Sensitized or desensitized to bullying? How hostile work climates shape target identification and outcome

The experience of workplace bullying is largely subjective, and what is perceived as bullying may therefore vary from one person to another depending on their individual characteristics and previous experiences. Exposure to bullying behaviours at work may also be interpreted differently depending on their context, such as the work climate in which they occur and broader workplace conditions (Rosander and Nielsen, 2026c; Rosander and Salin, 2023). Although previous studies have established associations between both supportive climates (Dollard et al., 2017) and hostile work climates (Rosander and Salin, 2023; Zahlquist et al., 2023) and bullying, no study has examined whether the work climate itself shifts the threshold at which targets identify as bullied. Nor has previous research examined whether such internalization affects outcomes. The present study addresses this gap by investigating whether perceptions of bullying are construed and experienced differently across workplace contexts – from relatively benign to more hostile work environments. As a second objective, we examine how acknowledging that one’s negative experiences constitute bullying may affect mental health outcomes.

Workplace bullying is defined as systematic, prolonged exposure to negative treatment from others in the workplace (Einarsen et al., 2020). This negative treatment can take many forms, from passive behaviours, such as withholding action when it would normally be expected, to active expressions of mistreatment, whether direct or indirect – all referred to as bullying behaviours. An important aspect of the concept involves a psychological power imbalance, that is, when a target finds it difficult to ward off or stop the mistreatment. The imbalance may be present from the start, such as when the perpetrator holds formal power, for example through abusive supervision (Tepper et al., 2017), or it may develop out of a conflict initially involving two equal parties where the informal power slowly shifts in favour of one party (Einarsen, 1999). Regardless of how it arises, the perception of being unable to defend against the exposure marks the transition to a second phase where a target comes to perceive themself as a victim of bullying (Nielsen et al., 2026).

While it has been suggested that this identification as a victim – also referred to as self-labelling or victimization – may involve a form of internalization of the treatment (Rosander and Blomberg, 2019), little is known about the factors that cause and influence it. However, in the literature on workplace bullying, victimization is assumed to be explained either by the personality characteristics of those exposed or by factors in the working environment. Meta-analytic evidence suggest that the strongest associations are found for the latter (Nielsen et al., 2017; Bowling and Beehr, 2006). Recent longitudinal research has further highlighted hostile work climate as a mechanism linking organizational stressors, such as role ambiguity, to workplace bullying (Blomberg et al., 2024). As for specific working conditions that are likely to influence the process leading up to victimization, it has been suggested that the social climate plays an important role in the occurrence of bullying (Einarsen et al., 1994). Already in his seminal work on bullying, Brodsky (1976) emphasized that, to reduce its occurrence, it is important to establish “a culture that exposes, discourages and censures harassment”, that is, formally condemning and sanctioning such behaviour. One type of work climate associated with an increased risk of bullying is commonly referred to as a hostile work climate (Blomberg et al., 2024; Rosander and Salin, 2023; Stapinski and Gamian-Wilk, 2023), defined as an affective organizational climate characterized by distrust, suspicion and antagonism (Mawritz et al., 2014). Rather than reflecting the negative behaviours of a few individuals, it denotes a general tendency among employees to interact in a more antisocial manner. It is plausible, and can be argued, that working in such a climate may make employees either more sensitized or desensitized to negative behaviours that could amount to bullying. Furthermore, being exposed to such behaviours in a hostile climate could amplify or attenuate the outcomes. The present study investigates these two opposing perspectives. Below, we outline why there may be a sensitizing or desensitizing process in hostile climates, and whether the outcomes may be amplified or attenuated.

In workplaces marked by a general level of hostility, individuals may be more sensitive to negative behaviours and quicker to interpret them as bullying. According to Lazarus and Folkman’s (1984) transactional stress theory, the impact of potentially stressful experiences depends on how they are cognitively appraised. In hostile work climates, negative behaviours are more likely to be appraised as threatening, as the broader environment amplifies their salience and significance. Primary appraisal thus could become sensitized, with employees more readily interpreting negative acts as harmful. At the same time, secondary appraisal may reinforce this process, because hostile climates typically limit perceived coping resources, making it more difficult to manage such behaviours. Based on the conservation of resources theory (Hobfoll, 1989), resources in hostile work environments may already be strained, and individuals must remain vigilant and invest energy to prevent additional resource loss. This heightened state of alertness, aimed at avoiding full depletion, can itself be taxing and could make employees more sensitive to negative acts. Together, this suggests a sensitizing process in which negative behaviours are more likely to be identified as bullying in hostile environments.

