Building a Resilient Organization Through Informal Networks: Examining the Role of Individual, Structural, and Attitudinal Factors in Advice-Seeking Tie Formation

Abstract

Modern organizations operate not only through formal structures but also through informal networks, which play a critical role in fostering a resilient organization. This study focused on informal advice networks within organizations as a key mechanism for strengthening contextual resilience, one of the core components of organizational resilience. By analyzing the activation of informal advice networks, this study conceptualized advice-seeking networks as a critical informal system that enhances contextual resilience and examined the individual, structural, and attitudinal factors influencing their formation. Specifically, we hypothesized that employees with higher levels of Machiavellianism are more likely to engage in advice-seeking behaviors, whereas the relationship between Machiavellianism and advice-seeking behaviors is moderated by betweenness centrality and organizational commitment, such that the positive effect of Machiavellianism on advice-seeking is weaker when betweenness centrality or organizational commitment is high. To empirically test these hypotheses, we conducted a network survey of employees at the headquarters of a life insurance company in Seoul, South Korea, and analyzed the data using an Exponential Random Graph Model (ERGM). The findings provide empirical support for all hypotheses. Based on these results, we discussed the theoretical contributions and practical implications of the study, along with its limitations.

1. Introduction

Nowadays, organizations need to transform into resilient organizations to achieve survival and success in the face of unexpected adversities such as market fluctuations, natural disasters, terrorism, and technological failures [1,2,3,4,5,6]. A resilient organization possesses organizational resilience, which helps it absorb strain, maintain stability, and even enhance its functioning despite challenges [7]. These organizations actively collect and analyze real-time information, exploring complex situations to identify potential risks [8]. This ability to navigate uncertainty is rooted in the collective ‘intelligent wariness’ of a resilient organization, which fosters proactive risk assessment and adaptive decision-making [9]. Through this organizational intelligence, the organization functions as a central hub for processing information, managing resources, and making strategic decisions to handle unexpected challenges effectively [7,8,10,11].

Most research on organizational resilience has primarily examined how organizations respond to external threats and uncertainty [12,13,14,15,16]. These studies generally assume that organizations function as a unified entity when navigating crises. Under this assumption, key resilience factors are emphasized at different levels: individual competencies at the micro level, an organization’s adaptive capabilities at the meso level, and relationships with external stakeholders at the macro level.

However, crises do not always emerge as sudden, singular events; rather, they often accumulate over time. In such cases, organizations may not respond as a cohesive whole. Instead, specific teams, departments, or functional units within the organization may take the lead in addressing these challenges [17]. This suggests that resilience is not solely an organization-wide capability but can also manifest itself through decentralized, localized responses within the firm. This internal dimension of resilience is further underscored by Kahn et al. who introduced the concept of the “geography of strain”, arguing that tensions and conflicts among teams, departments, and hierarchical levels within an organization may pose a greater risk to resilience than external disruptions [8]. Similarly, Hollnagel and Woods highlight the phenomenon of organizational fragmentation, emphasizing that understanding and managing internal adversity is critical for strengthening resilience [18].

Recent research indicates that organizational resilience depends on both formal and informal structures [19,20,21]. However, formal procedures and regulations handle predictable challenges well but may not adequately address unexpected situations or internal crises such as interdepartmental conflicts. Given the nature of emergencies, organizations should adopt a decentralized decision-making structure, rather than relying solely on hierarchical authority [19]. While hierarchical command structures may suffice in stable environments, they often fail to support prompt responses in rapidly changing and uncertain contexts. Consequently, distributed decision-making and flexible collaboration become essential for effective crisis management. In this respect, as the most fundamental components of an organization, employees’ skills, abilities, and social networks collectively determine the organization’s overall resilience. In particular, informal networks foster information exchange, relationship-building, and creative problem-solving, thereby strengthening organizational resilience during crises.

According to the definition of organizational resilience, it highlights cognitive elements (shared purpose, core values, constructive sensemaking), behavioral elements (learned resourcefulness, counterintuitive actions, prepared routines), and contextual elements (internal and external relationships that include informal networks) [15,22]. Among these, the contextual element particularly emphasizes informal networks rooted in internal and external relationships, which can serve as non-traditional bridges between organizations when formal ties are unavailable in times of crisis [23]. While formal structures may be sufficient in stable environments, dynamic interactions among employees form informal relationships that support the emergence of collective capabilities for resilience [24]. This also aligns with contingency theory, which argues for the need for a more organic organizational form to effectively respond to crises in rapidly changing contexts.

In these cases, the more active use of informal relationships among employees allows for more flexible and creative responses [8,17,25,26]. Because employees are the most fundamental part of an organization as a complex system, essential capabilities for organizational resilience can be found based on the characteristics of its employees [19]. Within an organization, there are generally two main types of informal networks among employees: advice networks and friendship networks. This study focused on advice networks, which revolve around seeking and giving work-related advice.

