Mass murder, defined as four or more victims killed excluding the offender, perpetrated by one or multiple offenders within a single event (Duwe, 2000, 2004, 2007; FBI, 2005; Fox & Levin, 1998, 2015), is an extremely rare form of violence, occupying less than 1% of all U.S. homicides (Fridel, 2021; Krouse & Richardson, 2015; Levin & Fox, 1996). Yet in recent years, public mass murder incidents have gained extensive attention within large-scale news reports and in scholarly literature (Croitoru et al., 2020; Duwe, 2007; Petee et al., 1997). Though coverage on public mass murder events is important, extensive coverage on select mass murders excludes discussions of the most common typology of mass killings.

A persistent finding in research suggests that over 50% of mass murder incidents lie within the familial unit (Bowers et al., 2010; Dietz, 1986; Duwe, 2000; 2004; 2007; Fox & Levin, 1998; Fridel, 2021; Levin & Fox, 1996; Petee et al., 1997; Taylor, 2018; Zeoli & Paruk, 2020). In one of the earliest attempts to create distinct typologies of mass murder incidents, Dietz (1986) introduced the term family annihilation within the psychiatric literature. These were cases where the entire nuclear family was killed by one or two other family members. The family annihilator, otherwise referred to here as family mass murderer, was identified as the male head of household who killed at least four family members within a short time and at one location (Fox & Levin, 1998; Levin & Fox, 1996; Ressler et al., 1988). Despite an increase in attention to public mass murder events, there is a surprising dearth of research examining the most common type of these events: family mass murders (FMM).

Research on FMM incidents generally appears as a byproduct within studies focusing on broader multiple-victim homicide trends (Bowers et al., 2010; Duwe, 2007; Fox & Levin, 1994; Karlsson et al., 2021; Liem & Reichelmann, 2014) and are limited to descriptive statistics (Fegadel & Heide, 2017; Liem & Reichelmann, 2014). Although the mass murder literature in general has grown in scope by beginning to examine differences among specific mass murder types (Duwe, 2016; Petee et al., 1997; Siegel et al., 2020), there remains a dearth of literature that examines possible heterogeneity among FMM incidents. As such, it is currently unknown if, and to what extent, offender, victim, and incident characteristics differ among FMM offender types.

To advance research on mass murder, the current study quantitatively analyzes FMM¹ event characteristics. Specifically, we searched for differences among offender types by examining if family mass murder case characteristics are significantly different across gender and victim-offender relationship. To meet this aim, data from USA Today’s Behind the Bloodsheddataset were utilized to gather information on 163 cases of family mass killing incidents from 2006 to 2017. Each case was then matched with news articles through Nexis Uni and Google to obtain a more detailed overview of the offenders, victims, and event characteristics. The results from the present study add to familial and mass murder literature by providing a quantitative analysis of family mass murder types beyond descriptive approaches.

Literature Review

Past Research on Mass Murder 

Most research on mass murder has limited the methodological approach to descriptive case summaries with small sample sizes due to the rarity of these type of events. As such, mass murder offender typologies that have been identified by early scholars (see Fox & Levin, 1994; Holmes & Holmes, 1994; Levin & Fox, 1985; 1996; Petee et al., 1997) have seldom been empirically assessed in research, with the exception of a small number of recent studies (Fridel, 2021; Taylor, 2018). In one multivariate analysis comparing family, public, and felony mass murder events using the USA Today database, defined by incidents including 4 or more victims, Fridel (2021) found a number of significant differences across offender, victim, and incident characteristics. Fridel’s (2021) analyses suggest that incidents involving a history of domestic violence, romantic and familial difficulty, a greater number of child victims, occurring in the South, and the offender dying by suicide are more likely to be associated with a family mass murder than a felony mass murder. Additionally, Fridel’s (2021) findings suggest that family mass murder victims are more likely to be younger in age and of the same race as their victims (children or family members) when compared to both public and felony mass murders. The most significant differentiating factor between family, public, and felony mass murders were the type of victims involved.

In a separate study incorporating a multivariate analysis to examine mass murder incidents, Taylor (2018) examined possible differences in victim, offender, and incident characteristics based on an offender’s motivation to commit mass violence. Similar to Fridel’s (2021) results, Taylor (2018) found that offenders motivated by a relationship issue were more likely to kill a family member, die by suicide following the event, and be older in age than other mass murder offender types. Furthermore, her findings identified that relationship issues were present in over 40% of all mass murder cases and among these cases, a firearm was most often the lethal weapon of choice.

