Skip to main content
SearchLoginLogin or Signup

Exploring the Nature and Prevalence of Targeted Violence Perpetrated by Persons Found Not Criminally Responsible on Account of Mental Disorder

Published onMay 08, 2023
Exploring the Nature and Prevalence of Targeted Violence Perpetrated by Persons Found Not Criminally Responsible on Account of Mental Disorder
·

Conflict of Interest

The authors have no conflict or declaration of interest to report.

Corresponding Author

Correspondence concerning this article should be addressed to Madison F. E. Almond, Department of Psychology, Simon Fraser University, 8888 University Drive, Burnaby BC, V5A 1S6; Email: [email protected]; Phone Number: 1-778-782-7239.

Keywords

targeted violence, NCRMD, mental illness, warning behaviors, forensic, psychosis, crime

Introduction

There is relative consensus that targeted violence can be defined as an intentional violent or aggressive action committed by an individual or group against a predetermined target (Meloy et al., 2013; Shouten & Brennan, 2016). It is premeditated and goal-directed in nature and can be distinguished from general violence wherein the decision to act aggressively may take place impulsively or reactively (Shouten & Brennan, 2016). Targeted violence is an important area of study within the broader research domain of violence risk and threat assessment and management, as its premeditated nature may result in a longer timeframe between when the perpetrator makes the decision to commit the act and when they attempt to perpetrate the act, making the crimes potentially foreseeable and thus, preventable (Meloy et al., 2013).

The concept of targeted or premediated violence has been overwhelmingly studied in samples of individuals who have committed serious violent offenses. The term was first popularized by researchers who examined exceptional cases, such as people who attacked public figures (Fein & Vossekuil, 1999) or perpetrated mass murders (Meloy et al., 2001), and has continued to focus on serious physical violence (e.g., school shootings, terrorism, and mass murder; Drysdale et al., 2010; Gill et al., 2017; Meloy et al., 2019). Violence of this severity is infrequent compared to other categories of crime which also can be goal-directed or premeditated (e.g., threats, criminal harassment, assaults), and these acts of lesser severity met the definition of targeted violence given for inclusion purposes in a systematic review of threat assessment literature (Mitchell & Palk, 2016). However, non-severe targeted violence has received less research attention, and it is unknown whether the frameworks proposed to assess for serious targeted violence are applicable to perpetrators of less severe targeted acts. There has also been comparably little examination of targeted violence among individuals with serious mental illness.

Mental Illness and Acts of Violence

Research has shown an association between mental illness and violence, but it is highly complex (Douglas et al., 2009; Sariaslan et al., 2020; Whiting et al., 2021) and the exact nature of the association remains debated in the literature. Although some studies have found evidence that mental illness is associated with a greater likelihood of approach behaviors (perpetrators approaching their prospective victims in-person) and recidivism (e.g., Scalora et al., 2003; Warren et al., 2008), other risk factors such as substance use, prior history of victimization, and firearm possession have been found to predict future targeted violence more strongly than presence of a mental disorder (Aitken et al., 2008; Metzl & MacLeish, 2015).

The authors of a recent systematic review of threat assessment guidelines found that consideration of the mental health of the perpetrator was the most commonly recommended factor when assessing risk for targeted violence (Mitchell & Palk, 2016). A recent structured review demonstrated that the odds of perpetration of any violent behavior increased substantially when a variety of mental disorders (e.g., schizophrenia, bipolar disorder, ADHD), personality disorders, or substance use was present (Whiting et al., 2021); a specific link between psychosis and violence has also been supported (Douglas et al., 2009). As individuals who are found NCRMD in Canada (Crocker et al., 2015a) and internationally (Dirks-Linhorst & Kondrat, 2012) are most frequently diagnosed with a psychotic disorder, the role of the different symptoms of psychosis in targeted violence is an important area of inquiry.

Psychosis and Acts of Violence

It is difficult to assess the specific role of active symptoms of mental illness in violent offenses. One possible exception is symptoms of psychosis, which can provide “symptom-consistent” motivation where rational action is taken based on irrational beliefs (Junginger & McGuire, 2004). Certain types of delusions, such as threat-control override delusions (Hodgins et al., 2003), persecutory delusions more broadly (Ullrich et al., 2014), and misidentification syndromes (i.e., Capgras; Bourget & Whitehurst, 2004) have been found to precede and potentially drive violence. These effects have been proposed for targeted violence broadly (Meloy, 1988) and observed for specific acts of targeted violence (e.g., stalking; Kienlen et al., 1997) as early as several decades prior. More broadly, although some research has observed that large proportions of individuals with schizophrenia have reported that their primary delusional system motivated an action (Patel et al., 2019; Wessely et al., 1993), in other literature, only 5% of offenders reported that psychotic symptoms (e.g., delusions or hallucinations) played an active, motivating role in their criminal behavior (Peterson et al., 2010; Peterson et al., 2014).

A common, related misconception that may limit the ability of professionals to assess and research targeted violence is that logical thought and planning cannot occur simultaneously with severe mental illness. Some experts have asserted that if logical thought and planning is required for targeted violence, then individuals with mental illness may not be capable of targeted violence (Corner et al., 2018). Contrary to this perception, research demonstrates the ability of individuals to commit targeted violence while suffering from psychotic disorders or depression (Capellan, 2015; Meloy et al., 2004; Nielssen et al., 2007, Scalora et al., 2003). Nevertheless, there remains the risk that clinicians may inadvertently dismiss threats and behavioral indicators that severely mentally ill patients express while in care as inconsequential and not worthy of further investigation when in fact they may offer critical clues to future risk and insights into prevention and risk management.

The Role of Warning Behaviors in Targeted Violence

Meloy and colleagues (2012) first proposed specific dynamic behaviors that, when demonstrated, may signal that the individual’s risk of violence is “increasing or accelerating” (p. 260). Preliminary research has demonstrated that people who perpetrate targeted violence may present more warning behaviors as they come closer to committing violence (Meloy et al., 2012). Warning behaviors include pathway behavior (planning, research, or other offense preparation), fixation on the victim or a related cause, identification with a “warrior” mentality or with other perpetrators of violence, novel acts of aggression, a burst of energy prior to the offense, leakage (“communication to a third party of an intent to do harm to a target”; Meloy & O’Toole, 2011, p. 514), “last resort” behavior (i.e., increasing desperation which posits no alternative to violence), and directly threatening the victim. Leakage, the most frequently studied behavior, can be expressed through varied types of communication (e.g., spoken, letters, internet forum posts) to various recipients (e.g., friends, family, the general public; Burnette et al., 2018). The prevalence of leakage in diverse cases of targeted violence has ranged from 13-88% in samples of people who committed mass murders, public figure attacks, lone-actor terrorism, and school-based targeted shootings (Drysdale et al., 2010; Fein & Vossekuil, 1999; Gill et al., 2014; Silver et al., 2018; Vossekuil et al., 2004). There has been a lack of research examining leakage in persons with mental illness who have committed acts of targeted violence, as well as the relevance of other warning behaviors. There are also methodological limitations in the few studies that do exist, such as amalgamation of the presence of the behaviors into a single variable.

