Centralized or Decentralized? Communication Network and Collective Effectiveness of PBOs—A Task Urgency Perspective

Abstract

In the construction industry, there are a large number of project-based organizations (PBOs), where the efficiency of communication and collaboration among organizational members greatly impacts the success of projects. For PBOs employing both centralized and decentralized communication networks, it is worth delving into the question of under what circumstances which type of network will yield better results. Based on the IMO model and organizational learning theory, this paper conducts a grouped communication experiment involving 598 engineering management personnel to explore the differences in collective effectiveness of varying communication networks from the perspective of task urgency. Beyond task performance assessments, we have included organizational member perception to form evaluation criteria for collective effectiveness. Our research results show that under conditions of weak task urgency, decentralized networks yield higher collective effectiveness. Conversely, under conditions of strong task urgency, centralized networks demonstrate superior collective effectiveness. Furthermore, this study also verifies the mediating role of knowledge sharing behavior when task urgency is strong. This research provides significant managerial insights for the establishment of appropriate communication networks for PBOs in the construction industry.

1. Introduction

In industries such as construction, high-tech manufacturing, management consulting, and professional services, a temporary legal project-based enterprise or firm is often established based on a specific output goal [1]. These are referred to as Project-Based Organizations (PBOs). Participants with different professional backgrounds and various resources are integrated through complex connections. The organizational structure is constructed based on knowledge sharing and organizational coordination to achieve the success for the project [2]. Existing research, based on the characteristics of project types, explores the formation and performance of various organizational structures for PBOs. For example, in specific situations like defense and national missions, organizational members tend to communicate or collaborate in a more centralized manner, forming a centralized structure [3,4]. In industries like software development and consulting, where high flexibility and innovation are encouraged, a more decentralized organizational cooperation structure is common [5,6]. In the construction industry, organizational structures are often built based on contractual relationships, influencing the communication relationships among organizational members.

To accurately measure the communication relationships among members of PBOs in the construction industry, Social Network Analysis (SNA) is widely used to capture and explore the structure and performance of these communication relationships, i.e., research based on communication networks. For example, indices, including those for centralization, connection strength, and network density, are used to measure the structural features of communication networks [7]. The interactions of stakeholders and collective decision-making can be predicted in a confined communication network [8,9]. Further, from the perspective of network centralization, communication networks can be divided into centralized networks and decentralized networks [10]. In decentralized networks, organizational members can directly communicate with any other member, adapting to faster unique information flows and improving performance in knowledge-based work [11,12]. In contrast, centralized networks include a core leadership or leadership team and a peripheral group capable of obtaining all critical information in a constantly changing task environment, thereby improving collective learning and problem-solving capabilities [13].

It is noteworthy that recent studies have discovered that the communication network of PBOs in the construction industry undergo adaptive evolution as the project progresses [14]. This implies that project progress, a key factor in project management, may cause the evolution and development of the communication network structure [15]. It is well-known that as a project nears completion, the urgency of tasks often increases. Previous research has confirmed that the urgency of tasks enhances the capability of knowledge sharing among members [16]. Therefore, during different stages of construction project development, what structure of communication network should we adopt to enhance management value? Moreover, what exactly signifies the management value of a PBO? Is it just project performance? Answers to these questions will help improve the management of complex projects in the VUCA era, but related studies are still somewhat lacking.

Hence, based on many observations and empirical studies of past projects, we propose hypotheses concerning the structural measurement of communication networks and their collective effectiveness for PBOs. Most PBOs we surveyed are in different spatiotemporal realms, and their collective effectiveness might be influenced by external environments. Laboratory behavioral experiments can be used to achieve the goals of this study and provide clear explanations for correlations and trends in an experimental scenario setting [17,18]. Thus, we conducted a grouped communication experiment involving 405 engineering project management personnel, comparing the differences in collective effectiveness under two levels of task urgency within two types of communication networks to provide theoretical guidance for establishing appropriate communication network structures throughout the entire lifecycle of a project. Additionally, we discuss scenarios where knowledge sharing plays a mediating role in this process. This study explores strategies for enhancing the effectiveness of PBOs from the perspective of communication networks, supplementing current research that often focuses on improving communication performance through technical means to provide new insights for complex project management.

