A Review of the Reality of Violent Offending and the Administration of Justice

Abstract

Policy makers have recently been exploring methods to reduce incarceration. Most current proposals for reducing incarceration exclude people in prison who were convicted of violent crimes. This article considers violence exclusion from criminal justice reform by examining a sample released from New Jersey’s prisons (n = 375). We assess the hidden nature of the violence in the crime through a review of the characteristics behind the criminal label, with particular focus on the displayed aggravation and mitigation. Postrelease outcomes for this sample are compared with a sample released for nonviolent crimes (n = 1,185). Findings reveal people released for violent offenses frequently display mitigation, as opposed to aggravation. Those released for violent crimes are less likely to be rearrested than those released on nonviolence. The degree of severity of the violence itself has no bearing on the risk of rearrest. Accordingly, policy reform for those convicted of violence is considered.

Keywords

violence recidivism risk criminal justice policy reform

Introduction

The United States has entered a period of prison reform, and a consensus has developed that prisons are overcrowded. With less than one twentieth of the world’s population, the United States has more than one fifth the world’s incarcerated population. However, this was not always the case. In 1972, the U.S. incarceration rate was comparable with that of other Western democracies and had been semi-stable for as far back as national prison population data were collected. At the time, this stability was explained by a “homeostatic theory of imprisonment,” which held that societies demand semi-stable levels of punishment (Blumstein & Cohen, 1973; Blumstein et al., 1976). After the eruption in the prison population that began in the early 1970s, that theory was abandoned in favor of a more apt recognition of American Penal Exceptionalism (Pratt, 2011; Reitz, 2017). This theory rests on the belief that when compared with other industrialized nations, the United States exhibits disproportionately higher rates of imprisonment, while also maintaining use of the death penalty (Reitz, 2017). Today, the United States is the world’s leader in incarceration with 2.2 million people in the nation’s prisons and jails; a 500% increase during the last 45 years of imprisonment (Carson, 2018). Today, the United States experiences what is commonly referred to as “mass incarceration.”

For much of the last 40 years, prison growth was accepted as a necessary consequence of high rates of crime. Recently, however, analyses of the connection between crime and imprisonment have questioned the value of high rates of incarceration as crime prevention (for a review, see National Research Council, 2014), while at the same time recognizing a broad range of undesirable consequences of prison growth (Clear, 2007; Western, 2007). As this kind of evidence accumulates, there is a newly emerging consensus that U.S. prison policy has been dysfunctional and that we should become less unique in our corrections policy. Even among the typical political divide, there appears to be a shared critical view of mass incarceration.

As doubts about high rates of incarceration have grown, policy makers have turned their attention to look for ways to cut the number of people in prison. A range of strategies have been proposed, and many implemented, with the aim of reducing the number of people behind prison bars (Schrantz et al., 2018). In general, lawmakers and the public are open to reducing sentences for drug crimes and other types of nonviolent property crimes; however, violent crimes are most often exempt from these reform discussions (Pfaff, 2019).

Numerous criminologists have pointed out that it is difficult to make headway in reducing mass incarceration without including people who are in prison for violent crimes (Courtney et al., 2017; Pfaff, 2017; Wilson, 1987). People convicted of violent crimes represent more than half of those in state prison (Carson, 2018; Wagner & Sawyer, 2018). Insiders know that significant reductions in prison numbers cannot occur if the people serving time for violent crimes are excluded from the conversation. However, this has not been recognized by those who are in charge of making policy decisions. Recently, for example, the Brennan Center published a collection of 20 essays calling for prison reform, written by prominent policy makers ranging from Bernie Sanders and Elizabeth Warren to Jared Kushner (Chettiar & Raghavan, 2019). The authors drafted passionate pleas for changing the manner by which drug laws are enforced, the services provided to those in prison, and the support received for reentry services. None of the essays called for a review or an amendment to the way violent crime is addressed. President Obama, for his part, granted 1,715 sentence commutations during his presidency. The commutations were given to people serving time for nonviolent drug offenses, but excluded individuals convicted of violent crimes. Stating the obvious, observers note that the case for sentence clemency grows even more difficult when you speak about violence (Mahdawi, 2018).

The reality that deep prison population cuts are not feasible without reducing the number of people in prison for violent crimes is one of the lessons derived from the so-called Iron Law of Prison Populations (Clear & Austin, 2009). The Iron Law holds that prison populations are entirely dependent on how many people are sent to prison and how long they stay. While the number of people sent to prison for violent crime has been declining, their proportion of the daily prison count has been increasing due to longer sentences. It follows, then, that length of stay for violent crime must be adjusted, to substantially change the number of people in prison on any given day.

Reform advocates point out that there are good reasons to include people serving time for violence as a part the overall reform effort (Eisen et al., 2016). People convicted of violence are older at prison release, and follow-up studies show them to be a lesser risk to the community than people serving time for other crimes (Alper et al., 2018; Langan & Levin, 2002; Nagin et al., 2009; Snodgrass et al., 2011). As moderate differences in length of stay seem to have no impact on the likelihood of new criminality (Drago et al., 2009; Meade et al., 2012; Mears et al., 2016; Rhodes et al., 2018; Roach & Schanzenbach, 2015), public safety will not necessarily be compromised by reducing the time-served for relatively large cohorts of people convicted of violent crime.

These are not new arguments; it has long been known that many people serving time for violence can be released to the community without putting the community at risk (Day & Doyle, 2010; Goldstein, 1975; Whitehead et al., 2007). The problem, of course, is not risk but stakes. Individuals with violent crime histories may rarely commit new violent crimes (Goldstein, 1975; Putkonen et al., 2003), but these events, while rare, have powerful consequences for the legitimacy of the justice system and its authorities when they occur. The problem is that while repeat violence may be unlikely, the dire costs to the system, when it does occur, are certain. When political leaders say their ideas for reform do not apply to those who are in prison for violence, what they really mean is that they do not apply to those for whom the stakes are elevated.

