Introduction

In spite of increasing knowledge and awareness spurred by more than three decades of research, workplace bullying continues to be a pervasive and detrimental problem that affects employees and organizations worldwide (Nielsen & Einarsen, 2018). Bullying at work has been established as a significant threat to the health and well-being of both targets (Mikkelsen et al., 2020) and bystanders (Nielsen et al., 2020b). Moreover, organizations suffer greatly from workplace bullying (Hoel et al., 2020). In particular, research has demonstrated relatively high turnover rates among targeted employees (e.g. Glambek et al., 2015; Hogh et al., 2011), which further impairs organizational efficiency, competitiveness and productivity (Kline & Lewis, 2019; Laschinger & Fida, 2014). At the same time, reports of these and similar tendencies are often general and do not delineate the mechanisms involved, nor consider the notion that certain employees may be more vulnerable to adverse outcomes. As such, although much is currently known about the individual-level outcomes of exposure to workplace bullying, there is still a need for knowledge about explanatory mechanisms as well as vulnerability and resilience factors among those exposed (Rai & Agarwal, 2018b).

In the present study, we investigate whether length of service moderates the association between exposure to workplace bullying behaviors and work-related outcomes. Specifically, we propose a prospective, conceptual model of exposure to bullying as a precursor of turnover intention developing through job dissatisfaction, wherein length of service denotes a moderator of this mediation chain. The rationale for the model draws on an integration of tenets from the conservation of resources theory (COR; Hobfoll, 1989) and the organizational socialization literature (e.g., Nelson, 1987). Specifically, in line with research demonstrating that organizational entry is associated with uncertainty and pressures to adapt, adjust, and fit in (Jones, 1986; Saks et al., 2007), we contend that short length of service is associated with poorer individual ability to mobilize resources among targets of bullying behaviors. Conversely, we argue that with long length of service, targets will retain greater resource availability and stronger individual ability to apply and benefit from the resources they possess. Hence, we propose that short length of service denotes a vulnerability factor regarding job dissatisfaction and turnover intention among targets of bullying exposure, while long length of service is associated with increased resilience regarding the same outcomes. This represents the main contribution of our study, and may hold significant implications for our understanding of why and when exposure to bullying is related to common work-related outcomes.

Exposure to bullying and work-related outcomes from a resource perspective

Exposure to workplace bullying behaviors involves being targeted with psychological and social mistreatment and harassment from superiors and/or colleagues. Bullying scenarios often endure over some time, with the target finding it increasingly difficult to defend against the mistreatment (Einarsen et al., 2020). One approach to workplace bullying research has been addressing antecedents and outcomes of long-term victimization. However, due to the variance of experiences and developmental trajectories across different cases, a more apt approach with regard to e.g. vulnerability and coping possibilities is to address exposure to bullying behaviors at work on a continuum, from occasional exposure to more severe forms of victimization (see also Nielsen et al., 2020a). The term exposure to workplace bullying behaviors is thus frequently employed to investigate adverse individual outcomes (e.g., Nielsen et al., 2016; Reknes et al., 2014), including job dissatisfaction (Giorgi et al., 2015), turnover intention (De Clercq et al., 2021) and various moderators (e.g., Reknes et al., 2019). Extending previous research on the association between bullying and turnover intentions, we will employ a resource perspective based on the COR theory’s tenets (Hobfoll, 1989) to explain why it is reasonable to expect a forward association between exposure to bullying behaviors and turnover intentions via length of service -dependent development of job dissatisfaction.

The COR theory has been applied as a theoretical framework for making predictions regarding exposure to bullying behaviors previously (Rai & Agarwal, 2018a), essentially based on the tenet that individuals use resources to obtain and protect “centrally valued assets” – things of high individual importance (Hobfoll, 2011). Resources may include work, financial reserves, skills, competence, reputation, self-efficacy or anything else perceived by an individual as a means of reaching goals associated with important assets (Halbesleben et al., 2014). In one sense, this implies that the COR theory is a theory of motivation, as individuals are highly motivated for resource conservation due to the resources’ role in attaining and preserving things of central importance. Concurrently, it is also a theory of resilience and vulnerability, centered on how events that threaten or deplete individual resources – thus jeopardizing those things that are of central value – result in strain reactions, and especially so when resources are already scarce.

