Studying psychological contracts as networks of exchanges: introducing the individual psychological contract network approach

Introduction

Decades of research have firmly established that employees develop psychological contracts (PCs) with their employers (Rousseau et al., 2018). These PCs consist of employees' beliefs regarding the mutual obligations between themselves and their employers (Rousseau, 1989). When employees perceive that their employers have fulfilled these obligations, they tend to reciprocate with positive attitudes and behaviors, such as increased job satisfaction and organizational commitment, and reduced turnover intentions (Zhao et al., 2007). In contrast, perceptions of PC breach, or the belief that the employer has failed to fulfill its obligations, have clear detrimental effects on employee attitudes and behaviors (Zhao et al., 2007). Yet, despite the well-documented importance of PCs, researchers still lack a comprehensive understanding of how employees cognitively structure and organize their PCs. As pointed out by Conway and Briner (2009) over a decade ago, we lack an understanding of the actual exchanges of obligations that employees perceive in their PCs. This prevents us from fully grasping how breach or fulfillment of specific obligations impact employee attitudes and behaviors, and limits the practical implications we can derive from PC research. After all, for practitioners, it is crucial to know what specific inducements an employee expects from an employer, and what they are willing to reciprocate with, to effectively manage the employment relationship.

A close inspection of the PC literature reveals that several important questions remain unanswered, partly due to methodological constraints. Traditional survey-based approaches aggregate employees' perceptions into a single score or category, which obscures the idiosyncratic content and structure of individual PCs. Moreover, existing measures cannot capture how specific obligations are interconnected or how the overall pattern of exchanges, rather than any single obligation, shapes employee attitudes and behavior.

Our goal is to apply network theory to the study of PCs, as network methods offer a powerful toolkit to address precisely these limitations. Vantilborgh (2019) introduced the individual psychological contract (iPC) network approach, which conceptualizes an employee's PC as a network of exchanges between the employee and their employer. By representing PCs as networks of nodes (i.e., obligations in the PC) connected by edges (i.e., the perceived exchange between these obligations), the iPC approach enables researchers to study the structure, content, and dynamics of individual PCs in a way that is simply not possible with conventional measurement strategies. However, this approach remained largely theoretical and has not yet been empirically tested. We therefore aim to operationalize the iPC network approach and demonstrate its utility in understanding PCs. As such, we will illustrate how employees' perceptions of specific exchanges of obligations can be mapped as networks, both at the group and individual level.

Our work makes several important contributions to the PC literature. First, we advance PC theory by empirically testing the iPC network approach. This approach has several advantages over traditional approaches to studying PCs, as it allows us to capture the complexity and idiosyncrasy of employees' PC perceptions (Vantilborgh, 2019). For example, rather than estimating average levels of perceived obligations across a sample, we can now examine how specific obligations are interconnected within an individual's PC. Knowing how specific obligations are interconnected can help us understand how breach or fulfillment of one obligation may impact perceptions of other obligations, and ultimately influence employee attitudes and behaviors. Fundamentally, this approach allows us to study PCs as dynamic systems of exchanges, rather than static constructs, thereby capturing the element of exchange that is central to PC theory yet often overlooked in empirical research (Conway and Briner, 2009). Second, we illustrate the utility of the iPC network approach by linking network characteristics to important employee outcomes, such as job satisfaction, organizational commitment, and turnover intentions. A large body of research has established that breach of the PC is associated with negative employee outcomes (Zhao et al., 2007). However, the iPC network approach allows us to go beyond this general finding and examine how the structure of an individual's PC network influences outcomes. Importantly, this approach allows us to capture characteristics of the psychological contract—such as reciprocity (i.e., the extent to which there is a reciprocal exchange between pairs of employee contributions and employer inducements) and balance (i.e., the extent to which there is equity in the exchange relationship between employee and employer)—directly, rather than inferring them from scale scores. Finally, our work has important practical implications. The iPC network approach provides a more nuanced understanding of employees' PC perceptions, which can inform interventions aimed at managing the employment relationship. We will illustrate how researchers and practitioners can use the iPC network approach to identify the perceived obligations in an employee's PC and how these obligations are interconnected. This information can help organizations tailor their HR practices to better meet employees' expectations and avoid unintentional breaches of the PC. At a more advanced level, organizations can use the iPC network approach to predict potential breaches and their impact on employee attitudes and behaviors.

Psychological contracts as networks of exchanges

The core idea of PCs is that employees and employers are engaged in a reciprocal exchange relationship (Rousseau, 1995, 1989), grounding PCs in social exchange theory (Blau, 1964). Employees offer contributions, such as effort and loyalty, and in return expect inducements from their employers, such as job security and career development opportunities (Rousseau, 1990). The PC forms a cognitive schema that organizes these perceived obligated inducements and contributions, thus providing a mental model of the employment relationship (Conway and Briner, 2009; Rousseau, 2001). Employees rely on this cognitive schema to interpret and evaluate their employment relationship, monitoring whether their employers fulfill their obligations so that they obtain the expected inducements and reach the goals they desire (Rousseau et al., 2018).

This description of PCs as cognitive schemas is widely accepted in the PC literature, yet implies a number of characteristics that are often overlooked in empirical research. First, PCs are inherently idiosyncratic, as they are based on individual goals, perceptions, and interpretations of the employment relationship (Conway and Briner, 2009). Put simply, no two employees will have the same PC, even when working for the same employer. Not only may employees perceive different obligated inducements and contributions, but they may also organize the exchanges between these obligations differently. Unfortunately, this idiosyncratic character has been difficult to capture empirically. Existing survey instruments ask employees to rate a fixed set of obligations that may not be relevant to their particular employment situation, and they aggregate responses into composite scores that mask individual variation. More fundamentally, the interdependencies between obligations (i.e., whether a particular inducement is perceived as linked to a particular contribution, and how strongly) cannot be captured by any standard measurement approach. The iPC network approach addresses all of these limitations simultaneously. Second, PCs are dynamic, as they are formed, maintained, and changed through ongoing interactions between employees and employers (Rousseau et al., 2018). Employees may start the exchange relationship with a number of perceived obligations based on pre-hire expectations, but these perceptions may change over time as they gain more experience with their employer. Moreover, repeated fulfillment of obligations may solidify the exchange relationship, while breach of obligations may lead to a reevaluation–or even dissolvement–of the PC. Third, the element of exchange is central to PC theory, yet often overlooked in empirical research (Conway and Briner, 2009). Most empirical studies either focus on the content of PCs (i.e., what obligations are perceived) or on the overall perception of fulfillment or breach (i.e., whether the PC is perceived as being fulfilled or breached). Hence, these studies fail to capture the actual exchanges of obligations that employees perceive in their PCs, and how these exchanges are interconnected (Vantilborgh, 2019). In summary, while the PC is widely conceptualized as an idiosyncratic, dynamic schema of exchanges, empirical research has yet to fully capture these characteristics.

