LIFE WITHOUT PAROLE SENTENCING IN NORTH CAROLINA

Brandon L. Garrett,* Travis M. Seale-Carlisle,** Karima Modjadidi,***

and Kristen M. Renberg****

Duke University School of Law

210 Science Drive
Durham, NC 27708
(919) 613-7090
bgarrett@law.duke.edu

September 22, 2020

Forthcoming North Carolina Law Review

© Brandon L. Garrett, 2020

LIFE WITHOUT PAROLE SENTENCING IN NORTH CAROLINA

*
*

*
*
*
* L. Neil Williams, Jr. Professor of Law, Duke University School of Law and Director, Wilson Center for Science
and Justice at Duke University School of Law.

** Post-doctoral fellow, Wilson Center for Science and Justice, Duke University School of Law.

*** Research Triangle Institute.

**** J.D. Candidate and PhD. in Political Science, Duke University.

*

*
*
We thank for their comments on earlier drafts, Sara Sun Beale, Emily Coward, Ben Finholt, Catherine Grosso, Ben
Grunwald, Carissa Hessick, Alex Jakubow, Eisha Jain, Ashley Nellis, and Barbara O’Brien, as well as participants
at works-in-progress breakfast at Duke Law organized by the Center for Science and Justice. Many thanks to James
Markham for assistance in locating North Carolina Department of Corrections data, to Duke Law Digital Resources
Librarian Sean Chen for his invaluable work scraping and formatting these data, and to North Carolina Prisoner’s
Legal Services for sharing their data concerning juvenile life without parole sentencing. We thank Barbara O’Brien
and Catherine Grosso for sharing data concerning death sentencing and life without parole sentencing in North
Carolina. Many thanks to Alex Jakubow for collecting and sharing data concerning prosecutor elections and
assisting with methods. Payton Bock, Madison Mumma, Andrew Smith, and Emma Studdard provided invaluable
research assistance. Empirical research presented in this Article was supported by a grant from the Charles Koch
Foundation and the Wilson Center for Science and Justice at Duke Law. The views expressed are those of the
authors and do not necessarily represent those of the funders.
*
*
 ABSTRACT

What explains the puzzle of life without parole (LWOP) sentencing in the United States? In the
past two decades, LWOP sentences have reached record highs, with over 50,000 prisoners
serving LWOP. Yet during this same period, homicide rates have steadily declined. The U.S.
Supreme Court has limited the use of juvenile LWOP in Eighth Amendment rulings. Further,
death sentences have steeply declined, reaching record lows. Although research has examined
drivers of incarceration patterns for certain sentences, there has been little research on LWOP
imposition. To shed light on what might explain the sudden rise of LWOP, we examine
characteristics of the more than 1,627 cases in which LWOP was imposed from 1995 to 2017, in
North Carolina, one of the states that imposes the largest numbers of these sentences. We begin
by analyzing defendant race, crime, and sentence patterns by county. We associate LWOP with
homicide rates, and examine interactions between homicide, victim race, and prior LWOP
sentencing. This first empirical analysis of adult LWOP sentences finds important local
variations in its imposition. We find that as the homicide rate increases within a county, we
observe fewer LWOP sentences. We find that fewer LWOP sentences are predicted to occur as
the number of black victim homicides increase in a county, but no such relationship is found
when considering the number of white victim homicides. Finally, we find a strong path
dependency and concentration of LWOP sentences in counties, where counties that have
imposed LWOP sentences in the past are more likely to continue to do so. These findings have
implications for efforts to reconsider the most severe sentences in the U.S., and they suggest that
prosecutorial discretion in seeking long sentences will be important subjects for future research
and policy.

TABLE OF CONTENTS

INTRODUCTION 1

I. The Rise of LWOP Sentencing 3

A. Theory and Policy Concerning LWOP 4

2
 B. Adoption of LWOP Statutes 5
C. Empirical Literature on LWOP 8
D. North Carolina Adoption of LWOP 11

II. Analysis of North Carolina LWOP Data, 1994-2018 15

A. Sources of Data 15
B. County-Level Analysis 16
1. Homicide Rates 23
2. Race and LWOP Sentencing 24
3. Population Density
4. Inertia
5. Death Sentencing

III. Implications 29
A. Prosecutors, Race, and Crime Rates 29
B. Constitutional Implications 31
C. Death Sentencing and Life Sentencing 33

CONCLUSION 34

APPENDIX A: Prosecutorial Districts 36

APPENDIX B: LWOP Sentences in North Carolina 40

3
 LIFE WITHOUT PAROLE SENTENCING IN NORTH CAROLINA

INTRODUCTION

Life without parole (LWOP) sentencing confounds the broad trends in both crime and
sentencing in the United States. During the last two decades, homicides have steadily declined. 1
Further, death sentences have reached record lows. 2 However, LWOP sentences have reached
record highs.3 We have never had more than 4,000 people on death row at a given time in this
country, and after two decades of steady decline, there are currently 2,700 people on death row. 4
Yet over 50,000 inmates are currently serving LWOP sentences, with the numbers steadily
rising.5 Why has this happened? This Article is the first to explore case-level LWOP sentencing
patterns to address this question. We focus on North Carolina, one of the leading LWOP-
sentencing states, where LWOP is a mandatory sentence for first-degree murder convictions. 6
Our findings suggest that in the shadow of the declining death penalty, LWOP has emerged as a
far more common, easily-imposed, and pervasive form of punishment, and yet it suffers from
distinct racial biases and prosecution incentives.7
Today, policymakers and the public increasingly have reconsidered criminal sentencing
practices in the United States, such as mandatory minimums, and drug sentencing—but have not
done so with LWOP.8 Incarceration in the United States has only slightly receded from record
highs, at about 1.5 million people incarcerated. 9 As the National Academy of Sciences put it in
1
BRANDON L. GARRETT, END OF ITS ROPE: HOW KILLING THE DEATH PENALTY CAN REVIVE CRIMINAL JUSTICE
86 (2017) (“The best available data show a stunning decline in homicides in the early 1990s and continuing for more
than two decades . . . .”).
2
Id. at 9–10, 97 (describing record lows in death sentencing, the forces explaining that decline, and the rise in
LWOP sentencing accompanying these trends).
3
ASHLEY NELLIS, THE SENT’G PROJECT, STILL LIFE: AMERICA’S INCREASING USE OF LIFE AND LONG-TERM
SENTENCES 5 (2017) [hereinafter NELLIS, STILL LIFE], https://www.sentencingproject.org/publications/still-life-
americas-increasing-use-life-long-term-sentences/ [https://perma.cc/6WB4-K8TL] (describing rise in LWOP
sentencing); see also MARC MAUER & ASHLEY NELLIS, THE MEANING OF LIFE: THE CASE FOR ABOLISHING LIFE
SENTENCES 15–16 (2018) (noting LWOP sentencing trends over time and presenting updated data).
4
Size of Death Row, DEATH PENALTY INFO. CTR., https://deathpenaltyinfo.org/death-row/overview/size-of-death-
row-by-year [https://perma.cc/Y7NP-7JPE].
5
NELLIS, STILL LIFE, supra note 3, at 5 (describing results of corrections survey documenting over 53,000 serving
LWOP as of 2016, as well as over 44,000 serving “virtual life sentences” of 50 years or more, and 162,000 serving
life sentences).
6
Id. For an overview of North Carolina statutory framework, see infra Section I.B.
7
Our data is publicly available. See North Carolina Life Without Parole, CTR. FOR OPEN SCI.,
https://osf.io/m7gk8/files/ (providing the data used in this study).
8
See, e.g., PEW CHARITABLE TRS., 35 STATES REFORM CRIMINAL JUSTICE POLICIES THROUGH JUSTICE
REINVESTMENT (2018), http://www.pewtrusts.org/-/media/assets/2018/07/pspp_reform_matrix.pdf
[https://perma.cc/H9M5-QZKG] (describing Justice Reinvestment approach towards reducing reliance on
incarceration and decline in incarceration in those states); NAT’L RSCH. COUNCIL., T HE GROWTH OF
INCARCERATION IN THE UNITED STATES: E XPLORING CAUSES AND CONSEQUENCES 2 (Jeremy Travis, Bruce
Western, & Steve Redburn eds., 2014) (“Between 2006 and 2011, more than half the states reduced their prison
populations, and in 10 states the number of people incarcerated fell by 10 percent or more.”); see also John
Monahan & Jennifer L. Skeem, Risk Redux: The Resurgence of Risk Assessment in Criminal Sanctioning, 26 FED.
SENT’G REP. 158, 158 (2014).
9
See, e.g., DANIELLE KAEBLE & MARY COWHIG, BUREAU OF JUST. STAT., U.S. DEP’T OF JUST., CORRECTIONAL
POPULATIONS IN THE UNITED STATES, 2016, at 2 tbl.1 (2018), https://www.bjs.gov/content/pub/pdf/cpus16.pdf
[https://perma.cc/GF3X-2Q4C] (describing decline of about six percent from 1.6 million to 1.5 million prisoners in
the U.S. from 2009 to 2016).
its report on the “historically unprecedented and internationally unique” growth of incarceration:
the best explanation for this rise is not crime rates, “but the policy choices made by legislators to
greatly increase the use of imprisonment as a response to crime.” 10 Longer prison sentences,
including LWOP sentences, are important drivers of incarceration.11
Before the 1970s, LWOP sentences did not exist in the United States; life sentences
included the possibility of parole after a term of years. 12 LWOP statutes were enacted, largely
beginning in the 1970s, in response to the concern that there be a certain “assurance to juries and
victims’ family members that perpetrators will not be set free” as well as in response to
constitutional challenges to death sentencing statutes.13 In many of these states, LWOP was
adopted as a way to ensure true “life” sentences, as the focus of state sentencing reforms shifted
to retribution, or punishment based on moral culpability and just deserts. 14 By the 1990s, a wave
of states adopted “truth-in-sentencing” legislation that abolished or limited parole more generally
for all sentences.15 Today, all states except Alaska have adopted LWOP. 16
The number of persons serving LWOP sentences is growing, with a current population of
over 50,000 persons serving LWOP.17 In addition to those serving LWOP, others still serve life
sentences, including those who are serving sentences so long that parole is not possible in their
lifetime. Although, about one in nine people currently serving prison time—over 160,000
prisoners—is serving a life sentence.18 Ashley Nellis at the Sentencing Project has conducted a
series of surveys of the LWOP population in the U.S. and has found that most of these tens of
thousands of prisoners who are serving life were convicted of murder; sixty percent of those
sentenced to LWOP were convicted of first-degree murder. 19 But many (over 15,000) lifers were
convicted of nonviolent crimes, like property offenses or drug offenses, and others (over 30,000)
were convicted of non-homicide violent crimes such as sexual assault, robbery, or kidnapping. 20
Moreover, for homicide cases, death sentencing has reached record lows in the U.S., while at the

10
NAT’L RSCH. COUNCIL., supra note 7, at 2–3.
11
Id. at 344–45 (listing other factors including truth-in-sentencing laws, three-strikes enhancements, and parole or
probation revocation).
12
See GARRETT, END OF ITS ROPE, supra note 1, at 95–96 (“Before the 1970s there were no true ‘life’ sentences for
crimes, since lawmakers adopted the view that all prisoners should be redeemable. Thus, in practice, a ‘life’
sentence usually meant that after ten or fifteen years parole was at least a possibility.”).
13
Richard C. Dieter, The Future of the Death Penalty in the United States, 49 U. RICH. L. REV. 921, 924–25 (2015).
14
See GARRETT, END OF ITS ROPE, supra note 1, at 96 (“By 1990 thirty-three states and the District of Columbia had
adopted LWOP. By 2012 all the remaining states had done so except Alaska . . . .”); see also MICHAEL TONRY,
Introduction to WHY PUNISH? HOW MUCH? A READER ON PUNISHMENT 3, 6–7 (Michael Tonry ed., 2011); Michele
Cotton, Back with a Vengeance: The Resilience of Retribution as an Articulated Purpose of Criminal Punishment ,
37 AM. CRIM. L. REV. 1313, 1315–16 (2000; Note, A Matter of Life and Death: The Effect of Life-Without-Parole
Statutes on Capital Punishment, 119 HARV. L. REV. 1838, 1839 (2006).);
15
See Thomas Davidson, Year That States Adopted Life Without Parole (LWOP) Sentencing, DEATH PENALTY INFO.
CTR. (Aug. 2, 2010), https://deathpenaltyinfo.org/stories/year-that-states-adopted-life-without-parole-lwop-
sentencing [https://perma.cc/H52Y-X82Q] (providing a national summary of statute adoption); PAULA M. DITTON
& DORIS J. WILSON, BUREAU OF JUST. STAT., U.S. DEP’T OF JUST., TRUTH IN SENTENCING IN STATE PRISONS 2–3
(1999), https://bjs.gov/content/pub/pdf/tssp.pdf [https://perma.cc/6ZDL-JFVX].
16
See Davidson, supra note 14; see also Ankur Desai & Brandon L. Garrett, The State of the Death Penalty, 94
NOTRE DAME L. R EV. 1255, 1299–1300 tbl.4 (2018) (providing statutory citations and year of LWOP adoption).
17
NELLIS, STILL LIFE, supra note 3, at 9 (describing over 53,000 prisoners serving LWOP as of 2016, based on
corrections survey of state and federal prisons).
18
Id. at 11 fig.4 (depicting life-sentenced prisoners as a percentage of all prisoners).
19
Id. at 12 tbl.3 (displaying data concerning crimes of conviction for persons convicted of LWOP, life, and virtual
life sentences).
20
Id.

