SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Background
  4. STUDY 1
  5. Method
  6. Results
  7. Discussion
  8. STUDY 2
  9. Method
  10. Analysis
  11. Results
  12. Discussion
  13. GENERAL DISCUSSION
  14. Acknowledgement
  15. References

In this article, we investigate the impact of engaging in ruminative-style thoughts after exposure to workplace violence. Rumination is a form of self-focused thinking characterized by abstract and passive negative thoughts. In an experimental study in which student volunteers were exposed to simulated violence using a video manipulation, the unpleasant affect of participants instructed to ruminate about the violence persisted, while the affect of participants in a distraction condition was quickly repaired. In a field study of violence experienced by social workers in their everyday working lives, employees who had a high tendency to engage in ruminative thinking exhibited a stronger negative relationship between exposure to violence and poor well-being and health complaints compared with those who had a low tendency to ruminate. Together, our findings suggest that ruminative thinking may exacerbate the negative effects of workplace violence.

Practitioner Points

  • Being subjected to violence at work can have negative implications for employees' health and well-being, but it is not always possible for organizations to prevent violent attacks (e.g., from members of the public).
  • Our findings indicate that the negative consequences of violence may be intensified when the victim engages in rumination, a negative type of thinking about the self that involves passive, abstract thoughts about what happened and the effects on one's life.
  • Organizations should therefore seek to discourage ruminative thinking among victims of workplace violence.

Background

  1. Top of page
  2. Abstract
  3. Background
  4. STUDY 1
  5. Method
  6. Results
  7. Discussion
  8. STUDY 2
  9. Method
  10. Analysis
  11. Results
  12. Discussion
  13. GENERAL DISCUSSION
  14. Acknowledgement
  15. References

A growing number of people are subjected to violent attacks in their places of work. According to the British Crime Survey, 1.4% of working adults in England and Wales were victims of violent workplace attacks in the 2009/2010 period (Packham, 2010). When exposed to a traumatic event like workplace violence, many victims experience negative psychological and somatic consequences, including depression, anxiety, gastrointestinal problems and sleep disturbances (Hogh & Viitasara, 2005). The associated financial costs are high: an International Labour Organization report estimated that losses from workplace violence account for between 0.5% and 3.5% of gross domestic product internationally, with the United Kingdom losing 18 million working days annually from violence at work (Hoel, Sparks & Cooper, 2001).

Although organizations and their employees may try to reduce the occurrence of workplace violence, it is rarely possible to prevent all attacks, especially those from organizational outsiders like clients, patients or members of the public, which tend to be unpredictable (Grandey, Kern & Frone, 2007). As such, there is impetus on researchers to identify ways in which the negative effects of exposure to violence can be minimized. In the present study, we draw on response styles theory (Nolen-Hoeksema, 1991) in addition to the traditional stressor-strain model (Lazarus, DeLongis, Folkman & Gruen, 1985) to argue that the way in which victims respond to workplace violence affects the impact of that violence. Specifically, we propose that whether or not the victim engages in rumination – a form of self-focused thinking characterized by repetitive thoughts about what happened, what caused the event and the effects on one's life – will play a central role in influencing the effects of violence. We test this proposition in both an experimental laboratory study and a field study in a work organization.

Workplace violence

Workplace violence refers to high-intensity, severe behaviour directed towards another person while at work, with intent to harm (Hershcovis, 2011), that is either physical (e.g., kicking or punching) or psychological (e.g., threatening violence or swearing; LeBlanc & Kelloway, 2002). To date, most workplace violence research has focused on examining the effects of exposure, and it is now well established that being subjected to violence at work has ‘serious acute and long-term consequences for the health and well-being of the victim’ (Hogh & Viitasara, 2005, p. 305). A small body of research suggests that the detrimental consequences of violence may not be inevitable, however. In particular, the way in which victims respond to violence is thought to influence its effects; for example, studies have suggested that the negative effects of exposure to violence are buffered amongst people who have a trait preference for reinterpreting negative events (i.e., reappraisal; Niven, Sprigg & Armitage, in press).

However, such findings about the effects of workplace violence have yet to be placed into a wider theoretical framework. As such, the reasons why violence typically leads to negative consequences, and why some people are more or less severely affected, are currently poorly understood. In the present study, we aim to provide insight into these two issues by proposing and testing a theoretical framework that combines the stressor-strain model (Lazarus et al., 1985) with response styles theory (Nolen-Hoeksema, 1991) to situate the effects of workplace violence in relation to victims' response types. Given this focus, it is important to note although a victim's response might influence the effects of violence, our framework does not imply that victims themselves are responsible for those effects. Rather, it offers a more nuanced understanding of the effects of violence, which can provide employees and organizations with insight into how to manage the aftermath of violent attacks in a way that best promotes the recovery of victims.

