Abstract
Research has established policing as one of the most stressful occupations and that work-induced strain can lead to various criminal and negative outcomes. This study extends existing literature in criminology and criminal justice by applying general strain theory to police stress. This study examines the influence of prior strain—namely, child abuse and interparental violence—on officer critical incident strain, psychological–physiological stress responses, and officer-on-officer aggression. Data analyzed the stress responses of 1,104 police officers from the Police Stress and Domestic Violence in Police Families in Baltimore, Maryland, and results show that prior strain events influence respondents in three key ways. First, those reporting exposure to childhood strain had higher work-related strain scores. Second, youthful strain events and work-related critical incident strain are associated with increased stress. Finally, child abuse exposure and critical incident strain increases the odds of officer-on-officer violence among participants. The sample offers several important implications for strain theory because it includes only adult respondents and offers some degree of control for serious adolescent delinquency.
The American public generally considers policing as one of the most stressful professions due to the risk of engaging with hostile suspects, working long hours, or investigating a violent crime scene, and a significant body of research finds that work-related stress can produce negative outcomes for law enforcement personnel and police agencies (Anderson, Litzenberger, & Plecas, 2002; Harpold & Feenster, 2002; Howard, Donofrio, & Boles, 2004; Kurtz, 2008; Liberman et al., 2002; Lott, 1995; Maguen et al., 2009). Most literature focuses on the psychological, physical, or psychiatric stress reactions to “critical incidents” generated by exposure to life-threatening events as part of the policing environment (Anderson et al., 2002; Liberman et al., 2002; Loo, 2004; Mashburn, 1993). Other research links police stress to numerous behavioral problems including excessive alcohol consumption, inadequate performance at work, unstable intimate relationships, and even domestic violence (Bartollas & Hahn, 1999; Haarr & Morash, 1999; Gibson, Swatt, & Joicoeur, 2001; Swatt, Gibson, & Piquero, 2007).
Officer violence is a particular area of focus within the police deviance literature, and several studies indicate that work-related stress contributes to domestic violence in police families (Anderson & Lo, 2011; Erwin, Gershon, Tiburzi, & Lin, 2005; Gershon, Barocas, Canton, Li, & Vlahov, 2009; Johnson, Todd, & Subramanian, 2005; Lonsway, 2006) or directed at the public in the form of excessive force. However, because these studies only examine violence within the family, it is unknown whether work-related stress contributes to physical aggression toward other people in proximity to a strained officer. Officer-on-officer violence presents a unique example when exploring the dynamics of police officer aggression because much of what we know about officer’s aggression remains descriptive in nature (Stinson & Liederbach, 2013; Stinson, Liederbach, & Freiburger, 2012). Few studies directly connect police misconduct to existing criminological theories, notwithstanding the recent work examining police stress within the theoretical framework of Agnew’s (2006) general strain theory (Anderson & Lo, 2011; Gibson et al., 2001; Swatt et al., 2007) and some research arguing critical incident stress as theoretically consistent with various types of strain (Anderson & Lo, 2011).
As such, the purpose of this study is to investigate police officer stress and officer-on-officer physical aggression, a subject that is vastly understudied. Using general strain theory as a theoretical guide, we explore the influence of child abuse and interparental violence on later stress and officer aggression. Research on police officers as perpetrators of violence (see Erwin et al., 2005; Lonsway, 2006) generally focus on family violence and the belief that negative relationships with criminals and other harmful factors associated with police work contribute to officer aggression. These studies commonly rely on various forms of Agnew’s (2006) general strain theory to explain officer misconduct (Anderson & Lo, 2011; Gibson et al., 2001; Swatt et al., 2007). Exposure to violence during childhood is one of the most commonly studied risk factors for physical aggression, and many existing studies fail to explore this as a conditioning factor for officer-on-officer violence (Anderson & Lo, 2011). This study adds to the literature by exploring the potential influence of early strain in the form of experiencing child maltreatment and witnessing interparental violence on later strain, officer stress, and violent behavior. We hope that the results of this study can inform policies and practices that could improve police officers’ stress reactions.
To accomplish this goal, we first review the literature on general strain theory and police stress. This is followed by a literature review on the connection between prior strain and criminal/violent offending and a review of studies on criminal/deviant behavior by police officers. Hypotheses established by the literature review are provided and followed by detailed descriptions of the data, methods, and findings. The last section in this article provides a discussion of the findings and the study’s implications and limitations.
Literature Review
General Strain Theory and Police Stress
According to Agnew (1992), negative relations and negative affect are important causal steps in the pathway to crime. Stressors, such as parental rejection, harsh physical discipline, child maltreatment, and criminal victimization (Agnew, 2006), lead individuals to experience negative emotions which, in turn, may lead to crime, which is thought of as an adaptation to stress (Agnew, 2006). An impressive body of criminological literature indicates that individuals exposed to various types of identified strain are more likely to commit criminal behavior (Agnew, Brenzina, Wright, & Cullen, 2002; Froggio, 2007; Mazerolle & Maahs, 2000; Mazerolle & Piquero, 1997; Piquero & Sealock, 2000). However, individuals with a strong social support system can generally avoid criminal reactive behavior and alleviate the strain through noncriminal coping strategies (Agnew, 1992, 2006). Therefore, it is important to include measures of social support systems, such as support from coworkers and family, in studies of general strain theory.