The Problem of “Violence”

At the heart of the problem is that the term, “violent criminal,” connotes images that make reform unpalatable. When most individuals hear the term, “violent criminal,” there is an image of someone who has no compunctions about hurting other people—someone who indeed enjoys inflicting pain. The term “violence” invokes and image of people who jump out from dark, hidden places, to attack people who are perfect strangers and to hurt them, perhaps gratuitously, to get what they want. These images show precisely why talk of prison reform leaves “the violent” out of the reform picture.

Yet, violent crime labels are often misleading. While the titles the law gives to crimes, armed robbery, aggravated assault, sexual assault, certainly denote harmful events, each criminal event also has its own circumstances. We have been long aware of some of these circumstances. People hurt known individuals more often than they hurt complete strangers. Crimes of violence arise from spontaneous events more often than they are planned ahead of time. The larger point, though, is that every crime of violence is a story. They are typically comprised of precipitating events, and they are often engaged in by people in some kind of current and (often) prior relationship (Wortley, 2001, 2016). The violence within the criminal event can also be restrained or excessive (Laajasalo & Häkkänen, 2006; Porter et al., 2003). For example, if two people get in an argument at a bar, and the argument continues outside the bar, this may eventually lead to an assault. The person who is “mostly responsible” for the assault often gets the prison sentence. Should this person be ineligible for sentencing reform, solely because an assault occurred?

This example demonstrates how a categorical ban on prison reform for people convicted of violent crimes can be based on faulty assumptions about what the violence in a violent crime event actually looks like. To the degree that situations such as these are representative among the population of those serving time for violence, then fencing these individuals off from the rest of the mass incarceration agenda rests on defective inferences about their crimes. It may be that people convicted of violent crimes present neither the risk nor the moral atrocity that individuals assume applies to the term, “violent criminal.”

In this article, the material nature of the criminal event that earned imprisoned individuals the label, “violent criminal,” is explored. For each event, questions are asked about the circumstances, motivation, victim, nature, and degree of violence contained in that violent crime event. Our purpose is to investigate the features of each specific criminal violent act so as to better understand what occurred. The aim is not to challenge the label that was imposed by the legal system, but rather to obtain a better understanding of the dynamics that permitted the label of violence or the nature of the criminal violence. In addition, we then consider the postrelease behavior of those with and without the violent label.

Violent Events as Situations

This approach is broadly consistent with a movement, popular among criminologists several decades ago, to think about crime in terms of criminal behavior systems as opposed to criminal law labels (Clinard et al., 2014; Clinard & Quinney, 1967). This approach held that behavioral labels used in the prosecution of crime, built as they were from a foundation of mens rea and the methods of legal classifications, were mostly concerned with blameworthiness and had limited value in understanding the nature of the crime. The idea was that criminal behavior is, above all, human behavior and not merely legal behavior. If crime is to be understood, then it should be investigated as any type of behavioral event that requires an interdisciplinary focus, merging the tools of the social sciences (sociology, psychology, social-psychology, and political science) to understand not so much why crimes occur, but to know more precisely what kind of human behavior crime is.

There are multiple ways that violent acts may be nuanced. For example, the victim and perpetrator may know each other or be perfect strangers. They may be involved in a situation of conflict, or the violence can seemingly come from nowhere. The perpetrator may be cruel, inflicting gratuitous harm or the individual may take steps to limit, or mitigate, the violence displayed. Other nuances include whether the act was planned or spontaneous, the level of instigation, and whether a weapon was utilized. It is easily seen that different answers to these questions signify extremely different events. A situation in which a person hits someone with a stick he picks up as part of an escalating argument is entirely different from one in which a person brings a baseball bat to an event expecting trouble. Both are assaults; however, they represent different violence situations.

These types of differences matter in two respects. One way in which they are different is in levels of “blameworthiness.” Legal scholars have cited blameworthiness as a key justification for punishment (Feeley, 1992; Heumann, 1978; von Hirsch, 1976; Wishman, 1982/2013). In the examples above, a person who plans a crime and carries it out is more blameworthy than a person whose emotions got the better of him, even though both have broken the law. A person who chooses a more vulnerable victim, such as a child, is more blameworthy. Presumably, judges take blameworthiness into account when sentencing, imposing harsher sanctions on those who are believed to be more deserving.

A second way these situational nuances matter is that some provoke greater dread or shock in the general public. People may be appalled by the person who stabs another during a drunken poker game, but they are terrified by someone who stabs a perfect stranger. It is the premise of this article that situational nuances deserve greater study, because they matter. Both of these indications of “blameworthiness” and “shock” are considered here after the violent events are coded into categories, specifically when the individual is determined to aggravate or mitigate the violence.

A Strategy for Understanding “The Violent Situation”

To obtain a deeper understanding of the circumstances of the crime, we used the official description of the offense, as recorded by the police at the time of the crime, as a baseline account of the occurrence. There may be inaccuracies in this description, certainly, and the process of jurisprudence may uncover evidence of which the police were unaware. Likewise, the restrictions of evidentiary due process exclude some of evidence referred to in the police report. There may be aspects of the criminal event that do not get recorded in the official police description at all. To be sure, the police report of the criminal event is by no means a conclusive description of what occurred.

For our purposes, however, using the police description of the criminal event has several advantages. We are interested in examining the underlying actions in a criminal event that the legal system defines as “a violent crime.” As initial step of our study was to sample only individuals currently serving time for a violent crime, we ended up excluding all cases where (a) the legal processes excluded evidence of violence initially reported by the police or (b) the prosecutor’s investigation questioned the accuracy of a police report. Thus, these two types of police inaccuracies are avoided using our sampling method. Finally, there were limited instances when the police failed to identify an aspect of violence in a crime that ends up resulting in a conviction for violence. These scenarios were rare, but when encountered, the description of the crime in the presentence investigation was used to code the underlying “official” description of the criminal event.

Method

Study Overview

The strategy for more deeply investigating the nature of the violence in criminal events resulting in incarceration was to examine the official police descriptions of the criminal event for a sample of people serving time for a violent crime (if needed, we also reviewed the official description of the event provided to the court at sentencing). This provided us the majority of the information necessary for the purpose of this study—to explore the nuances behind the criminal violence label and determine how much re-offense potential people who have been convicted of violent crime pose to the public upon release from prison, compared with those convicted of a nonviolent crime.

