Am J Crim Just (2015) 40:205–224 DOI 10.1007/s12103-015-9293-7 A Theoretical Integration of Social Learning Theory with Terror Management Theory: Towards an Explanation of Police Shootings of Unarmed Suspects Jon Maskaly & Christopher M. Donner Received: 9 December 2014 / Accepted: 5 February 2015 / Published online: 17 February 2015 # Southern Criminal Justice Association 2015 Abstract Each year in the United States, an unknown number of individuals are shot by law enforcement officers. Many of the suspects shot by officers are suspects who pose a lethal threat to officers or others, and thus the officers were legally justified in using deadly force. However, some estimates indicate that as many as 40 % of those shot by law enforcement each year are unarmed at the time of the encounter (Roy, 2004). Here, we present a theoretical model that integrates a traditional criminological theory (social learning theory) and a social psychological theory (terror management theory) in an effort to explain police shootings of unarmed suspects. Independently, neither theory can effectively or consistently explain the phenomena. However, when the two theories are integrated, a strong conceptual base is developed for explaining law enforcement shootings of unarmed suspects. Keywords Deadly force . Terror management theory . Social learning theory . Theoretical integration One of the most controversial topics in law enforcement is the use of force against members of the general public. This controversy is further exacerbated when considering the use of deadly force. Although police shootings are believed to occur infrequently, it is difficult to determine exactly how many police shootings there are annually as no national data are available. Some research indicates that in 2007 there were 391 fatal officer-involved shootings (Johnson, 2008). Fatal shootings are likely the easiest type of law enforcement shooting to record because there is a body J. Maskaly (*) Department of Criminology, Law, and Justice, University of Illinois at Chicago, 1007 W. Harrison St., M/ C 141, Chicago, IL 60607, USA e-mail: jonmaskaly@gmail.com C. M. Donner Department of Criminal Justice & Criminology, Loyola University Chicago, 820 North Michigan Ave., Chicago, IL 60611, USA e-mail: cdonner@luc.edu 206 Am J Crim Just (2015) 40:205–224 associated with the shooting. Other researchers have attempted to create a rate-ofoccurrence index for police shootings, and they have determined that there are approximately 4.1 shootings1 for every 1000 officers in a department (Fridell & Binder, 1992). Although these statistics were generated in 1992, if the general trend were to hold through time this would translate to approximately 3500 total police shootings annually.2 Other research indicates that within one region in Florida, approximately 40 % of all persons that police officers shot at were unarmed (Roy, 2004). While this statistic represents a very small portion of the United States, if a similar trend were observed nationwide it would translate into more than 1000 unarmed people shot at by the police annually. There have been several recent high-profile incidents of police officers shooting unarmed suspects in the United States. Two prominent examples of police shooting unarmed suspects occurred in 2014. The first example is the shooting death of Michael Brown by a police officer in Ferguson, Missouri on August 9, 2014. While the events that led up to the shooting are widely contested, it is clear that Michael Brown was shot six times by Officer Darren Wilson. The second example involves the case of Tamir Rice who was shot and killed by an officer from the Cleveland, Ohio Police Department on November 22, 2014. The officers received a call indicating that there was an African-American male in a city park brandishing a firearm, although the caller noted that the firearm was likely fake. Again, the sequence of events after the officers arrived on the scene is unclear, but what is clear is that one of the officers fired two shots at Tamir Rice within seconds of arriving on scene. In the wake of the shooting, investigators noted that Tamir Rice was carrying a toy gun—one that had the orange safety tip removed—with him at the time officers made contact with him in the park. Further, the officer stated that Rice was reaching for the gun in his waistband as the officers arrived on scene. It is worth noting that these are not the only two high profile incidents of police shooting unarmed suspects. Two of the most infamous examples involve the New York City Police Department. The first is the case of African immigrant Amadou Diallo, who was shot at 41 times—with 19 rounds striking him—by NYPD officers on February 4, 1999. Diallo was shot because he was reaching into his pockets for something and refused to follow the officers’ verbal commands. A subsequent investigation revealed that Diallo was reaching for his wallet and likely did not comply with the officers commands because he did not speak English (Morrison, 2000). The second example is the case of Sean Bell, who was shot and killed on November 25, 2006. After exiting a night club, several plain-clothes detectives attempted to prevent the car that Bell and two friends were riding in from leaving the club, believing there was a gun in the vehicle. Officers claimed that the driver of the vehicle tried to run them over (Fernandez, 2008). Officers opened fire, striking the three occupants of the 1 This includes shooting and killing a suspect, shooting a wounding a suspect, and shooting and missing a suspect. 2 Obtained using Bureau of Labor Statistic (2008–2009) Occupational Outlook Handbook estimates that there were 861, 000 police officers in the United States in 2009. Am J Crim Just (2015) 40:205–224 207 vehicle a total of 50 times, and subsequently killing Bell. No gun was located in the vehicle or on the three men. Although both of these examples stem from the New York City Police Department, the problem is far from being isolated to this city or this agency. Similar shootings have occurred in other locations, such as California (McKinley, 2009) and Florida (Anderson & Brink, 2005). When law enforcement executives and members of the public attempt to understand why police officers shoot unarmed suspects, there seems to be a general disconnect in the logic. There are stringent legal standards and agency policies in place dictating the necessary elements for officers to discharge their weapons. Many departments have adopted, or incorporated, the so-called Bdeadly force triangle 3^ into their use of deadly force policy (Olson, 1998). The triangle specifies that, in order to use deadly force, an officer needs a suspect who has: 1) the opportunity and 2) the ability to cause great bodily harm or death to the officer or others. Additionally, the suspect must 3) pose an Bimminent danger.^ Yet, as can be seen from the Diallo and Bell cases, not all of these elements were satisfied. This begs the question Bwhy are police officers using deadly force in these situations?^ A potential explanation lies within the police subculture. The police subculture is designed to serve a multitude of functions; one of the more important functions of the subculture, for the discussion at hand, is self-preservation. Researchers have noted that police officers are trained to see every person and every situation as potentially dangerous (e.g., Herbert, 1998; Skolnick & Fyfe, 1993; Van Maanen, 1974). This mentality leads to the documented Bus versus them^ phenomenon, in which law enforcement officers are seen as inherently good while the remainder of society is seen as potentially dangerous. This belief system is reinforced each time that a police officer is injured or killed in the line of duty, strengthening the officers’ belief in the dangerousness of society. This reinforcement also serves to emphasize the legitimacy of the police subculture. Next, we outline SLT and TMT, as well as discuss how an integration of these two theories could explain police shootings of unarmed suspects. Social Learning Theory Akers devised social learning theory (SLT) as an extension of differential association theory (DAT) by condensing the concepts from DAT into two concepts and adding an additional two concepts from behavioral learning theory (Akers, 2009). The two condensed concepts taken from Sutherland were broadly defined as differential associations and definitions. The two concepts that were added from other learning theories (e.g., Miller & Dollard, 1941) were differential reinforcement and imitation (Akers, 2009). The basic assumption of SLT is that the same learning process which is used to learn socially acceptable behavior is also utilized when people learn criminal/deviant behavior (Burgess & Akers, 1966). The full effect of each these concepts are explored below. 3 Olson is the first to spell out the policy in publicly available literature, and he admits that he does not remember the exact origins of the use of force triangle concept. 208 Am J Crim Just (2015) 40:205–224 Differential Associations SLT posits that a person will be more likely to commit a crime if he 4 differentially associates with persons who commit, model, and support violations of laws or social norms (Akers, 2009). Association is loosely defined by communication and interaction between two or more people. Akers (2009) identifies two primary types of interactions to which people might be subjected: behavioral (or interactional) and normative. Behavioral interaction is described as direct exposure to people with deviant behavioral tendencies. Normative interactions are described as a type of passive exposure where people are exposed to values and norms that are conducive to delinquent behavior. Recall that the effects of differential associations are believed to be moderated by four modalities of the interactions (Sutherland, 1947). Specifically the associations are moderated by, 1) frequency, 2) duration, 3) priority, and 4) intensity. Akers (2009) highlights that early in life these modalities are primarily associated with peer and friend groups. Later in life, these modalities expand to include spouses and members of working groups. Differential Reinforcement Akers (2009) indicates that when deviant behavior is differentially reinforced over conforming behavior, deviance is more likely to occur. It is believed that there are three primary reinforcement factors that will increase the likelihood of a behavior: 1) the higher the amount of the perceived reward, 2) the more frequently the behavior is rewarded and, 3) the greater the probability that the behavior will be rewarded. Akers (2009) identifies two distinct primary types of reinforcers: nonsocial and social. Nonsocial reinforcers (either positive or negative) occur within one’s self. Social reinforcers, however, are not exclusively the positive or negative reaction of others present when one engages in a particular behavior, but also the whole gamut of things that society and other subgroups value (e.g., money), including things that are highly symbolic (e.g., status and respect; Akers, 2009). Imitation A person who observes more delinquent than non-delinquent models is also more likely to be delinquent (Akers, 2009). Modeling behavior occurs when a person observes and then emulates the behavior of others. People who observe the behaviors of others that they admire or respect are particularly susceptible to the influence of the model and are more likely to engage in similar behaviors (Bladwin & Baldwin, 1981). Modeling has its strongest influence on a person’s initial decision to engage in delinquent behavior, and although it has an influence on continuation and cessation of behaviors, the effect is not as pronounced (Akers, 2009). 