Effects of Violence Prevention Behavior on Exposure to Workplace Violence and Threats: A Follow-Up Study

Abstract

This longitudinal study investigates the relationship between prevention behaviors, that is, enacted violence prevention policies, and exposure to workplace violence and threats across four different high risk work sectors: psychiatry, special schools, eldercare, and the prison and probation services. Logistic regression analysis of a 1-year follow-up sample of 3.016 employees from these four sectors shows that prevention behaviors are significantly and negatively associated with self-reported exposure to workplace violence and threats—in the prison and probation services, eldercare, and in psychiatry, while no significant associations are found for special schools. The results therefore show clear sector differences with regard to the preventive effect of violence prevention behaviors. Furthermore, this multisector comparison suggests that prevention behaviors are more effective in relation to a moderate frequency of violence and threats, and that only top management prevention behavior can prevent very frequent incidents (odds ratio [OR] = 0.58). This study contributes to the literature by use of a longitudinal design and acceptable response rates, while also simultaneously investigating several high risk sectors. The results imply that when managing workplace violence in high risk areas of human service work, there should be emphasis on the use of violence prevention behaviors from top management, supervisor, and among coworkers. However, type of sector and the frequency of workplace violence should be analyzed to evaluate the potential impact of prevention behaviors.

Keywords

threats physical violence human service work follow-up study prevention behavior

Introduction

Workplace violence and threats of violence are considered one of the principal occupational health hazards for many people at work (Leather & Zarola, 2010). Reviews have shown that exposure to workplace violence and threats is particularly high in service and human service sectors, such as health care, education, public safety, retail, and justice industries (Hogh & Viitasara, 2005; Piquero, Piquero, Craig, & Clipper, 2013; Spector, Zhou, & Che, 2014). The literature shows exposure rates as high as 66.9% for nonphysical violence and 36.4% for physical violence for nurses (Spector et al., 2014), and among special educators, 42.3% for nonphysical violence and 21.7% for physical violence (Tiesman, Konda, Hendricks, Mercer, & Amandus, 2013). Furthermore, among U.S. correctional officers, from 1999 to 2008, there were 125.200 nonfatal injuries and 113 fatal injuries of which 38% to 40% were due to assaults and violent acts (Konda, Reichard, & Tiesman, 2012).

These exposure estimates all warrant preventive action; however, a review of the literature on intervention effectiveness shows that there is much variability among studies in the types and effectiveness of intervention (Wassel, 2009). In fact, training has both been related to 50% decreases in violent interaction at 3-month follow-up (Fernandes et al., 2002) and twofold increases in the frequency of assaults when comparing data 12 months prior to the training and 11 months after training (Wilkinson, 1999). A study on organizational policies showed significant reduced risk (odds ratio [OR] = 0.5) related to “zero tolerance” for violence and a list of prohibited violent behaviors (Nachreiner et al., 2005). However, the policy of “zero tolerance” has also been related to an increase of reporting incidents by 85% during the following year (www.arbejdsmiljoviden.dk, 2013). Thus, more research is needed to uncover effective prevention of workplace violence.

Over more than 30 years, a strong theoretical and empirical base has been developed under the rubric of safety climate, which refers to shared perceptions among members of the organization regarding safety policies, procedures, and practices (Zohar, 1980, 2010; Zohar & Luria, 2003, 2005). Evidence shows that safety climate is important for predicting individual safety behavior, industrial accidents, and treatment errors in health care (Cooper & Phillips, 2004; Johnson, 2007; Naveh, Katz-Navon, & Stern, 2005; Silva, Lima, & Baptista, 2004). Spector, Coulter, Stockwell, and Matz (2007) were the first to extend the idea of safety climate to the context of workplace violence, and they coined the term perceived violence prevention climate (Spector et al., 2007). Their survey of 198 female nurses from a U.S. hospital showed that violence prevention climate was significantly and negatively related to violence, verbal aggression, injury, and perceived danger. Kessler, Spector, Chang, and Parr (2008) refined the violence prevention climate construct by developing it as a multidimensional construct: policies and procedures, practices and responses, and pressure for unsafe practice (Kessler et al., 2008). The policies and procedures dimension refers to an employee’s awareness of the formal regulations and formal distribution of information, such as through training. The practices and responses dimension refers to how management adheres to these formal regulations and their response to violent incidents. The pressure for unsafe practice dimension refers to employees’ perception of pressure to ignore the violence prevention policies and procedures to meet other demands. Thus, policies and procedures represent espoused priorities and desired behaviors, whereas practices and responses represent the enacted counterpart. Kessler and colleagues suggested that practices may supersede policies when management chooses to ignore or even contradict policy. Their web-based survey of 216 full-time employees in the United States, from a variety of jobs and work sectors, showed that the dimension of practices and responses was the most important predictor of physical violence, whereas the dimension of policies and procedures was more relevant for exposure to verbal aggression. They concluded that the response of supervisors to violent behaviors might well be more effective in curtailing violence than prevention policies alone. Chang, Eatough, Spector, and Kessler (2012) conceptualized poor violence prevention climate as a stressor associated with increased strain and reduced prevention motivation (Chang et al., 2012). The results from their web-based survey of 172 employee–coworker dyads from a variety of organizations in the United States suggested that the practices and responses dimension is the most important element, considering that it was linked to compliance with and participation in prevention, through both strains and motivation. Again, this finding was related to the importance of management reactions to assaults. The above studies on violence prevention climate are limited due to the use of cross-sectional designs, questionable heterogeneity, and lack of adequate information about response rates. However, one longitudinal survey on violence prevention climate has been conducted (Yang, Spector, Chang, Gallant-Roman, & Powell, 2012). This study found that only the dimension of “Pressure for unsafe practice” was related to risk of being exposed to physical violence over 6 months (OR = 1.69). This result contradicts findings from the former studies, where the dimension of practices and responses is seen as the more important element of violence prevention climate. However, there are important limitations to this longitudinal study. Although they invited 1565 nurses to participate, only 176 nurses completed both surveys. Thus, there is still a need for a longitudinal study examining the effects of violence prevention climate on exposure to workplace violence, particularly investigating the potential effects of prevention behaviors inherent to the dimension of practices and responses.