The next section of this paper introduces the literature review and hypothesis development. Section 3 describes the experimental design, samples, procedures, and measures. Section 4 presents data analysis and test results. Section 5 mainly discusses the results, outlining theoretical contributions and managerial implications.

2. Literature Review and Hypotheses Development

2.1. PBOs and Communication Network Structure

PBOs are organizational forms engaged in temporary work to create innovative products or services. As temporary organizations, they are characterized by project duration, unique tasks, and teamwork to achieve project goals [1,19,20]. Increasingly, people are proposing methods based on SNA to address the relational issues among members of PBOs [21,22]. Quantitative analysis based on the topology of social network relationships can assess the relational ties and the overall network structure. That is, by studying the attributes of nodes and the structure of connections between them, one can explore the interdependencies among network members and how their positions in the network influence their constraints and behaviors. On this basis, individual responses and collective outputs can be analyzed and predicted [23,24,25].

Currently, researchers in the field of construction project management have conducted considerable theoretical and practical exploration into integrating social networks with project organization structure management. Each engineering project contains a social interaction and collaboration network, where the flow of knowledge within the network is constrained by its topological structure. The connections and the overall structure can be quantitatively assessed by SNA. Further, the complex relationships among project members at different stages of project can be displayed using social graphs [26,27] to, for example, indicate the dynamic power of stakeholders in the implementation of social responsibility issues in construction projects [28]. Moreover, in recent years, discussions on the organizational structure of large-scale PBOs such as public–private partnership projects and major engineering projects have attracted more attention. For example, by identifying the network status and relationships of project members, there are discussions on the dynamics of stakeholders in implementing social responsibility issues [29], exploring the characteristics of relationship exchange behaviors among stakeholders in mega-projects from the perspective of stakeholder value networks [30] and organizational collaboration relationships [31].

Drawing upon the existing research on communication networks of engineering project teams featured in mainstream peer-reviewed papers and case studies [32,33,34], this study summarizes four types of communication networks. These networks include the fully connected network, the subgroup network, the core-periphery network, and the locally clustered network. Referring to existing classification methods [10], the first two are classified as decentralized networks, while the latter two are considered centralized networks.

2.2. Organizational Effectiveness in Different Communication Networks

Currently, there is a consensus that different types of communication networks exhibit performance variations in information transferr. A classic communication experiment manipulated the communication patterns among group members by controlling who could send information to whom and measured the impact of various communication patterns on group operation and performance. It was found that the degree of centralization—i.e., the extent to which one person acts as a communication hub—has a significant impact on both individual and group performance. The complexity of the task has also been proven to be a key moderating variable: centralization is beneficial for simple tasks, but harmful for complex tasks, according to Bavelas and Barrett [35]. A decentralized structure is the best choice when information is unevenly distributed among group members or when information is unclear [36,37]. Furthermore, within a defined communication network, the ability to transfer information between network neighbors affects the quality of collective decision-making [17], which may also be due to differences in knowledge transfer capabilities [38]. Relevant research has outlined common interest in solving communication problems in different network structures.

So, how should communication network structures be adopted in PBOs? The answer depends on the unique goals of each PBO with different customary standards. Recent research suggests that in addition to basic factors such as the cost, time, and quality of project, other factors like organizational characteristics and stakeholder interests should also be carefully considered [39,40]. But the current conclusions are not yet unanimous, as they are extremely difficult to measure.

Based on the Input-Mediator-Output (IMO) model, this study—proposes that the following indicators should be used to measure the organizational effectiveness of PBOs. Specifically speaking, problem-solving performance indicators representing organizational usefulness and member perception should also be considered. In the IMO model, interaction processes at the organizational level, environmental level, and individual level impact organizational performance and other outcomes (such as member satisfaction and group cohesion) [41]. Taking PBOs in the construction industry as an example, although task performance has been used for a long time as the main criterion for measuring project management success, it has been gradually realized that team members’ positive perceptions in team cooperation can affect both the completion of the current project and future project cooperation. Therefore, we believe that member perceptions including job satisfaction and participation, tacit understanding, organizational commitment, and willingness to continue cooperation should be added to constitute the category of collective effectiveness.