4 Note, he is used throughout the paper; this is not to suggest that only men will commit crime or are affected by SLT, but rather is used for simplicity sake. Am J Crim Just (2015) 40:205–224 209 Definitions A person is more likely to commit delinquent acts when he has learned definitions favorable towards delinquent acts; likewise, the more that one’s own attitudes reflect disapproval of a particular behavior, the less likely one is to engage in it (Akers, 2009). Each of the three previous concepts works to create definitions that are more favorable toward delinquent acts. Akers (2009) identifies three types of definitions that people may form which make offending more or less likely; only two, however, have bearing on the present discussion. The first is a positive definition, where a person has attitudes that are favorable towards deviant acts. The second definition is a neutralizing definition, whereby a person engages in delinquent acts, not because they see them as positive, but rather because they can justify or excuse the behavior. This type of definition is commonly associated with negative reinforcement schemes. Here, a person learns to escape the negative consequences of their behavior by justifying the behavior as morally correct. Connecting Policing and SLT Although the Chappell and Piquero (2004) study is the only study which directly tests SLT on law enforcement officers, there is a multitude of support for why the four different SLT concepts apply to police officers in general. Each of the four SLT variables work to create a police subculture. This subculture then serves as a platform for teaching and reinforcing police behavior. The behaviors that are taught and reinforced using this subculture are designed to protect law enforcement officers from harm, both from those wishing to do physical harm and harmful decisions of a punishment-centered upper-management. It is important to note that although these variables can be linked to the creation of a (deviant) police subculture, the variables alone cannot explain police shootings of unarmed suspects. Differential Associations Both police officers within organizations and researchers alike acknowledge the existence of a police subculture (Conser, 1980). The police subculture is designed to help officers deal with the sense of isolation that they feel from society (Alpert & Dunham, 1997). The sense of isolation is created by the powerful Bus vs. them^ mentality that is prevalent among many law enforcement officers (Rubenstein, 1973; Sparrow, Moore, & Kennedy, 1990; Tauber, 1970). As such, police officers often turn to other officers for moral support (Alpert & Dunham, 1997). The social interactions between officers and the subculture lead many officers to act in a manner that the peer group will approve of in order to develop a positive self-concept (Conser, 1980). Officers, thus, are likely to adopt cultural norms, of which, other officers will approve. Research indicates that the police subculture is so powerful and prevalent that officers are bound to adopt it (Skolnick, 1966).5 5 It is important to note that some researchers do not believe that all police officers adopt the same subculture; in one specific study, results indicate that only 25 % of law enforcement officers become what are termed Bsub-cultural adherents^ (Cochran & Bromley, 2003). 210 Am J Crim Just (2015) 40:205–224 Definitions Police officers hold several definitions that may be favorable to shooting a suspect given the right circumstances. Many officers mistakenly hold the belief that policing is about action, adventure, excitement, and the use of force (Brown, Maidmont, & Bull, 1993; Fletcher, 1996; Holdaway & Parker, 1998). The policing subculture, thus, respects aggressive, authoritative, and Btakecharge^ personalities of officers (Cochran & Bromley, 2003). Furthermore, nearly every police situation is seen as potentially dangerous, and every person is viewed with suspicion (Van Maanen, 1974). The confluence of the allegorical beliefs as to the role of police officers, the action orientation that is respected by other officers, and the inherent danger and untrustworthiness of all citizens work to create positive definitions towards using force against members of society. The definitions that officers create are passed down from one Bgeneration^ of police officers to the next. Research indicates that once officers adopt the police subculture, it provides an ideal medium to pass attitudes, beliefs, and values to the incoming generation of law enforcement officers (Kappeler, Sluder, & Alpert, 1994). Research also indicates that the police subculture is designed to create and maintain a normative order for the officers (Herbert, 1998). This normative order has been found to be similar to the definitional component of SLT, as originally conceived by Burgess and Akers and some researchers have used it as such (e.g., Ahern, 1972). Differential Reinforcement The subculture of policing has such a powerful effect on officers that it can reinforce deviant behavior (Conser, 1980). In fact, some researchers have gone so far as to say that one of the most powerful motivators within a police agency is the peer influence exerted by other members of the subgroup (Alpert & Dunham, 1997). One of the roles of the police subculture is to protect officers from punishment-centered upper-management (Waddington, 1999). Officers often perceive that upper-management is attempting to punish them for Bminor^ procedural errors (Walker, 1977). This fear of a punishment-centered upper-management is contrasted with the reinforcing structure of other police officers and the police subculture. New police officers gain status, reputation, and respect from other officers for their use of force (Brown & Sargent, 1995). The police subculture is one of masculinity through which aggression, violence, danger, risk taking, and engaging in courageous acts is rewarded by one’s peers (Waddington, 1999). Therefore, officers are differentially reinforced by different groups within the organization for the same behavior. They are positively and socially reinforced (as defined by Akers, 2009) by their peers within the police subculture, which is likely to lead to positive definitions for engaging in delinquent acts (e.g., shooting unarmed suspects). This is juxtaposed against the heavy-handed and punishmentcentered policies of upper-management, which, as Akers (2009) states, is likely to lead to neutralizing definitions that are favorable toward deviance. Am J Crim Just (2015) 40:205–224 211 Imitation This is the most challenging of the SLT concepts to apply to police officers – so much so that Chappell and Piquero (2004) did not include imitation as a variable in their study. However, based on the training methods of law enforcement, it is plausible to assume modeling behavior occurs at work. Currently, most police agencies in the United States utilize some form of Field Training Officer (FTO) program (Pitts, Glensor, & Peak, 2007). Many of these training programs are based on the original San Jose, CA FTO program, designed by Zwemke and Roberts in 1972 (San Jose Police Department, 2009). In an FTO program, a rookie officer begins his career by observing the actions of the training officer and assisting where possible or acting under the guidance of the trainer where appropriate. Throughout the course of the program, the new officer is progressively given more responsibility to act as a police officer with the Bsafety net^ of an experienced officer. Although the academic literature seems to neglect the fact that FTO’s can act as role models, the professional literature does not. For example, the California Police Officer Standards and Training Commission requests recruits who are finishing an FTO program to provide an overall evaluation of the program. One of the evaluation components asks officers to rate their FTO on the quality of role model that they believe him to be for new officers (California Police Officer Standards and Training Commission, 2004). Furthermore, the FTO programs allow and encourage new officers to integrate into the policing culture. New officers are frequently rated by their FTO (in many cases once per shift) by the subjective perceptions of the FTO. That is, if the new officer does not alter his behavior to the FTO’s subjective expectations, which are likely based on the way the FTO would have handled the situation, the new officer will be poorly evaluated. It is necessary to have positive evaluations throughout the FTO program in order to maintain employment as a police officer. Summary of SLT and Policing The information presented herein makes a case for the applicability of SLT to police officers and, perhaps, even to police shootings of unarmed suspects. The method of transmission of learning is through the policing subculture. The officers are differentially reinforced by the members of their subculture who reward them for behaving in a particular manner. Officers obtain their definitions about many things in their world from the subculture, especially in regard to the trustworthiness and dangerousness of members of the public. Lastly, at the inception of their career, officers are strongly encouraged to model their behavior after that of a more seasoned officer under the threat of termination for non-conformity. The ultimate question that remains is not why law enforcement officers shoot unarmed suspects, as this can be reasonably explained through SLT. The bigger question is why officers only shoot some of these suspects and not others. The explanation for this lies in Greenberg, Pyzczynski, and Solomon’s (1986) Terror Management Theory. 212 Am J Crim Just (2015) 40:205–224 Terror Management Theory Terror Management Theory (TMT), an extension of much of the work of Ernest Becker (1962), posits that much of what happens in human life is based on the denial of one’s own mortality (Pyszczynski, Greenberg, & Solomon, 2003). All creatures, including humans, engage in actions that are directed toward self-preservation; however, humans are the only animal that is uniquely aware of their own inevitable death (Pyszczynski et al., 2003). In fact, humans are the only animal that can project and imagine their own death, the thought of which causes B…potentially overwhelming terror at virtually any given moment^ (Pyszczynski et al., 2003, p.16). In order to protect from the terror of the inevitability of our own death, humans form a cultural worldview (Pyszczynski et al., 2003). The purpose of the cultural worldview is to distract humans from the terror that is essentially caused by the thought of their inevitable death. Greenberg, Pyszczynski, and Solomon (1986) advise that the cultural worldview created serves to inform humans that they live in a world of value and are, thus, differentially situated than other creatures. This cultural worldview also serves to allow humans to transcend their own mortality by allowing a person to feel as if he is part of a meaningful world and is contributing to that world in an important manner (Pyszczynski et al., 2003). There are two primary beliefs upon which TMT is founded. First, subscribing to a cultural worldview gives people a sense of order, stability, and permanence in their lives. Second, people must feel that they are valuable contributors to that worldview, which creates their reality (Pyszczynski et al., 2003). If these two elements are present, individuals are able to deny that they are merely creatures who are on Earth for a finite amount of time and are simply awaiting their own death. There are two major problems not accounted for by the creation of a worldview. First, there are several different cultures and competing worldviews. The threat posed by others possessing a different cultural worldview makes a person question one’s own worldview and undermining it on some level (Pyszczynski et al., 2003). Second, no cultural worldview can completely insulate a person