With research showing that gun availability substantially increases lethality in domestic violence situations (Campbell et al., 2003), scholars have begun to examine the potential link between domestic violence perpetration, firearm access, and mass murder. In an analysis of 89 offenders of mass shootings from 2014 to 2017, Zeoli and Paruk (2020) found about 30% of offenders had a suspected domestic violence history and 61% of these offenders, or 18.3% of the total, had come in contact with the criminal justice system for domestic violence. Among these individuals, only six offenders were eventually convicted of domestic violence-related charges. However, despite a conviction and qualifying for firearm restrictions, these offenders were still able to obtain a lethal weapon and execute mass killings.

In subsequent research examining 128 mass shootings from 2014 to 2019, Geller and colleagues (2021) found that about 68% of mass shootings were domestic violence related or the offender had a known history of domestic violence, and that these types of mass shootings resulted in a greater total of fatal and non-fatal victims (582 total victims) than non-domestic violence related incidents (352 total victims; see Geller et al., 2021). From these findings, some scholars have suggested that implementing stricter gun restriction policies against individuals with a domestic violence restraining order or a conviction of domestic violence could be one way to potentially lessen the opportunity of committing acts of mass violence (Geller et al., 2021; Taylor, 2018; Zeoli & Paruk, 2020). Others have similarly noted that individuals with personalities more likely to accept aggression behaviors as an appropriate solution to stress are more likely to also have a history of convictions for violent offenses, including intimate partner abuse (Peterson & Densley, 2019; Silver et al., 2018) and lethal forms of violence (Huff-Corzine & Corzine, 2020; Madfis, 2014).

Family Mass Murder

Literature on FMM is often included within more generalized research on mass murder as well as within more specific research analyses oriented on defining categories of multiple victim family homicides (e.g., Bowers, et al., 2010; Delisi & Scherer, 2006; Duwe, 2007; Karlsson et al., 2021; Liem & Koenraadt, 2008; Liem & Reichelmann, 2014; Liem et al., 2013). An important distinction between mass murder and the different types of multiple victim family homicides is that the former includes a broad definition of victims not limited to a specific type and includes cases with four or more victims. The latter employs categorical methods to analyze distinct types of multiple family homicides such as familicide (e.g., the killing of a spouse and one or more children), parricide (e.g., the killing of one’s parents), and siblicide (e.g., the killing of one’s siblings), and may include cases with one or more victims. Current mass murder literature, however, limits thorough discussions of possible heterogeneity among family mass murder instances as studies most often only examine how FMM cases are different from other types of mass murder (Dietz, 1986; Duwe, 2007; Fridel, 2021; Taylor, 2018). As such, studies examining multiple victim family homicides are useful to gain a more comprehensive understanding of specific offender, victim, and case characteristics. Notably, not all multiple victim family homicides fit the commonly accepted definition of mass murder events (4+ victims). Though recent work has suggested including multiple victim family homicide cases with 2+ or 3+ victims may offer additional insight to further understand acts of lethal violence in the family unit (Huff-Corzine & Corzine, 2020).

Cases involving the killing of multiple family members are often perpetrated by the husband or male intimate partner (Liem et al., 2013). The earliest attempt to further understand this phenomenon was completed by Frazier (1975), who identified two distinguishing motivators leading to the event as “murder by proxy” and “suicide by proxy.” “Murder by proxy” includes offenders who are motivated by anger and revenge aimed at their intimate partner following a threat of or actual withdrawal or estrangement (Websdale, 2010). In this scenario, children may also be killed because the offender sees them as an extension of the intimate partner (Liem & Reichelmann, 2014; Sillito & Salari, 2011; Wilson et al, 1995). Although the children are not regarded as the primary object of aggression, they are seen to be equally responsible for the offender’s feelings of betrayal caused by the intimate partner (Dietz, 1986; Fridel, 2021; Fox & Levin, 2011, Liem & Koenraadt, 2008; Liem & Reichelmann, 2014). Websdale (2010) further described these violent men as “livid coercive” because these offenders act out on profound feelings of anger and shame.

“Suicide by proxy” describes an offender who commits the murderous act to “protect” his family from the fate that would ensue without his financial support (Frazier, 1975; Websdale, 2010). These cases often involve offenders who have recently lost a job, face continuous unemployment, or increasing amounts of debt (Ewing, 1997; Fox & Levin, 2011; Liem & Reichelmann, 2014; Scheinin et al., 2011). Offenders most commonly view themselves as the provider and central figure of the family. Thus, when their ability to provide for their family is threatened, they may become suicidal and kill their family unit because they are seen as extension of themselves, in order to save the family from a life without the offender (Liem et al., 2013; Websdale, 2010; Wilson et al., 1995). Websdale (2010) classified this type of familicide offender as “civil reputable” and one who frequently dies by suicide following the event (Liem et al., 2013). Both distinctive familicidal offenders share a motivation marked by a sense of loss of control (Ewing, 1997; Liem & Koenraadt, 2008; Wilson et al., 1995).