2
same time, LWOP sentencing has reached record highs. 21 Thus, the incarcerated population itself
increasingly consists of people who by statute can never be released, absent clemency or
pardon.22
While research has increasingly documented the rise in LWOP sentences in the U.S., this
Article is the first to empirically analyze case and local-level data to examine the rise in such
sentencing. While LWOP is available in every state except Alaska, LWOP sentences are
concentrated in a subset of states.23 Researchers have carefully documented death sentencing
patterns and collected county-level data permitting analysis of sentencing patterns in the context
of country level crime and demographic information.24 What has not been studied are the
geographic differences at the county level for LWOP sentencing. In this Article, we examine the
rise in LWOP sentencing by conducting novel case-level and county-level analysis of LWOP
sentencing in North Carolina.
In Part I, we describe when and how states adopted LWOP sentencing schemes. We
describe the theory and rationales for adopting LWOP sentencing. We summarize prior empirical
research on LWOP sentences. We then describe the adoption of LWOP in North Carolina and
explain how the statute operates to mandate LWOP for first-degree homicide sentences.
In Part II, we set out our findings. This study builds on a prior piece that examines
individual and county-level data for a much smaller population of juvenile LWOP sentences. 25
That study examined the cases of ninety-four people sentenced to LWOP as juveniles in North
Carolina, noting that juvenile LWOP sentences have sharply declined, and almost half of the
sentences have been reversed on appeal.26 However, homicide rates were not predictive of
LWOP sentences, while county-level inertia effects (or the effect of past LWOP sentences)
were.27
This Article focuses on North Carolina case-level data concerning 1,627 people serving
adult LWOP sentences, having been sentenced from 1995-2017, and federal data concerning
21
See GARRETT, END OF ITS ROPE, supra note 1, at 98–100 (noting that not only has LWOP sentencing increased as
death sentencing has declined, but finding that availability of LWOP as an alternative does not provide a strong
explanation for the decline in death sentencing).
22
NELLIS, STILL LIFE, supra note 3, at 7 fig.1 (showing growing share of prisoners serving life sentences).
Regarding the role of executive clemency, see, e.g., Michael A.G. Korengold et al., And Justice for Few: The
Collapse of the Capital Clemency System in the United States, 20 HAMLINE L. REV. 349, 355–57 (1996), for a
discussion on the declining use of clemency in capital cases.
23
ASHLEY NELLIS, THE SENT’G PROJECT, LIFE GOES ON: THE HISTORIC RISE IN LIFE SENTENCES IN AMERICA 5,
25–30 (2013), https://sentencingproject.org/wp-content/uploads/2015/12/Life-Goes-On.pdf [https://perma.cc/VB9Z-
ALWJ] (finding five states—Florida, Pennsylvania, Louisiana, California, and Michigan—account for over half of
all LWOP sentences nationwide); NELLIS, STILL LIFE, supra note 3, at 7 tbl.1.
24
See GARRETT, END OF ITS ROPE, supra note 1, at 139–42. For more information on data concerning death
sentences in the United States from 1991 to 2019, see Data on Death Sentencing, END OF ITS ROPE,
https://endofitsrope.com/ [https://perma.cc/WR24-3WES]. See also David McCord & Talia Roitberg Harmon,
Lethal Rejection: An Empirical Analysis of the Astonishing Plunge in Death Sentences in the United States from
Their Post-Furman Peak, 81 ALB. L. REV. 1, 1–4 (2018) (comparing patterns in 1990s and subsequent death
sentencing); Brandon L. Garrett, Alexander Jakubow, & Ankur Desai, The American Death Penalty Decline, 107 J.
CRIM. L. & CRIMINOLOGY 561. 565–66 (2017) (presenting county-level analyses); Lee Kovarsky, Muscle Memory
and the Local Concentration of Capital Punishment, 66 DUKE L.J. 259, 264 (2016) (analyzing concentration of
death sentences at the county level); Robert J. Smith, The Geography of the Death Penalty and Its Ramifications, 92
B.U. L. REV. 227, 230–34 (2012) (studying local level imposition of death sentencing).
25
Ben Finholt, Brandon L. Garrett, Karima Modjadidi, & Kristen Renberg, Juvenile Life Without Parole in North
Carolina, 110 J. CRIM. L. AND CRIMINOLOGY 141, 141–42 (2020).
26
27
Id. at 165–66.

3
county-level demographics and homicide rates.28 Thus, we look at path dependency and LWOP
sentencing, for the first time in the non-capital sentencing literature. 29 We ran regressions to see
if a county’s homicide rate, population density, poverty rate, the Black population proportion,
and number of death penalty sentences were related to the number of LWOP sentences in that
county. We examined the effects of the racial fragmentation of a county and separated the White
and Black homicide rates in these regressions.
We find, in short, that homicide rates do not explain LWOP sentences. In fact, counties
with higher homicide rates have fewer LWOP sentences. However, we troublingly find that
counties with more Black victims of homicide do have statistically significantly fewer LWOP
sentences, and that this is not the case for counties with more White victims of homicides. This
race-of-victim effect is consistent with research on death sentencing patterns. 30 Second, we
investigate if there is a relationship between a county using an LWOP sentence in the past and
using it in the future. We find strong inertia or muscle memory effects, consistent across our
models. Further, we find much stronger effects when one looks at prosecution districts and
LWOP sentencing over time, as opposed to looking at individual counties. This provides
stronger evidence that it is not other county-level trends, but rather the preferences of
prosecutors, that are driving LWOP sentencing.
In Part III, we conclude by describing the implications of our findings for understanding
prosecution incentives and behavior and for future efforts to improve sentencing policy. We
conclude that the geographic disparities, race-of-victim effects, and inertia effects observed all
show a lack of uniformity and concerns of bias in the use of LWOP. These results suggest
reasons to be concerned with the use of LWOP for adults; at the same time, constitutional and
legal challenges based on these empirical findings are not likely to be successful. Further
research should investigate whether similar patterns exist in other states, and why there is
variation in state use of LWOP. In addition, these findings can inform policy efforts directed at
reconsidering severe sentencing and improving uniformity in criminal sentencing.

I. Understanding THE RISE OF LWOP

Why has the use of LWOP skyrocketed in the United States, especially during a time of
declining crime, and homicide rates? In this Part, we first develop national trends in LWOP
sentences, which have grown a great deal in the past two decades. We describe the rise in the
adoption of LWOP as a sentence, largely since the 1970s, in all states except Alaska. Second, we
summarize the theory and policy rationales and the debates concerning those rationales, for
LWOP sentencing. Third, we describe the empirical literature regarding LWOP, including
surveys of the LWOP population, research on juvenile LWOP, and recidivism research. Finally,
we describe the background regarding the adoption of LWOP in North Carolina in 1994, as part
of a 1990s move towards ending parole in the federal government and many states.

A. Theory and Policy Concerning LWOP Sentencing

28
The sources of these data are described infra Section II.A. We also look at prosecution districts, which often
include several counties in North Carolina. See discussion infra Appendix A.
29
For work looking at this phenomenon in death sentencing, see, e.g., the studies cited in GARRETT, END OF ITS
ROPE, supra note 1, at 98–100.
30
See infra Sections III.A and III.B.

4
 Proponents of LWOP argue that the sentence fulfills the goals of incapacitation,
deterrence, and retribution or “just deserts.”31 However, critics believe LWOP either fails or is
misguided in attaining each of those possible goals.
First, regarding the goal of incapacitation, LWOP ensures permanent incapacitation.
Incapacitation can be effective in preventing additional crimes by the individual that is serving
the punishment, at least in society (but not necessarily in prison). However, incapacitation also
relies on potentially faulty predictions and assumptions. For example, Paul Robinson states, “for
incapacitation to be effective as a distributive principle, one must be able to identify persons who
will commit offenses in the future, preferably with a minimum of ‘false positives’ (persons
predicted to be dangerous who in fact would not commit an offense.)” 32 However, the justice
system relies on prior convictions as a measurement to predict future criminality, . 33 This is a
highly overinclusive measure since the data suggests that criminality is highly correlated with
gender and age, with 25 years of age considered the peak of one’s criminal career.34
Second, regarding deterrence, effective deterrence requires, first, in order to deter, that
people be aware of the rule, and when it applies. 35 Second, even if people are aware of the rule,
deterrence is only effective if people are able to determine what actions are in their best
interests.36 Lastly, even if the first two conditions are met, deterrence will only be effective if
people conclude that the cost of being caught exceeds the potential benefits. 37 All three
prerequisites for effective deterrence are troublesome but the first requisite, that the potential
offender to be cognizant of the law, is especially problematic. LWOP statutes vary among
jurisdictions, eligibility can depend on broad, vague, or complex criteria, and the likelihood that
people know their own state’s law is unlikely. This is especially true of those who may receive
LWOP for a non-violent crime due to a habitual offender statute. Potential offenders cannot be
deterred from an action if they do not know the consequences.
A third justification for LWOP is fitting the sentence to the crime or giving the offender
their just deserts. However, many people serving LWOP sentences did not commit the most
severe offenses. As previously stated, habitual offender statutes make it possible for a person to
receive LWOP even if none of their crimes are violent. 38 Felony murder charges can result in
LWOP even if the person is not the one who committed the crime. 39 For example, in a North
Carolina case, Sethy Seam was sentenced to LWOP as a 16-year-old under the felony murder
31
Paul H. Robinson, Life Without Parole under Modern Theories of Punishment, in LIFE WITHOUT PAROLE:
AMERICA’S NEW DEATH PENALTY? 138, 138 (Charles J. Ogletree, Jr. & Austin Sarat, eds. 2012).
32
Id. at 142.
33
Id. at 142–43.
34
Jeffery T. Ulmer & Darrell Steffensmeier, The Age and Crime Relationship: Social Variation, Social
Explanations, in THE NURTURE VERSUS BIO-SOCIAL DEBATE IN CRIMINOLOGY: ON THE ORIGINS OF CRIMINAL
BEHAVIOR AND CRIMINALITY 377, 377 (Kevin M. Beaver, J.C. Barnes & Brian B. Boutwell, eds., 2014) (“The
relationship between aging and criminal activity has been noted since the beginnings of criminology. . .  Today, the
peak age-crime involvement (the age group with the highest age-specific arrest rate) is younger than 25 for all
crimes reported in the FBI’s UCR program except gambling, and rates begin to decline in the late teenage years for
more than half of the UCR crimes.”).
35
Id. at 140-42.
36
Id.
37
Id.
38
JENNIFER TURNER, ACLU FOUND., A LIVING DEATH: LIFE WITHOUT PAROLE FOR NONVIOLENT OFFENSES 98
(2013), https://www.aclu.org/sites/default/files/field_document/111813-lwop-complete-report.pdf
[https://perma.cc/7MGN-A26P].
39
See Finholt et al., supra note 24, at 154–55. More recently, North Carolina enacted legislation barring felony
murder juvenile LWOP sentences. N.C. GEN. STAT. §§ 14-17(a), 15A-1340.19B (2013)

5
statute.40 Seam was in the car while his friend robbed a convenience store and shot the clerk three
times leading to his death.41 Seam claimed he did not even know that his friend had a gun and
never entered the store, yet he received LWOP (his co-defendant took a plea offer and received a
lesser sentence).42 Drug offenses invoke a similar problem: while they may be serious, rarely is a
drug offense as severe a crime as murder. Yet, under some drug sentencing statutes, people can
receive LWOP.43 Such cases call into question the just deserts justification for LWOP.44
Indeed, the international view is that there is no retributive or other justification for a
LWOP sentencing practice, and that, to the contrary, such sentences violate human rights. Most
industrialized countries have abolished LWOP sentencing as contrary to human rights
obligations. The European Court of Human Rights banned LWOP sentences for countries under
its province.45 In its decision, the Court interpreted LWOP as inhumane and in conflict with the
goals of their justice system.46 The Court also noted that rehabilitation was constitutionally
required for any “community that established human dignity as its centerpiece.” 47 The
International Criminal Court, which tries cases such as genocide, war crimes, and crimes against
humanity, does not have an LWOP option and even the harshest sentences require review after
twenty-five years.48 Although people in the United States may see LWOP as a more lenient
sentence than the death penalty, the international community considers the punishment a human
rights violation.