from the thoughts of his own eventual death (Becker, 1975); this fear that can never really be removed is then projected onto an outside group or object (Pyszczynski et al., 2003). However, there are strategies that have been identified as possible methods of resolving the problems associated with competing worldviews and the fact that the thought of mortality can never be fully removed. Pyszczynski et al. (2003) identify five distinct types of solutions, which someone could implement in order to resolve the problems of uncertainty in their worldview. The authors do not indicate which solution is most likely, although it seems that certain solutions are more likely than others. The solutions seem to be presented in two different forms: passive (requiring little or no direct action by the person) and active (requiring a substantial undertaking of action). The first passive solution is conversion, whereby a person will simply adopt the worldview of another. This is most likely to occur when a person is not getting the validation that he needs and desires from his original worldview. Examples of this are religious conversions, in which people adopt a distinct religion based on the higher appeal of the new religion over the old one. The second passive solution is assimilation, whereby a person is asked by other people to set aside his Boutlandish^ worldview and Am J Crim Just (2015) 40:205–224 213 accept the dominant worldview. This is typically the result of physical proximity to another dominant cultural worldview. Pyszczynski et al. (2003) indicate there is a psychological boost given to one’s worldview when a large number of people accept it; in essence, people judge the ultimate validity of their worldview based on the number of others that subscribe to it as well. The final passive solution is accommodation, whereby a person accepts the appealing parts of a contrary worldview into his own worldview while simultaneously excluding those parts that they do not care for.6 The active solutions, however, require a person to engage in more drastic behavioral changes. The first active solution is derogation, whereby members of a worldview disparage those who ascribe to different worldviews. By demeaning others with dissimilar views, the threat they pose is somehow neutralized. Examples of this are seen in the sweeping pejorative generalizations made by one cultural group towards another (e.g., Ball black people are lazy^). The second type of active solution is the most extreme: annihilation. This solution indicates that a person will try to eliminate the threat posed toward his worldview by eliminating those with competing worldviews. Pyszczynski et al. (2003) discuss this solution to the presence of multiple worldviews; although it is an extreme solution, it is also the most effective solution and one of the most frequently observed. Lifton (1979) writes, BWars and prosecutions are, at bottom, expressions of rivalry between contending claims to immortality and ultimate spiritual power^ (p. 315). This type of solution can be seen throughout the history of the world, including the Civil War in the United States, as well as World War II. Modern examples include some of the violent gang wars in South Los Angeles (among other urban areas) and the War on Terror currently being fought in the Middle East. Annihilation is the principle response for the purposes of this integration. There are two distinct hypotheses that are derived from TMT. The first is that B… when people believe that they are objects of value in a world of meaning they should be able to function securely^ (Pyszczynski et al., 2003, p, 39). Essentially, Pyszczynski and his colleagues argue that self-esteem insulates a person from the terror he feels from knowing he is inevitably going to die. This hypothesis has been tested and sustained in various contexts (for a review see Pyszczynski et al., 2003). The second hypothesis states that reminders of one’s own death (through a process known as a mortality salience prime) should make a person rally around, and more zealously defend, his worldview (Greenberg et al., 1986). This hypothesis is the more popular of the two; with more than 100 studies testing and further sustaining the veracity of the hypothesis (see generally Pyszczynski et al., 2003). For the purposes of this integration, we will exclusively focus on the second hypothesis. General Research on TMT The very first test of TMT was conducted utilizing 22 municipal court judges in Tucson, Arizona (Rosenblatt, Greenberg, Solomon, Pyszczynski, & Lyon, 1989). Specifically, the judges were asked to set the bond amount for a prostitute that came into court; half the judges were in a control condition and half were in a mortality 6 The primary example given by Pyszczynski et al. (2003) is the Hippie Movement of the 1960’s and 1970’s, where some members of the mainstream society adopted the music and fashion of the competing cultural worldview. 214 Am J Crim Just (2015) 40:205–224 salience prime condition. The mortality salience prime was accomplished through the use of a new open-ended measure in which participants were asked to describe, in their own words, what physically happened to them when they died as well as the emotions that they associated with their own death. The results revealed that the judges in the mortality salience prime condition were substantially more punitive (setting an average bond of $455) than the judges in the control condition (setting an average bond of $50). Furthermore, results from a mood inventory filled out by the judges in the mortality salience prime condition revealed that the judges in this condition were not emotionally distraught about thinking about their own death. Rosenblatt et al. (1989), in an attempt to make the results more generalizable, conducted the same experiment but substituted undergraduates for judges. The results were consistent with those observed for the judges with one exception. The higher bonds were set by persons who had a negative view of prostitution within the mortality salience prime condition; yet, these participants still acknowledged that they were not upset after thinking about their own death. In order to determine if the results that Rosenblatt et al. (1989) observed were an effect of the negative perceptions of prostitutes, and criminals in general, they conducted an additional study. In their third study, Rosenblatt et al. (1989) asked participants to recommend a reward for a person who went out of their way to assist the police in catching a mugger and also asked them to rate the prostitution scenario again. The results for the prostitute scenario validated the earlier results. Moreover, the results from the new scenario revealed that participants recommended a substantially higher reward for the Bhero^ in mortality salience prime conditions (approximately $3500) than in the control (approximately $1100). The results from these three studies laid the groundwork upon which much of the additional TMT research has been conducted and strengthened. For instance, many scholars assumed that the results seen in other TMT research (i.e., individuals from other worldviews judged more harshly and individuals from the same worldview judged more positively) were a product of thinking about any terrifying event (e.g., taking an exam, dental pain, or general anxiety; Greenberg et al., 1995). This hypothesis was tested, and subsequently dispelled, as these events failed to produce the same results (for a more thorough discussion see, Greenberg et al., 1997). Other studies have examined the mortality salience hypothesis on more of a macro scale by having American college students read and rate essays that either represented pro- or antiAmerican sentiments (Greenberg et al., 1990). As expected, the results indicated that students in the mortality salience prime condition rated the pro-American essay significantly higher than the anti-American essay and higher than those in the control condition. These studies support the fact that mortality salience can be observed under laboratory conditions, but there was no indication whether it is a naturally occurring phenomenon. In an effort to answer this question, a series of studies were undertaken in which the researchers asked participants to indicate their position on an immigration reform measure in Germany (Pyszczynski et al., 2003). Participants were interviewed either in front of a funeral home (mortality salience condition) or more than 100 m away from a funeral home (control condition). Results indicated that the salienceprimed participants in the minority position of the immigration reform debate Am J Crim Just (2015) 40:205–224 215 substantially over-estimated the amount of support they would have in the reform;, however, this effect was not observed in the participants away from the funeral home or with the majority group in either location. Although this body of research generally supports the mortality salience prime, there have been a few problems associated with it. Specifically, it appears that mortality salience primes are most effective when a participant is distracted after the prime and the evaluation of the dependent variable is delayed (Greenberg et al., 1994). This is thought to occur for two reasons. First, the accessibility of a primed concept deteriorates over time (Fiske & Taylor, 1991). Second, experiencing goal-obstruction produces increased goal-related thoughts only after a distraction and delay (Martin & Tesser, 1993). This signifies that when a morality salience prime occurs, it is viewed as a goalobstruction because of the presence of an alternative worldview. However, when a mortality salience prime deteriorates, an individual has increasing goal-related thoughts (i.e., worldview reaffirming thoughts). Connecting Policing and TMT As one can see from the types of experiments reviewed above, TMT has never been directly applied to law enforcement officers or police use of force directly. However, TMT has been applied to situations in which a person might invoke force. It is important to note that some people believe the reactions observed in TMT research might be a product of the research itself. Specifically, some research indicates participants might be more willing to advocate extreme responses because they are responding to hypothetical vignettes in which they are not responsible for the consequences of the decisions they are advocating (Cohen & Insko, 2008). The key area of this integration is the use of violence. Mortality salience primes (and TMT) significantly affect people’s beliefs in the acceptability of the use of violence in certain situations. Specifically, research has demonstrated that, after receiving a mortality salience prime, participants indicated increased support for the extreme antiterrorism policies of President George W. Bush (Landau et al., 2004). Additionally, mortality salience has led to political conservatives expressing higher levels of Bacceptable^ collateral damage caused by the use of extreme military tactics in the capture of Osama Bin Laden (Pyszczynski et al., 2006). These types of attitudes are not only prevalent for Americans in regard to the Middle East, but also in other countries as well. For example, Israelis who support never returning the Gaza Strip to the Palestinians were more supportive of violent anti-terrorism policies (Hirschberger & Ein-Dor, 2006). Based on the results of these studies, it has been well documented that mortality salience increases prejudice, intergroup hostility, aggression, and support for war (for review see Pyszczynski et al., 2003). It is important to note that mortality salience primes do not affect all people in the same way. Evidence suggests that those most susceptible to respond to mortality salience primes with hostility and violence are those defined as rigid authoritarians; these individuals believe in absolute good and evil, and that they know the absolute Btruth^ in the world (Jost et al., 2003; Jost & Hunyady, 2005). Examples of people who are likely to fall into this category are religious fundamentalists, those who score high on measures of authoritarianism (e.g., The Right Wing Authoritarian Scale; Altemeyer, 1996), and those who are politically conservative. 