Although both Kessler and colleagues (2008) and Chang and colleagues (2012) emphasize the response to incidents of violence to effectively curtail violence, the dimension of practices and responses does not include social support after incidents of violence. This is surprising, considering findings on the important role of social support from supervisor (and coworkers) after exposure to violence and threats (Leather, Lawrence, Beale, & Dickson, 1998; Schat & Kelloway, 2003). For example, the cross-sectional survey study by Schat and Kelloway (2003), conducted among 225 employees in a Canadian health care setting, found that supervisor and coworker support after incidents of violence can alleviate the negative effects on employees’ emotional well-being, somatic health, and job-related affect. Findings from follow-up studies on workplace violence show that prior exposure significantly increases the risk of reoccurrence (Beale, Clarke, Cox, Leather, & Lawrence, 1999; Hogh, Sharipova, & Borg, 2008; Nachreiner et al., 2012). Thus, a history of workplace violence is of concern, which warrants preventive attention. In theory, social support may reduce future exposure by alleviating the victim’s symptoms of strain. These various strains may reduce capabilities to comply with prevention policies (Chang et al., 2012) and may increase aggressive outbursts due to a lack of mental resources (Aquino & Thau, 2009; Felson, 1992). In sum, theory and evidence suggests that effective violence prevention policies and the enacted counterpart, prevention behaviors, should include social support after incidents of workplace violence.

Kessler and colleagues (2008) and Chang and colleagues (2012) also emphasize the role of management to effectively curtail violence. However, the violence prevention climate construct does not distinguish between levels of management, for example, supervisor or top management. Evidence from research on safety climate has shown that both top-level management and supervisor-level management can influence employee behavior (Zohar & Luria, 2003, 2005). In particular, supervisory safety practices should be supported by higher level management through communication of safety priorities, even under increased work pressure. In the context of workplace violence, top management may prevent future employee exposure to violence directly by, for example, considering prevention in decisions concerning staffing and intake of clients. In contrast to top management, supervisors have day-to-day interaction with employees and may thus directly affect future exposure by encouraging formal reporting of incidents and taking reports seriously, thereby informing specific prevention strategies. Moreover, the above results concerning social support also provided evidence for the potential preventive role of coworkers (Schat & Kelloway, 2003). The influence of coworkers can be of considerable strength if informal groups or informal leadership have developed; these informal work groups serve as guides to correct behavior and can exercise pressure to conform to group standards and norms (Hussein, 1989; Schein, 2010). Thus, if an informal group of coworkers value violence prevention, this may directly affect the frequency of engaging in prevention behaviors, thereby also influencing the potential preventive effects of enacted policies.

No studies on violence prevention climate have been stratified to test for sector differences, which is likely to overlook structural and organizational differences. Viitasara and Menckel (2002) described a structural level of risk factors associated with workplace violence, which involves type of local-government unit to which the organization belongs, management, direction and control, policy, financing, both physical and psychosocial work environments, personnel, and education/training. Although human service sectors broadly are related to increased risk of workplace violence (Hogh & Viitasara, 2005; Piquero et al., 2013), structural factors in these sectors may vary considerably as a function of specific interpersonal challenges related to type of client and treatment goals. Indeed, a baseline study of the current sample of four high risk sectors showed significant differences in the frequency of threats and violence (Rasmussen, Hogh, & Andersen, 2013), thus implying plausible effects related to sector. The current study, therefore, contributes to the literature by investigating more specific prevention effects by stratifying for sector differences.

In sum, while the practices and responses dimension has much to offer with regard to analyzing enacted polices and the potential preventive effects, there are empirical and theoretical reasons for refining categories; this concerns specifying prevention behaviors at different levels within an organizational hierarchy, for example, coworkers, supervisors, and top management, and including the aspect of social support. The current study allows for a multisector comparison investigating the following research questions:

Research Question 1: What is the association between violence prevention behaviors and self-reported exposure to violence or threats at follow-up?

Research Question 2: Which differences in the relationship between violence prevention behaviors and self-reported exposure to violence or threats exist across different work sectors?

Method/Procedure

Participants

The study population consists of a follow-up sample. The baseline study, described in Rasmussen and colleagues (2013), included 5,497 employees from psychiatry, special schools, eldercare, and the prison and probation services. Altogether, 3,584 employees participated in the follow-up study with an overall follow-up response rate of 65% (see Figure 1). However, seeing that this study revolves around the perception of management, we further excluded all those with supervisory responsibilities at both line and top management level. Thus, the study sample consists of psychiatry (n = 617), special schools (n = 511), eldercare (n = 577), and the prison and probation services (n = 1,311): a total of 3,016 participants.

Figure 1.

Flow chart of data collection and sample.

