory, there is no ideal method to execute a system that results in optimum performance and thus system efficiency is instead attained through a variety of internal and external conditions and settings. Thus, this conceptualised CDE centre model theorises the enablers of a well-performing system as follows: training and policy development serving as external factors, employing quality assurance team as mediators, and quality technology serving as the structure and internal factor [29,39].

The measures identified as enablers using the concepts and contingency theory are discussed below:

Training: According to a study by Lai et al. [40], training program significantly increased technology usage by fostering a positive attitude towards utilisation through a clear purpose and detailed instructions on various aspects of the tools. Furthermore, training should be provided to encourage the acceptance and deployment of novel technologies [41]. Fathima et al. [42] asserted that to successfully adopt building information modeling, training should be undertaken. Despite these qualities, the study of Kamau [43] concluded that training alone cannot drive the adoption of technology and there is a need to add other extensive systematic factors.

Policy: Moreover, the employment of a technology policy technique is considered as a crucial task targeted at convincing team members to integrate the tools [44]. According to Qin et al. [45], policy is a critical factor that acts as an incentive tool to increase the perceived use and acceptance of the building information modeling. Nevertheless, the study of Phiri and Kanjo [46] found that the employment of policy alone does not drive technology utilisation.

Quality assurance team: The client, senior management, and technical personnel should offer prompt assistance to promote the adoption and utilisation of an information sharing environment [47]. Ebert et al. [48] asserted that the management team ensures the continuous processing, monitoring, and adequate performance of the platform without losing control of technology operations. Moreover, managers should be employed to make sure that the model is operating effectively to maximise its benefits and raise awareness about the tools’ applications amongst the participants [25]. Contrarily, Lin et al. [49] reported that though a FM engineer was appointed in their case study to manage the platform, the CDE and the team were unable to perform their job because the engineer was not available when needed for the project.

Quality technology: The development of quality technology, managed by qualified professionals, can significantly increase the implementation of the tool [25]. According to Batty et al. [50], the adoption of quality technology expedites the development of a computing infrastructure capable of addressing complexity. Furthermore, Wang et al. [51] assert that reducing the complexity of the technology is essential to ensure it is easy for participants to utilise. According to Ren [52], despite the technical quality of a technology, adoption depends heavily on other organisational cultures.

1.3. Characteristics of the Common Data Environment

The CDE facilitates efficient data management by creating open information sharing between various parties through the connection of information and documents [53]. The foundation of the CDE information management process presents all the needed data in a single source of truth, which promotes smooth data sharing and enables continuous collaboration [27]. Moreover, the CDE acts as a hub for communication, task delegation, and information sharing amongst participants [19]. Furthermore, the CDE makes it possible to manage and access data from several participants to guarantee the flow of quality information [27].

1.3.1. Information Management

A CDE system for information management enhances asset management, data transmission, and visualisation throughout the whole project lifecycle [54]. Moreover, the CDE is essential to project information management because it facilitates information sharing and manages procedures that guarantee the timely and economical completion of projects [55]. According to Losev [56], the CDE can manage information by gathering, organising, and disseminating data through a predetermined structure. However, Lin et al. [49] reported that in their project, the participants were having difficulties in operating the model and were not updating construction information, leading to missing nongeometric information in the BIM models and hardcopy documents.

1.3.2. Communication

The CDE platform is an environment in which the parties involved can coordinate and communicate with each other to prevent data or document duplication [57]. The platform enhances the communication between client, consultant, and contractor according to the planned format [58]. According to Patacas et al. [59], the creation of a CDE allows for more flexibility in the structure of communication plan, making it easier for parties to communicate with one another. In contrast, there was lack of interface communication in this case, and it was not clear who is responsible for what and there was no guidance [60].

1.3.3. Collaboration

A CDE platform offers a chance to improve collaboration and productivity while managing project information, facilitating data interchange performance and efficiency [61]. Furthermore, the platform was established as a set of data for any type of project that demands collaboration and information sharing among different participants [17]. Özkan and Seyis [21] asserted that the CDE improves effective collaboration by offering centralised communication that facilitates correct information management and sharing among all stakeholders. In contrast, in this project, it was discovered that it was difficult to build collaborations between client and participants because there was no clear definition of the procedure on how to protect ownership and control on the platform, and participants were not aware who was responsible for what [62].

1.3.4. Transparency

A CDE platform is developed to enhance information planning and decisions that are traceable and transparent [20]. According to Abegaz [63], using the CDE is crucial for fostering confidence and guaranteeing strong data protection in a safe way that promotes transparency. Jaskula et al. [64] asserted that the CDE reduces disagreements and guarantees a fair and impartial environment for all parties concerned through the provision of an unchangeable and transparent project history. Nevertheless, Aibinu and Papadonikolaki [65] found that the interference of the mediator was not available; thus, the different groups worked on their own data and updated it on the system once completed, and this created an opportunity for inadequate collaboration and communication, thus increasing production cost.

1.3.5. Payment System

Ahmadisheykhsarmast and Sonmez [66] asserted that on-time progress payments are important for the efficient management of a contractor’s cash flow and the project’s success. Additionally, it is important to provide information in a transparent manner to improve and process payments on time [24]. According to Parn and Edwards [67], in a CDE context, digital information can be exchanged to verify and maintain a competitive payment process that produces trackable transactions. The CDE creates a system that uses data monitoring to automate the payment process for progress payment management [18]. This is because the platform authorises traceable, high-quality information regarding the ongoing project procedures [68]. This common environment makes it possible to follow recorded data, which allows payment transactions to be communicated through the platform with all parties [69]. Additionally, the CDE platform facilitates the communication of payment processes with all parties through the shared environment, enabling easy tracking of recorded transactions. Ye et al. [70] emphasised that CDE smoothens the payment certificate acceptance process between contractors and clients by managing and maintaining all payment-related files and documentation. Furthermore, the CDE improves payment performance speed, security, traceability, and immutability [24]. Nevertheless, Sattineni and Macdonald [35] reported that in their case the platform was not accurately built, and it was not sufficient to extract accurate quantities for cost estimates. Additionally, since the platform was not developed properly in this case, some project elements were missing, and there was incompetency when developing the bill of quantities for the project [71].