Strain theory appears directly applicable to police deviance as prior findings link the stressful nature of police work directly to the risk of committing various forms of police misconduct (Anderson & Lo, 2011; Swatt et al., 2007). The criminology and criminal justice literature clearly establishes policing as a stressful profession, and many research articles investigate this topic (Anderson & Lo, 2011; Anderson et al., 2002; Harpold & Feenster, 2002; Howard et al., 2004; Liberman et al., 2002; Lott, 1995; Kurtz, 2008). Most studies draw connections between various types of stressors and a number of negative health and behavioral outcomes, focusing primarily on psychological, physical, or psychiatric stress responses to life-threatening events commonly referred to as critical incidents (Anderson et al., 2002; Harpold & Feenster, 2002; Liberman et al., 2002). A critical incident may include any situation when a police officer confronts the potential for serious injury or death or perceives an event as beyond his or her personal control (Anderson et al., 2002; Liberman et al., 2002). In addition, critical incidents may include making a violent arrest, responding to a gruesome crime scene, or dealing with a fatal accident (He, Zhao, & Archbold, 2002). Officers frequently deal with hostile citizens and other work-related stressors; however, truly life-threatening events remain fairly rare in law enforcement, with some research indicating that critical events are not a daily source of stress (Hart, Wearing, & Headey, 1993). Despite the rarity of these events, they could still elevate anxiety and stress among officers.
Much of the police stress literature focuses directly on officer psychological and physical responses, and these physiological stress reactions include increased heart rate, high blood pressure, muscle tension, indigestion, and fatigue (Anderson et al., 2002) and psychological concerns such as burnout and depression (Harpold & Feenster, 2002). Research indicates that many officers employ negative coping mechanisms to manage stress such as use of alcohol, tobacco, and social isolation (Anderson et al., 2002; Harpold & Feenster, 2002; Richmond, Wodak, Kehoe, & Heather, 1998; Violanti, 2004), and it is possible that they utilize violence to manage adverse emotional reactions to stress.
An important factor to consider with police officer stress is the potential influence of officer interpersonal relationships. Some research shows that family support reduces stress among officers; however, other studies indicate that social support may operate differently according to the level of stress (Anderson et al., 2002; Kurtz, 2012). Another potential source of stress or support centers on workplace dynamics. Some researchers have found that work–family conflicts can reduce job satisfaction and increase emotional exhaustion and burnout (Howard et al., 2004). Morash, Kwak, and Haarr (2006) found that workplace problems explained less of the variation in stress levels for women officers compared to men. They concluded that outside sources of stress, including caretaking and household responsibilities, could represent a unique stressor for female officers. Conversely, peer support within police organizations may buffer stress responses. Peer influences, an exceptionally important social force in policing, are an important context for understanding police behaviors (Brooks & Piquero, 1998). Shared work experiences allow officers to develop a mutual understanding of work stressors that can serve as a protective factor. The positive functions of shared peer experiences and acceptance of one’s peers may provide “a defensive function which allows the officer to tolerate high levels of anger, hostility and abuse” (Brooks & Piquero, 1998, p. 602). As such, the policing workplace culture may promote both prosocial and antisocial means of coping with job-related stressors. Within this study, we include specific measures of the support of family and friends and cooperation between units as a proxy measure of officer work environment. In sum, this literature highlights the importance of work-related strains on aggression.
Prior Strain and Criminal/Violent Offending
It is also important to assess the impact of strain exposure during the formative childhood years on adult aggressive behavior. Two of the strongest correlates of physical aggression are child maltreatment (Gelles, 1997; Gover, Jennings, Tomsich, Park, & Rennison, 2011; Gover, Kaukinen, & Fox, 2008; Schwartz, Hage, Bush, & Burns, 2006; Widom, 1989b; Widom & Ames, 1994) and witnessing interparental violence (Clarey, Hokoda, & Ulloa, 2010; Foshee, Buaman, & Linder, 1999; Kerley, Xu, Sirisunaluck, & Alley, 2010; Milletich, Kelly, Doane, & Pearson, 2010). Several studies indicate that children who experience abuse or violent forms of punishment are more likely to perpetrate acts of aggression than children without these experiences (Langhinrichsen-Rohling, Hankla, & Dostal Stormberg, 2004; Melander, Noel, & Tyler, 2010; Mihalic & Elliot, 1997; Renner & Whitney, 2012; Straus & Yodanis, 1996). For example, Widom (1989b) reported a link between experiencing child abuse and neglect and later arrests for violent offenses such as attempted homicide, rape, and battery. Similarly, Heyman and Smith Slep (2002) found that being the victim of child abuse is related to both later victimization and perpetration of violence. As such, individuals who experience maltreatment within the family of origin may engage in violence within their later interpersonal relationships.