Working from the official description of the criminal event, we coded the following: (a) the role the person played in the violent crime, (b) the relationship to the victim of the violence, (c) the level/severity of the physical violence, (d) the weapon used in the violent act, (e) the degree of planning that went into the completion of the violent act, (f) how the violence was initiated, and (g) the motivation for the violence. We collapsed these variables into an overall measure of the degree of aggravation and mitigation displayed during the violent event. Our analysis considers whether any of these situational characteristics are associated with sentence length, in-prison disciplinary reports, and subsequent arrests and convictions upon release, particularly for violent crimes.

Study Data

The data for the study consists of two random samples drawn from a cohort of all cases released from the New Jersey Department of Corrections in 2014 (N = 9,934). Of this cohort, 2,279 were released from prison while serving time for a violent crime. A random sample of 375 males from the violent crime cohort (16.5%) was drawn; they were known as the violent sample. A proportionate sample size of males (16.5%) who were released for a nonviolent crime was also pulled (n = 1,185). This yielded a working stratified random sample release cohort of 1,650 men released from prison in 2014.¹

A subject was eligible for the violence subsample if that person’s “instant offense” (the crime for which the person was convicted and sentenced) was designated as a “violent crime” in the New Jersey Criminal Code. Violent crimes include the offenses of murder, robbery, assault, rape (both completed and attempted), and terroristic threats. The nonviolent subsample was drawn from all other releases; their crimes included weapons offenses, property crimes, drug crimes, supervision offenses (i.e., returning after curfew), and administrative offenses (i.e., failure to pay fines).

For every case in the study sample, access was gained to the New Jersey Department of Corrections administrative database. This provided approximately 100 additional variables, including demographic characteristics, prior criminal history, and in-prison history (disciplinary records and prison programs). Three years of follow-up criminal justice contact information for both in-state and out-of-state records including arrests, convictions, and incarcerations were then added. Each case, regardless of the release date in 2014, was standardized to a 1,095-day follow-up period.

The Samples

Table 1 compares the violent and nonviolent samples on demographics. The cases are remarkably similar on race, age, and education. On criminal justice variables, there are understandable differences between the samples. The violent sample served more than twice the time in prison, even though they are significantly less likely to have prior arrests, convictions, and incarcerations. Upon release, the violent sample was more likely to be placed on parole and had a slightly larger proportion of both high- and low-risk groups. While the nonviolent sample has more justice system experience, the violent group was 12 times more likely to receive a mandatory minimum sentence, with 2.5 longer average time-served, reflecting the seriousness of the offense. While the violent group experienced many more in-prison disciplinary infractions, some of this difference was due to the longer time in prison. Their monthly rate of infractions was also 27% higher than that of the nonviolent group (p < .05).

Table 1.

Sample Descriptives (N = 1,560).

Variable	Frequency violent sample (n = 375)	Frequency nonviolent sample (n = 1,185)
Race
White	109 (29.1%)	339 (28.6%)
Non-White	266 (70.9%)	846 (71.4%)
Marital status**
Married	48 (12.8%)	79 (6.7%)
Nonmarried	327 (87.2%)	1,106 (93.3%)
Minimum sentence requirement**
Yes	182 (48.5%)	49 (4.1%)
No	193 (51.5%)	1,136 (95.9%)
Children
Yes	187 (49.9%)	—
No	170 (45.3%)	—
Supervised at release**
Yes	275 (73.3%)	637 (53.8%)
No	100 (26.7%)	548 (46.2%)
Grade level upon release
Less than high school	99 (26.4%)	348 (29.4%)
High school diploma	204 (54.4%)	686 (57.9%)
College credits	43 (11.5%)	104 (8.8%)
Prior arrest record**
Yes	283 (75.5%)	1,107 (93.4%)
No	92 (24.5%)	78 (6.6%)
Prior conviction record**
Yes	235 (62.7%)	1,040 (87.8%)
No	140 (37.3%)	145 (12.2%)
Prior incarceration record**
Yes	104 (27.7%)	447 (37.7%)
No	271 (72.3%)	738 (62.3%)
Crime of conviction
Violent	375 (100.0%)	—
Weapon	—	168 (14.2%)
Drug	—	370 (31.2%)
Property	—	261 (22.0%)
Supervision	—	300 (25.3%)
Administrative	—	86 (7.3%)
In-prison disciplines
0	189 (50.4%)	797 (67.3%)
1	62 (16.5%)	164 (13.8%)
2+	124 (33.1%)	224 (18.9%)
Discipline ratio by time served*	0.0014	0.0011
LSI-R risk score at parole
Low	20 (8.1%)	27 (4.6%)
Medium	184 (75.1%)	502 (85.5%)
High	41 (16.7%)	58 (9.9%)
Not on parole	100 (26.7%)	548 (46.2%)
Missing data	30 (8.0%)	50 (4.2%)
Variable	M/Mdn/s	M/Mdn/s
Age at release	35.09/32.0/11.11	35.21/33.0/10.25
Time served**	4.2 years/3 years/1,589.45	1.7 years/1.3 years/669.25

Note. Numbers that do not add to sample size represent missing data. LSI-R = Level of Service Inventory–Revised.

p < .05. **p < .000.

The demographic similarity suggests that these samples are useful for comparing the differences that arise not from background characteristics or risk levels but from differences due to the purported violent and nonviolent instant offense. The difference in the release offense is, as might be expected, associated with significant differences in criminal justice experience, especially prior record (with violent cases having less extensive previous criminality). This suggests that thinking of the violent and nonviolent groups as being different groups for policy purposes has some basis in the record, even though they look similar on many other criteria.

Coding Violent Crime Situations

The purpose of the article is to explore how improved knowledge of the characteristics underlying a violent crime improves our understanding of the policy-relevant distinctions to be made between violent and nonviolent cases. This approach assumes that crime labels, by themselves, do not provide enough information about the underlying crime to support policy actions and that an improved understanding of those situations might lead to better justice-reform decisions.

