Adult and Youth Involvement in Gun-Related Crime: Implications for Gun Violence Prevention Interventions

Adult Versus Youthful Offenders of Gun Violence

Compiled statistics tend to show that victims and offenders of fatal and nonfatal shootings are disproportionately young Black males (Moore et al., 2013). While no single definition has been established to what constitutes a “young” or “youthful” victim or offender, those below the age of 25 have generally been considered “youthful” individuals by a number of criminological studies (Shulman, Steinberg, & Piquero, 2013). Other statistical bodies, such as the U.S. Census Bureau (2010), have likewise utilized age 24 as a distinct age cut point (Howden & Meyer, 2010). In criminology, this age distinction reflects the “age-crime curve” in which criminal and deviant behavior tends to manifest during the late teen years, peaking during young adulthood, and decreasing gradually thereafter (Loeber & Farrington, 2014).

An important issue researchers currently face is the changing demographics and distribution of young people in violent activity. During the crack epidemic of the late 1980s and 1990s, as overall homicide rates were on the decline, youth homicide and gun violence rates increased sharply (Blumstein, 1995). While the spike in gun violence during this time period was largely attributed to the growth of illicit drug markets, high rates of youth homicide continued to persist in many large urban centers (Blumstein, 1995; Braga, Kennedy, Waring, & Piehl, 2001). Understanding patterns of gun violence in this light has presented unique questions to policy makers and researchers, for instance, are adolescent victims simply younger versions of adult victims? An important step in answering this question is assessing the differences and similarities between these two groups.

In general, the literature suggests that patterns of youth gun violence are not significantly different from adult gun violence. Regardless of age, most gun violence tends to be concentrated in the same demographic categories, and within the same kinds of economically disadvantaged neighborhoods (Cheatwood & Block, 1990). Victims and offenders most often know each other prior to the incident and the reason for the offense generally relates to an argument or ongoing conflict between the two parties (Pizarro, Zgoba, & Jennings, 2011).

While age alone may not significantly discriminate between youth and adult offenders, there is a large body of research that suggests that individual victim characteristics and lifestyle habits may contribute to heightened levels of risk. For instance, involvement in gun crimes and violent offending has been found to increase the risk of subsequent victimization (Wells & Chermak, 2011). Victim characteristics may mirror those of offenders’ in terms of deviant activity, gang membership, and criminal history (Pizarro et al., 2011). Gang member involvement can increase the frequency of violence (Pizarro et al., 2011) and is a strong predictor of violent victimization (Spano, Freilich, & Bolland, 2008). Similarly, being a member of co-offending networks that experience gun violence significantly increases the risk of future gun-crime offending and victimization for all network members (Papachristos, Braga, Piza, & Grossman, 2015). Black youth involved in gangs and violent activity are at highest risk of violent victimization—what is often termed the victim–offender overlap (Pizarro et al., 2011). Few studies have definitively identified the frequency of youth victimization in dispute, family, or intimate-partner gun incidents; however, the research does tend to suggest a higher rate of youth victimization in gang-related incidents than other incident types (Pizarro et al., 2011); thus, involvement in gangs may provide one of the most significant differences between youth and adult victims, as many studies show that the average age of gang members is relatively young (Lasley, 1992).

Studies of violent offenders have identified some differences between younger and older individuals in terms of lifestyle, such as gang membership, gun carrying, and drug dealing. For instance, gang-related homicide is often confined to adolescence (Rosenfeld, White, & Esbensen, 2012), coinciding with the peak of gang membership between ages 11 to 16 (Howell & Egley, 2005). Drug dealing and gun carrying have later ages of onset, generally around 17 to 18 (Rosenfeld et al., 2012). While most juveniles tend to desist from criminal activity before age 18, those who continue to offend into early adulthood are more often involved in increasingly severe violent offending (Loeber, Farrington, & Petechuk, 2013). The development of violent offending trajectories tends to begin in early adulthood (Rosenfeld et al., 2012). Consequently, most homicides are committed between ages 19 to 24, while gang-related homicides occur primarily during adolescence (Loeber et al., 2013).

The literature tends to agree that gang membership and co-offending peak primarily in early adolescence, and that violent offending by very young offenders is sometimes, but not always, motivated by gang violence or the commission of other crimes (Cheatwood & Block, 1990). The most serious adult offenders have consistent patterns of offending reaching back to the early teenage years (Sampson & Laub, 2005). Miethe and Regoeczi (2004) found that perpetrators of juvenile and adult homicides were highly similar along a number of characteristics. The authors did note a higher proportion of juveniles committed homicides with co-offenders and more often against strangers. Similarly, Cheatwood and Block (1990) found juvenile and adult homicides highly comparable among a number of dimensions; however, juveniles were more likely to be involved in multiple offender incidents and concurrent felonies than adults. Consistent with other literature, the authors noted that the victim’s age and offender’s age tended to be relatively similar as well.