The iPC network approach by Vantilborgh (2019) offers a promising avenue to address these limitations. In essence, this approach conceptualizes an employee's PC as a network representing the perceived exchanges of inducements and contributions between the employee and their employer. The nodes in the network represent the specific obligations (employer inducements and employee contributions) that the employee perceives, while the edges represent the perceived exchanges between these obligations. The approach is grounded in network theory and builds on the idea of perceived causal networks, a concept that has been applied in clinical settings to understand how individuals perceive the relationships between symptoms and behaviors (Burger et al., 2024; Klintwall et al., 2021). Because the iPC network approach allows for the mapping of specific obligations and their connections for each individual employee, it captures the idiosyncratic nature of PCs (Vantilborgh, 2019). Moreover, by assessing the network at multiple time points, researchers can examine how the PC network evolves over time, thus capturing its dynamic nature. Finally, by focusing on the exchanges between obligations, the iPC network approach directly addresses the element of exchange that is central to PC theory. In the next sections, we will outline how the iPC network approach can be operationalized in practice, how network theory can be leveraged to study PCs, and how this approach can enhance our understanding of the employment relationship.

Operationalization of the iPC network approach

To illustrate how the iPC network approach can be operationalized, we will start by describing an example of a PC of a hypothetical employee, John. When John was hired, he expected his employer to provide him with job security, career development opportunities, and a supportive work environment. These perceived obligations were based on his pre-hire expectations and the information he received during the recruitment process. Based on conversations with his supervisor and colleagues, John realized that his supervisor was willing to support his career development in return for John demonstrating initiative and taking on challenging tasks. Moreover, John perceived that his colleagues expected him to contribute to a positive team atmosphere in exchange for their support. Vice versa, John believed that him supporting his colleagues would contribute to a positive team atmosphere. Overall, John felt that he had to demonstrate loyalty to his employer in return for job security, career development opportunities, and a supportive work environment. Simultaneously, John believed that job security would enable him to take on challenging tasks. This example illustrates the essential elements of a PC: perceived obligated inducements (job security, career development opportunities, supportive work environment) and contributions (initiative, challenging tasks, positive team atmosphere, loyalty), as well as the exchanges between these obligations.

Figure 1 presents a network representation of John's PC, showing the perceived exchanges between the obligated inducements and contributions. As can be seen, the network consists of nodes (circles) representing the specific obligations and directed edges (arrows) representing the perceived exchanges between these obligations. Importantly, the directed edges in this network do not imply causality in the statistical sense. Rather, they reflect the direction of perceived exchange: an arrow from node A to node B indicates that the employee perceives obligation A as part of the exchange for obligation B. For example, the edge from “Initiative” to “Career Development” indicates that John perceives that demonstrating initiative will lead to career development opportunities. Importantly, this network consists of two types of nodes–employer obligations (inducements) and employee obligations (contributions)–with edges only running between nodes of different types. This implies that the network is bipartite, which is a key characteristic of the iPC network approach (Vantilborgh, 2019). In essence, this means that there can never be exchanges between two employee or two employer obligations. This example illustrates how an iPC network corresponds to the mental schema that employees use to organize their PC perceptions (Rousseau, 2001). Before we explain how network theory can be leveraged to study PCs, we will first outline the procedure that can be used to construct iPC networks in practice.

Constructing iPC networks involves two sequential steps that correspond to the group-level and individual-level approaches. The first step involves identifying the specific obligations an employee perceives in their PC. Practically, this involves presenting employees with a list of potential obligations derived from prior research, covering both employer inducements (e.g., job security, career development) and employee contributions (e.g., effort, loyalty), and asking them to indicate which obligations they perceive as part of their own PC. Open-ended questions can supplement this list to capture any idiosyncratic obligations not covered by existing inventories. The selected obligations become the nodes in the employee's iPC network. Returning to our example, John would select the seven obligations he perceives as relevant (job security, career development, supportive work environment, initiative, challenging tasks, positive team atmosphere, and loyalty).

The second step entails either estimating a group-level network or individual-level iPC networks. The binary obligation data from Step 1 (i.e., coding whether each obligation is present (1) or absent (0) in each employee's PC) can be used to estimate a group-level network capturing which obligations tend to co-occur across employees. This is done using an Ising model (Isvoranu et al., 2022), a statistical network model specifically designed for binary variables. Named after physicist Ernst Ising, the Ising model estimates the conditional association between each pair of obligations after accounting for all other obligations in the network. The result is a group-level network in which edges reflect partial associations between obligations across employees, revealing which inducements and contributions tend to be perceived together. These group-level Ising networks differ from individual-level iPC networks in several ways. First, iPC networks reflect perceived exchanges of obligations between two actors in a social exchange relationship. As mentioned above, this implies that these iPC networks are bipartite. In contrast, Ising networks reflect partial associations between all obligations in the network. Hence, Ising networks can have associations between employee contributions and employer inducements, which could be due to an exchange process, but they also have associations between same type obligations (i.e., employee contribution—employee contribution or employer inducement—employer inducement). Such associations may arise because two obligations share certain PC features (e.g., they are both transactional in nature) or because they arise from the same source (e.g., they were both communicated by the same supervisor) (Vantilborgh, 2019). Second, iPC networks are idiosyncratic, meaning that one estimates a unique network for each individual. Ising networks, in contrast, are nomothetic, meaning that they are estimated at the group-level. Third, iPC networks have directional edges, showing the direction of the exchange, whereas Ising networks are undirected.

To estimate individual-level iPC networks, we use the perceived causal network approach (Burger et al., 2024; Klintwall et al., 2021). This involves presenting each employee with all possible pairs of the obligations they selected in step one and asking them to rate whether they perceive an exchange between each pair, and if so, in which direction and how strongly. For example, John would be asked whether demonstrating initiative leads to career development opportunities (an exchange he perceives strongly), whether career development opportunities lead to demonstrating initiative, and so on for all obligation pairs. The strength ratings form the edge weights, and the perceived directions form the arrows, yield a unique directed, weighted network for each employee.

Together, these two steps allow researchers to capture both the shared patterns in PC content across employees (group level) and the idiosyncratic exchange structure of each individual's PC (individual level). To illustrate this approach, we created a Shiny app¹, located at: http://bit.ly/3LGX0sU.

Leveraging network theory to study psychological contracts

In the following section, we will explain how network theory can be leveraged to study PCs using the iPC network approach. In particular, we will focus on three key features of PCs that can be studied using network characteristics: reciprocity, balance, and density. Moreover, we will illustrate how these network characteristics can be linked to important employee outcomes.

