Abstract
Drawing on conservation of resources (COR) theory, we conducted a study examining how leader departure influences changes in staff turnover and how these changes subsequently affect variability in leader–member exchange differentiation (LMXD), ultimately influencing variability in relationship conflict. We examined these relationships in a four-wave survey design comprising 1,512 observations from 845 employees spanning 40 business units. The findings revealed that leader departures are associated with elevated unit-level staff turnover rates, which are subsequently associated with heightened LMXD, and in turn, to increased relationship conflict.
Keywords
Recent research has increasingly focused on collective turnover—the aggregated departure of employees within work units—because of its significant effects on organizational functioning (Hancock et al., 2013; Hausknecht & Trevor, 2011; Park & Shaw, 2013). Although much of this work emphasizes upper management turnover, less attention has been given to lower-level manager departures, despite their closer proximity to frontline employees and greater influence on team dynamics (Ballinger & Schoorman, 2007; Sauer, 2011).
Building on Context Emergent Turnover theory (CET), recent studies have explored how human capital dynamics affect unit performance during turnover events (Call et al., 2015; Pieper et al., 2023). However, a key gap remains: few studies have examined relational mechanisms as outcomes of turnover, despite their critical role in understanding and managing its impact (Hausknecht, 2017). The present study addresses this gap by examining the dynamic interplay between managers and staff turnover, leader–member exchange differentiation (LMXD), and relationship conflict. Although previous research has examined LMXD (the extent to which leaders develop relationships of varying quality with different team members) and relationship conflict (defined as interpersonal tensions and incompatibilities among team members), it has primarily relied on cross-sectional or time-lagged designs (Boies & Howell, 2006; Seo et al., 2018). As a result, these studies often overlook the cumulative nature of turnover (Nyberg & Ployhart, 2013), limiting our understanding of how these processes evolve over time. This study draws on conservation of resources (COR) theory (Hobfoll, 1989) and the crossover perspective (Westman, 2001) to propose that manager departures trigger increased staff turnover (a within-unit contagion effect), which in turn heightens LMXD and contributes to rising relationship conflict within teams.
This article offers several theoretical contributions. First, we extend research on collective turnover by examining how supervisor departures influence staff turnover dynamics, highlighting the role of proximal leader behavior in triggering nonmanagers turnover. Second, by framing relationship quality as an outcome—rather than an antecedent—of turnover, we respond to calls for deeper exploration of unit-level relational processes in turnover phenomena (Hausknecht, 2017). Third, we show how manager departures and staff turnover drive changes in LMXD, which, depending on its degree, may either exacerbate or alleviate intra-unit conflict. Finally, through the lens of COR theory, we shed light on the mechanisms through which turnover influences unit dynamics, offering new insights that inform CET theory.
This study also holds important practical implications. Understanding how leader departures drive staff turnover can support more effective succession planning and retention efforts, helping to reduce disruptions and maintain team stability. Moreover, because turnover affects workplace relationships, recognizing its impact on LMXD variability is essential for managing group dynamics, trust, and communication. Finally, by examining how both manager and staff turnover contribute to relationship conflict, we provide insights that can guide targeted interventions to address communication breakdowns, trust issues, and team cohesion challenges.
Theoretical Background and Hypotheses
COR Theory as an Overarching Theoretical Framework
Individual turnover decisions are shaped by diverse relational perspectives, many of which emphasize supervisors and coworkers as critical resources influencing whether an individual stays or leaves (Becker et al., 2023; Jo & Ellingson, 2019). We draw on COR theory (Hobfoll, 1989; Hobfoll et al., 2018) to explain how departures alter the availability and distribution of resources within business units—particularly those exchanged between leaders and members. These shifts can influence variability in LMXD, potentially intensifying coworker conflict as a defensive response.
COR theory posits that individuals strive to acquire, maintain, and protect valued resources—defined broadly as objects, states, or personal conditions. The theory highlights four key principles. First, the primacy of resource loss, where individuals react more strongly to losses than to equivalent gains, prompting protective or restorative behaviors. Second, resource investment, wherein people expend resources (e.g., effort) to protect or acquire additional ones—such as adapting to a new leader or preserving gains from a former one. Third, the desperation principle, which suggests that when resources are depleted, individuals may react defensively or irrationally in an effort to safeguard themselves. A final relevant dimension is the crossover principle, which demonstrates how emotions and resources (e.g., energy, frustration) can transfer from one individual to another—similar to emotional contagion (Felps et al., 2009)—thus reinforcing the relational impact of turnover across the unit.
Leader Departure and Within-Unit Staff Turnover
We define leader departure as either a planned or unplanned change in the formal leader of a team, business unit, or organization (Ballinger & Schoorman, 2007; Gordon & Rosen, 1981), a unit-level shock that can induce uncertainty and disrupt internal dynamics, shaping collective responses or reactions (Pieper et al., 2023). Drawing on COR principles, we propose that leader departure signifies potential resource gain or loss in the workplace. First, leader departure may indicate resource loss for employees with high-quality leader–member exchange (LMX) with the outgoing leader (Becker et al., 2023), characterized by mutual trust, respect, and open communication (Graen & Uhl-Bien, 1995). Second, such departures may raise concerns among employees about their future within the organization because leaders often control access to valuable resources, such as status or promotion opportunities (Laulié & Morgeson, 2021). These concerns can weaken organizational attachment (Feng et al., 2024; Shapiro et al., 2016). Third, leader departure can generate uncertainty about how the new leader will treat employees (Ballinger et al., 2010; Pieper et al., 2023). Employees who anticipate dissimilarities in personal characteristics, values, goals, or personalities with the new leader may believe they will gain more by leaving for another organization than by staying. Those who are highly worried about these potential mismatches may experience increased negative emotions, such as insecurity, anxiety, or pessimism, which are likely to diminish their desire to remain with the organization. This argument is supported by Ballinger et al. (2010), who found that employees who imagined their current leader leaving reported weaker organizational attachment and more negative affect. Conversely, individuals with a low-quality LMX relationship with the departing leader, marked by limited resources and autonomy, may view staying as an opportunity to gain valuable resources and improve their career prospects under the new leadership. An additional dimension introduced by a COR perspective is the exchange of resources through the crossover model. Crossover involves the dyadic transmission of psychological states and experiences between individuals (Hobfoll et al., 2018; Westman, 2001). In the context of turnover, this perspective demonstrates how the emotions, perceptions, and work attitudes of departing managers may transfer to remaining employees, potentially affecting LMX relationship dynamics within the unit.