216 Am J Crim Just (2015) 40:205–224 There are also concerns that mortality salience primes are most effective in certain situations. Although mortality salience primes are effective at producing results some of the time, these primes should not be taken as an indicator of an inevitable outcome (Pyszczynski et al., 2006). Mortality salience primes direct people towards subjective attitudes, behaviors, and feelings of safety and security. Therefore, if a person’s beliefs, behaviors, or attitudes associated with violence making them feel more safe and secure, it is more likely the person will respond with violence. Integrating SLT and TMT to Explain Police Shootings of Unarmed Suspects Although there are distinct differences in the SLT and TMT models, there are also many similarities in how they function. This integration will show how TMT works as an ancillary process to SLT, potentially explaining why police officers shoot unarmed suspects. It appears that SLT cannot fully explain the problem of police shootings of unarmed suspects; otherwise, there would be a systematic pattern to the events. Specifically, officers would be killing every unarmed person that met a certain criteria; however, this does not appear to be the case. Additionally, TMT cannot fully explain police shootings of unarmed suspects either. Again, officers would be shooting more people that violate their cultural worldview supplied by the police subculture. Although neither SLT nor TMT can fully explain law enforcement shootings of unarmed suspects, together, they present a convincing explanation of the phenomenon. The specific method in which SLT and TMT function together is in an end-to-end linear fashion. Specifically, SLT and the police subculture work in tandem to teach that violence, in certain situations, is normative. It appears that SLT and the police subculture work with a feedback effect. The police subculture serves to teach officers via SLT, and, when officers experience an adverse event, the subculture is reinforced through the learning process. Unlike other situations (e.g., drug using or criminal offending) in which one can passively learn acceptable behavior of a culture, the police subculture actively forces behavioral conformity. Indoctrination into the police subculture likely begins on the first day of the police academy. At this time, officers are instilled with definitions of appropriate conduct. Some of the definitions are positive (e.g., it is acceptable to protect another officer by any means necessary), some are negative (e.g., you should not use drugs or steal), and many are neutralizing (e.g., it is acceptable to use deadly force against a suspect in certain situations). Once an officer has had ample opportunity to learn the definitional standards taught at the academy (upon which he will be repeatedly tested to ensure retention), the officer is transferred to a Field Training Officer (FTO). The officer’s FTO serves many legitimate purposes (e.g., ensuring the new officer’s safety and ensuring compliance with legal standards). Additionally, the FTO serves two primary functions for the learning of the police subculture. First, the FTO confirms that the officer complies with the definitions that were instilled in the police academy. Second, he models acceptable behavior for the new officer to emulate. In many departments, the FTO uses a daily observation report to rate the officer’s success in many different areas. These areas include: use of force, driving skill, interaction with peers and supervisors, and general appearance (Chappell et al., 2005). The officer is typically rated on a scale Am J Crim Just (2015) 40:205–224 217 and must obtain a certain score in each of the areas to maintain employment. Much of this grading procedure is subjective and based on the experiences and expectations of the FTO. The police subculture also teaches officers to work within a system of differential reinforcements. One could make the case that the FTO grading system serves as a differential reward for officers when first starting in law enforcement; however, the differential rewards do not end there. Officers are differentially reinforced by uppermanagement and by their peers within the police subculture. For any particular incident, officers may observe upper-management reacting in a stereotypically punitive manner, while the officer is socially rewarded by his peer group. An example of this might be that of an officer who crashes a patrol car as a result of over-zealously pursuing a fleeing traffic offender. In this instance, upper-management is likely to punish the officer for damaging the patrol car, whereas the officer’s peers are likely to reinforce his behavior. Although impossible to say for certain the method that the peers might choose to reinforce the behavior, it is likely to be a social reinforcement. The social reinforcement could come in the form of increased levels of perceived masculinity or toughness. This belief falls in line with previous research which indicates that the police subculture thrives on the concepts of toughness and masculinity (e.g., Herbert, 1998; Waddington, 1999), which would then increase the officer’s social capital amongst his peers. The final component of the SLT model is differential association. The differential association component of law enforcement creates a defined Bus versus them^ mentality, in which all people outside of law enforcement are treated skeptically because they are thought to be liars (Van Maanen, 1974). In this subculture, officers learn that people who are not law enforcement officers are potentially dangerous and will try to hurt them if given the opportunity (e.g., Rubenstein, 1973). This reinforces the belief that police officers need only associate with other police officers, which also strengthens the Bus versus them^ mentality. Moreover, by associating with a group of people that have markedly similar definitions and role models that are reinforced for similar behaviors, the police subculture itself is further strengthened. It is clear that police officers develop a subculture that is designed to protect them from the inherent physical dangers associated with the job (e.g., being killed or injured) as well as the political dangers (e.g., a punishment-centered upper management). The subculture also reinforces officers’ behavior through peer validation and other forms of social reinforcement. However, what is not clear is why police officers shoot unarmed suspects. 7 A likely answer is seen in the police subculture itself; specifically, in the striking resemblance to the concept of worldviews as described by Greenberg et al. (1986) in the original presentation of TMT. The police subculture is designed to serve a similar function for police officers as the worldviews concept does for entire societies. In addition to the police subculture acting as a worldview, the subculture also serves to act as a mortality salience prime by consistently reminding officers of their own death that could likely be associated with the job. The effects of each of these are explored below. 7 The effects of shooting an armed suspect likely could not be determined because the officer is shooting the suspect in defense of themselves or others in the immediate area; ergo only unarmed suspects are considered because the officer’s legal justification for the use of deadly force is substantially diminished. 218 Am J Crim Just (2015) 40:205–224 The effects of a mortality salience prime typically force a person to more fiercely defend his worldview (Greenberg et al., 1990). Recall that the effects of a mortality salience prime are most effective after a delay and a distraction (Greenberg et al., 1994). This makes the frequently presented mortality salience prime that police officers are exposed to through their subculture an effective tool for convincing a police officer to defend his worldview (i.e., the police subculture). It is important to note that there are a wide variety of mortality salience primes that can be regularly presented in police work. On the most simplistic level, officers are reminded of their own mortality each day when they go to work adorning bullet-resistant Kevlar vests and handguns. These items are worn for protection against death, while performing the duties of the job. Additional, mortality salience primes come in the form of training, especially in officer safety and street-combat training. This is important because police officers are mandated by law, in most states, to attend firearms training at least twice a year. In some states, this consists of traditional firearms training (e.g., shooting at paper targets), while agencies in other states use innovative training methods to simulate survival situations (e.g., FATS) . Furthermore, each officer-involved shooting, murder, or assault on a police officer is highly publicized and dissected within the law enforcement community under the pretense of preventability. The mortality salience prime directs the officer to seek out safety and security when his worldview is confronted. This safe and secure feeling is obtained by annihilating the threat to the officer’s worldview. The reason that annihilation is chosen over other methods of worldview defense is, again, a product of the police subculture. Research points to three distinct explanations that indicate why the police subculture would choose annihilation over other options. First, police officers mythically believe that police work revolves around adventure, action, and the use of force (Brown et al., 1993; Fletcher, 1996; Holdaway & Parker, 1998). Second, the police subculture idealizes aggressive and authoritative actions (Cochran & Bromley, 2003). Third, the police subculture has a fascination with weapons and conflict, which is deeply instilled in officers through the learning process (Brown & Sargent, 1995; Chan, 1996; Herbert, 2001). This body of research seems to indicate that law enforcement officers are quite comfortable using weapons and force in a myriad of situations. Therefore, in accordance with the Pyszczynski et al. (2006) claim, police officers would likely use these tactics to resolve a threat to their worldview. This would likely be the most frequently-used solution to threats against the officer’s worldview, especially considering that weapons often serve as a mortality salience prime, and that police officers are adept at, and comfortable with, using weapons. Although this integration has focused primarily on annihilation, it should be noted that it is not the only solution. Another potential solution that officers might use to protect their worldview, which has been supported in the literature, is derogation. Evidence suggests that police officers draw a distinction between themselves and the members of society (Rubenstein, 1973, Sparrow et. al, 1990, Tauber, 1970). This division creates an in-group and an out-group, which makes it easier to disparage the members of the out-group. While derogation is a possible solution to maintain an officer’s worldview, annihilation is likely more pertinent solution because of the Am J Crim Just (2015) 40:205–224 219 importance placed upon weapons, use of force, and societal dangerousness within the police subculture. The other three worldview defenses as described in Pyszczynski et al. (2003) are not applicable to law enforcement officers because of the nature of the police subculture. Law enforcement officers would likely