First, reciprocity is a fundamental principle of social exchange theory and is central to PC theory (Blau, 1964; Rousseau, 1989). PCs are based on the idea that employees and employers engage in reciprocal exchanges of obligations, where each party fulfills their obligations in return for the other party fulfilling theirs. In other words, employees rely on a “tit-for-tat” principle, meaning that fulfillment of inducements is reciprocated by fulfillment of contributions and breach of inducements is reciprocated by breach of contributions. This corresponds with the notion of balanced reciprocity, suggesting “quid-pro-quo” transactions between both parties in the exchange relationship (Sahlins and Sahlins, 2003). When employees notice that their PC is breached, they may switch to a negative norm of reciprocity, meaning that breach perceptions are reciprocated by employees not fulfilling their own obligations (Cropanzano et al., 2017; Mitchell and Ambrose, 2007). Several studies have confirmed that the norm of reciprocity governs reactions in psychological contracts (Conway and Coyle-Shapiro, 2012; Coyle-Shapiro and Kessler, 2002; Dabos and Rousseau, 2004). In general, these studies rely on scales to measure employees' general orientation toward reciprocity norms (e.g., Wu et al., 2006) or they infer reciprocity indirectly from aggregate fulfillment scores. Neither approach captures the degree to which reciprocity is present in the specific obligations that constitute an individual employee's PC, as they do not test whether inducements and contributions in the PC are mutually linked.

Interestingly, the iPC network approach offers a means to operationalize reciprocity in the PC directly (Vantilborgh, 2019). Network theory proposes that link reciprocity captures the extent to which pairs of nodes in a directed network form mutual connections (Squartini et al., 2013). In other words, two nodes A and B in a directed network are reciprocal when there is an edge going from node A to node B, and from node B to node A. This concept has been extended to networks with weighted edges, with reciprocity operationalized as the bidirectional weight shared by both edges (Squartini et al., 2013). The total reciprocity (r) of the network is then defined as the ratio of reciprocal weights between edges and total weight of edges (r = 1 implies perfect reciprocity in network, r = 0 implies absence of reciprocity in network). Given that this total reciprocity indicator captures the extent to which exchanges between obligations are reciprocated, it provides a direct measure of the norm of reciprocity in the PC. Although existing reciprocity scales do not capture the structure of an employees' PC, they do measure employees' general orientation toward reciprocal exchange. If the network-derived reciprocity measure is valid, employees who report stronger general reciprocity orientation should also have more structurally reciprocal iPC networks. We will therefore use the balanced reciprocity scale as a convergent validity criterion, relating the total reciprocity indicator obtained from the iPC networks to scores on the norm of reciprocity scale (Wu et al., 2006). Moreover, in line with prior research on reciprocity, we expect that high levels of reciprocity in the iPC network are linked to positive employee outcomes (i.e., higher job satisfaction and affective commitment and lower turnover intentions).

Hypothesis 1: There will be a positive relation between the total reciprocity in the iPC network and scores on a scale measuring balanced reciprocity.

Hypothesis 2: There will be a positive relation between the total reciprocity in the iPC network and job satisfaction, affective commitment, and a negative relation with turnover intentions.

Second, balance is another key principle in PC theory, as employees strive for a balanced exchange relationship with their employers (De Cuyper et al., 2008; Ruokolainen et al., 2018). This notion of balanced exchange is rooted in social exchange theory, which posits that individuals seek to maintain equity in their relationships (Blau, 1964; Shore and Barksdale, 1998). By comparing employee contributions to employer inducements in the exchange relationship, Shore and Barksdale (1998) constructed a typology to describe different patterns of balance in PCs. When employees perceive that their contributions are roughly equal to the inducements they receive, they are said to have a balanced PC. However, Shore and Barksdale (1998) point out that perceiving low to moderate levels of both contributions and inducements should be considered different from perceiving moderate to high levels of both contributions and inducements, labeling the former as mutual low obligations and the latter as mutual high obligations. In contrast, when employees perceive that their contributions exceed the inducements they receive, they are said to have an employee over-obligation PC. Conversely, when employees perceive that the inducements they receive exceed their contributions, they are said to have an employee under-obligation PC. Several studies have confirmed that balance in PCs, in particular mutual high obligations, is associated with positive employee outcomes, such as reduced turnover intentions, increased levels of affective commitment, and increased job satisfaction (De Cuyper et al., 2008; Shore and Barksdale, 1998). However, these studies typically rely on arbitrary cut-off scores (e.g., median splits) to classify employees into difference balance categories, which may not accurately capture the underlying balance in the PC (for an exception, see Vantilborgh et al., 2013).

The iPC network approach offers a means to operationalize balance in the PC directly at the individual level (Vantilborgh, 2019). The central idea is that balance can be captured by summing the weights of all incoming and outgoing edges for each node in the iPC network. Before summing the weights, we first need to distinguish between edges going from employer obligations (inducements) to employee obligations (contributions) and vice versa. The former represent the contributions that the employee perceives they need to make in return for the inducements they receive, while the latter represent the inducements that the employee perceives they will receive in return for their contributions. By assigning negative signs to the former and positive signs to the latter, we can create a summary score for the entire network that captures the overall balance in the PC. An overall negative score indicates that the employee perceives that their contributions exceed the inducements they receive (employee over-obligation PC). Conversely, an overall positive score indicates that the employee perceives that the inducements they receive exceed their contributions (employee under-obligation PC). A score close to zero indicates that the employee perceives a balanced exchange relationship (balanced PC).

To further distinguish between mutual low obligations and mutual high obligations, we can also compute the average absolute weight of all edges in the network (i.e., global strength of the network) (van Borkulo et al., 2015). The global strength captures the density of the network, which reflects the overall level of perceived exchanges between obligations in the PC. In other words, a low global strength indicates that the employee perceives few exchanges between obligations, while a high global strength indicates that the employee perceives many exchanges between obligations. A balance score close to zero combined with a low average absolute weight thus indicates mutual low obligations, while a balance score close to zero combined with a high average absolute weight indicates mutual high obligations. As shown in prior studies, this distinction matters, as mutual high obligations are associated with more positive employee outcomes than mutual low obligations (De Cuyper et al., 2008; Shore and Barksdale, 1998). In summary, the literature suggests that balance in the PC is associated with positive employee outcomes, especially when employees perceive mutual high obligations. In case of imbalance, employee over-obligation PCs are associated with more negative outcomes than employee under-obligation PCs. These effects can be tested directly using the balance score and global strength obtained from the iPC networks.

Hypothesis 3: There will be a positive relation between the balance score in the iPC network and job satisfaction and affective commitment, and a negative relation with turnover intentions.

Hypothesis 4: There will be a positive relation between the global strength in the iPC network and job satisfaction and affective commitment, and a negative relation with turnover intentions.