Time gaps may occur between a leader’s decision to resign, the announcement of departure, and the actual exit (Klotz et al., 2021). During this period, the leader can influence staff turnover through the transfer of both positive and negative experiences. The crossover model (Chen et al., 2015; Westman, 2001; Westman & Chen, 2017) explores how positive and negative experiences, emotions, and resources transfer from leaders to employees and among employees themselves. The crossover model outlines the following three processes by which these experiences are transmitted within dyads or social groups: direct crossover transfer, indirect transfer, and spurious crossover (Westman & Vinokur, 1998).
The direct crossover transfer process entails the transmission of positive or negative event experiences between individuals or within a group through empathic identification. It is termed direct crossover because emotions are transferred directly from one person to another. For example, a leader about to resign might share negative emotions (frustration with authorities) or convey de-energizing personal resources (pessimism about the organization’s future). Conversely, the leader may also share positive emotions, such as excitement about a new position or enthusiasm for a future organization. Employees might empathize with the leader’s feelings, which can influence their own attitudes and decisions and potentially increase their desire to follow the leader or seek employment elsewhere.
The second mechanism is an indirect process mediated or moderated by variables, such as social support or work disengagement. Empirical evidence demonstrates that crossover of resources can lead to important gains or losses, triggering a chain of (dis)engagement processes (Bakker & Xanthopoulou, 2009; Westman et al., 2009). For instance, a manager who decides to leave due to a lack of organizational resources, such as inadequate support from senior managers or limited advancement opportunities, might disengage months before their departure by providing less career support and recognition to their employees. This can affect those employees’ levels of vigor and dedication, potentially increasing their own desire to leave.
Finally, spurious crossover occurs when individuals share common stressors, such as a weak mutual LMX relationship, resulting in a shared emotional experience, such as mutual frustration or anger. This shared experience can intensify the collective desire to quit.
Furthermore, leader departure may indirectly increase collective turnover by initiating a turnover contagion phenomenon among employees. The crossover model provides evidence that positive and negative experiences, emotions, or resources can transfer among team members (Bakker et al., 2006; Bakker & Xanthopoulou, 2009). In the context of turnover, scholars have described this phenomenon as “turnover itself causes more turnover” (Castle, 2005). Felps et al. (2009) defined this process as “the tendency to quit spread[ing] throughout a work group” (p. 547), where the decision to stay or leave is influenced by one’s coworkers. Decisions to quit can result from various behavioral cues from coworkers indicating potential turnover. These cues include (a) expressions of a desire to leave, such as declining work attitudes and concerns about the leader’s departure or the future of workgroup relationships; (b) indications of the relative ease of leaving, such as invitations for interviews or receipt of job offers; and (c) actual departure decisions. These behaviors can trigger and facilitate turnover among other employees. Research shows that when a high proportion of coworkers leave, overall employee turnover tends to increase (J. Oh & Chhinzer, 2021; Sunder et al., 2017).
The contagious process of turnover can also arise from spurious crossover (Westman, 2001). Higher collective turnover may introduce shared stressors within the unit, such as resource scarcity or increased job demands, leading to common responses, such as anxiety, insecurity, and dissatisfaction. According to COR theory (Hobfoll, 1989), if employees believe that temporary higher job demands will require them to work longer hours and engage in extra-role behaviors to compensate for the lack of human resources, they are more likely to experience job overload and fatigue, which can intensify their desire to quit (Bolino et al., 2015; Lan et al., 2022). Bartunek et al. (2008) found that coworkers often discuss their dissatisfaction and frustration with their work and job resources before making turnover decisions. Such negative perceptions and emotions are more likely to be shared among unit members. Felps et al. (2009) also found that employees in high-turnover environments tend to discuss turnover more frequently.
To date, most studies examining the implications of leader departures on staff turnover rates have focused on upper management (H. Li et al., 2020; Pieper et al., 2023) and are conducted either at the between-persons level (e.g., Ballinger & Schoorman, 2007) or at the between-units level (e.g., Michele Kacmar et al., 2006). Consequently, it remains unclear how the departures of proximal unit leaders affect staff turnover within the unit and, indirectly, the quality of relationships within those units over time. Longitudinal research typically distinguishes two types of variance: within-level (e.g., within-firm, within-business-unit, within-team, within-person) and between-level (Certo et al., 2017). From a between-unit perspective, units that experience the departure of their leader are more likely to experience a higher staff turnover rate than those with no succession events over time. Conversely, a within-unit perspective suggests that when a unit experiences a departure event at a specific point in time (e.g., Time 3), it is more likely to exhibit elevated staff turnover compared to its own historical average, defined as the average turnover rate across multiple time periods within that business unit. Based on the above considerations, we propose the following:
Hypothesis 1: Business units experiencing a leader departure will subsequently exhibit elevated staff turnover rates compared to their historical average.
Within-Unit Staff Turnover and Within-Unit LMXD
Prior research demonstrates that the quality of LMX relationships significantly influences an individual’s desire or decision to quit (Ballinger et al., 2010; Becker et al., 2023). However, there is limited exploration of how LMXD changes following turnover events. LMXD is a collective construct traditionally defined as an environment in which high-quality relationships with the leader form among some group members and low-quality relationships with the leader form among others (Liden & Maslyn, 1998). Emerging evidence suggests significant variability in LMX at the dyadic level (Nahrgang et al., 2009). This variability can lead to gains for some employees while resulting in loss or no change for others (Dimotakis et al., 2022). When multiple changes occur in dyadic relationship exchanges over a given period, the average LMXD within units is likely to fluctuate as well. We posit that frequent staff turnover constitutes a critical factor that can influence changes in LMXD within units above and beyond staff turnover levels (Nyberg & Ployhart, 2013).