More broadly, the presence of more warning behaviors has been shown to discriminate between individuals who carried out violence and those who threatened, but did not perpetrate, violent acts. Warning behaviors were a significant predictor of school-based threat cases being classified as substantive (more serious) rather than transient (less serious) by school threat assessment officials via professional judgment; threats classified as substantive were 36 times more likely to be attempted than transient threats (Burnette et al., 2018). Warning behaviors were also demonstrated by most perpetrators of serious school violence and violent acts of terrorism in two additional studies (Meloy et al., 2014; Meloy et al., 2019). These behaviors have therefore been shown to have a preliminary association with targeted violence in the aforementioned samples but have not been widely studied in relation to less severe violence (e.g., targeted threats, harassment, physical assaults) or in individuals with severe mental illness.

To summarize, studies that have examined rates of targeted violence perpetrated by individuals with severe mental illness have investigated the prevalence of psychotic and other disorders within samples of individuals who have committed varying types of violence. To our knowledge, no study has looked at the inverse: Rates of targeted violence within a sample of forensic psychiatric patients, for whom psychotic or other symptoms of serious mental illness are assumed to be the primary precipitator of the offense they committed. Additionally, no research has examined the presence of delusions and hallucinations directly before or at the time of targeted offenses or investigated the specific content of the delusions and hallucinations participants experienced.

The Current Study

The current study had four primary objectives. First, to establish the rates and nature of targeted violence perpetration within the population of individuals who were found NCRMD within the province of British Columbia (BC), Canada, between 2001 and 2005. Second, to examine the positive psychosis symptomology participants were experiencing at the time of the offense. Third, to compare the participants who had and had not committed an act of targeted violence regarding socio-demographic variables and offense-related variables. Fourth, to conduct an in-depth examination of the offenses of the sub-group of participants who had committed targeted violence and the individual’s behavior preceding this offense. Specifically, we investigated diagnosis at the time of the NCRMD verdict in relation to the index offense, the content of any delusions or hallucinations reported to have been occurring at or directly before their offense, contact with victims or law enforcement prior to the offense, and the prevalence of warning behaviors (Meloy et al., 2012) prior to the offense.

Methods

This study was an extension of the first National Trajectory Project (NTP) conducted in Canada, which examined individuals found NRCMD in three provinces (Crocker, Nicholls, Seto et al., 2015a; Crocker, Nicholls, Seto et al., 2015b; Crocker, Charette et al., 2015; Charette et al., 2015; please see Crocker and colleagues [2015a] for a comprehensive description of study methodology). Therefore, each participant was legally deemed to be suffering from a mental illness that was significant enough to “[render] the person incapable of appreciating the nature and quality of the act or omission or of knowing that it was wrong” (Criminal Code of Canada, 1985, s 16(1)). Any individual charged with a crime in Canada is eligible to use the NCRMD defense if the above criteria apply. The most common index offenses among individuals who have been found NCRMD include assault and uttering threats and other crimes against the person (Crocker et al., 2015a). In addition to the original NTP data, participants’ files were reviewed using a supplemental data collection protocol specific to targeted violence. Ethics approval was obtained from the relevant institutions where the data were housed and the research was conducted.

Participants

From the BC portion of NTP (N = 293)1, a subgroup of participants who had committed targeted violence (n = 59) was identified by the lead author. We defined targeted violence based on well recognized definitions available in the literature: an act of intentional violence committed by an identifiable individual against a predetermined target (Meloy et al., 2013; Mitchell & Palk, 2016; Shouten & Brennan, 2016). Three of the 59 participants originally identified as having committed targeted violence were missing key files, producing a final sample of 56 people who had committed targeted violence (19.1%). This subgroup of participants ranged in age from 35 to 92 years at the time of the offense (M = 53.9); 80.4% (n = 45) were male and 19.6% (n = 11) were female. Most of the targeted violence group had never been in a relationship (85.7%; n = 48), and over half had completed high school at time of the offense (55.4%; n = 31) and lived autonomously prior to their NCRMD offense (57.1%; n = 32). Psychosis or schizophrenia (92.9%; n = 52) were the most common primary diagnoses for this group, followed by substance use disorder (26.8%; n = 15), mood disorder (25.0%; n = 14), and/or personality disorder (23.2%; n = 13); 66.1% (n = 37) displayed comorbid diagnoses. If participants were diagnosed with multiple disorders, they were included in each relevant category. Almost half (48.2%; n = 27) of those who committed targeted violence had a prior criminal history.

Procedure

A succinct description of the participants’ index offenses was previously recorded by trained graduate and postbaccalaureate research assistants for the larger NTP project. This description was reviewed to determine if participants’ offenses involved targeted violence or not, based on the definition specified previously. This decision was initially made by considering if there was any evidence that the participant had selected a specific target (e.g., person, group of persons, or organization) prior to committing the index offense. At this preliminary screening stage, all participants were categorized into three groups: targeted violence (n = 48), not targeted violence (n = 177; a combination of violent and non-violent offenses that were clearly not targeted in nature), and unknown (n = 68; a combination of violent and non-violent offenses that were unable to be classified as targeted violence or not from NTP data).

Hospital files of all patients whose cases were preliminarily categorized as targeted violence or unknown (n = 116) were then coded in detail. Of cases that were confirmed to be non-targeted offenses (n = 237) after detailed review, 206 were further categorized as violent (i.e., reactive or impulsive) and 31 as non-violent. Inter-rater reliability was assessed for decisions to include a case as targeted violence or not using over 10% of study cases and substantial interrater agreement was established (N = 18; kappa = .65; Cohen, 1960); in cases where raters disagreed, a consensus was reached. After an additional 8 cases of targeted violence with complete file information were identified from the unknown category (n = 56), all targeted violence case files were coded by the first author using an additional protocol. Case files included police reports of the index offense, written Review Board decisions and justifications, and expert reports written by licensed psychiatrists attesting to the participant’s mental state at time of offense. Inter-rater reliability was also assessed for quantitative variables in this additional coding protocol, examining over 10% of study cases (n = 6); kappa values ranged from .40 (fair) to 1.00 (excellent) with a mean value of .71 (substantial; Cohen, 1960). One variable, if police issued a warning to the participant prior to their targeted violence, was excluded due to low inter-rater reliability (kappa = .33). All remaining variables except one (participant engaged in negative contact with their victim prior to their offense) displayed a kappa of at least .57.

Measures

Sample characteristics. Most data for the present study were entered into a single bilingual secure online database developed for the original NTP study (Crocker et al., 2015). The online coding protocol included comprehensive information on many aspects of participants’ socio-demographics, criminal history, and interactions with the criminal justice and provincial health systems. For the purposes of this study, socio-demographic variables and information regarding mental illness from the NTP were included. Of particular importance were variables regarding the offense leading to the NCRMD finding (e.g., offense type, victim characteristics, suicidal ideation or self-harm at time of offense, weapon use during the offense).

Mental illness at time of offense. For the present study, all available clinical file information previously coded for the NTP was recoded into new variables to afford more detail as to the presence or absence of hallucinations and/or delusions before or during the offense. These clinical files included police reports and psychiatrists’ reports to the Review Board, which entailed extensive violence risk assessments in preparation for the NCRMD evaluations. Information about diagnosis, presence and content of delusions, and presence and content of hallucinations (all coded for presence at time of the offense) was collected from hospital records. The information used to code for symptoms of psychosis reflects the type of information accessible to an assessor of future risk for targeted violence.