1.4. Managing Payment Systems Through a Common Data Environment

The payment system management for progress payments involves gathering information and documents, submitting these data for payment certification, the authorization of the payment data and conclusion of amounts, and generating a tax invoice and processing the approved payment [72,73,74,75]. The management of this payment system can be made ease by integrating the CDE platform [18].

1.4.1. Gathering Data and Submitting for Payment Certificates

The parties involved gather all the information and documents regarding the works on site and share the data on the CDE platform [76]. In a CDE platform, all parties involved follow the set rules and collaborate in a single source of truth that shares all project data [24,77]. According to Taylor [78], the platform is capable of sharing big data of the whole project, and all parties involved can access the required data. Similarly, Ye et al. [79] asserted that payment data required can easily be acquired in the CDE and updated in the payment system.

Payment data can include lot of information related to the amount of works, time taken to complete the work, and quality of the work [80]. According to Ye and König [81], a CDE can bring together all site information, including geometrical and semantical data, building information, changes, obstructions, approvals, and claims. Additionally, payment information will include the measured work carried out with its amount [82]. The CDE is able to smoothly manage the processes of these information and documents in a timely manner, allowing communication and collaboration regarding the data from any location of the parties involved [21].

The CDE can assess the documents including variation orders in a collaborative and accurate manner and update data that is up to date and of quality [21]. According to Radl and Kaiser [83], the CDE improves control functions by enhancing quality, ensuring the timely update of reliable data, record the quantities of workers on site and the progress of works. Furthermore, according to Jaskula et al. [27] CDE ensures that less time is taken to manage requests for information and making decisions. Thus, the payment mechanism in the CDE conducts audits to identify payment disputes and certify payments in a short period of time [84].

1.4.2. Authorization of Payment Data and Conclusion of Certificate Amounts

Given that the sharing of data in the CDE is transparent amongst the relevant parties, the evaluation of works can be conducted. and payment can be authorized [81]. Once the evaluation of work is completed, the data is updated in the CDE platform and transactions for payments can be certified [24]. These data will also include any additional document, photos, and protocols observed and initiated during the evaluation of the works [81].

1.4.3. Submission of Invoice and Processing the Approved Payment

The approved payment is automatically sent to the transaction system in the CDE and the contractor uploads an invoice as a request for payment [85]. According to Ye and König [81], only the payment-related information is updated, and the transaction is made automatically from client to contractor as per schedule.

From the above literature, constructs and their indicator variables were established and are presented in Annexture 1. Furthermore, the following hypothesises were developed from the theory and concepts of enabling factors, characteristics of the CDE, and the CDE-based payment system.

H1. 

Training has a direct positive influence on an effective common data environment.

H2. 

Policy has a direct positive influence on an effective common data environment.

H3. 

Quality assurance team has a direct positive influence on effective common data environment.

H4. 

Quality technology has a direct positive influence on an effective common data environment.

H5. 

An effective common data environment has a direct positive influence on effective efficient payment systems.

2. Materials and Methods

The study set out to develop a common data environment centred model for payment system to improve the management of progress payments. This was achieved by conducting a systematic literature review, collecting quantitative data using questionnaire surveys, and analysing the data through structural equation modeling.

2.1. Systematic Literature Review

A systematic literature review was conducted on published materials from Scopus, Web of Science, and Google Scholar [86]. A literature review is the basis of research which reveals the existing knowledge, builds new ideas, and gives direction for a study [87]. A review was conducted to identify the current protocol of progress payments, the challenges faced during the process of progress payments, and the opportunities of integrating a CDE into the payment system. Given that implementing CDE technologies still faces challenges in the industry, concepts for enablers to drive effective implementation were evaluated. Furthermore, the benefits of integrating a CDE in a payment system were established.

The databases were used to retrieve journal articles, conference papers, and book chapters. Additionally, the exclusion and inclusion method was adopted to carefully select relevant papers. The keywords late payment, progress payment, and common data environment were used to search the databases. The results presented 304 documents from combined databases. The search was limited to documents written in English in the construction engineering and economics sector published between the years 2012 and 2024. The outcomes yielded 96 documents left for further analysis. Additionally, the abstracts, introductions, discussions, and conclusions of the qualifying articles were screened, and 80 articles were considered suitable and used for the study.

2.2. Questionnaire Survey

A closed-ended field questionnaire survey was adopted to collect quantitative data on the implementation of a CDE in construction projects. The questionnaire survey was used to collect data of the ratings regarding the relationships of the variables and the hypothesised factors in a 5-point Likert scale [88]. A field survey was considered the most adequate method to collect this kind of data as other methods could not provide an accurate picture of the different relationships in using a CDE in the construction industry [89]. Additionally, the ideology under which this research was based on was the contingency theory of testing relationships, which uses quantitative approaches and thus collection ofs data in this view is by questionnaire surveys [90]. Furthermore, the validation of the hypothesised model developed in this study was carried out by using structural equation modelling, which required quantitative data [91].

To ensure that the questionnaire was distributed to the right professionals, certain measures were put in place. The study adopted a purposive sampling technique because of its nature of collecting data from specific respondents based on the criteria that they used a CDE platform and they are involved in providing information needed or related to progress payments [92]. This measure includes identifying construction projects that used a CDE to manage project information and procedures. The researcher firstly contacted the construction companies which were using a CDE in their projects to identify the specific projects on which the platform was employed. The number of projects derived from the method was 31. Secondly, the researcher contacted professionals in their network who used a CDE at their workplace to identify projects on which the platform was used. And the number of projects identified with this method was 27. The identified projects within South Africa were 58, which had an average of 5 participants on the CDE platform.