Children are not only affected by directly experiencing violence; researchers have also found a link between witnessing interparental violence and committing acts of violence (Foshee et al., 1999 [among females only]; Sims, Noland Dodd, & Tejeda, 2008). Although some studies have not found a connection between witnessing interparental violence and later aggression (Lavoie et al., 2002; Simons, Lin, & Gordon, 1998), others have found an association between parental violence and partner violence perpetration (Ehrensaft et al., 2003; Gover et al., 2008; Rosen, Bartle-Haring, & Stith, 2001; Stith et al., 2000; Whitfield, Anda, Dube, & Felitti, 2003). For example, in their study of male undergraduate students, Carr and VanDeusen (2002) found that observing violence between parents predicted physical dating violence perpetration. Milletich, Kelly, Doane, and Pearson (2010) found similar results, such as witnessing mother-to-father and father-to-mother violence correlated with dating violence among a college sample. As such, exposure to violence within the family of origin may influence whether or not someone personally utilizes violence within their intimate relationships.
The relationship between childhood exposure to violence and perpetrating aggressive behaviors later in life has been explained by the cycle of violence theory (Widom, 1989a), which is also commonly referred to as the intergenerational transmission of family violence theory (Abbassi & Aslinia, 2010; Schwartz et al., 2006). According to this perspective, children learn how to behave both by experiencing how others treat them and observing how their parents behave toward each other. In this manner, those who either witness or experience violence in childhood learn that violence can be an effective and normative response to conflict and are thus more likely to imitate these behaviors in their own relationships than individuals from nonviolent homes (Gray & Foshee, 1997; Stith et al., 2000). For example, a meta-analysis conducted by Stith and colleagues (2000) revealed that growing up in an abusive family, which included witnessing interfamilial violence and personally being a child abuse victim, is positively related to becoming involved in a violent marriage. Among a sample of male and female offenders, those who experienced childhood physical abuse were more likely to commit more violent offenses than nonabused offenders (Teague, Mazerolle, Legosz, & Sanderson, 2008). Consequently, children who grow up in violent homes learn the techniques of being violent as well as the justifications for this behavior (e.g., it’s for his own good; Gelles, 1997). Overall, this body of literature demonstrates the importance of prior violent events consistent with the theoretical conception of negative strain as a factor for later aggression.
Criminal and Deviant Behavior by Police Officers
Studies of criminal and deviant behavior among police officers are growing, but a limited area of the criminological literature (Geoffrey & Dunham, 1997; Herbert, 1998; Manning, 1977; Sherman, 1974; Skolnick, 1994; Skolnick & Fyfe, 1993). Concerns of deviant behavior by distressed or strained officers can range from behaviors like excessive alcohol consumption (Anderson et al., 2002; Richmond et al., 1998), numerous forms of work-related misconduct (Skolnick & Fyfe, 1993), to various types of interpersonal violence (Gershon, Lin, & Li, 2002; Manzoni & Eisner, 2006).
One possible negative outcome of police stress and work-related strain is the use of violence by police officers (Gershon et al., 2002; Manzoni & Eisner, 2006). Officer violence represents one of the more potential disconcerting deviant behaviors by distressed officers and this may uniquely connect to violent aspects of the job. Hodgson (2001, p. 536) states that police officers “learn to regard violence positively as either an appropriate means to an end or as an end in itself.” As such, officers may use violent behavior inappropriately in a range of interpersonal situations including within the family and/or in work environments. Griggs (2004) reported that 40% of the police families experience at least one incident of family violence per year, and 8% of the male officer report severe violence within the home. Several studies find that work-related stress contributed to this unique form of violence in police families (Anderson & Lo, 2011; Erwin et al., 2005; Gershon et al., 2009; Johnson et al., 2005; Lonsway, 2006). Because these studies only examine violence within police families, it is unknown whether work-related stress contributes to physical aggression toward other people in proximity to a strained officer. For example, some research finds that distressed and burned out officers are more likely to use violence during interactions with citizens (Stearns & Moore, 1993).
Another area of the police misconduct literature on police violence focuses primarily on excessive force by police officers within the confines of normal policing behavior. The extent of the excessive force problem is unclear because police organizations are closed systems, and loyalty and secrecy among police officers prevents accurate research (Cancino & Enriquez, 2004; Hickman, Piquero, Lawton, & Greene, 2001; Hodgson, 2001). Despite intuitive beliefs about the link between work-related stress and excessive force, existing literature provides contradictory findings. For example, Kop, Euwema, and Schaufeli (1999) found a significant link between officer stress and more frequent use of force among a sample of Dutch police officers; however, Manzoni and Eisner (2006) failed to find a direct connection in their analysis of Swiss officers.