Beyond the existence of various trajectories for violent offending, research examining the patterns of offending among juvenile and young offenders has suggested they are more likely to commit crimes in the presence of other co-offenders (Hochstetler, Copes, & DeLisi, 2002; McGloin & Stickle, 2011). Early adolescence represents a time in human development where peer influences are especially important (McGloin & Stickle, 2011). Individuals are often introduced into delinquency and criminal activity through the influence of their friends, which can develop into offending groups (Hochstetler et al., 2002). Often, these co-offending groups are a loosely knit band of friends or acquaintances, and tend to be comprised of relatively few individuals.

Similar to the age-crime curve observed in many studies, offenders have been observed to “age out” and become less likely to commit crime in groups. For instance, research by Carrington (2009) found a negative relationship between the age of offenders and the prevalence of co-offending, suggesting that youthful offenders are more likely to co-offend. Other studies have also identified an aging-out effect of co-offending, showing a peak at age 18 and dropping off by the end of young adulthood (van Mastright & Farrington, 2009). The study of adult co-offending has generally been sparse, however. A small body of research has indicated that certain offenses, including robbery, burglary, and some property offenses were more likely to be committed in groups (Carrington, 2009; van Mastright & Farrington, 2009). Chronic offenders have shown versatility in their offending habits, often committing crimes alone or with others (Carrington, 2009).

Although research has suggested differences between the adult and youthful offender populations, which could have an effect on the design of targeted intervention tactics, the literature on violence prevention has largely been silent to these distinctions. Some interventions do not distinguish their targeted population based on age (i.e., Cure Violence), while others such as the Boston Operation Ceasefire model do by often focusing on youth. The following section presents a discussion of what is known about successful violence prevention interventions, and how it relates to the literature on youth and adult offending.

Violence Prevention and Targeted Intervention

Criminological research has consistently found that a small number of individuals and places are responsible for a disproportionately large amount of crime (Sherman, 2007). Much of this research suggests that targeted enforcement on specific individuals and places has a significant impact on overall crime rates, while minimizing the impact on community residents (Braga, Papachristos, & Hureau, 2012). For instance, Sherman and colleagues (1995) found police raids on crack houses resulted in a block-level decrease in crime. Similarly, McGarrell et al. (2001; see also Cohen & Ludwig, 2003; Sherman & Rogan, 1995) found that directed police patrols, focusing on gun carrying in specific regions of the city, resulted in an overall decrease in violent firearm crime. Place-based strategies, such as “hotspot” policing, have been shown to be effective in reducing gun violence in high crime areas (Rosenfeld et al., 2012). Recent research suggests that combining people- and place-based strategies, such as focusing on prolific offenders in hotspot locations, holds significant promise for violence reduction (Groff et al., 2015; Uchida & Swatt, 2013). Similarly, targeting specific problems at places (in line with problem-oriented policing) is found to be a more effective way of reducing crime than simply increasing traditional patrol (Braga et al., 2012).

The growing emphasis on targeted intervention strategies have had an effect on the development of current violence intervention programs—shifting the paradigm to highly focused, data-driven initiatives. The “pulling-levers” strategy implemented as part of the 1995 Boston Ceasefire is among the most notable examples. As part of an initiative to reduce record-high youth gun homicides, the Boston Ceasefire utilized a multi-agency working group that included members of federal, state, and local law enforcement along with academic researchers. Identifying gang-involved youth as the primary target, the program-focused intervention resources through the use of deterrent messages delivered in the form of “call-in” meetings (Kennedy et al., 2001). During these meetings, law enforcement officials notified youth they were being targeted due to their involvement in gun crime, and that violent activity would not be tolerated, while offering support to those willing to desist from violence.

The successful implementation of the Boston Ceasefire, which was associated with a 63% reduction in youth homicides, led to subsequent replications in a number of cities. Similar Ceasefire interventions in Stockton, Indianapolis, and Chicago were generally successful in reducing homicides and shootings (McGarrell, Chermak, Wilson, & Corsaro, 2006; Papachristos, Meares, & Fagan, 2012). Consistent findings later emerged in Cincinnati (Engel, Tillyer, & Corsaro, 2013), and a meta-analysis of Ceasefire programs in 10 cities showed statistically significant decreases in homicides in nine of the locations (Braga & Weisburd, 2012).