Theoretical framework

Stressor-strain model

The stressor-strain model is the dominant theoretical approach in understanding effects of workplace violence (Lazarus et al., 1985). The model posits that being subjected to violence at work is an objective stressor. When people are exposed to stressors, they seek to make sense of what has happened by appraising: (1) the stressor itself and (2) their ability to cope with it. This two-stage appraisal process may result in an activated stress appraisal (e.g., ‘the event I experienced was intensely negative and I cannot cope’), which causes the person to experience acute affective and physiological arousal (e.g., unpleasant affect, sympathetic nervous system activation). Continued activation of the stress appraisal will lead to chronic strain symptoms (e.g., poor well-being and health complaints). Thus, workplace violence negatively affects employees as a result of stress appraisals.

Response styles theory

Beyond the domain of workplace violence, response styles theory (Nolen-Hoeksema, 1991) has frequently been used to understand the negative effects of traumatic events. This theory proposes that the ways in which people respond to a traumatic event will be a strong determinant of the effects of that event. In particular, the theory focuses on rumination, which involves negative evaluative thoughts about the self that tend to be at an abstract level of construal and that are passive in nature (Nolen-Hoeksema, Wisco & Lyubomirsky, 2008). For example, after being subjected to workplace violence, a victim's ruminative thoughts might focus on how the violence was his or her own fault, the role he or she played in the incident and the likely negative effects on other areas of his or her life. Rumination can be seen as a relatively stable response style (‘trait rumination’) such that certain people are particularly prone to engaging in ruminative thinking (Nolen-Hoeksema, Morrow & Fredrickson, 1993). However, people can also be prompted to engage in ruminative thinking in the moment (‘state rumination’), as a result of exposure to a particularly severe trauma (Haslam & Mallon, 2003) or other external triggers (e.g., experimental instructions; Glynn, Christenfeld & Gerin, 2002).

According to response styles theory, while there are adaptive and instrumental alternative response types, such as distraction (Nolen-Hoeksema et al., 2008), ruminative thinking after a stressful event is maladaptive for three reasons. First, it prompts exclusively negative thoughts and emotions about the event. Second, because ruminative thoughts are passive, they obstruct effective problem solving and mood regulation. Third, the abstract and self-referential nature of ruminative thoughts encourages generalization beyond the event (see also Rimes & Watkins, 2005; Watkins, 2008). Together, these reasons suggest that engaging in state rumination may be at least partly responsible for the negative effects of stressors like workplace violence. In support of this, laboratory studies have shown that while people instructed to ruminate about stressors have higher physiological arousal, those who are unable to ruminate, because they have been either distracted from the stressor (Glynn et al., 2002) or instructed to reappraise the stressor (Ray, Wilhelm & Gross, 2008), do not show high levels of arousal and recover quickly. In addition, field studies have demonstrated that people with higher trait rumination experience poorer long-term well-being and health after experiencing traumatic events like earthquakes (Nolen-Hoeksema & Morrow, 1991) and terrorist attacks (Friedberg, Adonis, Von Bergen & Suchday, 2005), presumably because trait rumination prompts people to continue to engage in ruminative thinking in the aftermath of such events. Thus, response styles theory suggests the negative effects of stressful events like violence arise as a result of ruminative thinking.

The combined stressor-strain response styles model

Applying response styles theory to the stressor-strain model provides an explanation for why workplace violence typically has a negative impact on its victims and why certain people are more or less likely to suffer these negative consequences (see Figure 1). Specifically, the negative effects of workplace violence arise as a result of stress appraisals, which continue to be activated when the victim engages in maladaptive response types, such as rumination. Thus, the extent to which victims of workplace violence engage in ruminative thinking (or other, more adaptive response styles) will influence the consequences of violence.

image

Figure 1. The combined stressor-strain response styles model.

Download figure to PowerPoint

During a traumatic event like being subjected to workplace violence, stress appraisals are likely to be activated for most victims; hence, almost all victims are likely to experience short-term arousal during and immediately after the event. Ruminative thinking in the aftermath of the event continues the activation of stress appraisals, because it focuses victims' attention on the negative aspects of the situation, encourages generalization and obstructs solutions, meaning that the event will still be appraised as a stressor that one is unable to cope with. However, if a victim engages in a response that prevents rumination (e.g., distraction, reappraising the situation), stress appraisals may be curtailed. State ruminative thinking immediately after an attack should therefore extend the duration of acute arousal, and those people with a trait tendency to engage in ruminative thinking should exhibit a stronger relationship between exposure to violence and chronic strain. We therefore hypothesize the following:

Hypothesis 1: State rumination will cause unpleasant affect arising from workplace violence to persist.

Hypothesis 2: Trait rumination will strengthen the relationship between exposure to workplace violence and poor well-being and health complaints.