As one of the most prominent informal networks, advice networks naturally form within the organization without relying on any formal structures. These networks facilitate knowledge exchange among employees, allowing them to seek assistance, share ideas, and ultimately reduce tension and conflict. At the individual level, research suggests that the total number of advice relationships an employee maintains has a direct positive impact on their organizational knowledge, work performance, and job engagement [27,28]. Furthermore, advice networks improve the flow of work-related information and help coordinate activities among employees. This process reinforces organizational values and enhances stability [29]. In other words, as employees collaborate on interdependent tasks, they gain access to crucial resources—such as information, technical expertise, and psychological support—which ultimately contribute to the smooth functioning of the entire organization [30]. Thus, fostering informal relationships within organizations plays a crucial role in mitigating internal tensions and adversity while facilitating efficient knowledge exchange between employees. Ultimately, this strengthens the organization’s resilience, ensuring its ability to adapt and sustain itself in dynamic environments.

This study examined the factors that help enhance organizational resilience by looking at how employees in modern organizations form and activate informal advice networks, which allow them to share valuable experiences and knowledge. We used organizational resilience theory, which includes cognitive, behavioral, and contextual elements, with particular focus on the contextual element that is based on internal and external relationships. Our goal is to show how these informal networks help employees respond to challenges in a more flexible and creative way, ultimately strengthening the organization’s overall resilience. To investigate this process, the study conducted a network survey at the headquarters of a life insurance company based in Seoul, South Korea, collecting comprehensive data on the structure of advice networks. In addition to structural attributes derived from network data, such as betweenness centrality, the study also collected data on factors that could influence the formation of advice networks. These factors include individual characteristics such as Machiavellianism, which reflects aspects of social personality, and organizational attitudes such as organizational commitment. This study employed Exponential Random Graph Models (ERGMs) to analyze the factors driving the formation of advice networks. Specifically, it examined how individual characteristics, structural attributes of the network, and employees’ organizational attitudes contribute to the activation of advice networks in modern workplaces. By exploring these multidimensional factors, this study provides an in-depth understanding of how informal networks like advice networks facilitate information flow and relationship-building, ultimately enhancing employees to respond to challenges more flexibly and effectively.

This study offers the following contributions. First, it emphasizes the importance of informal networks in transforming modern organizations into resilient organizations by applying a social network perspective. Rather than focusing on formal structures and procedures, this study highlights the value of informal networks within organizations and their critical role in building organizational resilience. Second, unlike prior studies that mainly addressed adversities arising from intergroup relationships, this study explored the factors that drive the activation of informal networks to mitigate adversities at the individual level. By filling this gap in the literature, the study provides new insights and directions for research on informal networks. Third, this study focused on advice networks, a specific type of informal network in the workplace, and examined a range of factors, including individual characteristics (such as social personality and Machiavellianism), structural attributes of the network (such as betweenness centrality), and organizational attitudes (such as organizational commitment). By integrating these multidimensional factors, this study identified interactions among individual, structural, and attitudinal elements, demonstrating that these interactions activate informal networks, which, in turn, ultimately play a crucial role in strengthening both individual and organizational resilience.

2. Hypothesis Development

2.1. Informal Advice Network as a Crucial System of the Company

In modern organizations, informal systems are often built on relationships among colleagues. These systems emerge from employees’ voluntary social interactions, making it essential to explore the relational network structures they create.

According to Krackhardt and Hanson, organizations typically have three types of informal networks: advice networks, trust networks, and communication networks. In general, network researchers categorize networks into two main types. Expressive networks, which focus on social interactions and the exchange of emotional or social support, such as friendship networks. Friendship networks are trust-based exchange relationships in which individuals offer support with the expectation that others will reciprocate [31]. These networks are based on personal attraction and mutual positive emotions, so changes in professional ideas or practices do not threaten these relationships. At the same time, the general trust derived from such relationships tends to remain limited in scope and does not easily develop further [32]. On the other hand, instrumental networks, including advice networks and communication networks, facilitate the exchange of work-related advice, information, and resources [29,33,34]. Each network can serve as a key to solving organizational problems, and even within the same company, network structures and characteristics may differ. A trust network reflects the level of trust among members, making it essential to assess when the organization faces a crisis or implements major change initiatives. In contrast, a communication network should be examined when information flow is inefficient, resource utilization is suboptimal, or new ideas are not emerging effectively. Lastly, an advice network reveals the most influential individuals in the company’s daily operations, helping to identify the causes of political conflicts or obstacles to achieving strategic goals [34].

This study highlights the important role that advice networks play in disseminating job-related knowledge and norms to enhance organizational resilience. Within an organization, the trust network builds emotional bonds and trust among members, providing psychological security during crises. However, when it comes to organizational resilience, a network that efficiently exchanges work-related knowledge and information during a crisis is more important than merely having trust-based relationships. Although communication networks capture broad interactions and overall information flow, they fall short in addressing specific problem-solving or the exchange of specialized knowledge. In contrast, advice networks facilitate the exchange of information, advice, and opportunities for problem-solving among colleagues. Over time, advice networks within organizations evolve and expand, forming a foundational structure for professional interactions and collaboration [29]. Furthermore, these advice networks help identify true influence among members based on trust, aiding in the analysis of political conflicts or unmet strategic goals. Therefore, this study focused on the advice network, which best captures the flow of work-related knowledge and technical advice as a key factor in enhancing organizational resilience.