In contrast to prior research indicating that turnover results from higher levels of LMXD or different LMX configurations (e.g., Nishii & Mayer, 2009; Seo et al., 2018), our study builds on the existing collective turnover and COR theory literature to propose that when staff turnover exceeds typical levels, it leads to increased levels of LMXD within the unit. Current research on the impact of collective staff turnover has not yet provided clear evidence regarding this relationship. The effect of turnover on organizational outcomes can vary, showing patterns such as linear negative, increasing negative, attenuated negative, or even an inverted U-shape (Hancock et al., 2013; Park & Shaw, 2013). We anticipate that the impact of within-unit staff turnover on LMXD will exhibit an increasingly positive trend rather than a linear pattern.
When the staff turnover rate remains steady, regardless of its level, the unit leader has limited incentives to either increase or decrease disparities in their LMX relationships. This is because the job resources available to the leader (e.g., challenge, autonomy, power, time, support) and the job demands (e.g., workload, effort, stress) remain consistent. In such a context, proximal leaders are less likely to increase LMX disparities because doing so could foster perceptions of favoritism, potentially harming overall unit cohesion and collaboration. Conversely, a leader may also have limited motivation to reduce disparities when staff turnover remains constant if their personal resources (e.g., time, attention, or support) are limited or if they are unable to acquire additional supportive resources, such as financial support, technology, or structural independence from senior leaders (Tangirala et al., 2007). Furthermore, a leader is unlikely to reduce LMX disparities if doing so involves diminishing incentives and recognition for high-performing employees or their immediate lieutenants, which could potentially decrease their motivation and engagement.
However, when staff turnover escalates from low to moderate levels, it is likely that the departures primarily involve employees who have failed to develop close relationships with the leader (Dulebohn et al., 2012). These departures, combined with a moderate influx of newcomers, are unlikely to lead to significant changes in the magnitude of LMXD for two main reasons: (a) the quantity of resources available to distribute to the remaining employees will be modest, and (b) the leader is unlikely to substantially change the quality of their relationships with the remaining employees because of significant uncertainty regarding the competencies and personalities of the newcomers.
At the same time, a higher-than-typical turnover rate at a given point in time is likely to substantially increase the magnitude of LMXD. This is because the pool of resources available to the leader, such as autonomy, participation in decision-making, and recognition and reward, is larger than typical when both low- and high-LMX employees have left. Rational and effective business leaders are likely to concentrate these resources among their more proficient and loyal employees. Research demonstrates that longer-tenured employees are typically more productive because of their accumulated organizational knowledge (e.g., routines, norms, culture) and tacit knowledge (e.g., customer preferences), which are necessary for their jobs (Park & Shaw, 2013). Additionally, organizational tenure is positively associated with performance (McDaniel et al., 1988; Ng & Feldman, 2010; Steffens et al., 2014) and may help maintain unit performance when collective turnover increases (Simón et al., 2022). These valuable employees are thus more likely to receive a significant share of resources from the leader because their attributes are crucial in shaping LMX dynamics (Nahrgang et al., 2009; Nahrgang & Seo, 2015).
When staff turnover exceeds typical levels at a given time, managers face several challenges: (a) effectively managing the increasing temporary job demands in their unit, (b) facilitating the rapid integration of a large influx of newcomers, (c) motivating the most loyal and productive employees to accept additional workload and responsibilities, and (d) preventing further turnover among the most proficient employees. One of the primary reasons leaders have developed higher levels of LMXD is to enhance work efficiency and effectiveness (Graen & Scandura, 1987). When an influx of newcomers surpasses the usual level of recruitment, both the leader and remaining employees are significantly affected. This new context can strain the leader’s capacity to provide personalized attention and support, potentially hindering the development of high-quality LMX with most newcomers. Additionally, the higher influx of newcomers requires existing members to share resources, such as knowledge, expertise, and time, to support the integration of new members and temporarily handle additional job tasks arising from increased job demands (Nyberg & Ployhart, 2013). To motivate existing employees to accept higher workloads and socializing responsibilities and share valuable resources, leaders are more likely to concentrate the available resources freed up by departures in the hands of the most loyal and proficient employees. Based on these considerations, we posit the following hypothesis:
Hypothesis 2: Compared to historical averages, elevated staff turnover rates will be associated with elevated LMXD, with this association becoming stronger as the turnover rate rises.
Based on the integration of our arguments derived from COR theory for Hypothesis 1 and Hypothesis 2, we argue that variability in staff turnover provides the mechanism through which leader departures influence changes in LMXD because the departure of a leader triggers several shifts in resource allocation and organizational dynamics that affect remaining employees and the leader’s ability to manage the changes. First, when a leader departs, it may indicate potential resource loss for some employees and generate uncertainty about their future within the organization (Feng et al., 2024; Shapiro et al., 2016). Leader departures may create uncertainty about how the new leader will interact with existing employees. Second, the crossover model in COR theory suggests that emotional and resource dynamics associated with a leader’s departure and team member turnover can transfer to remaining employees, influencing staff turnover (Chen et al., 2015; Westman & Chen, 2017). Finally, in the context of high turnover, remaining employees must handle increased job demands and integrate newcomers, which may strain the leader’s ability to maintain high-quality LMX with all employees, amplifying LMXD, with staff turnover acting as a mediator that influences resource allocation and relationship dynamics within the unit.
Hypothesis 3: Leader departure will be associated with increased LMXD through elevated staff turnover, compared to the business unit’s historical average.