Indirect support for these propositions is provided by a study that found that thoughts of revenge (framed as an indicator of rumination) marginally intensified the relationship between exposure to workplace bullying and strain (Moreno-Jiménez, Rodríguez-Muñoz, Pastor, Sanz-Vergel & Garrosa, 2009). Here, the marginal result may have arisen because thoughts of revenge represent a relatively active and concrete form of thinking that have a conceptual end-point (i.e., the act of revenge) and so may lead to deactivation of the stress appraisal. In addition, the finding that trait reappraisal buffers the negative effects of workplace violence (Niven et al., in press) suggests that engaging in a response that deactivates the stress appraisal can lessen the negative impact of violence. However, no studies have directly tested whether state rumination is responsible for the persistence of acute negative effects of workplace violence or whether trait ruminators will suffer more severe chronic consequences when exposed to workplace violence.

The current research

In the present research, two studies investigated the role of rumination in determining the effects of workplace violence, drawing on the advantages of controlled and naturalistic approaches to data collection. Hypothesis 1 was tested in Study 1, an experimental study in which violence and rumination were manipulated among a student sample. Hypothesis 2 was tested in Study 2, a field study of employees' experiences of violence in an organization.

STUDY 1

  1. Top of page
  2. Abstract
  3. Background
  4. STUDY 1
  5. Method
  6. Results
  7. Discussion
  8. STUDY 2
  9. Method
  10. Analysis
  11. Results
  12. Discussion
  13. GENERAL DISCUSSION
  14. Acknowledgement
  15. References

In a laboratory study, we used an experimental approach to examine the acute effects of state rumination on unpleasant affect. Although experimental studies are relatively rare in the area of workplace violence, they are increasingly being used to study applied phenomena, for three key reasons. First, because key variables can be manipulated rather than measured, experimental studies help to disentangle issues of causality inherent in correlational research. Second, experimental studies allow greater control over the treatment experienced by participants, meaning, for example, that participants can be exposed to violence of the same severity. Third, such studies allow random allocation to conditions, meaning that any underlying differences in personality and prior experiences are likely to be balanced across conditions, helping to rule out potential alternative explanations for effects.

To ensure ethical standards, participants were exposed to simulated violence using a realistic video. We used a video because participants feel more engaged and respond more authentically to videos compared to other manipulations of violence (Noel et al., 2008). We conducted the study in a laboratory using student participants. To increase the realistic nature of the experiment, we told participants that the perpetrator in the video was a fellow student from their degree course (as a member of their working environment) and used a perpetrator of the same gender as the participants (as same gender workplace assaults are far more common than different gender assaults; Flannery, Marks, Laudani & Walker, 2007). We contrasted the response type of rumination with the response type of distraction, as distraction is used as the comparison condition in most laboratory investigations of the effects of rumination (e.g., Glynn et al., 2002; Lyubomirsky, Caldwell & Nolen-Hoeksema, 1998; Nolen-Hoeksema & Morrow, 1993). It is favoured because distraction manipulations are affectively neutral, cognitively undemanding, and divert participants' attention away from a stressor, thus ensuring that they are unable to ruminate.

Method

  1. Top of page
  2. Abstract
  3. Background
  4. STUDY 1
  5. Method
  6. Results
  7. Discussion
  8. STUDY 2
  9. Method
  10. Analysis
  11. Results
  12. Discussion
  13. GENERAL DISCUSSION
  14. Acknowledgement
  15. References

Participants

Data were collected from 32 female first-year undergraduate psychology students from a UK university (Mage = 19.32 years, SD = .95) who participated in return for course credits. Only female participants were used to reduce variability in participants' appraisals of and arousal responses to the video manipulation (Glynn et al., 2002; Noel et al., 2008), in line with procedures adopted in other laboratory studies examining the impact of responses like rumination to stressors (e.g., Denson, Grisham & Moulds, 2011; Memedovic, Grisham, Denson & Moulds, 2010; Ray et al., 2008). Participants were recruited via an advert in an online participation system for psychology students that asked for female volunteers to take part in a study about ‘how aggression affects people’. The sample size was informed by a power analysis calculated using G*Power (Faul, Erdfelder, Lang & Buchner, 2007), based on a medium effect size (estimated from the results of meta-analysis; Mor & Winquist, 2002) and a power of .80 (a value that Cohen, 1988, proposes as appropriate), with the design as specified below.

Design

The experiment had a mixed design. Participants were randomized to a between-participants factor of condition (rumination vs. distraction; = 16 in each); the within-participants factor was time. Measures of the dependent variable (affect) were taken at three time-points: at baseline (Time 1), immediately after the violence manipulation (Time 2) and after the rumination or distraction response strategy (Time 3).