To fully understand these dynamics, it is also necessary to consider how employees regulate their emotional responses to persistent hostility, as such adaptation may shape how behaviours are experienced and interpreted. Ashforth and Humphrey (1995) identified four general strategies – neutralizing, buffering, prescribing and normalizing – the latter being most relevant to exposure to bullying. Normalization of emotions in organizations may involve reframing the reason for their occurrence (Ashforth and Kreiner, 2002), for example by interpreting the behaviour as a typical or normal way of dealing with a particular issue at the workplace. Normalization can also result from adaptation to the situation. Ashforth and Kreiner described this as psychological processes through which reactions to stimuli are reduced – either through habituation or desensitization. In a hostile environment, repeated exposure to a general level of negative behaviours can lead to habituation that diminishes the cognitive shock of new incidents. In such an environment, a certain level of negative behaviours may be anticipated. Desensitization occurs when perceptions match expectations, resulting in little or no emotional response. Normalization may be understood as a form of emotion-focused coping (Lazarus and Folkman, 1984), where the appraisal of the situation is altered without changing the objective conditions. They referred to such cognitive manoeuvres as cognitive reappraisal, regardless of whether the altered construal reflects a realistic interpretation of cues or a distortion of reality. Recent research further suggests that hostile work climates shape how employees cognitively process and make sense of bullying-related experiences (Rosander and Nielsen, 2026b). The paradox is that while normalization of negative behaviours in a hostile work environment may help individuals avoid being overwhelmed, the very same process can entrench circumstances that might otherwise be open to change.

While there is limited research specifically addressing the normalization of bullying behaviours, findings from studies on workplace violence suggest that repeated exposure to violent acts or hostile attitudes can lead to their gradual acceptance as normal, routine or even expected within a particular organizational culture or context. For example, Andersson and Överlien (2018) showed how the “normalization of violence” can serve as a psychological coping mechanism, whereby violence is reframed as an ordinary aspect of the job. Over time, this normalization becomes embedded in workplace norms and values, ultimately leading to desensitization and diminished recognition of its harmful impact.

Taken together, the preceding arguments point to the possibility of both a sensitizing and a desensitizing process. As for a sensitizing process, behaviours in a hostile environment, whether directed at oneself or occurring more generally, may be appraised as harmful, as some damage may already have been sustained simply by being in such a setting (Lazarus and Folkman, 1984). This can heighten the anticipation of further threats and foster a state of increased readiness to act and react, thereby making it more likely that negative behaviours directed at oneself are perceived as bullying. Thus, we propose:

However, in hostile environments, it is also possible that negative behaviours are perceived as part of everyday life and therefore less likely to be labelled as bullying as a result of normalization (Ashforth and Kreiner, 2002) or cognitive reappraisal (Lazarus and Folkman, 1984). Thus, as an alternative and contrasting hypothesis, we propose:

Above, we have explained how a hostile work climate may influence the risk of victimization following exposure to negative behaviours from others in the workplace. However, a hostile work climate may also influence the magnitude of health outcomes following exposure to bullying. According to conservation of resources theory (Hobfoll, 1989), stress arises when individuals lose valued resources or when such resources are threatened. Hostile work climates can be seen as environments already characterized by depleted resources, where coping capacities and social support are limited. In such contexts, additional exposure to negative acts could be more damaging, as it triggers further resource loss and restricts recovery. Through this amplifying process, the association between exposure to negative acts and subsequent mental health problems is expected to be stronger in hostile climates. Furthermore, according to the cognitive activation theory of stress (Ursin and Eriksen, 2004), the stress response functions as a general alarm triggered by a discrepancy between what should be and what is. Such activation is not inherently harmful, but if prolonged it contributes to allostatic load, increasing the risk of illness and disease (Guidi et al., 2021). Crucially, the impact of stress depends on response outcome expectancies. When individuals perceive that coping responses will be effective, the stress response can be adaptive. In hostile climates, however, coping options are often constrained, and negative or absent outcome expectancies may dominate. This heightens the likelihood of helplessness and hopelessness, thereby increasing the risk of adverse mental health consequences from exposure to negative behaviours.

While conservation of resources theory (Hobfoll, 1989) emphasizes amplification of strain through resource loss spirals, other perspectives highlight potential attenuation in hostile climates. Repeated exposure to negative behaviours may, as noted above, over time result in habituation or desensitization, where the match between anticipated and actual experiences reduces emotional arousal (Ashforth and Kreiner, 2002). From this perspective, negative acts in hostile environments may be appraised as less novel or threatening, thereby diminishing their immediate psychological impact. Stress inoculation research provides another mechanism. Moderate levels of prior adversity – in this case being in a hostile environment over time – may build resilience by fostering coping resources and expectations that subsequent stressors can be endured (Seery et al., 2010). In addition, normalization of negative behaviours may help employees avoid being overwhelmed by reframing such experiences as a typical part of work life, although this normalization may also contribute to maintaining a hostile work climate. Taken together, these mechanisms suggest that in hostile work environments, the association between negative acts and subsequent mental health problems may be weaker than expected. However, research on workplace bullying has consistently shown that the long-term effects eventually accumulate (Boudrias et al., 2021), and that continued exposure over time is particularly harmful once individuals begin to perceive themselves as bullied.