According to systems theory, modern organizations are open systems that rely on other individuals and subsystems within the broader scope of society for their survival [35]. Rather than being static, they develop through dynamic interactions [30]. While friendship networks can play a certain role within informal systems, the exchange of expertise facilitated by voluntary social interactions among members becomes essential in the dynamic work environments of modern organizations. This makes advice networks even more influential. To derive greater value in work-related contexts, advice networks must be given due attention. Morrison found that newcomers with larger advice networks are in a more advantageous position to acquire organizational knowledge compared to those with smaller networks [28]. This demonstrates that forming advice relationships significantly enhances knowledge acquisition. Consequently, the overall number of advice relationships maintained by employees is positively associated with organizational knowledge, job performance, and work engagement [27,28]. Furthermore, such advice-exchange relationships facilitate the flow of work-related information, coordinate activities, and strengthen organizational values, thereby enhancing organizational stability [29]. Through these dynamic interactions necessary for work execution, employees maintain their individual resources for security and contribute to the smooth functioning of the organization [30].

Therefore, this study identified advice networks—relationships in which employees seek guidance and collaborate with colleagues to solve problems—as a key relational structure within informal systems. The aim was to understand how these networks, recognized as a form of social capital in the workplace, develop and evolve into a capability that helps organizations address various adversities arising internally. Knowledge transfer, which is essential for maintaining a competitive advantage in organizations, often occurs through the process of one individual seeking advice from another [36]. Seeking advice is a fundamental mechanism for accessing and sharing knowledge, playing a crucial role in overcoming various workplace challenges. This study focused specifically on advice-seeking relationships, examining how individual characteristics, structural attributes of relationships, and employees’ organizational attitudes shape these interactions. By doing so, it provides insights into the factors that drive the formation of advice networks and their role in enhancing organizational resilience.

2.2. Machiavellianism and Advice-Seeking Ties

In recent years, the role of personality has garnered considerable attention from many authors [37,38]. The personality approach explains why individuals form unique network relationship patterns and shows that personality influences social bonds in the workplace [39,40]. For example, prior research suggests that a social personality, such as self-monitoring, helps explain network connections in workflows [40], high self-monitoring is associated with friendship and advice-seeking [41], and the Big Five personality traits are linked to informal advice networks or friendship networks [42,43,44,45,46,47]. Additionally, as the need to examine how personality influences changes in social networks is being recognized [48], there is an increasing need for research on various types of personalities to understand individual differences [49].

Therefore, among the many personalities above, we are interested in the social personality, which is a network-related personality shaped by the social environment, with particular focus on Machiavellianism [50]. In this study, we stressed the definition of Machiavellianism to encompass two main characteristics. First, individuals with high levels of Machiavellianism place greater emphasis on acquiring power and adopt a more manipulative approach in their interactions with others [51,52]. This characteristic of Machiavellianism is based on the belief that the ends justify the means [53], prompting individuals to pursue their self-interest by manipulating others without regard for relational and ethical considerations, feelings of guilt, or remorse for their actions [54,55,56]. Second, Machiavellianism was found to be positively predicted by fluid intelligence, suggesting that individuals with high levels of Machiavellianism possess strong abstraction and inference capabilities. This high level of cognitive ability is associated with the capability for strategic thinking, resulting in superior planning, reasoning, and problem-solving skills, thus earning them the designation of ‘evil genius’ [57,58].

Machiavellian individuals tend to analyze social situations and extract useful information based on shared characteristics among individuals [59]. Moreover, instead of adhering to standard organizational norms, rules, and practices, they are inclined to strategically leverage social capital to maximize access to the information necessary for achieving their goals [51,60]. Notably, when Machiavellian individuals—who are part of the Dark Triad personality spectrum—seek help, it is not merely a sign of cooperation but rather a strategic choice aimed at achieving personal goals and fulfilling specific needs [61,62]. Prior research suggested that Machiavellian individuals are less likely to engage in genuine cooperation; instead, they tend to form relationships selectively, primarily as a means to acquire necessary resources and advance their own interests [63].

This tendency is also evident at the corporate level. Studies on CEO Machiavellianism indicate that firms led by Machiavellian CEOs are more likely to engage in strategic alliances [52], suggesting that Machiavellian individuals do not simply exchange information but strategically form and utilize relationships for their own advantage [64]. Consequently, individuals with high Machiavellian tendencies are particularly interested in maximizing their use of social networks—not merely for maintaining relationships, but as a strategic tool for securing valuable resources and optimizing their personal gains. Thus, those with high Machiavellian tendencies are likely to secure resources and gain benefits by strategically engaging with colleagues who can support their work and seeking advice that provides valuable experience. Based on this, we propose the following hypothesis:

Hypothesis 1.

Employees with higher Machiavellianism are more likely to seek advice from colleagues.

2.3. Employees’ Network Positions: Betweenness Centrality in Advice Exchange Network

Machiavellian individuals are primarily focused on maximizing their self-interest and employ manipulative or deceptive strategies only when it benefits them [65]. Thus, employees with high Machiavellianism tend to seek advice from colleagues to maximize their personal gains. However, this behavior goes beyond mere information sharing or collaboration and can manifest as a manipulative act driven by self-interest [60]. Such advice-seeking behavior can undermine trust among colleagues and disrupt a genuine culture of cooperative knowledge sharing [66]. As a result, it can lead to misinformation and unnecessary conflicts within the organization, ultimately acting as an adversity that reduces overall organizational efficiency and resilience.