2.3. Urgency of the Project Process

The establishment of PBOs is aimed at harnessing creativity, collaboration, and coordination to complete a new task within an urgent timeframe. Due to the temporary nature of PBOs, members might not have prior experience working together. To succeed, they must rapidly establish a common understanding of their task and develop plans throughout the project. Project members need to utilize their time to understand the task, identify necessary steps for its accomplishment, and devise mechanisms for coordinating multiple interdependent activities [42]. For a new project, which often includes different stages like initiation, execution, and final deadlines, organizational members might suffer from uncertainty about when certain events or tasks should occur, incompatible priorities, or different pacing styles, leading to wasted efforts or inertia in the first half of the time [43]. As deadlines approach and time pressures increases, the collective tends to reorganize and adopt new methods to complete tasks [44]. Therefore, in PBOs aimed at developing complex products, organizational performance management must focus on time-based control mechanisms, including overall deadlines and the synchronicity of activities within the project [4,45]. Detailed conclusions about task urgency can also be found in existing research; for example, in high-tech-oriented project organizations, high task urgency can negatively impact inter-project communication and knowledge transfer intentions [46,47].

In the construction industry, completing projects on schedule has always been one of the standards of project performance management [48]. Therefore, research on project urgency and its impact on organizational performance has never ceased. For instance, studies on leadership discuss how project deadlines affect the objective pressure on project managers [49,50]. In recent years, debates on the impact of urgency on information sharing and knowledge transfer have not stopped. Most studies believe that under limited deadlines and tight schedules, project teams do not have enough time for communication and knowledge sharing [51] and that under centralized structures, the decision-making process becomes more concentrated, reducing the time for communication and coordination. Conversely, some studies suggest that when the task urgency is not strong, project team members are willing to seek knowledge from other project teams when needed [52], thereby enhancing the collective decision-making ability.

Thus, we can affirm that the degree of urgency in project progress is crucial for the study of collective effectiveness in PBOs. To further clarify the nature of the impact, we propose the following hypotheses:

H1a. 

Under condition of weak task urgency, decentralized communication networks will yield higher collective efficacy.

H1b. 

Under condition of strong task urgency, centralized communication networks will yield higher collective efficacy.

2.4. The Mediating Role of Knowledge Sharing

For the successful achievement of organizational goals, members of an organization endeavor to share and disseminate their unique information and knowledge to others, externalizing internal knowledge into collective knowledge that is owned, shared, and agreed upon by PBOs [53,54]. The aforementioned processes of knowledge acquisition, sharing, transfer, and integration largely depend on the communication structure. In recent years, social networks have been seen as channels for organizational learning, playing a role in information sharing and knowledge transfer [55,56,57]. For example, in a global PBO, we found that each member relies heavily on networks for acquiring knowledge, communication, and learning from others [58]. Thus, social networks can reveal patterns of knowledge integration [59]. In centralized networks, close ties and social cohesion among organization members who are geographically closer are more likely to facilitate knowledge transfer and learning among members [60]. On the other hand, decentralized networks, due to their heterogeneity of knowledge, can effectively enhance the effectiveness of shared knowledge [61,62].

Furthermore, organizational learning theory and group polarization theory can provide a theoretical explanation for communication networks having a key impact on the quality of collective task completion through the degree of knowledge sharing [63,64]. On the one hand, based on organizational learning theory, PBOs can benefit from integrating different types of knowledge, as people tend to use shared information as the basis for collective task completion, and a lack of non-shared information often leads to detrimental task completion outcomes [65,66]. On the other hand, according to group polarization theory, in interpersonal communication network structures where knowledge sharing is less than ideal, biased group polarization is more likely to occur, leading to lower satisfaction with collective task completion [67].

The aforementioned literature primarily demonstrates that communication network structures influence collective decision-making through their potential impact on organizational knowledge sharing and transfer. In urgent situations, quickly and effectively solving problems often requires stimulating collaboration and knowledge sharing among team members, as each individual may hold key information or expertise necessary for solving the problem. Faced with urgent tasks, team members may need to think and innovate quickly. Under such pressure, people are often more willing to share and explore new ideas, thereby promoting the flow of knowledge and the generation of new solutions, which also helps reduce knowledge hoarding behavior. Therefore, regarding the conditions and mechanisms of knowledge sharing behavior in PBOs, we propose the following hypothesis:

H2. 

Under condition of strong task urgency, collective communication will stimulate knowledge sharing behavior, thereby enhancing collective effectiveness.

The theoretical model of this study is shown in Figure 1.

