Workplace violence experienced by paramedics

Effective organisational interventions addressing workplace violence should be based on a thorough aetiology of the actual violent experiences (Leather, Cox, & Farnsworth, 1990). Research to date has relied on a variety of definitions and measures of workplace violence. Much of the workplace violence literature relies on the interpretation of the meaning of violence according to each individual. This subjective evaluation imparts a high degree of face validity but often causes problems with data predictions (Rogers & Kelloway, 1997). Mezey and Shepherd’s (1994) definition of violence as originating from either physical and/or verbal assaults (actual or the perceived threat of) is generally considered to be an appropriate measurement base, and this definition is therefore adopted by the current research.

Although the degree of exposure to violence varies considerably across occupations, jobs that entail close contact with clients and customers pose a greater risk (Hogh, Borg, & Mikkelsen, 2003). A substantial proportion of the workplace violence literature investigates the consequences of violence experienced by police officers, when the officers are exposed to assault and injury during the course of their duties (e.g., Anderson & Bauer, 1987). Mezey and Shepherd (1994) extended these discussions to evaluate the occurrence of violence in health care professionals including ambulance officers, emergency hospital unit workers, and general doctors. Mezey and Shepherd concluded that the experiences of violence at work (or the enduring threat of violence) were associated with adverse psychological outcomes including burnout and post-traumatic stress amongst these professionals. Walsh and Clarke (2003) also demonstrated that health care workers rated verbal threats and minor physical incidents of violence as significantly distressing. These findings are an important contribution to the literature that traditionally relies on the measurement of physical assaults.

The recognition that health care professionals are increasingly exposed to workplace violence is of serious concern. The UK National Health Service Executive (NHSE) for example, recently investigated the exposure to violence and its consequences. Approximately 65,000 violent incidents were reported as occurring each year and produced serious consequences including: poor morale, staff recruitment and retention problems, staff absenteeism, and high levels of occupational stress and psychological strain (NHSE, 1998). Pozzi (1998) also concluded that from a sample of US pre-hospital service providers, the majority (90%) had experienced verbal and/or physical violence during their careers. The consequences of these encounters were identified as adversely affecting levels of both individual health and well-being and various organisational costs. Similarly, Shakespeare-Finch, Smith, and Obst (2002) concluded that the exposure to workplace violence was significantly associated with high levels of occupational stress and anxiety and diminished levels of work-well being.

A possible moderator of the psychological consequences of violence experiences is the variable of perceived social support. The receipt of optimum types and levels of social support has been established as moderating the occupational stress-strain relationships, primarily through Johnson and Hall’s (1988) Job Demand-Control-Support model (JDCS). Recent investigations of both the sources and the types of workplace social support have considered the implications for organisational supervisor/management training (e.g., Brough & Frame, 2004). Such training interventions have been found to significantly reduce occupational stress consequences, as predicted by the JDCS. Beaton, Murphy, Pike, and Corneil, (1997) investigated the influence of workplace social support in samples of paramedics and firefighters. Beaton et al. demonstrated that these workers cope with inherently dangerous and stressful occupational demands by seeking social support from colleagues, friends, and family members. Furthermore, because of the shift-work nature of their job, these emergency service workers commonly develop a high degree of dependence on their colleagues, relying on them for both practical and emotional social support when encountering violent situations.

The literature therefore suggests that the daily exposure to (the threat of) violence by ambulance service personnel is of increasing concern. Some interventions such as the use of stab vests and police escorts in recognised danger spots are under consideration by the UK Health Services (Sherwell, 2001). The current research intends to empirically identify the extent and the characteristics of violent incidents experienced by a sample of Australian operational paramedics. The extent to which any such violent workplace incidents contribute to the experiences of perceived social support, job satisfaction, and psychological strain, for this occupational group is evaluated. The research tests two hypotheses:

H1: Verbal and physical workplace violence will predict similar adverse levels of both job satisfaction and psychological strain.
H2: The predictive relationship between workplace violence and job satisfaction and/or psychological strain will be moderated by workplace social support; such that high levels of support will reduce the adverse consequences of workplace violence.

Method

Five hundred paramedics from one Australian ambulance service were invited to participate in this research. The statewide sample consisted of a random selection of various ranks of paramedics from all of the geographic regions encompassed by the Service. Within each region, ambulance stations reporting violent incidents with either a high or low frequency were selected for research participation. The participants were invited to complete an anonymous self-report questionnaire and to return the completed questionnaires directly to the University researchers. A total of 119 completed questionnaires were returned to the researchers, producing a response rate of 24%.