Positive findings from the Boston Ceasefire led to a number of similar targeted intervention programs. Utilizing a similar paradigm to the Boston Ceasefire, the Strategic Approaches to Community Safety Initiative (SACSI) was attributed to significant decreases in gang homicides and gun violence in several cities (McGarrell et al., 2006; Tita, Riley, Ridgeway, Grammich, & Abrahamse, 2011) and in a comparison of SACSI cities with other U.S. cities (Roehl et al., 2008). Project Safe Neighborhoods (PSN) utilized similar focused deterrence strategies to address gun crime in a number of cities with observed declines in violence in PSN target cities (McGarrell, Corsaro, Hipple, & Bynum, 2010). Like the Boston Ceasefire and SACSI, PSN leveraged partnerships between local police departments, prosecutors, and federal agencies to focus on high-impact gun offenders (MacDonald, Wilson, & Tita, 2005). Other targeted intervention programs during this time, such as Project Exile, prosecuted gun carrying felons in federal court, giving them long mandatory sentences with no bail or early release (Rosenfeld et al., 2005). The Cure Violence program in Chicago built upon one component of the original Boston Ceasefire model, using street-level “violence interrupters” to mediate disputes between feuding parties, and to assist law enforcement in reducing gun-related violence (Skogan et al., 2009).

Thus, a considerable body of research has illustrated the effectiveness of focusing enforcement resources on those most involved in gun-related violence. In particular, the focused deterrence model has shown to be an effective method of addressing gun crime (Braga & Weisburd, 2012); yet, there has been little assessment of the relative impact on youthful and adult offending. The Boston Ceasefire evaluation focused specifically on youth homicides and youth gang homicides (Braga, Hureau, & Papachristos, 2014), but most Ceasefire and related focused deterrence studies have examined trends in violence generally without distinguishing between youth and adult offending (e.g., Corsaro & McGarrell, 2010; McGarrell et al., 2006; Rosenfeld et al., 2005). Others have focused on the impact on gang or group member violence but without distinguishing the impact on youth versus adult violence (Corsaro & McGarrell, 2009; Engel et al., 2013), and focused deterrence research on parolees have included adults but with age distributions crossing the youth–adult classification (Braga et al., 2014). Indeed, whereas this line of research suggests that focused deterrence and targeted enforcement hold promise for reducing violence, little attention has considered variation in efficacy for youth versus adult gun offending and victimization.

Similarly, there remain some unknowns regarding desistence from violent behavior. A meta-analysis of 17 longitudinal studies on youth crime desistence suggests that the core, underlying processes are not yet fully understood or conceptualized by researchers (Bastro-Pereira, Começanha, Ribeiro, & Maia, 2015). This limited understanding of desistance, as well as persistence and escalation, is particularly the case for serious violent crime such as gun-related violence. Consequently, understanding the group of interest (in this case, gun offenders) is increasingly important to ensure that intervention strategies match the characteristics of the at-risk offender population. Relatedly, it is important to understand whether these types of interventions are better suited for youth, adults, or both population groups as this type of knowledge would enable the development of more focused, efficient, and effective violence prevention strategies.

Theoretical Framework

There are several foundations for these gun violence reduction strategies. These include focused deterrence and situational crime prevention (Braga, Apel & Welsh, 2012; Clarke, 1995; Kennedy, 2009). These perspectives assume that offenders are rational actors who consider risks and rewards. By focusing on specific behaviors (e.g., illegal gun carrying) and high-risk locations (hotspots), strategies such as directed police patrol seek to increase the risks for illegal gun carrying and use. Strategies that focus on prolific offenders in hotspot locations similarly seek to increase the risks and reduce the awards for those believed at highest risk of being involved in gun crime. That is, consistent with deterrence and situational crime prevention theory, directed police deployment can increase the certainty of arrest (increased risk). Prosecution strategies focused on illegal gun possession and use, as in PSN and Exile, can increase the certainty and severity of punishment. Such strategies may also benefit from incapacitation effects, at least if utilized with prolific offenders at peak offending stages of the life course. The focus on prolific offenders is also suggested from deterrence research that experienced offenders have significantly lower perceptions of the risk of arrest and sanction (Loughran, Piquero, Fagan, & Mulvey, 2012).

Problem solving at hotspot locations seek to disrupt the connections between motivated offenders, vulnerable victims, in high-risk contexts, utilizing a variety of strategies consistent with the situational crime prevention framework and the rational choice, routine activities, and crime pattern theories upon which it is based (Eck & Guerette, 2012). Drawing upon these theories broadens the strategies beyond the offender-focused shifting of perceived risks, to strategies that seek to change the conditions at crime-generating and crime-attracting locations. Along with steps to increase the risk, situational crime prevention suggests steps to increase the effort, reduce rewards, reduce provocations, and remove excuses associated with decisions on whether to engage in crime (Eck & Guerette, 2012).

The theoretical basis for the Ceasefire focused deterrence strategy also includes deterrence but expands the theoretical foundation to include procedural justice and social support (Kennedy, 2009). Specifically, focused deterrence strategies are based on the assumption of deterring potential offenders by directly communicating them the consequence (i.e., punishment) of their actions. In essence, what these strategies attempt to do is shift the perceived certainty and severity of sanctions and dissuade would be offenders from engaging in crime. For example, the Boston Ceasefire Project used a “pulling-levers” component, which consisted of a threat to use every method available should the gang or group violence continue (often through the use of home checks for probationers, warrant service, misdemeanor citations, and arrests for minor infractions). Sources of support from social service agencies were combined with this enforcement message to support alternative, legitimate choices, and support re-entry into the community as law-abiding members. Individuals who did not adhere to the deterrence message and were apprehended by law enforcement were then made an example, in an effort to exemplify to other offenders the consequence of continued criminal involvement in gun violence.