Procedure

All participants completed their Time 1 measure of affect immediately upon arriving at the laboratory and were told that they would be shown a video of a fellow student from their course and that they should imagine the person was currently in the room with them. All participants then watched a 90-s video created for the purposes of the study, in which a female drama student acted in a violent manner directly towards the viewer. The video was scripted and involved the perpetrator performing the acts and threats of violence included in a widely used measure of workplace violence (Rogers & Kelloway, 1997). So, for example, the perpetrator threw an object at the viewer, threatened to enact violence, used aggressive language such as swearing and gender-specific insults and pushed the camera over to imply actual violence. The script was designed to encourage the belief that the perpetrator was from the participants' own course, and the video was shot in the same laboratory room in which the experiment took place. After watching the video, participants completed their Time 2 measure of affect.

All participants were then given the following message: ‘Imagine this incident just happened to you, and the person just left the room’, and received instructions to complete a ‘response task’ for a period of 2 min, which differed depending on condition. We chose a period of 2 min informed by prior research, which has reported that rumination can be successfully induced in this time period (Ray et al., 2008). Those in the rumination condition were asked to think about how the incident made them feel and what implications it might have for them (mirroring items from a well-established measure of rumination; Robinson & Alloy, 2003). Those in the distraction condition were given a task that has been shown not to influence people's affect (McQueen & Klein, 2006). In this task, originally devised by Steele and Liu (1983), participants are given a list of personal values (e.g., sense of humour, creativity) and asked to think about why the value that is least important to them might be important to other people. In both conditions, participants were told they would need to make notes after the 2 min were over. This encouraged participants to perform the task, but avoided the potentially confounding effects of writing about their feelings (Baikie & Wilhelm, 2005). After the 2-min period, participants completed their Time 3 measure of affect. They then filled out a series of scales relating to their previous exposure to violence and violent media. Finally, participants were debriefed, and a pleasant mood induction (Velten, 1968) returned all participants to a good mood.

Measures

Affect

Our key dependent variable was the valence (i.e., pleasantness) of participants' affect, measured using a single item in which participants rated the valence of their affect on a scale ranging from 1 ‘extremely unpleasant’ to 9 ‘extremely pleasant’. This item has been shown to correlate well with multi-item indicators while avoiding response fatigue (Russell, Weiss & Mendelsohn, 1989).

Previous exposure to violence and violent media

We assessed previous exposure to violence and violent media to check for pre-existing differences between participants. Previous exposure to violence was assessed using a modified version of the eight-item workplace violence scale of Rogers and Kelloway (1997), with items amended to relate to exposure to violence in any context. Items asked participants to indicate how frequently they had been exposed to violent acts (e.g., ‘Having objects thrown at you’) on a five-point scale from ‘never’ (1) to ‘daily’ (5) (α = .85). Previous exposure to violent media was measured using a three-item use of violent media content scale (Slater, Henry, Swaim & Anderson, 2003), which asked about frequency of playing violent video games, watching violent movies and visiting violent internet sites, with responses on a five-point scale ranging from ‘not at all’ to ‘very often’ (α = .61).

Results

  1. Top of page
  2. Abstract
  3. Background
  4. STUDY 1
  5. Method
  6. Results
  7. Discussion
  8. STUDY 2
  9. Method
  10. Analysis
  11. Results
  12. Discussion
  13. GENERAL DISCUSSION
  14. Acknowledgement
  15. References

Randomization check

We first checked for pre-existing differences between participants in the distraction and rumination conditions. One-way ANOVAs with the between-subjects factor of condition (rumination vs. distraction) confirmed that the two groups were equivalent with respect to previous exposure to violence (rumination = 1.38, SD = .16; distraction = 1.65, SD = .62), F (1, 30) = 2.97, = .10, inline image = .09, previous exposure to violent media (rumination = 1.52, SD = .36; distraction = 1.53, SD = .50), F (1, 30) = 0.01, = .94, inline image < .01, and baseline affect at Time 1 (see Table 1 for means), F(1, 30) = 1.54, = .22, inline image = .05.

Table 1. Means and standard deviations of affect by time and condition in study 1 (N = 32)
 DistractionRumination
Mean SD Mean SD
Note
  1. Higher affect scores indicate more pleasant affect.

Affect at Time 1 (baseline)5.781.476.501.79
Affect at Time 2 (post-violence manipulation)4.031.403.561.67
Affect at Time 3 (post-response strategy)5.501.264.001.67

Effects of the violence manipulation and response strategies

Hypothesis 1 stated that state rumination would cause unpleasant affect arising from workplace violence to persist. To test this hypothesis, we conducted a 3-within (time of measure: Time 1 vs. Time 2 vs. Time 3) by 2-between (condition: rumination vs. distraction) ANOVA with affect as the dependent variable. Means and standard deviations for measures of affect at all time-points are displayed in Table 1. The results of the ANOVA revealed a main effect of time, F(2, 29) = 28.52, < .01, inline image = .66, which was qualified by an interaction between time and condition, F(2, 29) = 5.34, < .05, inline image = .27 (see Figure 2). There was no main effect of condition, F(1, 30) = 0.99, = .33, inline image = .03.

image

Figure 2. Interactive effects of time and condition on affect in study 1.