The motivational- and the resilience mechanisms tied to resource conservation are both relevant for the purpose of the present research. First, to the degree that targets cannot efficiently cope with perceived mistreatment, exposure to bullying behaviors should engender motivation to evade the source of that exposure (e.g., the perpetrator, or the context itself). Specifically, exposure to bullying demonstrably threatens individual resources (Tuckey & Neall, 2014), be it in terms of lowered job security (Glambek et al., 2018), thwarted needs (Trépanier et al., 2015), lowered self-esteem (Iglesias & Vallejo, 2012) or health impairment (Nielsen & Einarsen, 2012). As threats to valued resources entail strong incentives to prevent further loss (Halbesleben et al., 2014), targets should reasonably display elevated motivation to avoid the situation that enables and characterizes those threats. For example, in the event of abusive supervision, employees who experience resource loss in the form of higher strain have been shown to avoid situations involving supervisory feedback, possibly in order to assure reduced exposure to the abusive leader (Whitman et al., 2014). This logic has also proven valid across numerous studies on workplace bullying. Targets exposed to bullying often feel compelled to permanently avoid the workplace as a “last resort” (Rosander et al., 2022), a tendency most often documented using turnover intention as an outcome (e.g. Berthelsen et al., 2011; Djurkovic et al., 2008; Salin & Notelaers, 2017). Additionally, according to the COR theory, resource loss often initiates vicious cycles of further resource depletion (Chen et al., 2015). Hence, the associated avoidance motivation is likely to increase over time, a notion also documented through previous research (Hogh et al., 2011).

Concurrently, as understood from the COR theory’s motivational perspective, the association between exposure to bullying behaviors and turnover intention is not necessarily direct. Specifically, when resource threats engender avoidance motivation, it is likely because the source of that threat is perceived as aversive or otherwise negative (Elliot, 2006). In line with this assumption, the association between aversive working conditions and intention to leave has been found to be mediated through job dissatisfaction (Jin et al., 2018; Rayton & Yalabik, 2014), which is possibly the most immediate outcome of aversive, work-related experiences. This mediation effect has also been found using workplace bullying as a predictor (e.g., Glasø et al., 2011), leading us to assume that the impact of bullying exposure on turnover intention is mediated by job dissatisfaction. The first two hypotheses of this study are thus based on the motivational dimension of the COR theory as well as on a significant body of extant literature. In order to test these hypotheses, we employ a six month time-lag in line with previous research on these associations (e.g., Glambek et al., 2014). This time-lag is equivalent to that commonly used to exemplify the time-aspect in the unfolding of bullying scenarios (Einarsen et al., 2020), and simultaneously allows for development in the outcome measures over and beyond construct stability. In addition, it likely restrains the probability of actual turnover or exclusion from working life before the second measurement (see e.g., Berthelsen et al., 2011).

In line with the theoretical rationale presented thus far, we summarize the first two hypotheses of the current study as follows:

  • H1: Exposure to workplace bullying behaviors will be associated with increased stability-adjusted turnover intention over a six-month time lag.

  • H2: The prospective association between workplace bullying and turnover intention will be mediated through increased stability-adjusted job dissatisfaction.

Second, regarding the resilience- and vulnerability dimension of the COR theory, we propose that the mediation chain suggested in hypothesis 2 is dependent on a particular demographic employment factor. Specifically, based on an integration of tenets from the organizational socialization literature and the COR theory’s proposed resilience mechanism, we hold that exposure to workplace bullying behaviors is more detrimental to targets relatively new at the workplace regarding the work-related outcomes here investigated, and less so to those with a longer length of service.

The moderating role of length of service

Length of service refers to how long one has been employed in one’s organization. In the present study, we construe length of service as a resource that buffers the relationship between exposure to workplace bullying behaviors and work-related outcomes, based on an integration of the COR theory, organizational socialization literature and research on length of service and work attitudes. According to the COR theory, at least two resource conditions must be met in order to alleviate stressor-strain effects (Chen et al., 2015). The first condition is resource availability. Specifically, when resources are threatened or depleted, for example due to exposure to workplace bullying behaviors (see e.g., Tuckey & Neall, 2014), resilience will be higher if the target can draw from relevant resources to aid them in coping with or resolving the situation. Consequently, when resources are available, it is reasonable to assume that a threat such as workplace bullying will be perceived as less overwhelming and less aversive (thus having a weaker impact on job dissatisfaction), and therefore less likely to engender avoidance motivation (thus having a weaker impact on turnover intentions).