On the other hand, when Machiavellian individuals occupy a structurally advantageous position, they are less likely to seek advice based on self-interest. They are highly motivated to gain power and are skilled at navigating organizational politics to secure advantageous positions [60,67,68,69]. Individuals in brokerage positions play a crucial role in connecting otherwise unlinked individuals or groups, allowing them to control the flow of information [25,70]. Individuals with high betweenness centrality leverage their privileged access to unique, non-redundant information to exert influence and solidify their structural advantages [71,72]. Given these tendencies, Machiavellians are more likely to occupy structurally advantageous positions within organizational networks. Moreover, their strong networking abilities and social insight enable them to quickly recognize and exploit structural advantages, maximizing their access to valuable information [73].

However, Machiavellians in strong network positions act cautiously to protect their reputation instead of actively seeking advice. Because they already benefit from strong network positions, they do not need to engage in frequent advice-seeking to access valuable information. Additionally, they recognize that excessive advice-seeking may expose their strategic tendencies and potentially undermine their cultivated image. Machiavellians strategically leverage their networks to maintain an informational advantage while ensuring a favorable reputation. Their pragmatic approach to relationships prioritizes personal gain over emotional bonds, meaning they selectively engage with others only when it serves their interests [55]. They also avoid behaviors that could create a negative impression among colleagues, ensuring that their reputation remains intact [74].

Thus, they are less likely to engage in advice-seeking behavior when they already hold a structurally advantageous position, as doing so may not only be unnecessary but could also weaken their strategic standing. Therefore, we propose the following hypothesis:

Hypothesis 2.

Employees with higher Machiavellianism will be less likely to seek advice, if they have higher betweenness centrality.

2.4. Employee’s Job Attitude Toward Their Organization: Organizational Commitment

Machiavellian individuals tend to seek advice from colleagues as a strategic tool to advance their self-interest [74]; they utilize advice-seeking behavior to achieve personal objectives, such as expanding their influence within the organization, which may come at the cost of organizational efficiency and collaboration. Such behavior can disrupt others’ task performance, negatively impacting the organization [75], ultimately creating a form of adversity at the organizational level. Disruptions of this nature typically occur most frequently through face-to-face interactions within advice networks [76]. While such interactions may strengthen social connections, they can also negatively affect workflow and hinder task progress [77]. Over time, repeated disruptions impose increasing time pressures on individuals who are frequently approached for advice, leading to frustration, burnout, and a decline in organizational efficiency [78,79,80]. Notably, individuals with high Machiavellian tendencies are more likely to leverage organizational networks to acquire information, using it as a tool to enhance their personal power and influence. While this strategy may be advantageous in certain contexts, excessively self-centered approaches can undermine collaboration within the organization.

However, when Machiavellian individuals exhibit higher levels of organizational commitment, they are more likely to adjust their advice-seeking strategies in alignment with organizational goals. Organizational commitment serves as a key factor that connects employees to the organization, motivating them to align their behavior with organizational objectives and encouraging cooperative actions. Employees with high organizational commitment tend to integrate organizational goals and values into their own identity [81], and they are strongly motivated to pursue long-term careers within the organization [82,83,84,85]. Organizational commitment represents an attitudinal attachment, wherein individuals perceive their personal success as closely linked to the organization’s development and actively seek to contribute to its goals [86,87].

A critical aspect of organizational commitment is the willingness to contribute beyond personal interests [87]. Employees with high organizational commitment recognize that their decisions should align with the organization’s success and, as a result, are less likely to engage in advice-seeking behaviors solely for personal gain [86]. This is because they are more aware of how frequent advice-seeking may impose a burden on colleagues, leading them to regulate their behavior more carefully.

Therefore, highly Machiavellian individuals with strong organizational commitment are more likely to modify their advice-seeking behaviors by considering organizational goals. Rather than focusing solely on personal benefit, they may adopt a more strategic approach to navigating networks within the organization. This suggests that organizational commitment mitigates the negative effects of Machiavellian tendencies and serves as a moderating factor that reduces instrumental advice-seeking behavior. Therefore, we propose the following hypothesis:

Hypothesis 3.

Employees with higher Machiavellianism will be less likely to seek advice, if they have higher Organizational commitment.

Figure 1 depicts the study’s research model, in which the informal system (advice-seeking network) is influenced by individual social personality (Machiavellianism), particularly moderated by structural position aspects (betweenness centrality) and organizational attitude aspects (organizational commitment). Specifically, the advice-seeking network, as a representative form of informal advice networks—a specific type of social network within organizations—facilitates employees’ exchange of work-related knowledge, experience, and emotional support, ultimately contributing to enhanced organizational resilience.

3. Methods

3.1. Data

As network analysis typically requires a complete view of relational structures—which cannot be easily captured through sampling—it is a common research practice to select a single site and conduct a full-network survey. Following this convention, to test our hypotheses, we collected network data from the headquarters of a life insurance company located in Seoul, South Korea. More specifically, to collect complete network data, we use a roster method in which each employee is asked to answer whether he or she has built the types of relationships with other employees who are listed in the given roster. Among the types of social network data that are frequently used in social network studies, e.g., complete network or ego-centric network, we chose the complete network because the ERGM requires the complete network data to examine how a specific informal network structure may emerge from employees’ individual characteristics as well as the social features that they have in interactions with their colleagues.