Download figure to PowerPoint

We broke down the significant interaction into stages to determine whether the results supported the hypothesis. Our initial exploration examined the impact of the violence manipulation. Repeated measures ANOVAs comparing Time 1 and Time 2 affect for each condition separately showed that participants' affect became significantly less pleasant after watching the violent video compared with baseline in the distraction condition, F(1, 15) = 17.93, < .01, inline image = .54, and in the rumination condition, F(1, 15) = 42.32, < .01, inline image = .74. Simple effects analyses confirmed that there were no differences between the two conditions in affect at Time 2 after watching the video, F(1, 30) = 0.74, = .40, inline image = .02. Thus, participants' affect worsened as a result of violence manipulation to a comparable degree in both conditions.

We then examined whether engaging in the response strategy to which they were assigned influenced participants' affect. A repeated measures ANOVA comparing Time 2 and Time 3 affect indicated that the affect of those in the distraction condition was significantly more pleasant after completing their distracter task, F(1, 15) = 16.32, < .01, inline image = .52 (in fact, a comparison between Time 1 and Time 3 indicated that their affect was restored to baseline levels, F(1, 15) = 0.91, = .36, inline image = .06). In contrast, equivalent analyses showed that the affect of participants in the rumination condition remained unchanged between Time 2 and Time 3, F(1, 15) = 1.00, = .33, inline image = .06 (and was significantly less pleasant at Time 3 compared with baseline levels, F(1, 15) = 17.44, < .01, inline image = .54). Simple effects analyses further confirmed that distraction participants' affect was significantly more pleasant than rumination participants' affect after the response strategy at Time 3, F(1, 30) = 8.18, < .01, inline image = .21. These findings suggest that while distraction participants' affect improved after using their response strategy, recovering to their original levels, rumination participants' unpleasant affect persisted. Hypothesis 1 was therefore supported.

Discussion

  1. Top of page
  2. Abstract
  3. Background
  4. STUDY 1
  5. Method
  6. Results
  7. Discussion
  8. STUDY 2
  9. Method
  10. Analysis
  11. Results
  12. Discussion
  13. GENERAL DISCUSSION
  14. Acknowledgement
  15. References

Our results showed that after being exposed to simulated violence, the affect of participants who were distracted was restored to pre-violence levels, while the unpleasant affect of participants instructed to ruminate persisted. These findings suggest that state rumination prolongs acute unpleasant affect after exposure to simulated violence. However, as we contrasted rumination to an ‘active’ comparison condition (as is the norm with laboratory studies of rumination, e.g., Nolen-Hoeksema & Morrow, 1993), it is possible that the effects were driven by positive gains in our distraction condition only. We expect that this is not the case, because the distracter task we used has negligible effects on people's affect (McQueen & Klein, 2006). Nevertheless, future studies using a non-manipulation control condition or a manipulation check would help to increase confidence in our theoretical explanation.

Despite the strengths of our experimental study design, two key limitations restrict the generalizability of our results. First, our study was somewhat artificial in nature. For ethical reasons, we were only able to expose participants to simulated violence; although the purpose-designed violent video manipulation did worsen participants' affect, exposure to real workplace violence would likely elicit stronger stress appraisals. Similarly, actual rumination in response to workplace violence might be prompted unexpectedly as a result of lasting reminders of the attack (e.g., injuries), which could give rumination an intrusive feel. As such, the true effects of ruminative thinking following exposure to violence may have been underestimated in the present study. Second, our use of a female-only sample means that we cannot be sure that our findings would apply to men. A second study was therefore conducted to address these limitations.

STUDY 2

  1. Top of page
  2. Abstract
  3. Background
  4. STUDY 1
  5. Method
  6. Results
  7. Discussion
  8. STUDY 2
  9. Method
  10. Analysis
  11. Results
  12. Discussion
  13. GENERAL DISCUSSION
  14. Acknowledgement
  15. References

In our second investigation, we conducted a field study exploring how trait tendency to spontaneously engage in rumination would influence the effects of exposure to real incidents of workplace violence on well-being and health complaints. We recruited a sample of male and female social workers, selected because social workers are at increased risk of violent assaults by clients (Packham, 2010). As such, this second study addressed the two key weaknesses of Study 1.

Method

  1. Top of page
  2. Abstract
  3. Background
  4. STUDY 1
  5. Method
  6. Results
  7. Discussion
  8. STUDY 2
  9. Method
  10. Analysis
  11. Results
  12. Discussion
  13. GENERAL DISCUSSION
  14. Acknowledgement
  15. References

Sample and design

Data were collected from social workers employed at a regional council in the UK. An advert for a study about ‘unacceptable behaviour at work’ was sent out by a senior manager in an email that included the link to an online survey. The survey was completed by 78 volunteers (62 women and 16 men, Mage = 47.58 years, SD = 9.72). Respondents had been employed in the department for at least 6 months (Mtenure = 12.87 years, SD = 8.51), which was important because we examined participants' exposure to workplace violence and their well-being and health complaints over the previous 6 months. They worked an average of 35.71 hrs a week (SD = 7.62), and spent on average 35% of their work time dealing directly with clients.