Regarding length of service, a large body of research suggests that resources are especially scant in the earliest phases of employment. Organizational entry and short length of service are often characterized by social vulnerability (Saks & Gruman, 2012), and have even been found to denote stressors in and of themselves, imbued with feelings of surprise (Louis, 1980), uncertainty (Mignerey et al., 1995), anxiety (Saks, 1994) and pressures to quickly learn and adapt (Ashforth et al., 2007). Importantly, early socialization is associated with even greater difficulty accommodating if one cannot find and develop social resources in the work environment (Fang et al., 2011). In line with this, coworker undermining appears detrimental to the socialization process of newcomers, and especially so in the earliest phases (Kammeyer-Mueller et al., 2013). Hence, upon entry and until the socialization process has stabilized (see e.g., Ashford et al., 2007), employment is often characterized by attempts to secure resources, rather than to benefit from them. This suggests that short length of service denotes a vulnerability factor in the present rationale, as scantiness of resources will be directly related to lower resilience. In contrast, longer length of service is associated with increased availability of relevant work-related resources such as job security (Hellgren & Sverke, 2003; Kuhnert & Vance, 1992), more central positions in social networks (Rollag, 2004) and more stable levels of work engagement and psychological contracts (Bal et al., 2013). Consequently, longer length of service possibly represents a “caravan” of resources, which, according to the COR theory, denotes a conglomerate of lower-order resources (see Hobfoll, 1998; Wright et al., 2007) associated with increased resilience against work-related stressors (Chen et al., 2015).

Importantly, uncertainty regarding resource availability is also a key characteristic of the socialization process, regardless of the actual resources at the employee’s disposal. Even given equal levels of relevant resources, it is our position that newcomers will retain higher levels of uncertainty regarding what those resources signify and how robust they are, compared with employees with a longer length of service. This notion is related to the second condition for individual resilience in a resource perspective. According to the COR theory, available resources are only relevant as resilience factors insofar as they fit with the demands of the situation (Chen et al., 2015; Hobfoll, 1989). Put differently, the ability to apply resources is imperative to individual resilience. In an organizational frame of reference, this is likely highly dependent on the individual’s level of implicit knowledge and familiarity with the idiosyncrasies of the organizational context. The ability to ensure the best possible fit between resources and demands in working life is, in other words, likely contingent upon context-specific experiences related to ensuring such a fit. Reasonably, short length of service allows both for fewer relevant experiences anchored in the given organizational setting, as well as other sources of such implicit knowledge, including observations of how others might have handled similar situations in the past. Hence, as alluded to earlier, short length of service should be associated with lower confidence regarding which resources are available, relevant and trustworthy, and how they can be mobilized. With longer length of service, on the other hand, the pool of relevant experiences is greater, allowing for better and more precise resource investment, and, consequently, more effective coping, even given equal resource availability.

In line with this reasoning, length of service may be associated both with resource availability and with adaptive ability developing gradually from an initial vulnerability among organizational newcomers into a source of resilience among employees with a long length of service. The linearity of this assumption is theoretically supported by the notion that resources tend to spur further resource gain through so-called gain cycles (Halbesleben et al., 2014). Specifically, from the stage where socialization processes have stabilized, it may become self-reinforcing as a higher-order resource, possibly over fairly long time-lags. This is supported by several strains of research. Gibson and Klein (1970), for example, found that length of service affects work attitudes in a linear fashion up until about 12 years of service, after which the curve flattens. Similarly, work-related attitudes such as intention to leave are found to be far more stable and thus less sensitive to external influence among employees with high tenure (Bal et al., 2013), indicating that length of service denotes a resource condition that may become increasingly significant over several years.