Within-Unit LMXD and Within-Unit Relationship Conflict
Relationship conflict refers to disagreements among group members that stem from differences in personality, values, or interpersonal styles. Although individual studies and meta-analyses on conflict have identified its consequences, only a handful of studies consider intragroup relationship conflict as a dependent variable (De Dreu et al., 2004; Jehn & Bendersky, 2003; Pelled, 1996). We seek to address this gap by enhancing the current understanding of LMXD variability as a specific antecedent of intragroup conflict dynamics. Drawing on the COR “desperation” principle, we propose that LMXD levels exceeding typical rates can create greater divisions or tensions within a business unit. The desperation principle suggests that resource deprivation drives individuals to become defensive, aggressive, or even irrational when their resources are overstretched or exhausted (Hobfoll et al., 2018). According to COR theory, individuals seek to acquire resources from others to enhance their own effectiveness and success in their jobs or to recover lost resources. Because these resources are often not voluntarily provided, employees and groups may adopt various strategies to obtain them from coworkers or supervisors. For instance, individuals may actively seek resources through proactive help-seeking actions (Lim et al., 2020), engage in defensive behaviors, such as disengaging from experiences or events that hinder resource availability (Halbesleben et al., 2014; Hobfoll & Shirom, 2001), use defensive cognitive mechanisms by denying the immediate need to act or downplaying the significant impact of resource loss (Hobfoll, 2001), or resort to maladaptive problem-focused coping strategies, such as alcohol abuse (Bacharach et al., 2008). We posit that units experiencing a higher average LMXD than typical will employ conflict tactics as a relational defense mechanism.
Group leaders grapple with inherent tensions regarding resource allocation, while teams are acutely aware of how these allocations can foster or hinder differentiation. If the leader’s intention is to enhance task productivity, the leader should allocate resources among a select few based on contributions or investments. Conversely, if the primary goal is to promote social cohesion or prevent intragroup conflict, the leader should adhere to an equality principle by treating each member in the unit equally, allocating the same resources or rewards regardless of individual contributions or investments (Leventhal, 1976a, 1976b, 1980). This suggests that an increase in LMXD may lead to a greater intra-unit division because heightened differentiation can potentially foster political and confrontational behaviors among both low- and high-LMX team members (Galinsky et al., 2003). Team members who receive significant resources from their leader are motivated to protect their gains and maintain the status quo (Anderson & Brion, 2014). Conversely, individuals who experience significant losses in LMX benefits during a given period are motivated to recover these resources (Tarakci et al., 2016). According to the first principle of COR theory, resource loss is disproportionately more salient than resource gain; additionally, gaining resources becomes more important when other resources have been lost (Hobfoll, 2001). One way to acquire resources is to engage in both direct and indirect intragroup conflict behaviors such as power struggles or personal attacks (Bendersky & Hays, 2012; Greer & van Kleef, 2010; Tremblay, 2023). There is compelling evidence that LMXD adversely influences various team processes, including relational conflict (Choi et al., 2020; Hooper & Martin, 2008; Mayer et al., 2008). However, much of this research relies on the assumption that LMXD is invariant. A recent study suggests that stable LMX relationships are less likely to occur at the dyadic level (Dimotakis et al., 2022), suggesting that LMXD may also be unstable at the unit level, influencing variability in conflict intensity. Therefore, we propose the following hypothesis:
Hypothesis 4: Elevated LMXD, relative to a unit’s historical average, will be associated with increased relationship conflict.
Hypothesis 2 proposes that higher staff turnover in a business unit during a given period will lead to greater LMXD than typically experienced. This is because high turnover requires remaining employees to handle increased job demands and integrate newcomers, prompting leaders to concentrate resources on motivating long-tenured employees to take on temporary responsibilities and to stay. Hypothesis 4 posits that a higher LMXD during a given period will be linked to an increase in intra-unit relationship conflict as a defense mechanism. Higher-than-typical LMXD can foster political and confrontational behaviors among both low- and high-LMX unit members. Those who gain resources seek to protect them and maintain the status quo (Anderson & Brion, 2014), while individuals who lose LMX benefits will seek to recover them (Tarakci et al., 2016), creating internal divisions. This suggests that within-level LMXD mediates the relationship between staff turnover and intra-unit relationship conflict.
Hypothesis 5: Compared to its historical averages, elevated staff turnover will be associated with increased relationship conflict, mediated by heightened LMXD.
Methodology
Sample and Procedure
This study used a sample from a Canadian retailer specializing in the sale of artistic and creative materials. This chain of stores employs an average of 625 people across 40 business units. Store managers have significant freedom to manage internal climate and wage conditions. We collected longitudinal data using a survey of frontline employees, which included four assessment waves over a 14-month period. For each repeated survey, the human resources department sent invitations to store managers and employees to request their participation in the study. The invitations described the objective of the study, assured participants of confidentiality, and contained a link to the web survey. We sent reminder memos 1 week after the initial invitations. At each measurement time point, we sent invitations to employees who had a minimum of 3 months of tenure with the organization and were active employees at the time of the surveys. In the first wave, 594 individuals participated in the survey, with a response rate of 95%. In the second wave, 497 individuals participated with a response rate of 80%. The third wave included 520 individuals with a response rate of 83%. In the fourth wave, 525 participants had a response rate of 84%. Across the four waves, the sample comprised 2,111 employee observations from 1,166 employees across 160 business-unit year observations, with an average of 31 employees per store (SD = 17). The average employee was 33 years old, had 4 years of service, and worked 27 hours per week. Of these respondents, 62% worked full time, and 72% were female. For longitudinal analyses, we included participants who completed two consecutive surveys, resulting in 1,512 observations from 845 employees in 40 business units, with a final average response rate of 70.4%.