Measures

Exposure to workplace violence

We assessed the frequency with which participants had been exposed to acts of physical violence or threats of violence from clients in the previous 6 months while at work using the established eight-item workplace violence scale (Rogers & Kelloway, 1997) described in Study 1 (α = .81).

Psychological well-being

Participants' psychological well-being was assessed using a six-item scale developed by Warr (1990). The scale measures the extent to which people have experienced unpleasant affective states associated with anxiety (‘Tense’, ‘Worried’, ‘Uneasy’) and depression (‘Miserable’, ‘Depressed’, ‘Gloomy’) over the past 6 months. Responses are made on a five-point scale from ‘not at all’ (1) to ‘all of the time’ (5) (α = .93).

Health complaints

We assessed health complaints using an abridged version of Schat, Kelloway and Desmarais's (2005) physical health questionnaire (PHQ). The original scale assesses four types of somatic symptoms (gastrointestinal problems, headaches, sleep disturbances and respiratory illness), and we used the two highest loading items assessing each of these symptom types. All items asked participants to indicate how much of the time they had experienced a symptom over the past 6 months (e.g., ‘How often have you suffered from an upset stomach?’) on a seven-point scale from ‘not at all’ (1) to ‘all of the time’ (7) (α = .79).

Trait rumination

We used the nine-item stress-reactive rumination scale (Robinson & Alloy, 2003) to assess participants' trait ruminative thinking style (e.g., ‘I often think about how negative events will affect other areas of my life’). Participants responded on a five-point scale ranging from ‘never’ (1) to ‘always’ (5) (α = .91).

Analysis

  1. Top of page
  2. Abstract
  3. Background
  4. STUDY 1
  5. Method
  6. Results
  7. Discussion
  8. STUDY 2
  9. Method
  10. Analysis
  11. Results
  12. Discussion
  13. GENERAL DISCUSSION
  14. Acknowledgement
  15. References

Hypothesis 2 stated that trait rumination would strengthen the relationship between exposure to violence and poor well-being and health complaints. We tested this hypothesis using moderated regression analyses, following Aiken and West's (1991) procedures. At Step 1, the dependent variable (either poor well-being or health complaints) was regressed onto the standardized independent variable (exposure to workplace violence). The standardized moderator (trait rumination) was entered at Step 2. Moderation was indicated by significance at Step 3, when the interaction between these two standardized variables was entered. Simple slopes analyses examining the relationship between violence and the outcomes at low (−1SD) and high (+1SD) levels of rumination were then used to plot significant interactions.

Results

  1. Top of page
  2. Abstract
  3. Background
  4. STUDY 1
  5. Method
  6. Results
  7. Discussion
  8. STUDY 2
  9. Method
  10. Analysis
  11. Results
  12. Discussion
  13. GENERAL DISCUSSION
  14. Acknowledgement
  15. References

Means, standard deviations and correlations between the study variables are displayed in Table 2. The demographic characteristics of gender, age and tenure were largely unrelated to the variables of interest, and so these were not used as controls in subsequent analyses.1 Although mean rates of exposure to workplace violence were low (= 1.23 on a 1–5 scale), suggesting that few people experienced very frequent (e.g., daily) violence, just 23% of the sample reported not having been exposed to any of the acts of violence we measured in the prior 6 months. Consistent with previous literature, both exposure to workplace violence and trait rumination were associated with poor well-being and health complaints.

Table 2. Means, standard deviations, intercorrelations and Cronbach's alphas of study 2 variables (N = 78)
 MeanSD1234567
Note
  1. *< .05; **< .01.

1. Gender (1 = female, 2 = male)         
2. Age47.589.72.21*(–)     
3. Tenure12.878.51.16.48**(–)    
4. Exposure to violence1.23.30.12.13.07(.81)   
5. Trait rumination2.75.73.01.05.09.05(.91)  
6. Poor well-being2.23.82−.18.13.13.45**.48**(.93) 
7. Health complaints2.991.12−.27*−.07.06.27*.46**.59**(.79)

Rumination moderated the association between exposure to workplace violence and poor well-being, with the interaction between rumination and violence explaining a significant proportion of the variance in employees' well-being at Step 3 of the analysis, ΔR2 = .04, ΔF(1, 78) = 4.19, < .05 (Table 3). As can be seen in Figure 3, the relationship between violence and poor well-being was only significant at high levels of rumination (= .58, < .01) and not at low levels (= .08, = .55). Rumination also moderated the association between exposure to violence and health complaints, with the interaction between rumination and violence explaining a significant proportion of the variance in health complaints at Step 3 of the analysis, ΔR2 = .10, ΔF(1, 78) = 10.58, < .01. Figure 4 shows that the relationship between violence and health complaints was significant at high levels of rumination (= .54, < .01) but not at low levels (= −.29, = .14). Thus, people exposed to violence who typically ruminate about stressful experiences reported poorer well-being and more health complaints. Hypothesis 2 was therefore supported.

image

Figure 3. Interactive effects of violence and trait rumination on poor well-being in study 2.