Earlier, we argued that the informal structure of a company is a crucial social system that enables the company to function according to its organizational purposes. Furthermore, such an informal organizational structure can be effectively identified by the advice exchange network, in which employees voluntarily reach out to their colleagues to obtain relevant information to fulfil their tasks. Therefore, in our attempts to collect the network data, we asked the respondents to answer how often they ask for advice from each specific employee whose name is provided in the survey questionnaire.

In so doing, we obtained the complete advice exchange network data by which we can identify who reaches whom to seek for the advice that is relevant to their task completion. The final advice exchange network data to use in our ERGM analysis was arranged as a square matrix sized 99 by 99 through 100% response rate from the whole employee group. Along with the network data, we also collected individual level variables of interest, such as Machiavellian personality, organizational commitment, role ambiguity, and social stress, as well as individual demographic variables, such as age, gender, employment type, and working position. These individual level variables are all included in our ERGM analysis.

3.2. Exponential Random Graph Modeling

Given our aim to understand interdependent social systems, the Exponential Random Graph Model (ERGM) is well-suited for our purposes, especially in modeling advice networks [88,89]. Scholars empirically and theoretically assert that social networks exhibit several interdependent characteristics [90]. ERGM explicitly models network relationships and represents the complete network as a collective outcome of various local network processes. It considers the interdependence between network ties and node attributes and treats network structures as endogenous [91,92]. Specifically, since our study aimed to identify which structural and individual attributes contribute to the formation of advice networks, traditional statistical approaches that assume independence, such as t-tests, ANOVA, and regression analysis, should not be used [93,94]. Therefore, we performed ERGM analysis using the statnet and ergm packages in R, and estimated ERGM parameters using the Markov chain Monte Carlo maximum likelihood method [95].

Historically, network researchers in organizational studies have been adept at using regression models such as logit and probit models. However, these approaches have significant limitations when investigating network formation for various reasons [96]. Traditional social network analysis methods, which predict dependent variables based on individual attributes, often overlook issues arising from the interdependence of observations and struggle to predict the antecedents of network formation [88,93]. Therefore, recognizing these limitations, contemporary scholars have begun to use the ERGM to study network formation within organizations. Additionally, the Exponential Random Graph Model (ERGM) has been receiving increasing attention in recent research [91,92,97,98].

3.3. Measurement

3.3.1. Dependent Variable

The dependent variable in the ERGM is the advice network itself. Using the names of employees obtained through document retrieval, we collected network data by measuring whether and to what extent employee i seeks advice from employee j. This was categorized on a scale from “0 = Never” to “4 = A lot”. However, in this study, we constructed a binary matrix, coding employees who sought advice four times as 1 and those who did not seek advice or sought advice up to three times as 0. The network data collected for this study included a total of 102 employees, but the final analysis included a total of 99 nodes, resulting in a 99 × 99 adjacency matrix. It was deemed appropriate to exclude three executives from the analysis, so only the data of 99 employees were included in the analysis. The three executives were excluded from the analysis, not on the basis of their organizational status, but due to missing data in key variables required for this study. Some of these variables included items that the executives were unable to respond to during the data collection process. To maintain analytical consistency and ensure the integrity of the results, these cases were excluded to avoid potential bias caused by incomplete data.

3.3.2. Independent Variables

For explanation variables, they generally contain purely structural effects, which are network relation types dis-interdependent of a node’s attributes, and actor–relation effects [93]. In this paper, we added these effects while also including the nature of the actor node’s attributes. However, in this study, the pure structural effect and actor–relation effect are added in the control variable dimension and to the basic model as well. The independent variables used in this paper are Machiavellianism, betweenness centrality in advice network, and organizational commitment, which are attributes of nodes.

Machiavellianism. This study used the Kiddie Mach, a measure of Machiavellian tendencies proposed by Christie and Geis, to measure employees’ Machiavellian tendencies in dealing with their colleagues [53]. The instrument has a total of 20 items, including items such as “It is good to be nice to people in power even if you don’t like them”, “It is wise to think that people can be mean when the opportunity arises”, “You should always be honest no matter what (R)”, and “Trusting others completely will get you into trouble”. Specifically, this study followed the method proposed by Christie and Geis and used a four-point scale without a “neutral” option, with statements ranging from strongly agree (4), somewhat agree (3), somewhat disagree (2), and strongly disagree (1) [53].

Betweenness Centrality. In this study, the betweenness centrality of individuals in the advice network was calculated using the sna package in R. Betweenness centrality measures how often a particular node lies on the shortest path between other nodes in the network. Nodes with high betweenness centrality are often seen as indicators of power and influence within a group or organization due to the reliance of others on these nodes [99,100,101]. Traditionally, betweenness centrality has been defined and measured using theories and algorithms designed specifically for undirected graphs [102]. Although Gould suggests that betweenness centrality can also be measured for directed graphs, the lack of standardized units of measurement and a unique definition for a maximally central graph in directed networks make these measurements difficult to interpret [103]. Therefore, in this study, we removed the directionality from the advice network and calculated betweenness centrality for our analysis.