Contact with law enforcement and victim(s) prior to offense. The additional protocol coded by the first author separate from the NTP also assessed if there was any written, verbal (e.g., via telephone), or in-person contact with the victim(s) prior to the offense and if law enforcement had issued the participant a warning, a restraining order, or had arrested them before the offense leading to the index NCRMD finding. If contact had been made with the victim(s) or law enforcement prior to the offense, the time from first contact to offense was also coded. A time limit was not given for this contact, but reliance on archival file-based coding may have limited the availability of information that was substantially old (e.g., ten years prior to the index offense). However, this is information the expert would be expected to have included in their description of the relationship of the accused to the victim, their criminal history, and future risk.

Warning behaviors. Mirroring past research (Burnette et al., 2018; Meloy et al., 2012; 2014; Meloy et al., 2019), a coding sheet was developed to identify the presence or absence and qualitative details of eight warning behaviors. All clinical files were reviewed to determine if each warning behavior was displayed by participants who committed targeted violence. First, pathway behavior was considered present if the documentation indicated the accused had researched, planned, or prepared in other ways for the offense, such as obtaining a weapon or writing out details of the planned offense. Second, fixation was coded if persistent thoughts of the victim or consistent efforts to contact them were recorded. Identification necessitated documentation of participants associating themselves with organizations such as the police or military, as well as past violent offenders, while novel acts of aggression required participants to have committed an aggressive act unrelated to other pathway behavior in the month prior to their targeted violence. This warning behavior was not coded if the participant possessed a substantial history of aggression. Fifth, energy burst behavior was coded as present if a participant’s activities related to their victim noticeably increased within the month before their offense (e.g., increased thoughts of the victim, increased attempts to contact the victim). Sixth, leakage (see Meloy & O’Toole, 2011) was documented if the participant communicated their intent or plan to harm the victim to any other person prior to the action that precipitated criminal charges. Seventh, for last resort behavior to be present, the participant must have expressed overt desperation, desire for injury or death, and/or posited their injury/death as the alternative to their offense. Eighth, directly communicating a threat to the target of the offense included in-person, telephone, and online threats; this was not coded as present if the participant’s index offense was uttering threats. The presence of each warning behavior was recorded for each participant; the total number of behaviors exhibited by each participant was also calculated.

Data Analysis

Descriptive statistics were generated to report the rate of targeted violence within the population of individuals found NCRMD and describe the patient characteristics. Inferential analyses were conducted to compare the socio-demographic characteristics and the NCRMD offense for the subsample who had committed targeted violence with the subsample who had perpetrated non-targeted violent or non-violent offenses. Cases in the targeted violence subsample were not matched with non-targeted violence cases given the relatively small total population of NCRMD cases in BC during the study period (n = 293) and the many unique aspects of each case (e.g., age, education, victim, diagnosis, etc.). Chi-square tests were conducted, with phi coefficients calculated from Wald chi-square values to provide effect size estimates (Rosenthal, 1991). Cohen’s (1988) guidelines informed cut-offs for effect sizes reported (.10 = small, .30 = medium, .50 = large). Fisher’s exact tests were performed when cell size violations were present. Associations significant at the bivariate level were included in a logistic regression model and odds ratios (OR) were calculated to further elucidate which factors demonstrated the strongest relationship with targeted violence. Considering that non-targeted cases included participants who perpetrated violent and non-violent offenses, all analyses that compared targeted and non-targeted cases were also conducted with the small subsample of non-violent cases (10.6%; n = 31) excluded to determine if results were impacted when only violent offenses were compared.

Results

Prevalence and Nature of Targeted Violence among people found NCRMD

The index offense of 19.1% (n = 56) of individuals found NCRMD in British Columbia over a five-year period involved an act of targeted violence. The index offense committed most often by individuals who perpetrated targeted violence was assault (35.7%, n = 20), followed by criminal harassment (30.4%, n = 17), uttering threats (14.3%, n = 8), and murder (7.1%, n = 4). Other participants had committed another type of crime against a person (e.g., unlawful confinement) or a breach or failure to comply with a previous no-contact order (5.4%, n = 3); 7.1% (n = 4) of participants committed unclassified crimes.

Victims of offenses involving targeted violence most often included family members, romantic partners, or ex-romantic partners (39.3%, n = 22), followed by professionals (14.3%, n = 8), acquaintances or friends (14.3%, n = 8), neighbours (8.9%, n = 5), a co-patient, a roommate, or an individual unknown to the accused (e.g., a person who passed by the participant’s house regularly; 8.9%, n = 5), a public figure (5.4%, n = 3), or a co-worker (5.4%, n = 3). There was no individual victim in two cases (3.6%) which involved the accused targeting an organization such as the UN or Canadian government. Victims of the targeted violence offenses were predominantly female (n = 33, 62.3%; male: n = 20, 37.7%); the victim’s gender was unspecified in one case. In most cases, the victim’s ethnicity was unknown (76.8%, n = 43). When victims’ ethnicities were documented (n = 13), the victims were White (7.1%, n = 4), Asian/Pacific Islander (7.1%, n = 4), or reported another ethnicity (8.9%, n = 5). The victims’ sexual orientation was mostly unknown (67.9%, n = 38) or heterosexual (32.1%, n = 18). In 15 cases (26.8%), the accused used a weapon during perpetration of the offense, generally, a knife (14.3%, n = 8), or a blunt object (12.5%, n = 7).

Mental Illness at the Time of Offenses Involving Targeted Violence

All 56 individuals found NCRMD who had perpetrated targeted violence displayed evidence of delusions at the time of the offense; 71.4% (Table 1) were experiencing persecutory or paranoid delusions, 37.5% were experiencing grandiose delusions, 21.4% were noted to have erotomatic delusions, and 7.1% had psychotic symptoms classified as other (e.g., delusions of a somatic nature). The delusional content was described as a motivating factor to commit the offense in 80.4% of cases (45 of the 56).

Hallucinations were present in 37.5% of cases, with auditory hallucinations being the most common (32.1%); visual hallucinations (14.3%) were less common. In five of the 21 cases (23.8%) in which hallucinations were present during the index offense, the hallucination was noted to have played a motivating role in the offense (e.g., accused experienced a threatening visual hallucination of a co-inmate and assaulted that person the next day). Suicidal ideation was evident in 7.1% of cases, a suicide attempt occurred in 1.8% of cases, and self-harm was not present in any case of targeted violence.

Victim Contact Prior to Offenses Involving Targeted Violence

There was a record of the accused engaging in negative or harassing contact with the victim prior to the offense in 38 of the 56 cases (67.9%) (Table 1). Of this prior contact, 48.2% of the cases involved verbal or written communication (e.g., phone calls, one or multiple written letters) and 50.0% of the cases had a history of direct, in-person contact. There was an average of 10.3 months between any type of first contact with the victim and the offense; prior contact length spanned one day to almost seven years (83 months). In 21.4% of cases, the participant was stalking the victim before the offense was committed.

Warning Behaviors and Contact with Police Preceding the Index Offense

There was at least one warning behavior (M = 1.9) present prior to the offense in 92.9% of targeted violence cases. Fixation was the most frequent warning behavior (60.7%; Table 1), followed by planning or preparation (41.1%), and then direct threatening behaviors (35.7%). The five other warning behaviors examined appeared in less than ten cases respectively: novel aggression (16.1%), leakage (10.7%), last resort behavior (8.9%), energy burst behavior (7.1%), and identification behavior (7.1%).

Police had some form of contact with the accused prior to the offense in 25.0% of cases. Of these, police issued a no-go, no-contact, or restraining order to the participant in 7.1% of cases and arrested the participant in 10.7% of cases (see Table 1).