To analyze the crime situations, official offense descriptions were examined. A data collection tool was developed to code situational characteristics of the violent event. Seven facets of the violent crime were included to help provide a truer picture of what occurred; in essence, we worked to provide a more nuanced understanding of the crime than the offense category itself provides. Within each of these characteristics or each case in the violence sample, the following variables were coded:

Role played: This assesses whether the assailant acted alone or as part of a group. If the individual acted as one of a group, it was determined whether the assailant led the violence or followed someone else’s lead. The role options included Sole Actor, Leader, Participant/Follower, and Role Unclear.

Event planning: This assesses the degree of premeditation involved in the violent act; in particular, we determined whether the criminal event was planned or spontaneous. The options included Planned, Circumstantial (the violence happened during the course of another crime), Spontaneous, and Planning Unclear.

Motivation for violence: This assesses the main goal or objective of the violent act; in particular, we identified the instrumental need the violence sought to meet. The options included Sadism (desire to inflict pain), Pedophilia (sexual gratification with a child), Humiliation (interpersonal domination), To Resolve Conflict (get even with someone or win a dispute), To Get/Take Something (steal an item of value), and Motivation Unclear.

Level of violence: This assesses the level of needlessness or cruelty in the violence; in particular, we examined whether there was overwhelming violence that appeared to be gratuitous, or whether there was evidence of self-restraint. The options included Enjoyed Inflicting Violence, Excessive Violence but Ceases When Successful, Controlled/Restrained Violence, Accidental Violence, and Violence Unclear. Because these categories might be deemed as subjective, coders were trained to decipher cues in the inmate files. Individuals who were believed to “Enjoy inflicting violence” were described in the file by the police or clinician to have no remorse, receive pleasure from the violence, and/or laughed at the violence they inflicted. Those receiving a designation of “Controlled/restrained violence” exhibited that they attempted to mitigate their physical or psychological harm by demonstrating reduced violence or victim/offender feedback stated that they had done so. Cases demonstrating “Excessive violence that ceases when successful” fulfilled scenarios where the researchers witnessed extreme levels of violence prior to the victim either providing a tangible item or becoming passive to an attack. No further violence took place after that. Accidental violence was violence that occurred but was not intentional (i.e., the victim fell and hit their head while the offender grabbed their bag and ran away, automotive manslaughter, etc.).

Relationship: This established the relationship between the victim and the perpetrator. The relationship between victims and offenders has been subject to much study, including standard definitions for what constitutes a “stranger” (Truman & Langton, 2015). Our options included: Complete Stranger, Acquaintance/Friend, Loved One/Dependent, Multiple Relationships, and Relationship Unclear.

Initiation status: We determined how the violence was initiated. Information was coded on who initiated the crime and whether the violence was a response to a preceding interaction. The options included Instigator, Responder/Reactor, and Initiation Unclear.

Instrument: We identified the weapon used to complete the violent act and whether it was carried or improvised at the scene. The options included Weapon Carried to Event, Improvised Weapon, Body Part (fist, feet, etc.), Verbal Threats Only, and Unclear Weapon or Instrument.

The coding tool was developed by reviewing a sample of criminal incident cases. Prior to coding the full 375 cases, the authors piloted approximately 35 cases to determine the overarching violence characteristics and the options within them. The options within each of the seven characteristics were mutually exclusive and provided no overlap. We used descriptions of coding conventions based on common examples of offense descriptions and developed illustrations of the operational meaning for coding options for each variable. Three research assistants were trained on the coding strategy by the research unit of the New Jersey Department of Corrections. To obtain a reliability estimate, two researchers coded each file. If there was disagreement on any of the answers to the seven categories, the case was discussed by lead researchers and the research assistants. There were 42 cases of disagreements in coding (11% of the sample), all of which involved the “Level of Violence” category. The inter-rater reliability score using Cohen’s kappa demonstrated 89% agreement.

Table 2 shows the results of the violent event coding. Generally, the cases in our sample included a sole actor (63%) and most individuals were the instigators of a crime (89%), which was committed with some degree of planning (60%) rather than completely spontaneous (26%). The most common motivation for the violence was to get or take something (39%), though the violence also commonly took place to resolve a dispute or other conflict (20%). While nearly two fifths of the victims were complete strangers (39%), a clear majority of victims were either acquaintances (33%) or loved ones (21%). While the person committing the violence was often carrying a weapon (35%), the perpetrator most often either kicked or hit the victim (37%). Nearly one fourth of the time the criminal act was motivated by sadism (11%) or pedophilia (14%), while humiliation (3%) was a rare motivation. In one third of the events, the person showed evidence of enjoying inflicting pain (33%), and even though the violence generally ended as soon as the person was successful at achieving his desired result, the violence in many cases was excessive (37%). Carefully controlled or obviously restrained use of violence occurred only about one sixth of the time (17%).

Table 2.

Coding Tool Descriptives and for Violent Offenders (n = 375).

Variable	Frequency	Percent	Coding schema
Violent crime of conviction
Sexual assault	85	22.9	—
Assault	99	26.5	—
Homicide	39	10.1	—
Robbery	124	33.0	—
Terroristic threats	28	7.5	—
Role
Sole actor	237	63.2	Aggravating
Leader	41	11.0	Aggravating
Participant/Follower	53	14.1	Mitigating
Role unclear	44	11.7	Neutral
Event planning
Planned	223	59.5	Aggravating
Circumstantial	28	7.5	Mitigating
Spontaneous	99	26.4	Mitigating
Planning unclear	24	6.4	Neutral
Motivation for violence
Sadism	41	10.9	Aggravating
Pedophilia	54	14.4	Aggravating
Humiliation	12	3.2	Aggravating
To resolve conflict	76	20.3	Mitigating
To get or take something	147	39.2	Neutral
Motivation unclear	44	11.7	Neutral
Level of violence
Enjoyed inflicting violence	124	33.1	Aggravating
Excessive violence that ceases when successful	134	35.7	Neutral
Controlled/Restrained violence	65	17.3	Mitigating
Accidental	13	3.5	Neutral
Level of violence unclear	38	10.1	Neutral
Relationship
Complete stranger	147	39.2	Aggravating
Acquaintance/Friend	124	33.1	Neutral
Loved one/Dependent	80	21.3	Neutral
Multiple relationships	10	2.7	Neutral
Relationship unclear	13	3.5	Neutral
Initiation status
Instigator	333	88.8	Aggravating
Responder/Reactor	26	6.9	Mitigating
Initiation unclear	15	4.0	Neutral
Instrument
Weapon carried to event	131	34.9	Aggravating
Improvised weapon	52	13.9	Neutral
Body part	140	37.3	Neutral
Verbal threats only	28	7.5	Neutral
Weapon unclear	23	6.1	Neutral
Coding type
Aggravating score	Range = 0–7	Mdn/Mode = 4/3	Distribution = 82%, Score 3+
Mitigating score	Range = 0–5	Mdn/Mode = 1/1	Distribution = 81%, Score 0–1

Note. Numbers that do not add to 375 represent missing data.