As outlined in our rationale for H1 and H2, we believe that exposure to workplace bullying behaviors affects turnover intention through increased job dissatisfaction, because aversive and resource threatening stimuli spurs avoidance motivation. In line with our conception of length of service as a higher-order resource, we further believe that length of service will act as a moderator of this mediation chain. Specifically, we hold that the relationship between bullying exposure and job dissatisfaction will be stronger for employees with a short length of service, further resulting in higher intention to leave. Conversely, we expect that long length of service comprises a resilience condition that buffers the same relationship, most notably as targets with a long length of service are able to maximize the utility of their resources and to remain confident in how and when to invest them effectively. Hence, we propose the third and final hypothesis of our study as follows:

  • H3: The mediating role of job dissatisfaction in the prospective association between workplace bullying and turnover intention will be moderated by length of service.

The conceptual model outlining these associations is depicted in Fig. 1.

Fig. 1
figure 1

Conceptual model of the relationship between workplace bullying and intention to leave, as mediated through job dissatisfaction and moderated by organizational length of service

Full size image

Method

Sample

The data were collected in 2015 from a random and representative probability sample of Norwegian workers (N = 5000), drawn by Statistics Norway (SSB) from the Norwegian Central Employee Register (the official register of all Norwegian employees, as reported by employers). Sampling criteria were adults aged between 18 and 60 years employed in a Norwegian enterprise. Questionnaires were distributed through the Norwegian Postal Service at both baseline (T1) and at follow-up six months later (T2). At T1, 32% of the sample returned the questionnaires (N = 1608). Only T1-respondents were invited to participate at T2, with 1003 respondents returning the questionnaires at the second measurement, yielding a 62% response rate. Attrition analyses indicated that attrition was random and that the T2 sample was representative of the T1 sample.

Instruments

Exposure to workplace bullying behaviors

We measured exposure to workplace bullying behaviors using the nine-item version of the Negative Acts Questionnaire Revised (S-NAQ; Notelaers et al., 2018), a standardized and validated short version of the commonly used NAQ-R measure of workplace bullying (Einarsen et al., 2009). Participants were presented with a list of nine common bullying behaviors and were asked to indicate the degree to which they had been exposed to each behavior during the preceding six months, using a five-point scale ranging from 1 (never) to 5 (daily). The internal stability of this measure was satisfactory at T1 (α = .86) and T2 (α = .87).

Turnover intention

We measured turnover intention using a scale of three items (Sjöberg & Sverke, 2000). Example items include “I am actively searching for a new job” and “If I could choose freely, I would quit this job”. All items were scored by respondents on a five-point Likert-type scale ranging from 1 (fully disagree) to 5 (fully agree). The internal stability of this measure was satisfactory at both T1 (α = .88) and T2 (α = .89).

Job dissatisfaction

We measured job dissatisfaction using a scale of four items developed by Hetland et al. (2008) based on the job satisfaction index of Brayfield and Rothe (1951). Item examples include “I am quite happy with my current job” (reverse-coded) and “I think my job is quite unpleasant”. All items were scored by respondents on a five-point Likert-type scale ranging from 1 (fully disagree) to 5 (fully agree). The internal stability of this measure was satisfactory at both T1 (α = .86) and T2 (α = .87).

Length of service

We measured length of service at T1 as a continuous variable by asking respondents to indicate how many years they had been employed in their current workplace. For the hypothesis test concerning length of service, simple slopes were defined by the 16th, 50th and 84th percentiles, representing short (approximately 2.5 years), average (approximately 8.5 years) and long (approximately 22 years) length of service, respectively.

Control variables

Age and leader responsibility were used as control variables. Age was measured as a continuous variable at T1, and was employed as a control variable to ensure that any moderating effect of length of service could not be ascribed to the respondents’ age at T1. Leader responsibility was measured as a dichotomous variable by asking respondents to indicate whether their position involved leader responsibility at T1 (coded 1) or not (coded 0), and was included as a control variable to ensure that any moderating effect of length of service could not be ascribed to formal power positions.