--------------------------- INSERT TABLE 1 HERE --------------------------------

Comparison of Targeted and Non-Targeted Violence Perpetrators

Compared to people who did not, people who perpetrated targeted violence were significantly more likely to utter threats or commit criminal harassment (44.6%; Table 2) than assault or other bodily harm (42.9%; p < .01), to offend against a female victim (62.3%; p = .015), to be diagnosed with schizophrenia or another psychotic disorder (92.9%; p < .01) and/or a personality disorder (23.2%; p < .01), and display delusions at the time of the offense (100.0%, p < .001) (Table 2). The subgroup who perpetrated targeted violence was less likely than the non-targeted group to be diagnosed with a substance use disorder (26.8%; p = .005). All the aforementioned results remained statistically significant when non-violent cases were excluded.

--------------------------- INSERT TABLE 2 HERE --------------------------------

The largest phi coefficients were observed for a diagnosis of a psychotic disorder (ϕ = .42), offense type (ϕ = .22), and presence of delusions at time of offense (ϕ = .22). As diagnoses of a psychotic disorder, personality disorder, substance use disorder, and delusions at time of offense were significantly associated with targeted violence at the bivariate level, these variables were included in a multivariate logistic regression model. The model was statistically significant (Wald χ2 = 32.83, p < .001, Pseudo R2 = .19; see Table 3) and delusions at the time of the offense (OR = 11.96), diagnosis of a psychotic disorder (OR = 4.04), and absence of a substance use disorder (OR = 2.79) remained significantly associated with targeted violence (all p ≤ .02). The model and its components remained significant when non-violent cases were excluded.

--------------------------- INSERT TABLE 3 HERE --------------------------------

Discussion

The present study examined the prevalence and nature of targeted violence perpetrated by individuals with a psychotic disorder or other severe mental illness by examining data on all persons found NCRMD in British Columbia within a five-year period. Our results suggest that the presence of a severe mental illness does not preclude the perpetration of targeted violence, as approximately one in five individuals (19%) found NCRMD had committed targeted violence. In fact, this may underestimate the rate of targeted violence in the sample given that only the index offenses were examined. It is possible that some patients had committed targeted violence in the past, or since their NCRMD finding, meaning the prevalence rate of targeted violence in samples of people with severe mental illness may be underestimated. This is consistent with other studies that found it is not rare for severe mental illness and targeted violence perpetration to co-occur (Capellan, 2015; Gill et al., 2017; Meloy et al., 2001). Taken together, the extant literature suggests that even if an individual shows signs of severe mental illness such as psychosis or depression, they should not be discounted for risk of targeted violence. Similarly, the current study found certain symptoms of psychosis are common among people who perpetrated targeted violence. In 79% of cases where the participant was experiencing delusions, those delusions appeared to have played at least some motivating role in the offense leading to the NCRMD finding. Forensic clinicians are encouraged to consider how psychiatric symptoms may have motivated past violent behavior and/or could increase risk of future offending and weigh these factors more heavily than diagnosis, which has been shown to have a weaker association with violence regarding psychotic disorders (Douglas et al., 2009) and may generally be less relevant within time-sensitive threat assessments (National Center for the Analysis of Violent Crime; 2019).

Relationship of Delusions and Targeted Violence

Some light has been shed on how delusions may relate to violent offending. Ullrich and colleagues (2014) reanalyzed the MacArthur Violence Risk Assessment Study data and found that, while there was no relationship between delusions and violent offending overall, specific delusions such as being spied on, plotted against, or followed, thought insertion, and possessing special gifts or powers significantly predicted an outcome of violence. This reflects the fact that persecutory delusions are a proposed mechanism that can trigger violence in individuals with delusional misidentification syndromes (Bourget & Whitehurst, 2004) and aligns with the finding that persecutory or paranoid and grandiose delusions were the most prevalent types of delusions that occurred directly before instances of targeted violence within the current study. Future research in this area should use stronger methodology to investigate whether paranoid or grandiose delusions are consistently found to be precursors to violence among mentally ill and justice-involved individuals, and if so, these could be considered additional warning behaviors.

The prevalence of hallucinations was comparatively lower than delusions in the current study and when present, hallucinations typically did not appear to play a motivating role in the perpetrator’s targeted violence. This parallels the findings of some prior research, in which hallucinations did not contribute to risk of violence to others regardless of their nature (Bjørkly, 2002; Weltens et al., 2021) but is contrary to other findings (McNiel et al., 2000) and clinical wisdom (Bjørkly, 2002) that command hallucinations in particular increase the risk of violence. Command hallucinations have been found to more consistently contribute to risk of self-directed violence (Wong et al., 2013), which is an equally serious concern for this population. As such, the clarification of the role of hallucinations in violence, and especially targeted violence, should be a priority for future research.

Warning Behaviors among Individuals with Mental Illness who Commit Targeted Violence

It was very common for one or more warning behaviors to be present (93%) before an act of targeted violence was committed by an individual who was found NCRMD within our sample. However, the rates of individual warning behaviors are lower than seen in previous research with different populations and more serious incidents of targeted violence. For example, Meloy et al. (2019)’s methodologically similar retrospective file review study found all but two of the eight warning behaviors were present in most cases, while only one behavior (fixation) was present in over half of cases in the current sample. Leakage was present in just over 10% of our cases, while other studies have reported it was present in over half of their samples (Meloy et al., 2019; Silver et al., 2018). This result contrasts Silver and colleagues’ (2018) supposition that individuals with mental illness may display leakage more frequently due to difficulty with communication regulation; however, these differences must be qualified by the disparity in severity of targeted violence between prior study samples and the current one. It is possible that warning behaviors are less applicable to samples with high rates of harassment and/or threat offenses, as these cases may not have progressed far enough or had enough time to allow opportunity for these behaviors to emerge, meaning that perpetrators’ behaviors may have been noticed and curtailed beforehand. Fixation, the most prevalent warning behavior shown by this sample, was indeed the first typically observed in a recent time-sequence analysis of warning behaviors (Meloy et al., 2021), which further supports this possibility. Warning behaviors may be perpetrated in comparable rates by mentally ill persons who perpetrate serious violent offenses such as mass murder and terrorism. As there is a paucity of research on several other warning behaviors, it is unknown whether behaviors like last resort and energy burst behaviors are occurring at lower rates than in other populations who commit targeted violence offenses.

It is important to recognize the limitations in the extant literature, which has primarily examined people who perpetrated extreme acts of targeted violence with low base rates (e.g., mass shooters), forming the majority of the population for research on warning behaviors and the prediction of targeted violence. Further, most studies have merely identified the prevalence of the behaviors without use of any comparison group (e.g., Allwin et al., 2019; Meloy et al., 2012; Silver et al., 2018;). The current study is one of the first to examine warning behaviors in a sample where criminal harassment and minor assaults were substantially more prevalent than homicide and other severe forms of violence. The findings provide preliminary support for the generalization of warning behaviors to less serious acts of targeted violence, although lower severity of violence may be associated with fewer behaviors and future research should endeavour to separate warning behaviors from other potentially confounding variables.

Differences Between Perpetrators of Targeted and Non-Targeted Violence

Some differences emerged between individuals found NCRMD who had and had not committed an act of targeted violence. Participants who committed targeted violence were more likely to have been diagnosed with a psychotic disorder and to have displayed delusions at the time of their index offense. These results align with the link between delusions, fixation, and targeted violence observed in prior research (Pathé et al., 2016; Meloy & Rahman, 2020). While acknowledging that most people with mental illness are never violent, and violence associated with mental illness accounts for a small proportion of all violence and crime, these findings add to the existing strong evidence that, compared to individuals with other mental illnesses, professionals should be particularly vigilant about risk of future violence for those with psychotic disorders (Douglas et al., 2009; Fazel et al., 2009; Whiting et al., 2021).