Whereas focused deterrence can serve as a blanket theory for offender-based strategies, the mechanisms of how they might work and their crime prevention efficacy may differ based on age. Indeed, the Life Course perspective in criminology suggests that offender-focused strategies are likely to have varying influence based on age. Life Course criminology focuses on offending continuity and change during an individual’s life (Sullivan, Piquero, & Cullen, 2012). This theoretical perspective suggests that the “pushes” toward crime may differ by age (Sampson & Laub, 2005). For example, while the social capital incurred by friendship with peers may be an important source of delinquency during adolescence, the formation of bonds with family members are more important and often represent a transition stage during adulthood (Sampson & Laub, 2005).

Changes in one’s life circumstances and social groups could also be considered an indication of changes in one’s lifestyle or daily routines. Lifestyle theory as originated by Hindelang, Gottfredson, and Garofalo (1978) focused on the explanation of why some demographic groups (i.e., males, youth) are more at risk of victimization than others; however, it has also been used to explain offending patterns (Pizarro et al., 2011). The crux of the theory rests on the assumption that role expectations are dependent on the individual’s demographic group, and that such expectations are conducive to lifestyle choices that may contribute to future victimization and offending, such as going out at night and spending time away from the residence. Related to age, younger individuals whose social capital is tied around peers and friend networks may engage in leisure activities that may put them at risk such as frequenting geographic areas that have more crime attractors and generators such as alcohol establishments, or may be more likely to go out at night. On the other hand, older individuals who are more bonded with family (due to marriage or child rearing) might be less likely to frequent these areas, or to be out at night during high-risk hours.

Given that the types of relationship and social capital groups change throughout an individual life course, and that these relationships influence an individual’s lifestyle, it is plausible to hypothesize that patterns of offending may differ by age. Indeed, if younger individuals are more likely to frequent high-risk areas and have stronger bonds with peers who share similar characteristics, one can expect a difference in the activities they undertake. Relatedly, if the patterns and covariates of offending differ, then the types of “levers” practitioners can rely on in focused deterrence tactics may also be different. For instance, youthful offenders may be better reached with social strategies that focus on peer groups, while adults may be better reached via strategies that focus on strengthening families and employment opportunities.

Life course theory also distinguishes between life course persistent and adolescent limited offending (Moffitt, 1993). Most youth are adolescent limited, meaning that while they may engage in crime and delinquency during adolescence, they eventually mature, establish new prosocial bonds, and stop offending. Adolescent limited offending is thus consistent with the long observed age-crime curve. Life course persistent offending, in contrast, comprises individuals who engage in antisocial and criminal behavioral throughout the life course. The persistence of these behaviors has a cumulative effect whereby dropping out of school, developing a criminal history, and incarceration limit legitimate opportunities and thus contribute to the persistence in offending. When considering strategies aimed at gun violence, typically committed by a small group of chronic offenders, it seems likely that the target audience comprises a larger proportion of life-course-persistent offenders but also that strategies may have a differential impact on youths (more likely a mix of adolescent limited and life course persistent) and adults (likely comprising more life course persistent offenders). Consequently, understanding the age and related characteristics of gun offenders may shape selection of gun violence prevention strategies.

The Present Study

Two research questions are explored in this study. The first addresses whether significant differences exist in the temporal, situational, and spatial characteristics of adult and youthful gun-crime offenders. The importance of this question lies on building a knowledge base on the similarities and differences among these two groups. If the groups are similar, then comprehensive strategies that target both groups simultaneously may yield positive results. If the groups are dissimilar, however, then an appropriate next step may be to develop specific prevention strategies for each group. Consequently, the second research question explored in this study dovetails from the first in that it examines what are the implications of the research findings for gun violence intervention programs. Based on the research reviewed and the theoretical framework, it is hypothesized that while similarities exist between these two groups, important differences exist, which could inform the design and implementation of violence reduction strategies.

In answering these research questions, the goal of the present study is to assess whether violence intervention type should be based on offender/victim age. While research has shown the effectiveness of targeted intervention violence prevention strategies, it is still unclear whether some interventions would be best suited for youthful offenders versus adults, or whether blanket interventions that focus on offense type regardless of age would be equally effective. The questions explored in this study are important at the practical level not only for practitioners and criminologists who focus on the area of crime prevention and desistance but also to criminologists who study the life course. As mentioned in the outset, knowledge on whether existing interventions are valid for adults as well as youth have the potential save time and money, and ultimately lives, as it will enable researchers to design and invest in efficacious prevention strategies. Indeed, cities suffering from high rates of gun-related crimes and budget cuts to public safety institutions can benefit from this knowledge as it would allow them to engage in data-driven and informed crime prevention spending. In addition, exploring these questions can shed light into the propositions of Life Course Criminology, and further elucidate patterns of offending based on age.