Analyses

Statistical procedures were carried out using the statistical package IBM SPSS version 27 and version 3.4 of the PROCESS macro supplement (Hayes, 2018). First, scale and descriptive statistics were produced (see Table 1). Hypothesis 1 was subsequently tested using linear regression analyses with pairwise deletion. Age and T1-levels of turnover intentions were entered in the first step, and exposure to workplace bullying behaviors was entered in the second step. Then, to address the possibility of a reverse causation effect, turnover intentions at T1 was tested as a predictor of stability-adjusted exposure to workplace bullying behaviors at T2. Hypotheses 2 and 3 were tested simultaneously using model 7 in the PROCESS macro SPSS supplement. Specifically, exposure to workplace bullying behaviors was entered as a predictor of T2-levels of turnover intentions, with job dissatisfaction at T2 entered as a mediator, and length of service entered as a moderator of the relationship between workplace bullying and T2 job dissatisfaction, controlling for age, leader responsibility, and T1-levels of turnover intentions and job dissatisfaction. Additionally, in order to ascertain the value of the moderator where the relationship between exposure to workplace bullying behaviors and job dissatisfaction becomes non-significant in the present moderated mediation model, we used the Johnson-Neyman technique for establishing significance regions (Hayes, 2018). Moderated mediation analyses were conducted using heteroscedasticity-consistent standard error- and covariance matrix estimators with 5000 bootstrap resamples and 95% bias-corrected confidence intervals for the indirect effect at different levels of the moderator.

Table 1 Descriptive and scale statistics and bivariate correlations for variables included in hypotheses tests
Full size table

Results

Hypothesis 1 was tested using a linear regression analysis, with age and T1-levels of turnover intention entered in the first step, and exposure to bullying behaviors at T1 entered in the second. The variance explained by the full model was 41% (F (3, 972) = 227.51, p < .001), and the inclusion of exposure to bullying behaviors at T1 in Block 2 contributed significantly to turnover intentions at T2, R2 change = .01, F change (1, 972) = 18.81, p < .001. Hence, H1 was supported. Reverse causation was addressed using a linear regression analysis. The results showed that turnover intentions at T1 could not predict stability-adjusted exposure to bullying behaviors at T2 (R2 change = .00, F change (1, 969 = 1.2, β = .03, p = .27), in line with previous studies (e.g., Glambek et al., 2014).

Hypotheses 2 and 3 were tested simultaneously using model 7 in Hayes’ (2018) PROCESS macro supplement to the SPSS package. Specifically, exposure to bullying behaviors was employed as a predictor of stability-adjusted turnover intentions at T2, mediated through stability-adjusted job dissatisfaction at T2 as moderated by length of service, with age and leader responsibility as additional covariates. As predicted, exposure to bullying behaviors at T1 was associated with higher job dissatisfaction at T2 (B = .35, p < .001), which further predicted increased turnover intention at T2 (B = .87, p < .001) such that the direct effect of exposure to bullying behaviors at T1 on turnover intentions at T2 was no longer significant (B = .11, p = .172). This mediation effect was, however, dependent on the target’s length of service, which moderated the effect of exposure to bullying behaviors at T1 on job dissatisfaction at T2 (B = −.02, p = .021) in the mediation chain. As depicted in Fig. 2, this interaction indicates that exposure to bullying behaviors at T1 predicts stability-adjusted turnover intentions through stability-adjusted job dissatisfaction at T2, yet only for respondents with short or average length of service (16th and 50th percentiles). Moreover, a pairwise contrast analysis revealed that all simple slopes were significantly different from each other, as detailed in Table 2. We thus conclude that the effect was strongest for employees with the shortest length of service. Additionally, using the Johnson-Neyman technique for ascertaining significance regions, we found that the moderation effect becomes non-significant at the 69th percentile (14.6 years of length of service) in the present sample, indicating that the development of length of service as a resilience condition is a gradual and long-range process. The results further showed that the index of moderated mediation was statistically significant, Index = −.01, SE (Boot) = .01, Boot CI = −.03 – - .002. Hence, H2 and H3 were supported (see Table 2 for additional details and overview).

Fig. 2
figure 2

Interaction between exposure to workplace bullying behaviors and length of service at T1 on stability-adjusted job dissatisfaction at T2

Full size image
Table 2 Results of moderated mediation analysis with exposure to bullying behaviors as a predictor
Full size table

Discussion

The present study found that length of service moderates the mediating role of stability-adjusted job dissatisfaction in the relationship between exposure to workplace bullying behaviors and a subsequent increase in turnover intention. More specifically, the effect of bullying exposure on job dissatisfaction was strongest for employees with short length of service, thus spurring higher levels of intentions to leave. Furthermore, while the mediation chain was significant, albeit weaker, among employees with average length of service, it was non-significant among those with long length of service, specifically from 14.6 years and onwards (69th percentile). Short length of service thus seems to represent a significant vulnerability factor regarding the effects of bullying on job dissatisfaction in a manner that further intensifies its effect on intention to leave the organization. Conversely, long length of service seems to denote a gradually developing resilience factor, ultimately nullifying the generally robust association between bullying and intention to leave. Importantly, this moderation effect was independent of related variables such as age and leadership positions.