B. Adoption of LWOP Statutes

Before the 1970s, the most severe term-of-years sentences imposed were life sentences. 49
Life sentences were imposed through an indeterminate sentence with fixed minimum years of
imprisonment, or in a mandatory form without any minimum term fixed. 50 In both situations, the
parole board could permit prisoners parole release and there was a set minimum term after which
the parole authority could consider early release.51 In 1970, only seven states prohibited parole
eligibility for life sentences (Massachusetts, Michigan, Mississippi, Nebraska, Nevada,
Pennsylvania, and West Virginia).52
In the 1970s, a period in which the constitutional status of capital punishment was in flux,
LWOP emerged as an alternative to the death penalty. In 1972, the Supreme Court ruled that the
application of the death penalty, as it then was being applied, was unconstitutional. 53 Four years
later, the Supreme Court ruled that if states satisfied constitutional requirements, then they could
40
See Finholt et al., supra note 24, at 154–56.
41
Id. at 155.
42
See id. at 154–55. .
43
See NELLIS, STILL LIFE, supra note 3, at 13.
44
See id.
45
Joseph Tutro, Eliminating the Effective Death Sentence of Life Without Parole, 1 F.: TENN. STUDENT LEGAL J. 11,
18–19 (2014).
46
Id. at 24–25.
47
Vinter v. United Kingdom, Eur. Ct. H.R. 41 (2013); Tutro, supra note 45, at 20.
48
Marc Mauer, Long-Term Sentences: Time to Reconsider the Scale of Punishment, 87 UMKC L. REV. 113, 128
(2018).
49
MARC MAUER, RYAN KING, & MALCOLM YOUNG, THE SENT’G PROJECT, THE MEANING OF “LIFE”: LONG
PRISON SENTENCES IN CONTEXT 4 (2004), https://www.sentencingproject.org/publications/the-meaning-of-life-long-
prison-sentences-in-context/ [https://perma.cc/LR2P-HR4G].
50
Id.
51
See id.
52
Id. at 5–8.

6
continue to impose death sentences.54 Furthermore, the Supreme Court held that mandatory death
penalty statutes (like the one adopted in North Carolina) were unconstitutional and that states
must have an alternative sentence available.55 The alternative sentence typically selected was
LWOP.56 In addition, some death penalty abolitionists became proponents of LWOP as an
alternative to capital punishment.57 Thus, the status of LWOP as alternative to the death penalty
contributed to its broader adoption and use during this time period.58
By the mid-1990s, a broader driver of LWOP adoption was that states increasingly
rejected parole entirely, for many or all types of sentences. LWOP statutes reflected a new
legislative desire to focus on retribution and just deserts rather than rehabilitation. 59 Although the
death penalty could only be applied in homicide cases, following the Supreme Court’s ruling in
Coker v. Georgia,60 LWOP did not have such constraints.61 Many non-homicide felonies such as
kidnapping, armed robbery, and sexual battery could result in an LWOP sentence.62 Additionally,
New York’s Rockefeller Drug Laws enabled the state to sentence serious drug offenders to life
sentences.63 Michigan’s “650-Lifer Law” mandated LWOP sentences for sale, manufacture, or
possession of 650 grams of cocaine or heroin.64 Three-strikes laws were adopted under the theory
that if a person commits a crime three times, they are unable to be rehabilitated. 65 In 1994,
Congress passed a version of three-strikes laws in the Violent Crime Control and Law
Enforcement Act.66 The federal three-strikes law mandated LWOP for people convicted of a
federal offense if they had two prior offenses that resulted in state or federal convictions. 67 States
quickly followed suit and currently all 50 states and the federal government use enhanced
sentences for habitual offenders, and of those, thirty states and the federal government use
LWOP as the enhanced punishment.68
By 2014, all American jurisdictions except for Alaska adopted LWOP as a sentence. 69
The structure of these statutes varies. For example, six states (Maine, Pennsylvania, South

53
Furman v. Georgia, 408 U.S. 238, 239–40 (1972) (reversing death penalty sentences as required by state statutes).
Soon after, Georgia legislature enacted law consistent with the Supreme Court ruling. GA. CODE ANN. § 26-1101(a)
(1972).
54
Gregg v. Georgia, 428 U.S. 153, 169, 206–07 (1976).
55
Woodson v. North Carolina, 428 U.S. 280, 304–05 (1976).
56
Garrett, End of its Rope, supra note 1, at 96-97.
57
Id. at 96; see also Craig S. Lerner, Who’s Really Sentenced to Life Without Parole?: Searching for “Ugly
Disproportionalities” in the American Criminal Justice System, 2015 Wis. L. Rev. 789, 796.
58
David McCord, What’s Messing with Texas Death Sentences?, 43 TEX. TECH L. REV. 601, 611–12 (2011)
(discussing speculation that LWOP might explain the Texas decline and rejecting that explanation).
59
See supra notes 13 and 30.
60
433 U.S. 584 (1977).
61
See id. at 592.
62
NELLIS, STILL LIFE, supra note 3, at 21; Jing Cao, Commuting Life Without Parole Sentences: The Need for
Reason and Justice Over Politics(2015) (S.J.D. Dissertations, Fordham University School of Law).
63
Act of May 8, 1973, ch. 276 §§ 220.21, 220.43, 1973 N.Y. Laws 1040, 1052–54 (codified as amended in N.Y.
Penal Law). For additional background on Rockefeller Drug Laws, see generally Peter A. Mancuso, Resentencing
After the “Fall” of Rockefeller: The Failure of the Drug Law Reform Act of 2004 and 2005 to Remedy the Injustices
of the New York Rockefeller Drug Laws and the Compromise of 2009, 73 ALBANY L. REV. 1535 (2010).
64
Cao, supra note 61, at 11.
65
Id. at 12.
66
Pub. L. No. 103-322, 108 Stat. 1796 (1994) (codified as amended at 18 U.S.C. § 3559).
67
18 U.S.C. 3559(c)(1) (1994).
68
TURNER, supra note 38, at 35–36.
69
Lerner, supra note 56, at 797. New Mexico’s stance on LWOP is unclear, but its law at least recognizes the
possibility of LWOP as a sentence. Id. at 797 n.37.

7
Dakota, Illinois, Iowa, and Louisiana) and the federal system have an LWOP statute but no life
with parole alternative; in other words, parole is not available to anyone serving a life sentence. 70
Other states, as noted, require LWOP sentences for certain crimes, but require or permit life
sentences with parole for others, including through repeat or habitual offender provisions.71
We note that in contrast to adult LWOP, juvenile LWOP has been impacted by
constitutional rulings. The U.S. Supreme Court’s ruling in Miller v. Alabama72 forbade
mandatory LWOP for juvenile homicide offenses and mandated that sentencing judges consider
such offenders’ “youth and attendant characteristics” before imposing “the harshest possible
penalty.”73 Following the Miller ruling,74 North Carolina lawmakers enacted a new statute
requiring sentencing court to consider “all the circumstances of the offense” as well as the
“particular circumstances of the defendant,” and “any mitigating factors.”75 Further, the North
Carolina Supreme Court ruled that the statute creates no presumption in favor of LWOP. 76
Factfinders should select a sentence “in light of the United States Supreme Court’s statements in
Miller and its progeny” which state that LWOP sentences “should be reserved for those juvenile
defendants whose crimes reflect irreparable corruption rather than transient immaturity.” 77 In
addition, North Carolina lawmakers in 2013 removed juvenile LWOP for felony murder.78

C. Empirical Research on LWOP Sentencing

Nationwide, the number of prisoners serving LWOP has steadily grown over the past two
decades.79 The rise in LWOP during that time period has increased four times as quickly as
indeterminate life sentences (sentences with a maximum of life imprisonment); LWOP sentences

70
ASHLEY NELLIS & RYAN S. KING, THE SENT’G PROJECT, NO EXIT: THE EXPANDING USE OF LIFE SENTENCES IN
AMERICA 4 (2009), https://www.sentencingproject.org/wp-content/uploads/2016/01/No-Exit-The-Expanding-Use-
of-Life-Sentences-in-America.pdf [https://perma.cc/28SU-XY6G].
71
See id.
72
567 U.S. 460 (2012).
73
Id. at 469, 479, 483, 489.
74
The resulting statute was titled “An Act To Amend The State Sentencing Laws To Comply With The United
States Supreme Court Decision In Miller v. Alabama.” 2012 N.C. Sess. Laws 713 (codified as amended at N.C.
GEN. STAT. §§ 15A-1476 to 1479). The prior statute made LWOP sentences mandatory. N.C. GEN. STAT. § 14–17
(2009) (“[A]ny person who commits [murder in the first degree] shall be punished with death or imprisonment in the
State’s prison for life without parole as the court shall determine pursuant to [N.C. GEN. STAT.] [§] 15A-2000,
except that any such person who was under 18 years of age at the time of the murder shall be punished with
imprisonment in the State’s prison for life without parole”); N.C. GEN. STAT. § 14-17(a) (2013) (removing the
LWOP requirement for persons under 18 years of age); § 15A-1340.19B (2013) (“If the sole basis for conviction of
a count or each count of first degree murder was the felony murder rule, then the court shall sentence the defendant
to life imprisonment with parole.”).
75
N.C. GEN. STAT. § 15A-1340.19C (2020). The mitigating factors to be considered in sentencing include: (1) the
offender’s age at the time of offense; (2) immaturity; (3) ability to appreciate the risks and consequences of the
conduct; (4) intellectual capacity; (5) prior record; (6) mental health; (7) familial or peer pressure exerted upon him;
(8) likelihood that he would benefit from rehabilitation in confinement; and (9) other mitigating factors and
circumstances. § 15A-1340.19B.
76
State v. James, 371 N.C. 77, 89, 813 S.E.2d 195, 204 (2018).
77
Id. at 93–94, 813 S.E.2d at 207.
78
N.C. GEN. STAT. § 14-17(a) (2013) (providing that any person who commits first degree murder shall be punished
with death or LWOP unless the person was under eighteen years of age at the time of the murder); § 15A-1340.19B
(2013) (“If the sole basis for conviction of a count or each count of first-degree murder was the felony murder rule,
then the court shall sentence the defendant to life imprisonment with parole.”).
79
See NELLIS, STILL LIFE, supra note 3, at 19.