Method

Variables

Offenders of homicide and nonfatal shootings are the unit of analysis. Incidents in which an offender could be identified were analyzed for this study, with approximately 62% of homicide cases and 38% of nonfatal shooting cases having identified offenders. ¹ In total, 188 offenders were examined for this study. Offenders were disaggregated into youthful offenders (aged 24 years or younger), which were coded as 1, and adult offenders (aged 25 years and older), coded as 0. One hundred five (56%) were youthful offenders and 83 (44%) were adult. Consistent with the Detroit population, the offender group was highly homogeneous, with 98% Black and 88% male. The mean age for youthful offenders was 19.8 years and 34.7 years for adults. Females were roughly equally distributed among both groups, comprising 11% of youthful and 13% of adult offenders.

Three sets of variables were analyzed—temporal, situational, and spatial. Temporal variables were calculated from the date and time reported on the police report that the incident occurred. Time of day was coded into day and evening incidents. Evening incidents occurred between the hours of 6 p.m. and 5 a.m. and were coded as 1. Day incidents occurred between the hours of 5 a.m. and 6 p.m. and were coded as 0. Day of the week was coded as weekday (0) if the incident occurred Monday through Friday or weekend if the incident occurred on a Saturday or Sunday (1). Corresponding with cooler and warmer times of the year, month was coded as 0 if the incident occurred between October and March or 1 if it occurred between April and September.

Narrative data from the police report was coded to generate five situational variables: motive, the relationship between victim and offender, the number of offenders, the mean age of victims, and whether the victim(s) were cooperative in the follow-up investigation (in the case of nonfatal shootings). Motive refers to the primary reason the offender acted against the victim. Three motive categories were coded: argument/conflict, drug/gang, and robbery. Argument/conflict referred to incidents in which the shooting was precipitated by an argument or fight, or due to an ongoing feud, and was coded as 0. Drug/gang referred to incidents in which the primary motive behind the shooting was related to either the sales or purchasing of drugs, or in the furtherance of the goals of a known street gang (i.e., fighting over territory, assassinations), and was coded as 1. Finally, robbery included incidents that occurred during or after the carrying out of an armed robbery, and was coded as 2. Argument\conflict was used as the reference category. Relationship was coded as 0 if the incident involved strangers or 1 if the incident involved parties that knew each other. Offenders known to the victim prior to the incident included friends, intimate partners, family members, neighbors, street acquaintances, or known rivals. Co-offending captured whether the incident involved only one offender (coded as 0) or multiple offenders (coded as 1). Victim cooperation was coded as (0) if the victim cooperated with the police and (coded as 1) if the victim was not cooperative (i.e., declining to prosecute or answer questions).

Spatial variables captured incident location, distance from crime facilitators and attractors, social disorganization, and economic deprivation. For incident location, the variable was coded as 1 if the crime occurred within or directly in front of a residence, or 0 if it occurred on a public street, sidewalk, or adjacent to a business. Distance from crime facilitators captured feet from the incident location to the nearest liquor establishment, gas station, and convenience store. These variables were generated using the address of the incident and geocoding the location, of which all incidents were successfully matched to addresses. The geo-coordinates of crime facilitators were obtained through InfoGroup. Using ArcGIS, the near-proximity function was utilized to measure the Euclidian distance (in feet) between the location of the shooting incident and the nearest gas station, liquor store, and convenience or grocery store.

Social disorganization and economic deprivation measures were obtained from the 2013 American Community Survey. A principal components analysis was carried out on a set of nine primary variables (percent of individuals unemployed, percent of individuals in poverty, percent of individuals with no high school degree, of individuals who moved in the past 3 years, median household income, percent of household receiving government assistance, percent owner occupied households, percent vacant properties, and median house age) to generate orthogonal measurements of poverty and disorder in the community. Two components were extracted with eigenvalues greater than 1 and subjected to a varimax rotation. Component 1, with high loadings (>.50) on percent unemployed, percent in poverty, percent on government assistance, median income, and percent owning a home, was conceptualized as an economic deprivation measure. Component 2, which loaded highly on percent of properties vacant and median house age, was conceptualized as neighborhood disorder measure. Percent with no high school degree and percent moved in the past 5 years did not load highly on either component. Together, these two components explained approximately 52% of the total variance (see Table 1). The data were then merged to incidents that fell inside the census block group. ²

OPEN IN VIEWER

Table 1.

Principal Components Analysis American Community Survey—2013, 5-Year Estimates.