not use the assimilation, accommodation, or conversion tactics. Specifically, assimilating or accommodating another worldview into the police subculture is not a likely solution because the police subculture is riddled with independence and insensitivity (Brown & Sargent, 1995, Chan, 1996, Herbert, 2001). The tactics that are used in these worldview defenses quickly discount police officers from using them. Furthermore, conversion is also likely to be ruled out as a solution because it does not logically fit into what is known about the worldview of police officers and the police subculture. The police subculture creates the divide of Bus versus them^ (Rubenstein, 1973, Sparrow et. al, 1990, Tauber, 1970), which has the implicit judgment that Bwe^ (the police) are better than Bthem^ (everyone else). A conceptual model of the causal process outlined above is depicted in Fig. 1. Examples of the Process Evidence suggests that there might be a connection between mortality salience primes and police shootings of unarmed suspects. For instance, the National Law Enforcement Memorial indicates prior to several of the well-known shootings of unarmed suspects discussed above, a law enforcement officer was killed in the line of duty near the location of the shooting of the unarmed suspect. For instance, on November 16, 2014— 6 days prior to the shooting of Tamir Rice—an officer with the Akron Police Department was shot and killed in the line of duty. Akron is approximately thirty miles from the city of Cleveland. This incident was likely in the forefront of the minds of the officer who shot and killed Tamir Rice. Additionally, 3 days prior to the Amadou Diallo shooting, a police officer in the New York City Health and Hospital Police Department Fig. 1 Conceptual model of the causal process 220 Am J Crim Just (2015) 40:205–224 was killed in the line of duty by an unarmed suspect who kicked the officer in the chest—stopping his heart. This incident would have likely been reported to the officers that ultimately shot and killed an unarmed Diallo. The mortality salience prime likely made the officers hold to their authoritarian and aggressive worldview taught through, and reinforced by, the police subculture. Simultaneously, it validated the assumption that the public is dangerous and out to kill or injure law enforcement, and, thus cannot be trusted. Therefore, when Diallo failed to respond to verbal commands, the officers saw him as a potential threat to their worldview and reacted in the manner that made them feel most secure. Unfortunately, this is not an isolated incident. A similar string of events unfolded prior to the Sean Bell shooting. A few weeks prior to the shooting of Sean Bell, other New York City Police officers were engaged in an extensive shootout with a suspect who was assaulting another man. Officers in this case fatally shot and killed the suspect. Again, it is likely that this information was passed onto the officers who shot at Sean Bell and his friends. This information could have served as a mortality salience prime, which increased the officers’ perceptions and general paranoia of armed suspects. It is important to note that not all cases of police shooting unarmed suspects are preceded by the type of mortality salience prime described above. There are other types of mortality salience primes (e.g., violent attacks that do not result in death) that could have similar effects, although these data are not readily available. Although the events that potentially led to the Rice, Diallo, and Bell cases were substantially different, each had a similar effect. The events immediately preceding the officer-involved shootings all served as mortality salience primes to the involved officers, telling them that they needed to protect their worldview. The events had another effect in that they served as reinforcement of the assumptions that the police subculture makes about members of the out-group. However, it likely does not take such drastic and tragic incidents to reinforce the subculture or serve as a mortality salience prime. Specifically, it is possible that police officers go to a training class or a tactical briefing about another situation, which could serve the same purpose. Specific types of training could vary from so called Bforce on force training^ in which officers react to real scenarios with real suspects who are really shooting back at the officers. Research Implications Although this theoretical integration is well founded, there are some difficulties associated with research design. It would be improbable, and most likely unethical, to convince police departments to allow some officers to be trained utilizing a mortality salience prime while other officers are not trained with the mortality salience prime. Some police administrators would be reticent allow this kind of training because it could potentially put officers in the non-primed category in danger; additionally, the officers in the prime condition could pose an undue risk to members of the public. Utilizing scenario-based experimentation (i.e., vignettes) could be a potential solution to this problem, although there is a loss of verisimilitude between the events of real shooting situations conveyed in this manner. One possible solution to this problem would be implement force-on-force training (either FATS® or Simunnitions ®), whereby officers are randomly assigned to each group. This training would present a greater degree of similarity to events on the street; however, there would still be a loss of some Am J Crim Just (2015) 40:205–224 221 realism. In order to convince law enforcement administrators to permit this type of research, it will likely be necessary to empirically substantiate the theoretical model. There is no information currently available that would allow this model to be quantitatively assessed; however, there are many research designs that could be utilized for qualitative analysis. Any qualitative analysis that would be conducted must be able to particularly address the training of the officers, the timeline of events, and any historical events that may have altered the outcome. Policy Implications This theoretical model, in addition to other research, supports modifications to existing policies within law enforcement agencies. One of the key policy implications is related to the training of new law enforcement officers. Currently, research estimates that police officers are trained an inordinate number of hours in the use of firearms (including practical scenarios) and defensive tactics (including weapons retention). Kappeler et al. (1994) indicate that fully 15 % of all academy training is devoted exclusively to the use of firearms. These researchers have figures for defensive tactics and scenario training as well; however, these figures cannot be disaggregated to allow for accurate representation of how much time is devoted to officer survival and fending off attacks from suspects on the street. However, a conservative estimate of an additional 5 % of academy training being devoted to tasks related to officer survival means that one-fifth of valuable academy training is devoted to officer survival. We must concede that the use of deadly force is an important training topic largely because of the adverse consequences that are derived from the use of force. Questionable police uses of force can lead to consequences for the victim, the officer, the department, and the police/community relationship (White, 2003). Despite the fact that law enforcement shootings are quite rare—many officers could go an entire career without using deadly force—these consequences are still ever present. Body cameras are one of the primary mechanisms thought to control police use of force and to improve police-community relations. In one of the few peer-reviewed studies that has examined the issue, Jennings, Fridell, and Lynch (2014) conclude that the police are generally supportive of body-worn cameras. The authors suggest that officers believe that body-worn cameras will serve to improve the behavior of citizens and police officers alike. Further, the use of body cameras will likely increase the transparency of the events because of the presence of video footage. The video and the improved behavior of both parties will likely diminish the incidents of police shootings of unarmed suspects and may ultimately serve to improve police-community relations. Conclusion Law enforcement officers create a subculture conducive to certain types of behaviors via the SLT process defined by Akers (2009). Police officers differentially associate with other officers, which creates an exclusionary in-group. The officers within the subculture are instilled with definitions of Bacceptable behaviors^ through academy and 222 Am J Crim Just (2015) 40:205–224 field training. Furthermore, the police subculture differentially reinforces those behaviors conforming to established definitions and punishing those that do not. This same process is used to teach law enforcement officers about all facets of their job, including the use of deadly force. The law enforcement culture, in conjunction with legal standards and departmental policies, dictate acceptable situations to use deadly force. Police officers use their subculture to create a worldview; moreover, officers protect their worldview from opposing worldviews through derogation and annihilation as described through TMT. Annihilation is the primary mode of protection of the worldview, especially after experiencing a mortality salience prime. Mortality salience primes are frequently occurring events in law enforcement; they are provided by the death of another officer, a dangerous shoot-out with suspects, and through training. These mortality salience primes remind police officers of the perils of the job, which could quite possibly lead to their inevitable death. Furthermore, the reliance upon annihilation as a plausible solution feeds back into the police subculture and reinforces the extant beliefs of the subculture. References Ahern, J. F. (1972). Police in trouble. New York: Hawthorn. Akers, R. L. (2009). Social learning and social structure: A general theory of crime and deviance. New Brunswick: Transaction Publishers. Original work published 1998. Alpert, G. P., & Dunham, R. G. (1997). Policing urban America. Illinois: Waveland Press. Altemeyer, B. (1996). The authoritarian specter. Cambridge: Harvard University Press. Anderson, L. B., & Brink, G. (2005). Deputy cleared in drug raid shooting. St. Petersburg Times, p. 1B. Bladwin, J. D., & Baldwin, J. L. (1981). Behavior principles in everyday life. Englewood Cliffs: Prentice Hall. Becker, E. (1962/1971). The birth and death of meaning. New York: Free Press. Becker, E. (1975). Escape from evil. New York: Free Press. Brown, J., Maidmont, A., & Bull, R. (1993). Appropriate skill-task matching or gender bias in deployment of male and female police officers? Policing and Society, 32, 209–223. Brown, J., & Sargent, S. (1995). Policewomen and firearms in British police service. Policing, 18, 1–16. Burgess, R. L., & Akers, R. L. (1966). A differential association-reinforcement theory of criminal behavior. Social Problems, 14, 128–147. California Police Officer and Standards and Training Commission (2004). FTO Critique Form. Sacramento, CA. Available online via http://lib.post.ca.gov/Publications/FTG-A9-ftocritique.PDF. Chan, J. (1996). Changing police culture. British Journal of Criminology, 36, 109–134. Chappell, A. T., Lanza-Kaduce, L., & Johnston, D. H. (2005). Law enforcement training: Changes and challenges. In R. G. Dunham & G. P. Alpert (Eds.), Critical issues in policing: Contemporary readings (5th ed., pp. 71–88). Long Grove: Waveland Press. Chappell, A. T., & Piquero, A. R. (2004). Applying social learning theory to police misconduct. Deviant Behavior, 25(2), 