Next, we expect that the network characteristics of reciprocity, balance, and density are linked to breach perceptions and feelings of violation. Several studies have shown that perceptions of breach are associated with negative affective reactions (i.e., feelings of violation), which in turn lead to negative employee outcomes (Morrison and Robinson, 1997; Zhao et al., 2007). Importantly, perceptions of breach are likely to be influenced by the structure of the PC, but may also influence the structure of the PC itself (Vantilborgh, 2019). In other words, breach perceptions–and by extension feelings of violation–may relate to the network characteristics of reciprocity, balance, and density. First, we expect that low reciprocity in the iPC network is associated with higher breach perceptions and feelings of violation. When employees perceive a lack of reciprocity in their PC, they may feel that the exchange relationship is unbalanced and unfair, leading to perceptions of breach and feelings of violation. Alternatively, perceptions of breach may lead employees to re-evaluate their PC and perceive a lack of reciprocity. Second, we expect that imbalance in the iPC network is associated with higher breach perceptions and feelings of violation. When employees perceive that their contributions exceed the inducements they receive (employee over-obligation PC), they may feel that they are being exploited, leading to perceptions of breach and feelings of violation. Third, we expect that low density in the iPC network is associated with higher breach perceptions and feelings of violation. When employees perceive few exchanges between obligations, they may feel that the exchange relationship is weak and unstable, leading to perceptions of breach and feelings of violation. This is akin to the notion of a transitional psychological contract, which is characterized by low levels of perceived obligations and a lack of long-term commitment (Rousseau and Tijoriwala, 1998).

Hypothesis 5: There will be a negative relation between the total reciprocity in the iPC network and breach perceptions and feelings of violation.

Hypothesis 6: There will be a negative relation between the balance score in the iPC network and breach perceptions and feelings of violation.

Hypothesis 7: There will be a negative relation between the global strength in the iPC network and breach perceptions and feelings of violation.

Finally, to illustrate the individual and the group-level approaches to iPC networks, we will compare the networks of full-time and part-time employees. The group-level approach implies estimating an Ising network model for both groups. The individual-level approach requires estimating an iPC network for each individual, and then comparing the distributions of network characteristics between the full-time and part-time employee groups. Conway and Briner (2002) argue that psychological contracts may differ between full-time and part-time employees. For example, part-time employees may be less likely to receive certain inducements (e.g., promotion opportunities, task variety, autonomy) than full-time employees. Moreover, full-time employees may have different career goals and could have more expectations from their employer. Conway and Briner (2002) also suggest that because part-time employees spend less time in the organization, they may perceive less obligations in the exchange agreement. While they could not confirm that contract type (full-time vs. part-time) moderated the effects of breach perceptions on outcome, their arguments suggest that the structure of the psychological contract may be influenced by contract type. Based on this, we explore if there are differences in reciprocity, balance, and global strength between full-time and part-time employees, based on both the individual-level and group-level approaches to iPC networks.

Research question 1: Are there differences in reciprocity, balance, and global strength between full-time and part-time employees, when using the individual-level and the group-level approach to iPC networks.

Methods

Procedure

We collected data using a three-wave prospective design with a 1-week interval between each wave. The purpose of this design was to reduce self-report biases (e.g., response tendencies or social desirability) by separating the measurement of the iPC network from that of other variables. The interval was deliberately kept short to minimize attrition and ensure that the iPC network structure remained relatively stable across measurement moments. We recruited participants through the participant pool of Prolific. Respondents could participate in the study if they met the following criteria: age between 18 and 65 years, employed full-time or part-time in a paid position, and having a direct supervisor. These inclusion criteria ensured that participants were actively engaged in the workforce and had a formal employment relationship with their employer, allowing them to form a psychological contract. Participants first completed an informed consent form, after which they completed the first wave of the online survey, which was programmed in Qualtrics. We assessed job satisfaction, affective commitment, turnover intentions, breach perceptions, and feelings of violation at time 1. One week later, participants were invited to complete the second wave of the survey, in which we assessed the iPC network. Finally, 1 week later, participants were invited to complete the third wave of the survey, in which we assessed the norm of reciprocity. In line with Prolific's guidelines, participants were compensated for each completed survey. To remove careless responders, we excluded participants who completed the survey in less time than 2 s per item (Ward and Meade, 2023).

Participants

We recruited a total of 558 participants for the study. The average age of respondents was 30.03 years (SD = 8.5) and ages ranged from 18 to 58 years. The sample contained 232 female (41.58%) and 277 male (49.64%) participants, with 49 participants not disclosing their gender. About half of the participants were employed full-time (n = 267 or 47.85%) with the remainder employed part-time (n = 291 or 52.15%). Participants were recruited from various countries, resulting in a diverse sample containing employees from 32 different countries. Most participants currently lived in the United Kingdom (n = 117), Poland (n = 114), Portugal (n = 50), Greece (n = 42), Italy (n = 46), and Spain (n = 22).

Measures

At time 1, we assessed sociodemographic variables, breach perceptions, feelings of violation, job satisfaction, affective commitment, and turnover intentions. The iPC network was assessed at time 2, 1 week later. Finally, at time 3, we assessed the norm of reciprocity.

iPC network

Our measure of the iPC network was based on the PC scale developed in the context of the PSYCONES (Psychological Contracting Across Employment Situations) project (Isaksson, 2005; Isaksson et al., 2010). Participants were presented with a list of 17 employee contributions and 15 employer inducements that are commonly found in PCs (see Appendix for a full list of items). For each item, participants indicated whether they perceived this obligation in their PC (0 = no, 1 = yes). After selecting the obligations they perceived in their PC, participants were presented with all possible pairs of selected obligations (both contributions and inducements) and asked to indicate whether they perceived an exchange between these obligations on a four-point rating scale ranging from 1 (“Not at all”) to 4 (“To a large extent”). This allowed us to construct an edgelist and weighted adjacence matrix for each participant, representing the perceived exchanges between the obligations in their PC, serving as the basis for constructing the iPC network.

We computed three network characteristics from each individual iPC network. All three were derived from the weighted adjacency matrix of each participant's network using custom R functions.

Total reciprocity measures the extent to which perceived exchanges in the network are bidirectional (Squartini et al., 2013). For each pair of obligations, we checked whether the employee reported a perceived exchange in both directions. When edges existed in both directions, the reciprocal weight of that pair was set to the smaller of the two edge weights, reflecting the strength of the shared mutual exchange. We summed these reciprocal weights across all pairs of obligations and divided by the total sum of all edge weights in the network. The resulting score ranges from 0 (no exchanges are reciprocated) to 1 (all exchanges are fully reciprocated at equal strength).

Balance measures the degree to which the perceived exchange structure favors one direction over the other. For each edge in the network, we added its weight to a running total if the edge ran from an employee contribution node to an employer inducement node (reflecting the employee's sense that their contributions generate inducements from the employer), and subtracted its weight if the edge ran in the opposite direction (i.e., from an employer inducement node to an employee contribution node). A negative balance score reflects an employee over-obligation pattern, a positive balance score reflects an employee under-obligation pattern, a score of zero indicates a fully balanced exchange.

Global strength measures the overall intensity of the perceived exchanges in the iPC network (Borsboom et al., 2021) and is also referred to as density in the network literature. It was computed as the mean of the absolute edge weights across all edges present in the network. Higher values indicate that the employee perceives exchanges as generally strong and. Lower values indicate weaker, more tentative perceived connections between obligations, characteristic of a PC with little perceived exchange structure.