The coding for these situational variables provides categorical level measures rather than ordinal measures. While these categories are not ordinal, for each situation, some categories represent more severe levels of violence than others. To facilitate the analysis, the sub-situational categories were further collapsed into those that represent “aggravating” aspects of the violent event and those that may be seen as “mitigating.” The terms “aggravating” and “mitigating” were chosen intentionally, because these are legal terms that establish blameworthy and relevant nuances on which judges rely to impose sentences. The classification of characteristics as aggravating or mitigating was based on whether each of the seven violence categories would be viewed to incur blame or risk, or whether they would be looked at as attempting to diminish the violent event. While our classification does not intend to reproduce those legal elements, our nomenclature is meant to place in context the implication of our coding of these violent criminal events.² The term aggravating was chosen to represent situations where the individual antagonized or amplified the event, while mitigation denotes de-escalation or neutralization. The seven violence categories are collapsed into the following codes:

Role played—Aggravating: Sole Actor, Leader and Mitigating: Participant/Follower and Neutral: Role Unclear.

Event planning—Aggravating: Planned and Mitigating: Circumstantial, Spontaneous and Neutral: Planning Unclear.

Motivation for violence—Aggravating: Sadism, Pedophilia, Humiliation and Mitigating: To Resolve Conflict and Neutral: To Get/Take Something and Motivation Unclear.

Level of violence—Aggravating: Enjoyed Inflicting Violence and Mitigating: Controlled/Restrained Violence and Neutral: Excessive Violence that Ceases When Successful, Accidental Violence and Level of Violence Unclear.

Relationship—Aggravating: A Complete Stranger and Mitigating: (none) and Neutral: Acquaintance/Friend, Loved One/Dependent, Multiple Relationships, and Relationship Unclear.

Initiation status—Aggravating: Instigator and Mitigating: Responder/Reactor and Neutral: Initiation Unclear.

Instrument—Aggravating: Weapon Carried to Event and Mitigating: (none) and Neutral: Improvised Weapon, Body Part, Verbal Threats Only, and Weapon Unclear .

These categorizations produce two scales: an “aggravation” scale with possible values ranging from 0 to 7 and a “mitigation” scale with possible values ranging from 0 to 5. For each crime event, then, and individual receives both an aggravation and a mitigation score. Table 2 shows the distributions of codes for the seven characteristics, as well as the summary mitigation and aggravation scores. The mode and median values for the mitigation score were both one, while the mode and median for the aggravation score were three and four, respectively. The total aggravation and mitigation scores are concentrated: 81% of the mitigating scores fell between 0 and 1, while 82% of the aggravating scores fell between 3 and 7. The correlation between these aggravating and mitigating scores is −.56, as would be expected. For the crimes committed by this violent sample of prison releases, aggravation was more frequent than mitigation. This result is not also unexpected, because the crimes they committed were deemed serious enough to result in a prison sentence (it may be that more mitigated versions of these offenses result in nonprison sentences). That said, the criminal events were generally more nuanced than we might think. Overall, 99.7% of the sample had at least one aggravating factor and yet 63.2% had at least one mitigating factor.

The pattern of aggravation and mitigation is generally consistent with what might be expected. Individuals convicted of sexual assault showed more aggravation during the course of the violent event than any other crime, with approximately 80% scoring between 4 and 6 on the aggravation scale. Those convicted of robbery showed the next highest aggravation rates, with just over half scoring between 4 and 6. For the remaining violence types, aggravation and mitigation were more equally distributed. Over 60% of those who were convicted of sexual assault scored 0 on the mitigation scale, indicating they did nothing to temper the violence involved in their criminal act. Almost 40% of those who committed robbery displayed no mitigation. Individuals committing assault demonstrated the highest degree of mitigation during their offense.

From this analysis, we can see that simple offense categories such as robbery or assault can conceal considerable variability in the underlying crime situation. Criminal events bearing the same statutory title can differ significantly in their underlying seriousness—a fact well known to jurists who seek discretion in imposing punishments, for exactly that reason. Research studies that use rely solely upon the offense categories fail to take account of nuances in the seriousness of criminal acts. Just as significantly, prison reform policies that impose categorical exclusions using those same offense titles treat underlying conduct equivalently. This study attempts to uncover just how important those nuances may be.

Analysis

The first step of the analysis compares the violent cases to nonviolent cases on correctional outcomes. The predictive power of current violent offending on rearrest is then tested, while controlling for background variables. The analysis concludes by examining whether the specific crime event characteristics, as measured by the mitigation and aggravation scores, contribute to the prediction of continued postrelease offending. Subjects who scored highest on the aggravation scale are also examined separately to determine whether increased levels of aggravation during the act of violence lead to increased re-offending.

Nonviolent and Violent Samples as a Comparison

Table 3 compares the violent and nonviolent samples on correctional outcomes in the 36 months following release from prison. These samples are compared on the number of rearrests and re-convictions, the number and type of re-incarcerations, and the time to the first violation postrelease. The nonviolent sample had more than twice the rate of rearrests and re-convictions in the follow-up period (p < .001). Barely 4% of the violence sample is returned to prison for a new crime conviction. The comparable rate for the nonviolent sample was also low (11%) but is 2.5 times higher than the violent sample (p < .001). Despite this lower level of repeat criminality, the violence cases were just as likely to be readmitted to prison overall, during the follow-up period (likely due to a return to prison for supervision rule violations). It is plausible that cases released from prison on a violent crime are subjected to greater scrutiny than other cases, and misbehavior that falls short of a new crime can result in removal from the community.³ Looking at arrest outcomes alone, it seems that the people preleased from New Jersey prisons on violent crimes may not represent a greater risk to the community than other releasees.