Measurements
Unit Manager Departure
For each period studied and each business unit, we assessed whether the unit leader was new or the same as in the previous period. The organization provided an identifying code for the leader of each business unit and for each survey wave. Consistent with previous research (e.g., Pieper et al., 2023), we coded 0 for business units that had not experienced a leader departure event and 1 for business units that had experienced a departure event. Notably, each business unit had only one manager. Our analysis indicates that, on average, 24% of managers left the organization completely, whereas the remaining 76% either stayed in their current position or moved to another position within the same organization. We decided to focus our analysis exclusively on leaders who had permanently left the organization rather than those who remained in the same business unit during a given period because a permanent departure is more likely to create a significant disruption in a team’s stability and a more substantial loss of team resources. A permanent leader departure can indicate broader organizational instability, which may contribute to increased employee uncertainty and anxiety. By focusing on leaders who left the organization permanently, we could more clearly isolate the effects of leadership departures on changes in staff turnover, excluding leaders who moved to other positions within the same organization. Those who had moved to other positions could introduce confounding variables because they could continue to influence their former teams indirectly, making it harder to attribute changes in staff turnover, LMXD, or conflict solely to their departure.
Unit LMXD
Each employee independently evaluated the quality of their relationship with their unit manager using the LMX-MDM scale developed by Liden and Maslyn (1998), using seven-point Likert-type scales varying from 1 (strongly disagree) to 7 (strongly agree). Similar to some previous studies (e.g., Walumbwa et al., 2008), we collapsed the measures into single constructs with highly acceptable reliability (α = .82). Consistent with previous studies on group-level LMX (e.g., Erdogan & Bauer, 2010; A. N. Li & Liao, 2014), we operationalized LMXD as the within-group standard deviation (SD) of the LMX-MDM scores. The mean (0.97) and SD (0.46) of our LMX differentiation are comparable to those of studies conducted in North America (e.g., Nishii & Mayer, 2009).
Intra-Unit Relational Conflict
We used four items from Jehn (1995) to assess relationship conflict. Participants responded on a 7-point Likert scale ranging from 1 (none) to 7 (a great deal) to the following items: (a) “How much personal friction is there among the employees in your store?”; (b) “How evident are personality conflicts in your store?”; (c) “How much tension is there among employees in your store?”; and (d) “How much emotional conflict is there among the employees in your store?” The average Cronbach’s alpha for relationship conflict was .91. Consistent with the referent-shift consensus composition model (Chan, 1998), we averaged the individual responses. The median rwg(j) for this construct was .83, the ICC (1) was .03, and the ICC (2) was .86. The data provided sufficient support for aggregation.
Staff Turnover Change
We assessed turnover data from company records as the percentage of employees who voluntarily left each unit. As in previous research (e.g., Makarius et al., 2017), we measured collective turnover rate by dividing the number of employee departures during the period by the workforce size in the same period and multiplying the result by 100 to calculate a percentage value of turnover rates. The turnover data were provided 6 months after each survey. We calculated changes in staff turnover by subtracting the turnover rate at Time 2 from the subsequent average turnover rates at Times 1, 3, and 4. For example, if the turnover rate at Time 2 was 42% and the average turnover rates in other periods were 30%, we calculated the change in staff turnover by subtracting 30% from 42%, resulting in a turnover rate change of +12% (indicating a higher-than-typical turnover rate). We repeated this process across all subsequent survey waves.
Control Variables
We included unit-level data on store size, age diversity, and gender diversity as control variables because they can influence turnover and relationship conflict (Guenter et al., 2016). We also controlled for average LMX quality (LMX mean (LMXM)]) and LMXD in business units because LMXM quality may influence LMXD and relationship conflict, and to isolate the influence of LMXD change. However, we removed gender diversity because it had no significant effect on outcomes.
Analytical Approach
To explore the dynamic relationships between the predictors and outcomes of our model over time and given the nested structure of occasions within persons and persons within groups, we employed a multilevel longitudinal modeling (MLM) technique using R package version 8.3. MLM enables investigation of within-person or within-group relationships, capturing fluctuations (Curran & Bauer, 2011). Models at the within-unit level assess the relationships among unit-centered variables or deviations from each unit’s mean score (Certo et al., 2017; Curran & Bauer, 2011). To estimate the effect of the leader’s departure on turnover change proposed in Hypothesis 1, we assessed whether the leader left the organization between Time 1 and Time 2 (or between Times 2 and 3 or 3 and 4). Because we collected staff turnover data 6 months after each survey, we estimated the effect of leader departure on changes in staff turnover by subtracting the turnover rate at Time 1 (6 months later) from that at Time 2 (6 months later). We repeated this process for subsequent time periods. To assess the impact of changes in staff turnover on LMXD proposed in Hypothesis 2, we calculated the turnover rate change as described above and then estimated its effect on the difference between the LMXD score at Time 2 and average LMXD scores from other time periods. We then repeated the process for the subsequent time periods. Regarding the association between changes in LMXD and changes in relationship conflict (within-unit effect) proposed in Hypothesis 4, we calculated the LMXD change as described above and estimated its effect on the difference between the conflict score at Time 2 and average conflict scores from other time periods. We followed the same process for all subsequent time periods. For the hypothesized indirect effects in Hypotheses 3 and 5, we used the bootstrapping procedure proposed by Preacher and Hayes (2008). Bootstrapping is a nonparametric resampling procedure used to test the significance of hypothesized indirect effects in mediation analysis. In this approach, new samples (20,000 in the present study) are drawn with replacement from the original sample, and the indirect effect (the product of the path from the independent variable to the mediator and the path from the mediator to the dependent variable) is estimated in each resample. This generates a sampling distribution of the indirect effect, from which a confidence interval (here 95%) is derived. If the confidence interval does not include zero, the indirect effect is considered statistically significant.