This study expands the literature on the possible link between police stress and violence by incorporating a unique type of police aggression, officer-on-officer violence. We are not aware of any other studies that explore violence between officers, thus this project offers new insight into the influence of strain and stress on officer behavior. Based on the literature review, we explore the importance of prior strain and exposure to violence within the officer’s family of origin and test the following hypotheses:
Method
Data for this research are drawn from the cross-sectional study The police and domestic violence in police families in Baltimore, Maryland, 1997–1999 (Gershon, 2000). The original questionnaire utilized a five-page survey that assessed a number of variables relevant to this research, including background and demographic characteristics, police work–induced strains, officer stress reactions, and support levels. Of particular relevance to this study, the data included questions regarding prior strain exposure in the form of childhood maltreatment and interparental domestic violence within the officer’s home of origin. These data also included questions about current violent behavior. These data are open to academic use and available for downloading at the Inter-university Consortium for Political and Social Research (ICPSR) website and frequently used by published scholars to explore police stress, burnout, and violence (Anderson & Lo, 2011; Gershon et al., 2009; Kurtz, 2008, 2012; Zavala, 2013).
Sample
Gershon’s (1999) research team originally distributed the survey to officers in the Baltimore Police Department during roll calls in each precinct and their headquarters at the start of each shift. During the survey time frame, the Baltimore Police Department employed approximately 2,500 sworn officers and more than 1,200 surveys were distributed. Ninety-two percent of the officers receiving the survey completed it and the final sample included 1,104 officers.
Measures
For the purposes of this project, we use several sets of variables including demographic characteristics, support variables, measures of exposures to strain events in childhood, an index measuring strain from work-related critical incidents, an index measuring psychological and physiological stress responses, and a variable measuring officer use of violence. We utilize comparative cross tabulations and quantitative analyses with these variable groupings. For more detail on variable construction and the reliabilities of all indices, refer to Appendix A.
Support and control variables
As noted in the literature review, social supports can mediate negative strains and also offer the potential to reduce officer stress responses. Several variables measure forms of support and other important controls identified in the existing literature. To access the level of interpersonal support, we use variables from an existing survey question. Officers were asked their level of agreement with the following statement: “I feel that I can rely on support from my family, friends, etc.” Possible responses to this 5-point ordinal scale ranged from strongly disagree to strongly agree (mean = 4.194, SD = .879). We measure unit cooperation in an identical fashion (mean = 3.224, SD = 1.034). We reverse coded the numerical assignments from the original instrument that scored strongly agree as 1 and strongly disagree as 5.
The control variables included in these analyses are officer gender and education level. Some research indicates that women officers have higher overall stress levels than male officers (He et al., 2002; Kurtz, 2008) and may have appreciably higher levels of psychosomatic stress formations (He et al., 2002). In order to explore potential gender differences, a dummy variable was created with women serving as the reference category in regression models. An additional control variable accessed education level of officers and those with a four-year college degree or better were coded as 1 with all other groups coded as 0.
Police stress measures
The questionnaire utilized several 4-point scaled measures to examine elements of psychological and physiological stress, with possible answers ranging from never to always. Many of these items appear to measure the same latent traits, and for the purposes of this analysis, we combined them into a single index measuring both physical and psychological stress (α = .82). The measures assessed whether respondents ever experienced the following seven signs of psychological stress in the last 6 months: restlessness, feeling hopeless, panic attacks, irritability, withdrawal, depression, and emotional depletion. The physiological portion of this index used five questions assessing whether respondents experienced nausea, trouble getting their breath, a lump in the throat, pains or pounding in the chest, and faintness or dizziness. Index scores ranged from 12 to 48, and the variable serves as a dependent variable in a model addressing factors influencing stress and as an independent variable in a model assessing police violence. Appendix A details variable construction and descriptive statistics for all indexed variable in this study.
Strain variables
Several variables included in the survey measure aspects of youthful strain exposure and an indexed variable measures work-related critical event strain. One form of childhood strain exposure captures child maltreatment and respondents were asked the following question: “Have your parents (when you were a child) ever gotten physical with you?” Within the sample 32.2% (356) of the respondents reported experiencing child maltreatment strain. A second form of strain measures the presence of interparental violence within the officer’s home of origin and 20.1% (223) of the officers reported experiencing this form of strain. In this survey, the following question is asked: “Did your parents ever get physical with each other?” (0 = no, 1 = yes). Both the child abuse strain and the interparental violence strain variable are included as independent variables in all regression models.