89–108. Cochran, J. K., & Bromley, M. L. (2003). The myth (?) of the police subculture. Policing: An International Journal of Police Strategies and Management, 26(1), 88–117. Cohen, T. R., & Insko, C. A. (2008). War and peace: Possible approaches to reducing intergroup conflict. Perspectives on Psychological Science, 3, 87–93. Conser, J. A. (1980). A literary review of the police subculture: Its characteristics, impact and policy implications. Police Studies, 2, 46–54. Fernandez, M. (2008). In Bell case, black New Yorkers see nuances that temper rage. The New York Times. Fiske, S. T., & Taylor, S. E. (1991). Social cognition. New York: McGraw-Hill. Fletcher, C. (1996). ‘The 250 1b. man in an alley’: Police storytelling. Journal of Organizational Change Management, 9, 36–42. Fridell, L., & Binder, A. (1992). Police officer decision-making in potentially violent confrontations. Journal of Criminal Justice, 20(5), 385–399. Am J Crim Just (2015) 40:205–224 223 Greenberg, J., Pyszczynski, T., & Solomon, S. (1986). The causes and consequences of a need for self esteem: A terror management theory. In R. F. Baumeister (Ed.), Public Self and Private Self (pp. 189– 212). New York: Springer. Greenberg, J., Pyszczynski, T., Solomon, S., Rosenblatt, A., Veeder, M., Kirkland, S., et al. (1990). Evidence for terror management theory 11: The effects of mortality salience on reactions to those who threaten or bolster the cultural worldview. Journal of Personality and Social Psychology, 58, 308–318. Greenberg, J., Pyszczynski, T., Solomon, S., Simon, L., & Breus, M. (1994). Role of consciousness and accessibility of death-related thoughts in mortality salience effects. Journal of Personality and Social Psychology, 67, 627–637. Greenberg, J., Simon, L., Porteus, J., Pyszczynski, T., & Solomon, S. (1995). Evidence of a terror management function of cultural icons: The effects of mortality salience on the inappropriate use of cherished cultural symbols. Personality and Social Psychology Bulletin, 21, 1221–1228. Greenberg, J., Solomon, S., & Pyszczynski, T. (1997). Terror management theory of self-esteem and cultural worldviews: Empirical assessments and conceptual refinements. In M. Zanna (Ed.), Advances in Experimental Social Psychology (Vol. 29, pp. 61–139). Orlando: Academic. Herbert, S. (1998). Police subculture reconsidered. Criminology, 36, 343–369. Herbert, S. (2001). ‘Hard charger’ or ‘station queen’? Policing and the masculinist state. Gender, Place, and Culture, 8, 55–71. Hirschberger, G., & Ein-Dor, T. (2006). Defenders of a lost cause: Terror management and violent resistance to the disengagement plan. Personality and Social Psychology Bulletin, 32, 761–769. Holdaway, S., & Parker, S. (1998). Policing women police: Uniform patrol, promotion, and representation in the CID. British Journal of Criminology, 38, 40–60. Jennings, E. G., Fridell, L. A., & Lynch, M. D. (2014). Cops and cameras: Officer perceptions of the use of body-worn cameras in law enforcement. Journal of Criminal Justice, 42, 549–556. Johnson, K. (2008). Justifiable homicides at highest in more than a decade, FBI says; Analysts point to changing attitudes on the street. USA Today, p. 1A. Jost, J. T., Glaser, J., Kruglanski, A. W., & Sulloway, S. J. (2003). Political conservatism as motivated social cognition. Psychological Bulletin, 129, 339–375. Jost, J. T., & Hunyady, O. (2005). Antecedents and consequences of system-justifying ideologies. Current Directions in Psychological Science, 14, 260–265. Kappeler, V. E., Sluder, R. D., & Alpert, G. P. (1994). Forced of deviance: Understanding the dark side of policing. Prospect Heights: Waveland Press. Landau, M. J., Solomon, S., Greenberg, J., Cohen, F., Pyszczynski, T., Arndt, J., et al. (2004). Deliver us from evil: The effects of mortality salience and reminders of 9/11 on support of President George W. Bush. Personality and Social Psychology Bulletin, 30, 1136–1150. Lifton, R. J. (1979/1983). The broken connection: On death and continuity of life. New York: Simon and Schuster. Martin, L. L., & Tesser, A. (1993). Thought processes following the attainment and none attainment of goals. Unpublished Manuscript, University of Georgia, Athens, Georgia. McKinley, J. (2009). Oakland looks warily ahead after killings. The New York Times, p. A15. Miller, N., & Dollard, J. (1941). Social learning and imitation. New Haven: Yale University Press. Morrison, B. (2000). N.Y. police shooting case heads to jury Four defendants could face 25 years to life in prison. USA Today, p. 3A. Olson, D. T. (1998). Improving deadly force decision making. FBI Law Enforcement Bulletin, 67(2), 1–9. Pitts, S., Glensor, R. W., & Peak, K. J. (2007). The police training officer (PTO) program: A contemporary approach to postacademy recruit training. The Police Chief Magazine, 74(8). Pyszczynski, T., Abdollahi, A., Solomon, S., Greenberg, J., Cohen, F., & Weise, D. (2006). Mortality salience, martyrdom, and military might: The great Satan versus the axis of evil. Personality and Social Psychology Bulletin, 32, 525–537. Pyszczynski, T. A., Greenberg, J., & Solomon, S. (2003). In the wake of 9/11: The psychology of terror. Washington: American Psychological Association. Rosenblatt, A., Greenberg, J., Solomon, S., Pyszczynski, T., & Lyon, D. (1989). Evidence for terror management theory I: The effects of mortality salience on reactions to those who violate or uphold cultural values. Journal of Personality and Social Psychology, 57, 681–690. Roy, R. (2004). Killings by police under-reported : Special report: Deadly but legal. Orlando Sentinel. Rubenstein, J. (1973). City police. New York: Farrar, Strauss, and Giroux. San Jose Police Department (2009). Field training and evaluation program purpose and description. In Field Training Officer (FTO) Program. Retrieved from http://www.sjpd.org/BFO/FieldTraining. Accessed 12 Oct 2014. 224 Am J Crim Just (2015) 40:205–224 Skolnick, J. H. (1966/1994). Justice without trial: Law enforcement in democratic society. New York: John Wiley. Skolnick, J. H., & Fyfe, J. J. (1993). Above the law: Police and the excessive use of force. New York: Free Press. Sparrow, M., Moore, M. A., & Kennedy, D. (1990). Beyond 911: A new era for policing. New York: Basic Books. Sutherland, E. (1947). Principles of criminology (4th ed.). Philadelphia: J. B. Lippincott. Tauber, R. K. (1970). Danger and the police. In N. Johnson, L. Savitz, & M. E. Woldgang (Eds.), The sociology of punishment and correction (pp. 95–104). New York: Wiley. Van Maanen, J. (1974). Working the street: A developmental view of police behaviors. In H. Jacob (Ed.), The potential for reform of criminal justice (pp. 83–130). Beverly Hills: Sage. Waddington, P. (1999). Police (canteen) subculture: An appreciation. British Journal of Criminology, 39, 287– 309. Walker, S. (1977). A critical history of police reform: The emergence of professionalism. Lexington: Lexington Books. White, M. D. (2003). Hitting the target (or not): Comparing characteristics of fatal, injurious, and noninjurious police shootings. Police Quarterly, 9, 303–330. Jon Maskaly Ph.D., is a Visiting Assistant Professor in the Department of Criminology, Law, and Justice at the University of Illinois at Chicago. He received his doctorate degree in criminology from the University of South Florida. His resent research has been published Social Science Research, Group Processes & Intergroup Relations, and Policing: An International Journal of Police Strategies & Management. Christopher M. Donner Ph.D., is an Assistant Professor in the Department of Criminal Justice and Criminology at Loyola University Chicago. He received his doctorate degree in criminology from the University of South Florida. His recent research has been published in Policing: An International Journal of Police Strategies & Management, the Journal of Criminal Justice, Police Quarterly, Computers in Human Behavior, and Social Science Computer Review. Copyright of American Journal of Criminal Justice is the property of Springer Science & Business Media B.V. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. AGGRESSIVE BEHAVIOR Volume 40, pages 56–68 (2014) Moral Disengagement Among Children and Youth: A Meta‐Analytic Review of Links to Aggressive Behavior Gianluca Gini1*, Tiziana Pozzoli1, and Shelley Hymel2 1 2 Department of Developmental and Social Psychology, University of Padua, Padova, Italy Faculty of Education, University of British Columbia, Vancouver, Canada .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. A growing body of research has demonstrated consistent links between Bandura’s theory of moral disengagement and aggressive behavior in adults. The present meta‐analysis was conducted to summarize the existing literature on the relation between moral disengagement and different types of aggressive behavior among school‐age children and adolescents. Twenty‐seven independent samples with a total of 17,776 participants (aged 8–18 years) were included in the meta‐analysis. Results indicated a positive overall effect (r ¼.28, 95% CI [.23, .32]), supporting the hypothesis that moral disengagement is a significant correlate of aggressive behavior among children and youth. Analyses of a priori moderators revealed that effect sizes were larger for adolescents as compared to children, for studies that used a revised version of the original Bandura scale, and for studies with shared method variance. Effect sizes did not vary as a function of type of aggressive behavior, gender, or publication status. Results are discussed within the extant literature on moral disengagement and future directions are proposed. Aggr. Behav. 40:56–68, 2014. © 2013 Wiley Periodicals, Inc. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. Keywords: aggression; bullying; cyberbullying; moral disengagement; meta‐analysis INTRODUCTION Aggressive behavior toward peers during childhood and adolescence has been studied for decades (Dodge, Coie, & Lynam, 2006) and has been shown to be a significant correlate, both concurrently and longitudinally, of poor health and maladjustment in both perpetrators and victims (e.g., Card, Stucky, Sawalani, & Little, 2008; Gini & Pozzoli, 2009, 2013; Ttofi, Farrington, Losel, & Loeber, 2011). Although the majority of aggressive children display temporary or desisting aggressive behavior, about 10% of the general population are persistently aggressive over the years and can follow a deviant “career” path (Pepler, Jiang, Craig, & Connolly, 2008). Personal correlates and risk factors for youth aggressive behavior include positive attitudes toward (Carney & Merrell, 2001; Rigby & Slee, 1993) and high self‐efficacy for (Andreou & Metallidou, 2004) the use of aggression, low empathy (e.g., Gini, Albiero, Benelli, & Altoè, 2007a; Jolliffe & Farrington, 2011), and high masculinity (Gini & Pozzoli, 2006). Individual differences in aggression have also been attributed to biases in morality, with aggressive behavior linked to distorted moral reasoning that helps to minimize guilt (Arsenio & Lemerise, 2004; Caravita, Gini, & Pozzoli, 2012; Hymel, Schonert‐Reichl, Bonanno, Vaillancourt, & Rocke Henderson, 2010; Malti, Gasser, & Gutzwiller‐ Helfenfinger, 2010; Menesini, Nocentini, & Camodeca, 2013; Tisak, Tisak, & Goldstein, 2006). The present study focused on one particular type of moral reasoning, moral disengagement (henceforth MD), as described in Bandura’s social cognitive theory of moral agency (1986, 1990). Specifically, this study reports on the first meta‐analytic synthesis of developmental research on the relation between MD and aggressive behavior in school‐age children and youth. We also explored the factors that might moderate the effect of MD on aggression, including the type of aggressive behavior, participant characteristics (age, gender), and methodological features of the studies conducted to date. Social Cognitive Theory of Moral Behavior