Results
Hypothesis Testing
Table 1 displays the descriptive statistics and intercorrelations among the study variables, and Table 2 presents the results of the MLM analyses. Hypothesis 1 proposed that leader departure would positively be associated with changes in staff turnover, implying that when a unit undergoes a leadership departure, the unit turnover rate is more elevated compared to its historical average. The findings provide robust support for Hypothesis 1 (b = 0.08, p < .01). Furthermore, we found support for Hypothesis 2. The association at the within-unit level between staff turnover and LMXD is significant and positive (linear: b = 0.40, p < .01; quadratic: b = 1.99, p < .05). The quadratic effect, illustrated in Figure 1, shows that the relationship is stronger at higher turnover rates. Business units that experienced no change in turnover rates showed the smallest change in LMXD. Interestingly, LMXD increased in a nonlinear fashion, even at low levels of staff turnover. These findings offer only partial support for Hypothesis 2, as we initially hypothesized that low to moderate levels of staff turnover would not lead to significant changes in LMXD. Furthermore, the effect of rising staff turnover on increasing LMXD is stronger than the effect of declining turnover on decreasing LMXD, highlighting the principle that negative events have a greater impact than positive ones (Baumeister et al., 2001). Notably, our findings demonstrate that changes in staff turnover have a more potent effect on shifts in LMXD than leader departure does.
Correlations and Descriptive Statistics.
Note. 845 employees and 40 business units.
p < .05. **p < .01.
Within-Level Staff Turnover and Within-Level LMX Differentiation.
Note. 1,512 responses from 845 employees and 160 units observations from 40 stores.
p < .05. **p < .01. ***p < .001.
Effect of store turnover square change on LMXD change.
To assess the indirect effect of leader departure on LMXD through staff turnover, we used the online utility developed by Selig and Preacher (2008) to generate 95% Monte Carlo confidence intervals through 20,000 replications. This indirect association was not statistically significant (b = 0.02, CI = −0.02, 0.05) because the confidence interval includes 0, providing no support for Hypothesis 3.
Regarding relationship conflict, the results presented in Table 3 show no significant direct association between leader departure and changes in relationship conflict (Model 2: b = 0.08, p > .05). However, we observed a significant positive relationship between shifts in staff turnover and changes in intra-unit relationship conflict (Model 2: b = 0.62, p < .01). These results demonstrate that when a business unit experiences elevated staff turnover within a given period, it also reports higher levels of intra-unit relationship conflict.
Predictors of Within-Level Relationship Conflict.
Note. 1,512 responses from 845 employees and 160 units observations from 40 stores.
p < .05. **p < .01.
Consistent with Hypothesis 4, we found a significant relationship between shifts in LMXD and changes in relationship conflict (Model 3: linear: b = 0.19, p < .05). However, the indirect effect at the within-unit level of staff turnover on relationship conflict through LMXD was positive but not significant (b = 0.04, CI = −0.10, 0.18), providing no support for Hypothesis 5.
Supplemental Analysis
We conducted supplemental analyses to verify whether our results were homologous across levels. The analysis showed that the effect of leader departure on staff turnover was not significant at the between-unit level (b = 0.04, p > .05), whereas this relationship was significant at the within-unit level. Moreover, the link between staff turnover and LMXD was not significant at the between-unit level (b = 0.05, p > .05), whereas the same relationship was significant at the within-unit level. With respect to the relationship conflict model, we observed a significant negative association between leader departure and relationship conflict at the between-unit level (b = −0.53, p < .05), but not at the within-unit level. This relationship suggests that business units with fewer leader departures over time tend to experience, on average, lower levels of relationship conflict than those with more frequent leader departures. Additionally, our analyses show a significant negative relationship between staff turnover and relationship conflict at the between-unit level (b = −0.66, p < .05), whereas an opposite association emerged at the within-unit level. This contrasting pattern suggests that business units with higher average staff turnover over time experience less relationship conflict compared to units with lower turnover. However, we found a homologous positive association between LMXD and relationship conflict across levels. Overall, these supplemental analyses show that the effects of leader departures on staff turnover vary across levels, whereas the association between LMXD and relationship conflict remains consistent, supporting isomorphism with this relationship (Certo et al., 2017).
Although our longitudinal study did not establish clear causality among the variables, it explored reciprocal relationships. First, we examined whether leader departure was influenced by shifts in staff turnover, LMXD, and relationship conflict. The analyses show that elevated staff turnover at a given time leads to a subsequent departure of the unit leader (b = 0.45, p < .05). By contrast, higher LMXD in a specific period is associated with a subsequent retention of the same leader (b = −0.17, p < 0.01). Furthermore, higher relationship conflict than typically experienced was not significantly associated with a subsequent leader departure. These results suggest a reciprocal relationship among leader departure, staff turnover, and LMXD at the unit-business level. In other words, we cannot rule out the possibility that a leader’s departure could have a contagious effect on staff turnover, and a higher staff turnover might prompt the unit leader to leave as well, creating a vicious turnover cycle.
We also tested whether the relationship between shifts in staff turnover and LMXD was reciprocal. We found that elevated LMXD at a given time was significantly associated with increased staff turnover (when control variables and leader departure were included in the model). This reciprocal relationship suggests that higher LMXD leads to a subsequent increase in the staff turnover rate. This finding is consistent with cross-sectional studies and individual-level analyses (Harris et al., 2014; Martin et al., 2018; Seo et al., 2018). Finally, we found that changes in relationship conflict were associated with changes in LMXD (b = 0.17, p < .05). This reciprocal association suggests that in a situation of elevated relationship conflict among employees or between subgroups, leaders may respond by allocating resources, such as time, support, and attention—more selectively. They are likely to prioritize employees who align closely with their goals or those who show the greatest potential to help resolve interpersonal conflict. Overall, these supplemental analyses show that relationships among leader departure, staff turnover, LMXD, and relationship conflict are not causally unidirectional. Instead, they appear to be reciprocal, with each variable capable of influencing the others, thereby creating a dynamic feedback loop.