To sum up, working in a hostile environment may involve an initial state of strained or depleted resources (Hobfoll, 1989), which could magnify the stressfulness of exposure to negative acts and contribute to allostatic load (Ursin and Eriksen, 2004), thereby making their health consequences more severe. Thus, we propose:

However, experiencing, and observing others being exposed to, varying degrees of negative behaviours in a climate marked by suspicion and animosity may to some extent prepare employees for more systematic mistreatment (Seery et al., 2010). At least initially, the consequences may therefore be less severe than if the exposure occurred suddenly and without warning in an otherwise benign environment. Thus, as an alternative hypothesis, we propose:

Learned helplessness theory (Abramson et al., 1978; Seligman, 1975) proposes that perceiving oneself as unable to control or escape aversive experiences leads to passivity and psychological harm. Central to the theory is the concept of noncontingency, the perception that one’s actions have no influence on outcomes. Repeated experiences of noncontingency create an expectation of uncontrollability, which may generalize across situations and develop into a state of global helplessness (Abramson et al., 1978).

Applied to bullying, the repeated exposure to negative behaviours – and the increasing difficulty in warding them off, a defining feature of bullying (Einarsen et al., 2020) – may foster feelings of being unable to manage the situation. At the point where individuals self-label as bullied, negative acts are internalized as personal victimization. This may reinforce perceptions of uncontrollability and helplessness, and thereby constitutes a key mechanism in explaining why bullying is so detrimental to health. While exposure to negative behaviours may exert some direct effect on mental health, we propose that it is the internalization of being a victim, together with the sense of helplessness, that explains the severe psychological outcomes consistently reported in previous research (Boudrias et al., 2021). This assumption is in line with previous longitudinal research which has shown that the association between exposure to negative behaviours and mental health problems is explained by victimization (Nielsen et al., 2012). Thus, we hypothesize that the main pathway from exposure to bullying behaviours to mental health problems runs through the identification of oneself as a victim of bullying:

A direct effect of exposure on mental health may be especially evident when the treatment is unexpected. This may be particularly likely in work environments not marked by general hostility. Such experiences may shatter assumptions of the world as benevolent and meaningful, thereby increasing the risk of mental health problems (Janoff-Bulman, 1992). In hostile climates, however, exposure to negative acts may be less surprising, as similar behaviours are perceived as part of everyday interactions. In such contexts, victimization may become a particularly central pathway through which exposure translates into mental health problems. Possibilities for perceived control or coping may already be heavily reduced, making helplessness a probable response. Thus, we hypothesize:

This study draws on a prospective probability sample representing the Swedish workforce, excluding employees younger than 18 years and those working in establishments with fewer than ten staff members. Sampling and data collection were carried out by Statistics Sweden (the national statistical agency). The first wave of data was gathered during a two-month period in autumn 2024 (n = 3,307), with the second wave conducted in spring 2025, approximately seven months later. In total, 2,024 individuals participated in both waves. For each wave, all sampled individuals with a digital mailbox received information about the study and details required for informed consent electronically, whereas those without a digital mailbox were contacted by post. Up to three reminders were issued, with the second reminder always sent by post to those who had not yet responded. Following the completion of data collection, Statistics Sweden added register information from the Swedish National Population Register to the data set (e.g. sex, age and educational level). Researchers were granted access only after anonymization, which further safeguarded the ethical soundness of the study. Ethical approval was obtained from the Swedish Ethical Review Authority (Protocol No. 2023–06603–01).

Of the final sample, 58% were women and 42% men. The mean age was 49.44 years (SD = 10.50). Participants had, on average, worked 10.11 years (SD = 9.62) at their current workplace, and the vast majority (96%) held permanent employment. With regard to education, 16% had completed only compulsory school, 33% had finished upper secondary education and 51% had obtained a university degree.

Workplace bullying was measured with the Negative Acts Questionnaire–Revised (NAQ–R; Einarsen et al., 2009), which has been validated in a Swedish setting (Rosander et al., 2024). The scale comprises 22 items capturing a range of unwanted and hostile behaviours that employees may encounter at work. When such behaviours – referred to as bullying behaviours – are experienced repeatedly and systematically over time, they may amount to bullying. Participants responded on a five-point scale indicating how often each behaviour occurred, from never to daily. Internal consistency was high (Cronbach’s α = 0.89 at T1).