Variable	Component 1	Component 2
% Unemployed	0.55	0.30
% In poverty	0.85	0.06
% On government assistance	0.72	0.31
Median income	−0.86	−0.07
% No high school degree	0.43	0.24
% Vacant properties	0.25	0.74
% Moved in past 3 years	−0.24	0.07
% Own house	−0.78	0.24
Median house age	−0.19	0.74
Proportion of total variance	36%	16%

Findings

Descriptive Results

Table 2 presents descriptive statistics of the data and the chi-square test results. Time and day of the incident did not vary significantly between the two groups. All offenders were most often active during the evening hours, and during the week. While not statistically significant, however, the data suggest that adult offenders were more active during weekday evenings while youthful offenders offending patterns were evenly distributed during the day and evening hours. Incidents were roughly evenly split between warmer and cooler months. Some differences were observed in the analysis of situational variables. Arguments and conflicts were primarily the motivating factor for both youth and adult offenders; however, a slightly higher proportion of youth were involved in drugs, gangs, or robbery-related shootings. In this case, the difference was not significant at the .05 level (p = .10; V = .156), and the measure of effect size indicated a relatively weak association. Similarly, a statistically significant, but small difference (p = .04; V = .132), was observed between the victim–offender relationship, with more youth involved in incidents against strangers than adults.

OPEN IN VIEWER

Table 2.

Adult and Youth Offenders.

		Adult (25+)	Youth (>25)	p *	Cramer’s V
Temporal
Hour of day	Morning hours	32 (39%)	50 (48%)	n.s.	.079
	Evening hours	51 (61%)	55 (52%)
Day of week	Weekday	46 (55%)	59 (56%)	n.s.	.001
	Weekend	37 (45%)	46 (44%)
Month	October-March	42 (51%)	49 (47%)	n.s.	.028
	April-September	41 (49%)	56 (53%)
Situational
Motive	Argument/conflict	52 (63%)	50 (48%)	.10	.156
	Drug/gang	19 (23%)	30 (29%)
	Robbery	12 (14%)	25 (24%)
Relationship	Stranger	30 (36%)	53 (50%)	.04*	.132
	Acquaintance	53 (64%)	52 (50%)
Mean age of victim(s)	Youth victim	31 (37%)	59 (56%)	.008**	.181
	Adult victim	52 (63%)	45 (43%)
Co-offending	Single offender	52 (63%)	39 (37%)	.000**	.243
	Multiple offenders	31 (37%)	66 (63%)
Victim cooperation	Cooperative	65 (78%)	77 (73%)	n.s.	.045
	Uncooperative	18 (22%)	28 (27%)
Spatial
Location type	Residence	43 (52%)	43 (41%)	n.s.	.097
	Street\business	40 (48%)	62 (59%)
					Cohen’s d
Economic deprivation	Component mean	0.140	0.190	n.s.	.051
Neighborhood disorder	Component mean	−0.122	−0.148	n.s.	.030
Distance to liquor establishment	Mean distance (feet)	1,599.8	1,729.6	n.s.	.131
Distance to gas station	Mean distance (feet)	2,304.4	2,350.8	n.s.	.040
Distance to convenience store	Mean distance (feet)	1,599.9	1,628.8	n.s.	.031

+

p<.10 *p < .05 ** p < .01.

Differences were observed in the mean age of victims between youthful and adult offenders (p = .008; V = .181). While proportionally more youthful offenders were involved in incidents against individuals older than themselves, both groups tended to offend against those roughly within their same age group. More than half (56%) of youthful offenders offended against individuals under the age of 25, while 63% of adult offenders offended against individuals aged 25 or older. Co-offending among youth and adult offenders showed significant differences (p = .000; V = .243), with the largest effect size among the tested variables. More than half of youthful offenders committed their offense with another individual (63%), compared with 37% of adult offenders.

No significant differences were observed between both groups in regard to cases involving uncooperative victims, or spatial distribution. Both groups tended to offend in roughly similar areas of economic deprivation and neighborhood disorder, and at similar distances from potential crime attractors and crime generators.

Multivariate Results

Results from the logistic regression mirrored the chi-square tests (see Table 3). None of the independent variables in Models 1 or 3, which represented temporal and spatial differences, reached statistical significance. In Model 2, which represented situational differences, both the youth victim and co-offending variables were significant. These two variables remained significant at the .05 level in the full model. Odds ratios indicated that homicides and nonfatal shootings with a victim below the age of 25 were roughly 2.8 times (95% confidence intervals [CI] = [1.45, 5.68]) more likely to also have an offender below the age of 25. Shooting incidents with more than one offender were roughly 2.4 times (95% CI = [1.23, 4.71]) more likely to have an offender below the age of 25.

OPEN IN VIEWER

Table 3.

Binary Logistic Regression Predicting Youth Offender.