Organizational Commitment. To measure organizational commitment, this study used the organizational commitment questionnaire (OCQ), an instrument proposed by Mowday, Steers, and Porter [104]. The organizational commitment questionnaire (OCQ) is a measurement tool that describes global organizational commitment, and this study used a shortened version of the organizational commitment questionnaire (OCQ) that reduced the original 15 items to 9 items. This instrument has a high correlation with existing instruments and is considered appropriate for measuring organizational commitment. The shortened version of the organizational commitment questionnaire (OCQ) includes nine items that represent the feelings an individual may have about the company or organization they currently work for, including the following: “I am willing to put in a great deal of effort beyond that normally expected in order to help this organization be successful.”, “I talk up this organization to my friends as a great organization to work for.”, “I would accept almost any types of job assignment in order to keep working for this organization.”, “I find that my values and the organization’s values are very similar.”, “I am proud to tell others that I am part of this organization.”, “This organization really inspires the very best in me in the way of job performance.”, “I am extremely glad that I chose this organization to work for over others I was considering at the time I joined.”, “I really care about the fate of this organization.”, and “For me, this is the best of all possible organizations for which to work.”. In particular, this study used reversed measures for two of the nine statements, “I am proud to tell others that I am part of this organization.” and “For me, this is the best of all possible organizations for which to work.” to determine the accuracy of respondents’ statements. The scale for the items was a 5-point scale with a value of 1 for “completely disagree” and a value of 5 for “strongly agree”.

3.3.3. Control Variables

Purely structural effects. Apart from the main effects we are interested in examining in this paper, purely structural variables serving as controls are included in the model by default. Many studies have already included these pure structural effects, and many of them have used reciprocity and twopath [91,96,105,106]. Therefore, in this study, we will include these two structural effects in line with previous studies. Mutual indicates the tendency of the advice network to be reciprocal, meaning that if node i seeks advice from node j, then node j also seeks advice from node i. Twopath measures the degree to which actors sending ties receive ties as well, controlling for the correlation between in-degree and out-degree. Specifically, it refers to the extent to which i receives advice from f as much as it seeks advice from j.

Actor-relation effects. In addition to structural variables, this study has included homophily variables—employment type, gender, position, and role ambiguity—to control for variations in the homophily effects on employees’ advice-seeking behavior based on prior research [107,108,109,110]. These variables are incorporated using the “nodematch” and “absdiff” commands from the statnet package. Specifically, they mitigate homophily tendencies that determine whether employees are likely to seek advice from colleagues of the same employment type, gender, or position. Furthermore, they account for how employees with ambiguous roles are inclined to seek advice from peers facing similar role ambiguities.

Node’s attributes. In the attributes of the nodes, we first added the variable social stress, which indicates that stress in relationships or social life is likely to affect the likelihood of seeking advice. Employees experiencing interpersonal conflicts at work may encounter social stress, leading to negative emotions [111]. Such stressors can result in individuals avoiding specific individuals within the workplace and exhibiting negative effects on group interactions such as seeking advice [112]. Therefore, this study controlled for the impact of social stress stemming from interpersonal relationships on seeking advice. We also controlled for the role ambiguity variable. Role ambiguity refers to the ambiguity of the roles that employees share. We also controlled for the effects of role ambiguity based on existing research suggesting that role ambiguity provides opportunities for employees to engage in advice-seeking behavior [113].

4. Results

4.1. Descriptive Statistics

Before examining the statistical results, we first discussed the general characteristics of the advice network. Although not depicted in Figure 2 below, our study’s advice network consists of a total of 99 nodes. In the advice network, there is a 2.29% probability that an employee seeks advice from a colleague, indicating a network density of 0.02. Additionally, we identified 12 isolated nodes—employees who did not report advice relationships based on our criteria. Because the extent to which some nodes remain isolated is also considered a characteristic of an organization’s informal network, we included these isolated nodes in the ERGM analysis to accurately represent the network structure. Therefore, the overall advice network structure is depicted in the figure below.

As shown in

Table 1. Descriptive statistics results.

Count Variables	Counts		1	2	3	4	5	6	7	8	9	10	11
edges	223		1
mutual	17		0.48	1
twopath	501		0.93	0.48	1
nodematch. employment	106		0.71	0.37	0.66	1
nodematch. gender	106		0.65	0.30	0.61	0.51	1
nodematch. position	42		0.39	0.22	0.36	0.48	0.37	1
absdiff. role ambiguity	137		0.75	0.36	0.70	0.50	0.50	0.29	1
Continuous Variables	Mean	SD
social stress	2.21	0.70	0.97	0.46	0.88	0.70	0.63	0.38	0.71	1
Machiavellianism	2.27	0.28	0.04	−0.02	0.02	0.03	0.03	0.06	0.08	0.08	1
betweenness centrality	26.83	63.21	0.15	−0.01	0.15	0.10	0.08	−0.01	0.11	0.17	0.00	1
organizational commitment	3.25	0.61	0.00	0.03	0.03	−0.03	−0.01	−0.05	−0.02	−0.08	−0.16	0.11	1

Note: All correlation coefficients presented in Table 1 are statistically significant at the p < 0.05 level.