Our findings are in line both with the COR theory’s tenets (e.g., Hobfoll, 1989; Hobfoll et al., 2018) and with literature on organizational socialization (e.g., Ashforth et al., 2007; Louis, 1980; Nelson, 1987). First, the COR theory suggests that individuals are motivated to avoid resource threats that are impervious to individual coping attempts, not least when the source of that threat is perceived as aversive. As one of the most detrimental psychosocial hazards an employee may face (Hauge et al., 2010), in part due to the targets’ difficulty in defending against the mistreatment (Einarsen et al., 2020), exposure to bullying behaviors is hence a plausible precursor of turnover intention as mediated through job dissatisfaction (see also Glasø et al., 2011). Second, and most seminal to the contribution of our study, the COR theory’s tenets imply that targets’ resilience and vulnerability is dependent both on individual resource availability and the fit of those resources with the demands that the bullying situation represents (see Chen et al., 2015). In light of findings from research on organizational socialization, it is fair to assume that employees with a short length of service possess relatively fewer resources and simultaneously remain less able to mobilize the resources they do possess. Hence, theoretically, it is plausible that employees with a relatively short length of service are more vulnerable to detrimental, work-related outcomes of negative psychosocial experiences (Kammeyer-Mueller et al., 2013), such as exposure to bullying behaviors. For example, relatively new employees are generally in greater need of attaining, developing and maintaining social resources in the work environment as part of their socialization process (Fang et al., 2011), which is likely already to be imbued with significant uncertainty (Mignerey et al., 1995; Saks, 1994), regardless of interpersonal problems. With longer length of service, on the other hand, this social vulnerability is likely to wane, not only due to higher resource availability, but also due to a likely increased ability to ensure the best possible fit between available resources and the demands of the situation. When exposed to workplace bullying behaviors, relatively new employees may for example become uncertain about the value and robustness of the resources they do in fact possess, as well as how and when they should and could be mobilized. Employees with a long length of service, on the other hand, retain a broader range of experiences with investing resources to meet demands and problems in their particular organizational setting. Hence, length of service may, at least in general, be construed as a higher-order resource that accumulates over and beyond the reduction of the vulnerability associated with newcomer status. This interpretation seems likely not least in light of the significance regions established via the Johnson-Neyman technique in the present study, which showed that the point where the workplace bullying-slope is no longer a significant predictor of job dissatisfaction within the moderated mediation model is at 14.6 years of service in the organization. This implies that the relevancy of length of service as a moderator does not cease immediately following the vulnerability of the initial socialization phases, even though the slope was steepest earliest on, but rather that length of service continues to develop as a higher-order resource far beyond this point.

Having controlled for leader positions and age, the moderated mediation effect shown in our study is not attributable to formal power positions or traits that co-vary directly with older age. Importantly, however, any resilience associated with longer length of service cannot alter the destructive nature of workplace bullying in itself, even if prolonging the path towards turnover. This study has not addressed the psychological toll of bullying across length of service levels, and little is known about the long-term effects of workplace bullying among those who choose to stay because they hold the resources enabling them to do so. However, what may be ascertained from our study is that when bullying occurs, organizations risk quickly losing organizational newcomers before having had a chance to handle the problem.

The interaction between exposure to workplace bullying behaviors and length of service may still have alternative explanations not accounted for in our study. For example, it is possible that employees with long length of service in general hold certain attitudes that may account for the findings. For example, commitment to the organization possibly increases with length of service, as may organizational identification, perhaps heightening the threshold for turnover. Concurrently, while these and similar attitudinal constructs may be plausible moderators of turnover intention, it is less certain that job dissatisfaction would be affected to the same degree. In the case of higher commitment and a more salient organizational identification, for example, it could also be that targets feel that there is more at stake (implying that bullying exposure represents a greater resource threat), and possibly, that the exposure is more personal as they are more identity invested. Further research may possibly provide some answers regarding these possibilities, for example by replicating the present findings while controlling for additional attitudinal constructs.