Victimization was assessed with a single item following a standard definition (Einarsen and Skogstad, 1996). Respondents were first presented with the text: “Bullying occurs when a person, repeatedly and over time, is subjected to negative treatment by one or more individuals, in situations where the person has difficulty defending themselves. A conflict between two equally strong individuals is not considered bullying”. They were then asked: “How often have you been subjected to bullying at your workplace during the past six months?” Responses were given on the same five-point frequency scale as for the NAQ–R, but were recoded into a binary variable representing self-labelling as bullied to at least some degree. This recoding reflected our theoretical focus on the transition from not identifying as bullied to beginning to perceive oneself as a victim. The acknowledgement that the definition applies to one’s own situation at least to some extent marks a qualitative shift in how the situation is construed and interpreted.

Hostile work climate was measured using five items capturing the overall social atmosphere at the workplace (HWC; Rosander and Salin, 2023). The scale reflects perceptions of suspicion, conflict, misunderstandings and mistreatment among co-workers. To capture the construct comprehensively, the scale also includes two reverse-coded items assessing feelings of safety and a positive atmosphere. As such, the measure represents psychological climate, that is, the individual’s perception of the work environment rather than an aggregated group score (James and James, 1989; Kuenzi and Schminke, 2009). This is particularly relevant in the context of workplace bullying, where perceptions of the same environment can differ sharply between targets, witnesses and non-witnesses. Aggregating across individuals would risk producing an average climate score that represents none of these groups (Rosander and Nielsen, 2024). Responses were given on a seven-point Likert scale, and internal consistency was 0.88 at T1.

Mental health problems were measured using the Hospital Anxiety and Depression Scale (HADS; Zigmond and Snaith, 1983). The instrument includes 14 items – seven assessing anxiety symptoms and seven assessing depression symptoms – referring to experiences during the past week. Each item is rated on a four-point scale (0–3). An example anxiety item is “I feel tense or ‘wound up’”, with response options ranging from not at all to most of the time. A corresponding example for depression is “I have lost interest in my appearance”. Analyses yielded very similar results when anxiety and depression symptoms were examined separately, therefore the full scale was used in the study. Internal consistency was 0.91 at both T1 and T2.

Perceived helplessness was measured with a single item, answered only by those who had indicated that they were exposed to negative behaviours to at least some degree: “No matter what I do—nothing will ever change”. The item was presented after questions about how respondents handled the negative treatment and whether they perceived their strategy as effective. Responses were given on a seven-point Likert scale.

Analyses were conducted in Stata version 19.5. To examine whether the NAQ–R and the HWC should be considered two distinct constructs, structural equation modelling was used. Model fit was evaluated with the root mean square error of approximation (RMSEA), the comparative fit index (CFI), the Tucker–Lewis index (TLI) and the standardized root mean squared residual (SRMR). RMSEA values below 0.05 were considered indicative of good fit and values below 0.08 of acceptable fit, whereas CFI and TLI values close to 0.95 and SRMR values below 0.08 were considered indicative of good fit (Hu and Bentler, 1998).

To test the hypotheses, regression and structural equation modelling approaches were applied. Logistic regression was used to examine the association between exposure to negative acts and subsequent victimization (H1a/b). Both main effects and their interaction were included in the model. Because interaction effects in logistic regression are not straightforward to interpret from coefficients alone, predicted probabilities were plotted to illustrate the nature of the interaction.

Linear regression (moderation) was used to investigate whether exposure to negative acts predicted later mental health problems, and whether this effect depended on the level of hostile work climate (H2a/b). To probe significant interactions, simple slopes were calculated and Johnson–Neyman intervals (Hayes, 2022) were used to identify the range of values for which the effect of exposure was statistically significant.

A mediation analysis was conducted using generalized structural equation modelling (gsem) to assess whether victimization mediated the link between exposure to negative acts and subsequent mental health problems (H3a). To test H3b, the same model was extended with an interaction term to examine whether the indirect effect varied as a function of hostile work climate. We used gsem as it allows a binary mediator (victimization) to be combined with a continuous outcome (mental health problems) in a single model. As an additional test of the theoretical assumptions underlying the mediation model, we compared levels of perceived helplessness between individuals who self-labelled as bullied and those who did not. For H3b, we also computed the index of moderated mediation (Hayes, 2022) to formally assess whether the indirect effect significantly differed across levels of the moderator.

All models were initially estimated with sex, age and managerial position as control variables. Age was retained in models predicting mental health problems, where it reached significance, whereas in all other models, none of the control variables were significant and they were therefore excluded from the final models reported in the Results, in line with the recommendations by Cohen et al. (2003). In addition, baseline measures of all T2 variables were included in the corresponding models to control for their initial levels, thereby estimating change over time rather than static associations.