The majority of the respondents were male (79%) and aged between 28-37 years. Over half (67%) the respondents were married or in a de-facto relationship, while the remaining 33% were single or divorced. A University degree or TAFE diploma was the highest formal educational qualification for 60% of the respondents. Exactly half (50%) of the respondents had been working as a paramedic for between 10 to 19 years, while 37% had worked as a paramedic for up to nine years. Three-quarters of the respondents had been working at their current station for five years or less (78%), with 5% working there 15 years or longer. Finally, most of the respondents identified themselves as Advanced Care Paramedics (40%) and 21% were Intensive Care Paramedics.

Experiences of violence. The experiences of workplace violence were assessed with a six-item measure designed for the purpose of this research. The items were elicited from previous research with New Zealand ambulance officers (Brough & Mansell, 2001). The measure was tested for face validity in qualitative interviews with a small sample (N = 15) of the current research population and produced unanimous approval. The measure contains three items pertaining to verbal violence and three items pertaining to physical violence and requests the respondents to rate how often they had actually experienced any of the items during their work duties within the past six months. Example items include: “minor verbal abuse” (verbal violence) and “actual physical assault” (physical violence). The respondents indicated their responses on a Likert scale from 1 = Never to 5 = Virtually all of the time. High scores therefore indicate a high experience of workplace violence. The measure produced acceptable internal reliability estimates (Cronbach’s alpha coefficient): .81 (verbal) and .69 (physical).

Social support. The receipt of social support was estimated with Caplan, Cobb, French, Harrison, and Pinneau’s (1975) Social Support Scale. This instrument evaluates social support received from three sources: work supervisor, work colleagues, and family and/or friends. Only the workplace support subscales (supervisors and colleagues) are included in this paper. The respondents evaluated the extent to which four items of support were received from each source. Example items include: “How easy is it to talk to the following people?” and “How much are each of the following people willing to listen to your personal problems?” The participants responded on a five-point Likert scale from 0 = Very much to 5 = Don’t have any such person. All items were reverse scored, with high scores indicating high levels of social support. The measure produced acceptable internal reliability estimates: .89 (supervisor) and .88 (colleagues).

Job satisfaction. Job satisfaction was measured with the 15-item Warr, Cook, and Wall (1979) instrument. The measure requires the respondents to indicate how satisfied they are with each workplace characteristic (item). Example items include: “The recognition you get for good work” and “The way the organisation is managed”. The participants responded to each item on a seven point Likert scale, from 1 = Extremely dissatisfied to 7 = Extremely satisfied. High scores therefore suggest a high level of satisfaction. The measure produced an acceptable internal reliability estimate of .86.

Psychological strain. Context-free psychological strain was measured with the General Health Questionnaire (GHQ-12; Goldberg, 1972). The respondents answered each item via a consideration of their recent health behaviours and indicated their responses on a four point scale where 0 = Not at all to 4 = Much more than usual. High scores indicate high levels of psychological strain. Example items include: “Been losing confidence in yourself?” and “Felt constantly under strain?”. The measure produced an acceptable internal reliability estimate of .77.

Results

The six items of the experiences of violence measure were subject to an exploratory factor analysis (EFA: Principal Components Analysis) employing direct oblimin rotation to identify the structure of the measure. The EFA produced two factors with eigenvalues greater than 1. Examination of the scree plot confirmed the existent of two distinct factors, prior to levelling off. The six items produced acceptable factor loadings for the factors (minimum of .66) and the two factors accounted for 68.98% of the variance within this measure; these results are illustrated in Table 1. It can be observed that the first factor contains items pertaining to verbal violence (actual or threat of) whilst the second factor contains items pertaining to the threat or actual experience of physical violence. The two factors were moderately correlated with each other (r = .32) and each factor produced an acceptable level of internal reliability (.81 and .69, respectively). These two factors (Verbal Violence and Physical Violence) were therefore deemed suitable for inclusion within the subsequent analyses.