It was also found that individuals who committed targeted violence were less likely to commit physical violence than individuals who did not commit targeted violence; instead, criminal harassment or other means of verbal violence was more likely. While this finding may be an artifact of the relatively low severity of violence exhibited by the entire population in comparison to most previous research on targeted violence, other possible explanations include the fact that participants may not be able to physically see others they have relationships with due to constraints of liberty (e.g., supervised housing), geography, or because they may not possess the organization necessary to perpetrate planned violence in person. Furthermore, it is difficult to assess premeditated violent intent in the aftermath of an assault; for instance, when a person may have planned the attack but given no evidence of it. This may mean targeted physical violence is underreported; this underscores the importance of future prospective designs. Hypothesized deficits in patient organizational ability can be refuted by evidence that extremely psychotic individuals can plan complex offenses (see Corner et al., 2018) and this research question is best tested within a comparison group of non-mentally ill perpetrators of targeted and non-targeted violence to further understand these findings.

Participants who committed targeted violence were also more likely to have a diagnosis of a personality disorder, although this relationship was not longer statistically significant when included in the multivariate model with other associated variables. This finding also warrants further research, as personality disorders could not be tested by specific type, only by presence or absence, due to methodological limitations such as sample size. While some evidence exists that antisocial and borderline personality disorders are predictive of general violence (de Brito & Hodgins, 2009; Newhill et al., 2009), there is less reported association between other personality disorders and violence, with traits like impulsivity partially mediating the relationship (Rogier et al., 2019). Clarifying the relationship between personality disorders and targeted violence is important, as personality disorders are notably absent as a risk factor in established targeted violence assessment tools (e.g., the TRAP-18; Meloy, 2018; the WAVR-21; Meloy et al., 2013).

Substance use is a well-established risk factor for violent offending (Bonta et al., 2014; Whiting et al., 2021), however, substance use disorders were significantly less common among people found NCRMD for targeted violence than for non-targeted violent offenses. Although substance use disorders were prevalent in samples who perpetrated targeted violence (Corner et al., 2018), they may be more strongly associated with physical violence versus the non-contact offenses (i.e., harassment, threats) that this sample commonly committed. Substance use may have also reduced the likelihood of a finding of NCRMD: at the time of NTP-1 data collection, there was no accommodation for substance use with respect to criminal responsibility in Canada (Lawrence, 2017). Alternatively, it may be that a substance use disorder limited the accused’s ability to plan a targeted offense, as substance use has been linked with impulsivity (Verdejo-Garcia & Albein-Urios, 2021). Cases in which the patient used substances may have been less likely to be considered targeted violence by raters due to perceived incompatibility of substance use and planning. In sum, previous findings are equivocal and there is a need for additional research to elucidate the relationship between substance use and targeted violence perpetration.

Limitations

The current study relied on a retrospective, secondary analysis of file-based data and therefore, limited variables could be coded with confidence and history of targeted violence prior to the accused’s index offense could not be explored. The cohort of individuals found NCRMD which the present study examined was also finite, limiting our sample size; therefore, these findings should be interpreted with caution and may not be generalizable to larger and more heterogeneous samples of targeted violence perpetrators (e.g., in prisons or the community). In addition, the current sample is not representative of individuals suffering from severe mental illness, as a finding of NCRMD may be more likely for certain kinds of mental illness (i.e., psychosis) and less likely for people with other mental illness (e.g., substance use) and certain kinds of criminal offenses. However, these participants were identified from a population of all individuals found NCRMD within a Canadian province over five years. The number of participants in the current study is also larger than previously published research on targeted violence (Allwin et al., 2019; Meloy et al., 2019) and sheds unique light on psychosis symptoms and warning behaviors displayed by individuals who have committed less serious violent offenses that were still targeted in nature. While this may decrease the validity of comparisons with prior studies of targeted violence, it provides the first empirical investigation into the applicability of these concepts for more common offenses.

Similar to previous targeted violence research, we did not have a comparison group of individuals who committed non-targeted violence or did not commit violence. Similarly, non-targeted violence was compared with other non-violent offenses, leaving it unclear if the variance in the variables studied is due to the perpetration of targeted violence or violence in general. Examining comparison groups is necessary to assess if warning behaviors are truly predictive of targeted violence and can reliably differentiate impending targeted violence from other situations where warning behaviors are displayed but violence does not follow. Currently, to our knowledge, only a few studies have examined warning behaviors in relation to violence with a comparison group of participants who issued threats or other concerning behaviours but were ultimately not violent (e.g., Burnette et al., 2018; Meloy et al., 2019). The former study observed warning behaviors to be significantly predictive of violence, but it is unclear how “warning behaviors” were defined and all warning behaviors were amalgamated. Five of eight behaviors were associated with targeted violence in the latter study of individuals who committed terrorism acts, but it is unknown if this finding is generalizable to other offense types.

Lastly, a low number of cases were coded for inter-rater reliability, and it is also acknowledged that when cases were identified as targeted violence, there was potential for the coder’s decision to be influenced by the characteristics of the case. For example, a case in which the participant explicitly fixated on the victim before the offense may have been more likely to be designated targeted violence than a case where this did not take place. This limitation was mitigated by coders classifying the cases based solely on our operationalization of targeted violence (see Methods). Future studies could use coders who classify cases as targeted violence or not who are blind to the study hypotheses and/or who do not also code warning behaviors.

Directions for Future Research

A crucial next step in threat assessment research is to further examine the predictive validity of warning behaviors and other behaviors found to be prevalent among people who engaged in targeted violence in the current study, such as persecutory or paranoid delusions. Prospective studies that allow for the collection of richer data about the accused’s behavior before the offense, symptoms at the time of the offense, and their motivations for the offense are necessary. In addition, future research on targeted violence should employ adequate control groups of people who are not mentally ill and did not commit targeted violence. There have been previous calls for the threat assessment field to move from case studies and exploratory research with small samples to larger studies with more empirically rigorous methodology (Corner et al., 2018); this is necessary to improve best practices in threat assessment. Until this takes place, the external validity of these potential behavioral precursors to violence will be justly challenged.

Future work should also seek to clarify the operationalization of targeted violence: different studies have used conspicuously different definitions (e.g., Meloy et al., 2013; Mitchell & Palk, 2016; Schouten & Brennan, 2016) and the present study included non-physical violence (e.g., threats, harassment), which some researchers may disagree with. This term is used frequently in threat assessment literature and a consensus on this term should be a key priority to improve the internal validity of future research and the development of policy and practice.

Examining larger and more diverse samples would also contribute to further determining the prevalence of warning behaviors in all types of targeted violence. Critically, future research comparing cases of targeted and non-targeted violence is needed to assess if warning behaviors are indicative of the former. Ideally, these behaviors should be studied within a large sample of diverse offenders to determine if the presentation of these behaviors differs by type of offense. These results suggest persecutory and grandiose delusions may be common in individuals found NCRMD before an act of targeted violence; future studies should investigate if the specific content of the delusion (e.g., whether it is the government or a known acquaintance the individual believes is monitoring them) plays a role in future violence perpetration.