Less frequent forms of multiple victim family homicides include a combination of parricides, the killing of one or more parents, and siblicides, the killing of one or more siblings. Current literature on these rare phenomena is mainly limited to case studies and descriptive statistics to explain offender, victim, and event characteristics. Findings show that most parricides and siblicides are committed by male offenders (Heide, 2013; Heide & Frei, 2010; Liem & Reichelmann, 2014; Marleau et al., 2006; Peck & Heide, 2012). In a recent study analyzing multiple family homicides, Liem and Reichelmann (2014) identified an extended parricide typology that involved cases of offenders killing their parents and siblings. They found that, similar to previous literature, most offenders were male, white, and did not typically die by suicide after the event. Compared to the other typologies identified in their study, these cases were the least likely to be premeditated and most often the perpetrator’s problems were not primarily related to the victims. The source of aggression stemmed from outside circumstances, such as intimate partner problems, unemployment, or drug/alcohol problems for which parents may be seen as partially responsible. Siblings became fatal victims as they were seen as extensions of the parents or simply witnesses that needed to be removed (Liem & Reichelmann, 2014).

To summarize, what is currently known about FMM offenders is that these instances overwhelmingly involve male perpetrators (Dietz, 1986; Duwe, 2000; 2007; Fegadel & Heide, 2017; Fox & Levin, 2011; Fridel, 2021; Liem et al., 2013), currently or formerly in a romantic relationship with at least one of the victims (Fox & Levin, 2011; Fridel, 2021; Liem et al., 2013), where firearms are frequently the weapon of choice (Fridel, 2021; Taylor, 2018), and many mass murder offenders have a recorded history of domestic violence (Fox & Levin, 2011; Fridel, 2021; Zeoli & Paruk, 2020). Additionally, among mass murder offenders with a recorded history of domestic violence, they are often not successfully restricted from acquiring firearms (Zeoli & Paruk, 2020) and involve greater fatal victim counts compared to mass murder offenders without a history of domestic violence (Geller et al., 2021).

Current Study

Previous literature on mass murder has primarily focused on the most sensationalized typologies (i.e., public mass murder) and research on multiple family homicides is generally limited to a small sample of case studies aimed to further understand only distinct subsets of familicide. Although recent scholars have begun to apply quantitative approaches beyond descriptive statistics to examine significant difference between unique types of mass murder incidents (Fridel, 2021; Taylor, 2018; Zeoli & Paruk, 2020), no studies to date have exclusively assessed the most frequent typology of mass murders. Consequently, there is an important gap in the literature that has yet to closely examine FMM cases to identify possible heterogeneity of victim, offender, and incident characteristics across offender types. As such, using the USA Today dataset, the purpose of the current study is to begin addressing this gap in literature by examining 163 U.S. family mass murder cases from 2006-2017 to assess significant differences by offender gender and victim-offender relationship.

Methods

To examine family mass murder incidents, the present study utilizes the open-access mass murder incident data from USA Today’s Behind the Bloodshed database. Based on the availability of data presented within this data source, the time period for the current study was limited to cases that occurred from 2006 to 2017 (see Overberg et al., 2016). The USA Today dataset is an interactive report that has been a preferred source by some scholars of mass murder events in recent research (Fridel, 2021; Taylor, 2018) as it overcomes known data reporting limitations of multiple victim homicide cases within the FBI’s Supplementary Homicide Report (SHR) database (see Fox & Pierre, 1996; Overberg et al., 2016). Compiled by researchers, this dataset was created through a corroborated process involving verification among the FBI’s SHR, local police records, and news media reports. Mass murder cases in the USA Today dataset were defined as a mass killing of 4 or more victims, excluding the offender, in a single event. Events may have occurred throughout hours, days, or more as long as there was no definite cooling off period. For each case, USA Today used information on the social and familial distance between offender(s) and victims to separate cases into four distinct mass murder categories: family killings, public killings, felony killings, and other. For the present study, only cases documented as family killings were collected to confirm the validity of the information of each FMM incident.

After each case listed under family killings was collected from the USA Today report, similar to previous research (Liem & Reichelmann, 2014; Liem et al., 2013; Petee et al., 1997; Taylor, 2018), Nexis Uni was utilized to find more information on the case. Specifically, data were collected to document the age of the primary offender, accomplice(s) present, victim information, the location of the event, the weapon(s) used, whether the offender died by suicide, and the motivation or trigger for the mass murder event. Together with the location, a range of a week from the date detailed on the USA Today report was first used to find each case in the Nexis Uni database. If no articles were found or there were over 50 results from the search, the articles were further filtered by keywords, such as “family,” “murder,” “killings,” or the perpetrator’s last name if that information was available on the USA Today report. If, after this approach was complete, no article was located through Nexis Uni, Google was used to identify a news source that offered more details about the case (Liem & Reichelmann, 2014).