Table 1. Exploratory Factor Analysis Structure Matrix for the Experiences of Violence (N = 119)

	Component
Items	1	2
1. Serious verbal abuse (e.g. heavy swearing)	.87	.21
2. Minor verbal abuse (e.g. name calling)	.85	.18
3. Minor verbal threats (e.g. threatening to harm)	.85	.24
4. Threat of sexual assault	.41	.83
5. Sexual assault	-.05	.83
6. Physical assault with a dangerous weapon	.29	.68

Variance explained	44.50%	24.48%
Cumulative variance explained	44.50%	68.98%

The bivariate relationships (correlations) between the research variables are depicted in Table 1. It is pertinent to note that none of the demographic variables were significantly related to the experience of workplace violence, job satisfaction, or psychological strain. Verbal violence was significantly associated with low levels of supervisor support and job satisfaction and high levels of psychological strain. Physical violence produced insignificant associations with all of the research variables. Both supervisor and colleague support were significantly associated with job satisfaction but not with psychological strain. Finally, job satisfaction and psychological strain produced a significant negative association as expected.

Table 2. Scale Descriptives and Intercorrelations (N = 119)

	1	2	3	4	5	6	7	8	9	10	M	SD
1. Gender	-										-	-
2. Marital status	-.15	-									-	-
3. Age	-.30***	.22*	-								-	-
4. Tenure	-.28**	.02	.31**	-							-	-
5. Verbal violence	.03	-.08	.02	.01	(.81)						2.02	1.70
6. Physical violance	-.05	-.06	-.01	.02	.32**	(.69)					3.31	0.69
7. Supervisor support	-1.0	.00	.14	.18	-.29**	-.02	(.89)				3.63	0.97
8. Colleague support	.02	-.03	.10	-.07	-.10	.10	.34***	(.88)			3.70	0.79
9. Job satisfaction	.07	-.08	.09	-.01	-.31***	-.07	.44***	.33***	(.86)		4.46	2.14
10. Psychological strain	-.06	.08	-.04	.05	.18*	.05	-.07	-.13	-.25**	(.77)	1.14	0.04

Note. *p <.05; **p < .01;***p < .001; Cronbach's alpha coefficients are depicted in parentheses along the diagonal;
Gender dummy coded 0 = Male, 1= Female; M Status dummy coded 0 = Single/Divorced, 1 = Married/Defacto.

The ability of the violence and work support variables to predict levels of job satisfaction and psychological strain were evaluated in two individual moderated hierarchical multiple regression analyses. The results of these regression analyses are summarised in Table 2. Verbal violence was a significant predictor of both job satisfaction and psychological strain, although the influence of verbal violence was reduced to non-significance with the subsequent entry of the moderator terms at Step 3 of the psychological strain equation. Physical violence was not a significant predictor of either criterion variable. Both supervisor and colleague support significantly predicted job satisfaction but not psychological strain.

Finally, the verbal violence x supervisor support and verbal violence x colleague support interactions were tested in the prediction of each criterion variable. The interaction terms were constructed with standardised variables prior to entry in the equations. The entry of the two moderator terms accounted for no significant proportion of additional variance in job satisfaction. The predictor variables in total accounted for 26% of the variance in job satisfaction (F (6, 105) = 5.91, p < .001). For psychological strain, the verbal violence x supervisor support moderator term significantly accounted for a moderate proportion of additional variance. The predictor variables in total accounted for a small and statistically insignificant proportion of variance in psychological strain (R² = .11, F (6, 107) = 2.10; NS).

Table 3. Moderated Hierarchical Multiple Regression for Job Satisfaction and Psychological Strain (N = 119)

		Job satisfaction			Psychological strain
	Predictors	Step 1β	Step 2 β	Step 3β	Step 1β	Step 2 β	Step 3β
Step 1	Verbal violence	-.33**	-.21*	-.21*	.18*	.19*	.08
Step 1	Physical violence	.04	.01	.01	-.01	-.00	-.01
Step 2	Supervisor support		.28**	.28**		.08	.04
Step 2	Colleagues support		.21**	.22*		-.11	-.08
Step 3	V violence x Sup support			-.02			-.30**
Step 3	V violence x Coll support			.06			.10
	R	.32**	.50***	.51***	.18	.21	.32
	R²	.10**	.25***	.26***	.03	.05	.11
	R²change	.10*	.15***	.00	.03	.01	.06*

Note. *p <.05; **p < .01;***p < .001; β = standardised beta weight.

Moderating influence of supervisor support

The significant moderation term produced within the psychological strain equation is illustrated in Figure 1. The supervisor support and verbal violence variables were converted into categorical variables by including scores falling one standard deviation below (low category) and above (high category) the mean. Respondents receiving both high and low levels of supervisor support show a similar response to increasing experiences of verbal violence. However, the high support group display significantly higher levels of psychological strain when experiencing verbal violence as compared to the respondents receiving low levels of supervisor support.