3. Methodology

3.1. Research Design and Experimental Treatment

Based on the Bavelas–Leavitt–Guetzkow series experiment [36,68,69] and a recent communication network experiment [13], this study presents an experimental platform to instrument the connection between communication networks and collective effectiveness for PBOs for two levels of urgency. The requirements of the platform were as follows: first, maximum verisimilitude, which means both that the presented networks had real-world analogues and that the means for accomplishing the task similar to real information delivery work in PBOs; second, maximum accessibility, which required the task to be easily understandable and implementable by subjects; third, maximum instrumentation, which required that results obtained by the task and feedback from participants be captured as richly as possible in subsequently analyzable data. Referring to the reality of PBOs in the construction industry, participants in the organization often have different information dimensions of previous cases due to their different professional backgrounds and experiences.

Therefore, this paper uses group-based experiments to explore the mechanism of how communication networks affect the collective effectiveness of PBOs. Specifically, we aim to investigate the interaction effects between communication networks (fully connected/subgroup/core-periphery/locally clustered) and two levels of task urgency on task performance and member perception (H1a and H1b) as well as the emergent scenarios of the mediating effect of knowledge sharing (H2). To complete a theater construction project, a PBO was formed consisting of owners, designers, builders, and supervisors. In this experiment, twelve participants were designated to play the roles of Owner A, Designer B, Builder C, and Supervisor D. Each role was represented by three members: the Owners (A1, A2, A3), Designers (B1, B2, B3), Builders (C1, C2, C3), and Supervisors (D1, D2, D3). Members within the same role operated independently of each other, with no formal relationship beyond their association as team members, as depicted in Figure 2.

Referring to the studies of Mason and Watts [70] and Enemark, et al. [71], we adopted a messaging task to measure the communication of PBOs who were in different network structures. Specifically, participants were faced with the following tasks: (a) each member initially had 33 unique messages for role category; (b) team members performed a messaging task with an optional communication partner during a designated communication time; and (c) 99 four-dimensional messages were integrated one by one to form a complete message. The overall task goal of the project was for each member to complete effective information transfer and integration within the time limit and for the individual with decision-making authority to aggregate valid four-dimensional information and make collective decisions accordingly. The main experiment was divided into two parts, each lasting ten minutes, with the results of the first part being carried over to the second part.

3.2. Participants and Procedure

Through expert interviews and recommendations, we adopted a snowball sampling method and recruited 598 individuals with over two years of experience in engineering project management from fifteen construction companies and engineering firms in Shanghai, China, to participate in this experiment. Those who successfully completed the experiment were each rewarded with a ¥100 supermarket coupon. To avoid the interference from learning effects and legacy effects due to the increasing familiarity of the experiment participants as well as the empirical summaries of the experimental steps or experimental order effects, this experiment adopted a between-group design method, thus each person could only participate in the experiment once. Before the main experiment, we played a video for all participants about the experiment rules and examples. Afterwards, participants were required to answer five questions regarding their willingness to participate and the rules of the experiment. Those who passed were then randomly assigned to groups for the main experiment. We used a random number method to group participants, striving to ensure a balanced number of participants and groups in each experimental setting. Next, participants entered their respective groups, with each sitting in front of a computer in a private booth, filling out personal information, and logging into the experimental platform. All activities were conducted through an online platform, with data being automatically collected and scanned by the system. The total duration of the experiment was 20 min. Ten minutes into the main experiment, the system paused to allow participants to fill out a questionnaire on ‘perception’ (including job satisfaction and communication experience). At the end of the main experiment, each participant filled out the questionnaire on perception again and provided feedback on the experiment. The feedback form was used to measure whether the subjects correctly executed the experimental procedures and rules, thus conducting an operational check of the experimental data [72]. Participants had to correctly answer all six questions. For instance, do you agree that passing information is to support decision-making activities? When information is shared, it is necessary to confirm whether the recipient had the information before. According to the rules, who can you communicate with?