Data Collection

A web-based questionnaire was used for participants in the prison and probation services; participants from the other work sectors completed written questionnaires during a planned meeting at the worksite. It was stated in the cover letter of the questionnaire that participation in the study was voluntary and that the data would be treated confidentially. The baseline data collection took place in the period between May and October 2010, and the follow-up was conducted in the same period in 2011. The study was approved by the Danish Data Protection Agency, and followed the regulations for data storage and protection. Respondents were identified by questionnaire numbers, which only members of the research group could link to civil registration numbers. This procedure was to ensure accurate matching of questionnaires from the two rounds of data collection.

Measures

Prevention behaviors measured at baseline

Three items inspired by Zohar and Luria’s (2005) scale for measuring Organizational-Level Safety Climate were used to measure employees’ perception of top management prevention behavior. Perception of supervisor prevention behavior was measured with four items from Spector and colleagues (2007). We added an item concerning supervisory support inspired by Vegchel, Jonge, Söderfeldt, Dormann, and Schaufeli (2004). Coworker prevention behaviors were construed similar to items on supervisor behavior (Spector et al., 2007), with the exception of this one item “. . . takes reports of workplace violence seriously.”

Items were translated into Danish and supplemented with a context-specific introduction using organizational terms for the relevant supervisor or top management. Respondents were asked to rate how much they agreed with the statement; using a 5-point response scale ranging from not at all to very high degree supplemented with do not know. The do not know response category was used in Spector and colleagues (2007) and scored in between No” and “Yes.” Conceptually, this means that being unaware of prevention behaviors represents a weak practice concerning violence prevention. For the current study, however, respondents primarily replying do not know were excluded from the main analyses to avoid confounding between being unaware of violence prevention and being aware that there is a lack of violence prevention. The former may be a mere lack of awareness on the part of the individual (possibly about prevention initiatives that are occurring). The latter involves an organization that is failing to engage in violence prevention. The former (lack of awareness), would be a concern, but a worker in an environment where prevention is taking place, but of which he or she is unaware, may still be protected from violence by virtue of the organization taking steps to prevent it. However, in the latter context, there is a lack of prevention, which may put the worker at risk. The exclusion of do not know followed the general procedure for missing, in that respondents must answer more than half of the items in a scale to be included in that scale. Thus, the three scales were scored from 0 (not at all) to 4 (very high degree). Items and Cronbach’s alpha for the three scales are shown in Table 1.

Table 1.

Violence Prevention Behaviors.

Top management	Requires each manager to help improve violence prevention in his or her departmentInvests a lot of time and money in violence prevention training for workersConsiders violence prevention in decisions concerning staffing and intake of clients

Supervisor	Your supervisor encourages staff to report physical violenceYour supervisor encourages staff to report threats Your supervisor takes reports of workplace violence seriouslyYour supervisor gives sufficient help and support after a violent or threatening incident

Coworker	Among coworkers you are encouraged to report physical violenceAmong coworkers you are encouraged to report threats Your coworkers give sufficient help and support after a violent or threatening incident

Note. Items were supplemented with a context-specific introduction using organizational terms for the relevant supervisor or top management. Cronbach’s alpha across sectors: .7 to .9 for top management, .8 to .9 for supervisor, and .8 and .9 for coworker.

Threats of violence and physical violence measured at follow-up

Respondents were asked whether they had experienced the following acts of threatening behaviors at their current worksite within the past 12 months: threats of beatings, written threats, threats in a scolding manner, threats in an insulting manner, threats over the phone, threats involving objects, and indirect threats (toward family). Similarly, respondents were asked whether they had experienced the following acts of physically violent behaviors: spitting, hitting, hitting with object, scratching/pinching, shoving, being held, punching with a fist, kicking, biting, having a hard object thrown at you, and use of a weapon or weapon-like object (Menckel & Viitasara, 2002). Response categories were 1 = yes, daily; 2 = yes, weekly; 3 = yes, monthly; 4 = yes, now and then; and 5 = no, never. Using this as a continuous measure, we saw problems with the statistical assumptions of normality and homoscedasticity; therefore, the scale was dichotomized. We chose to dichotomize at the 75th percentile recognizing that in these high risk work sectors, some exposure to violence or threats is widespread. Thus, the analyses would identify associations with high exposure relative to low exposure. However, due to low frequencies of physical violence in the prison and probation services, the violence scale was here collapsed into two categories: 1 = yes, exposed (daily, weekly, monthly, now and then), and 0 = not exposed (never). A descriptive account of the relative frequencies of workplace violence across sectors is given in Table 1, using types of violent and threatening behaviors as two continuous measures scored from 0 to 100. Cronbach’s alphas across sectors were between .7 and .9 for both scales.

The exposed participants were asked who the perpetrator(s) were: “A client,” “A relative to the client,” “A coworker,” “A superior,” “A subordinate,” and “Other people.” This item was made to fit the specific work sector by exchanging “client” with either “pupil,” “patient,” “elderly citizen,” or “inmate.” It was possible to report several perpetrators.

Covariates measured at baseline

While the analyses were stratified for sector differences, we included controls for individual factors such as gender and seniority, as these have been related to risk of workplace violence (Hogh & Viitasara, 2005; Lawoko, Soares, & Nolan, 2004). Seniority, in particular, may also affect engaging in violence prevention due to more experience concerning policies and participation in various training programs. We further considered controlling for working permanent night or evening shifts, thus controlling for supposed differences in staff–supervisor interaction rates. However, correlation analyses (not shown) revealed no significant correlations with any of our three predictors (prevention behaviors). Therefore, time of work was not included in the main analyses.