In sum, this study contributes to the extant literature as the first to explore the prevalence and nature of targeted violence perpetrated by individuals with a psychotic disorder or other serious mental illness. We demonstrated that the presence of serious mental illness does not preclude an individual from perpetrating targeted violence; in fact, certain diagnoses and symptoms (i.e., persecutory and grandiose delusions) were commonly displayed by the participants who committed targeted violence. While less prevalent than in other samples, warning behaviors were still present within the current sample. These results could serve to inform future research within the intersection of mental illness and threat assessment and inform clinicians on potential precursors to violence and develop prevention strategies.

References

Aitken, L., Oosthuizen, P., Emsley, R., & Seedat, S. (2008). Mass murders: Implications for mental health professionals. International Journal of Psychiatry in Medicine38(3), 261–269. https://doi-org.proxy.lib.sfu.ca/10.2190/PM.38.3.c

Allwinn, M., Hoffmann, J., & Meloy, J. R. (2019). German mass murderers and their proximal warning behaviors. Journal of Threat Assessment and Management6(1), 1–22. https://doi-org.proxy.lib.sfu.ca/10.1037/tam0000122

Bjørkly, S. (2002). Psychotic symptoms and violence toward others—A literature review of some preliminary findings: Part 1. Delusions. Aggression & Violent Behavior, 7(6), 617. https://doi.org/10.1016/S1359-1789(01)00049-0

Bonta, J., Blais, J., & Wilson, H. A. (2014). A theoretically informed meta-analysis of the risk for general and violent recidivism for mentally disordered offenders. Aggression and Violent Behavior19(3), 278–287. https://doi-org.proxy.lib.sfu.ca/10.1016/j.avb.2014.04.014

Bourget, D., & Whitehurst, L. (2004). Capgras syndrome: a review of the neurophysiological correlates and presenting clinical features in cases involving physical violence. The Canadian Journal of Psychiatry49(11), 719–725. https://doi.org/10.1177/070674370404901102

Burnette, A. G., Datta, P., & Cornell, D. (2018). The distinction between transient and substantive student threats. Journal of Threat Assessment and Management5(1), 4–20. https://doi-org.proxy.lib.sfu.ca/10.1037/tam0000092

Capellan, J. A. (2015). Lone wolf terrorist or deranged shooter? A study of ideological active shooter events in the United States, 1970–2014. Studies in Conflict & Terrorism, 38(6), 395–413. https://doi.org/10.1080/1057610X.2015.1008341

Charette, Y., Crocker, A. G., Seto, M. C., Salem, L., Nicholls, T. L., & Caulet, M. (2015). The National Trajectory Project of individuals found not criminally responsible on account of mental disorder in Canada Part 4: Criminal recidivism. The Canadian Journal of Psychiatry60(3), 127–134. https://doi-org.proxy.lib.sfu.ca/10.1177/070674371506000307

Criminal Code, R.S.C. (1985) c C-46

Cohen, J. (1960). A coefficient of agreement for nominal scales. Educational and Psychological Measurement20(1), 37-46.

Cohen, J. (1988). Statistical Power Analysis for the Behavioral Sciences (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum Associates, Publishers.

Corner, E., Gill, P., Schouten, R., & Farnham, F. (2018). Mental disorders, personality traits, and grievance-fueled targeted violence: The evidence base and implications for research and practice. Journal of Personality Assessment100(5), 459–470. https://doi-org.proxy.lib.sfu.ca/10.1080/00223891.2018.1475392

Crocker, A. G., Charette, Y., Seto, M. C., Nicholls, T. L., Côté, G., & Caulet, M. (2015). The National Trajectory Project of individuals found not criminally responsible on account of mental disorder in Canada Part 3: Trajectories and outcomes through the forensic system. The Canadian Journal of Psychiatry, 60(3), 117–126. https://doi.org/10.1177/070674371506000306

Crocker, A. G., Nicholls, T. L., Seto, M. C., Charette, Y., Côté, G., & Caulet, M. (2015a). The National Trajectory Project of individuals found not criminally responsible on account of mental disorder in Canada Part 2: The people behind the label. The Canadian Journal of Psychiatry60(3), 106–116. https://doi.org/10.1177/070674371506000305

Crocker, A. G., Nicholls, T. L., Seto, M. C., Côté, G., Charette, Y., & Caulet, M. (2015b). The National Trajectory Project of individuals found not criminally responsible on account of mental disorder in Canada Part 1: Context and methods. The Canadian Journal of Psychiatry60(3), 98–105. https://doi.org/10.1177/070674371506000304

de Brito, S. A., & Hodgins, S. (2009). Antisocial personality disorder. In M. McMurran & R. Howard (Eds.), Personality, personality disorder and violence: An evidence-based approach (pp. 133–153). Wiley-Blackwell.

Dirks-Linhorst, & Kondrat, D. (2012). Tough on Crime or Beating the System. Homicide Studies, 16(2), 129–150. https://doi.org/10.1177/1088767912438711

Douglas, K. S., Guy, L. S., & Hart, S. D. (2009). Psychosis as a risk factor for violence to others: A meta-analysis. Psychological Bulletin135(5), 679–706. https://doi-org.proxy.lib.sfu.ca/10.1037/a0016311

Drysdale, D., Modzeleski, W., & Simons, A. (2010). Campus attacks: Targeted violence affecting institutions of higher education. Washington, DC: U.S. Secret Service, U.S. Department of Homeland Security, Office of Safe and Drug-Free Schools, U.S. Department of Education, and Federal Bureau of Investigation, U.S. Department of Justice.

Fazel, S., Gulati, G., Linsell, L., Geddes, J. R., & Grann, M. (2009). Schizophrenia and violence: systematic review and meta-analysis. PLoS Medicine6(8). https://doi.org/10.1371/journal.pmed.1000120

Fein, R. A., & Vossekuil, B. (1999). Assassination in the United States: An operational study of recent assassins, attackers, and near-lethal approachers. Journal of Forensic Sciences, 44, 321–333.

Gill, P., Horgan, J., & Deckert, P. (2014). Bombing alone: Tracing the motivations and antecedent behaviors of lone‐actor terrorists. Journal of Forensic Sciences59(2), 425–435. https://doi-org.proxy.lib.sfu.ca/10.1111/1556-4029.12312

Gill, P., Silver, J., Horgan, J., & Corner, E. (2017). Shooting alone: The pre-attack experiences and behaviors of US solo mass murderers. Journal of Forensic Sciences, 62(3), 710–714. https://doi.org/10.1111/1556-4029.13330

Hodgins, S., Hiscoke, U. L., & Freese, R. (2003). The Antecedents of Aggressive Behavior Among Men with Schizophrenia: A Prospective Investigation of Patients in Community Treatment. Behavioral Sciences & the Law21(4), 523–546. https://doi-org.proxy.lib.sfu.ca/10.1002/bsl.540

Junginger, J., & McGuire, L. (2004). Psychotic motivation and the paradox of current research on serious mental illness and rates of violence. Schizophrenia Bulletin30(1), 21–30. https://doi-org.proxy.lib.sfu.ca/10.1093/oxfordjournals.schbul.a007064

Kienlen, K. K., Birmingham, D. L., Solberg, K. B., O’Regan, J. T., & Meloy, J. R. (1997). A comparative study of psychotic and nonpsychotic stalking. Journal of the American Academy of Psychiatry and the Law25(3), 317–334.