8
increased by 59% whereas indeterminate life sentences increased by 17.8%.80 Interestingly,
“while serious crimes, including murder have generally declined for the past 25 years nationwide
the number of lifers in prison has continued to rise.”81 If crime has generally declined, then we
must look to a number of contributing factors that explain the growth in LWOP sentences. These
include inflexible statues, misinformed assumptions about recidivism, and rejection of the idea of
rehabilitation.82 The frequency with which LWOP is used also varies state to state. In Louisiana,
10.9% of people in prison are serving LWOP sentences. 83 In Pennsylvania, 9.4% of the prison
population is serving an LWOP sentence.84 However, there is an opposite trend in 15 states
where less than 1% of the prison population is serving LWOP.85
As noted, one rationale for the adoption of LWOP is the assumption that individuals who
commit crimes that are worthy of such a sentence are highly likely to re-offend and are incapable
of rehabilitation.86 However, emerging science indicates that age at the time of offense is one of
the most accurate predictors of criminal behavior. 87 The social and psychosocial developments
that occur during a person’s early 20s make them the most likely to reoffend. 88 Criminal offenses
decline as a person ages into their mid-20s and flattens out by one’s late 30s. 89 These age trends
are true across crime types. The Sentencing Project found that people who were released from
life sentences were less than one-third as likely to be rearrested within three years compared to
all released prisoners.90 Specifically, while two-thirds of people who were released in 1994 were
rearrested within three years, only 20% of people who were released from a life sentence were
rearrested.91
Prior empirical research has examined juvenile LWOP sentencing, including in a prior
study examining such sentences in North Carolina. 92 That study described a rise in such
sentences following adoption of LWOP in 1994, but a decline in more recent years, as well as a
rise in reversals on appeal.93 The study examined the cases of the 94 people in North Carolina
who were sentenced to LWOP as juveniles from 1994 to present. 94 Their ages at the time of the
offense ranged from 13 to 17.95 Of those, forty-eight are currently serving LWOP sentences (one
more is currently pending retrial).96 Several patterns stand out in the juvenile data. First, juvenile
LWOP sentencing has declined markedly since its late-1990s height in North Carolina;

80
Id.
81
Id. at 20.
82
See NELLIS & KING, supra note 69, at 2–4, 36.
83
Id. at 7.
84
Id. at 8.
85
Id. at 7–8.
86
See infra Part I.A.
87
See Ulmer & Steffensmeier, supra note 37, at 393–94.
88
See id. at 378–79.
89
See id. at 389.
90
NELLIS & KING, supra note 69, at 36.
91
Mauer et al., supra note 50, at 24.
92
Finholt et al., supra note 24, at 141.
93
Id. at 157, 163–64.
94
Id. at 146.
95
Id.
96
Id. In contrast, 203 offenders sentenced for crimes committed when 17 or younger are serving life with parole
sentences and 63 are serving terms of over forty years. N. C. DEP’T OF PUB. SAFETY, LIFE AND 40+ YEAR
SENTENCES FOR THOSE SENTENCED WHEN 17 OR YOUNGER (2018). Given prison life expectancies, terms of over
forty years may often consist in de facto or virtual life without parole sentences, given prison life expectancies, if
they are no reconsidered prior to the end of the term.

9
beginning in 2011, there have been either one or no such sentences each year.97 Second, juvenile
LWOP sentences were highly concentrated in a handful of counties. 98 Third, race disparities in
juvenile LWOP sentences mirror race disparities in juvenile homicide offending in North
Carolina.99
There are reasons to think that adult LWOP sentences would not exhibit the same
patterns as found in juvenile LWOP cases, and that instead, the use of adult LWOP would be far
more common, across more counties, but perhaps also reflecting disparities due to the use of
prosecutorial discretion. Both U.S. and North Carolina Supreme Court regulation of juvenile
LWOP sentences has resulted in a decline in its use and in the reversal of many of the prior
sentences imposed.100 Not only are adult LWOP sentences more common, but they have not been
constitutionally regulated.101 There is far more prosecutorial discretion to seek such sentences,
given the broad definition of first-degree murder, and as a result, they might exhibit different
geographic patterns in their use. Further, for adults, the death penalty is available for some first-
degree murders, and there may be more of a connection between adult LWOP sentencing and
death sentencing patterns. We examine each of those questions in this study. 102 First, however,
we turn to the adoption of LWOP in North Carolina.

D. North Carolina Adoption of Life Without Parole

North Carolina adopted LWOP in 1994, as part of the change from the prior sentencing
scheme, termed “Fair Sentencing” (after the Fair Sentencing Act (“FSA”)), to a new scheme
termed “Structured Sentencing.”103 The goals of the Structured Sentencing Act (“SSA”) were to
provide rational, truthful, and consistent sentences.104 The Act applies to most misdemeanor and
felony crimes committed on or after October 1, 1994. 105 Under the SSA, first-degree homicide is
punishable by either by a death sentence or life without parole, but not a term of years.106
Not only are the possible punishment alternatives extremely severe, but at the same time,
first-degree homicide is defined extremely broadly in North Carolina. 107 The statute, in addition
to specific types involving poison, torture, or weapons of mass destruction, defines first-degree
murder as any other kind of “willful, deliberate, and premeditated killing.”108 Further, any murder
that is committed as part of the commission or attempted commission of “arson, rape a or sex
offense, robbery, kidnapping, [or] burglary,” or any “other felony committed or attempted with
97
Finholt et al., supra note 24, at 157.
98
Id. at 160.
99
Id. at 158–59.
100
See id. at 173.
101
See id. at 150.
102
See infra Part II.
103
1993 N.C. Sess. Laws 2299 (codified as amended at N.C. GEN. STAT. § 15A, art. 81B). For four years, until the
provision was repealed in 1998, the North Carolina statute also provided a safety valve in the form of judicial
review of LWOP sentences after twenty-five years of imprisonment; sentences entered during that window will be
eligible for review beginning in 2019. N.C. GEN. STAT. § 15A-1380.5 (repealed 1998).
104
THE N.C. SENT’G AND POL’Y ADVISORY COMM’N, A CITIZEN’S GUIDE TO STRUCTURED SENTENCING 1 (2014),
https://www.nccourts.gov/assets/documents/publications/citizenguide2014.pdf?
QUy2UMcGsNAKtUWMbLQnK004OLlEsYwd [https://perma.cc/HM3F-YE6V].
105
N.C. GEN. STAT. § 15A-1340.10 (LexisNexis current through Session Laws 2020-35).
106
1993 N.C. Sess. Laws 229, 2307–09 (codified at N.C. GEN. STAT. § 15A-1340.17).
107
Compare, e.g. N.Y. PEN. L. § 125.27(1) (defining all first degree murder as having been committed “[w]ith intent
to cause the death of another person”).
108
N.C. GEN. STAT § 14-17(a) (LexisNexis current through Session Laws 2020-35).

10
the use of a deadly weapon,” qualifies as first-degree murder under North Carolina’s felony
murder provision.109 Thus, any murder, regardless of premeditation, that involved the use of a
gun or knife could be charged as a first-degree murder. The statute gives prosecutors wide
discretion, then, to seek LWOP sentences in murder cases.
More broadly, the SSA eliminated parole for all offenders, meaning that inmates
sentenced under the SSA were no longer able to achieve early release through good behavior or
other means and it defined all life sentences as “natural life” sentences with no possibility for
parole.110 The SSA had several consequences, both intended and unintended. One study showed
that in comparison to inmates sentenced under previous sentencing law in North Carolina, those
sentenced under the SSA had higher overall in-prison infraction rates—25% higher for males
and 55% higher for females. 111 This increase may be due to the lack of incentive to be compliant
under determinate sentencing schemes, as there is no early release for good behavior. 112 Another
report found that while the use of prison for violent crimes remained the same under the SSA, the
length of prison terms went up substantially. 113 The prison population in North Carolina,
however, is no longer rising as rapidly as it was pre-FSA. 114 Scholars believe this slowdown is
due to lighter sentences given for non-violent and non-drug related crimes, as well as a reduction
in crime statewide.
As of the end of 2017, 1,627 offenders were serving LWOP sentences in North
Carolina.115 African-Americans make up the majority of the population with 1,005 people,
followed by 496 Whites, and 126 Other.116 There were 1,543 men serving LWOP sentences and
84 women.117 The vast majority of these offenders (1,582) were convicted of first-degree
murder.118 The second most frequent offense is violent habitual felonies (64). 119 The average age
at conviction in North Carolina is thirty-two years old and the current average age of someone
serving an LWOP sentence is forty-three years old. The county in which an offender is convicted
varies and closer analyses are described in Part II.
The overall trend in adult LWOP sentences in North Carolina is shown in Figure 1.
Following adoption of LWOP in 1994, sentencing rose sharply and has remained at a fairly
steady level since 2000. In contrast to the rise in LWOP, Figures 2 and 3 show the steady decline
in the number of death sentences and homicide rates in North Carolina across that same time
period.
109
Id.
110
See 1994 North Carolina Laws 1st Ex. Sess. Ch. 7 (mandating offenders to serve at least 100% of minimum
sentence and 85% of maximum sentence).
111
JAMES J. COLLINS ET AL., EVALUATION OF NORTH CAROLINA’S STRUCTURED SENTENCING LAW 61 (1999). The
report also found age inversely related to infractions—as age increased, the likelihood of involvement in infractions
decreased. Id. at 66–67.
112
See id. at 6; cf. CRIME AND JUSTICE AT THE MILLENNIUM: ESSAYS BY AND IN HONOR OF MARVIN WOLFGANG
285 (Robert A. Silverman et al. eds., 2002).
113
Ronald F. Wright, Counting the Cost of Sentencing in North Carolina, 1980-2000, 29 CRIME & JUST. 29, 87
(2001). For example, the average prison term served during 1993 (the year before the implementation of the SSA)
for personal injury crimes was 21 months. Id. at 88. Under the SSA, the minimum term imposed was 60 to 67
months. Id.
114
In 2016, North Carolina’s rate of incarceration was the 21st lowest in the nation. See E. ANN CARSON, BUREAU
OF JUST. STAT., U.S. DEP’T OF JUST., PRISONERS IN 2016 at 9 tbl.7 (2018).
115
See infra Table 1.
116
See infra Table 1.
117
See infra Table 1.
118
See infra Table 1.
119
See infra Table 1.

11
Figure 1: Adult LWOP Sentences in North Carolina, 1995-2017

Figure 2: Death Sentences in North Carolina, 1995-2017

Figure 3: Homicide Rates in North Carolina, 1995-2017

12
By way of comparison, the trends in juvenile LWOP in North Carolina are markedly different;
juvenile LWOP sentences experienced a similar rise upon adoption in 1994, but have since
sharply declined.120 Moreover, as noted, almost half of the sentences that were imposed have
been reversed on appeal.121
Somewhat similar to juvenile LWOP, death sentencing in North Carolina also exhibits a
very different trend from adult LWOP sentencing. In the 1990s, twenty-five or more death
sentences were imposed in some single-year periods, as shown in Figure 2. Like juvenile LWOP
sentencing though, since 2001, less than five death sentences per year have been imposed in
most years, and in some years, none have been imposed.122
In 2001, a state law created a statewide Indigent Defense Services office to support
defense representation, and granted prosecutors the discretion whether to seek the death
penalty.123 Prior to 2001, prosecutors were required to seek the death penalty in first-degree
homicide cases in which they found death-eligible aggravating circumstances present. However,
after this act was adopted, prosecutors received discretion in capital charging decisions, and
could instead, for example, seek LWOP for an otherwise capitally-eligible first-degree murder
charge.124

II. ANALYSIS OF NORTH CAROLINA LWOP S ENTENCING , 1995-2017

In the sections that follow, we analyze data concerning LWOP sentences in North
Carolina. The first section describes the sources of these data. The next three sections describe
our exploratory analyses and findings.