Data Analyses

The data analyses aims to explore the association between violence prevention behaviors and self-reported exposure to violence or threats at follow-up, while also allowing for a multisector comparison. Thus, all analyses were conducted separately for each sector, but statistical procedures were the same. As we sought to identify predictors of binary measures of workplace threats and violence, we analyzed our data using logistic regression. Associations were estimated by OR and 95% confidence intervals. We were particularly interested in whether the OR was less than 1 indicating that strong prevention behavior—as opposed to weak behavior—was negatively related to high violence or threats—as opposed to low exposure. Gender, seniority, and baseline exposure were included in all analyses. SPSS 20 was used to conduct the statistical analysis.

Results

In eldercare, psychiatry, and special schools, 99% to 100% of threats and violence were perpetrated by clients, while in the prison and probation services 94% of threats and 98% violence were perpetrated by clients. Thus, across sectors, exposed participants reported clients to be responsible for almost all incidents of workplace violence. The descriptive account of the relative frequency of threats and violence across sectors shows that special schools have the highest mean values of both threats and violence (Table 2). Psychiatry has the second highest overall mean levels of threats and violence. The prison and probation services have a higher mean value for threats in comparison with eldercare, while eldercare has a higher mean value of violence in comparison with the prison and probation services. This pattern is similar to baseline findings (Rasmussen et al., 2013). Descriptive results on prevention behaviors show that top management behavior has the lowest mean level in all sectors (see Table 2). In psychiatry, special schools, and the prison and probation services, coworker prevention behavior has slightly higher means than supervisor behavior.

Table 2.

Descriptive Statistics for Main Study Variables.

	Eldercare	Psychiatry	Special Schools	Prison and Probation Services
Total N (follow-up study sample)	577	617	511	1,311
Female: n (%)	97.0	79.2	72.0	50.0
Age: M (SD)	47 (10)	45 (11)	43 (9)	43 (10)
Seniority: M (SD)	9 (9)	8 (8)	6 (5)	9 (8)
Violence and threats of violence
Violence scale, follow-up: M (SD)	2.7 (6.6)	6.1 (8.0)	13.7 (13.9)	0.9 (2.9)
Threat scale, follow-up: M (SD)	4.9 (8.4)	14.9 (13.0)	15.8 (14.2)	7.5 (9.3)
High level, violence at follow-up^a: n (%)	108 (19.3)	128 (22.5)	113 (22.6)	210 (16.2)^b
High level, violence at baseline^a: n (%)	117 (21.6)	118 (19.8)	115 (23.0)	194 (15.1)^b
High level, threats at follow-up^a: n (%)	159 (28.3)	144 (24.0)	122 (24.4)	292 (22.5)
High level, threats at baseline^a: n (%)	126 (23.4)	144 (24.5)	97 (19.7)	274 (21.5)
Violence prevention behaviors
Top management: M (SD)	2.3 (1.2)	2.5 (0.9)	2.2 (1.0)	1.2 (1.2)
Supervisor: M (SD)	3.3 (0.7)	3.3 (0.7)	3.1 (0.7)	3.3 (0.8)
Coworker: M (SD)	3.3 (0.8)	3.5 (0.7)	3.3 (0.7)	3.5 (0.7)

Note. Data on gender, seniority, and education were retrieved from the baseline survey. For analytical purposes, the dependent variable was dichotomized. We used the 75th percentile split for these variables measured both at T1 and T2.

Due to missing values, the total N for each variable may deviate from the total N of the study population.

Due to low frequencies of physical violence in the prison and probation services, the physical violence scale was dichotomized at yes-no, in contrast to the 75th percentile.

The logistic bivariate analyses of the odds of physical violence by prevention behavior, gender, seniority, and baseline violence yielded some statistically significant associations (Table 3). In psychiatry, all three types of prevention behaviors were significantly associated with lower self-reported exposure to violence (ORs from 0.52 to 0.72). In the prison and probation services and eldercare, both supervisor and coworker prevention behaviors were significantly associated with lower self-reported exposure to violence (ORs from 0.53 to 0.78). No significant associations were found for special schools.

Table 3.

Logistic Regression Analyses of the Association Between Prevention Behaviors and Exposure to Physical Violence and Threats of Violence.

	Psychiatry		Special Schools		Prison and Probation Services^a		Eldercare
	OR	95% CI	OR	95% CI	OR	95% CI	OR	95% CI
Physical violence
	n = 413		n = 230		n = 708		n = 186
Top management prevention behavior	0.69	[0.52, 0.92]	1.01	[0.71, 1.44]	0.90	[0.74, 1.10]	0.94	[0.91, 1.03]
Model χ²(4)	83.37***		51.32***		159.37***		29.54***
R² (Nagelkerke)	.27		.31		.33		.24
	n = 525		n = 410		n = 1,026		n = 390
Supervisor prevention behavior	0.52	[0.30, 0.89]	1.05	[0.74, 1.48]	0.78	[0.63, 0.98]	0.53	[0.35, 0.80]
Model χ²(4)	105.01***		64.70***		234.89***		71.60***
R² (Nagelkerke)	.27		.22		.34		.32
	n = 559		n = 468		n = 1,136		n = 460
Coworker prevention behavior	0.72	[0.52, 0.98]	1.00	[0.72, 1.39]	0.77	[0.60, 0.99]	0.61	[0.44, 0.84]
Model χ²(4)	112.53***		72.53***		239.67***		81.95***
R² (Nagelkerke)	.27		.22		.32		.31
Threats of violence
	n = 383		n = 230		n = 701		n = 189
Top management prevention behavior	0.58	[0.43, 0.80]	0.94	[0.66, 1.34]	0.81	[0.67, 0.98]	0.75	[0.57, 0.996]
Model χ²(4)	107.71***		63.09***		226.87***		44.29***
R² (Nagelkerke)	.36		.35		.40		.28
	n = 487		n = 404		n = 1,013		n = 393
Supervisor prevention behavior	1.30	[0.91, 1.86]	1.10	[0.76, 1.59]	0.58	[0.47, 0.72]	0.64	[0.47, 0.89]
Model χ² (4)	112.15***		91.83***		345.03***		68.84***
R² (Nagelkerke)	.31		.30		.43		.22
	n = 519		n = 462		n = 1,123		n = 466
Coworker prevention behavior	1.01	[0.70, 1.42]	1.33	[0.94, 1.88]	0.62	[0.49, 0.78]	0.68	[0.53, 0.87]
Model χ²(4)	120.83.3***		99.58***		365.84***		85.00***
R² (Nagelkerke)	.31		.29		.42		.22