Upon identifying the projects, the entities involved in the projects were recognized, along with their respective participants. In some cases, the principal agent and/or the project manager/director preferred to share the questionnaire with the participants so that they respond in an appropriate time that will not interfere with their working hours because of their busy schedule.

The study adopted a rule of thumb to calculate the sample size using the 10 times rule [93]. Given that the highest number of indicators for the constructs was 19, a 20:1 ratio was adopted. Therefore, the minimum sample size needed for analysis was 200. Additionally, a minimum sample size of 200 is required for SEM to achieve stable parameter estimates and good model fit [94].

The questionnaire was first developed on a word document and transferred to a Google form to easily share it online. These platforms were used because it was convenient and easy to contact and follow up with the respondents, and it saved more time compared to face-to-face administration. All the questionnaires were filled by the professionals in their own time as it was convenient for them. Considering that questionnaire surveys were distributed to all identified projects and there were no more respondents coming in, the data collection was concluded since the minimum sample of 200 required was reached [95]. A sample of 201 valid questionnaires received from the respondents was used for the analysis.

Table 1. Background of participants.

Group	Item	Percentage
Type of organisations	Client	14.4
	Consulting firm	57.7
	Construction entity	27.9
Roles of participants	Project manager-Client	7.5
	Project manager-Consultant	10.9
	Principal agent	1.5
	Designer-Architect	5.5
	Designer-Civil/Structural	14.4
	Designer-MEP	2.5
	Site engineer-Civil/Structural	8.0
	Site engineer-MEP	2.5
	Quantity surveyor-Consultant	14.4
	Quantity surveyor-Contractor	15.9
	Contracts manager/Director-Contractor	1.5
	Construction manager	4.0
	Estimator-Contractor	1.0
	Site manager	2.5
	Health and safety agent	0.5
	Health and safety officer	2.0
	Others-Project administrators, Technologist, Surveyor	5.5
Type of clients	Private Property Developer	33.3
	Parastatal Organisation	10.4
	Public/Government	36.8
	Mining Organisation	19.4

below presents the background information of respondents.

The type of organisations the respondents were working for included the client, consulting firms, and construction entities. Most of the respondents were working for consulting firms (116), followed by construction entities (56), and lastly the client (29). Most of the respondents who participated in the survey belonged in the groups: Quantity surveyor-Contractor (15.92%), Quantity surveyorConsultant (14.43%), Designer-Civil/Structural (14.43%), and Project manager-Consultant (10.95%). The least participation was found amongst the groups: Health and safety agent (0.5%), Estimator-Contractor (1%), Principal agent (1.49%), Contracts manager/Director-Contractor (1.49%), and Health and safety officer (1.99%). Additionally, the groups involved in most projects in this study were from Public/Government (74) and Private Property Developer (67), and those involved in the least were from Mining Organisation (39) and Parastatal Organisation (21).

2.3. Structural Equation Modeling

The data collected with the questionnaire was used to test the model fit and hypothesis using structural equation modeling (SEM). SEM is widely recognised as the most comprehensive statistical process in social and scientific research, for all operations of general linear modelling operations such as analysis of variances, multivariate analysis of variance, and multiple regression [96]. According to Kamaruddin and Abeysekera [97], SEM evaluates the significance and strength of a relationship within a complete posited model by estimating all the model’s coefficients. This was carried out using Analysis of Moment Structures (AMOS) and Advance Analysis of Composite (ADANCO).

Data screening was necessary prior to analysing the fit for the hypothesised model. Therefore, a pre-analysis was first conducted with the aim to check for missing data, characteristics of the data distribution, outliers, and the model identification. Furthermore, Raykov et al. [98]) asserted that outliers and missing values are common in ageing research, and when they are present in the raw data, they can have a negative impact on SEM results. Moreover, the identification of the model and multivariate distribution was crucial in determining if the data is normal and require the maximum likelihood technique [99].

Lei and Wu [100] asserted that identifying a model is complex and entails determining whether it is over-identified (with fewer parameters to estimate than the number of variances and covariances), just-identified (with the same number of parameters as the number of variances and covariances), or under-identified (having fewer variances and covariances than the number of parameters). Furthermore, in order to identify a model, the degrees of freedom (DF) must be greater than zero, meaning that there must be at least as many observations as parameters to estimate [95].

3. Results

The model’s main aim was to establish how a CDE can be implemented effectively to ensure an efficient payment system.

Upon receiving the field data from the questionnaires on the Google form, the data were exported to an Excel spreadsheet. The data were examined to identify missing values, and there were no missing values identified. These data were used to analyse the model fit and test hypotheses. A number of 201 samples was used to analyze the data.

3.1. Distribution Characteristics of the Data

When employing Structural Equation Modelling (SEM) analysis, it is important to ensure that the observed data are normally distributed. To achieve this criterion, both the univariate and multivariate normality were assessed. To assess univariate normality, the skewness and kurtosis of the observed variables were tested. According to Hair et al. [101], an observed variable with a skewness between −2 and 2 with its kurtosis between −7 and 7 is generally acceptable. Furthermore, a Mardia multivariate kurtosis of p(p + 2) is acceptable, where p is the total number of observed variables [101,102]. The number of observed variables were 68, thus the threshold for the Mardia multivariate kurtosis is 68(68 + 2) = 4760.

From

Table 2. Univariate and Mardia’s normalised multivariate estimates.