Covariates

Job satisfaction

Job satisfaction was measured with the Generic Job Satisfaction Scale (Macdonald and Maclntyre, 1997). This scale consists of 10 items that assess overall job satisfaction (example item: “I feel good about working at this company”). Respondents rated the items on a Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). The scale demonstrated good internal consistency in our sample (α = 0.81).

Affective commitment

We measured affective commitment with the 8-item subscale of the three-component model of commitment scale developed by Meyer et al. (1993). An example item is “I would be very happy to spend the rest of my career with this organization.” Items were rated on a Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). The scale demonstrated good internal consistency in our sample (α = 0.86).

Turnover intention

Turnover intention was measured with a two-item measure by Bentein et al. (2005). Both items were rated on a Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). An example item is “I often think about quitting this organization”. The scale demonstrated acceptable internal consistency in our sample (α = 0.81).

Breach perceptions

We measured general breach perceptions with a five-item scale (Robinson and Wolfe Morrison, 2000) that assesses the extent to which employees perceive that their employer has failed to fulfill its obligations (example item: “My employer has broken many of its promises to me even though I've upheld my side of the deal”). Items were rated on a Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). The scale demonstrated good internal consistency in our sample (α = 0.91).

Feelings of violation

We also assessed feelings of violation with a four-item scale by Robinson and Wolfe Morrison (2000). This scale measures the emotional impact of perceived breach (example item: “I feel betrayed by my organization”) and items were rated on a Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). The scale demonstrated good internal consistency in our sample (α = 0.91).

Reciprocity

We measured the norm of reciprocity using the norm of reciprocity scale developed by Wu et al. (2006). This scale consists of three subscales that measure generalized reciprocity (4 items; e.g., “My organization would help me develop myself, even if I cannot make more contributions at present”), balanced reciprocity (5 items; e.g., “My organization takes care of the organization's interests as much as my interest”), and negative reciprocity (7 items; e.g., “I have the impression that my organization is up to something that could hurt me”). Whereas, generalized reciprocity captures the extent to which employees perceive that their organization is willing to help them without expecting immediate returns, balanced reciprocity captures the extent to which employees perceive a balanced exchange relationship governed by a simultaneous exchange of equivalent resources (Sahlins and Sahlins, 2003). Finally, negative reciprocity captures the extent to which the exchange relationship is governed by self-interest and maximizing ones own interests at the expense of others (Sahlins and Sahlins, 2003). All items were rated on a Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). The internal consistency was good for the generalized (α = 0.88), balanced (α = 0.8), and negative reciprocity (α = 0.89) subscales.

Analyses

To interpret effects in regression models, we estimated and compared predicted values at different levels of the predictor (25th and 75 quartile), using the marginaleffects package in R (Arel-Bundock et al., 2024). The R code used to analyze the data can be consulted at https://osf.io/r983j?view_only747cee2b054449eeb001ec283aab7466.

Results

Descriptive statistics

Figure 2 shows how frequently each employer inducement and employee contribution was perceived as obligated in the PC (see Appendix for an overview of all obligations together with their abbreviations). The most frequently selected employee contributions were “Respect the norms and regulations of the company” (e1), “Meet the performance expectations for the job” (e4), and “Be punctual (prompt)” (e2). The least frequently selected employee contributions were “Volunteer to do tasks outside my job description” (e10), “Accept an internal transfer if necessary” (e16), and “Go to work even if I don't feel particularly well” (e17). The most frequently selected employer inducements were “Provide me with a safe work environment” (o1), “Ensure fair treatment by my managers and supervisors” (o3), and “Provide me with an environment free from violence and harassment” (o5) The least frequently selected employer inducements were “Help me in dealing with problems I encounter outside work” (o4), “Provide me with a job that is challenging” (o10), and “Be flexible in matching my demands of non-work roles with work” (o11).

Figure 3 shows the Ising network model estimated for the entire sample. In other words, this network represents the group-level approach to iPC networks, showing the associations between the perceived obligations in the PC across all employees. This network is estimated based on the binary presence or absence of each obligation in the PC, with edges representing the conditional dependencies between obligations after controlling for all other obligations in the network. It is important to note that this network is undirected, meaning that edges imply associations between obligations, but not the direction of these associations. Moreover, this network contains edges between obligations of the same type (i.e., between employee contributions or between employer inducements), which are not part of the iPC network approach, but are included when estimating a psychometric Ising network. When we zoom in on the edges between employee contributions and employer inducements, we find that the strongest edges are between “Develop new skills and improve my current skill” (e11) and “Improve my future employment prospects” (o14), between “Be a good team player” (e3) and “Provide me with a good working environment” (o2), and between “Meet the performance expectations for the job” (e4) and “Provide me with opportunities to advance and grow” (o7) (see Appendix B for a full list of edges). As can be seen in Figure 3, the strongest edges are typically between obligations of the same type, such as: “Provide me with a safe work environment” (o1) and “Ensure fair treatment by my managers and supervisors” (o3). This suggests that some obligations of the same type tend to cluster together, for example because they share similar PC features (Vantilborgh, 2019).

Figure 4 shows four randomly selected individual iPC networks from the sample. This illustrates the idiosyncratic nature of PCs, as each network is unique and reflects the specific obligations and perceived exchanges between these obligations for each employee. Not only are there differences in the specific obligations that are perceived in the PC, but also in the structure of the network itself, including the number and strength of edges between obligations. For example, the top two networks show a relatively simple structure with few obligations and edges, while the bottom two networks show a more complex structure with many obligations and edges. Moreover, whereas the top left network puts a selection of organizational inducements at the center of the network, the bottom right network puts employee contributions at the center of the network.

Next, we computed the network characteristics of reciprocity, balance, and density for each individual iPC network. Whereas, each individual's network is unique (as illustrated in Figure 4), we can still summarize the distribution of network characteristics in the sample. Table 1 shows the descriptive statistics and correlations between these network characteristics and the covariates. The total reciprocity in the iPC networks correlates positively with job satisfaction and affective commitment. Likewise, the global strength of the iPC networks correlates positively with job satisfaction and affective commitment, and negatively with breach perceptions. These two network characteristics also correlate positively with each other, which is not surprising since both are based on the weights of edges in the iPC networks. The balance network characteristic correlates positively with job satisfaction, but is not significantly correlated with any other covariate. It is worth pointing out that participants, on average, had a negative balance score (M = −16.52), which is consistent with prior work showing that employees are more likely to perceive that they contribute more to the exchange relationship compared to their employer (De Cuyper et al., 2008).

Table 1. Means, standard deviations, and correlations between iPC network characteristics and covariates.

M and SD are used to represent mean and standard deviation, respectively. Values in square brackets indicate the 95% confidence interval. The confidence interval is a plausible range of population correlations that could have caused the sample correlation (Cumming, 2014).

^* indicates p < 0.05.

^** indicates p < 0.01.