Discussion
Drawing on COR theory, this article examines the intricate dynamics between leader departure and staff turnover and their influence on shifts in LMX differentiation and intra-unit relationship conflict. We rigorously tested the within-unit models through four repeated measures in a Canadian retail organization. First, we demonstrate that leader departure predicts changes in staff collective turnover rates, leading to elevated LMXD. Second, our results indicate that units experiencing higher LMXD also report increased levels of intra-unit relationship conflict.
Theoretical Implications
This study makes several contributions to literature. First, consistent with previous research (Feng et al., 2024; Hendricks et al., 2024; Pieper et al., 2023), we demonstrate that departures of proximal managers in a small retail business unit are associated with an elevated staff turnover rate. Building on the resource loss tenet and crossover process in COR theory, we propose that team members may choose to leave when their leader departs due to the anticipated loss of resources, fears about the ability to protect existing resources under new leadership, and the emotional influence of the departing leader’s positive or negative emotions. Future research should empirically test these theoretical explanations.
Second, applying a relational lens, our study contributes to the LMXD literature by identifying a leader’s departure and shifts in staff turnover as key antecedents to LMXD variability. Although previous research suggests that LMXD often results from varying mutual exchanges between leaders and employees, the role of departures in LMXD receives insufficient attention. Drawing on COR theory, we emphasize the importance of employees’ efforts to acquire, protect, and invest resources in the context of LMXD during leadership transitions. Our findings show that LMXD tends to increase more in units led by long-tenured leaders. A possible explanation for this relationship is that, over time, long-tenured leaders allocate resources more strategically, focusing on high performers or critical employees, which leads to a concentration of resources among a select few. In contrast, the observed decrease in LMXD following a leader’s departure may reflect a shift toward more equal treatment in LMXs. This shift may occur because new leaders are highly motivated to earn respect, demonstrate fairness, and avoid potential friction and conflict (Ballinger et al., 2009).
Notably, shifts in staff turnover emerged as a stronger predictor of LMXD than leader departure, whereas the average turnover rate showed no significant effect on within-unit LMXD. This highlights the critical role of workforce instability in driving shifts in LMXD. Additionally, we found a curvilinear relationship between changes in staff turnover and LMXD. Although this relationship was stronger when the staff turnover rate was higher, even a modest increase contributes to a significant shift in LMXD. One possible explanation is that when it takes time to replace departing employees or when newcomers lack experience, unit leaders are more likely to respond to workforce instability, even modestly, by concentrating their resources on the most loyal and capable contributors. This aligns with the turnover literature, which suggests that leaders often focus on mitigating the impact of new staff and enhancing unit efficacy.
Notably, units with stable staff turnover rates exhibited the lowest variability in LMXD, suggesting that stability in human resources reduces the need for altering leader–employee relationship patterns. Stable turnover may reflect effective leadership practices, reinforcing consistent LMX resource allocation. We do not rule out the possibility that stable staff turnover promotes uniform LMX perceptions, given that employees are more likely to share common LMX experiences when they consistently interact with the same colleagues.
Building on the desperation principle from COR theory, and in line with research at the between-unit level (e.g., Hooper & Martin, 2008; Mayer et al., 2008; Yu et al., 2018), we show that increases in LMXD during a specific time period are associated with escalation of relationship conflict. We propose that shifts to higher LMXD may lead to greater engagement in defense mechanisms, such as relationship conflict, because team members are more likely to perceive higher inequality when LMXD is elevated. Conversely, when LMXD is significantly reduced, the need for desperation tactics, such as engaging in conflict, diminishes because team members perceive more equal treatment and access to resources. This greater sense of equality is likely to lessen perceptions of favoritism and reduce negative emotions, such as jealousy and resentment, as well as in-group/out-group dynamics, all of which are common sources of relationship conflict within teams. These interpretations require further investigation in future research.
As noted earlier, staff turnover dynamics are directly associated with intra-unit relationship conflict levels, regardless of average LMX or LMXD. One explanation for this link is that when staff turnover exceeds typical levels, the influx of newcomers may lead to more personal confrontations with existing team members, the formation of subgroups, and the emergence of an “us vs. them” mentality. Although Kuypers et al. (2018) found no significant relationship between changes in collective turnover and task conflict, our findings demonstrate that increased staff turnover significantly contributes to the escalation of intragroup relationship conflict, highlighting the need for further investigation of turnover and types of conflict.
Unexpectedly, Mediation Hypotheses 3 and 5 were not supported. One possible reason is that the size of the changes might be relatively small. Small effects, when multiplied together, can result in a nonsignificant indirect effect. Another explanation is that shifts in LMXD after a leader leaves or changes in relationship conflict after staff turnover occurs can be shaped by other factors beyond staff turnover or LMXD, such as the leadership style of the new leader or the loss of trust in new colleagues. In other words, multiple pathways may exist; however, because only one pathway has been tested, the mediation effect might not show up as strong enough.
Practical Implications
The results of this study yield several practical implications. First, senior managers in small service businesses should recognize that unit manager departures directly contribute to increased employee turnover and may also disrupt the quality of exchanges between leaders and employees. Given the costs associated with turnover (Park & Shaw, 2013), understanding how leader departures influence staff turnover variability can aid in developing effective retention strategies, minimizing disruptions, and maintaining workforce stability. Consequently, investing in retaining effective leaders is crucial. Human resources departments should prioritize enhancing the selection process for store managers, identifying profiles associated with lower turnover. Research demonstrates that leaders who exhibit transformational or servant leadership styles (Herman et al., 2013; Jaramillo et al., 2009; Waldman et al., 2015), quickly build high LMX rapport, and maintain low levels of LMX differentiation (Harris et al., 2014) are more likely to retain their employees. When retaining leaders is not feasible, providing ample notice, clear communication, and promoting replacements from within the organization can mitigate the negative impact of leader departures (Ballinger & Schoorman, 2007). Internal promotions, in particular, may lessen the risk of broader group turnover triggered by manager departures (Pieper et al., 2023). To prevent a potential turnover spiral, human resources departments should proactively assess the strength of relationships among key employees and respond promptly to emerging concerns (Porter & Rigby, 2021).