Download figure to PowerPoint

image

Figure 4. Interactive effects of violence and trait rumination on health complaints in study 2.

Download figure to PowerPoint

Table 3. Beta and t-values for interactions between violence and trait rumination in study 2 (N = 78)
 Effects on poor well-beingEffects on health complaints
βtβ t
Note
  1. *< .05; **< .01.

Step 1
Exposure to violence.454.27**.272.34*
ΔR² .20** .07*
Step 2
Exposure to violence.424.45**.242.33*
Trait rumination.454.82**.484.73**
ΔR² .21** .23**
Step 3
Exposure to violence.353.51**.121.14
Trait rumination.515.31**.575.77**
Interaction: Violence × trait rumination.212.05*.343.25**
ΔR² .04* .10**

Discussion

  1. Top of page
  2. Abstract
  3. Background
  4. STUDY 1
  5. Method
  6. Results
  7. Discussion
  8. STUDY 2
  9. Method
  10. Analysis
  11. Results
  12. Discussion
  13. GENERAL DISCUSSION
  14. Acknowledgement
  15. References

The findings of Study 2 indicate that when people are exposed to violence at work, those who have a high tendency to ruminate experience significantly poorer well-being and greater health complaints. The study built on Study 1 by examining people's spontaneous ruminative thinking and their exposure to real incidents of violence in a working context and by including both female and male participants. The findings of the present study therefore provide a more ecologically valid and generalizable test of our theory.

However, there were limitations apparent in this second study. Notably, it was correlational in design with the key variables measured using self-reports. Although common-method variance would usually be a concern with such data, the moderation effects we observed suggest that our data were not strongly affected by this bias, because the relationships between our independent and dependent variables were not significant at low levels of rumination; if common-method variance was a problem, all relationships would be artificially inflated. However, the correlational design does mean that our data do not permit causal inferences. The design of the study also means that we cannot rule out some alternative explanations for the results we observed; for example, trait ruminators might be more likely to share personality characteristics like neuroticism that are thought to predict poorer health and well-being.

GENERAL DISCUSSION

  1. Top of page
  2. Abstract
  3. Background
  4. STUDY 1
  5. Method
  6. Results
  7. Discussion
  8. STUDY 2
  9. Method
  10. Analysis
  11. Results
  12. Discussion
  13. GENERAL DISCUSSION
  14. Acknowledgement
  15. References

Workplace violence can have a devastating impact on victims. In this article, we argue that ruminative thinking is a maladaptive response style that can worsen the prognosis of employees who are exposed to workplace violence. The results of two studies confirmed this proposition. In Study 1, the effects of simulated workplace violence on unpleasant affect continued among a group of people instructed to ruminate about the violence, but ceased among a group who were distracted from the violence. In Study 2, reported exposure to real workplace violence was more strongly associated with poor well-being and health complaints among people who had a higher tendency to engage in ruminative thinking.

The present research contributes to our understanding of workplace violence in three important ways. First, Study 1 shows that the acute negative effects of violence are only likely to persist when the victim engages in ruminative thinking, as opposed to a more adaptive form of response type, such as distraction. Thus, state rumination may at least partly explain the negative effects of workplace violence, most likely because it continues the activation of stress appraisals. Second, Study 2 identifies a key individual characteristic likely to affect the impact violence has on victims: trait rumination. Those people who tend to respond to stressful events with a ruminative thinking style may therefore be most susceptible to the negative effects of workplace violence. Third, our research provides a general theoretical framework, which proposes that it is the interplay between exposure to workplace violence and victims' response types that influences the effects violence has on its victims. This framework not only offers a theoretical explanation for the current findings about rumination, but can also be used to interpret the results of prior research concerning other response types (e.g., thoughts of revenge; Moreno-Jiménez et al., 2009; and reappraisal; Niven et al., in press).

The present research also contributes methodologically. Although experimental designs are increasingly used to study applied phenomena, thus far there have been relatively few experimental studies of workplace violence. Those available typically adopt a vignette approach in which participants are given written scenarios and asked to indicate on a rating scale how they would respond to the situation (e.g., Hershcovis & Barling, 2010), but research has revealed that participants feel more engaged and respond more authentically to videos of simulated violence (Noel et al., 2008). Our laboratory study therefore provides a more ecologically valid design for examining workplace violence experimentally and could be adapted to study other types of responses to violence, for example, thoughts of reconciliation or revenge. Future studies may wish to design equivalent videos for female and male participants for such purposes.