Much research on police stress and deviant maladaptive behaviors focuses on critical incidents in the work environment (Anderson et al., 2002; Harpold & Feenster, 2002; He et al., 2002; Liberman et al., 2002), and we argue that these events are theoretically consistent strain exposures. Gershon (1999) included many of these strain situations in the original data collection process and we construct an indexed variable to measure the critical incident strain. We adapted our measurement technique from Anderson and Lo (2011) and this variable captured strain by asking police officers if they ever participated in a number of work-related traumatic events. In particular, police officers were asked whether they have ever (1) made a violent arrest, (2) shot someone, (3) were the subject of an internal affairs investigation, (4) responded to a call related to a chemical spill, (5) responded to a bloody crime scene, (6) personally know the victim, (7) involved in a hostage situation, (8) attended a police funeral, and (9) experienced a needle stick injury or other exposure to blood and body fluids. If a police officer experienced any of these events, they were asked to indicate how much that experience affected them emotionally. Response categories included 0 = did not experience, 1 = not at all, 2 = a little, and 3 = very much. These 9-items were then combined to create an index, with scores ranging from 0 to 27, with higher scores indicating a higher level of strain. Previous reported reliability analysis (α = .79) of these indexed items indicated a moderate consistency (Anderson & Lo, 2011; Gershon et al., 2009). We used this measure as a dependent variable in an exploratory model of factors influencing strain responses and as an independent variable in regression models for stress and officer violence. Appendix A details variable construction and descriptive statistics for the critical incident strain variable.
Violence dependent variable
The core area of interest in this research relates to deviant responses to various forms of strain and law enforcement stress, and respondents were asked if they had “ever gotten out of control and been physical (e.g., pushing, shoving, and grabbing) with a fellow officer” (Gershon, 2000). Respondents were coded 1 = yes and 0 = no. Because the violence variable examined only one serious form of maladaptive officer behavior, only a minority of the respondents engaged this officer-on-officer violence. Seventy-eight officers reported officer-on-officer violence, representing 7.1% of the sample. Because this dependent variable is binary, logistic regression analysis was used for this model to determine whether prior and current strain, controlling for other variables and stress, influences the odds of respondents reporting violent behavior. With regard to variable measurement and distribution, the independent and control variables can take any level of measurement and logistic regression has no assumptions regarding normal variable distribution. For a detailed discussion of logistic regression assumptions, see Hosmer and Lemeshow (2000) and/or Menard (1995).
Analytical Plan
The analysis is conducted in several stages. First, bivariant correlations were generated, and Table 1 displays a correlation matrix among the study’s variables. Bivariate correlations are useful to determine the direction and strength of the relationship between variables and also provide preliminary test for multicollinearity between the variables. Second, cross tabulation and tests of mean differences of key dependent variables differentiated by exposure to childhood strain are included in Table 2. Finally, a series of regression models are produced that examines police officer strain and use of violence by prior exposure to childhood strain and key control variables (Tables 3 and 4).
Correlations Matrix of Study Variables.
Note. a N = 1,104, b N = 1,060, c N = 1,077.
*p ≤ .05; **p ≤ .01 (two-tailed).
Self-Reported Violent Aggression, Mean Strain, and Stress Scores by Childhood Strain Events.
Note. aThe t-test used for comparison of means for strain and stress and χ2 with Fisher’s test utilized to test for statistical significance among officer-on-officer violence. b N for stress and strain means slightly smaller due to missing data on some scale items.
*p < .05; **p < .01; ***p < .001.
OLS Regression Analysis for Officer Stress.
Note. n = 1,016. VIF = variance inflation factor; OLS = ordinary least squares.
*p < .05; **p < .01; ***p < .001.
Logistic Regression Analysis for Officer Self-Reported Violence.
Note. N = 1,024. CI = confidence intervals.
*p < .05; **p < .01; ***p < .001.
Bivariate Results
The results of the bivariate matrix show several significant relationships among the main independent and the dependent variables. Stress (r = .144, p ≤ .05), child abuse (r = .127), and critical incident (r = .132) strains were all positively correlated with peer violence, while interparental violence was weakly correlated with peer violence (r = .064). Additionally, the bivariate findings indicate that no control variables have a statistically significant correlation with peer violence. The correlation matrix was further examined to identify any problems with multicollinearity. No two independent variables were highly correlated with each other, indicating no substantial problems with multicollinearity. To further assess model stability and regression assumptions, several diagnostic statistics are calculated. The Variation Inflation Factor (VIF) and Tolerance statistics are displayed in Table 3 with the corresponding ordinary least squares (OLS) regression model. The VIF statistics are all below 3.0 and a Tolerance above .20, further indicating that collinearity is not a problem in regression models. Furthermore, the indexed dependent stress variable appears to have a fairly normal distribution and the acceptable skewness statistic is 1.375. These diagnostic statistics in coordination with only minor variable correlations, results of plotted residual against fitted values, and acceptable standard errors that seem to indicate stable regression models with no major violations of regression assumptions (for detailed descriptions of regression assumptions and diagnostic test, see McClendon, 1994). With regard to variable measurement and distribution for the officer-on-officer violence dependent variable model, collinearity is also not a problem although logistic regression has no assumptions regarding normal variable distribution.