Variable	Skewness	Kurtosis
TRN1	−0.349	−0.601
TRN2	−0.618	−0.103
TRN3	−0.828	0.804
PLC1	−0.578	−0.239
PLC2	−0.625	0.153
PLC3	−0.531	−0.387
QAT1	−0.730	0.354
QAT2	−0.615	0.153
QAT3	−1.020	1.355
QAT4	−0.743	0.285
QTC1	−0.647	0.458
QTC2	−0.828	0.703
QTC3	−0.731	0.320
CDE1	−1.115	1.081
CDE2	−1.139	1.180
CDE3	−0.695	0.204
CDE4	−0.569	0.131
CDE5	−0.130	−0.749
CDE6	−0.345	−0.436
CDE7	−0.586	−0.099
CDE8	−0.438	−0.614
CDE9	−0.755	−0.242
CDE10	−0.561	−0.354
EPS1	−0.257	−0.398
EPS2	−0.127	−0.534
EPS3	−0.373	−0.196
EPS4	−0.821	0.245
EPS5	−0.645	−0.197
EPS6	−0.292	−0.378
EPS7	−0.434	−0.374
EPS8	−0.686	0.283
EPS9	−0.364	−0.648
EPS10	−0.526	−0.036
EPS11	−0.434	−0.005
EPS12	−0.435	−0.266
EPS13	−0.457	−0.160
Multivariate		286 199

, the univariate skewness ranges between −1.139 and −0.130, and kurtosis ranges between −0.749 and 1.355. The univariate skewness and kurtosis fall under the acceptable threshold for normality. Furthermore, the Mardia multivariate kurtosis was 286 199, which is above the threshold of 4760. Thus, while the univariate coefficients were acceptably normal, the multivariate coefficient was non-normal. Therefore, the data distribution is non-normal and robust maximum likelihood was adopted.

The robust maximum likelihood (RML) estimation method was chosen due to the non-normality of the data, which provides multiple robust fit indices [103]. A RML estimation approach produces a robust chi-square statistic (Satorra-Bentler scaled statistic) and robust standard errors that account for non-normality [104]. AMOS structural equation modeling software was chosen for its user-friendliness and ability to produce standard errors for non-normal data with the robust maximum likelihood method [105]. The confirmatory factor analysis (CFA) method was firstly used to analyse the measurement model and thereafter the complete structural model. Moreover, the CFA was used to confirm the significance of the covariances, fit indices, reliability, and validity of the sample data on the hypothesised models.

3.2. Identifiability of the Model

The identifiability of the model is carried out through SEM analysis as it is a requirement. Furthermore, for the model to be analysed, it must meet the conditions of model identification. Additionally, model identification needs to be checked before the model can be analysed. Researchers are strongly advised to ascertain whether a conceptualised model is identified [106]. Identification is concerned with whether it is possible to estimate each parameter’s value in a model in a unique manner [107].

The model can be under-identified with degree of freedom less than zero, just identified with a zero degree of freedom or over-identified with a positive degree of freedom [108]. The results from AMOS showed that the value of the degree of freedom ranges between 97 and 595. The values designate a positive value for the degree of freedom; therefore, the model is over-identified. An over-identified model gives freedom for the model to be rejected when there is disagreement in the data [109].

3.3. Fit Statistics on Measurement Models

The number of cases that were analysed for the enablers (training, policy, quality assurance team, and quality technology) and the constructs’ effective common data environment and effective efficient payment system was 201. The latent variable training is made of three groups with nine indicator variables, policy is made of three groups with eight indicator variables, quality team assurance has four groups from nineteen indicator variables, and quality technology has three groups from nine indicator variables. The construct’s effective common data environment has ten indicator variables, and the effective efficient payment system has thirteen indicator variables.

Annexture 1 presents the explanation of the indicator variables. The model postulates that the latent variables may be explained by the following six factors: training, policy, quality assurance team, quality technology, effective common data environment, and effective efficient payment system.

• Training is explained by grouped indicator variables TRN1 to 3;
• Policy is explained by indicator variables PLC1 to 3;
• Quality assurance team is explained by indicator variables QAT1 to 4;
• Quality technology is explained by indicator variables QTC1 to 3;
• Effective common data environment is explained by indicator variables CDE1 to 10;
• Effective efficient payment system is explained by indicator variables EPS1 to 13.

The descriptions for these indicator variables are presented in

Table 3. Correlations of the common data environment-centred model.

	TRN	PLC	QAT	QTC	ECDE	EEPS
Training (TRN)	1
Policy (PLC)	0.791	1
Quality Assurance Team (QAT)	0.746	0.852	1
Quality Technology (QTC)	0.690	0.734	0.797	1
Effective Common Data Environment (ECD)	0.536	0.483	0.543	0.597	1
Efficient Effective Payment System (EPS)	0.314	0.432	0.575	0.687	0.490	1

. The goodness-of-fit statistics and statistical significance of parameter estimations at a 5% probability level were analysed to determine how well the model fits the sample data and the strength of the hypothesised relationships between variables. Furthermore, the reliability internal consistency score was tested by examining the composite reliability and Dijkstra–Henseler’s rho coefficients. The construct validity was tested by determining the parameters’ coefficients, the convergent validity, and discriminant validity. The discriminant validity was tested by determining the HTMT2 correlation coefficients. Additionally, the correlation of the constructs is presented in Table 3.

The model was run in AMOS and yielded outputs of the indexes’ results. Upon examining the results of the indexes, they did not match the cut-off values. According to Teo et al. [108], a model can be modified by firstly examining the parameter coefficients and secondly by applying the modification indices. According to Hair et al. [110], parameter estimates of a value of 0.5 and higher are considered practically significant. All the parameter estimates values were more than 0.5, then modification indices were applied at 10 thresholds, and the model still did not fit.

The third modification method was applied, which examines the values of residuals and those that are far from zero are removed. Schermelleh-Enge et al. [111] asserted that it is important to look for patterns in the residual matrix that indicates an ill fit; the values of the standardised residuals should be near zero. In the results’ output, the values of QAT4 were far from zero, with some patterns being far at 3458. Thus, QAT4 was removed. Additionally, from the matrix, the values of EPS4 were far from zero, with some patterns far at −1821. Thus, EPS4 was removed, and the model reached the required fit indices. Figure 1 presents the modified measurement model. Following measurement theory and confirmation factor analysis practices, it is significant to remove these items to improve construct validity, unidimentionality, and model fit, which gives an acceptable structural model [112,113].