The experiment was conducted from April to September 2023 in Building No. 2 of the Shanghai Lixin University of Accounting and Finance, yielding 49 sets of experimental data. After excluding one set of erroneous data, 48 sets of valid experimental data were obtained, ensuring 12 groups for each of the four networks. From a basic demographic analysis, there were 267 males (46.3%) and 309 females (53.7%), indicating a relatively balanced gender distribution. It was found that 368 persons (63.9%) had two to five years of work experience, 141 persons (24.5%) had five to ten years of work experience, and 67 persons (11.6%) had more than ten years of work experience. In addition, 62 persons (10.8%) held positions of department manager or above, 289 persons (50.2%) were project managers, and 225 persons (39.0%) were engineers and related technical personnel. Additionally, Harman’s single-factor test was used to check for potential common method biases. The highest variance of all member-reported variables was 24.328%, less than half of the total explained variance. This suggests that our analysis did not suffer from severe common method bias issues.

3.3. Measures

This study measured collective effectiveness in terms of task performance and member perception. Referring to existing methods [73], task performance in this experiment was defined as the total amount of complete information obtained by the group within a set time—i.e., the cumulative number of complete pieces of information obtained by decision-makers after excluding duplicate data. Additionally, referencing the measurement methods of subjective experiences in existing communication experiments [10], this experiment measured organizational member perception through a questionnaire with five items (Appendix A). Knowledge sharing is manifested as effective private information transmission between members through individual-based behaviors. Following the methods used in behavioral experiments to measure information sharing [18], we collected the total amount of information shared and acknowledged by others within a limited time as the data source for knowledge sharing. Data was collected through records of information transmission by participants during the experiment. Considering the gradually increasing urgency of engineering construction project tasks and the results of the pilot experiment, the 10th minute of the formal experimental process was used as the critical point for dividing urgency.

To eliminate the influence of confounding factors on the experimental results, this study considered the following covariates: the participants’ professional backgrounds, work experience, industry characteristics, knowledge related to organizational structure, communication experience, and previous collaboration experience. We categorized the samples into two groups based on the average scores of these six items at the collective level. We compared intergroup task performance, member perception, and knowledge sharing, with p-values of 0.456, 0.865 and 0.558, respectively, and found no significant differences (p > 0.05). Therefore, we disregarded these covariates in subsequent analyses.

4. Results

4.1. Manipulation Check

Following the recommendations of existing research [10], we conducted a manipulation check on the structure of communication networks and task urgency. Initially, by examining the occupancy of available communication paths among team members, there was no significant difference in occupancy between members in decentralized networks and centralized networks under two levels of task urgency (p > 0.1). However, in core-periphery networks and locally clustered networks, the core nodes had a slightly higher occupancy of available communication paths compared to other members (p < 0.1), indicating that the manipulation of communication network structures was effective. On the other hand, analysis of the frequency of transmissions among members in the four types of communication networks showed that under conditions of strong urgency, the average transmission frequency was significantly higher than under conditions of weak urgency (p < 0.001). This result indicates that the manipulation of urgency was effective.

4.2. Reliability and Validity Test

Firstly, regarding the reliability of the member perception scale, the Cronbach’s α value was 0.965, indicating good reliability of the questionnaire. Next, the validity of the scale was tested through exploratory factor analysis (EFA) and confirmatory factor analysis (CFA). The KMO value for member perception was 0.823, with p-values at 0.001, suggesting that the structural validity meets the requirements. Furthermore, the composite reliability (CR) value exceeded 0.7, the average VVariance extracted (AVE) value was greater than 0.5, and the factor loading (FL) values were above 0.6, indicating good convergent validity of the constructs [74]. FL values of all items under their respective variables were significantly higher than those under other variables. The square root of AVE was greater than the inter-construct correlations. Additionally, IFI, TLI values were above 0.9, and the RMSEA value was 0.031, indicating good discriminant validity [75]. Therefore, this study demonstrates good reliability and acceptable validity in the measurement of member perception.

4.3. Hypothesis Testing

Table 1. Results of descriptive statistics.

Task Urgency	Network Structure	Task Performance		Members Perception
		Mean	SD	Mean	SD
Weak	Locally clustered	5.250	0.968	3.250	0.661
	Core-periphery	6.250	1.089	3.000	0.408
	Subgroup	7.750	1.198	3.375	0.857
	Fully connected	16.350	2.057	3.875	0.331
Strong	Locally clustered	78.375	5.521	4.750	0.433
	Core-periphery	36.000	2.121	3.750	0.661
	Subgroup	32.625	2.176	3.500	0.645
	Fully connected	28.875	2.571	2.750	0.433

shows the descriptive statistical results for task performance and member perception within four communication networks under two conditions of urgency.