Second, senior managers should also take proactive steps to prevent negative crossover effects before a leader departs. Reducing the time between a leader’s resignation and departure can minimize the likelihood that employees will consider leaving as well. Managers should also clearly communicate the reasons for the leader’s departure and outline steps to ensure continuity and stability within the team. Reassuring employees about their roles and career paths can help alleviate anxiety during leadership transitions. To support retention, organizations should foster friendships among team members through team-building activities or social events. Providing access to counseling services or employee assistance programs can also help employees cope with the emotional strain of colleague departures. Additionally, ensuring a fair distribution of workload among remaining staff and, if necessary, bringing in temporary support can reduce fatigue and buffer the broader impact of turnover (E. J. Oh et al., 2025).
Third, organizations should recognize that both leadership and staff stability influence how leaders interact with employees. Stable leadership in small business units often contributes to increased LMXD, whereas new leaders typically reduce it. Senior managers should carefully assess the most appropriate resource allocation strategy based on the business context, ensuring clear communication of goals and anticipating employee reactions to changes in resource exchanges. If the previous leader departed due to poor performance, decision-makers may encourage the incoming leader to adopt an equity-based approach by sharing insights into employees’ strengths and potential. Conversely, if the leader left voluntarily, it is crucial to guide new leaders in fostering more egalitarian LMX relationships, especially early in their tenure, to support a smooth transition and effective resource management.
Finally, our study indicates that rising staff turnover often prompts unit managers to increase LMXD at an accelerating rate, an approach that rarely produces positive outcomes. To prevent the contagion effects of higher turnover and elevated LMXD within units, senior managers should support unit leaders in managing these challenges effectively. Providing adequate resources helps resolve the tension between growing job demands and the need to motivate employees to remain and perform. Regular monitoring of workload distribution is essential to ensure that no team members are overburdened or underutilized, thereby promoting balanced leader–member dynamics during transitions. Proactively managing these dynamics can also help prevent the escalation of relationship conflict and the onset of a turnover spiral. However, the benefit of containing turnover contagion may diminish if the unit leader exacerbates LMXD. When staff turnover rises, senior managers should encourage consistent interactions with all team members, avoid favoritism, involve the entire team in decision-making processes, and ensure that resources and rewards are distributed based on clear and transparent criteria.
Strengths, Limitations, and Future Research
To reduce potential same-source bias, we collected information through four surveys, employing various sources to assess the variables and relationships. We obtained information on leader departure and collective turnover from company records and assessed LMXD using an objective measure. Additionally, our hypothesis tests for nonlinear effects have helped mitigate some common method biases (Siemsen et al., 2010). Despite these strengths, this study has several limitations, pointing to opportunities for future research.
First, our results are based on data from a Canadian retailing company, which may limit generalizability to other populations. To increase generalizability, future studies should explore various industries and cultural contexts. However, focusing on a single employer has advantage to control for additional sources of variance, such as pay policies and human resources staffing practices. Second, despite employing a longitudinal research design, this study cannot establish causation, so experimental studies should be conducted.
Third, we were unable to identify the reasons for leader departure or whether staff turnover was influenced by the transfer of positive or negative experiences before their departure. The effect of leader departure on LMXD and employee turnover may differ based on factors such as the reason for departure (poor performance or strained relationships with employees), whether the new leader comes from outside or inside the organization (Pieper et al., 2023), and whether they consciously or inadvertently prompted employees to leave before their departure. Further research should delve into these aspects. Additionally, future research should identify factors that can mitigate the effect of transferred emotions or experiences on staff decisions to quit before the leader’s departure. Fourth, our examination focused on the effect of departure solely at the within-unit level. Future researchers should explore the multilevel effects of leader departures, considering both within-unit and within-person dynamics, to clarify whether the effects of leader departures are consistent at the individual and group levels. For instance, do the LMX relationships formed by outgoing and incoming leaders rely on similar bases at the dyadic versus the group level? Fifth, a valuable avenue for future research is whether information about employees provided by the previous leader affects how the incoming leader establishes relationships with employees. Sixth, we did not definitively explain why employees in business units react more negatively to an increase in LMXD. The role of a justice climate is likely significant in understanding the impact of LMXD. Hence, future researchers should investigate why distributive justice and procedural justice criteria either strengthen or weaken group processes when LMXD changes. Finally, exploring whether relational and informational justice components can mitigate the effects of a substantial shift toward egalitarian relationships would be a valuable avenue for investigation.
Conclusion
Drawing on a relationship perspective and principles of COR theory, we examined the dynamics between leadership departures and staff turnover, focusing on how the exits of proximal managers and nonmanagers influence exchanges between leaders and employees within small business units. Our study demonstrates the role of leader departure in triggering staff turnover contagion and indicates that new leaders tend to adopt less differentiated LMX relationships than their predecessors, treating employees more equally. Additionally, we found that higher staff turnover prompts unit leaders to increase LMXD, which in turn intensifies relationship conflict. However, when leaders maintain consistent LMX patterns while minimizing employees’ resource loss, the risk of conflict escalation is reduced. We strongly recommend that senior managers closely monitor employees’ responses to new leadership and actively support unit leaders in managing newcomer integration and sustaining motivation among remaining employees. Such proactive measures are essential to mitigating the adverse consequences of escalating relationship conflict within and across small business units.
Finally, supplemental analyses indicate that the dynamics between leader departures, staff turnover, LMXD, and relationship conflict are more complex than initially anticipated. Rather than following a simple unidirectional path, these relationships appear reciprocal and intertwined. Importantly, the effects are not fully homologous across within- and between-unit levels, indicating level-specific processes at play. This complexity emphasizes the need to consider multilevel and reciprocal influences in understanding workplace dynamics.
Footnotes
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Social Sciences and Humanities Research Council of Canada (Grant No. 435-2018-1531).