Relationships between iPC network characteristics and covariates

Next, we tested our hypotheses regarding the relations between the iPC network characteristics and the covariates. For this, we estimated a series of regression models in which we regressed each covariate on the three iPC network characteristics (total reciprocity, balance, and global strength), while controlling for socio-demographic variables (age, gender, and contract type). For the regression models predicting job satisfaction, affective commitment, and turnover intentions, we also controlled for feelings of violation, given its established relationship with these outcomes and because it is a more proximal predictor than breach perceptions (Zhao et al., 2007).

Balanced reciprocity

Table 2 shows the results of the regression model predicting balanced reciprocity. The model explained a significant amount of variance in balanced reciprocity [R² = 0.10, 90% CI (0.05, 0.14)]. In line with Hypothesis 1, total reciprocity in the iPC networks was positively related to balanced reciprocity [b = 2.16, 95% CI (1.02, 3.30)]. Increasing total reciprocity from its 25th percentile (0.81) to its 75th percentile (0.91) is associated with an estimated increase of 0.21 points in balanced reciprocity [95% CI (0.1, 0.33), p < 0.001], controlling for other predictors.

Table 2. Regression results predicting balanced reciprocity.

Breach perceptions and feelings of violation

Table 3 shows the results of the regression model predicting breach perceptions, with the model explaining a significant amount of variance in breach perceptions [R² = 0.04, 90% CI (0.01, 0.06)]. We found no support for Hypothesis 5, as total reciprocity in the iPC networks was not significantly related to breach perceptions [b = 0.91, 95% CI (−0.37, 2.18)]. In line with Hypothesis 6, balance was significantly and negatively related to breach perceptions [b = −0.002, 95% CI (−0.004, −0.0001)]. Increasing balance from its 25th percentile (−28.25) to its 75th percentile (1) is associated with an estimated decrease of −0.06 points in breach perceptions [95% CI (−0.12, 0), p = 0.038], controlling for other predictors. In line with Hypothesis 7, global strength was negatively related to breach perceptions [b = −0.28, 95% CI (−0.51, −0.04)]. Increasing global strength from its 25th percentile (2.78) to its 75th percentile (3.33) is associated with an estimated decrease of −0.15 points in breach perceptions [95% CI (−0.28, −0.02), p = 0.021], controlling for other predictors.

Table 3. Regression results predicting breach perceptions.

As can be seen in Table 4, the regression model explained a significant amount of variance in violation feelings [R² = 0.04, 95% CI (0.02, 0.07)]. In contrast with Hypothesis 5, total reciprocity in the iPC networks was significantly and positively related to feelings of violation [b = 1.62, 95% CI (0.21, 3.03)]. Increasing total reciprocity from its 25th percentile (0.81) to its 75th percentile (0.91) is associated with an estimated increase of 0.16 points in feelings of violation [95% CI (0.02, 0.3), p = 0.024], controlling for other predictors. Balance was not significantly related to feelings of violation, contrary to Hypothesis 6 [b = −0.002, 95% CI (−0.004, 0.001)]. Finally, in line with Hypothesis 7, global strength was negatively related to feelings of violation [b = −0.27, 95% CI (−0.53, 0.01)]. Increasing global strength from its 25th percentile (2.78) to its 75th percentile (3.33) is associated with an estimated decrease of −0.15 points in feelings of violation [95% CI (−0.28, −0.02), p = 0.021], controlling for other predictors.

Table 4. Regression results predicting violation feelings.

Job satisfaction, affective commitment, and turnover intentions

As can be seen in Table 5, the regression model explained a significant amount of variance in job satisfaction [R² = 0.35, 90% CI (0.28, 0.39)]. We found no support for Hypothesis 2, as total reciprocity in the iPC networks was not significantly related to job satisfaction [b = 0.73, 95% CI (−0.03, 1.48)]. In line with Hypothesis 3, balance was positively related to job satisfaction [b = 0.001, 95% CI (0.0001, 0.0002)]. Increasing balance from its 25th percentile (−28.25) to its 75th percentile (1) is associated with an estimated increase of 0.04 points in job satisfaction [95% CI (0.003, 0.07), p = 0.031], controlling for other predictors. Global strength was unrelated to job satisfaction, contrary to Hypothesis 4 [b = 0.10, 95% CI (−0.04, 0.24)].

Table 5. Regression results predicting job satisfaction.

Table 6 shows the results of the regression model, with the model explaining a significant amount of variance in affective commitment [R² = 0.21, 90% CI (0.14, 0.25)]. In line with Hypothesis 2, we found that total reciprocity in the iPC networks was positively related to affective commitment [b = 1.12, 95% CI (0.05, 2.19)]. Increasing total reciprocity from its 25th percentile (0.81) to its 75th percentile (0.91) is associated with an estimated increase of 0.11 points in affective commitment [95% CI (0.005, 0.21), p = 0.04], controlling for other predictors. Balance was not significantly related to affective commitment, contrary to Hypothesis 3 [b = 0.001, 95% CI (−0.001, 0.003)]. Finally, global strength was unrelated to affective commitment, contrary to Hypothesis 4 [b = 0.04, 95% CI (−0.16, 0.24)].

Table 6. Regression results predicting affective commitment.

Finally, Table 7 shows the results of the regression model, with the model explaining a significant amount of variance in turnover intentions [R² = 0.31, 90% CI (0.25, 0.36)]. In line with Hypothesis 2, total reciprocity in the iPC networks was negatively related to turnover intentions [b = −1.92, 95% CI (−3.38, −0.46)]. Increasing total reciprocity from its 25th percentile (0.81) to its 75th percentile (0.91) is associated with an estimated decrease of −0.19 points in turnover intentions [95% CI (−0.33, −0.04), p = 0.01], controlling for other predictors. Balance was not significantly related to turnover intentions, contrary to Hypothesis 3 [b = 0.001, 95% CI (−0.001, 0.003)]. Likewise, global strength was unrelated to turnover intentions, contrary to Hypothesis 4 [b = 0.13, 95% CI (−0.14, 0.40)].

Table 7. Regression results predicting turnover intentions.

Exploring differences between full-time and part-time employees

To explore differences between full-time and part-time employees using the group-level approach, we first estimated a separate Ising network for each group using the IsingFit package in R (see Figure 5). The resulting networks represent the co-occurrence structure of perceived obligations for each group separately, allowing visual inspection of potential differences in network topology. To formally compare the two networks, we used the Network Comparison Test (NCT; van Borkulo et al., 2023), implemented in the NetworkComparisonTest package in R. The NCT is a permutation-based hypothesis test that evaluates whether two networks differ significantly from one another. We specified 10,000 permutation iterations to ensure a stable null distribution and treated the two groups as independent samples. The NCT tests two aspects of network invariance: (1) global strength invariance, which tests whether the overall sum of absolute edge weights differs between the two networks, and (2) edge weight invariance, which tests whether any individual edge weight differs significantly between groups. Results suggested no significant difference in global strength between the full-time and part-time employee networks (S = 6, p = 0.668). Comparing individual edge weights revealed only one significant difference: the edge between “Protect my company's image” (e12) and “Provide me with a reasonably secure job” (o6) was significantly stronger for part-time than for full-time employees (p = 0.020), suggesting that part-time employees more strongly perceive job security as contingent on their organizational loyalty.