3.3.1. Model Goodness-of-Fit Statistics—RML

The study evaluated the robust statistic indexes for the Comparative Fit Index (CFI), Incremental Fit Index (IFI), and absolute fit index of the root mean square error of approximation (RMSEA) to determine the fit of the measurement model. Moreover, to compliment the conclusion on model fit, the Satorra-Bentler scaled chi-square (S—BΧ²) and the standard root mean squared residual (SRMR) were examined.

The model was run in AMOS and yielded outputs of the indexes results. The results yielded a chi-square of 953.976 with degrees of freedom of 498 and a p-value of 0.000. This chi-square result showed that there was a significant difference between the sample data and the postulated model. The normed chi-square divided by the degree of freedom was 1.916, which is significant.

Additional fit test results are presented in

Table 4. Robust fit indexes of the overall structural model.

	Cut-Off Value	Estimate	Comment
S—Bx2		953.976
Df	0 ≥ Acceptable	498	Acceptable
CFI	0.9 ≥ Acceptable 0.95 ≥ Good fit	0.919	Acceptable
IFI	0.9 ≥ Acceptable 0.95 ≥ Good fit	0.920	Acceptable
SRMR	0.08 ≤ Acceptable 0.05 ≤ Good fit	0.061	Acceptable
RMSEA	0.08 ≤ Acceptable 0.05 ≤ Good fit	0.068	Acceptable

below. The CFI found was acceptable at 0.919, and the IFI was at 0.920, which was above the cut-off of 0.900. The SRMR was found acceptable at 0.061, which was below the cut-off criterion of 0.08. Furthermore, the RMSEA was found acceptable at 0.068, which was below the cut-off criterion of 0.08. The overall results of this postulated model have reached a good fit.

3.3.2. Statistical Significance of Parameter Estimates

The model’s effectiveness was examined by assessing the parameter estimates for magnitude, signs, and statistical significance. Raykov and Widaman [114] suggested the additional analysis of standard errors, test statistics, and parameter estimates, in addition to fit statistics to ascertain whether the postulate model is adequate.

Table 5. Parameter estimates, Z-statistic, Reliability, and Construct validity.

Latent Variable	Number of Indicators/Groups	Indicator Variable (Grouped)	Parameter Estimates	Z-Statistics	Composite Reliability	Dijkstra–Henseler’s Rho	AVE
Training	3	TRN1	0.909	45,450	0.937	0.944	0.641
		TRN2	0.940	47,000
		TRN3	0.886	44,300
Policy	3	PLC1	0.838	27,933	0.905	0.942	0.650
		PLC2	0.904	45,200
		PLC3	0.874	29,133
Quality Assurance Team	3	QAT1	0.870	43,500	0.920	0.960	0.551
		QAT2	0.948	67,714
		QAT3	0.851	29,344
Quality Technology	3	QTC1	0.890	37,083	0.887	0.938	0.610
		QTC2	0.870	29,000
		QTC3	0.780	20,000
Effective Common Data Environment	10	CDE1	0.683	14,532	0.902	0.909	0.488
		CDE2	0.730	15,869
		CDE3	0.682	13,918
		CDE4	0.666	12,109
		CDE5	0.725	18,125
		CDE6	0.706	15,012
		CDE7	0.630	10,327
		CDE8	0.740	17,209
		CDE9	0.698	15,174
		CDE10	0.648	11,172
Effective efficient payment system	12	EPS1	0.782	20,237	0.941	0.949	0.549
		EPS2	0.711	14,580
		EPS3	0.773	16,041
		EPS5	0.690	14,020
		EPS6	0.734	15,122
		EPS7	0.759	23,941
		EPS8	0.799	22,243
		EPS9	0.765	18,341
		EPS10	0.747	14,800
		EPS11	0.770	17,116
		EPS12	0.774	15,872
		EPS13	0.755	1400

below presents the estimates and the significance of each variable.

Covariance or correlation coefficients that are not positive definite, with correlations greater than 1.00, and negative variances are described as unreasonable estimates [115]. Furthermore, Byrne [115] emphasised that Z-statistics is critical for determining if the estimate is statistically different from zero at 0.05 level, and a test statistic greater than 1.96 indicates null hypothesis rejection. This study uses a Z-statistic, which is a parameter estimate divided by its standard error, to test if the estimate is statistically different from zero. The output results shows that all correlation values are not greater than 1 and have positive signs. Furthermore, the test statistics results yielded Z-statistic values greater than 1.96. Therefore, the estimates are reasonable and statistically significant at 5% level.

The correlation values between the six latent variables demonstrates their interrelationship, with values ranging from 0.314 (training and effective efficient payment system) to 0.852 (policy and quality assurance team). These results presented in Table 3 are significant.

3.3.3. Construct Validity of the Measurement Model

The construct validity and internal reliability of the proposed model determine the validity of the measurement models’ constructs based on parameter estimates. A strong relation between the latent and indicator variables was vindicated by the fact that all the parameter estimates in Table 5 were greater than 0.5.

Furthermore, the composite reliability was significant in all latent variables with the lowest at 0.887. Additionally, the Dijkstra–Henseler’s rho was satisfied in latent variables with lowest at 0.909. Moreover, all the AVEs were significant above the 0.5 threshold, expect for the effective common data environment. The AVE of this latent variable was 0.488. Omar and Abdul-Karim [116] suggested that because the difference is too little at 0.012 from the threshold of 0.5, the AVE of 0.488 can be ignored; therefore, the items do not necessarily need to be deleted. Moreover, ref. [117] emphasised that the construct’s convergent validity is adequate if the AVE is less than 0.5 but composite reliability is greater than 0.6. The authors developed the AVE concept and note that an AVE of 0.50 or higher indicates adequate convergent validity; however, they also observed that when the AVE is less than 0.50 and composite reliability is sufficiently high, the construct may still have acceptable convergent validity. In this model, the results indicated a composite validity of 0.902 for this construct, which is higher than 0.6 and closer to 1. Therefore, this construct’s validity is significantly acceptable.