It is worthwhile noticing that the moderation effect reported in our study is not strong in terms of its ability to explain stability-adjusted levels of turnover intention. Hence, in a probability sample such as that employed in the present study, several alternative sources of turnover intentions are present. At the same time, workplace bullying exposure is a fairly low-frequent phenomenon, at least in Norwegian working life. Hence, for the targets of bullying exposure in isolation, the effect may still be considerable. The finding that long length of service eliminates the well-established relationship between bullying exposure and both of the attitudinal outcomes investigated in our study would, at least, indicate that. Hence, in populations with higher prevalence of workplace bullying, stronger effect sizes could be more likely. Moreover, despite our rationale for employing a six-month time-lag between the predictor and the outcomes, it is possible that job dissatisfaction and intention to leave may keep developing beyond this time-lag, at least for targets who do not act upon their intention to quit. Additionally, a possible source of bias to be noted is that exposure to workplace bullying behaviors may be part of a longer victimization process, which would not be accounted for in our analyses. Specifically, for targets with a short length of service, long-term exposure is not possible, while employees with a longer length of service may, as far as we know, be targets of recurrent or longer-lasting victimization. If so, it could imply an underestimation of the true moderation effect size.

As this is one of the first studies to address the role of length of service as a resource in the workplace bullying literature, several questions remain unanswered. As alluded to earlier, we still know little about whether length of service buffers against other known outcomes of bullying and related phenomena, such as health impairment and employment security. The question of actual turnover also remains, and to the degree that length of service acts as a buffer against separation from the organization, the long-term consequences of staying is yet a possible venue for further research. In these and related senses, initiatives to investigate length of service as a moderator would be an interesting and practically relevant future endeavor, in the workplace bullying literature, in the organizational stressor–strain literature more generally, and in the literature on organizational socialization.

Methodological considerations

Among the strengths of the present study is the use of prospective data, allowing for causal interpretations with respect to an explanatory model regarding workplace bullying and turnover intentions. While the main association between bullying and turnover intention has been established in previous studies, there is a shortage of prospective studies on potential explanatory mechanisms. Hence, from our results, we may infer not only the strength of the association between bullying and turnover intention, but also how and when bullying may heighten such intentions over time, over and above those turnover intentions present at T1. As the main relationship investigated in this study is held to develop over short time spans, we would also hold the relatively short six-month time lag as a methodological strength. Together with our exclusive use of validated scales for all psychological constructs and a relatively large and representative sample, the present results are likely reliable and generalizable to working populations outside the Norwegian workforce, such as in European, North American, and Australasian countries.

With respect to limitations, we would highlight the use of single-source data, as there is always a danger of common-method variance when respondents provide all information individually (Podsakoff & Organ, 1986). Yet, length of service represents a rather objective measure even when using self-report data. Furthermore, the research model tested in this study may not be particularly well-suited for cross-level analyses, as only individual-level constructs are included. In addition, the use of a cross-lagged design with stability-adjusted outcomes is generally regarded as significantly more reliable with respect to avoiding common-method biases than are cross-sectional data (Podsakoff et al., 2003). Finally, at the time of writing, labor markets worldwide are facing new and unpredicted uncertainty spurred by a global pandemic. While psychological and interpersonal phenomena measured at the individual level may not be significantly altered by this development or potentially other future, unforeseen sources of global uncertainty, the labor market in which they occur may be affected. Hence, readers should be aware that the present data were collected before the pandemic outbreak, and therefore, caution should be exercised upon large-scale generalizations before markets have re-stabilized.

Theoretical implications

From a theoretical viewpoint, the present study extends the literature on bullying, turnover in organizations and organizational socialization. First, a large body of empirical work presently exists with regard to outcomes of workplace bullying, yet only a small portion of it addresses the mechanisms that might explain such relationships (Rai & Agarwal, 2018b). As a contribution to the literature in that regard, the present study offers a theory-based, prospective conceptual model shedding new light on the interrelationships between workplace bullying, job dissatisfaction, and intention to leave. Second, and relatedly, the present study highlights length of service as a resilience-related, higher-order resource within this conceptual model. Scholars suggest that demographics, often used as control variables in analyses of workplace bullying and turnover intention, including length of service, should be more thoroughly tested as explanatory factors (Coetzee & van Dyk, 2018). In line with this, the reasoning put forth in the present study suggests that the tenets of the COR theory may be integrated with organizational socialization literature to construe length of service as a higher-order resource, at least when comparing newcomers to employees with a longer employment relationship. An important aspect of this reasoning is that length of service as a continuous variable may denote a vulnerability condition imbued with the uncertainty of initial socialization in the lower end of the scale, and a resilience condition associated with resource-related gain cycles in the other. The demonstration that the COR theory’s tenets and key findings from organizational socialization literature may be complimentary in predicting work-related outcomes of bullying exposure thereby represents yet a possible theoretical implication of this work.