Note. All analyses are adjusted for seniority, gender, and baseline exposure. OR = odds ratio; CI = confidence interval.

Due to low frequencies of physical violence in the prison and probation services, the violence scale was dichotomized at yes-no, in contrast to the 75th percentile

p < .05. **p < .01. ***p < .001.

The logistic bivariate analyses of the odds of threats by prevention behavior, gender, seniority, and baseline threats also yielded some statistically significant associations (Table 3). In the prison and probation services and eldercare, all three types of prevention behaviors were significantly associated with lower self-reported exposure to threats (OR from 0.58 to 0.81), although top management prevention behavior in the eldercare was borderline significant (OR = 0.75, CI = [0.57, 0.996]). In psychiatry, top management prevention behavior was significantly associated with lower self-reported exposure to threats (OR = 0.58). No significant associations were found for special schools.

Discussion

The results of this longitudinal study demonstrate that violence prevention behaviors are significantly and negatively associated with lower self-reported exposure to violence and threats of violence at follow-up (Research Question 1). This corresponds well with and extends results from existing cross-sectional studies (Chang et al., 2012; Gimeno, Barrientos-Gutierrez, Burau, & Felknor, 2012; Kessler et al., 2008; Lipscomb et al., 2012; Spector et al., 2007). Furthermore, the current results show similar and different trends within and across sectors with regard to the preventive effects of prevention behaviors (Research Question 2). In psychiatry, prevention behaviors overall had a more preventive effect on physical violence in comparison with threats of violence. This is in line with results from Kessler and colleagues (2008) who found that the dimension of practices and responses was a more important predictor for physical violence than verbal aggression. Furthermore, in psychiatry, top management prevention behavior was the only prevention measure that affected both violence and threats, while in eldercare and the prison and probation services, top management was the only prevention measure that did not affect both violence and threats. Moreover, results on eldercare and the prison and probation services showed that prevention behaviors overall had a more preventive effect on threats in contrast to violence. This differed from results on psychiatry. No results on special schools were significant and only by controlling for the baseline measure of self-reported exposure did model-fit indices reach significant χ². This suggests that prevention behaviors do not fit the hypothesized model indicating that this work sector is qualitatively different from the other three sectors.

The different effects of prevention behaviors may reflect sector-specific profiles with regard to structural or organizational qualities and frequencies of threats and violence. The high frequencies of threats and violence in special schools, implying that most employees are at high risk (Rasmussen et al., 2013), may explain why the current analyses did not find significant associations with high exposure contrary to low exposure. Moreover, at the time of data collection, there was limited knowledge on the scope of workplace violence in special schools, which may correspond to a less distinct tradition for prevention strategies in comparison with the other three sectors. Later studies, however, have shown that special education teachers are at a significant higher risk in comparison with other education workers (Gerberich et al., 2011; Tiesman et al., 2013). Furthermore, employees in special schools may regard themselves as caregivers in a sort of parenting way, and the student as someone to protect and not to be protected from. Labeling or pushing aside incidents as violence has been studied by Åkerström (2002), who describes a tendency to determine what a phenomenon is according to who is involved instead of what he or she does. How clients are typified may influence the interpretation of actions thereby also affecting (preventive) reactions. Thus, responding to threatening or violent behavior by a special school student may be perceived in terms of learning objectives in contrast with violence prevention.

In psychiatry, prevention was overall more effective with regard to violence than it was with regard to threats, which may reflect that it is harder to prevent frequently occurring threats. It may be that the frequency of threats is related to staffing norms and the intake of patients, thereby amenable only to top management prevention behavior (OR = 0.58). Overcrowding and staffing norms have been widely debated in Denmark due to many financial cutbacks in hospitals and long wait-list for psychiatric treatment. Another possible explanation may be that formal reporting of very frequent incidents is very time consuming, and therefore, it is tacitly accepted that this is not feasible or necessary. Researchers have suggested that if there is a high frequency of incidents, then it might be appropriate to periodically have “reporting weeks,” where employees are requested to report all incidents (Beale, Cox, & Leather, 1996), thus giving a truer picture of the day-to-day frequency of threats and thereby informing relevant prevention strategies.

In psychiatry, supervisor prevention behavior, surprisingly, was positively associated with self-reported exposure to threats (OR = 1.30), albeit this association was not significant. Thus, although, supervisors may be attentive to violence prevention by encouraging reporting and giving support, it does not prevent exposure to threats. Instead, those employees experiencing supervisor prevention behaviors may also be those consistently being exposed to threats. In contrast, safety research suggests that more frequent supervisory safety-oriented interaction will increase workers’ safety behaviors (such as use of protective gear), and that failure to use protective gear accounts for 40% of work accidents (Zohar & Luria, 2003). Thus, the differential effect of supervisory prevention behavior may relate to available intervention and effectiveness of this intervention. The prevention of workplace threats, in contrast to industrial accidents, does not have a technical or an easy accessible solution, such as using protective gear, but involves more complex and time-consuming efforts with varying degrees of effectiveness (Wassel, 2009). The positive association between prevention and the reporting of incidents has also been found in an intervention study by Arnetz and Arnetz (2000), in which the intervention group reported 50% more violent incidents than their control group. The authors suggested that this increase in reporting was related to a heightened awareness and possibly a supportive environment to sustain reporting of incidents. This reflects the complexity in measuring the effectiveness of violence prevention.