Additionally, HTMT2 has satisfied the discriminant validity with correlation values at 0.9 and below [118]. The HTMT2 values are presented in

Table 6. Discriminant validity of structural equation model HTMT2.

Construct	Training	Policy	Assurance	Technology	Common	Payment
Training
Policy	0.847
Assurance	0.779	0.900
Technology	0.730	0.786	0.844
Common	0.570	0.516	0.575	0.648
Payment	0.311	0.444	0.587	0.719	0.513

, with values between 0.311 and 0.900. Therefore, the construct validity of the measurement model has been satisfied. Moreover, the internal consistency criterion was satisfied since the composite reliability and Dijkstra–Henseler’s were both determined to be significant, with values close to 1.00.

The model fit indices were all acceptable; the CFI and IFI are above the threshold of 0.9. Additionally, SRMR and RAMSEA are below 0.08. The division of chi-square and degree of freedom is less than the threshold of 3. Furthermore, the construct validity and reliability of the model are significant. Therefore, the model fits the sample data which allows for the evaluation of the statistical significance of the hypothesised parameters.

3.4. Model Hypotheses Testing

The hypotheses were rejected based on the relevance of parameter estimations, including magnitude, sign and statistical significance. The estimates should have correlation values less than 1, with positive correlations and variances, to be considered significant. The estimates should be statistically different from zero, with a test statistic greater than 1.96 at a 5% level, which indicates rejection of the null hypothesis [115]. This study’s test statistic, which was the parameter estimates divided by its standard error, served as a Z-statistic to determine whether the estimate deviated statistically from zero. The test statistics were used to examine the hypotheses H1 to H5.

3.4.1. Testing the Direct Influence of Enabling Factors on Effective Common Data Environment

The general hypotheses are that the enabling factors (training, policy, quality assurance team, and quality technology) have a positive influence on effective common data environment. The results of the parameters are presented in

Table 7. Parameter estimates and test statistics of the complete model.

Label	Hypothesis Enabling Factors Have Direct Positive Impact on Effective Common Data Environment	Parameter Estimates	Z-Statistics	p-Value	Significant at 5% Level?
H1	Training	0.393	3.083	0.001	Yes
H2	Policy	0.373	2.082	0 025	Yes
H3	Quality Assurance Team	0.463	2.544	0 010	Yes
H4	Quality Technology	0.472	3.799	0.0001	Yes
	Effective common data environment has a direct positive impact on efficient effective payment system
H5	ECD	0.589	12.117	0.000	Yes

below, and Figure 2 presents the relationships of the constructs being tested. The results showed that there is a positive influence from training, policy, quality assurance team and quality technology on effective common data environment. This means that when training increase by a unit, effective common data environment increases by 0.393. Secondly, when policy increase by 1 unit, effective common data environment increases by 0.373. Thirdly, when quality assurance team increase by a unit, effective common data environment increases by 0.463. Lastly, when quality technology increase by 1 unit, effective common data environment increases by 0.472. Therefore, all the null hypotheses were rejected.

3.4.2. Testing the Direct Influence of Effective Common Data Environment on Effective Efficient Payment System

The general hypothesis is that effective common data environment has a positive influence on effective efficient payment system. The results of the parameters are presented in Table 7 above. These results revealed that there is a positive influence of effective common data environment on the effective efficient payment system. This means that when the effective common data environment increases by 1 unit, the effective common data environment increases by 0.589. Therefore, the null hypothesis was rejected.

4. Discussion

The study aimed to validate the measurement model and explore the structural relationships affecting the implementation of a Common Data Environment (CDE) and its impact on creating an efficient payment system in South African construction projects. The analysis confirmed the reliability and validity of all latent constructs, indicating meaningful causal pathways aligned with contingency theory, with high composite reliability among the variables.

Contingency theory asserts that no single organisational structure or strategy is deemed universally effective; instead, a system’s effectiveness varies based on its operating conditions. This study illustrates that the implementation of a Common Data Environment (CDE) and enhancements in progress payment systems were influenced by various organisational, technological, and contextual factors.

4.1. The Influence of Training on Effective Common Data Environment (H1)

The hypothesis was that training has a positive influence on effective common data environment which could not be rejected. The findings showed that 64.1% of the indicator variables explained training. The indicator variables that explained training included TRN1 to 3.

The findings were that training has a positive influence on the effective common data environment. These results are in line with the conclusions of Vitente et al. [41], which emphasised that training should be in place to stimulate the adoption and implementation of innovative models. Likewise, the study of Fathima et al. [42] recommended that training should be initiated in order to succeed in implementing building information modelling. These findings may provide the basis of understanding the case study results of Migilinskas et al. [32], which stated that participants were not willing to implement the BIM platform because they did not know how to use it and lack understanding of the model. Thus, training is crucial in facilitating the adoption and implementation of CDE platform.

Contingency theory posits that systems can only function optimally when they align with the participants’ skills, capabilities, and readiness. The study indicates that when participants lack understanding of a digital platform, implementation becomes limited, despite technology availability. Consequently, training is identified as a vital contingency factor that facilitates system optimisation and enhances participant acceptance.

4.2. The Influence of Policy on Effective Common Data Environment (H2)

The hypothesis was that policy has a positive influence on effective common data environment which was accepted. The results revealed that 65% of the indicator variables defined the latent variable policy. These indicator variables included PLC1 to 3.