Findings
Table 2 displays key study variables differentiated by prior exposure to childhood strain events. Roughly half of the sample (n = 564) reports no incidence of childhood strain, with 356 respondents experiencing child abuse and 223 observing interparental violence within the household. The mean critical event strain score for those with no prior youthful strain events is 14.914. For those exposed to child abuse strain, the mean score on the strain index increases by more than 1 unit to 15.994, and t-test indicates a statistically significant difference. For this analysis, t-tests are calculated between the no exposure group and each type of childhood strain. The t-tests are not calculated between the two types of strain exposure subgroups. Those indicating youthful interparental violence strain also have mean critical incident strain index scores elevated by more than 1 unit at 15.968. This difference is also statistically significant. The influence of prior strain is also observed with the psychological–physiological stress variable as well. Those reporting childhood abuse have mean stress scores of 23.329 and those with interparental strain have mean scores of 23.481—both of these statistically significant mean scores are roughly two units greater than the means of those with no reported childhood strain. The subgroup reporting no childhood strain has the lowest percentage of officer-on-officer violence with only 4.6% reporting peer-directed violent behavior. Conversely, 11.8% of those indicating exposure to child abuse strain also reported officer-on-officer violent behavior, and 10.3% of the respondents who witnessed interparental violence also engaged in this violent behavior.
Table 3 displays an OLS regression model with officer stress as the dependent variable, 1 and results generally support the theoretical framing of this article. Several background and demographic variables exert a statistically significant influence on stress scores. Compared to women, men (b = −1.322) have more than a unit decrease in the stress index, supporting some prior literature finding that women have higher stress in policing (Haarr & Morash, 1999; He et al., 2002; Kurtz, 2012). Additionally, increases in the levels of interpersonal support (b = −1.376), a college education (b = −.827), and interunit cooperation (b = −1.044) are all associated with the expected decrease in stress scores.
The findings of this regression model support the Agnew’s (1992) general strain theory, as all the three subtypes of strain are statistically significant and associated with increased levels of stress. According to β weights, the variable with the strongest influence on officer stress is the critical incident strain variable. Each unit increase in the critical incident strain index (b = .406) corresponds with an increase of nearly half a unit on psychological–physiological stress index. The child abuse strain variable (b = .676) is associated with more than half a unit increase in stress as compared to respondents reporting no child abuse. The β score for this variable indicates that it exerts the fourth strongest influence on the stress scores of the included variables. Compared to those with no interparental violence in the childhood home, those experiencing this type of strain event have almost a one-unit increase in the stress measure (b = .852); however, the β weight indicates this is not a particularly strong relationship within the model. The adjusted r 2 statistic indicates moderate explanatory power of this model (r 2 = .238). Given the complexity of the psychological–physiological stress measures and the innumerable number of personal health and lifestyle factors that could influence this variable, the findings are important.
Table 4 displays the logistic regression model for officer-on-officer violent behavior. In this model, only four of the independent and control variables are statistically significant. Men have a 286% increased odds of self-reported officer-on-officer violence compared to women. This control variable demonstrates the strongest relationship with officer-on-officer violence within this research model and fits with existing research on the willingness of men to use violence (Miller, 1998). None of the other control or support variables within this model reach statistical significance, but generally display the expected directional relationship. The focal strain variables are each statistically significant and correlate with increases in the odds of officer-on-officer violence except for the interparental violence strain variable. Officers reporting exposure to child abuse associates with a 133% increased odds of officer-on-officer violence compared to those without this history of exposure. Experiencing emotional responses to critical incident strain events also increases the odds of violence and each unit increase results in an approximately 10% increase in the odds of officer-officer violence. The psychological–physiological stress variable shows a similar relationship with violence odds, and each unit increases relative odds of officer-on-officer violence by more than 7%. Furthermore, the strain variables have a secondary indirect relationship with offending in addition to the direct observed relationship because they increase stress, which in turn increases the odds of officer-on-officer violence.
Discussion
This study offers several important findings relevant to criminology and criminal justice literature in both police stress and general strain theory. First, we probe factors that influence officer stress levels pertinent to police stress literature, such as the influence of critical incidents on officer stress and negative reaction, and we further conceptualize critical incident responses as a form of strain in a manner similar to research conducted by Anderson and Lo (2011). Second, we explore the influence of childhood strain on current strain, stress levels, and officer-on-officer violence. Third, the unique sample of police officers presents important findings for strain theory as we test a purely adult sample of individuals with no significant history of involvement with the adult or juvenile justice system. Moreover, the nature of police work exposes this population to types of adult strain that are not experienced in most occupations, offering a unique caveat for strain theory.
This study examines the factors influencing officer self-reported psychological–physiological stress, and our findings are generally supportive of existing literature (Anderson et al., 2002; Gershon et al., 2009; Harpold & Feenster, 2002). Social support and background factors such as interpersonal support, interunit cooperation, and level of education correlate with reduced stress scores among survey participants. Men also have lower stress scores than women in the sample and this further supports an emerging area of literature regarding the influences of gender on stress outcomes in law enforcement (Haarr & Morash, 1999; He et al., 2002; Kurtz, 2012).