All analyses were conducted with 5,000 bootstrap replications and percentile bootstrap confidence intervals (BootCI; Hayes, 2022) are reported throughout. As robustness checks, bias-corrected (BC) and normal-based intervals were also examined. In addition, heteroskedasticity-consistent HC3 standard errors were computed for the OLS model. Across all approaches, the conclusions remained unchanged. Furthermore, as expected in a general workforce sample, the outcome variable, mental health problems, showed some positive skewness, although at a moderate level (T2 skewness = 1.10, kurtosis = 3.91). Given the large sample size, the use of summed multi-item scales, bootstrapped confidence intervals and HC3 robust standard errors, these deviations from normality were not considered severe enough to compromise the linear modelling approach. Nevertheless, as a robustness check, we also estimated the focal models using ordered logistic regression with a quartile-based version of the mental health variable (see Supplementary Materials). All interaction terms were estimated using mean-centred variables.

Two of the central variables in the study may capture overlapping aspects. The NAQ–R assesses individual exposure to negative acts directed at oneself at work, whereas the HWC reflects a more general perception that the workplace is characterized by a social climate marked by suspicion, antagonism and interpersonal tension. Previous research has indicated a reciprocal relationship between the two (Rosander and Salin, 2023). To examine whether these measures reflected a single underlying construct or should be regarded as two correlated but distinct constructs, we compared two models. A two-factor model, in which NAQ–R and HWC were specified separately, showed acceptable fit, χ²(316) = 3514.81, RMSEA = 0.057, CFI = 0.921, TLI = 0.912, SRMR = 0.047, whereas a one-factor model yielded clearly inferior fit, χ²(317) = 7456.03, RMSEA = 0.084, CFI = 0.823, TLI = 0.804, SRMR = 0.065. The difference was significant, Δχ²(1) = 3941.22, p < 0.001, and the two-factor solution was also favoured by both AIC and BIC. These results indicate that hostile work climate cannot be reduced to individual exposure to negative acts and should be treated as related but empirically distinct constructs in the subsequent analyses.

Descriptive statistics and correlations between the study variables are presented in Table 1. Hostile work climate and exposure to negative acts at T1 were associated with victimization at T2 (r = 0.29, and 0.36, respectively), and with mental health problems at T2 (r = 0.37 and 0.40, respectively).

Testing whether there is a sensitizing (H1a) or desensitizing (H1b) process in hostile climates, a logistic regression controlling for victimization at T1 showed that exposure to negative acts at T1 strongly predicted victimization at T2 (OR = 8.92, p < 0.001). Hostile work climate also predicted victimization (OR = 1.84, p < 0.001). Importantly, the interaction between exposure to negative acts and hostile work climate was significant (OR = 0.56, p < 0.001), indicating that the effect of negative acts on victimization weakened as the level of hostile climate increased, that is, a desensitizing effect consistent with H1b, whereas H1a was not supported (see Table 2).

Figure 1 illustrates this desensitizing effect, showing the predicted probability of victimization with increasing exposure to negative acts across levels of hostile climate. At lower levels of exposure, the probability of victimization was somewhat higher in hostile climates. As exposure increased, this difference diminished, and beyond a tipping point, the probability even became lower in hostile than in less hostile climates. The tipping point occurred at 1.04 above the mean of the NAQ–R, representing the 1.5% most exposed individuals.

Testing whether there is an amplification (H2a) or attenuation (H2b) effect on mental health as an outcome in hostile climates, we conducted a moderation analysis. Controlling for baseline mental health problems the results showed that exposure to negative acts at T1 was positively associated with mental health problems at T2, b = 0.13, p = 0.007, while hostile work climate showed no significant main effect, b = 0.02, p = 0.088. Importantly, the interaction between exposure to negative acts and hostile work climate was significant and negative, b = −0.05, p = 0.004, indicating that the association between bullying exposure and later mental health problems was weaker in workplaces characterized by a more hostile climate, that is, an attenuation effect consistent with H2b, whereas H2a was not supported (see Table 3).

To further probe the interaction, we examined the simple slopes. At low HWC (p10), exposure to negative acts significantly predicted subsequent mental health problems (b = 0.06, 95% BootCI [0.02, 0.09], p = 0.002). The slope was also significant at the median HWC (p50; b = 0.04, 95% BootCI [0.01, 0.07], p = 0.005). At high HWC (p90), however, the slope was nonsignificant (b = 0.01, 95% BootCI [−0.01, 0.04], p = 0.285). A Johnson–Neyman analysis indicated that the simple slope of exposure to negative acts on mental health problems was significant for HWC values up to about 0.8 SD above the mean, and non-significant at higher levels (see Figure 2).

To test victimization as a mechanism linking exposure to negative acts with subsequent mental health problems (H3a), we conducted a mediation analysis with victimization at T2 as mediator. Controlling for victimization and mental health problems at T1, as well as age, the results showed that victimization fully mediated the association between exposure to negative acts at T1 and mental health problems at T2. The indirect effect was significant (b = 0.29, 95% BootCI [0.10, 0.49]), whereas the direct effect was nonsignificant (b = 0.02, 95% BootCI [−0.06, 0.09]). These results supported H3a (see Figure 3).