The findings indicated that policy has a positive influence on an effective common data environment. This result concurs with the study of Qin et al. [45], which stipulated that policy is a key factor which acts as an incentive tool to stimulate the perceived use and adoption of building information modeling platforms. This makes sense of the conclusion of Migilinskas et al. [32] where, in their case study, though the company had policy in place, some participants did not implement building information modeling platform due to the fact that they were not aware of the policy standards and rules. This might be because the participants did not want to do things in the wrong way since they were unsure. Therefore, policy employment has a positive influence in implementing a successful CDE.

Contingency theory posits that processes should align with guidance and strategic intent. The findings support this by showing that policy creates an environment that fosters compliance, decreases uncertainty, and stimulates the constant use of the digital system.

4.3. The Influence of Quality Assurance Team on Effective Common Data Environment (H3)

The set hypothesis was that quality assurance team has a positive influence on effective common data environment which was accepted. Some of the indicator variables could not be confirmed as they were unable to define quality assurance team. These indicator variables were removed as they were weakening the confirmation factor analysis process. These variables included, ensured that the progress payments information is transferred and updated, ensured that progress payments were processed on time, maintained a progress payment process that is free of conflicts and made sure transparency in the CDE is sustained. A closer evaluation of these variables’ reasoning might point to the misalignment of the CDE’s current requirements and weak capability in digital skills within South African construction projects.

The remaining indicator variables explained 55.1% of the latent variable quality assurance team. The indicator variables included that a quality assurance team will ensure QAT1 to 3.

The findings revealed that quality assurance team has a positive influence on effective common data environment. The findings are in line with the study of Xu et al. [47], which stipulated that to facilitate acceptance and use of a data exchange platform, the company owner, top management, and technical staff should provide immediate support. Furthermore, the results confirm the conclusions of Nguyen et al. [25], which stated that managers need to ensure that the model is working properly to its advantages. In addition, these findings might make sense of the study of Migilinskas et al. [32], which explained that the main obstacle preventing project participants from utilising the newest technologies is the absence of support from top level managers. Thus, the employment of a quality assurance team is important in the successful implementation of a CDE.

Contingency theory posits that management structures should align with organisational technology and complexity. The findings indicate that lack of support, monitoring, and technical guidance from dedicated management teams leads to increased resistance and inconsistent in system implementation.

4.4. The Influence of Quality Technology on Effective Common Data Environment (H4)

The hypothesis was that quality technology has a positive influence on the effective common data environment which could not be rejected. The results indicated that 61% of indicator variables explained the latent variable quality technology. The indicator variables demonstrated that quality technology is explained by QTC1 to 3.

The findings disclosed that quality technology has a positive influence on an effective common data environment. The findings are in agreement with the study of Wang et al. [51], which stated that it is crucial to ease technology complexity in order to ensure that it is clear to be used by participants. Moreover, the study of Nguyen et al. [25] explained that an increase in the use of these platforms can be facilitated by building a quality technology operated by skilled technicians. This might make sense of the case study of Sattineni and Macdonald [35], which stated that the model was not developed correctly and did not meet its intended capacity, thus failing to perform the tasks. Therefore, it is very important to ensure that the adopted CDE is built to its maturity to ensure its successful implementation.

Contingency theory emphasises the need for technology to match task demands and user capabilities. Findings indicate that a mature, reliable, and user-friendly platform enhances implementation and improves outcomes, while poorly developed or complex technology serves as a barrier, revealing significant contingency dependency.

4.5. The Influence of Effective Common Data Environment on Effective Efficient Payment System (H5)

From the above findings, it has been discovered that for a common data environment to be effective, training, policy, quality assurance team, and quality technology should be employed first. Thus, it was hypothesised that an effective common data environment has a positive influence on effective efficient payment system. The findings revealed that 48.8% of the indicator variables with a composite reliability of 90.2% explained the latent variable effective common data environment. This is significant because 48.8% was closer to the 50% cut-off and reliable above 60% [91]. Therefore, there was no need to remove some of the variables to reach 50%. The indicator variables included that an effective common data environment is explained by CDE1 to 10.

The results revealed that an effective common data environment has a positive influence on effective efficient payment system. This is in agreement with the study of Parn and Edwards [67], which concurred that digital data shared within a CDE can be leveraged to validate and sustain a competitive payment process, enabling transparent and trackable transactions. Furthermore, the findings are agree with the studies of Tao et al. [18] and Elghaish and Abrishami [69], which stated that CDE develops an automated payment system for progress payments, enabling tracking of transactions and improving communication among all parties, thereby promoting transparency and efficiency of the system. Moreover, the study of Lu et al. [22] presented that the CDE can significantly enhance management of progress payments by facilitating the generation of bill of quantities, supporting cash flow management, and enabling automated payments. Therefore, when the CDE is implemented properly it will eliminate challenges faced by construction projects of late payments on completed work.

The results indicate that the effectiveness of the payment system hinges on the effective functioning of the CDE, illustrating the following hierarchical contingency chain: Training + Policy + Management Support + Technology leads to an effective CDE, which results in an efficient payment system. This supports the contingency theory perspective that system performance arises from the alignment of various organisational factors rather than from the independent implementation of technology.

4.6. The Effective Efficient Payment System

From the above it can be observed that an effective common data environment led to an effective efficient payment system. The effective common data environment should be enabled by factors such as training, policy, quality assurance team, and quality technology. It was proven in several case studies that without these factors the CDE will not be effective.

The objective was to validate the relationship of the indicator variables with the latent variable effective efficient payment system. From the results, it was discovered that 54.9% of the indicator variables with composite reliability of 94.1% define the latent variable effective efficient payment system. The indicator variables confirmed their relationship with the latent variable expect for one variable that was weak and had to be removed. The indicator variables that explained the latent variable included EPS1 to 3 and EPS5 to 13.