120
Finholt et al., supra note 24, at 157.
121
Id. at 163.
122
Brandon L. Garrett, The Decline of the Virginia (and American) Death Penalty, 105 GEO. L. J. 661, 720 fig.10
(2017).
123
N.C. GEN. STAT. § 7A-498.2 (2000) (creating Indigent Defense Services) and N.C. GEN. STAT. § 15A-2004
(2001) (providing district attorneys with discretion whether to seek the death penalty).
124
See N.C. GEN. STAT. § 15A-2004 (LexisNexis current through Session Laws 2020-35).

13
 First, we explored whether county-level characteristics predict the frequency of LWOP
sentencing. Specifically, we assessed whether a county’s homicide rate, population density,
poverty rate, the Black population percentage, and the number of death penalty sentences were
related to the number of LWOP sentences in that county. We also explored whether the racial
fragmentation of a county was related to the number of LWOP sentences in that county. 125
Furthermore, we separated the homicide rate by White and Black victim rates by county (the FBI
does not collect data on homicide rates for other racial or ethnic groups) in order to see if the
victim’s race related to the number of LWOP sentences.126
Most inmates serving LWOP have been convicted of first-degree murder, for which an
LWOP sentence is mandatory (if there is not a death sentence). Although the LWOP sentence is
mandatory for first-degree murder, prosecutors have discretion whether to charge that offense
(and it is quite broadly defined, as noted, including with a broad felony murder theory.) For this
reason, we also explored whether county-level characteristics predict the frequency of LWOP
sentencing. We explored whether a prosecutor district’s homicide rate, poverty rate, the
percentage of a district’s Black population, and the number of death penalty sentences were
related to the number of LWOP sentences in that county.
Lastly, we examined the inertia effect in each county. That is, we investigated if there is a
relationship between a county using an LWOP sentence in the past and using it in the future. We
aimed to assess whether there is path dependency in county-level charging patterns over time.

A. Sources of Data

Data concerning life without parole sentences was obtained from the North Carolina
Department of Public Safety website, and verified against data supplied directly from the
Department of Public Safety.127 This allowed us to capture cases in which persons were
sentenced to LWOP in the past, but are not currently serving such a sentence due to a successful
appeal, clemency, or because they are deceased. Data concerning death sentences was obtained
from a prior research collection concerning all death sentences from 1990-2017.128
The homicide rate, measured as the number of homicides per 100,000 inhabitants in each
county-year was provided by the FBI’s Supplemental Homicide Reports. 129 We also include the
county homicide rate for White victims and the homicide rate involving Black victims in
additional analyses.
The percent of the Black population in each county was provided by the U.S. Census
Bureau.130 The density of each county’s population, measured as the number of people per square
mile of land, was also provided by the U.S. Census Bureau, as was the poverty rate of each
125
A racial fragmentation measure reports the probability that two randomly selected individuals in a jurisdiction
belong to different racial groups. Alberto Alesina et al., Fractionalization, 8 J. ECON, GROWTH 155, 156 (2003).
126
ERICA L. SMITH & ALEXIA COOPER, BUREAU OF JUST. STAT., U.S. DEP’T OF JUST., HOMICIDE IN THE U.S.
KNOWN TO LAW ENFORCEMENT, 2011, at 16 (December 2013) (“Due to the lack of reporting of ethnicity by
submitting law enforcement agencies, homicide rates by Hispanic or Latino origin were not calculated.”).
127
See infra Appendix B (summarizing the county level data). Many thanks to Professor James Markham for
assistance in contacting the North Carolina Department of Public Safety, Division of Adult Correction and Juvenile
Justice, and to Duke Law Digital Resources Librarian Sean Chen for his invaluable work scraping and formatting
these data.
128
See Garrett et al., supra note 23, at 616–42 (analyzing death sentence data from the Bureau of Justice Statistics,
U.S. Department of Justice).
129
Easy Access to the FBI’s Supplementary Homicide Reports: 1980–2016, OFF. OF JUV. JUST. & DELINQ.
PREVENTION, https://www.ojjdp.gov/ojstatbb/ezashr/ [https://perma.cc/K7KH-P6TN].

14
county, defined as the percent of families within the county living in poverty, based on the
applicable poverty thresholds.131
In North Carolina, there are currently 43 prosecutor districts, 132 but, depending on the
year, that number fluctuates; since 1990, lawmakers have altered the prosecutor district map
several times. In order to obtain district-level data throughout this time period, data for each
county belonging to a district during a particular year from 1990 to 2017 was aggregated. For
some of the independent variables, this was a straightforward process. For example, the number
of LWOP sentences and the number of death sentences were simply summed across counties for
each district for each year. We describe district-level findings in Appendix A.

B. County and District Characteristics and the Application of LWOP Sentences

A series of maps were created by aggregating all adult LWOP sentences across all NC
counties (Figure 4) during the period from 1995 to 2017. The shade of each county corresponds
to the number of observed LWOP sentences. Here, darker colors represent more LWOP
sentences and lighter colors indicate fewer LWOP sentences.

Figure 4. Number of LWOP Sentences in North Carolina across Counties and Time

130
Datasets, U.S. CENSUS BUREAU, https://www.census.gov/data/datasets.html#.html [https://perma.cc/NTV8-
RLJT].
131
Poverty Thresholds, U.S. CENSUS BUREAU, https://www.census.gov/data/tables/time-series/demo/income-
poverty/historical-poverty-thresholds.html [https://perma.cc/A3TB-QMFC]. In alternative specifications we
included a count of previous death penalty sentences, a one-year lag of the homicide rate, and a count of previous
adult LWOP convictions.
132
For the current districts, see N.C. GEN. STAT. § 7A-60(a) (LexisNexis current through Session Laws 2020-35).

15
16
Note: G is Guilford County, W is Wake County, C is Cumberland County, and M is
Mecklenburg County.

Table 1. Descriptive Profile of Adult LWOP Population in North Carolina, 1995-2017

    Frequency Percentage

Crime First Degree Murder, Second Degree Murder 1,551 96%

Sexual Assault, Rape 14 1%

Violent Habitual Felon 64 3%

 
Race White 496 30%
Black 1,005 62%
Asian 16 <1%
Native American 42 3%
Other 63 5%
Unknown 5 <1%
 
Gender Male 1,543 95%
Female 84 5%

As shown in Table 1, 1,551 of the 1,627 persons who were sentenced to LWOP from
1995 to 2017 in North Carolina, were listed as having committed murder (first or second degree)
as their primary crime.133 Observing the counties which most frequently applied LWOP
sentences between 1995 and 2017 (Cumberland, Guilford, Meckenblerg and Wake counties),
they had on average, a 5.9 homicide rate and an average of 14.8% families living in poverty. By
contrast, the average county in North Carolina had 13.3% of families living in poverty and an
average homicide rate of 5.8 in 2010.

Figure 5. Number of LWOP Sentences in North Carolina in county-year format and

prosecutorial district-year format

133
North Carolina Life Without Parole, CTR. FOR OPEN SCI., https://osf.io/m7gk8/files/ [perma TBD] (providing the
data used in this study).

17
 The data concerning LWOP sentences for each of the 1,627 cases was transformed into a
county-year dyad format in Figure 5A. This transformation allowed us to conduct a series of
statistical analyses in order to determine which county-level characteristics were correlated with
more adult LWOP sentences. The LWOP sentences data was also transformed into a
prosecutorial district-year dyad format in Figure 5B, and those analyses are discussed in the
Appendix. Since there are 100 counties in North Carolina and the LWOP sentencing data covers
23 years (1995 to 2017), there are 2,300 observations under this arrangement. A number of
independent variables related to each county in North Carolina were also merged with our data
on LWOP sentencing.134
The dependent variable in Figure 5A is the frequency of LWOP sentences and measures
the number of LWOP sentences within each county for each year. Around 65% of the time, there
were zero LWOP sentences in a given county for a given year. The dependent variable has a
mean value of 0.71 and a standard deviation of 1.37.
A Poisson regression was chosen as the appropriate regression model for two reasons.
First, the dependent variable is a count variable (0, 1, 2, 3, etc.), which means a discrete
probability distribution such as the Poisson distribution or the negative binomial distribution is
needed to accommodate these data. Second, the Poisson distribution accommodated the data well
and required fewer parameters than other discrete probability distributions such as the negative
binomial distribution. 135 A simple robustness check was also performed by visually inspecting
the raw data. Figure 5 shows the county-year and district-year distributions plotted, collapsed

134
Fixed effects for counties were also included in each model to control for unobserved and heterogenous
relationships within the data. See infra Table 2.
135
The negative binomial distribution is used to model data that are heavily skewed, or, in other words, when the
variance of the distribution far exceeds the mean of the distribution. ALAN AGRESTI, CATEGORICAL DATA ANALYSIS
559–63 (2d ed. 2002). When the data are not heavily skewed, the Poisson distribution is used because it assumes the
mean of the data is approximately equal to the variance in the data. Id. at 663–65. The distribution of county-year
LWOP sentences (Figure 5A) and the distribution of district-year LWOP sentences (Figure 5B) are not so heavily
skewed that a negative binomial model is needed to fit the data.

18
across time. These distributions do not exhibit the extreme skewedness needed in order to reject
the Poisson distribution and explore alternative distributions (such as the negative binomial).
Before running the Poisson regressions, each continuous independent variable was
centered and scaled so that the distribution of values was set to have a mean of 0 with a standard
deviation of 1. This put all of the continuous independent variables onto a comparable scale. The
county-level results of a set of Poisson regressions are presented in Table 2. The district-level
results of a similar set of Poisson regressions are presented in Appendix A. The Poisson
regression estimates a coefficient for each variable (or predictor) in the model. If a variable is
estimated to have a negative coefficient, then increases in that variable predicts fewer LWOP
sentences. If a variable is estimated to have a positive coefficient, then increases in that variable
predicts more LWOP sentences. The extent to which LWOP sentences are predicted to increase
or decrease depends on the value of the coefficient. For example, a large coefficient means that
for every one unit increase in a predictor (and when holding all other predictors in the model
equal), LWOP sentences will increase by that large amount.

Table 2: County Poisson Regression Results for Adult LWOP Sentences (1995-2017)

Model 1 Model 2 Model 3

Predictors 1A 1B 1C 2A 2B 3A 3B

Homicides per 100k^ -0.15*** -0.15** -0.15*** -0.09* -0.11*

Black Victim
-0.17**
Homicides per 100k^

White Victim
-0.05
Homicides per 100k^

% in Poverty^ -0.12* -0.10 -0.09 -0.12* -0.10 -0.14** -0.13**

Population Density^ -0.28*** -0.33*** -0.37*** -0.32*** -0.32*** -0.19*** -0.27***

% Population Black^ -0.65

% Racial
-0.24
Fragmentation^

# LWOP Sentences in
0.05 **
prior year

# Death Sentences in
0.07
prior year

Intercept 0.09 0.12 0.14 -0.01 0.06 0.09 0.14

Observations 2300 2300 2300 2300 2300 2300 2300

19
 County Fixed Effects Yes Yes Yes Yes Yes Yes Yes

AIC 3981.3 3765.1 3775.2 3980.7 3982.6 3840 3847.1

Note: ^ indicates values were centered and scaled prior to running the regression. Values were
scaled so that the distribution of values had a mean equal to zero and a standard deviation equal
to one. AIC for the best fitting model is shown in bold. *p<0.1; **p<0.05; ***p<0.001