We adopted two-factor completely randomized multi-group design to test the hypotheses usingone-way ANOVA, two-way MANOVA and Tukey-HSD post hoc test, respectively [76,77]. First, the linearity of most groups was obvious, except for perception within the fully connected network and Subgroup network under weak task urgency. There was no multicollinearity in the correlation coefficient between task performance and member perception (|r| < 0.9). Next, according to the Box plot and Mahalanobis test, there were no univariate outliers and no multivariate outliers. Additionally, we used the Shapiro-Wilk method to test if task performance and member perception met the requirements of multivariate normal distribution (p > 0.05). The Box test indicated that the assumption of equal covariance matrices is valid (p = 0.009).

Further, there was a significant interaction effect between the variables task performance and member perception (F-value= 4.051, p = 0.003, Wilk’s λ = 0.752, partial η²= 0.122). Further, results for the one-way between-subject ANOVA test showed the interaction between network structure and task urgency had a significant effect on task performance (F-value = 8.426, p = 0.001, partial η² = 0.189), as well as member perception (F-value = 4.961, p = 0.002, partial η² = 0.136).

Next, the results of main effect test were completed (Figure 3). The fully connected network (M = 16.350, SD = 2.057) led to better task performance compared to the locally clustered network (M = 5.250, SD = 0.968, CI [9.987, 12.213], p < 0.001) and the core-periphery network (M = 6.250, SD = 1.089, CI [8.959, 11.240], p < 0.001) under the condition of weak task urgency. In addition, the fully connected network (M = 3.875, SD = 0.331) inspired more positive member perception compared to the core-periphery network (M = 3.000, SD = 0.408, CI [0.581, 1.169], p < 0.001) and the locally clustered network (M = 3.250, SD = 0.661, CI [0.262, 0.988], p < 0.001) under the condition of weak task urgency. Thus, H1a is validated.

However, under conditions of strong task urgency, the locally clustered network (M = 78.375, SD = 5.521) improved task performance compared to the subgroup network (Mean = 32.625, SD = 2.176, CI [42.841, 48.659], p < 0.001) and the fully connected network (M = 28.875, SD = 2.571, CI [46.497, 52.503], p < 0.001). Moreover, the locally clustered network (M = 4.750, SD = 0.433) also stimulated more positive member perception compared to the subgroup network (Mean = 3.500, SD = 0.645, CI [0.578, 1.922], p < 0.001) and the fully connected network (Mean = 2.750, SD = 0.433, CI [1.633, 2.367], p < 0.001). Therefore, H1b is supported.

Furthermore, we found a positive correlation between task performance and member perception (Figure 4). That is to say, as the level of member perception increases, task performance shows an upward trend under various levels of task urgency. Additionally, under higher task urgency, for any given level of member perception, task performance is highest in a decentralized structure. In a decentralized network, members with the same level of perception tend to achieve higher task performance. However, under conditions of weaker task urgency, although decentralized networks can achieve better member perception, they do not necessarily have the highest task performance; conversely, centralized networks exhibit significant advantages in task performance. Thus, H2 is verified.

To confirm the mediating role of knowledge sharing, we further used a bootstrap mediation method with 5000 samples by process [78]. Since the independent variable is categorical and there are four centralization levels of communication networks, we designed three dummy variables. As shown in

Table 2. The mediating role of knowledge sharing.

Independent Variable	Task Urgency	Network Structure	Effect	SE	95% Confidence Interval (CI)
Task performance	Weak	D1	0.186	0.088	[−0.022, 0.375]
		D2	0.310	0.144	[−0.039, 0.618]
		D3	1.120	0.515	[−2.106, 2.210]
	Strong	D1	0.407	0.443	[0.317, 1.015]
		D2	0.682	0.746	[0.647, 1.700]
		D3	2.460	2.718	[3.748, 6.205]
Member perception	Weak	D1	−0.138	0.225	[−0.606, 0.283]
		D2	−0.178	0.353	[−0.794, 0.354]
		D3	−0.215	0.447	[−0.973, 0.429]
	Strong	D1	1.469	0.756	[0.146, 3.167]
		D2	1.899	0.953	[0.183, 3.966]
		D3	2.301	1.143	[0.224, 4.755]

, the indirect effect of network structure and task performance via knowledge sharing was significant under conditions of strong task urgency because the 95% confidence intervals did not include zero. Similarly, the indirect effect of network structure and member perception via knowledge sharing was significant under the condition of strong task urgency. On the contrary, the mediating effect of knowledge sharing was not significant under condition of weak task urgency (all CIs include zero).