Fig 1

Figure 1. Interaction between supervisor support and verbal violence on psychological strain

Curvilinear relationships

It is important to acknowledge that although not presented formally in these results, curvilinear relationships between the four continuous variables and the two criterion measures were also tested. The inclusion of an additional four predictors exceeded the recommended parameters of the regression equations (i.e. 10 predictors) in consideration of sample size (N = 119; Cohen, Cohen, West, & Aiken, 2003). However, the potential of curvilinear relationships existing between the predictor and criterion measures was nevertheless tested to ensure the moderation results were not spurious (Lubinski & Humphreys, 1990) . Research reporting significant interactive effects without controlling for nonlinear components, may run the risk of insecure interpretations of their findings (Mansell & Brough, 2005). Quadratic terms for the four continuous variables were calculated by squaring the continuous variables (Aiken & West, 1991) and these four quadratic terms were entered into the (four step) regression equations as step 3 (i.e. after the linear tests and before the moderation terms (step 4)). The block of quadratic terms did not account for any significant increment in explained variance in either equation. No individual quadratic term was statistically significant in the prediction of either criterion measure. Comparable results for the block of moderation terms were produced regardless of whether the quadratic terms were included in the two equations or not (i.e. only the violence x supervisor support term was statistically significant). Therefore, the block of quadratic terms was not included in the final results reported in this paper (i.e., as reported in Table 3) due to the instability of these regression equations.

Discussion

Verbal violence had a direct impact on levels of both job satisfaction and psychological strain. Physical violence however, was not a significant predictor of either criterion variable. These findings offer some support for Hypothesis 1; the adverse impact of verbal violence was clearly demonstrated. The respondents indicated that they usually disregarded most forms of verbal abuse by justifying its occurrence to the ‘high stress’ characteristics of the operational situation. However, this research suggests that the consequences of exposure to verbal violence are actually more serious compared to the exposure to physical violence. One obvious reason for this result is the frequency of occurrence of these forms of workplace violence. Both the descriptive results (Table 1) and the respondent’s comments indicated that verbal violence occurs frequently (typically several times per work shift) whilst physical violence (especially the more serious incidents) occur more infrequent.

These findings reinforce Walsh and Clarke’s (2003) conclusion that verbal threats of violence experienced by health workers are significantly distressing and produce consequences comparable to experiences of physical violence. The results produced by the current research also reflect recent discussions suggesting that frequently occurring minor daily stressors (hassles) have a comparable or even greater influence on psychological outcomes, as compared to the more infrequent occurring traumatic stressful events (Brough, 2004; Hart, Wearing, & Headey, 1995). These results imply that verbal abuse should be recognised as a workplace hazard for paramedics and therefore, paramedic training courses should ensure that they specifically address the effective management of verbal abuse.

The inability of physical violence to directly predict either criterion variable is surprising and offers no support to this element of Hypothesis 1. Physical violence also produced no significant bivariate relationships with any of the research variables. These findings contradict recent reports of the increasing exposure to physical violence experienced by paramedics (e.g., Shakespeare-Finch et al., 2002). To eliminate the possibility that the findings reported here could be sample specific, further testing within other Ambulance Services is required. The current results otherwise imply that for this sample of Australian paramedics, physical violence is an infrequent occurrence and has minimal impact on two measures of psychological outcomes. Such results contradict findings reported in both the US (Pozzi, 1998) and the UK (Sherwell, 2001).

Both supervisor and colleague support directly predicted job satisfaction but were both insignificant predictors of psychological strain. Interestingly, supervisor support accounted for a larger proportion of variance in job satisfaction in comparison with the verbal and physical violence predictors. These results support the demonstrations of the positive influence of effective workplace supervisor support on job satisfaction and other psychological outcomes (Brough & Frame, 2004; Stephens & Long, 2000). These results add further weight to the implications noted above that the effective management of workplace violence (by supervisors, for example) can be expected to have a stronger influence on outcomes, as compared to the actual violence experiences alone.

The hypothesised moderating influence of support upon the violence-satisfaction/strain relationships was only apparent in one of the four moderation equations: Supervisor support significantly moderated the consequences of verbal violence upon psychological strain. However the moderation did not occur in the hypothesised direction, offering only mixed support for Hypothesis 2. The influence of verbal violence upon psychological strain was higher for the respondents who received high levels of supervisor support (Figure 1). These results imply that respondents experiencing higher levels of strain due to their experiences of violence seek more supervisor support, compared to their colleagues experiencing less strain. However, the statistical insignificance of the psychological strain regression equation as a whole, suggests this result should be subject to further consideration.