1. Homicide Rates

As was shown in Figure 3, since LWOP was adopted in North Carolina, homicide rates
have generally fallen. For example, in 2017, the state reported 637 murders, for a rate of 7 per
100,000 persons;136 in 1994, when the SSA was adopted, there were 759 murders for a rate of
11.2 per 100,000.137 In contrast, as noted in Figure 1, LWOP sentences have remained fairly
constant, since their rise following the adoption of the SSA.
That said, it is important to note that the homicide rate variable supplied by the FBI
includes homicides other than first-degree homicide. As such, even though the homicide rate
may be decreasing in the state or in a county over time, it does not imply the rate of first-degree
homicide is decreasing within the county. For this reason, we cannot more precisely specify the
relationship between homicide rates and LWOP sentencing, nor can we measure culpability.
That said, the homicides that consist of first-degree homicide still involve a degree of
prosecutorial discretion. First-degree homicide is defined as including certain specific types of
murders involving poison, torture, or weapons of mass destruction, but also as any other kind of
“willful, deliberate, and premeditated killing.”138 Those terms are not defined in the statute and
they involve some degree of interpretation by prosecutors and by jurors.
We find, as shown across all models in Table 2, a statistically significant negative
correlation between the homicide rate and adult LWOP sentences. This suggests that as the
homicide rate increases within a county, we should expect to observe fewer LWOP sentences. In
Model 1B, we replace homicide rate with a variable that represents the homicide rate for Black
victims in each county and again we observe a negative relationship. Here, the result implies that
as the homicide rate for Black victims increases, we expect to observe fewer LWOP sentences.
However, when we include the homicide rate for White victims in Model 1C, there is no
statistically significant relationship between this homicide rate and the use of LWOP sentencing.
In Models 2 and 3, there is also a statistically significant negative correlation between the
homicide rate and the number of LWOP sentences.
The finding of a negative correlation between the homicide rate in a county and the
number of LWOP sentences we observe in that county, is perhaps surprising. With LWOP as a
mandatory sentence for first degree homicide, one might expect LWOP sentences to be
positively correlated with the homicide rates (assuming that first-degree homicides are a fairly
consistent proportion of homicides in general). That is, one might expect there to be more LWOP

136
N. C. STATE BUREAU OF INVESTIGATION, CRIME IN NORTH CAROLINA – 2017, at 2 tbl.2 (2018),
http://crimereporting.ncsbi.gov/public/2017/ASR/2017%20Annual%20Summary.pdf [https://perma.cc/K27T-
LZ7M].
137
See North Carolina Crime Rates, 1960-2018, at http://www.disastercenter.com/crime/nccrimn.htm.
138
N.C. GEN. STAT. § 14-17(a) (2020).

20
sentences in counties with more homicides. In section III.A, we discuss one possible reason for
this finding and note that in the death sentencing literature, one observes the same negative
correlation, suggesting that serious sentencing is not as closely correlated with serious offending
in a manner that people might intuitively expect.
Further, the results imply that this negative correlation is driven by the homicide rate for
Black victims of a county or district. The correlation is not statistically related to the homicide
rate for White victims in a county or district. That suggests, then, that race matters, and in a way
connected to the race of the victim. This finding is consistent with findings in prior death
sentencing research, in which death sentencing has been observed to correlate with White
victimization among homicides (a stronger correlation than what we observe here), and similar to
what we observe here, Black victimization is negatively correlated with death sentencing. 139 One
further explanation for this result is that the counties and districts with the highest homicide rates
are generally less likely to use LWOP and instead rely on other forms of sentencing, including
death sentencing.140

1A. Four County Analysis – Homicide Rates

To further unpack the relationship between homicide rates and LWOP sentencing, in
Figure 6 we calculated the predicted probability of observing an LWOP sentence as well as the
predicted number of LWOP sentences for four counties: Durham, Guilford, Robeson, and Wake.
These four counties were chosen to provide a varied sample of counties found in North Carolina.
Guilford County, for example, is a small county, with an average household income of $42,618
and has a Black population which comprises around 29% of the county’s population. Wake
County on the other hand, is geographically larger and wealthier, and is also home to North
Carolina’s state capital, Raleigh.
Figures 6A, 6B, and 6C reflect the probability of observing an LWOP sentence in each of
these four counties. This data was generated by holding all other variables in Models 1A, 1B,
and 1C of Table 2 constant and allowing the homicide rate to vary from zero to forty-three.
Figures 6D, 6E, and 6F reflect the predicted number of LWOP sentences in each of these four
counties. Similar to the top panel of figures, the data in the bottom panel was generated by
holding all other variables in Models 1A, 1B, and 1C of Table 2 constant and allowing the
homicide rate to vary from zero to forty-three.

139
See infra Section III.A.
140
But see Theodore Eisenberg, Death Sentence Rates and County Demographics: An Empirical Study, 90 CORNELL
L. REV. 347, 358 (2005) (finding a negative relationship between county homicide rates and the rate of death
sentencing). Once again, the results imply that as the percent of the Black population in a county increases, the less
likely we are to observe adult LWOP sentences in that county. However, it is important to note the racial
fragmentation measure in Model 7 was determined to not be statistically related to observing one or more LWOP
sentences. Yet, Models 1, 2, 3, 4, 5 and 6 in Table 2 suggest there is a negative relationship between the Black
population share in a county and the likelihood we observe at least one LWOP sentence. For additional work
observing race of victim disparities in death sentencing in North Carolina, see, e.g., Barbara O’Brien et al.,
Untangling the Role of Race in Capital Charging and Sentencing in North Carolina, 1990-2009, 94 N.C. L. REV.
1997, 2023–26 (2016); Isaac Unah, Empirical Analysis of Race and the Process of Capital Punishment in North
Carolina, 2011 MICH. ST. L. REV. 609, 639–48 (2011). For additional work observing race of victim disparities in
death sentencing in North Carolina, see generally Amy R. Stauffer et al., The Interaction Between Victim Race and
Gender on Sentencing Outcomes in Capital Murder Trials: A Further Exploration, 10 HOMICIDE STUD. 98, 107–11
(2006)..

21
 Figures 6A and 6D were based on Model 1A, which revealed a significant, negative
correlation between homicide rate and LWOP sentencing. Each county in Figures 6A and 6D
reflect this significant, negative correlation. As the homicide rate increases, the probability of
LWOP sentence (6A) and the predicted number of LWOP sentences (6D) decreases. This
negative relationship is very pronounced in Durham, Guilford and Robeson counties and less
pronounced in Wake County.
Figures 6B and 6E were based on model 1B. These figures explore the relationship
between the Black-victim homicide rate and LWOP sentencing. Model 1B revealed a significant,
negative correlation between Black-victim homicide rates and LWOP sentencing. Each county in
Figure 6B and 6E reflects this significant, negative relationship. Generally speaking, a similar
pattern of results is found whether general homicide rates (Model 1A) or black-victim homicide
rates (Model 1B) were analyzed.
Figures 6C and 6F were based on Model 1C. These figures explore the relationship
between the White-victim homicide rate and LWOP sentencing. Unlike the general homicide
rate and Black-victim homicide rate, the White-victim homicide rate was not significantly related
to the probability of observing LWOP sentences, as shown in Table 2. Figures 6C and 6E show
that across these four counties, an increase in the White-victim homicide rate does not
significantly reduce the probability of an LWOP sentence (6C) or the number of LWOP
sentences (6F).
Together, these results highlight racial disparities in LWOP sentencing across a small, but
varied, collection of counties in North Carolina. Fewer LWOP sentences are predicted to occur
as the number of Black victim homicides increase, but no such relationship is found when
considering the number of White victim homicides.

Figure 6. Predicted probability and predicted number of LWOP Sentences in North

Carolina in county-year format given a range of homicide rates

22
Note: Thick lines represent model estimate. For 95% confidence intervals, please see the online,
colored version of this figure.

2. Race and LWOP Sentencing

The Black population share within a county is sometimes estimated to have a statistically
significant and negative relationship with the number of LWOP sentences applied in a county as
shown in Model 2A of Table 2. However, that relationship was not statistically significant. If, for
example, we were to compare two counties that were identical in every possible way, but one
county had a smaller Black population than the other, we would expect to observe fewer LWOP
sentences in the county that has the larger Black population. To further address the relationship
between the racial composition of a county and the use of LWOP sentencing within the given
county, a racial fragmentation measure was developed to serve as a proxy for the racial
demography of the counties in our study. This measure reports the probability that two randomly
selected adult individuals from a county would belong to different racial groups.141 Though the
effect is not statistically significant, the result in Model 2B implies that as racial fragmentation
increases within a county we are less likely to observe more LWOP sentencing in the county.142
141
A measure of racial fragmentation was previously utilized in empirical work by GARRETT, supra note 1, at 267.
142
It is important to note the racial fragmentation measure is collinear with the black population share variable–as
both are measures of racial composition. Following standard practice, model 7 included only racial fragmentation
and does not include black population share. Another general convention in the literature suggests the black
population share variable is less susceptible to incorrect interpretations when included in a statistical model. See

23
 3. Population Density

We find that the population density has a statistically significant negative correlation with
observing more LWOP convictions in Models 1, 2, and 3. Generally speaking, this suggests that
more rural counties are more likely to observe LWOP sentences than more urban counties. In an
additional robustness check we found the count of adults within a county as reported by the U.S.
Census, yielded similar results as population density measures in Table 1.

4. Death Sentencing

In another relative surprising finding, we find that death sentencing is not significantly
related to LWOP sentencing, in another relatively surprising finding. After all, one might expect
that the same prosecutors that are inclined to seek death sentences would also be more included
to seek LWOP sentences. As Table 2 displays, for Model 3B, we find there is no statistically
significant relationship between the number of previous death sentences within a prior county-
year and the number of LWOP sentences observed within that county in a current year. This
finding suggests that a county’s decision to apply a death penalty sentence is driven by a
different mechanism than the decision to apply an LWOP sentence. Indeed, there is considerable
discretion whether to seek a death sentence and only a very narrow group of cases are capitally
eligible. LWOP, on the other hand, is mandatory for all first-degree murder cases in North
Carolina.

5. The Inertia Effect

Prior research on both death sentencing nationally and juvenile LWOP sentencing in
North Carolina specifically, has shown strong inertia effects, in which prior sentences are
associated with subsequent sentencing patterns.143 County level “muscle memory” powerfully
explains the concentration of death sentences at the local level. 144 That path dependency can
occur of prosecutors develop a preference for imposing severe sentences, but also when
amenable judges and jurors, prosecution resources, and inadequate defense lawyers, all facilitate
such local patterns.145 We sought to examine whether the same type of path dependency can be
observed in LWOP sentencing in North Carolina. Where the sentence is mandatory, and unlike
death sentences does not involve jury decision-making, this inertia effect may more closely be
attributed to prosecutorial decision-making. However, because there is less discretion involved in
seeking LWOP for adult first-degree homicide than in deciding whether to seek LWOP for a
juvenile, we might expect there to be less inertia in such decision-making, particularly when
controlling for homicide rates. We relied on the statistical estimates, presented in the following
sections, to answer the question of whether there is inertia in LWOP sentencing.
A strong inertia effect, as we term it, is found in model 3A in Table 2. Here, we find the
number of previous LWOP sentences in a prior county-year is strongly, positively correlated

Chad R. Farrell, Bifurcation, Fragmentation or Integration? The Racial and Geographical Structure of US
Metropolitan Segregation, 1990—2000, 45 URB. STUD. 467, 476–77 (2008).
143
Finholt et al., supra note 24, at 165–67.
144
See Garrett, End of its Rope, supra note 1, at 149-50; see also Lee Kovarsky, Muscle Memory and the Local
Concentration of Capital Punishment, 66 Duke L. J. 259 (2016).
145
See Garrett, End of its Rope, supra note 1, at 149.

24
with observing more LWOP sentences in a county’s current year. This finding is statistically
significant and supports the prediction of an inertia effect in adult LWOP sentencing. Each
analysis we conducted observed institutional, social, and economic county-level characteristics
to understand under what conditions are adult LWOP sentences more likely to be observed. We
conclude from these empirical assessments that the number of previous adult LWOP sentences in
a prior county-year has one of the strongest relationships with an increased probability of adult
LWOP sentences. While our analyses are limited to just counties from North Carolina, we
suspect the observed inertia effect found within North Carolina is generalizable to other states,
and we plan to investigate that question in future work.

5A. Four County Analysis – Prior LWOP Sentences

To further unpack the inertia effect results, we calculated the predicted probability of
observing an LWOP sentence and the predicted number of LWOP sentences for four counties:
Durham, Guilford, Robeson, and Wake (see Figure 7). As mentioned previously, these four
counties were chosen to provide a varied sample of counties found in North Carolina. Both
Figure 7A and 7B were based on model 3A in Table 2. Figure 7A shows that as the number of
previous LWOP sentences increases we are more likely to observe more LWOP sentences–
however the strength of this relationship varies across counties. Overall, the results suggest
different counties have different propensities to implement LWOP sentences and as the number
of previous LWOP sentences increases so does their propensity to assign LWOP sentences.