5. Discussion and Conclusions

5.1. Main Findings

This study involving 598 engineering project management personnel participating in group experiments based on information sharing and collective decision-making found that both centralized and decentralized communication networks have advantages in enhancing collective effectiveness under different levels of task urgency. Specifically, under condition of weak task urgency, decentralized networks contribute to improved task performance and more positive member perception. Conversely, as task urgency increases, centralized networks show advantages in enhancing task performance and also bringing about more positive member perception. The possible explanation for this phenomenon is multifaceted.

Firstly, the equal participation of all organizational members is emphasized in decentralized networks. Through effective and rapid information flow among individuals with different experiences, they integrate shared skills and knowledge to improve project outcomes and achieve strategic objectives [79,80,81]. Therefore, in the early stages of a project with weaker task urgency, each member or node usually has more autonomy, enhancing the sense of responsibility and involvement of organizational members. From the perspective of systems engineering, the organization is more stable at this time. That is, when individual nodes encounter problems, other nodes can continue to operate independently. This redundancy and fault-tolerance mechanism is particularly important in non-urgent tasks, as it provides additional time to correct or optimize specific parts of the system [82,83]. In contrast, centralized networks can help to quickly coordinate organizational members and production resources under urgent and coordination-challenging situations, improving the production efficiency of the project organization. The case of the 2010 World Expo shows that establishing a project command composed of experienced executives from relevant government departments can integrate various resources, accelerate project progress, and improve completion quality through efficient communication and coordination [84]. Studies of the large-scale 2012 London Olympics project also support this conclusion, suggesting that project managers play a key role in strengthening organizational coordination and team integration [31,85].

Additionally, this study measured PBOs members’ perceptions with regard to work involvement and job satisfaction. The results indicate that under weaker task urgency, decentralized networks inspire more positive member perception, while under stronger task urgency, centralized networks lead to more positive member perception. In other words, as task urgency changes, organizational members’ experiences of different communication networks also change. This conclusion also confirms the complementary relationship between task performance and member perception. The reason is that enhanced motivation and participation can promote better cooperation and work output. The achievement of task performance, in turn, feeds back into the perception of the work process, thus forming a virtuous cycle. Our research results are consistent with the literature emphasizing that job satisfaction enhances work performance [86,87].

Furthermore, this study also explored the internal mechanism by which the interaction of communication network and task urgency affects collective effectiveness. Under different level of task urgencies, the mediating effect of knowledge sharing in the communication network and task urgency is evolutionary. That is, when project processes are under weak urgency, the mediating effect of knowledge sharing is not significant. However, when tasks become more urgent, the transformation of collective behavior based on communication and decision-making will enhance task performance and member perception through knowledge contribution. A possible explanation is that knowledge sharing is a fundamental organizational capability needed to promote the integration of expertise, whose implementation relies on stronger linkage relationships [88,89]. Consistent with related literature, the process of knowledge sharing aims to break down barriers between knowledge owners, achieving a certain degree of free knowledge flow within a certain range, which plays a crucial role in enhancing organizational learning, knowledge creation, and organizational performance [64,90]. In situations of close relationships and high cohesion, effective knowledge sharing will help organizational members gain access to others’ unique knowledge, avoid resource waste due to repetitive knowledge production, thereby strengthening the collective belief in completing tasks and improving the perception of collaboration [91,92,93].

5.2. Theoretical Contributions

First, although there is an increasing number of studies on PBOs management focusing on inter-member relationships from a social network analysis perspective [31,94], discussions from the task urgency viewpoint are still somewhat lacking. This study advances our understanding of organizational structure relationships of PBOs by revealing how the impact of communication network structure on collective effectiveness is moderated by task urgency. Specifically, for PBOs in the construction industry in which long construction cycles are prone to turbulence, complexity, uncertainty, and ambiguity, centralized networks can provide stronger resilience with increasing task urgency, thus enhancing collective effectiveness [95].