This is to say that when a payment system is effective and efficient, the process of progress payments will be transparent, on time, and without conflict, as it should be. This concurs with the emphasis of Ahmadisheykhsarmast and Sonmez [66], which stated that timely progress payments are crucial for the smooth operation of a contractor’s cash flow and the success of the project. Additionally, the results affirm the study of Sigalov et al. [24], which concurred that transparency in information sharing is fundamental to enhance progress payment, ensuring they are made on time and as per agreement. Therefore, it is crucial to integrate a productive CDE to establish a payment system that is effective and efficient to avoid late payments, conflicts, and corruption activities during progress payments.

The findings indicate that the benefits of an effective payment system emerge only when proper organisational contingencies are present. This supports contingency theory, demonstrating that the success of the system relies on the alignment of operational processes, technology, organizational capabilities, and governance structures.

5. Conclusions

The main aim of this study was to develop a common data environment model to improve the management of progress payment and most importantly to establish the enabling factors for an effective CDE. To achieve this aim, the study conducted a robust systematic literature review and a field questionnaire survey. The questionnaire survey aimed at validating the indicator variables and the conceptualised model from literature review.

A questionnaire survey was analysed using structural equation modelling to validate the indicators and test the relationship of the enabling factors on effective CDE and the CDE on effective efficient payment system. The findings revealed that the indicator variables defined the latent variables namely, enabling factors (training, policy, quality assurance team, and quality technology), effective common data environment and effective efficient payment system. The enabling factors, quality assurance and quality assurance team, emerged as the strong influencers.

The study concluded that a successful and productive payment system can be ensured by integrating an effective CDE. This effective CDE can be possible through the employment of enabling factors, namely, training, policy, quality assurance team, and quality technology.

5.1. Practical Implication

The knowledge of the influence of effective CDE on effective efficient payment system can help project owners (clients), consultants, and contractors to realise the importance of CDE in the management of construction information, communication, collaboration, and transparency. Moreover, the realisation that an effective CDE is positively influenced by enabling factors will help the client to plan and organise policy development and the type of CDE technology to adopt. Furthermore, this knowledge can help the client in the allocation of financial resources and personnel for the implementation of a CDE. Additionally, the knowledge can help consultants understand the part they play in making sure that the CDE system is implemented and bring awareness on the requirements and terms of use. Furthermore, the contractors understand that they can be part of the system and the importance of their roles, requirements, and terms of use.

5.2. Recommendations

5.2.1. Theoretical

It has been discovered from literature that progress payments are paid late because of improper information management, poor communication, lack of collaboration, and corruption challenges. Furthermore, it was identified that these challenges can be managed by CDE platform.

However, it has been established from literature that the implementation of CDE is facing challenges, limiting the platform to reach utilisation maturity in the construction industry.

The results revealed that CDE projects fail to reach maturity because of lack of training, the lack of policy deployment, absence of top management, and the fact that the adopted technology is not built right.

Therefore, there is considerable hindrances to the successful implementation of CDE. These include the primary challenges of the external and internal factors needed to be considered before implementing a CDE which was identified using contingency theory. These factors are identified as the enabling measures for a successful CDE implementation, which include training, policy, quality assurance team, and quality technology. These enabling factors enhance a successful CDE, which influence an effective and productive payment system that ease the management of progress payments.

It is recommended that the influence of enabling factors should be integrated in CDE implementation frameworks that have been already proposed in other studies but did not include the influence of enabling factors.

5.2.2. Methodological

It is suggested that a combination approach of questionnaire survey and SEM be encouraged in research studies where a test and retest method may not be appropriate to validate an analysis. This practice is normal in Architecture, Engineering, and Construction (AEC) research and most end in questionnaires and case studies. Therefore, it is recommended to adopt the method of questionnaire surveys and SEM to validate a study and improve its generalisation.

Studies in AEC research try to identify the effect relationship between latent variables. Most studies employ inadequate analytical methods like ANOVA and multiple regressions which are standard statistical approaches, that do not effectively test hypotheses at a higher level of abstraction. Therefore, for similar studies to this one, SEM is recommended for a more robust analysis.

Furthermore, it is recommended that similar studies should be conducted with larger sample size from a different population to keep improving the implementation in the construction industry.

5.2.3. Practical

It is recommended that each party (client, consultant, and contractor) identify which indicator variables serves them when ensuring an effective CDE and an efficient payment system.

It is further recommended that clients should understand consultant and contractor’s participation indicators to make a proper planning at the initial stage to include each one’s expected output in ensuring a successful implementation of the CDE. Additionally, knowing what factors to include in ensuring a working CDE to enhance the management of progress payments helps prepare realistic estimates for the project.

Moreover, it is recommended that all parties understand which indictors are effectual towards a successful CDE implementation and a productive payment system in order to plan, organise, and manage a CDE centred payment system effectively.

Preparing for enabling measures—providing training and developing policy for implementing CDE is a prerequisite to regulate practices and warrant utilisation. Furthermore, the employment of a quality assurance team and technology to ensure the workability and the mediation of management to participants for a successful collaboration and communication are recommended.

5.3. Limitations

The study is based on data from professionals within construction industry projects in South Africa, and a larger population would be preferable to conduct similar with sufficient resources. Additionally, the study analyses were limited to a sample size of 201, but a larger sample size would benefit the study considering that the model had three levels of measurement models. Furthermore, the model in this study only validated the indicators and the overall model conceptualised for this research and did not include other construction payment model initiated by other researchers.

5.4. Future Research

Future research on understanding the relationship, influence, and performance between training, policy, quality assurance team, and quality technology is encouraged.

It is recommended that an evaluation should be carried out on the direct influence of training, policy, quality assurance team, and quality technology on effective efficient payment system analysed with SEM would give more insight on the level of positive influence of each of the enablers.

The CDE platform requires the involvement of professionals as they will implement technology to manage construction project information. As a result, it is advised that further research be devoted on the ability and relations of professionals from construction and technology and the possibility of collaboration amongst them.

Furthermore, it is recommended that research on establishing strategies to employ enablers in CDE implementation should be conducted. This will help in raising financial resources.