Table 3.

Difference of Mean/Proportion—t-test and χ² Results for Postrelease Behavior.

	Violent sample (n = 375)	Nonviolent sample (n = 1,185)	t/χ² value	df	p
Subsequent criminal offending
Number of rearrests	0.68	1.54	8.82	922	.000
Number of re-convictions	0.31	0.73	8.10	894	.000
Number of re-incarcerations	0.38	0.35	−0.79	589	.407
Re-incarceration type			23.01	2	.000
Not readmitted	249 (66.4%)	809 (68.3%)	—	—	—
New crime commitment	16 (4.3%)	130 (11.0%)	—	—	—
Supervision violation	110 (29.3%)	246 (20.8%)	—	—	—
Time to first violation	128 days (SD = 253)	224 days (SD = 291)	−8.23	702	.000

One of the central questions of the analysis is explored in this final section. Particularly, we examine how much re-offense potential people who have been convicted of violent crime pose to the public upon release from prison, compared with those convicted of a nonviolent crime. To test this question, a prediction model for the dependent variable (DV) “rearrest” was constructed. There exists considerable dispute over the best measure of recidivism to evaluate correctional outcomes. For the current purposes, this study most notably seeks to understand what people released from prison do, rather than those who are responsible for supervising them. Re-incarcerations would arguably be a good measure, but for both violence and nonviolence cases, a large proportion of the re-incarcerations result from rule violations rather than new crimes. These violations represent correctional supervision policies as much as they do acts that impinge upon public safety. By comparison, police arrests most certainly reflect police discretion, but provide a closer approximation of conduct that imperils public safety.⁴

Table 4 presents the results of a logistic regression model predicting rearrest, along with the control variables of race (White/non-White), age at release, the number of prior arrests, violent offense as the index offense (violent/nonviolent offense), time served in days and supervision status upon release (y/n). Based on the importance of these variables in criminological research, particularly in recidivism research, their inclusion in the model aids in the determination of how much of the differences in criminal justice outcomes between those convicted of violent and nonviolent crimes can be attributed to crime conviction type.

Table 4.

Logistic Regression Predicting Rearrest for Full Sample (N = 1,560).

	B	Wald	p	Exp(B)
Race (White/Non-White)	0.132	1.12	.290	1.14
Age at release	−0.078	117.03	.000	0.926
Time served	0.000	0.000	.283	1.00
Number of prior arrests	0.109	106.89	.000	1.12
Violent index offense (y/n)	−0.802	29.55	.000	0.449
Supervision at release (y/n)	−0.383	10.80	.001	0.682
Constant	2.27	0.249	.000	9.68
Nagelkerke R²	.227

The results indicate that race and time served are nonsignificant in the model, while age, has the expected significant, inverse relationship to postrelease rearrest. As individuals age, their likelihood of rearrest decreases. The number of prior arrests has the expected positive relationship, indicating as the number of prior arrests increases, so do the odds of rearrest. In addition, being released from prison for a violent crime reduces a person’s odds of rearrest, as does being released to parole supervision. The results presented in Table 4 indicate that being convicted of a violent crime, while controlling for a number of variables, does not enhance the risk a person represents to the community upon release.

As distinctions between violent and nonviolent offenses do not support the rationale of violence begetting violence, we subsequently explored a more nuanced understanding of the underlying nature of the violent act. To this end, we constructed a logistic regression model for rearrest on the violence sample only, adding the aggravating and mitigating characteristics of the offense. The results, presented in Table 5, now show that younger age, prior arrests, and supervision status at release all have the expected, significant impact on rearrest probability. There is also a significant effect of race, with non-Whites more likely to be rearrested. Neither mitigation nor aggravation have a significant impact on the odds of rearrest.⁵

Table 5.

Logistic Regression Predicting Rearrest for Violent Sample (n = 375).

	B	Wald	p	Exp(B)
Race (White/non-White)	0.887	7.90	.005	1.19
Age at release	−0.119	37.13	.000	0.888
Number of prior arrests	0.159	37.99	.000	1.17
Supervision at release (y/n)	−0.616	4.06	.044	0.540
Mitigation present (y/n)	0.173	0.345	.557	1.19
Aggravation score	−0.101	0.811	.368	0.904
Constant	2.24	8.09	.004	9.40
Nagelkerke R²	.312

Discussion

This study is offered as an entry into the crucial debate on prison reform for people convicted of violent crimes. There is broad consensus that America’s prison system is too large and that too many individuals are incarcerated. As a result, at the local, state, and federal levels, a wide range of legislative and programmatic efforts aim to reduce the number of people in prison. Despite a growing awareness among criminologists that prison reform cannot proceed very far without addressing the large number of people serving time for violent crime, most stakeholders indicate their efforts are not intended to apply to “violent offenders.” This reluctance to extend the prison reform agenda to include people who have committed violent crimes reflects two different concerns. First, many feel that violent crimes are so appalling that prison reform is not called for, and second, concern exists that people who commit these crimes represent such a risk to the public that it would be unwise to include them in the prison reform agenda. The purpose of this article was to investigate both of these concerns. Partial support is found for the first claim, while no support is found for the second.

To obtain a better understanding of the cruelty of violent crimes, arrest reports for the violent crime were examined. Factors associated with the serious of the criminal incidents were coded. While factors aggravating the seriousness of the criminal conduct were frequently present, they were not uniformly present. In fact, in 63.2% of the cases, there is at least one mitigating factor present in the manner by which the crime was carried out. The underlying behavior that leads to the criminal conviction is nuanced in various ways, and it is often not in accord with public stereotypes of the crime, as well as those based solely on the legal conviction.