Lawrence, M. S. (2017). From defect to dangerous: Has the door opened for recognition of an addiction-based defence in Canadian criminal law? Canadian Journal of Criminology and Criminal Justice59(4), 572–582. https://doi-org.proxy.lib.sfu.ca/10.3138/cjccj.2016.0006.R1

McNiel, D. E., Eisner, J. P., & Binder, R. L. (2000). The relationship between command hallucinations and violence. Psychiatric Services51(10), 1288–1292. https://doi-org.proxy.lib.sfu.ca/10.1176/appi.ps.51.10.1288

Meloy, J. R. (1988). Violent and Homicidal Behavior in Primitive Mental States. Journal of the American Academy of Psychoanalysis, 16(3), 381-394.

Meloy, J. R. (2018). The Operational Development and Empirical Testing of the Terrorist Radicalization Assessment Protocol (TRAP–18). Journal of Personality Assessment100(5), 483-492. https://doi-org.proxy.lib.sfu.ca/10.1080/00223891.2018.1481077

Meloy, J. R., Goodwill, A., Clemmow, C., & Gill, P. (2021). Time sequencing the TRAP-18 indicators. Journal of Threat Assessment and Management8(1–2), 1–19. https://doi-org.proxy.lib.sfu.ca/10.1037/tam0000157

Meloy, J. R., Goodwill, A. M., Meloy, M. J., Amat, G., Martinez, M., & Morgan, M. (2019). Some TRAP-18 indicators discriminate between terrorist attackers and other subjects of national security concern. Journal of Threat Assessment and Management6(2), 93–110. https://doi-org.proxy.lib.sfu.ca/10.1037/tam0000119

Meloy, J. R., Hempel, A. G., Gray, B. T., Mohandie, K., Shiva, A., & Richards, T. C. (2004). A comparative analysis of North American adolescent and adult mass murderers. Behavioral Sciences & the Law22(3), 291–309. https://doi.org/10.1002/bsl.586

Meloy, J. R., Hempel, A. G., Mohandie, K., Shiva, A. A., & Gray, B. T. (2001). Offender and offense characteristics of a nonrandom sample of adolescent mass murderers. Journal of the American Academy of Child & Adolescent Psychiatry40(6), 719–728. https://doi-org.proxy.lib.sfu.ca/10.1097/00004583-200106000-00018

Meloy, J. R., Hoffmann, J., Guldimann, A., & James, D. (2012). The role of warning behaviors in threat assessment: An exploration and suggested typology. Behavioral Sciences & the Law30(3), 256–279. https://doi-org.proxy.lib.sfu.ca/10.1002/bsl.999

Meloy, J. R., Hoffmann, J., Roshdi, K., & Guldimann, A. (2014). Some warning behaviors discriminate between school shooters and other students of concern. Journal of Threat Assessment and Management1(3), 203–211. https://doi-org.proxy.lib.sfu.ca/10.1037/tam0000020

Meloy, J. R., Mohandie, K., Knoll, J. L., & Hoffmann, J. (2015). The concept of identification in threat assessment. Behavioral Sciences & the Law33(2–3), 213–237. https://doi-org.proxy.lib.sfu.ca/10.1002/bsl.2166

Meloy, J. R., & O’Toole, M. E. (2011). The concept of leakage in threat assessment. Behavioral Sciences & the Law29(4), 513–527. https://doi-org.proxy.lib.sfu.ca/10.1002/bsl.986

Meloy, J. R., & Rahman, T. (2020). Cognitive‐affective drivers of fixation in threat assessment. Behavioral Sciences & the Law. https://doi-org.proxy.lib.sfu.ca/10.1002/bsl.2486

Meloy, J. R., White, S. G., & Hart, S. (2013). Workplace assessment of targeted violence risk: The development and reliability of the WAVR‐21. Journal of Forensic Sciences, 58(5), 1353–1358. https://doi-org.proxy.lib.sfu.ca/10.1111/1556-4029.12196

Metzl, J. M., & MacLeish, K. T. (2015). Mental illness, mass shootings, and the politics of American firearms. American Journal of Public Health105(2), 240–249. https://doi-org.proxy.lib.sfu.ca/10.2105/AJPH.2014.302242

Mitchell, M., & Palk, G. (2016). Traversing the space between threats and violence: A review of threat assessment guidelines. Psychiatry, Psychology and Law, 23(6), 863–871. https://doi-org.proxy.lib.sfu.ca/10.1080/13218719.2016.1164638

National Center for the Analysis of Violent Crime. (2019). Making Prevention a Reality: Identifying, Assessing, and Managing the Threat of Targeted Attacks. Federal Bureau of Investigation.

Newhill, C. E., Eack, S. M., & Mulvey, E. P. (2009). Violent behavior in borderline personality. Journal of Personality Disorders23(6), 541–554. https://doi-org.proxy.lib.sfu.ca/10.1521/pedi.2009.23.6.541

Nielssen, O. B., Westmore, B. D., Large, M. M., & Hayes, R. A. (2007). Homicide during psychotic illness in New South Wales between 1993 and 2002. Medical Journal of Australia, 186(6), 301–304. https://doi.org/10.5694/j.1326-5377.2007.tb00905.x

Patel, A., Barot, C. K., Vankar, G., & Pal, S. (2019). Acting on delusions in patients suffering from schizophrenia. Archives of Psychiatry and Psychotherapy, 21(4), 52–61. https://doi-org.proxy.lib.sfu.ca/10.12740/APP/109009

Pathé, M. T., Lowry, T. J., Haworth, D. J., Winterbourne, P., & Day, L. (2016). Public figure fixation: Cautionary findings for mental health practitioners. Behavioral Sciences & the Law34(5), 681–692. https://doi-org.proxy.lib.sfu.ca/10.1002/bsl.2252

Peterson, J., Skeem, J. L., Hart, E., Vidal, S., & Keith, F. (2010). Analyzing offense patterns as a function of mental illness to test the criminalization hypothesis. Psychiatric Services61(12), 1217–1222. https://doi-org.proxy.lib.sfu.ca/10.1176/appi.ps.61.12.1217

Peterson, J. K., Skeem, J., Kennealy, P., Bray, B., & Zvonkovic, A. (2014). How often and how consistently do symptoms directly precede criminal behavior among offenders with mental illness? Law and Human Behavior38(5), 439–449. https://doi-org.proxy.lib.sfu.ca/10.1037/lhb0000075

Rogier, G., Marzo, A., & Velotti, P. (2019). Aggression among offenders: The complex interplay by grandiose narcissism, spitefulness, and impulsivity. Criminal Justice and Behavior46(10), 1475–1492. https://doi-org.proxy.lib.sfu.ca/10.1177/0093854819862013

Rosenthal, R. (1991). Meta-analytic procedures for social research (rev. ed.). Newbury Park: Sage.