Second, this study expands the emergent mechanisms of knowledge sharing in PBOs. The results confirm that under certain communication networks, knowledge sharing simultaneously drives task performance and member perception, reemphasizing the importance of organizational network structure indicators in this study. Moreover, from the perspective of organizational learning theory, the important mediating role of knowledge sharing between communication networks and collective effectiveness depends on the increased strength of connections among organizational members, with task urgency also enhancing these linkages. The results of this study reveal the dynamic nature of knowledge sharing channels in PBOs [14].

Third, this paper supports the current expansion of project management value beyond just focusing on expected outcomes and benefits from a reductionist viewpoint [96,97,98]. Combining task performance and member perception, this study sets an expanded definition and measurement standards for the management success of PBOs. This aligns with the more complex, multidimensional technical project management characteristics of PBOs, reflecting the contextualized, strategic nature of a wider range of projects across various sectors and those undertaken for competitive advantage [99]. Additionally, our study also found an interactive relationship between task performance and member perception, which has rarely been explored in previous research. This finding highlights the positive correlation between organizational structure and member job satisfaction in non-structured tasks as suggested by motivation theories and self-efficacy theories [100,101,102]. Therefore, future research can enhance the management value of PBOs on a more solid theoretical and methodological basis.

5.3. Managerial and Practical Implications

This study also provides important insights for decision-makers and leaders in the construction industry. Specifically, it is necessary to design organizations that are adaptable to the environment through the coordination between member behavior and interactions [1,22]. In other words, the communication structure has a crucial impact on organizational management performance. Specific suggestions are as follows.

Communication and coordination among organizational members are significant in enhancing project outcomes. When executing projects, decision-makers should gather human resources with different knowledge backgrounds and establish appropriately centralized organizational structures according to the type of and environmental factors relevant to the project. Therefore, leaders should recognize that even if formal task interdependence channels have been formed among team members within a certain relationship structure, interactions may not effectively occur. Project organization leaders should investigate and recognize the actual needs and difficulties of knowledge senders and receivers and promote cohesion between them through network interventions. For example, the appointment of a central coordinator as a hub is crucial for large PBOs like the 2020 World Expo. It is advisable to implement coordination among sub-projects by building management teams and coordinating resources to mitigate the negative impacts of potential daily changes and disruptions, thus improving collective outcomes [31]. This is a solution from a central collaboration perspective. On the other hand, in situations with relatively relaxed schedules or high cohesion among organizational members, members with different professional expertise in a decentralized structure have more defined responsibilities and rights in the collective, promoting the exchange and sharing of knowledge, forming efficient task decisions based on lean construction concepts. This organizational structure also provides adaptive collaboration for frequently occurring work coordination issues in project progress. Additionally, decision-makers and leaders of public organization projects should avoid exclusively using economic or financial performance indicators and adopt more diversified evaluation standard. Of course, especially for large-scale projects like public facilities that require significant investment, have extensive coverage, and large social impact, assessing task performance is essential. However, paying attention to PBO members’ perceptions of work (participation, satisfaction, etc.) is very important. This not only helps to improve project task performance but also enhances the well-being of project members, contributing to the formation of an experiential knowledge base. This aligns with the long-term development needs of the construction industry.

6. Conclusions

This study, based on the IMO model and organizational learning theory and from the perspective of social network analysis, explores the relationship between communication networks, knowledge sharing, and collective effectiveness in PBOs. Utilizing a classic communication experiment design and the practice of contract management in construction engineering projects, we confirmed through implementation using an online platform that the interaction between communication networks and task urgency is an important way to enhance the collective effectiveness of PBOs. The research results show that under stronger task urgency, it is recommended to adopt a centralized communication network, which leads to higher task performance and member perception. Conversely, a decentralized communication network is recommended under weaker task urgency. The measurement indicators of collective effectiveness should include task performance and member perception.

However, due to the diversity of PBOs, this study cannot precisely simulate and measure all structures in reality. For example, some PBOs are combinations of multiple structures or large project plans, which could bring noticeable changes to the results. To improve the stability of the research, future studies can combine field experiments and expand the types of projects. Additionally, our study is applicable to the construction industry in China and does not consider the impact of regional cultural differences on the behavior of organizational members. Furthermore, whether its results are applicable to developed markets or even emerging markets still needs further verification. In the future, experiments can be conducted for different industrial backgrounds and from the perspectives of participants.