To compare full-time and part-time employees using the individual-level approach, we computed total reciprocity, balance, and global strength for each participant's iPC network and then compared the group means on each characteristic using independent samples t-tests. This approach is straightforward to implement: once individual network characteristics have been derived (as described in the Measures section), any standard group comparison method can be applied to them, just as one would compare scale scores between groups. The advantage of the individual-level approach is that it preserves the variability within each group and allows researchers to examine not only whether groups differ on average, but also to explore individual differences within groups or to use the network characteristics as predictors in further analyses. We found no significant differences between full-time and part-time employees in total reciprocity, ΔM = 0.01, 95% CI (−0.01, 0.02), t(509.37) = 0.83, p = 0.405, balance, ΔM = −2.03, 95% CI (−9.09, 5.03), t(507.19) = −0.57, p = 0.572, and global strength, ΔM = −0.01, 95% CI (−0.09, 0.06), t(515.00) = −0.33, p = 0.740.

Discussion

The goal of the current study was to demonstrate how iPC networks can advance our understanding of psychological contracts in organizations, and to validate key iPC network characteristics by examining their relations with established PC covariates. Results provided mixed support for our hypotheses. Total reciprocity showed the most consistent pattern of relations, correlating positively with balanced reciprocity, affective commitment, and feelings of violation, and negatively with turnover intentions, suggesting it captures a meaningful aspect of the exchange relationship in the PC. Balance and global strength showed more selective effects: balance was positively related to job satisfaction and negatively to breach perceptions, while global strength was negatively related to both breach perceptions and feelings of violation, consistent with the idea that denser PC networks are more resilient to disruption (Van De Leemput et al., 2024). Finally, comparing full-time and part-time employees using both the group-level and individual-level approaches revealed no significant differences in network structure or characteristics between the two groups, but illustrates how the iPC approach can be applied to group comparisons. Overall, these findings offer initial support for the validity of the iPC network approach and its proposed characteristics.

Theoretical implications

Several authors have argued that PCs should be studied as mental schema that guide employees' perceptions and interpretations of their employment relationship (Rousseau, 2001; Rousseau et al., 2018; Vantilborgh, 2019). Yet, a gap remains in how PCs are theorized and studied in practice. A major theoretical implication of our study is that the iPC network approach provides a concrete operationalization of the mental schema perspective on PCs. By conceptualizing PCs as networks of interconnected obligations, the iPC network approach captures the complexity and idiosyncratic nature of PCs as mental representations. Moreover, the iPC network approach brings exchange to the forefront of PC theorizing, as it focuses on the perceived exchanges between obligations in the PC. This is an important advancement, as exchange is a core element of PC theory (Rousseau et al., 2018).

The significance of this contribution becomes clearer when we consider what questions the PC literature has long been interested in but has struggled to answer empirically. First, researchers have long acknowledged that PCs are idiosyncratic, yet the field has relied almost exclusively on aggregate measures that reduce individual variation to a group-level mean (Conway and Briner, 2009). The iPC network approach addresses this directly by producing a unique network representation for each individual, making the idiosyncratic structure of PCs empirically tractable for the first time. Second, PC theory has always placed exchange at its core, yet empirical research has been unable to test whether specific obligations are perceived as interconnected or whether certain obligations drive others (Vantilborgh, 2019). The iPC approach operationalizes these exchange relationships directly as directed, weighted edges, enabling researchers to ask not only which obligations are perceived, but how they are linked. Third, a long-standing question concerns which specific obligations are most influential in shaping employees' reactions. Standard approaches treat all obligations as equivalent contributors to a total score; the iPC approach opens the door to examining obligation-level centrality using network measures (Bringmann et al., 2019). While these are not new questions, the iPC network approach for the first time provides the methodological infrastructure to pursue them empirically.

Our study sought to illustrate the iPC network approach and validate some of its proposed characteristics by examining their relations with established psychological contract covariates. The results provided mixed support for our hypotheses, suggesting that some iPC network characteristics capture important aspects of PCs, while others may require further refinement or reconsideration. First, if iPC networks are to be considered valid representations of PCs, we would expect that the reciprocity characteristic relates to established measures of reciprocity in exchange relationships. The positive relation between total reciprocity in the iPC networks and balanced reciprocity provides initial support for this notion. As such, the total reciprocity characteristic may form an indirect measure of the norm of reciprocity in psychological contracts. Importantly, as the total reciprocity characteristic does not depend on self-report measures, and is instead inferred from the structure of the iPC network, it may be less susceptible to biases such as social desirability and common method variance (Arthur et al., 2021; Podsakoff et al., 2003). Moreover, the total reciprocity characteristics was related in meaningful ways to various psychological contract outcomes, including affective commitment and turnover intentions, as well as feelings of violation. This suggests that the total reciprocity characteristic captures an important aspect of PCs that relates to employees' attitudes and behaviors in the workplace. Second, the balance and global strength characteristics offered more mixed results. In partial support of Hypotheses 3 and 6, balance was positively related to job satisfaction and negatively related to breach perceptions, consistent with the expectation that employee under-obligation (i.e., perceiving more inducements than contributions) is less detrimental for the employment relationship than employee over-obligation (De Cuyper et al., 2008; Shore and Barksdale, 1998). However, balance was not significantly related to affective commitment, turnover intentions, or feelings of violation. This suggests that the effects of balance may be most pronounced for outcomes that closely reflect the immediate experience of the exchange relationship, such as general satisfaction and breach perceptions, while more distal outcomes like commitment and turnover intentions may be primarily shaped by other factors. Global strength was negatively related to both breach perceptions and feelings of violation, suggesting that employees with stronger overall connections between obligations in their PC are less likely to perceive breach and experience violation feelings. This finding aligns with the notion that dense networks are more stable and resilient to disruptions (Van De Leemput et al., 2024). Put differently, employees with many strong connections between obligations in their PC may be better able to withstand perceived breaches, as the overall structure of their PC remains intact. While this protective nature of global strength is promising, its lack of relations with job satisfaction, affective commitment, and turnover intentions suggests that further research is needed to fully understand its implications for psychological contract outcomes.

Overall, our findings offer some initial evidence supporting the validity of the iPC network approach and its proposed characteristics. Importantly, the iPC network approach introduces new theoretical constructs that can enhance our understanding of PCs in organizations and open up new avenues for research. Building on network theory, this approach allows us to leverage network characteristics to better understand the structure and dynamics of PCs. For example, future research could explore how centrality measures (e.g., strength, betweenness, closeness) can be used to identify key obligations in the PC that have a disproportionate impact on employees' perceptions and behaviors. This could help explain why certain obligations are more salient or influential in shaping employees' attitudes and behaviors. An important consideration is that different centrality measures may capture different aspects of importance within the iPC network (Bringmann et al., 2019). Moreover, recent studies suggest that nodes in the periphery of networks may actually be more important than central nodes in changing network structure and dynamics (Van De Leemput et al., 2024).