Inclusion Criteria

We define family mass murder as any event where four or more victims are murdered, not including the offender, in one or more locations in close proximity to each other, without a cooling off period, where a majority of the victims were related to each other and have an intimate or familial relationship with the offender. Cases were excluded if there was no familial relationship between the offender and the victims, even if all of the victims were related to one another.²

Of the 185 cases originally documented under family killings in the USA Today report, twelve cases were excluded from the study because they did not fit the definition of family mass murder (i.e., the killings occurred between cooling off periods, there were less than three victims, or there was no familial relationship between offender and a majority of the victims). An additional ten cases were removed because the offender was unknown. This left a total of 163 family mass murder cases with 163 primary offenders, 397 adult victims, and 361 victims under the age of 18.

Variables

Data were collected to obtain a comprehensive look at family mass murder incidents from 2006 to 2017 based on event characteristics, offender/victim characteristics, and motivations. The event characteristics of each case were documented by the year, the month, and the state where the family mass murder occurred. Following Fox and Levin’s (1998) structure of situational characteristics, cases were further broken down to the number of incidents that occurred in each Census region of the United States. The incident locations were grouped into three categories: residential setting, multiple locations, and public settings. Additionally, information on the weapon used for the killings was gathered and categorized by gun, other close contact weapon (e.g., knife, blunt force object), multiple methods (e.g., gun and fire, gun and knife), arson, and other/unknown if the method of killing did not fit into any of these categories or no information could be found.   

Offender and victim characteristics collected for the current study included offender type, the age of the offender, offender gender, the number of victims per incident, the number of minor victims per incident, whether the offender died by suicide, and whether the offender had an accomplice present. Of the 163 cases included in the study, offenders were grouped into four distinct types based on their relationship to the victims: current intimate spouse or partner, former intimate spouse or partner, direct family member, and distant family member. Cases were grouped under current intimate spouse or partner if at least one of the victims was currently married or in a relationship with the offender and the other victims were related to each other and/or to the offender in some way. The former intimate spouse or partner category included cases where at least one of the victims was once in a relationship with the offender but was estranged at the time of the mass murder incident. The remaining victims were either related to the offender or the former intimate partner. Offenders grouped under direct and distant family members included offenders who did not kill anyone they were or had been involved with in a romantic relationship with. Instead, FMM direct family member offenders only killed individuals within their immediate family unit, including parents, stepparents, siblings, and children, whereas FMM distant family member offenders killed individuals within their extended familial unit, including grandparents, aunts, and cousins.

As discussed above, there are several ways that prior research has categorized motives and triggers of offenders that commit multiple murders. The current study uses Taylor’s (2018) categories that build upon Petee et al.’s (1997) classifications of offender motivations. Although there were a handful of cases where no information could be found from our data sources on the offender’s motivation, most of the cases were categorized under one or more of the following: mental illness, financial problems, emotional triggers, and relationship issues. The mental illness of an offender was documented using explicit statements found in news sources made by law enforcement, family, or friends about specific mental health issues the offender had exhibited or was previously diagnosed with (Taylor, 2018). Financial problems were determined by information included in news sources that indicated that the offender was going through a financial hardship at the time of the event and was perceived to be one of the leading motivations for the killings. Emotional triggers were grouped by events that were motivated by a loss of a job or a documented dispute shortly before the mass murder incident. Relationship issues included cases where the offender experienced a recent loss of a relationship, had engaged in a fight with their current or former intimate spouse or partner, and/or had a history of familial disorder, or domestic violence accusations illustrated through statements by the offender or victim’s family.

Analytic Strategy

First, descriptive statistics were computed to examine the total number of family mass murders that occurred between 2006 to 2017 across all of our independent variables (Table 1). Next, to differentiate family mass murder cases across the offender’s gender and each of the four offender types, we conducted a series of bivariate analyses. Chi-square (χ²) tests were conducted for independent variables measured at the ordinal or nominal level and ANOVAs were conducted for independent variables measured at the interval/ratio level of measurement. Fisher’s exact tests were utilized when low expected counts were observed in bivariate chi-square analyses. Finally, to further examine if there are unique differences among family mass murder incidents across offender type, a multinomial logistic regression was performed. Considering that there were too few cases in the distant family member offender category, our dependent variable (FMM offender types) for the multivariate analysis was recoded to measure three distinct offender groups: current intimate partner, former intimate, and non-intimate, which combined the direct and distant family offender groups. Findings from the multinomial logistic regression will be used to determine relative risk ratios to understand which, and to what extent, victim, offender, and incident characteristics can predict the type of family mass murder offender over another. Unfortunately, due to the small sample size of female offenders, differences of family mass murder incidents across gender were not able to be analyzed. Given the exploratory nature of this study and the lack of statistical power as a result of the small sample size of FMM cases, coefficients with p ≤ 0.10 are reported as significant and discussed.