Practical implications

From a practical viewpoint, the results of our study first underscore the importance of managing commencing workplace bullying incidences swiftly. Among targets, job dissatisfaction escalates in a matter of months among those exposed, and also denotes a key mechanism in the relationship between bullying and subsequent turnover intention. Ensuring fair and resolute intervention early on may help to restore job satisfaction among employees exposed to workplace bullying behaviors, thereby preventing the unnecessary cost – for targets and organizations alike – of undue turnover. There is, however, a need to be particularly attentive to turnover intention when ill-treatment befalls relatively new employees, with the strongest escalation in turnover intention seen among the more recent organizational members. Possibly, job dissatisfaction and turnover intention may originate long before the problem has been handled, or before it is even recognized as bullying. In that case, at least two adverse organizational outcomes may ensue. First, organizations risk losing new employees to turnover, with any cost and detriment associated with turnover in itself. Second, there is a risk of creating work environments where bullying keeps re-occurring, or become embedded in the organizational culture as the root causes of the problem is disguised as a turnover problem and thus remains unhandled. This possibility is in line with the COR-theory-based crossover model (Westman, 2001), which posits that stressors and strains may be transferred between individuals, implying that organizational anti-social behaviors can become mutually reinforcing if not addressed (Hobfoll et al., 2018). It is also in line with empirical accounts of workplace mistreatment “regeneration” (see e.g. Lutgen-Sandvik, 2003), which, in light of the present results, may be prevented by taking length of service into account. For example, managers and organizational practitioners could assure resource availability among relatively new employees. Individual consideration is necessary, as new organizational members may vary greatly with respect to resource needs. However, whether focused on social thriving, mentorship with respect to job tasks, perceptions of organizational voice and justice processes or other important resources, practitioners should remain attentive to the relatively more vulnerable position of employees with short length of service. Additionally, as suggested in the present paper, practitioners should be attentive to the possibility that employees with a short length of service not only display lower levels of resource availability, but also retain a poorer ability to recognize and successfully invest resources when and where they are needed. As the vulnerability of early socialization is tied in part to the process of securing resources rather than benefiting from them, a higher level of uncertainty regarding the resources’ worth, robustness and durability is likely among the newest members of the organization. To that end, providing information about formalized ethical infrastructures (e.g., Einarsen et al., 2019) such as conflict management procedures and complaint systems may be an effective remedy against possible perplexity and coping inability of bullying targets, and possibly mostly so among targets with a short length of service. Similarly, informal ethical infrastructures such as a strong conflict management climate and norms against workplace mistreatment (Einarsen et al., 2016; Glambek et al., 2020) should not only be pursued, but also effectively communicated to relatively new employees.

Conclusion

The present study has applied the COR theory framework to predict and test a resource-based rationale for the relationship between workplace bullying and turnover intention. Specifically, we have suggested that targets of workplace bullying behaviors, due to the aversive and resource-depleting nature of the phenomenon, are strongly motivated to withdraw from and permanently avoid the workplace, to prevent further loss of resources. As supported by the present findings, this avoidance motivation manifests as an increased intention to leave over time, a tendency further demonstrated to be mediated through job dissatisfaction, again in line with the tenets of the COR theory. The mediating role of job dissatisfaction, though, has also been shown to be dependent on the target’s length of service, with the strongest association prevailing among relatively new employees, and long service acting as a buffer that nullifies the impact of bullying on job dissatisfaction and turnover intention. In sum, during the workplace bullying process, organizations may particularly risk losing newcomers to turnover, before the problem has been recognized and addressed, potentially with significant implications for organizational productivity, efficiency, and competitive ability.