Bandura (1986, 1990, 1991) focused on moral reasoning and its relation to social behavior in an attempt to  Correspondence to: Gianluca Gini, Department of Developmental and Social Psychology, University of Padua, via Venezia 8, Padova 35131, Italy. E‐mail: gianluca.gini@unipd.it Received 10 November 2012; Accepted 24 July 2013 DOI: 10.1002/ab.21502 Published online 13 September 2013 in Wiley Online Library (wileyonlinelibrary.com). © 2013 Wiley Periodicals, Inc. Moral Disengagement and Aggressive Behavior explain how “good” people can behave “badly”. With age, children develop standards of right and wrong that serve as guides for their conduct. Through this self‐ regulatory process, individuals usually act in ways that give them satisfaction and a sense of self‐worth, and tend to avoid behaviors that violate their moral standards in order to prevent or minimize self‐condemnation. However, Bandura (2002) argued that the self‐regulation of behavior involves more than just moral reasoning, and that moral reasoning is linked to moral behavior through a series of self‐regulatory mechanisms through which moral agency is exercised. Moreover, the development of self‐regulation does not create an invariant control system within a person, and there are many psychological and social processes by which self‐sanctions can be disengaged. Selective activation and disengagement of internal control permit different types of conduct— sometimes very negative—with the same moral standards. Specifically, Bandura described eight mechanisms, clustered into four broad categories through which moral control can be disengaged (see Hymel et al., 2010, for a more detailed discussion). The first, cognitive restructuring, operates by framing the behavior itself in a positive light, by (i) portraying immoral conduct as warranted (moral justification); (ii) contrasting a negative act with worse conduct (advantageous comparison); or (iii) using language which palliates the condemned act, thus diminishing its severity (euphemistic labeling). The second set of disengagement strategies operates by obscuring or minimizing one’s agentive role in the harm caused (displacement or diffusion of responsibility). The third set of strategies operates by minimizing, disregarding or distorting the consequences of one’s action, allowing individuals to distance themselves from the harm caused or to emphasize positive rather than negative outcomes (minimizing or misconstruing consequences). Finally, negative feelings can be avoided by stripping the recipients of detrimental acts of human qualities (dehumanization) or considering aggression as provoked by the victim (attribution of blame). These mechanisms can lead to aggressive behaviors through a process of MD, that is a partial gap between the “abstract” personal idea of moral behavior and the individual’s real life behavior. In this way, the individual protects him/ herself from negative feelings, such as guilt or shame, that usually follow immoral conduct (Bandura, 1991). Measures of Moral Disengagement Bandura et al. were the first to develop self‐report scales to measure proneness to MD (Bandura, Barbaranelli, Caprara, & Pastorelli, 1996; Caprara, Pastorelli, & Bandura, 1995). A short version (14 items) of his moral disengagement scale has been adapted for use with elementary school children, a 24‐item version has been 57 developed for adolescents, and a longer scale consisting of 32 items is used with adults (Caprara et al., 1995). The scale has also been adapted to specific populations (i.e., American minority youth; Pelton, Gound, Forehand, & Brody, 2004) and it is sometimes used in revised versions including subsets of the original items (e.g., Ando, Asakura, & Simons‐Morton, 2005; Barchia & Bussey, 2011). The scale has shown good reliability (a ¼ .81; Bandura et al., 1996), and is by far the most commonly used measure of MD across countries. A few studies have used different scales of MD. Hymel, Rocke‐Henderson, and Bonanno (2005) also utilized self‐reports to assess MD in children and youth, although their survey focused specifically on MD regarding peer bullying. The original 18 items of the scale were identified “post hoc” from a larger survey about bullying as reflecting the four broad categories of MD outlined by Bandura (2002). However, factor analytic results failed to distinguish the four different types of MD and instead yielded a single, 13‐item scale tapping overall MD with regard to bullying (a ¼ .81). Nevertheless, this measure has been adopted by other researchers (Almeida, Correia, Marinho, & Garcia, 2012; Vaillancourt et al., 2006). Although the Bandura and Hymel et al. scales show a significant degree of conceptual overlap, a recent study considering a subset of items from each scale indicated a moderate association between the two (r ¼ .51; Ribeaud & Eisner, 2010), with the Bandura scale tapping a broader range of MD beliefs and the Hymel et al. scale tapping a more restricted set of MD beliefs about peer bullying.1 Moral Disengagement and Different Types of Aggressive Behavior Starting from early age, individuals who morally disengage may perceive some types of antisocial behavior as reasonable or justified, at least under some circumstances, even if they have internalized moral rules that prohibit such behavior. Indeed, research has shown that children and youth who endorse these mechanisms are more likely to engage in both general aggression (e.g., Bandura et al., 1996; Caprara et al., 1995) and peer bullying (e.g., Gini, 2006; Hymel et al., 2005). Importantly, the link between MD and aggressive behavior remains significant even after other predictors of such behavior, such as aggression efficacy, rule perception, or parenting, are controlled (e.g., Barchia & Bussey, 2011; Caravita & Gini, 2010; Pelton et al., 2004). Interestingly, 1 Menesini et al. (2003) have investigated MD by assessing children’s attributions of morally engaged emotions (guilt, shame) versus disengaged emotions (pride, indifference) in response to hypothetical moral transgressions. In this meta‐analysis we did not include that study because its methodology differs importantly from those considered here. Aggr. Behav. 58 Gini et al. MD has been shown to be a significant correlate of these behaviors in juvenile delinquents samples (Hodgdon, 2010; Kiriakidis, 2008; Shulman, Cauffmann, Piquero, & Fagan, 2011), representing extremely violent individuals, as well as community samples, thus confirming that MD mechanisms operate within the “normal” range of psychological functioning (Bandura, 1986). Of recent interest is the degree to which MD is associated with cyberbullying, defined as aggressive behavior perpetrated via information and communication technologies, such as the Internet and mobile phones (Smith et al., 2008). Several authors have suggested that MD might be less evident with cyberbullying, albeit for different reasons. Pornari and Wood (2010), for example, suggest that online aggression may not demand the same level of rationalization and justification as traditional aggression because youngsters might consider cyberbullying as less serious than traditional forms of aggression and “the anonymity, the distance from the victim, and the consequences of the harmful act do not cause so many negative feelings (e.g., guilt, shame, self‐condemnation), and reduce the chance of empathizing with the victim” (p. 89). Others (e.g., Bauman, 2010; Perren & Sticca, 2011) argue that the inability to observe the immediate reaction of the victim may allow the aggressor to minimize the impact of the negative behavior; this would make MD less necessary. Indeed, the “online disinhibition effect” (Suler, 2004), which refers to a loosening of social/moral restrictions and inhibitions during online interaction that would otherwise be present in face‐to‐face interaction, itself can represent a variation of MD, allowing the individuals to behave in ways that are contrary to their moral code. Drawing upon this literature, our first aim was to evaluate the strength of the association between MD and any form of peer‐directed aggressive behavior among school‐age children and youth. Of additional interest was whether the magnitude of this effect varied as a function of the behavior considered (e.g., aggression vs. bullying vs. cyberbullying). Testing Potential Moderators Three categories of potential moderators were hypothesized to influence the relationship between MD and various forms of aggressive behavior. First considered are participant characteristics, specifically age and sex. Previous longitudinal research by Paciello, Fida, Tramontano, Lupinetti, and Caprara (2008) examined stability and change in MD and its relation to aggressive behavior among 366 Italian adolescents, followed at four time points from 14 to 20 years. Although generally MD appeared to decline with age, especially between ages 14 and 16, four distinct trajectories were identified: (i) non‐ disengaged adolescents (37.9% of the sample) who initially showed low levels of MD followed by a Aggr. Behav. significant decline, (ii) a normative group (44.5%) with initially moderate levels that later declined, (iii) a “later desistent” group (6.9%) that started with initially high‐ medium levels followed by a significant increase from ages 14 to 16 and an even steeper decline from ages 16 to 20, and (iv) a “chronic” group (10.7%) that maintained constant medium‐high levels of MD. Importantly, youth who maintained high levels of MD were more likely to engage in aggressive acts in later adolescence. At least one study has reported age differences in mean levels of MD favoring older students (e.g., Barchia & Bussey, 2011), whereas others reported no significant age differences (e.g., Bandura et al., 1996; Pornari & Wood, 2010). However, these studies did not assess directly whether age‐groups differed in the link between MD and aggressive behavior. This meta‐analysis tested whether the relation between MD and aggression varied as a function of age, comparing studies of children versus adolescents. Consistent with Bandura’s idea that moral disengagement develops over time as a result from behaving in contrast to internal moral values, it was expected that the relation between MD and aggression would be stronger in adolescence as compared to childhood. Regarding sex differences, higher levels of self‐ exonerating mechanisms have consistently been found in male as compared to female samples from different cultural contexts, even after controlling for other demographic variables, such as ethnicity or socio‐ economic status (Bandura et al., 1996; Obermann, 2011; Yadava, Sharma, & Gandhi, 2001). Less clear is whether the magnitude of the relation between MD and aggression varies across boys and girls. Some authors have suggested stronger links for boys (Bussman, 2007; Paciello et al., 2008), others report the reverse (Yadava et al., 2001), and still others find no moderating effect of gender (Gini, 2006; Obermann, 2011). Accordingly, the present meta‐analysis tested directly whether sex moderated the link between MD and aggressive behavior. Discrepancies in findings across studies as a function of methodological differences were also considered. First, because different scales to measure MD exist, we tested whether effect sizes varied as a function of the type of instrument, by comparing studies that used the original scale devised by Bandura, studies that used a revised shortened version of that scale, and studies that employed other scales. Second, although MD is always assessed through self‐reports, studies differ in their assessments of children’s aggressive behavior, with self, peers, and adults (teachers, parents) used as sources of information. As demonstrated in a previous meta‐analysis by Hawker and Boulton (2000) on the relations between peer victimization and psychosocial