The similar trends of preventive effects of prevention behaviors between the prison and probations services and eldercare may also reflect their more similar frequencies of threats and violence. Both sectors have nonsignificant associations between top management prevention behavior and violence, suggesting that with mean exposure levels below 2.7 only supervisors and coworkers are involved in prevention behaviors. Furthermore, the mean levels of threats in eldercare (4.9) and the prison and probations services (7.5) were comparable to the mean level of violence in psychiatry (6.1), suggesting that prevention behaviors are more effective with a moderate frequency of incidents. Thus, supervisor and coworker prevention behavior may be effective with a relatively moderate degree of violence or threats, while only top management may affect relatively high levels of threats or violence and may not have any effect on relatively low levels of violence.

In the current study, we argue that respondents replying primarily do not know to items on the violence prevention scales should be excluded from the main analyses to avoid confounding between being unaware of violence prevention, and being aware that there is a lack of violence prevention. However, this negatively affected the size of the analytical sample, particularly with respect to the dimension of top management. It has been suggested that nonresponse to questions concerning management may be related to people working in loose networks with changing leaders, several leaders, or no leaders, implying that questions assuming the “normal” hierarchy at the workplace may face increasing problems with missing values (Kristensen, Hannerz, Hogh, & Borg, 2005). For the current study, we did some supplementary exploratory analyses (data not shown) to ascertain possible bias as a consequence of many missing due to the exclusion of do not know on prevention measures. A comparison on demographics between included and excluded respondents showed that the latter consistently had lower seniority. However, this may not have biased our findings seeing that we adjusted for seniority in our main analyses.

Our measures of prevention behaviors are closely related to the practices and response dimension used in studies on violence prevention climate (Kessler et al., 2008; Spector et al., 2007; Yang et al., 2012). The only longitudinal study did not find a statistically significant association between prevention practices at Time 1 and violence exposure at Time 2 (6-month follow-up). The current study, therefore, contributes to the literature by showing significant associations using a large heterogenic sample and follow-up data with acceptable response rates. Although our prevention behavior measures are closely linked to the measures used for the dimension of practices and responses, they are not identical. The current study specified prevention behaviors at different levels within an organizational hierarchy, for example, coworkers, supervisors, and top management, and included the aspect of social support. Descriptive results showed that coworker prevention behavior had slightly higher means, suggesting that employees perceive coworkers to be the most engaged in prevention behaviors. In terms of informal groups, this type of behavior would be considered a collaboration with the formal organization as a result of a common interest (Hussein, 1989). The preventive effect was parallel to the effect of supervisor behavior, although the latter had a slightly stronger influence on self-reported exposure at follow-up. This again emphasizes the role of supervisors to effectively prevent workplace violence and threats, in line with studies on violence prevention climate (Chang et al., 2012; Kessler et al., 2008). Furthermore, it is possible that this study has found more significant effects in comparison with Yang and colleagues (2012) due to the inclusion of social support. By including items on social support, this study expands on existing evidence for the important role of supervisor and coworker responses after exposure to violence and threats (Leather et al., 1998; Schat & Kelloway, 2003). Thus, the present results suggest that violence prevention policies and the enacted counterpart should involve social support from coworkers and supervisors after incidents of threats and violence.

Strengths and Limitations

A significant strength of the present study is the use of a follow-up sample with acceptable response rates at both baseline and follow-up. Another significant strength is that data allowed for a multisector comparison, which provided novel longitudinal evidence for sector differences with regard to the association between prevention behaviors and workplace violence and threats. Also, using a longitudinal design reduces bias among measures.

However, the study also has limitations. Using only the survey method induces risk of single source bias, however, this approach is consistent with studies on workplace violence and prevention climates (Gimeno et al., 2012; Kessler et al., 2008; Lipscomb et al., 2012; Spector et al., 2007; Yang et al., 2012). Future studies may further explore our findings by use of interviews and intervention designs. Also, dichotomizing measures means that we lose some information and therefore we are perhaps simplifying the relationships in question. However, dichotomizing also has some advantages (Farrington & Loeber, 2000). For example, it makes the interpretation of the results more intuitive because we do not have to translate what a step on a scale equals in practical terms. Furthermore, most research on workplace violence and safety or violence prevention climate has studied individual experiences of climate, despite climate being conceptualized as shared perception across individuals (Zohar, 1980, 2000, 2010; Zohar & Luria, 2005). This limitation also has some merit for our findings, though we only explore one aspect of a possible aggregate-level climate, namely, prevention behaviors. Finally, while the entire sector of the prison and probation services participated, data from the other three sectors were collected by use of nonrandom sampling, also described in Rasmussen and colleagues (2013). Although this implies that we cannot rule out selection bias, the results show similar findings between particularly eldercare (nonrandom sampling) and the prison and probation services (entire population), which suggests that sampling procedures may not have inferred substantial bias for the associations explored.