	Model 1			Model 2			Model 3			Model 4
	b	SE	exp(b)	b	SE	Exp(b)	b	SE	Exp(b)	b	SE	exp(b)
Temporal
Evening hours	−0.369	0.299	0.692							−0.190	0.333	0.826
Weekend	−0.045	0.298	0.532							−0.511	0.353	0.560
Warm months	0.142	0.295	1.153							−0.095	0.338	0.909
Situational
Drug or gang motivated				0.396	0.377	1.485				0.364	0.396	1.439
Robbery motivated				0.511	0.467	1.829				0.757	0.519	2.134
Victim acquaintance				−0.391	0.346	0.729				−0.402	0.360	0.668
Youth victim				0.974**	0.325	2.578				1.041**	0.346	2.834
Multiple offenders				0.896**	0.322	2.457				0.780**	0.342	2.389
Uncooperative victim				0.224	0.378	1.285				0.226	0.393	1.255
Spatial
Occurred at residence							−0.428	0.309	0.652	−0.094	0.358	0.910
Economic deprivation							0.022	0.155	1.023	−0.004	0.174	0.996
Neighborhood disorder							−0.059	0.186	0.942	−0.198	0.205	0.820
Feet to bar/liquor store							0.009	0.002	1.000	0.001	0.001	1.000
Feet to gas station							0.002	0.001	1.000	.0001	0.001	0.999
Feet to convenience							0.002	0.002	1.000	.0001	0.002	0.999
Intercept	0.392	0.314	—	−0.659	0.411	—	0.186	0.432	—	−0.178	0.706	—

**

p < .01.

Discussion and Conclusion

In an effort to aid in the development of effective and efficient violence prevention strategies, this study explored two research questions: (a) Do significant differences exist in the temporal, situational, and spatial characteristics of adult and youthful offenders in gun-related incidents? and (b) What are the implications for gun violence intervention programs? It is important to note that the present study does not directly examine the second research question: the efficacy of gun-crime prevention strategies for youth and adult gun offenders. Rather, the findings on the patterns of youth and adult gun offending allows us to speculate on the likely relevance of evidence-based and evidence-informed strategies for addressing youth gun violence, adult gun violence, and gun violence generally. ³

The findings suggest that youth and adult offenders of gun crimes in Detroit are not significantly different in the time or place in which they offend. Youthful and adult offender incidents involved victims and offenders who knew each other, and who were engaged in arguments or ongoing conflicts at the time of the crime. Minor differences were observed in the chi-square tests in terms of motive; however, these disappeared in the multivariate models. The most significant differences between youth and adult offenders involved the age of their victim and the presence of co-offenders, which is supported by the propositions of life course criminology (Sullivan et al., 2012). The bivariate and multivariate models indicated that incidents with co-offenders and with victims under the age of 25 were significantly more likely to have been victimized by youthful offenders. ⁴

Despite the minimal differences observed between adult and youth gun offenders, these findings should not be interpreted as contradictory to the tenets of life course criminology. While similarities exist related to the situational characteristics of offending between youthful and adult offending, the pushes and pulls toward onset and frequency of criminality were not explored in this study. Therefore, the extent to which social capital and bonds with peers and family affected the onset and frequency of offending between these groups is unknown. It is plausible that the similarities in offending patterns between the two groups may be more due to shared social structural and opportunity structures, but that the pushes and pulls related the onset and frequency of offending are different. Future research should examine this question with similar offending groups.

Although only modest differences were observed, this study has important policy implications. The finding that youthful offenders are more likely to co-offend suggests that disrupting gangs or street groups, and influencing the behavior of these street groups, may be a fruitful method of reducing youth violent crime. This supports gun violence reduction interventions such as Ceasefire that combine a focused deterrence message with offers of social support delivered to people involved in high-risk social networks (Kennedy et al., 2001). To the extent that older offenders are involved in such co-offending networks, such strategies may also affect gun crime committed by older offenders. The results suggest, however, that a significant portion of the violence committed by older offenders may not be affected by interventions that focus on co-offending groups. Adult gun violence offenders are much more likely to offend alone and to victimize an acquaintance. Consequently, so-called prolific offender strategies such as chronic violent offender programs (Great Britain Home Office, 2010), Project Exile (Rosenfeld et al., 2005), and parolee forums (Braga et al., 2009; Papachristos et al., 2015) may be called for with adult offenders. At the same time, these strategies are also likely to affect younger offenders who are carrying firearms and who have already accumulated significant criminal histories. In addition, these strategies are likely to complement the youth-focused Ceasefire strategy by reinforcing the focused deterrence message.