To further support the notion that feelings of helplessness are an important reason for considering victimization as central in explaining why exposure to negative acts is associated with mental health problems, we conducted a t-test. Employees who perceived themselves as exposed to at least some negative acts and as victimized by these behaviours reported significantly higher helplessness in dealing with the exposure compared to those not victimized, t(492) = 5.64, p < 0.001, ΔM = 1.02, 95% CI [0.67, 1.38], d = 0.62, representing a medium-to-large effect.

To investigate whether the indirect effect depended on the level of hostile work climate (H3b), a moderated mediation analysis was conducted. Controlling for age, victimization and mental health problems at T1, the results showed that both the indirect and the direct path varied with hostile climate. As shown in Table 4, hostile work climate significantly moderated the path from exposure to negative acts to victimization (b = −0.40, p = 0.010) as well as the direct path from exposure to negative acts to mental health problems (b = −0.06, p = 0.001). The index of moderated mediation was significant (b = −0.07, 95% BootCI [−0.14, −0.01], p = 0.026), indicating that the indirect pathway from negative acts to mental health through victimization depended on the level of hostile climate. In absolute terms, the indirect effect was strongest at low levels of hostile work climate. However, because the direct effect was substantially reduced as hostile climate increased, the proportion of the total effect mediated through victimization was largest at high levels of hostile climate (89%) compared to low levels (73%). Moreover, at high levels of hostile work climate, the direct effect was nonsignificant. These results provided support for H3b.

To examine whether the findings depended on treating mental health problems as a continuous outcome, supplementary sensitivity analyses were conducted using ordinal logistic analyses based on quartiles of the HADS. The overall pattern of results remained unchanged across analyses, including the interaction between exposure to bullying behaviours and hostile work climate, as well as the mediating role of victimization. These findings suggest that the substantive conclusions were robust to alternative modelling approaches for the outcome variable. The results are presented in Table S1–3 in the Supplementary Materials.

The present study aimed to investigate whether hostile work climates give rise to sensitizing or desensitizing processes in relation to workplace bullying, and whether exposure to bullying behaviours in such climates amplifies or attenuates mental health problems. Furthermore, we examined the role of victimization in explaining the association between exposure to bullying behaviours and subsequent mental health problems, and whether this process depends on the level of hostile work climate. The results showed evidence of desensitizing processes in hostile work climates (H1b), with exposure in such climates attenuating mental health problems (H2b). Victimization, together with feelings of helplessness, played a central role in explaining the well-documented association between exposure to bullying behaviours and subsequent mental health problems (H3a). In highly hostile climates, the effect was in fact fully mediated through victimization (H3b).

Thus, to answer the question posed in the title of this study, our findings indicate that hostile work climates foster desensitization to bullying behaviours rather than sensitization, attenuating rather than amplifying the negative outcome. Specifically, the desensitization effect of a hostile work climate appears to raise the threshold for labelling oneself as a victim of bullying. This helps explain the attenuating effect, as we also found that the link between exposure to bullying behaviours and subsequent mental health problems was mediated by victimization – an effect particularly pronounced in hostile climates. In other words, the likelihood of labelling one’s own negative treatment as bullying is lower in hostile climates, and because this self-labelling is the central route to mental health problems, the overall effect of exposure is attenuated, at least in the short run. These findings will be discussed further below.

Previous studies have indicated a variation in how and when exposure to bullying behaviours is perceived as bullying, based on, for example, personal characteristics (Nielsen et al., 2017; Rosander and Nielsen, 2026a) or demographic aspects such as gender (Rosander et al., 2020). Adding to the understanding of how bullying may be construed, the present findings show that factors in the working environment may also influence perceptions of being bullied.

The results point to a normalization process regarding bullying behaviours in hostile work environments. The framework provided by Ashforth and Humphrey (1995) is useful for understanding why the probability of perceiving oneself as bullied does not increase to the same extent with higher levels of exposure to bullying behaviours as it does in less hostile work environments. The important question is what psychological function normalization serves in this context. Being exposed to bullying behaviours – and acknowledging to oneself that one is being bullied – may threaten self-image and evoke a sense of reduced control over one’s situation. Normalization helps reduce the dissonance between this experience and the need to perceive the world as comprehensible and manageable, despite the negative treatment to which one is exposed. By perceiving the behaviours as something anyone at work may be subjected to, the situation becomes less of a cognitive shock and the work environment appears more predictable, at least as construed by the individual. Normalization, whether through habituation or desensitization (Ashforth and Kreiner, 2002), may thus create the impression that an otherwise chaotic situation is manageable. Viewing the behaviours as something we are exposed to, rather than being singled out as a deviant, provides a sense of belonging that may feel protective in the situation (Hogg and Terry, 2000). It is psychologically easier to endure a hostile environment together than to stand alone. In this way, normalization can function as a defensive coping strategy that reduces perceived threat, in terms of emotion-focused coping (Lazarus and Folkman, 1984), where the situation is reappraised without altering the objective conditions.