The approach of combining controlled and naturalistic study designs to test the central proposition of the research allowed us to overcome the main limitations associated with each type of design. In the controlled study, we manipulated key variables and so overcame the problem of disentangling causality associated with correlational research. Moreover, random assignment to conditions helped to rule out alternative explanations (e.g., relating to personality traits like neuroticism) that could have explained the findings we observed in our field study. In the naturalistic study, we examined spontaneous rumination tendencies, and investigated workplace violence in a sample directly affected by the phenomenon and so overcame the artificial nature of laboratory research.

Nevertheless, some limitations remain. The primary limitation concerns the applicability of our findings to men as well as women. Although gender was unrelated to the key variables of interest in Study 2 and the pattern of our findings was not affected when this variable was statistically controlled, Study 1 only involved women, and so it is possible that different findings would be observed in a male sample. A second limitation regards the role of underlying individual differences in explaining our findings. Although we randomized participants to conditions in Study 1, we did not explicitly measure potentially relevant traits like neuroticism; as such, it is possible that differences in such traits might partly explain the patterns of findings we observed across our studies. Finally, the duration of time over which ruminative thinking might prolong the negative effects of exposure to violence is unclear. In our field study, we found an association over a 6-month period, but this was a correlational rather than causal relationship, and in the laboratory study, we only studied rumination over a 2-min period. Future research should therefore investigate the longer-term consequences of rumination following exposure to workplace violence among men and women, controlling for potential confounding variables.

Another important direction for future research will be to consider other possible effects of rumination for victims of workplace violence. For instance, poor well-being and health complaints have been linked to withdrawal behaviours and poor performance (Darr & Johns, 2008; Wright & Cropanzano, 2000); might ruminative thinking in response to workplace violence then have wider-reaching implications? Future studies should also consider other possible response types that victims of workplace violence might engage with. In the present research, distraction is highlighted as one alternative that may be more adaptive (echoing earlier studies, e.g., Nolen-Hoeksema & Morrow, 1993), but our theoretical model implies that any response type that deactivates a person's stress appraisals might prove a functional alternative.

Notwithstanding the limitations we have highlighted, our findings have important practical implications. For employees who are subjected to violence at work, our findings suggest the need to refrain from rumination. As a relatively stable thinking style, it may not be easy for people who typically ruminate to prevent this type of cognition (Denson, 2009), but laboratory investigations suggest that focusing on distractions might help to reduce negative symptoms, even among those with a high tendency to ruminate (Nolen-Hoeksema et al., 2008). It should, however, be noted that suppressing thoughts and feelings about traumatic events may require psychological work and in itself lead to negative consequences (Pennebaker & Susman, 1988). Thus, while distraction may provide immediate relief, victims should not necessarily avoid all thoughts about violence, merely ruminative-style thoughts.

For organizations, the findings of the present research must be taken somewhat tentatively, as replication (especially among mixed-gender samples) is necessary before the findings are used to inform organizational policy. However, our findings do suggest that any interventions organizations introduce in the hope of improving victims' prognoses (e.g., counselling, debriefing) should attempt to promote thinking styles that are positive, active and concrete, as opposed to negative, passive and abstract, in order to avoid the potential traps of rumination.

Footnote
  1. 1

    Although gender was not correlated with the core study variables, with the exception of health complaints (= −.27, < .05), because it was a point of interest in the present study (as Study 1 involved a female-only sample), we repeated our moderation analyses with gender included as a control variable at Step 1. These additional analyses revealed the same pattern of results as those reported in the main text. The interaction between rumination and violence still explained a significant amount of variance in health complaints, ΔR2 = .07, ΔF(1, 78) = 7.55, < .01, although the effect on poor well-being was reduced to marginal significance, ΔR2 = .03, ΔF(1, 78) = 3.00, = .09. In both analyses, gender was a non-significant predictor at the final step of analysis in which the significance of the interaction was tested (βs < ±.19, ps > .05).

Acknowledgement

  1. Top of page
  2. Abstract
  3. Background
  4. STUDY 1
  5. Method
  6. Results
  7. Discussion
  8. STUDY 2
  9. Method
  10. Analysis
  11. Results
  12. Discussion
  13. GENERAL DISCUSSION
  14. Acknowledgement
  15. References

We gratefully acknowledge the support of the Economic and Social Research Council (ESRC) UK (RES-060-25-0044: ‘Emotion regulation of others and self [EROS]’) and the Institution of Occupational Safety and Health (IOSH) (R/0707/2: ‘The effect of work-related violence on employee health and well-being’). Thanks also to Ruth Giles for her help with data collection in Study 2.

References

  1. Top of page
  2. Abstract
  3. Background
  4. STUDY 1
  5. Method
  6. Results
  7. Discussion
  8. STUDY 2
  9. Method
  10. Analysis
  11. Results
  12. Discussion
  13. GENERAL DISCUSSION
  14. Acknowledgement
  15. References