, there are a total of 223 edges, with 17 reciprocal connections in purely structural relationships, and 501 connections where advice is exchanged bidirectionally. Regarding actor relations, there are 106 connections where permanent employees seek advice from other permanent employees, while only 18 connections exist where temporary employees seek advice from other temporary employees. This suggests that temporary employees do not exhibit a strong tendency towards homophily compared to permanent employees. Additionally, there are 106 connections seeking advice from the same gender, and 42 connections seeking advice from colleagues in the same position. Particularly, there are 137 pairs where advice is sought from individuals with similar levels of role ambiguity. This indicates a tendency for individuals to seek advice from others in similar situations. Among the node attributes, the average level of social stress, indicating stress from relationships or social life, is relatively low at 2.21. Machiavellianism, the primary focus of this study, is generally low at the individual level. The mean betweenness centrality in the advice network is 26.83, with a relatively high standard deviation, indicating significant variability.

We can also see that the correlation of the main variables in this model is low. As shown in the correlation results on the right side of the chart, the correlations between the main effects of this model, Machiavellianism, betweenness centrality, and organizational commitment, are all low and significant except for Machiavellianism and betweenness centrality.

4.2. Estimation

In this study, we employed Exponential Random Graph Models (ERGMs) to test the hypotheses proposed in this paper. According to prior research, after completing the ERGM estimation, the Akaike’s Information Criterion (AIC) value serves as the primary indicator of model fit. Generally, a lower AIC value indicates a better model fit [114]. Additionally, diagnostic statistics can be utilized to assess the appropriateness of the MCMC process and the estimated results [93]. By comparing the results from observed and simulated data, we can determine the degree of alignment between the statistics obtained from the observed network and the parameter estimates. Specifically, this is achieved by calculating the difference between the actual observation and the simulated observation, divided by the standard error. If the absolute value of this test statistic is less than 2, the match is considered good, supporting the hypothesis that the observed data and the simulated data are the same, resulting in a p-value greater than 0.1.

As shown in

Table 2. ERGM estimation results.

	Model 1			Model 2			Model 3
	Est.	p Value	Test Stat.	Est.	p Value	Test Stat.	Est.	p Value	Test Stat.
Purely structural
edges	−2.16	0.00	−0.47	−2.03	0.00	−0.06	−1.95	0.00	−1.69
mutual	2.06	0.00	−1.93	2.10	0.00	−0.61	2.08	0.00	−1.60
twopath	−0.08	0.09	−0.01	−0.07	0.09	0.29	−0.07	0.13	−1.92
Actor-relation
nodematch. employment. 1	0.49	0.00	−0.61	0.52	0.00	−0.85	0.51	0.00	−0.91
nodematch. employment. 2	−0.70	0.01	0.83	−0.73	0.01	0.93	−0.70	0.01	−1.51
nodematch. gender	−0.09	0.49	0.26	−0.10	0.43	0.14	−0.10	0.45	1.75
nodematch. position	0.04	0.83	−1.51	0.05	0.80	−0.41	0.02	0.89	−0.27
absdiff. role ambiguity	−0.15	0.28	−1.10	−0.17	0.22	1.89	−0.19	0.18	−0.22
Node’s attribute
social stress	−0.31	0.00	−0.64	−0.36	0.00	−0.02	−0.38	0.00	−1.29
Machiavellianism				0.93	0.00	0.06	0.70	0.01	−1.14
betweenness centrality				0.00	0.00	−1.90
Machiavellianism × betweenness centrality				−0.00	0.01	0.79
organizational commitment							−0.16	0.17	−1.51
Machiavellianism × organizational commitment							−0.92	0.04	−0.04
AIC	2065			2046			2059

Note: p-values indicate statistically significant contributions when p < 0.05.

, we built three models to test our hypotheses. Model 1 is a basic model that includes pure structural effects (mutual, twopath), actor–relation effects (nodematch. employment; nodematch. gender; nodematch. position; absdiff. role ambiguity), and node attribute (social stress). Model 2 includes the first main effect of the study, Machiavellianism, the moderator variable betweenness centrality, and their interaction effects. Model 3 includes the second main effect of the study, Machiavellianism, the moderating variable organizational commitment, and their interaction effects.

The coefficient estimation results of the ERGM are shown in the chart below. First of all, we can see that the coefficients of Models 2 and 3, including the main effect, are lower than the basic model, Model 1. The AIC value of Model 1 is 2065, but when the main effect is added, it is 2045 and 2059, which are smaller than the original 2065. The second way to judge the goodness of fit is to check the diagnostic statistics using the mcmc.diagnostics function to check whether the mcmc process and the estimated results are appropriate. Through this process, we can check the model fit by comparing the key statistics of the observed network, with the key statistics obtained from the simulated network using the ERGM. As shown in the results plot above, the absolute value of the test statistics is less than 2, indicating that the observed network statistics are not significantly different from the simulated network. This confirms that the results of the ERGM analysis can be interpreted without much difficulty.