adjustment, use of the same informant for both constructs can inflate the Moral Disengagement and Aggressive Behavior magnitude of the measured effects due to shared method variance. Accordingly, this meta‐analysis tested whether this feature accounted for differences in effect sizes across studies. Finally, publication bias is a threat to any meta‐analytic review, with concern that unpublished studies are more likely to have smaller or non‐significant results and less likely to be included in a meta‐analysis than published studies, yielding estimated effect sizes larger than those that actually exist. To reduce publication bias, efforts were made to include as many unpublished studies as possible (e.g., dissertations, conference papers). To check whether a significant difference existed between published and unpublished studies in the reported effect sizes, we also tested for the moderating effect of publication status. METHODS Literature Search Multiple methods were used to identify potentially eligible studies. First, computer literature searches from the year each database started until March 2012 were conducted using PsychInfo, Educational Research Information Center, Scopus and Google Scholar with “moral disengagement,” “aggressive behavior,” “aggression,” “bullying,” “school violence,” “antisocial behavior” used as keywords. Second, recent review articles and book chapters on aggressive behavior, bullying, or morality in children were reviewed for relevant citations. Third, reference sections of the collected articles were searched for relevant earlier references (i.e., “backward search” procedure). Finally, authors were contacted directly to obtain other relevant studies. With unpublished studies (conference papers, dissertations), principal investigators were contacted to ask for ad hoc analysis (if no response was received, a second e‐mail was sent 2–3 months after the first). A total of 70 potentially relevant journal articles, chapters, conference and dissertation abstracts were reviewed. Inclusion Criteria The most basic requirement for inclusion in the present meta‐analysis was consideration of measures of Bandura’s MD mechanisms and any form of aggressive behavior, or bullying, or cyberaggression/cyberbullying, including self‐report questionnaires, as well as peer‐, parent‐, or teacher‐reports. Studies were excluded if the aggression items were part of a wider measure (e.g., a scale measuring externalizing problems) and a separate effect size was not available. In one case (Hyde, Shaw, & Moilanen, 2010), the original author was able to calculate, upon request, the effect size for aggressive 59 behavior from a broader measure of externalizing problems at the age of 15 and this study was then included in the present meta‐analysis. Second, eligible studies were required to have enough quantitative information to calculate effect sizes. Therefore, studies based on interviews or open‐ended questions were excluded. Third, study participants were school‐age children or adolescents from the community, with studies involving clinical samples or incarcerated offenders, and studies of adults excluded. Finally, both published reports (i.e., journal articles) and unpublished studies (e.g., conference papers, doctoral theses) were considered. In the latter case, data were obtained from the principal investigator or his/her supervisor. When multiple reports (e.g., a conference paper or dissertation and a published article) presented results from the same sample, only one effect size was used in the meta‐analysis. Using these inclusion criteria, the final sample of the current meta‐ analysis included 27 studies; 12 examined the relation between MD and general aggression, 11 considered MD and bullying and four considered MD and cyberbullying (see Table I). All studies were coded independently by the first and the second author, using an a priori coding scheme, recording authors and year of publication, the type and form of MD and aggression measures used (self‐report vs. peer/adult reports), sample size, national setting, and demographic characteristics of participants (age, gender). Inter‐rater agreement was found to be very good; all Cohen’s kappas exceeded .92. Discrepancies were resolved by discussion. Data Analysis Pearson’s r was used as the effect size metric, because almost all studies provided zero‐order correlation coefficients between the constructs of interest. In three cases (Bacchini, Amodeo, Ciardi, Valerio, & Vitelli, 1998; Del Bove, Caprara, Pastorelli, & Paciello, 2008; Hymel et al., 2005), the effect size was calculated from the comparison between a group of aggressive children and a control (non‐aggressive) group, by first calculating the standardized mean difference and then converting it into r (for details see Borenstein, Hedges, Higgins, & Rothstein, 2009, p. 48; Card, 2011, p. 100–101). Most studies did not report data separately for boys and girls. Given that one aim of the study was to test the possible moderating role of gender, authors were contacted and asked for ad‐hoc analyses. This resulted in 45 correlation coefficients disaggregated by gender (Hyde et al.’s study only provided boys’ effect size). In four cases for a given study we had two independent effect sizes for each gender (e.g., for primary school boys/girls and for middle school boys/girls). However, the very small numbers of these subgroups did not allow for a more detailed Aggr. Behav. 60 Gini et al. TABLE I. Summary of Studies Included in the Meta‐Analysis Authors (Year) Almeida, Correia, Marinho, and Garcia (in press) Ando et al., (2005) Bacchini et al. (1998) Bandura et al. (1996) Barchia and Bussey (2011) Bauman (2010) Bussey and Quinn (2012) Bussman (2008, study 2) Caprara et al. (1995) Caravita and Gini (2010) Caravita, Gini, and Pozzoli, (2011) Del Bove et al. (2008) Fitzpatrick and Bussey (2012) Gini (2006) Gini et al. (2007b) Gini, Pozzoli, and Hauser (2011) Hyde et al. (2010) Hymel et al. (2005) Menesini, Fonzi, and Vannucci (1999) Obermann (2011) Paciello et al. (2008) Pelton et al. (2004) Perren and Sticca (2011) Pornari and Wood (2010) Qingquan, Zongkui, Fan, and Lei (2009) Stevens and Hardy (in press) Yadava et al. (2001) Sample Size (% of Girls) 499 (47.1%) 2,301 169 799 1,285 190 1,152 136 706 538 879 475 708 581 1,084 719 257 468 652 677 349 245 480 359 1,578 290 200 (49.8%) (46.9%) (45.2%) (53.8%) (54.2%) (37.2%) (52.9%) (43.6%) (46.6%) (47.4%) (45.1%) (57.1%) (49.2%) (50.9%) (48.5%) (0%) (43%) (48.2%) (47.6%) (53.3%) (49.4%) (48.9%) (53%) (48%) (60.3%) (50%) Behavior Measure Shared Method Variance Effect Size: r Spain Cyberbullying, SR Yes .28 Japan Italy Italy Australia United States Australia United States Italy Italy Italy Italy Australia Italy Italy Italy United States Canada Italy Denmark Italy United States Switzerland UK China Samoa India Bullying, SR Bullying, SR Aggression, SR, PN, TR, PR Aggression, SR Cyberbullying, SR Aggression Aggression, PN Aggression, SR, PN, TR Bullying, PN Bullying, PN Aggression, SR Bullying, SR Bullying, PN Bullying, PN Bullying, PN Aggression, PR Bullying, SR Bullying, PN Bullying, SR, PN Aggression, PN Aggression, SR, TR, PR Cyber/Bullying, SR Cyber/Aggression, SR Aggression Aggression Aggression Yes Yes Mixed Yes Yes Yes No Mixed No No Yes Yes No No No No Yes No Mixed No Mixed Yes Yes No Yes Yes .25 .16 .27 .27 .32 .47 .16 .20 .18 .20 .22 .31 .22 .27 .13 .20 .59 .14 .22 .17 .13 .42 .40 .23 .54 .30 Age Range National Setting 11–18 12–15 9–14 10–15 12–15 10–14 12–17 9–12 8–14 9–15 8–15 11–18 12–16 8–11 15–17 9–13 15 13–16 8–14 11–14 12–14 9–14 12–18 12–14 9–11 13–18 15–17 Note. Measures of moral disengagement were all self‐reports. SR, self‐report; PN, peer nominations, TR, teacher‐report; PR, parent‐report. analyses and effect sizes were thus pooled by gender group. In order to avoid violation of the assumption of independence, mean effect sizes for the total sample were calculated for those studies reporting multiple effect sizes (e.g., two or more informants for the same behavior) (Becker, 2000; Borenstein et al., 2009). Outlying effect sizes and sample sizes were identified on the basis of standardized z values larger than 3.29 or smaller than 3.29 (e.g., Tabachnick & Fidell, 2001). No outliers were detected for effect size, but there was one study with an outlying sample size (Ando et al., 2005). We winsorized this number of participants (i.e., reduced to the next largest sample size, following Barnett & Lewis, 1994; Lipsey & Wilson, 2000), resulting in an N of 1,578. Data from individual studies were pooled (with comprehensive meta‐analysis program—v.2.2) using a random effects model. To account for variations in sample size, which influences precision with larger samples yielding more precise estimates than smaller samples, each study was weighted by the inverse of its variance (Hedges & Olkin, 1985). Moreover, because the Aggr. Behav. use of correlation coefficients can result in problematic error formulation, the correlation coefficient for each study was converted to the Fisher’s z scale, and all analyses were performed using the transformed values (Lipsey & Wilson, 2000; Rosenthal, 1991). Then, the resulting summary effect and its confidence interval were converted back to correlations for ease of interpretation. A 95% confidence interval (CI) was computed around each mean effect size. Confidence intervals not including zero were interpreted as indicating a statistically detectable result favoring the association between MD and aggressive behavior. Heterogeneity was assessed using the Q statistic (which is distributed as x2 with df ¼ k  1, where k represents the number of effect sizes; Lipsey & Wilson, 2000), evaluating whether the pooled studies represented a homogeneous distribution of effect sizes. Significant heterogeneity indicates that variations in effect sizes are likely due to sources other than sampling error (e.g., study characteristics). Also reported is the I2 statistic, indicating the proportion of observed variance that reflects real differences in effect size (Higgins, Moral Disengagement and Aggressive Behavior Thompson, Deeks, & Altman, 2003). Moderator analyses were conducted to examine this variability. Even though we were able to include several unpublished studies, we evaluated the potential “publication bias” in different ways. We computed the “fail‐safe N” (Nfs) according to the method proposed by Orwin (1983), which is more conservative than the traditional Rosenthal’s Nfs (Rosenthal, 1978, 1979). Orwin’s Nfs determines the number of additional studies yielding null results that would be needed to reduce meta‐analytic results to a negligible result of .05 (Durlak & Lipsey, 1991). We also inspected the funnel plot, which displays effect sizes plotted against the sample size, standard error, or some other measure of the precision of the estimate. An unbiased sample of studies would ideally show a cloud of data points that is symmetric around the population effect size (Field & Gillett, 2010). Moreover, the association between the effect sizes and the variances of these effects was analyzed by rank correlation with use of the Kendall’s t method. If small studies with negative results were less likely to be published, the correlation between variance and effect size would be high. Conversely, lack of significant Study correlation can be interpreted as absence of publication bias (Begg & Mazumdar, 1994). RESULTS Association Between MD and Aggressive Behavior The 27 studies examining the association between MD and aggressive behavior reported data on 17,776 participants aged 8–18. The distribution of effect sizes is presented in Figure 1. Under the random effects model, the mean effect size for the association between MD and aggressive behavior was r ¼ .28, which was significantly different from zero (Z ¼ 11.06, P
Purchase answer to see full attachment