Conclusion and Implications

This longitudinal study shows that enacted prevention policies, that is, prevention behaviors, at top management level, supervisor level, and among coworkers are associated with lower self-reported exposure to workplace violence and threats—in the prison and probation services, eldercare, and in psychiatry, while no significant associations were found for special schools. This multisector comparison showed that prevention behaviors overall are more effective with a relatively moderate frequency of exposure, as seen for threats in eldercare and the prison and probation services and for violence in psychiatry. Supervisor and coworker prevention behaviors are also effective for infrequent exposure, such as for violence in eldercare and the prison and probation services. Top management, however, does not affect infrequent exposure, but is the only prevention behavior that affects very frequent incidents (OR = 0.58), such as threats in psychiatry.

In practice, these results imply that high risk sectors should pay special attention to the actual practice of policies, while also specifying prevention behaviors at different organizational levels including the aspect of giving support. The potential impact of prevention behaviors should be evaluated according to type of sector and frequency of workplace violence and threats. The overall results show that violence prevention is a collaborative effort; however, top management should be aware of their unique responsibility concerning preventing frequently occurring incidents.

Footnotes

Acknowledgements

The authors would like to thank the participating worksites for their involvement in the study.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The authors thank The Danish Working Environment Research Fund for financial support for the study.

Author Biographies

Charlotte Ann Gadegaard (former Rasmussen) is a psychologist and PhD fellow. She graduated from Aarhus University in the summer of 2008. She worked as a work and organizational psychologist from 2008 to 2009 and has been working at the Department of Environmental Medicine in Herning since the autumn of 2009; she was enrolled at the University of Copenhagen as a PhD fellow in January 2010.

Lars Peter Andersen is a psychologist and has a PhD in work and organizational psychology. He graduated from Aarhus University in 2001. He got his PhD from Aarhus University in 2006. He worked as a clinical psychologist and researcher at the Department of Environmental Medicine in Herning since 2004.

Annie Hogh is a professor and has a PhD in work and organizational psychology. She graduated from University of Copenhagen in 1994. She got her PhD from University of Copenhagen in 2005. She worked at The National Research Centre for the Working Environment from 1997 to 2009 and at the University of Copenhagen since 2009.

References

Åkerström

(2002). Slaps, punches, pinches—But not violence: Boundary-work in nursing homes for the elderly. Symbolic Interaction, 25, 515-536.

Aquino

Thau

(2009). Workplace victimization: Aggression from the target’s perspective. Annual Review of Psychology, 60, 717-741.

Arnetz

J. E.

Arnetz

B. B.

(2000). Implementation and evaluation of a practical intervention programme for dealing with violence towards health care workers. Journal of Advanced Nursing, 31, 668-680.

Beale

Clarke

Cox

Leather

Lawrence

(1999). System memory in violent incidents: Evidence from patterns of reoccurrence. Journal of Occupational Health Psychology, 4, 233-244.

Beale

Cox

Leather

(1996). Work-related violence—Is national reporting good enough? Work & Stress, 10, 99-103.

Chang

Eatough

E. M.

Spector

P. E.

Kessler

S. R.

(2012). Violence-prevention climate, exposure to violence and aggression, and prevention behavior: A mediation model. Journal of Organizational Behavior, 33, 657-677.

Cooper

M. D.

Phillips

R. A.

(2004). Exploratory analysis of the safety climate and safety behavior relationship. Journal of Safety Research, 35, 497-512.

Farrington

D. P.

Loeber

(2000). Some benefits of dichotomization in psychiatric and criminological research. Criminal Behavior and Mental Health, 10, 100-122.

Felson

R. B.

(1992). “Kick’em when they’re down”: Explanations of the relationship between stress and interpersonal aggression and violence. The Sociological Quarterly, 33, 1-16.

10.

Fernandes

C. M.

Raboud

J. M.

Christenson

J. M.

Bouthillette

Bullock

Ouellet

Moore

C. F

. (2002). The effect of an education program on violence in the emergency department. Annals of Emergency Medicine, 39, 47–55.

11.

Gerberich

S. G.

Nachreiner

N. M.

Ryan

A. D.

Church

T. R.

McGovern

P. M.

Geisser

M. S.

. . . Pinder

E. D.

(2011). Violence against educators: A population-based study. Journal of Occupational and Environmental Medicine, 53, 294-302.

12.

Gimeno

Barrientos-Gutierrez

Burau

K. D.

Felknor

S. A.

(2012). Safety climate and verbal abuse among public hospital-based workers in Costa Rica. Work, 42, 29-38.

13.

Hogh

Sharipova

Borg

(2008). Incidence and recurrent work-related violence towards healthcare workers and subsequent health effects. A one-year follow-up study. Scandinavian Journal of Public Health, 36, 706-712.

14.

Hogh

Viitasara

(2005). A systematic review of longitudinal studies of nonfatal workplace violence. European Journal of Work & Organizational Psychology, 14, 291-313.

15.

Hussein

R. T.

(1989). Informal groups, leadership and productivity. Leadership & Organization Development Journal, 10, 9-16.

16.

Johnson

S. E.

(2007). The predictive validity of safety climate. Journal of Safety Research, 38, 511-521.

17.

Kessler

S. R.

Spector

P. E.

Chang

Parr

A. D.

(2008). Organizational violence and aggression: Development of the three-factor violence climate survey. Work & Stress, 22, 108-124.

18.

Konda

Reichard

A. A.

Tiesman

H. M.

(2012). Occupational injuries among U.S. correctional officers, 1999-2008. Journal of Safety Research, 43, 181-186.

19.

Kristensen

T. S.

Hannerz

Hogh

Borg

(2005). The Copenhagen Psychosocial Questionnaire—A tool for the assessment and improvement of the psychosocial work environment. Scandinavian Journal of Work, Environment & Health, 31, 438-449.