Sariaslan, A., Arseneault, L., Larsson, H., Lichtenstein, P., & Fazel, S. (2020). Risk of subjection to violence and perpetration of violence in persons with psychiatric disorders in Sweden. JAMA Psychiatry77(4), 359–367. https://doi.org/10.1001/jamapsychiatry.2019.4275

Scalora, M. J., Baumgartner, J. V., & Plank, G. L. (2003). The relationship of mental illness to targeted contact behavior toward state government agencies and officials. Behavioral Sciences & the Law21(2), 239–249. https://doi-org.proxy.lib.sfu.ca/10.1002/bsl.525

Schouten, R., & Brennan, D. V. (2016). Targeted violence against law enforcement officers. Behavioral Sciences & the Law34(5), 608–621. https://doi-org.proxy.lib.sfu.ca/10.1002/bsl.2256

Silver, J., Horgan, J., & Gill, P. (2018). Foreshadowing targeted violence: Assessing leakage of intent by public mass murderers. Aggression and Violent Behavior38, 94–100. https://doi-org.proxy.lib.sfu.ca/10.1016/j.avb.2017.12.002

Ullrich, S., Keers, R., & Coid, J. W. (2014). Delusions, anger, and serious violence: New findings from the MacArthur Violence Risk Assessment Study. Schizophrenia Bulletin40(5), 1174–1181. https://doi-org.proxy.lib.sfu.ca/10.1093/schbul/sbt126

Verdejo-Garcia, A., & Albein-Urios, N. (2021). Impulsivity traits and neurocognitive mechanisms conferring vulnerability to substance use disorders. Neuropharmacology183. https://doi-org.proxy.lib.sfu.ca/10.1016/j.neuropharm.2020.108402

Vossekuil, B., Fein, R. A., Reddy, M., Borum, R., & Modzeleski, W. (2004). The final report and findings of the Safe School Initiative. Washington, DC: US Secret Service and Department of Education.

Warren, L. J., Mullen, P. E., Thomas, S. D. M., Ogloff, J. R. P., & Burgess, P. M. (2008). Threats to kill: A follow-up study. Psychological Medicine38(4), 599–605. https://doi-org.proxy.lib.sfu.ca/10.1017/S003329170700181X

Weltens, I., Bak, M., Verhagen, S., Vandenberk, E., Domen, P., van Amelsvoort, T., & kker, M. (2021). Aggression on the psychiatric ward: Prevalence and risk factors. A systematic review of the literature. PloS One16(10), e0258346–e0258346. https://doi.org/10.1371/journal.pone.0258346

Wessely, S., Buchanan, A., Reed, A., Cutting, J., Everitt, B., Garety, P., & Taylor, P. J. (1993). Acting on delusions: I Prevalence. The British Journal of Psychiatry163, 69–76. https://doi-org.proxy.lib.sfu.ca/10.1192/bjp.163.1.69

Whiting, D., Lichtenstein, P., & Fazel, S. (2021). Violence and mental disorders: A structured review of associations by individual diagnoses, risk factors, and risk assessment. The Lancet Psychiatry8(2), 150–161. https://doi-org.proxy.lib.sfu.ca/10.1016/S2215-0366(20)30262-5

Wong, Z., Öngür, D., Cohen, B., Ravichandran, C., Noam, G., & Murphy, B. (2013). Command hallucinations and clinical characteristics of suicidality in patients with psychotic spectrum disorders. Comprehensive Psychiatry54(6), 611–617. https://doi-org.proxy.lib.sfu.ca/10.1016/j.comppsych.2012.12.022

Tables

Table 1 : Frequency of Psychiatric Symptoms at Time of Offense, Prior Contact with Victims and Law Enforcement and Warning Behaviors

Variable

n

%

Psychiatric Symptoms at Time of Offense

Delusions

56

100.0

Persecutory/paranoid

40

71.4

Grandiose

21

37.5

Erotomanic

12

21.4

Other (e.g., somatic)

4

7.1

Hallucinations

21

37.5

Auditory

18

32.1

Visual

8

14.3

Suicidal ideation

4

7.1

Suicide attempt

1

1.8

Self-harm

0

0.0

Prior Contact

Any contact with victim prior to offense

38

67.9

Verbal contact with victim

27

48.2

In-person contact with victim

28

50.0

Stalking victim prior to offense

12

21.4

Any contact with police prior to offense

14

25.0

Police gave no-contact/restraining order

4

7.1

Police arrested participant

6

10.7

Warning Behaviors

At least one warning behavior prior to offense

52

92.9

Fixation

34

60.7

Pathway behavior

23

41.1

Directly communicated threat

20

35.7

Novel aggression

9

16.1

Leakage

6

10.7

Last resort

5

8.9

Energy burst

4

7.1

Identification

4

7.1

Note: Participants could be coded under multiple categories (e.g., if they experienced multiple types of delusions and/or hallucinations or engaged in more than one form of warning behaviors).

Table 2 : Comparisons Between Individuals Who Perpetrated Targeted and Non-Targeted Violence: Sociodemographics, Criminal History and Offense Characteristics

Targeted Violence

Non-Targeted Violence

Total

ϕ

effect size

Demographic

Category

n

%

n

%

n

χ2

Gender

Male

45

80.4

201

86.3

246

1.38

.07

Female

11

19.6

32

13.7

43

Ever in relationship

Yes

8

14.3

35

14.7

43

.06

.01

No

48

85.7

203

85.3

251

Education

Completed high school

31

55.4

112

51.1

143

1.21

.07

Did not complete high school

21

37.5

107

48.9

128

Employment

Employed

11

19.6

38

15.7

49

.56

.04

Unemployed

45

80.4

204

84.3

249

Autonomous

32

57.1

123

52.8

155

.14

.02

Not autonomous

24

42.9

110

47.2

134

Criminal History

Yes

27

48.2

103

43.3

130

.45

.04

No

29

51.8

135

56.7

164

Offense type

Bodily harm

24

42.9

98

41.5

122

8.54*

.22

Uttering threats/

criminal harassment

25

44.6

38

16.1

63

Gender of victim

Male

20

37.7

85

55.2

105

4.81*

.15

Female

33

62.3

69

44.8

102

Relationship to victim

Family member

20

35.7

57

24.2

77

3.17

.10

Not family member

34

60.7

179

75.8

213

Use of weapons

Yes

15

26.8

44

19.3

59

1.71

.08

No

41

73.2

184

80.7

225

Schizophrenia/

Psychotic Disorder

Yes

52

92.9

92

39.7

144

-

.42

No

4

7.1

140

60.3

144

Mood Disorder

Yes

14

25.0

55

28.6

69

.02

.01

No

42

75.0

137

71.4

179

Anxiety Disorder

Yes

0

0.0

8

4.4

8

-

.11

No

56

100.0

174

95.6

230

Personality Disorder

Yes

13

23.2

22

9.5

35

7.97*

.17

No

43

76.8

210

90.5

253

Substance Use Disorder

Yes

15

26.8

88

44.7

103

3.93*

.12

No

41

73.2

109

55.3

150

Delusions at time of offense

Yes

56

100.0

181

76.7

237

-

.22

No

0

0.0

55

23.3

55

Hallucinations at time of offense

Yes

21

37.5

98

41.5

119

.64

.05

No

35

62.5

138

58.5

173

Note. * = p < .05. Fisher’s exact tests were used for variables that do not display a test statistic. Some percentages do not add to 100% across categories due to missing data.

Table 3: Multivariate Logistic Regression of Variables Associated with Targeted Violence Perpetration

Independent Variables

Standardized Beta Coefficient

Wald x2

Odds Ratio

95% CI

Delusions at time of offense

.17*

5.70

11.96

1.56

91.80

Substance Use Disorder

-.09*

6.55

2.79

1.27

6.10

Personality Disorder

.03

.89

1.52

.64

3.65

Schizophrenia/Psychotic Disorder

.10*

6.21

4.04

1.35

12.14

Note: Model χ2 = 32.83, p > .05, Pseudo R2 = .19. * = p < .05.

Comments
0
comment
No comments here
Why not start the discussion?