To be sure, the criminal law recognizes that people who commit acts of violence deserve more severe punishments, if only from the framework of retribution (von Hirsch, 1976). This begs the question, how much more punishment is deserved for violent acts? Most would agree that the answer to this question should reflect nuances in the violence of the acts themselves. Our results show that within broad legal classifications criminal acts, there is a range of mitigating and aggravating conduct. If the legal theory of “just deserts” means greater punishments for violent acts, it also requires that this range of aggravation and mitigation in demonstrating violence be taken into account when determining punishment. In our sample, there were numerous instances in which the character of the act itself may have justified less onerous penalties than applies to the ordinary crime of that type.⁶

If there are reasons to doubt that the seriousness of the crime always precludes a shorter sentence for the person convicted, the case regarding public safety concern is even stronger. In this sample, those who were released from prison for a violent crime were significantly less likely to be arrested within 3 years after the release. Not only did these cases perform better, the aggravating factors of the instant crime had no relationship to risk of future rearrest. From our data, no public safety argument can be made for a categorical wall between prison reform and people serving time for violent crime.

Stakeholders and politicians may argue that the longer sentences imposed on those convicted for violent crimes explains their lower rate of recidivism, as they are deterred from new crimes. This notion is contradicted by recent studies finding that a reduction in length of stay in prison of a few months has no impact in risk of new crimes (Drago et al., 2009; Mears et al., 2016). In fact, in our sample, people released from prison on for violent offenses were far more likely to be returned to prison for violating the rules of parole, rather than for the commission of a new crime through an arrest. Findings such as these fuel a growing discussion that public safety claims alone cannot justify excluding people in prison for violent crimes from the prison reform agenda.

Our findings must be examined within the context of the limitations of our data and methods. First, official criminal records were the necessary means of measuring both the characteristics of the violent crimes and recidivism upon release from prison. While it is customary to use these types of data in recidivism research, we must acknowledge the potential pitfalls of only relying on these sources. Not all offenses come to the attention of official criminal justice stakeholders. Undoubtedly, official descriptions of the offense as recoded by police will vary from what actually occurred. There is always measurement error in the use of official data in studies such as ours. By choosing official records, however, we err conservatively. The narratives we rely on are, after all, the same narratives policy makers would use to decide whether a person should be eligible for prison reform.

Second, by selecting our sample on the basis prison release offense, we mask violence contained in a person’s criminal history. Within our comparison sample of people released for nonviolent crimes, there are clearly cases with prior records for violent crimes.⁷ We note, however, that most prison reform proposals that categorically exclude people for violence refer to the current offense and do not necessarily refer to a person’s total criminal record. In that respect, our study method matches the policy arena it critiques.

Policy Implications

The policy implications of these findings are both straightforward and subtle. The plain, straightforward implication is that people who are convicted of violent crimes should be included in policy discussions designed to reduce the size of the prison population. In fact, a strong case can be made that this subpopulation of people in prison should be at the center of reform efforts. People serving time for violent crime are a majority of today’s prison population. As reform efforts regarding drug crime take greater effect, diverting many people who might otherwise have gone to prison, the concentration of people convicted of violent crime will continue to increase. Without addressing those serving time for violence, the ceiling on attempts to shrink incarceration will get lower and lower. The policy converse is also true. Because people doing time for violence serve longer sentences, reasonable reductions in their time behind bars will lead to even larger reductions in overall prison populations. The results indicate that these reductions can be pursued without placing the public at undue risk.

The more subtle implication is just as significant, though less easily addressed. The public conversation about “the violent offender” is too often based on a stereotype that is often misfocused. This stereotype makes it more difficult for the people who receive this label to overcome the stigma of their crimes. These individuals stay longer than necessary in prison and they face significant impediments upon release—including closer scrutiny by authorities and significant limitations on housing, employment, and civic life (Hager, 2019).

The policy consequences, as we have shown, are unnecessary for many who have been convicted of violent crime. But more to the point, the public conception of violence is erroneous in ways that distort the public debate about prison reform. It is time for the public conversation about violent crime to reflect a more informed recognition of the facts that are often hidden behind the cloak of stereotype.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Kristen M. Zgoba

Notes

Author Biographies

Kristen M. Zgoba is an assistant professor at Florida International University. Prior to academia, she was employed as Supervisor of Research for the New Jersey Department of Corrections for approximately 15 years. She studies violence, correctional research and policy evaluation and was recently awarded a Fulbright Scholars award to study sexual crimes in the United Kingdom. Her recent research appears in Sexual Abuse: A Journal of Research and Treatment, Criminal Justice and Behavior and the Prison Journal.

Todd R. Clear is a University Professor of Criminal Justice at Rutgers, The State University of New Jersey. He has served previously as Provost of the University, and before that Dean of the School of Criminal Justice. Clear has also held professorships at John Jay College of Criminal Justice (where he held the rank of Distinguished Professor), Florida State University (where he was also Associate Dean of the School of Criminology and Criminal Justice) and Ball State University. Clear has authored 13 books and over 100 articles and book chapters. His most recent book is The Punishment Imperative, by NYU Press. Dr. Clear has also written on community justice, correctional classification, prediction methods in correctional programming, community-based correctional methods, intermediate sanctions, and sentencing policy.

References

Alper

Durose

Markman

(2018). 2018 Update on prisoner recidivism: A 9-year follow-up period (2005-2014) (NCJ 250975). U.S. Department of Justice, Office of Justice Programs, Bureau of Justice Statistics.

Blumstein

Cohen

(1973). A theory of the stability of punishment. The Journal of Criminal Law and Criminology, 64(2), 198–207.

Blumstein

Cohen

Nagin

(1976). The dynamics of a homeostatic punishment process. The Journal of Criminal Law and Criminology, 67(3), 317–334.

Carson

E. A.

(2018). Prisoners in 2016 (NCJ 251149). U.S. Department of Justice, Office of Justice Programs, Bureau of Justice Statistics.

Chettiar

Raghavan

(Eds.). (2019). Ending mass incarceration: Ideas from today’s leaders. The Brennan Center for Justice.

Clear

T. R.

(2007). Imprisoning communities: How mass incarceration makes disadvantaged neighborhoods worse. Oxford University Press.