The primary purpose of this article is to explore the effects of youthful and work-related strain on officer psychological–physiological stress and officer-on-officer violence. Work-related strain, as measured by critical incident responses, has a statistically significant relationship with officer stress scores, and the β weight of this variable indicates it is the strongest variable within stress regression model. This variable also has an influence on the odds of peer-directed violence. Although exposure to such events remains a component of working in the law enforcement fields, these findings, nevertheless, highlight the potential for negative stress and behavior responses (Anderson et al., 2002; Harpold & Feenster, 2002; Howard et al., 2004; Kurtz, 2008; Liberman et al., 2002; Lott, 1995; Maguen et al., 2009). Prior exposure to strain in an officer’s youth is also a factor in officer behavior and influences stress responses to critical incident strain.
Prior strain events appear to influence respondents in three key ways. First, prior strain events directly influence officer scores on the work-related strain index. Those respondents reporting exposure to child abuse or interparental violence had work-related strain scores more than a unit higher on this indexed variable. These prior strain events were also the only statistically significant variable in an exploratory regression model with police work strain serving as a dependent variable. Although we note the weak explanatory power of the model, the importance of prior strain on new strain responses may indicate a continuing caveat for strain theory. Second, youthful strain events are associated with increased stress scores as indicated by roughly a two-unit increase in mean psychological–physiological stress scores among those exposed to child abuse or domestic violence in their childhood. These variables were also statically significant and associate with higher officer strain scores in the regression model for stress. Finally, the child abuse exposure variable increases the odds of officer-on-officer violence among the sample participants. These findings expand upon a budding area of research on the importance of prior strain events on delinquent behavior (Mazerolle & Piquero, 1997) and provide yet another validation for strain theory (Agnew et al., 2002; Mazerolle & Maahs, 2000; Mazerolle & Piquero, 1997; Piquero & Sealock, 2000). Our test of strain theory also supports the small but growing literature that applies strain directly to police deviance (Anderson & Lo, 2011; Swatt et al., 2007).
In addition to the direct relationship between childhood strain and negative officer outcomes, childhood strain has an indirect influence on officer stress and officer-on-officer violence. Prior strain experiences elevate current strain scores, and the current strain scores are directly associated with higher stress and risk of officer-on-officer violence in distinct regression models. This offers some evidence of secondary effects of prior strain because of the work-related strain variable’s influence on stress and violence. A similar relationship is noted between the childhood strain exposure and the stress variable. Childhood strain events increase the stress scores, and higher stress scores increase the odds of officer-on-officer violent behavior in the logistic regression model. On average, those exposed to child abuse or domestic violence had stress scores elevated by approximately two units on this indexed variable. Each unit increase in the logistic regression model also increases the odds of violence by roughly 6%, indicating the indirect influence of these prior strain events is an important consideration for future research. As noted in the literature review, two of the strongest correlates of physical aggression are child maltreatment (Gelles, 1997; Gover et al., 2008, 2011; Schwartz et al., 2006; Widom, 1989b; Widom & Ames, 1994) and witnessing interparental violence (Clarey et al., 2010; Foshee et al., 1999; Kerley et al., 2010; Milletich et al., 2010).
This study has several limitations that caution some interpretations of the results. First, these data were not specifically designed to test criminological theories. Therefore, there may be other influences that could account for the variation of the dependent variables. In general, only about a fourth of the variance in officer stress levels is explained by the included independent variables in the stress model and a similar concern is noted in the officer-on-officer violence logistic regression model. Part of the weakness of explanatory power relates to a myriad of factors not measured in this research sample that could influence officer stress.
As mentioned earlier, the cross-sectional nature of the data does not allow us to clearly establish a causal relationship between study’s variables. However, the wording of the child abuse and interparental violence strain variables does establish these as prior occurring events and adds some ability to interpret causal ordering for those two variables. Furthermore, we are not currently aware of any longitudinal data available that examines police stress and officer-on-officer violence. The lack of longitudinal data makes the current cross-sectional data appropriate and consistent with other studies that have examined the prevalence and causes aggression in police families (Anderson & Lo, 2011; Gershon et al., 2009; Gibson et al., 2001). Future studies could attempt to resolve this problem by collecting detailed information on officer-on-officer violence longitudinally.
The original study by Gershon (1999), perpetration/victimization was captured by asking broad questions. For example, the child maltreatment measure could encompass acts of corporal punishment that may not clearly fit abusive behaviors consistent with strain events. Future studies should use multiple measures that not only include physical perpetration and victimization but also include measures of emotional abuse perpetration and victimization. We assume that police officers who participate in this survey were forthright in their responses; however, some respondents may have been reluctant to disclose childhood trauma and/or their use of violence which could indicate a degree of underreporting. Nevertheless, studies have shown that data on self-reported acts of violence can be reliable if respondents were assured confidentiality (Johnson et al., 2005).