The unexpected direction of these results occurred also within the bivariate relationships (Table 2); verbal violence and supervisor support produced a significant negative coefficient. This result implies that experiencing verbal violence is associated with low levels of supervisory support. As discussed below, a longitudinal research design with a larger respondent sample would enable cross-lagged analysis to determine the predictive relationship between these variables. Is the experience of violence a product of poor support (and thus remedied by supervisory support training), or do the paramedics who experience verbal violence not seek support? This latter point for example, could be related to the stigma associated with reporting minor distressing work experiences which has been documented elsewhere (e.g., within Police Services; Brough, 1998).

The moderate response rate experienced by this research qualifies the findings to some extent. It is unknown whether the non-respondents considered workplace violence to be an insignificant issue within their working lives (although the qualitative interviews suggested this was not the case) or whether their non-response was based on other reasons. Certainly attempts by an organisation to seriously address the issue of workplace violence require a more selective sampling method (based perhaps on workshops rather than self-report questionnaire completion alone).

A related limitation concerns the self-report design of this research. The measuring of workplace violence by subjective methods is recommended so as to gauge perceptions of the threat of violence as well as actual violent experiences (Mezey & Shepherd, 1994; Walsh & Clarke, 2003). Objective data collection methods (such as formal incident reports) generally do not capture experiences of the threat of violence and instead focus upon reporting physical (rather than lone verbal) assaults. Nevertheless the advantages of including both subjective and objective methods of workplace violence are acknowledged and should be considered by future researchers. One advantage of employing objective measures for example, is that the issue of common method variance that commonly arises when relying upon self-report data is avoided (Podsakoff, MacKenzie, Lee, & Podsakoff, 2003). The potential for common method variance to inflate the correlation results reported in this paper is acknowledged and these results therefore should be interpreted cautiously. It is recommended that obtaining a greater number of respondents would enable structural equation modelling to be employed and this is highly recommended for future researchers. Structural equation modelling combined with measuring method constructs is the recommended statistical technique for controlling method variance (Conway, 2002). This technique for example, is considered to be superior to the partial correlation technique as it allows unwanted method effects (i.e., respondent bias) to be removed whilst retaining any substantive variance (see Conway, 2002).

Finally, a third limitation is the cross-sectional nature of this research: a common limitation of many psychological health investigations. A longitudinal methodological design would have enabled stronger causal conclusions to be drawn from this data and also would enable tests for reverse causality to be conducted (e.g., via cross-lagged analysis; Brough, O’Driscoll, & Kalliath, in press). The impact of physical and verbal workplace violence on long-term psychological health outcomes is a particularly neglected area and is recommended as an especially worthwhile area for future investigations.

This research succeeded in addressing the primary point of effective workplace violence research, i.e. researching the actual aetiology of the violent experiences (Leather et al., 1990). This research identified three verbal and three physically violent incidents commonly experienced by operational paramedics. These experiences of workplace violence were directly predictive of adverse job satisfaction but had little direct impact on psychological health. The influence of workplace social support was also found to directly predict job satisfaction levels, with supervisor support in particular being a strong predictor. These findings have an important implication for the content of supervisory (support) training programs in managing workplace violence.

Finally, the high levels of intrinsic rewards achieved by this job were readily acknowledged by the responding paramedics and in many cases were used to justify their continued employment. This paper concludes with a particularly pertinent comment from one respondent which illustrates this point and places the experiences of workplace violence in context: “Verbal abuse is considered the norm, I get verbally abused once a week, I get sworn at once every three weeks, I get spat on or at once a year. I get threatened with violence once every three weeks and I get assaulted once a year. All the assaults I have experienced have resulted in minor injury due to my own awareness and experience and the patient’s actions have been slowed by alcohol or drugs. I have not experienced as much violence as some, so why do we tolerate this? We do because we love our job.” (A/O 30).

Acknowledgements

References

Beaton, R. D., Murphy, S. A., Pike, C. K., & Corneil, W. (1997). Social support and network conflict in firefighters and paramedics. Western Journal of Nursing Research, 19, 297-313.

Brough, P. (1998). The utility and value of in-service police welfare provision. Police Research and Management, 2, 27-34.