Clear

T. R.

Austin

(2009). Reducing mass incarceration: Implications of the iron law of prison populations. Harvard Review of Law and Policy, 3(2), 307–324.

Clinard

M. B.

Quinney

(1967). Criminal behavior systems: A typology. Holt, Rinehart and Winston.

Clinard

M. B.

Quinney

Wildeman

(2014). Criminal behavior systems: A typology (3rd ed.). Routledge.

10.

Courtney

Eppler-Epstein

Pelletier

King

Lei

. (2017, July). A matter of time: The causes and consequences of rising time served in America’s prisons. https://apps.urban.org/features/long-prison-terms/intro.html

11.

Day

Doyle

(2010). Violent offender rehabilitation and the therapeutic community model of treatment: Towards integrated service provision? Aggression and Violent Behavior, 15(5), 380–386.

12.

Drago

Galbiati

Vertova

(2009). The deterrent effects of prison: Evidence from a natural experiment. Journal of Political Economy, 117(2), 257–280.

13.

Eisen

L.-B.

Austen

Cullen

Frank

Chattier

I. M.

(2016). How many Americans are unnecessarily incarcerated? The Brennan Center for Justice.

14.

Feeley

M. M.

(1992). The process is the punishment: Handling cases in a lower criminal court. Russell Sage Foundation.

15.

Goldstein

(1975). Aggression and crimes of violence. Oxford University Press.

16.

Hager

(2019, April 3). When “violent offenders” commit nonviolent offenses. https://www.themarshallproject.org/2019/04/03/when-violent-offenders-commit-nonviolent-crimes

17.

Heumann

(1978). Plea bargaining: The experiences of prosecutors, judges, and defense attorneys. The University of Chicago Press.

18.

Hirsch

A. V.

(1976). Doing justice: The choice of punishments. Hill and Wang.

19.

Laajasalo

Häkkänen

(2006). Excessive violence and psychotic symptomatology among homicide offenders with schizophrenia. Criminal Behaviour and Mental Health, 16(4), 242–253.

20.

Langan

P. A.

Levin

D. J.

(2002). Recidivism of prisoners released in 1994 (NCJ 193427). U.S. Department of Justice, Office of Justice Programs, Bureau of Justice Statistics.

21.

Mahdawi

(2018, August 21). The case for releasing violent offenders early. The Guardian. https://www.theguardian.com/us-news/2018/aug/21/clemency-case-for-releasing-violent-offenders-early

22.

Meade

Steiner

Makarios

Travis

(2013). Estimating a dose-response relationship between time served in prison and recidivism. Journal of Research in Crime and Delinquency, 50(4), 525–550.

23.

Mears

Cochran

Bales

Bhati

(2016). Recidivism and time served in prison. Journal of Criminal Law and Criminology, 106(1), 82–122.

24.

Nagin

D. S.

Cullen

F. T.

Jonson

C. L.

(2009). Imprisonment and reoffending. University of Chicago Press.

25.

National Research Council. (2014). The growth of incarceration in the United States: Exploring causes and consequences. The National Academies Press.

26.

Pfaff

(2017). Locked in: The true causes of mass incarceration—And how to achieve real reform. Basic Books.

27.

Pfaff

(2019, August 14). What democrats get wrong about prison reform. Politico. https://www.politico.com/magazine/story/2019/08/14/what-democrats-get-wrong-about-prison-reform-227623

28.

Porter

Woodworth

Earle

Drugge

Boer

(2003). Characteristics of sexual homicides committed by psychopathic and nonpsychopathic offenders. Law and Human Behavior, 27(5), 459–470.

29.

Pratt

(2011). The international diffusion of punitive penality: Or, penal exceptionalism in the United States? Wacquant v Whitman. Australian & New Zealand Journal of Criminology, 44(1), 116–128.

30.

Putkonen

Komulainen

E. J.

Virkkunen

Eronen

Lönnqvist

(2003). Risk of repeat offending among violent female offenders with psychotic and personality disorders. The American Journal of Psychiatry, 160(5), 947–951.

31.

Reitz

(Ed.). (2017). American exceptionalism in crime and punishment. Oxford University Press.

32.

Rhodes

Gaes

G. G.

Kling

Cutler

(2018). Relationship between prison length of stay and recidivism: A study using regression discontinuity and instrumental variables with multiple break points. Criminology & Public Policy, 17(3), 731–769.

33.

Roach

M. A.

Schanzenbach

M. M.

(2015, December 12). The effect of prison sentence length on recidivism: Evidence from random judicial assignment. https://ssrn.com/abstract=2701549

34.

Schrantz

DeBor

Mauer

(2018). Decarceration strategies: How 5 states achieved substantial prison population reductions. The Sentencing Project.

35.

Snodgrass

G. M.

Blokland

A. A.

Haviland

Nieuwbeerta

Nagin

D. S.

(2011). Does the time cause the crime? An examination of the relationship between time served and reoffending in the Netherlands. Criminology, 49(4), 1149–1194.

36.

Truman

Langton

(2015). Criminal victimization, 2014. Bureau of Justice Statistics, U.S. Department of Justice.

37.

Wagner

Sawyer

(2018, March 14). Mass incarceration: The whole pie 2018. https://www.prisonpolicy.org/reports/pie2019.html

38.

Western

(2007). Punishment and inequality in America. Russell Sage Foundation.

39.

Whitehead

P. R.

Ward

Collie

R. M.

(2007). Time for a change: Applying the good lives model of rehabilitation to a high-risk violent offender. International Journal of Offender Therapy and Comparative Criminology, 51(5), 578–598.

40.

Wilson

W. J.

(1987). The truly disadvantaged: The inner city, the underclass, and public policy. University of Chicago Press.

41.

Wishman

(2013). Confessions of a criminal lawyer: A memoir. https://openroadmedia.com (Original work published 1982).

42.

Wortley

(2001). A classification of techniques for controlling situational precipitators of crime. Security Journal, 14(4), 63–82.

43.

Wortley

(2016). Situational precipitators of crime. In Wortley

Townsley

(Eds.), Environmental criminology and crime analysis (2nd ed., pp. 70–91). Routledge.