The risks associated with a delayed realization that it is actually bullying one is exposed to (I am being bullied), as opposed to simply working in a poor environment (we have to endure negative working conditions), may lead to postponement in seeking help, which in the long run could have devastating effects on health. As bullying is an escalating process (Einarsen et al., 2020), the later individuals acknowledge being victims rather than one of many in a hostile working environment, the more extreme the mistreatment may become and, consequently, the more severe the outcomes. Reaching the point at which one labels oneself as bullied can be understood as an attribution process (Martinko et al., 2006). Initially, in a hostile working environment, it is most likely that, through the process of normalization, the attribution is directed towards the situation. Although this may involve negative consequences, these are more likely to concern work-related stress, exhaustion, reduced job satisfaction and intentions to leave. In contrast, once an individual labels themselves as a victim, the attribution likely shifts to the person, often involving self-doubt, rumination and questions about one’s own shortcomings (Weiner, 1985). This highlights the link to mental health problems as the outcome through victimization, rather than directly from merely being exposed to negative treatment in a hostile environment while believing oneself to be one of many targets.

Our findings suggest that exposure to bullying behaviours in itself is not sufficient to explain mental health problems. Instead, the critical mechanism appears to be whether individuals see themselves as bullied (i.e. victimization) – not merely exposed to a number of negative acts. This was particularly evident in hostile work climates, where the direct effect of exposure on mental health problems disappeared and the association was fully mediated by victimization. However, this tendency was also apparent regardless of the level of hostile climate. This process may reflect a form of learned helplessness (Abramson et al., 1978), where perceiving oneself as a victim is associated with reduced perceived control and resignation, thereby exacerbating mental health problems. The pattern was further supported by the finding that perceived helplessness was significantly higher among those who labelled themselves as bullied. This reasoning is in line with the cognitive activation theory of stress (Ursin and Eriksen, 2004), which highlights the importance of response outcome expectancies. From this perspective, acknowledging that the negative treatment one is exposed to represents a situation beyond one’s control, with no constructive response available, may contribute to such feelings of helplessness. Moreover, in line with the theory, victimization may also give rise to fears that attempts to act could worsen the situation, thereby reinforcing hopelessness and further increasing the risk of mental health problems.

When the climate was not hostile, the results showed partial mediation, that is, there was a smaller direct effect between exposure to bullying behaviours and mental health problems (up to about a quarter of the effect was direct in non-hostile work environments). This could be understood in terms of Janoff-Bulman’s (1992) theory of shattered assumptions. Being exposed to bullying behaviours in an otherwise benign working environment may come as a shock. Such experiences may shatter fundamental assumptions about the world as basically benevolent, where events are perceived as predictable and people as deserving of respect and fair treatment. Hence, exposure to bullying behaviours in a generally benign working environment may be directly damaging because it undermines these core beliefs. Supporting this interpretation, a recent study similarly found stronger associations between bullying and impaired mental health in what they referred to as low-bullying climates (Steele et al., 2022) – a concept that may be seen as corresponding to a low hostile work climate. In addition, the discrepancy between expected and actual reality may in itself trigger a stress reaction with little expectancy of coping, thereby directly affecting mental health (Ursin and Eriksen, 2004).

The finding that so much of the effect on mental health outcomes operates through victimization is an important contribution to the research field of workplace bullying. Previous research has provided indications of such patterns, for instance showing that the association between exposure to bullying behaviours and psychological distress disappeared once victimization was accounted for (Nielsen et al., 2012). The present findings go beyond these indications by explicitly demonstrating that the effect is mediated through victimization, and by linking this mechanism to the perception of helplessness (Abramson et al., 1978).

The findings also carry important practical implications. Firstly, interventions should not only focus on reducing the occurrence of bullying behaviours but also on addressing the normalization processes that could delay self-labelling and help-seeking. Awareness-raising initiatives, such as training and communication campaigns, may lower the threshold for recognizing bullying behaviours and help employees distinguish them from generally poor working conditions. Such initiatives may be particularly important in organizational cultures where interpersonal hostility has become routinized or minimized as part of the occupational culture.

Secondly, organizations need to create supportive structures that facilitate early acknowledgement of victimization. Because self-labelling is a key mechanism linking exposure to mental health outcomes, it is essential that employees have safe channels to reflect upon and report their experiences without fear of stigmatization or retaliation. This could involve confidential counselling services, peer-support initiatives or HR practices that explicitly validate employees’ perceptions of negative treatment. Occupational health services may also play an important role by helping employees interpret and contextualize their experiences before the situation escalates into more severe psychological strain or prolonged sickness absence.