Figure 7. Predicted probability and predicted number of LWOP Sentences in North

Carolina in county-year format given a range of prior year LWOP Sentences

25
Note: Thick lines represent model estimates. For 95% confidence intervals, please see the online,
colored version of this figure.

III. Implications

This first empirical analysis of case, county, and prosecution district-level characteristics
of adult LWOP sentences has implications for our understanding of serious sentences and how
they are imposed in the U.S. While research has illuminated geographic and race disparities in
death sentencing, as well as juvenile LWOP sentencing, this study shows that similar concerns of
prosecutorial discretion, racial bias, and path dependency exist for the far larger numbers of adult
LWOP sentences. This research also has implications for litigation and policy efforts in other
states. This research should be replicated in other states, to determine whether the observed
patterns are similar or different. These findings suggest that it is important to pay attention not
only to statewide sentencing rules, but how local decisionmakers, and in particular prosecutors,
implement these rules. First, we discuss what implications this work has for prosecutorial
discretion, race, and crime rates. Second, we discuss state and federal constitutional implications.
Third, we discuss implications for patterns in the imposition of life and death sentences.

A. Prosecutors, Race, and Crime Rates

Though LWOP was adopted in an effort to respond to rising crime rates, with more
retributive sentencing options, we find that in practice, LWOP sentencing is not responsive to
crime rates.146 In particular, we observe that the results from each model imply there is a
negative correlation between the homicide rate in a county and the number of LWOP sentences
we observe in that county. This finding is easily observe just from our descriptive data. After
rising in the 1990s, LWOP sentencing has remained steady in North Carolina, even as homicide
rates have fallen. The descriptive data is puzzling, however, because prosecutors retain great
discretion whether to seek LWOP sentences. One might expect prosecutors to be responsive to
crime rates and incidence in a county, but instead we find that factors other than homicide rates
correlate with LWOP sentencing.
One explanation for geographic variation in LWOP sentences may be plea-bargaining
dynamics. As previously described, the broad definition of first-degree murder requires

146
While these results are consistent with analysis of death sentencing from 1991-2017, see Garrett et al., supra note
23, at 569–70, they are inconsistent with one study examining death sentencing from 1976-2001. Eisenberg, supra
note 137, at 354–55 (finding a statistically significant direct relationship between murder rates and death sentences).

26
prosecutors to make judgments during plea bargaining that may vary in each case. Plea
bargaining may depend on the defendant’s desire to avoid a death sentence, or conversely on the
defendant’s belief that they can obtain a conviction for a lesser offense than first-degree murder,
such as a second-degree or manslaughter conviction. Thus, prosecutors’ varying approaches to
plea bargaining and the threat to the defendant of a death sentence or LWOP sentence may
influence defense lawyers practicing in a particular geographic area and the subsequent sentences
imposed on their clients.
The path dependency that we observe in LWOP sentencing practices may reflect these
plea bargaining dynamics over time. In addition, inertia effects may also arise from prosecutorial
discretion. Prosecutors may have different policies regarding when they seek a death sentence or
when they seek a first-degree homicide sentence that would result in LWOP.
Still more troubling, the results of this study also imply that LWOP sentencing is not just
driven by past LWOP sentencing practices, but also by the homicide rate within the Black
population of a county. This correlation is not, however, statistically related to the homicide rate
within the White population of a county. We also observe that increased LWOP sentencing is
associated with a greater White population in a county. This may represent, as we noted, a race-
of-victim effect, although we also note that results these were not consistent across models. It
may be that murders involving White victims explain county-level LWOP sentencing patterns
better than murders involving Black victims.
A race-of-victim effect has been widely documented in studies of death sentencing,
together with geographic disparities.147 Specifically, the death penalty is sought significantly
more often in homicides when the victim is White compared to when the victim is Black. This
holds true when accounting for the number of aggravating factors and culpability.148
That the same pattern may be observed in LWOP sentencing adds still more concern
regarding race-of-victim bias in the manner in which severe sentences are imposed. That said,
state-level patterns vary considerably in death sentencing research, and they may similarly do so
in LWOP outcomes.149 Thus, researchers should study LWOP sentencing data in other states.
Prior research on death sentencing nationally shows strong inertia effects. However, that
inertia may reflect decisions of judges and jurors, and not just prosecutorial discretion. Similarly,
researchers have observed findings of inertia in juvenile LWOP sentencing. 150 We did not,
however, observe the same results in adult LWOP sentencing. In contrast to death sentencing and
juvenile LWOP sentencing, for which limited aggravating factors must now be present in order
to sentence, adult LWOP sentencing is more broadly available in first-degree murder cases.

147
See U.S. GEN. ACCT. OFF., DEATH PENALTY SENTENCING: RESEARCH INDICATES PATTERN OF RACIAL
DISPARITIES 5–6 (1990) (finding race of victim disparities in analysis of 28 capital charging and sentencing
studies); see also Catherine M. Grosso et al., Race Discrimination and the Death Penalty: An Empirical and Legal
Overview, in AMERICA ’S EXPERIMENT WITH CAPITAL PUNISHMENT 525, 525 (James Acker & Robert Bohm eds.,
2014) (reviewing the literature up to 2013); DAVID C. BALDUS, GEORGE WOODWORTH & CHARLES A. PULASKI,
JR., EQUAL JUSTICE AND THE DEATH PENALTY: A LEGAL AND EMPIRICAL ANALYSIS 266 (1990); John J. Donohue
III, An Empirical Evaluation of the Connecticut Death Penalty System Since 1973: Are There Unlawful Racial,
Gender, and Geographic Disparities?, 11 J. EMPIRICAL LEGAL STUD. 637, 637 (2014).
148
See O’Brien et al., supra note 137, at 2023, 2025 (finding that “white victim cases are 3.3 times more likely
(8.6%/3.4%) to receive a death sentence than black victim cases” and that after controlling for aggravating evidence,
“[t]he selection rates by race of victim in Column E document that white victim cases are more likely to receive a
death sentence at every level of aggravation.”) (emphasis in original).
149
See Garrett et al., supra note 23, at 570.
150
See Finholt et al., supra note 24, at 165.

27
Nevertheless, we did observe great differences in the concentration of LWOP sentences, with
some districts imposing far greater numbers of such sentences.

B. State and Federal Constitutional Implications

A second question is whether courts will act to remedy the observed disparities. State
courts may be more hospitable to challenges making use of such data than federal courts. The
observed racial disparities, together with inertia effects driving LWOP sentencing outcomes,
could also be addressed by legislation like the now-repealed Racial Justice Act in North
Carolina, in which disparate impacts can be studied and remedied.151
Federal courts are not likely to carefully scrutinize patterns in LWOP sentencing, even if
the exhibit racial disparities. In the death penalty context, the U.S. Supreme Court in McCleskey
v. Kemp152 rejected constitutional claims raised to challenge race-of-victim effects in death
sentencing.153 Because such challenges did not succeed in the death penalty context, they are less
likely to succeed in the adult LWOP context. To be sure, Justice Breyer, in his opinion in
Glossip v. Gross,154 took account of geographic disparities in death sentences. 155 However,
Justice Breyer was writing in dissent, and that opinion also emphasized how rare death sentences
have become, and how concentrated they have become in a narrow group of counties. 156 While
LWOP sentences are concentrated and imposed more frequently in some counties, they are
generally far more common than death sentences. Regardless, Justice Breyer’s approach towards
closer examination of county-level sentencing patterns does not appear likely to garner a
majority of the Justices in the death penalty context, much less other sentencing contexts.
State courts may be more amenable to such claims. State courts, in the death penalty
context, however, have followed Justice Breyer’s approach and examined geographic disparities
and race disparities in death sentencing patterns.157 Further research should examine whether
similar patterns in LWOP sentencing can be observed in other states, in order to improve our
understanding of these sentencing practices, and to potentially inform litigation and policy.
Empirical data concerning LWOP sentencing may inform more specific challenges to
types of LWOP sentences, rather than challenges to the entire enterprise. In its death sentencing
rulings, an important focus of the Court has been on the direction of change, as states have ended
death sentencing practices.158 Proponents of sentencing reform, then, might first focus on
challenges to particular LWOP sentencing practices, beyond juvenile cases, such as LWOP for
non-murder offenses. At least in North Carolina, the bulk of adult LWOP sentences are for first-
degree homicide. Thus, the most appropriate challenges to adult LWOP sentences might be to
challenge their mandatory imposition, or their application to mentally ill or intellectually
disabled populations, or the subject involving felony murder theories. Furthermore, it is likely
151
N.C. GEN STAT. § 15A-2011(b), 2012(a)(3) (repealed 2013) (providing that a person sentenced to death “may
seek relief . . . upon the ground that racial considerations played a significant part in the decision to seek or impose a
death sentence” and that statistical evidence can be used to support such claims).
152
481 U.S. 279 (1987).
153
Id. at 287–92. Additionally, in United States v. Bass, 536 U.S. 862 (2002) (per curiam), the Court found
empirical evidence regarding race disparity in federal death sentencing insufficient to state a claim. Id. at 863–64.
154
135 S. Ct. 2726 (2015).
155
Id. at 2768 (Breyer, J., dissenting).
156
Id.
157
See, e.g., State v. Gregory, 427 P.3d 621, 630 (Wash. 2018).
158
Roper v. Simmons, 543 U.S. 551, 567 (2005) (noting “consistency in the trend toward abolition of [the juvenile
death penalty]” in the states).

28
that legislative efforts addressing the cost savings and limited benefits of lengthy sentences may
be more promising than constitutional litigation.
State court challenges to LWOP sentencing could be brought under North Carolina’s
equal protection clause.159 In addition, there is a separate nondiscrimination clause in the North
Carolina Constitution.160 The nondiscrimination clause has not been interpreted by the North
Carolina Supreme Court or North Carolina Court of Appeals. 161 However, we do know that the
clause was the product of the North Carolina Constitutional Study Commission in 1968-69, an
era in which there were several high-profile Civil Rights Act decisions in North Carolina. 162 The
Study Commission Report suggests that the legislature intended this supplementary protection, in
addition to equal protection, providing “a prohibition of improper discrimination by the State.” 163
The nondiscrimination clause, since it was added to supplement the preexisting equal protection
clause, appears to “[do] more than protect individuals from unequal treatment.”164 Perhaps like
the Civil Rights Act of 1964, it prevents practices that invidiously discriminate on the basis of
race even in the absence of a specific intent to discriminate.165
Another way to address LWOP sentences individually in North Carolina, could be
through Consent Motions for Appropriate Relief (“MARs”), in which the parties agree that the
sentence was extreme or would not be imposed today. In 2012, lawmakers enacted a provision
which permits parties to enter into an agreement regarding “any aspect, procedural or otherwise”
of a motion for appropriate relief.166 This allows a MAR on any grounds and at any time, if both
parties consent.167 However, if LWOP sentences truly reflect current prosecution priorities, then
both parties may not be open to joining such motions to revisit sentences in the past. During the
COVID-19 pandemic, there were efforts in North Carolina, as in many states, to reconsider
lengthy sentences in order to reduce prison populations, including through the use of MAR
motions.168 To date, a number of such motions have proceeded, and resulted in release, in North
Carolina.169 The general research in this Article, however, could support the consideration of
such MARs in individual cases, or the use of analogous resentencing mechanisms in other states.