Practical implications

A key practical implication of our study is that organizations can use the iPC network approach to assess and understand the psychological contracts of their employees. In the past, organizations have typically relied on surveys that measure general perceptions of PC content or of breach and violation. While these measures provide useful information, they do not capture the complex and idiosyncratic nature of PCs. In practice, managers and HR professionals are often interested in understanding the specific obligations that a particular employee perceives in their PC, as well as how these obligations are interconnected. The iPC network approach offers a way to capture this complexity by allowing employees to map out their own PCs in a structured manner. For example, managers could use this structured approach in one-on-one meetings with employees to discuss their PCs and identify potential areas of misalignment or unmet expectations. This could lead to more realistic expectations and avoid unintentional breaches which can harm the employment relationship.

Moreover, the iPC network approach can help organizations identify potential areas of intervention to improve employee satisfaction and commitment. By analyzing the iPC networks of employees, organizations can identify key obligations that are central to the network and may have a significant impact on employees' perceptions of their PC. Network characteristics, such as centrality measures, can help identify these key obligations, although research is still needed to understand which centrality measures function well in the context of such networks (Bringmann et al., 2019). Interventions can then be targeted at these key obligations to enhance the overall PC experience for employees. For example, if an iPC network analysis reveals that many employees perceive a lack of opportunities for career advancement as a central obligation in their PC, the organization could implement programs to strengthen career development opportunities. This targeted approach can be more effective than generic interventions that may not address the specific needs and expectations of employees. This approach can also be used at the team or department level to identify shared obligations and expectations among employees, allowing for more tailored interventions that address the collective needs of the group.

To illustrate how organizations can use the iPC network approach in practice, we have developed a shiny app that allows users to create and visualize iPC networks based on their own data. The app can be accessed at http://bit.ly/3LGX0sU. After selecting the obligations perceived in their PC, users can enter the perceived exchanges between these obligations to construct their iPC network. The app then computes the network characteristics and provides visualizations of the iPC network, allowing users to explore the structure of their PC and identify key obligations.

Limitations and future research

We have to acknowledge a number of limitations of the current study. First, our approach to assessing iPC networks requires participants to evaluate all possible pairs of obligations they perceive in their PC. This can be a time-consuming and cognitively demanding task, especially for participants who perceive many obligations in their PC. This may lead to fatigue and reduced attention, potentially affecting the quality of the data. Future research could explore alternative methods for assessing iPC networks, such as the Perceived Causal Networks (PECAN) method (Klintwall et al., 2021). This method limits the number of pairs that participants need to evaluate by asking them to select no more than three incoming and three outgoing edges for each obligation they perceive in their PC. This results in sparser networks, which may be easier for participants to evaluate.

Second, we focused on a limited set of iPC network characteristics in the current study. However, iPC networks can be characterized in many other ways, such as through centrality measures (e.g., strength, betweenness, closeness) and community detection algorithms (Vantilborgh, 2019). For example, centrality measures allow us to identify which obligations have the largest influence within the iPC network or serve as key bridges between different parts of the network. More central obligations may have a greater impact on employees' perceptions of their PC and related outcomes, allowing for an alternative way to identify key obligations in the PC. Community detection algorithms can help identify clusters of obligations that are closely related within the iPC network, providing insights into how different types of obligations are organized within the PC. Future research could explore whether obligations cluster together based on their content (e.g., psychological contract types; Rousseau (1995)) or based on the source of the obligation [e.g., different parties in the exchange such as supervisors, coworkers, or the organization as a whole; Alcover et al. (2017)]. As such, we believe that the iPC approach opens up many avenues for future research to explore new characteristics of psychological contracts that have not been studied before.

Third, while we separated the measurement of the iPC networks and covariates in time, our data does not allow us to draw causal inferences regarding the relations between iPC network characteristics and covariates. For example, while certain iPC network characteristics (e.g., global strength) may lower the likelihood of perceiving breach, it is also possible that perceiving breach leads to changes in the iPC network (e.g., weakening the connections between obligations). Moreover, we did not control for unmeasured variables such as tenure, which could act as a confounder for certain relations. Future research could employ longitudinal designs to examine how iPC networks evolve over time and how changes in iPC network characteristics relate to changes in psychological contract outcomes. This would provide a more robust test of the causal relations between iPC network characteristics and important outcomes. In recent years, several techniques for longitudinal network analyses have been developed, such as graphical vector-autoregression models for time-series and panel data (Epskamp, 2020), dynamic time warp analysis (Hebbrecht et al., 2020), and cross-lagged network models (Wysocki et al., 2025). These techniques are well suited for group-level networks where the ties are estimated directly from empirical data. In addition, Burger et al. (2024) developed a longitudinal extension for perceived causal networks, which could be applied directly to iPC networks.

Finally, we chose to use OLS regression analysis to test our hypotheses, to avoid adding unnecessary complexity and to ensure that the focus remained on the iPC network methodology. However, other approaches could be used to relate iPC network characteristics to outcomes, such as structural equation modeling and path analysis. These approaches would allow for testing a set of associations with multiple outcomes simultaneously, while taking into account potential relationships between outcomes. Alternatively, one could apply person-centered analysis (e.g., latent profile analysis, mixture regression models) to explore if there are groups in the population with qualitatively different profiles of network characteristics or with uniquely different effects of network characteristics on outcomes (Woo et al., 2024).

Conclusion

This study aimed to illustrate how the iPC network approach can advance our understanding of psychological contracts in organizations. We demonstrated how iPC networks can be constructed and analyzed using both group-level and individual-level approaches, offering both practical and theoretical insights. Our findings provided initial support for the validity of certain iPC network characteristics, particularly total reciprocity and balance, in relation to established psychological contract covariates. However, other characteristics, such as global strength, yielded mixed results, highlighting the need for further research to refine this construct. Overall, the iPC network approach offers a promising framework for studying psychological contracts as complex, interconnected systems of obligations. By leveraging network theory, this approach opens up new avenues for research and practice, allowing for a deeper understanding of the dynamics of psychological contracts in the workplace.

Generative AI statement

The author(s) declared that Generative AI was used in the creation of this manuscript. The author(s) used generative AI tools (Claude) to assist with spell-checking and language improvement of this manuscript. The author(s) reviewed and take full responsibility for the final content.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/forgp.2026.1856493/full#supplementary-material

Supplementary material Supplementary material at the source

Footnotes

Edited by: Alexandre J. S. Morin, Concordia University, Canada

Reviewed by: Lia M. Daniels, University of Alberta, Canada

Matthew McLarnon, Mount Royal University, Canada

Footnotes

Please note that this Shiny app is meant to illustrate the core ideas and principles of the iPC network approach and showcases how data can be collected to construct an iPC network. The app is not meant as a tool to collect data for research or practical purposes.