Thirdly, the findings underscore the importance of work climate and leadership. Hostile work climates both foster normalization and raise the threshold for self-labelling, thereby attenuating the link between exposure and health outcomes in the short run but likely increasing risks in the long run. Organizations may therefore benefit from regular psychosocial risk assessments and from training managers and HR professionals to recognize subtle forms of normalized hostility and recurring interpersonal conflicts before more severe bullying develops. Leaders play a crucial role in setting clear behavioural standards, responding to early signs of mistreatment and promoting a culture where negative behaviours are not tolerated or dismissed as “the way things are done here”, echoing the pioneering contribution of Brodsky (1976), who highlighted the need for a work climate that clearly shows that bullying is not accepted and that this stance is firmly backed by the organization. By improving the psychosocial climate and strengthening conflict management practices, organizations may prevent negative behaviours from becoming routinized and protect employees from the harmful escalation of bullying (Einarsen et al., 2018).

Fourthly, the findings suggest that organizations should pay greater attention to the role of bystanders and informal social norms. Previous research show that bystander interventions are highly important to reduce both the occurrence and outcomes of workplace bullying (Nielsen et al., 2021; Rosander and Nielsen, 2023). When negative behaviours become normalized, co-workers may be less likely to intervene, validate targets’ experiences or report problematic conduct. Interventions should therefore not only target potential perpetrators and victims, but also strengthen collective responsibility and employees’ confidence in responding constructively to early signs of harassment. Encouraging active bystander behaviour may help disrupt normalization processes before hostile interaction patterns become embedded in the organizational culture.

An important strength of this study is the use of a prospective national probability sample. In addition, we controlled for baseline levels of the mediator and outcome, which allowed us to capture change over time rather than mere static associations. Controlling for prior levels typically absorbs much of the variance through stability paths, often reducing the strength of other effects. The fact that the hypothesized processes nevertheless emerged under these rigorous conditions underscores the robustness of the results and supports confidence in their practical relevance.

There are also limitations. The data relied on self-reports and are therefore susceptible to social desirability and common method variance (Podsakoff and Organ, 1986). An important step in addressing this concern was to investigate whether the measure of exposure to bullying behaviours and that of hostile work climate could be considered distinct constructs. The analysis clearly indicated that this was the case.

Workplace bullying is a highly skewed phenomenon, as most people are not exposed. For this reason, we applied bootstrapping in all analyses, a method that reduces the impact of deviations from normality. We also used robustness checks by estimating both bias-corrected and normal-based intervals, and in the OLS model we additionally used heteroskedasticity-consistent standard errors.

The time frame between baseline and follow-up was seven months. This may have influenced both the desensitization and attenuating effects, which are likely to be most evident in the short run. The results point to a raised threshold for victimization, while previous research suggests that prolonged exposure to bullying behaviours will eventually take its toll, that is, over a longer time frame, desensitization and attenuation effects may diminish. An important question, therefore, is how long such effects actually last – something that future research should investigate. Furthermore, with regard to the time frame, it is important to acknowledge that bullying is a process that often takes time to develop. However, a seven-month interval does not imply that respondents were not exposed to bullying behaviours prior to the baseline measurement; on the contrary, our data showed that only a small proportion of those who reported exposure at baseline had in fact started to experience it very recently.

Finally, although the present study demonstrated the importance of victimization in understanding the outcomes of exposure to bullying behaviours, we do not advocate relying on self-labelling instead of behavioural experience as the preferred way of measuring bullying. Rather, the findings highlight the value of using both approaches in combination. It should also be noted that victimization was measured with a single self-labelling item based on a standard bullying definition, which is common practice in workplace bullying research. However, single-item measures do not allow estimation of internal consistency and may be more susceptible to measurement error. Such error would typically attenuate associations and may therefore have resulted in conservative estimates of the indirect effects in the moderated mediation analyses.

The present study provides novel evidence that hostile work climates foster desensitization rather than sensitization, thereby attenuating rather than amplifying the immediate health consequences of exposure to bullying behaviours. Crucially, the findings demonstrate that victimization – self-labelling as bullied – is the central mechanism linking exposure to negative acts with mental health problems, and that this mechanism is particularly salient in hostile climates. While normalization processes may initially protect employees from the cognitive shock of exposure, they risk delaying recognition and help-seeking, which can exacerbate the long-term consequences of bullying. Taken together, the results underscore the need to consider both individual perceptions and workplace climate when assessing the risks and outcomes of bullying, and they highlight the value of integrating behavioural and self-labelling measures in future research.

The supplementary material for this article can be found online.