20.

Lawoko

Soares

J. J. F.

Nolan

(2004). Violence towards psychiatric staff: A comparison of gender, job and environmental characteristics in England and Sweden. Work & Stress, 18, 39-55.

21.

Leather

Lawrence

Beale

Dickson

(1998). Exposure to occupational violence and the buffering effects of intra-organizational support. Work & Stress, 12, 161-178.

22.

Leather

Zarola

(2010). In Mental capital and wellbeing, by Cooper

Field

Goswami

Jenkins

Sahakian

(Eds.), (pp. 717-724). Chichester, England: Wiley-Blackwell.

23.

Lipscomb

J. A.

London

Chen

Y. M.

Flannery

Watt

Geiger-Brown

. . . McPhaul

(2012). Safety climate and workplace violence prevention in state-run residential addiction treatment centers. Work, 42, 47-56.

24.

Menckel

Viitasara

(2002). Threats and violence in Swedish care and welfare—Magnitude of the problem and impact on municipal personnel. Scandinavian Journal of Caring Sciences, 16, 376-385.

25.

Nachreiner

N. M.

Gerberich

S. G.

McGovern

P. M.

Church

T. R.

Hansen

H. E.

Geisser

M. S.

Ryan

(2005). Relation between policies and work related assault: Minnesota nurses’ study. Occupational & Environmental Medicine, 62, 675-681.

26.

Nachreiner

N. M.

Gerberich

S. G.

Ryan

A. D.

Erkal

McGovern

P. M.

Church

T. R.

. . . Feda

D. M.

(2012). Risk of physical assault against school educators with histories of occupational and other violence: A case-control study. Work, 42, 39-46.

27.

Naveh

Katz-Navon

Stern

(2005). Treatment errors in healthcare: A safety climate approach. Management Science, 51, 948-960.

28.

Piquero

N. L.

Piquero

A. R.

Craig

J. M.

Clipper

S. J.

(2013). Assessing research on workplace violence, 2000–2012. Aggression and Violent Behavior, 18, 383-394.

29.

Rasmussen

C. A.

Hogh

Andersen

L. P.

(2013). Threats and physical violence in the workplace: A comparative study of four areas of human service work. Journal of Interpersonal Violence, 28, 2749-2769.

30.

Schat

A. C.

Kelloway

E. K.

(2003). Reducing the adverse consequences of workplace aggression and violence: The buffering effects of organizational support. Journal of Occupational Health Psychology, 8, 110-122.

31.

Schein

E. H

. (2010). Organizational culture and leadership (4th ed.). San Francisco, CA: Jossey-Bass.

32.

Silva

Lima

M. L.

Baptista

(2004). OSCI. An Organisational and Safety Climate Inventory. Safety Science, 42, 205-220.

33.

Spector

P. E.

Coulter

M. L.

Stockwell

H. G.

Matz

M. W.

(2007). Perceived violence climate: A new construct and its relationship to workplace physical violence and verbal aggression, and their potential consequences. Work & Stress, 21, 117-130.

34.

Spector

P. E.

Zhou

Z. E.

Che

X. X.

(2014). Nurse exposure to physical and nonphysical violence, bullying, and sexual harassment: A quantitative review. International Journal of Nursing Studies, 51, 72-84.

35.

Tiesman

Konda

Hendricks

Mercer

Amandus

(2013). Workplace violence among Pennsylvania education workers: Differences among occupations. Journal of Safety Research, 44, 65-71.

36.

Vegchel

N. V.

Jonge

J. D.

Söderfeldt

Dormann

Schaufeli

(2004). Quantitative versus emotional demands among Swedish human service employees: Moderating effects of job control and social support. International Journal of Stress Management, 11, 21-40.

37.

Viitasara

Menckel

(2002). Developing a framework for identifying individual and organizational risk factors for the prevention of violence in the health-care sector. Work, 19, 117-123.

38.

Wassel

J. T.

(2009). Workplace violence intervention effectiveness: A systematic literature review. Safety Science, 47, 1049-1055.

39.

Wilkinson

C. L.

(1999). An evaluation of an educational program on the management of assaultive behaviors. Journal of Gerontological Nursing, 25, 6-11.

40.

www.arbejdsmiljoviden.dk . (2013). Nul tolerance oeger voldsstatistik [Zero tolerance increases statistics on violence]. Retrieved from http://www.arbejdsmiljoviden.dk/Aktuelt/Nyheder/2013/05/17-Nul-tolerance-oeger-voldsstatistik

41.

Yang

L. Q.

Spector

P. E.

Chang

C. H.

Gallant-Roman

Powell

(2012). Psychosocial precursors and physical consequences of workplace violence towards nurses: A longitudinal examination with naturally occurring groups in hospital settings. International Journal of Nursing Studies, 49, 1091-1102.

42.

Zohar

(1980). Safety climate in industrial organizations: Theoretical and applied implications. Journal of Applied Psychology, 65, 96-102.

43.

Zohar

(2000). A group-level model of safety climate: Testing the effect of group climate on microaccidents in manufacturing jobs. Journal of Applied Psychology, 85, 587-596.

44.

Zohar

(2010). Thirty years of safety climate research: Reflections and future directions. Accident Analysis & Prevention, 42, 1517-1522.

45.

Zohar

Luria

(2003). The use of supervisory practices as leverage to improve safety behavior: A cross-level intervention model. Journal of Safety Research, 34, 567-577.

46.

Zohar

Luria

(2005). A multilevel model of safety climate: Cross-level relationships between organization and group-level climates. Journal of Applied Psychology, 90, 616-628.