These data only contain one major multifaceted form of strain—occupational stress. Future studies should attempt to test whether other forms of strain are predictive of perpetration/victimization. Zavala and Spohn (2013) recently reported a correlation between vicarious and anticipated strain with victimization and perpetration. An interesting study would be to determine whether this relationship also holds true for police families, given that police officers must assume danger when interacting with suspicious individuals (anticipated) or hear reports of fellow police officers being assaulted or shot at by suspects (vicarious). Sixth, the generalizability of the study’s results is limited to cities similar to the characteristics of Baltimore. Future studies should attempt to replicate the study’s findings using a nationally representative sample of police officers encompassing numerous types of law enforcement agencies at the local and federal levels.
Finally, the sample used in this research offers several unique and important implications for strain theory because it includes only adult respondents and offers some degree of control for serious child and adolescent delinquency. What makes our study unique is that the police sample means that these officers were never identified as seriously delinquent in their youths because the rigorous background and psychological tests involved in the hiring process for a major police department should offers fairly significant control for documented serious juvenile delinquency and prior mental health problems. We must conclude that either the prior strain exposure placed these officers at risk for significant negative reactions to police work or they were involved in delinquent acts but managed to evade detection. Furthermore, police officers’ work environments may offer exposure to strain in the form of critical events and difficult work conditions found in few other employment fields. It is possible that these findings may be limited to professions like law enforcement, social work, military service, and medicine that expose employees to significant forms of strain that may exist in few other fields. Furthermore, extensive background and psychological testing should indicate that the respondents in this sample have a limited history of mental illness of prior serious criminal behavior and the measured negative behaviors represent unique responses to these adult strains. Moreover, we find support for both a direct and an indirect influence of youthful strain on the critical incident strain scores, psychological–physiological stress levels, and officer-on-officer violent behaviors. This offers a crucial area for future research in both general strain theory and critical incident stress for law enforcement—the importance of youthful strain exposure on coping ability following adult strain.
The findings have several policy implications for police organizations pending further empirical validation. First, the finding that child abuse/interparental violence is significant in predicting officer stress and violence, suggesting that police organizations may need to consider programs that identify and assist officers at greater risk of negative stress responses because of their early life experiences. Ironically, these negative early experiences may also influence the selection of law enforcement as a career for these individuals that may seek employment in a helping profession. Second, critical incidents may be a frequent occurrence for some large police departments and therefore these departments should closely monitor their personnel and offer preventive services to those police officers that demonstrate high levels of stress. This may be particularly true for officers with prior strain events in their life course. It also suggests that these agencies should encourage their officers to report any problems they may encounter while on the job. This could include better identification of distressed officers or the potential introduction of more holistic wellness programs by police departments. At a minimum, department should consider the importance of childhood experiences on the potential for negative reactions to critical incident strains and psychological–physiological stress.
Footnotes
Appendix A
Construction of Indexed Variables.
| Psychological–Physiological Stress (Cronbach’s α = .82, mean = 22.112, SD = 5.371) |
| In the past 6 months, how often did you have … |
| Physiological symptoms of stress |
| Nausea |
| Trouble getting your breath |
| A lump in your throat |
| Pains or pounding in the chest |
| Faintness or dizziness |
| Psychological symptoms of stress |
| Restlessness |
| Feelings of hopelessness |
| Panic attacks |
| Irritability |
| Withdraw |
| Depression |
| Emotional depletion |
| Responses to these questions use a 4-point scale with possible answers ranging from never to always |
| Critical incident strain (Cronbach’s α = .78, mean = 15.240, SD = 3.767) |
| Officers were asked to indicate emotional impact of the following critical incidents: |
| Made a violent arrest |
| Shot someone |
| Were the subject of an internal affairs investigation |
| Responded to a call related to a chemical spill |
| Responded to a bloody crime scene |
| Personally knowing the victim |
| Involved in a hostage situation |
| Attended a police funeral |
| Experienced a needle stick injury or other exposure to blood and body fluids |
| Police officers were allowed to answer either 0 = did not experienced, 1 = not at all, 2 = a little, and 3 = very much |
Appendix B
Exploratory OLS Regression Analysis for Officer Strain Scores.
| Variables | b | SE | β |
|---|---|---|---|
| Men | .481 | .327 | .045 |
| College degree | .247 | .261 | .029 |
| Interpersonal support | −.145 | .233 | −.019 |
| Unit cooperation | −.778 | .475 | −.05 |
| Interparental violence/strain | .679 | .292 | .073* |
| Child abuse/strain | .993 | .25 | .124*** |
| F Significance = .000 | |||
| Adjusted R 2 = .025 |
Note. n = 1,055. OLS = ordinary least squares.
*p < .05; **p < .01.
Authors’ Note
An earlier version of this article was presented at the 2012 meeting of the American Society of Criminology in Chicago, IL.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