Then, the ERGM analysis shows that the model includes edges, which are not usually interpreted as a constant term in a regular regression model, as they can act like a constant term [93]. In Model 1, the estimated coefficient for reciprocity is 2.06 and is significant (p < 0.00), suggesting that the likelihood of forming an advisory network is higher when reciprocity is present. Furthermore, the estimated coefficient for twopath is negative and significant at the 0.1 level, indicating that employees who frequently provide advice are less likely to receive advice from their colleagues. This implies that most employees neither seek advice as much as they give it, nor do they give advice as much as they seek it. Additionally, an employee’s propensity to seek advice was influenced by whether they belonged to the same employment type. Specifically, regular employees were more likely to seek advice from other regular employees (0.49, p = 0.00), whereas non-regular employees exhibited less homophily in this regard (−0.70, p = 0.01). However, no significant effects were found for seeking advice based on the same gender, position, or experience of role ambiguity. In terms of personal attributes, employees experiencing high levels of social or interpersonal stress were found to be less likely to seek advice (−0.31, p = 0.00).

Upon testing Hypotheses 1 and 2, as shown in Model 2, Hypothesis 1, which posits that individuals with higher levels of Machiavellianism are more likely to seek advice, is supported (0.93, p = 0.00). This finding suggests that employees with high Machiavellian traits seek more advice to obtain the information necessary to achieve their goals. Similarly, as shown in Model 2, Hypothesis 2 regarding the moderating effect of betweenness centrality is also supported (−0.00, p < 0.00). Contrary to Hypothesis 1, however, the results show that employees with high Machiavellian traits are less likely to seek advice when their betweenness centrality in the advice network is high. This suggests that even employees with instrumental Machiavellian tendencies are less inclined to seek advice when they already occupy structurally advantageous positions characterized by high betweenness centrality.

The results for Hypothesis 3, as shown in Model 3, indicate that organizational commitment mitigates the positive relationship between Machiavellianism and the likelihood of seeking advice (−0.92, p = 0.04). Thus, Hypothesis 3 is supported. This suggests that employees with high organizational commitment are less likely to seek advice even if they have strong Machiavellian tendencies, as they may be less inclined to disrupt others or the organization.

5. Discussion and Conclusions

Building on our research objective of examining how employees in modern organizations form and activate informal advice networks, this study demonstrates that these networks significantly bolster the contextual dimension of organizational resilience—rooted in internal and external relationships. By employing Exponential Random Graph Models (ERGMs), we identified that Machiavellianism (as an independent variable) drives the activation of informal advice networks, while betweenness centrality and organizational commitment function as moderating factors that shape this influence. Our findings underscore that formal structures alone may be insufficient for effectively handling crises; instead, informal networks serve as a complementary mechanism that enhances both adaptability and overall resilience. Consequently, organizations should develop strategies to leverage these networks more systematically, thereby fostering flexible responses to change and a sustainable culture of collaboration.

This study offers valuable insights for both theory and practice by extending our understanding of informal networks as a core mechanism for fostering organizational resilience. Theorectically, although prior research has often centered on formal structures and top–down strategies for crisis management, our findings demonstrate the critical role of advice networks in enabling employees to share knowledge, coordinate efforts, and navigate internal adversities—aligning with the perspective that organizations function as dynamic systems rather than static hierarchies [8,18]. Furthermore, this study contributes to network research on organizational behavior by illustrating how individual personality traits, specifically Machiavellianism, influence the formation and maintenance of informal relationships: Machiavellian employees strategically engage in these networks to maximize their own benefits rather than to cultivate genuine collaboration [63]. In addition to these theoretical contributions, methodologically, we employed Exponential Random Graph Models (ERGMs) to capture entire network structures while controlling for both structural variables and node attributes, thereby treating the network as an endogenous system and providing a more comprehensive understanding of relational interdependencies. Practically, the findings highlight key factors necessary to establish and sustain a healthy informal advice network, offering strategic guidance on how organizations can facilitate constructive knowledge sharing, create stable and cooperative environments, and ultimately bolster overall resilience.

Despite these contributions, this study has several limitations that future research can address. First, because this study was conducted in the headquarters of a life insurance company in South Korea, there may be limitations in generalizing the findings to other industries and cultural contexts. Although the study established clear boundaries within a single organization to mitigate some inherent limitations of network analysis, future research should involve paired studies across two or more organizations in different industries and cultural contexts to more thoroughly examine the generalizability of the results. Second, the study provides a cross-sectional analysis of advice networks at a specific point in time. Future research could adopt a longitudinal approach to track how advice-seeking behaviors evolve over time, particularly in response to organizational changes or leadership transitions. Third, while this study focused on Machiavellianism, future research could explore how other dark personality traits, such as narcissism and psychopathy, influence advice networks. Examining multiple personality dimensions would provide a more comprehensive understanding of how individuals with different social strategies engage in workplace networks. Fourth, this study specifically examined advice networks, but future studies could investigate other informal network structures, such as trust networks, mentorship networks, or conflict networks. This would offer deeper insights into how different relational dynamics interact to shape organizational resilience. Fifth, this study did not fully incorporate structural network parameters that could help mitigate model degeneracy in Exponential Random Graph Models (ERGMs). Future research could integrate geometrically weighted degree (GWODEGREE), geometrically weighted dyad-wise shared partners (GWDSP), geometrically weighted edgewise shared partners (GWESP), and geometrically weighted in-degree (GWID) as structural factors to enhance model stability and capture network dependencies more effectively. Incorporating these parameters would provide a more robust representation of network structures in examining informal interactions and their impact on organizational resilience.