These findings also suggest that youthful and adult gun offenders are involved in drug selling and robbery incidents at relatively similar rates. Thus, prevention strategies aimed at these types of gun crimes may have an impact on both youth and adult gun offending. With respect to drug-related gun violence, prior research suggests the potential efficacy of the Drug Market Intervention (DMI; Corsaro, Brunson, & McGarrell, 2013; Corsaro, Hunt, Hipple, & McGarrell, 2013). Built on Ceasefire principles, the DMI focuses on discrete drug markets and uses street-level intelligence and crime data to identify all those involved in drug sales as part of the particular market. Traditional undercover operations are utilized to build cases on as many of the drug sellers as possible. At that point, however, the DMI strategy distinguishes between individuals with histories of violence (who are subject to prosecution) and nonviolent individuals who are informally diverted from prosecution but with the promise that the charges will be filed if the drug market continues to operate. Parallel components of the strategy involve efforts to link diverted drug dealers to services and to build neighborhood collective efficacy in an effort to reclaim public space and prevent the reemergence of the drug market. Given that drug market operations often involve both adults and juveniles, this may be an effective strategy aimed at both age categories.

Strategies aimed at robbery may similarly affect both youth and adult robbery-related violence. Problem solving at repeat robbery locations and directed police patrols are potential strategies (but see Rosenfeld et al., 2014). Similarly, dispute interruption and resolution strategies such as Cure Violence appear to be just as relevant for adult as youthful offenders (Skogan et al., 2009). In the context of domestic and intimate-partner violence, where adult offenders are more often the perpetrators of gun crime, offender-focused interventions, lethality assessments, and strategies aimed at protecting and supporting victims may be appropriate for both youth and adult offenders but with greater relevance for adult offenders (Maxwell & Garner, 2011; Messing et al., 2015).

Finding that most incidents occurred in roughly similar areas suggests that identifying relevant crime attractors and crime generators may present an intermediate method of addressing both adult and youth gun violence; hence, focused interventions that target specific crime hotspots and locations might be a fruitful avenue for prevention. The Los Angeles Police Department LASER Program may be another example of a highly focused strategy that could be expected to address both youth and adult gun violence. LASER utilizes ongoing analyses of gun-crime patterns to identify micro-places (e.g., a street block or small group of street blocks) as well as intelligence on chronic, violent offenders associated with these places. Initial research indicates LASER has generated violent crime reductions (Uchida & Swatt, 2013).

The central point is that there are likely to be gun violence prevention strategies most appropriately aimed at youth offenders, adult offenders, and at gun violence generally. These gun violence strategies could likely be fruitfully combined with a variety of developmental and situational crime prevention strategies (Welsh & Farrington, 2012). Research also suggests that selection of appropriate strategies be developed as part of a problem-solving framework whereby analyses of local gun-crime patterns inform the selection of evidence-based and evidence-informed strategies (McGarrell, 2010). Analyses of gun-crime patterns at the local level can establish patterns such as the prevalence of youth co-offending groups, geographic hotspots, drug markets, and domestic violence that should shape the selection of youth-focused, adult-focused, and general prevention strategies.

Despite the noteworthy findings, there are some notable limitations in this study. The sample population was highly homogeneous (primarily Black males), and the study region comprised some of the most economically disadvantaged and violent neighborhoods within a city that experiences very high levels of gun violence. In addition, because the data were collected as part of an ongoing violence reduction program, it is possible that these patterns of violence may not fully represent all urban gun violence. The intervention area was specifically chosen because of its high rates of violent gun crime, and connection to known offending groups. While these results likely cannot be generalized to all cities, they may be applicable to cities that are struggling with high crime rates and crime which is driven by group- and gang-related violence, drug markets, and inter-personal conflicts. Researchers should examine whether patterns of violence are similar in medium- and low-crime cities as compared with high crime cities. This study also only addressed incidents involving the use of a firearm. Other studies should expand upon this research by replicating the methods presented here, and by all aggravated assault types and unarmed robberies.

Finally, this study’s second research question raised the issue of varying efficacy of violence prevention strategies based on offender age. The findings suggest, but do not test, the proposition that some gun violence prevention strategies are likely to have an impact on youth offenders (e.g., gang- and group-focused), whereas others may be more appropriate for older offenders (e.g., parolee forums) or as general strategies (e.g., place-based problem solving). Future studies of these types of gun violence prevention strategies would benefit through analytic approaches that test the relative impact on youth versus adult offending and victimization.

In conclusion, the results of this study suggest that individuals who are involved in the most serious kinds of violent activity are quite similar in the time, place, and circumstances in which they offend, regardless of age. These findings are largely consistent with a body of literature suggesting that individuals who persist in violent activity into adulthood have patterns of criminal activity reaching back to adolescence and early adulthood (Moffitt, 1993; Sampson & Laub, 2005). While not explicitly tested here, it is possible that older gun violence offenders retain many of the characteristics of offending in their youth. Based on these findings, it is logical to suggest that similar intervention strategies may be used for both youthful and adult offenders. Despite these similarities, however, slight differences emerged regarding the prevalence of co-offending among these two groups, which could affect the types of strategies used. Future research should explore co-offending patterns that may affect prevention strategies. Researchers should also build on these findings by examining the generalizability of these findings in other cities, particularly in cities with more social structural variation to assess the importance of structural variables. Finally, researchers should also further explore the role of crime attractors and whether they affect these two populations differently.