Brough, P. (2004). Comparing the influence of traumatic and organisational stressors upon the psychological health of police, fire and ambulance officers. International Journal of Stress Management, 11(3), 227-244.

Brough, P., & Frame, R. (2004). Predicting police job satisfaction, work well-being and turnover intentions: The role of social support and police organisational variables. New Zealand Journal of Psychology, 33, 8-18.

Brough, P., & Mansell, A. (2001). The occupational well-being of New Zealand Emergency Service workers: Final report. Unpublished report, School of Psychology, Victoria University of Wellington, New Zealand.

Brough, P., O’Driscoll, M., & Kalliath, T. (in press). Evaluating the criterion validity of the Cybernetic Coping Scale: Cross-lagged predictions of psychological strain, job and family satisfaction. Work & Stress.

Cohen, J., Cohen, P., West, S., & Aiken, L. (2003). Applied Multiple Regression/Correlation Analysis for the Behavioral Sciences. Mahwah, NJ: Lawrence Erlbaum Associates.

Conway, J. M. (2002). Method variance and method bias in Industrial and Organizational Psychology. In S. G. Rodelberg (Ed.), Handbook of Research Methods in Industrial and Organizational Psychology (pp. 344-365). Oxford, UK: Blackwell Publishers.

Department of Health, (2000), New campaign to end violence against ‘high risk’ NHS staff, Department of Health press release, October 3rd.

Goldberg, D. P. (1972). The Detection of Psychiatric Illness by Questionnaire. Oxford University Press, Oxford.
Hart, P.M., Wearing, A.J., & Headey, B. (1995). Police stress & well-being: Integrating personality, coping, & daily work experiences. Journal of Occupational & Organisational Psychology, 68, 133-156.

Hogh, A., Borg, V., & Mikkelsen, K. L. (2003). Work-related violence as a predictor of fatigue: A 5-year follow-up of the Danish Work Environment Cohort Study. Work & Stress, 17, 182-194.

Johnson, J.V., & Hall, E.M. (1988). Job strain, workplace social support, and cardiovascular disease: A cross-sectional study of random sample of the Swedish working population. American Journal of Public Health, 78, 1336-1342.

Leather, P. J., Cox, T., & Farnsworth, W. J. F. (1990). Violence at work: An issue for the 1990s. Work & Stress, 4, 3-5.

Lubinski, D., & Humphreys, L. G. (1990). Assessing spurious “moderator effects”: Illustrated substantively with the hypothesized (“synergistic”) relation between spatial and mathematical ability. Psychological Bulletin, 107, 385-393.

Mansell, A., & Brough, P. (2005). A comprehensive test of the job demands-control interaction: Comparing two measures of job characteristics. Australian Journal of Psychology, 57(2),103-114.

Mezey, G., & Shepherd, J. (1994). Effects of assault on health-care professionals. In J. Shepherd (Ed.), Violence in health care: A practical guide to coping with violence and caring for victims (pp. 1-11). New York: Oxford University Press.

National Health Service Executive (NHSE). (1998), A survey of violent incidents against NHS staff. Unpublished report, Department of Health, UK.

Podsakoff, P. M., MacKenzie, S. B., Lee, J.Y., & Podsakoff, N. P. (2003). Common method biases in behavioral research: A critical review of the literature and recommended remedies. Journal of Applied Psychology, 88, 879-903.

Pozzi, C. (1998). Exposure of pre-hospital providers to violence and abuse. Journal of Emergency Nursing, 24, 320-323.

Rogers, K. A., & Kelloway, E. K. (1997). Violence at work: Personal and organizational outcomes. Journal of Occupational Health Psychology, 2, 63-71.

Shakespeare-Finch, J., Smith, S., & Obst, P. (2002). Trauma, coping resources, and family functioning in emergency services personnel: A comparative study. Work & Stress, 16, 275-282.

Sherwell, N. (2001). Zero Tolerance: Violence and aggression towards ambulance personnel. Unpublished report. East Anglican Ambulance National Health Service Trust.

Stephens, C., & Long, N. (2000). Communication with police supervisors and peers as a buffer of work-related traumatic stress. Journal of Organisational Behaviour, 21, 407-424.

Walsh, B. R., & Clarke, E. (2003). Post-trauma symptoms in health care workers following physical and verbal aggression. Work and Stress, 17, 170-181.

Warr, P., Cook, J., & Wall, T. (1979). Scales for the measurement of some work attitudes and aspects of psychological well-being. Journal of Occupational Psychology, 52, 129-148.