C. Death Sentencing and Life Sentencing

159
N.C. CONST. art. I, § 19 (“No person shall be denied the equal protection of the laws”).
160
N.C. CONST. art. I, § 19 (“[N]or shall any person be subjected to discrimination by the State because of race,
color, religion, or national origin.”).
161
The nondiscrimination clause, added in 1971, was “based on federal civil rights legislation.” JOHN V. ORTH &
PAUL MARTIN NEWBY, THE NORTH CAROLINA STATE CONSTITUTION 68 (2d ed., 2013).
162
N.C. STATE CONST. STUDY COMM’N, REPORT OF THE NORTH CAROLINA STATE CONSTITUTION STUDY
COMMISSION TO THE NORTH CAROLINA STATE BAR AND THE NORTH CAROLINA BAR ASSOCIATION at i
(1968); see ORTH & NEWBY, supra note 156, at 32–34.
163
N.C. STATE CONST. STUDY COMM’N, supra note 157, at 74.
164
State v. Cofield, 320 N.C. 297, 302, 357 S.E.2d 622, 625 (1987).
165
See Griggs v. Duke Power Co., 401 U.S. 424, 430–31 (1971) (enforcing Title VII of the Civil Rights Act of
1964).
166
N.C. GEN. STAT. § 15A-1420(e) (LexisNexis current through Session Laws 2020-35).
167
See State v. Chevallier, 264 N.C. App. 204, 213, 824 S.E.2d 440 (2019) (noting alleged error for multiple
convictions was not properly argued on appeal, but this did not bar defendant seeking relief by other means,
including an MAR by agreement).
168
For an overview of litigation and policy concerning prison and jail release during COVID-19, see Brandon L.
Garrett, Constitutional Criminal Procedure Post-COVID, HARV. L. REV. BLOG (May 19, 2020),
https://blog.harvardlawreview.org/constitutional-criminal-procedure-post-covid/ [https://perma.cc/3H9F-HDAG].
169
Virginia Bridges, Durham DA, judge OK early release of convicted drug traffickers over COVID-19 concerns,
News & Observer, April 9, 2020.

29
 Beginning in the 1970s, concerns about the constitutionality of the death penalty led to
adoption of LWOP statutes as an alternative. More recently, strict sentencing laws led to LWOP
adoption more broadly across the country. In North Carolina, LWOP was adopted both as part of
a statute eliminating parole, and as an alternative to the death penalty. In practice, as described in
Part II, in the bulk of cases in which LWOP has been imposed, it has been in first-degree
homicide cases. That said, not all first-degree murder cases can result in the death penalty; not
only must aggravating factors be present, but juveniles and intellectually disabled individuals
may not be sentenced to death. The relationship between the death penalty and LWOP has not
been closely examined empirically.170 There is a far larger body of research on death sentencing,
and studies of LWOP are few. We find that trends in death and LWOP sentencing do not track
each other. Prosecutors increasingly do not seek or obtain death sentences in states like North
Carolina, while LWOP sentencing has remained more stable or even increased. We also observe
that death sentencing and life sentencing separately exhibit inertia, meaning that counties with
prior LWOP sentences experience more LWOP sentences in the future. Again, further research is
needed across states to better understand these trends. These findings suggest, though, that at the
local level, LWOP sentencing has increased substantially.

Conclusion

During a time in which homicide rates continue to fall, and death sentences plummet,
LWOP sentencing persists at record levels. Although research has examined drivers of
incarceration generally, and death sentencing specifically, there has been little research on
LWOP sentences, despite their growing prominence. We examined the characteristics of the
1,627 LWOP sentences imposed in North Carolina from 1995 to 2017. We analyzed defendant
race, crime, and sentence patterns by county. We associated LWOP with homicide rates, by
county, and examined interactions between homicide, victim race, and prior LWOP sentencing.
This empirical analysis of adult LWOP sentences suggests that even for a penalty that can
be mandatory (for first-degree homicides) there are important variations in its imposition. We do
not find positive correlations with homicide rates. Instead, most notably, we find strong county-
level inertia effects, suggesting that path dependency in local prosecution practices affects
LWOP sentencing. We find that fewer LWOP sentences are predicted to occur as the number of
Black victim homicides increase in a county, but no such relationship is found when considering
the number of White victim homicides. We also find that less densely populated and more rural
counties are more likely to impose LWOP sentences than urban counties.
While the U.S. Supreme Court has not regulated adult LWOP under the Eighth
Amendment, this evidence begins to make the case that there are important arbitrariness and bias
concerns in adult LWOP sentencing, as with juvenile LWOP. Further research should examine
whether similar patterns in LWOP sentencing can be observed in other states. These findings
suggest that state legislative interventions do not always have uniform effects. These findings
also suggest that local patterns in sentencing will be important subjects for future research and

170
One exception is the examination of the relationship between death sentencing and state-level adoption of LWOP.
See Garrett et al., supra note 23, at 569–70.

30
policy. Far more attention is due to LWOP sentences, which persist at record-levels despite a
continued decline in homicide rates, likely because of stark differences in exercise of
prosecutorial. These findings have implications for efforts to reconsider the most severe
sentences in the U.S., beginning with LWOP sentences.
In many other countries, LWOP sentencing has been barred, having been the subject of
substantial human rights concerns since, like death sentences, LWOP sentences do not permit the
possibility of review or relief based on rehabilitation. 171 The U.S. is bucking a global trend in the
increasing use of LWOP sentences, doubling down on a practice that implicates grave human
rights concerns, even as crime and homicide rates have generally declined. This North Carolina
study suggests that LWOP sentences reflect not national or state, but chiefly local preferences as
well as preferences for severe sentences in cases with White victims, rather than a consistent
response to crime rates. That such severe sentences can flow from local preferences raises
constitutional, state law, but also policy concerns. These are costly sentences, even if one puts to
one side their human rights and dignitary costs, where they bring equivocal benefits to public
safety. More broadly, far more work must be done to examine the consequences of our growing
and aging population of life and long-term prisoners.172 We hope this Article provides an
empirical foundation for a more systematic re-examination of LWOP in the United States.

171
Vinter, Eur. Ct. H.R. 54 (Power-Forde, J., concurring) (holding that a life sentence that is irreducible violates
ECHR Article 3, and that such a sentence must offer a prospect of release and possibility of review).
172
Lila Kazemaian & Jeremy Travis, Imperative for Inclusion of Long Termers and Lifers in Research and Policy,
14 CRIMINOLOGY & PUB. POL ’Y 355, 356 (2015).

31
Appendix A: Prosecutorial Districts

In addition to the county-level data analysis, we attempted to analyze prosecutor district-

level data as well. In North Carolina, there are currently 43 prosecutor districts, but, depending
on the year, that number fluctuates. Since 1990, the prosecutor district map has changed many
times.173 In order to obtain district-level data throughout this time period, data for each county
belonging to a district during a particular year from 1995 to 2017 was aggregated. For some of
the independent variables, this was a straightforward process. For example, the number of
LWOP sentences and the number of death sentences were simply summed across counties for
each district for each year. However, to calculate the overall homicide rate, the white homicide
rate, the black homicide rate, and the poverty rate for each district and year required a series of
steps. The first step required calculating the raw number of homicides, the number of white
homicides, the number of black homicides, and the number of those in poverty for each county
belonging to a particular district in a particular year. Those numbers were then summed and that
sum was divided by the total population in that district. These rates were then transformed into a
rate per 100,000 inhabitants. The proportion of the population that is black required calculating
the number of black inhabitants for each county in each district and dividing that number by the
total population of the district. Other independent variables such as population density and racial
fragmentation were not calculated at the district-level because it was not feasible to calculate. In
total, the district-level data resulted in 1,506 observations under this arrangement.

173
For the current districts, see e.g., N.C. GEN. STAT. § 7A-60 (2019).

32
Table A1: Prosecutorial District Poisson Regression Results for Adult LWOP Sentences
(1995-2017)

Model 4 Model 5 Model 6

Predictors 4A 4B 4C 6A 6B

Homicides per 100k^ 0.09 0.09 0.03 0.05

Black Homicides per 100k^ 0.14

White Homicides per 100k^ 0.01

% in Poverty^ -0.03 -0.03 -0.03 -0.06 -0.04 -0.04

% Population Black^ 0.04

# LWOP Sentences in prior year 0.05 ***

# Death Sentences in prior year 0.03

Intercept -0.54 * -0.52 * -0.53 ** -0.55 * -0.53 * -0.49 *

Observations 1506 1506 1506 1506 1506 1506

District Fixed Effects Yes Yes Yes Yes Yes Yes

AIC 4552.8 4547.3 4555.4 4554.4 4227.2 4244.4

Note: ^ indicates values were centered and scaled prior to running the regression. Values were
scaled so that the distribution of values had a mean equal to zero and a standard deviation equal
to one. AIC for the best fitting model is shown in bold. *p<0.1; **p<0.05; ***p<0.001

The dependent variable for these regression analyses is the frequency of LWOP sentences
within each district for each year (see Figure 5B). Approximately 31% of the time, there were
zero LWOP sentences in a given district for a given year. This percentage is far lower than what
was observed for the county-level analyses. This is because districts aggregate the frequency of
LWOP sentences across multiple counties, making it more likely that a district in a particular
year observes at least one LWOP sentence. The dependent variable has a mean value of 1.79 and
a standard deviation of 2.05. Once again, a Poisson regression was considered to be the
appropriate regression model given the dependent variable is a count variable and the data are
not overly-dispersed.
Each continuous independent variable was centered and scaled to have a mean of 0 with a
standard deviation of 1 before running the Poisson regressions. The district-level results of a set
of Poisson regressions are presented in Table A1.

33
 1. Homicide Rates

Models 4, 5, and 6 in Table A1 yield no significant relationship between homicide rate

and adult LWOP sentences. This remains true even when we examine the race of the victim by
considering the black-victim homicide rate in a district (model 4B) and the white-victim
homicide rate in a district (model 4C).

2. Race and LWOP Sentencing

Model 5 in Table A1 yields no significant relationship between the black population

percentage in a district and the number of LWOP sentences observed in that district.

3. Death Sentencing

Model 6B in Table A1 yields no statistically significant relationship between the number

of previous death sentences within a prosecutor district and the number of LWOP sentences
observed in that district. This is consistent with the county-level regressions that we reported
previously.

4. The Inertia Effect

Model 6A in Table A1 yields a highly significant, positive relationship between the

number of LWOP sentences occurring in the district’s prior year and the number of LWOP
sentences occurring in the district’s current year. This finding is yet another example of the
inertia effect and replicates the county-level analyses.

Appendix B: LWOP Sentences in North Carolina

The table below presents the number of LWOP sentences per county, the number of death
penalty sentences, the county population size (based on 2010 Census), and the black population
share (based on 2010 Census).

Table A2. North Carolina County LWOP Data (1995 – 2017)

Black
Death
Population Size (in Population
County LWOP Count Penalty
2010) Share (in
Count
2010)
Alamance 34 151131 4 18.8
Alexander 5 37198 1 5.5
Alleghany 3 11155 0 1.3
Anson 5 26948 0 48.6

34
 Ashe 3 27281 2 0.6
Avery 3 17797 0 4.0
Beaufort 4 47759 4 25.6
Bertie 5 21282 3 62.5
Bladen 10 35190 3 34.9
Brunswick 15 107431 3 11.4
Buncombe 37 238318 13 6.4
Burke 14 90912 3 6.6
Cabarrus 18 178011 1 15.3
Caldwell 10 83029 0 4.9
Camden 2 9980 1 13.2
Carteret 7 66469 0 6.1
Caswell 3 23719 1 33.8
Catawba 19 154358 4 8.4
Chatham 8 63505 0 13.2
Cherokee 1 27444 0 1.3
Chowan 2 14793 1 34.3
Clay 0 10587 0 0.6
Cleveland 23 98078 1 20.7
Columbus 25 58098 2 30.5
Craven 13 103505 3 22.4
Cumberland 99 319431 17 36.7
Currituck 2 23547 0 5.8
Dare 2 33920 0 2.5
Davidson 25 162878 7 8.9
Davie 3 41240 3 6.3
Duplin 16 58505 4 25.3
Durham 48 267587 3 38.0
Edgecombe 26 56552 3 57.4
Forsyth 78 350670 28 26.0
Franklin 3 60619 0 26.7

35
 Gaston 36 206086 10 15.3
Gates 1 12197 2 33.2
Graham 3 8861 1 0.2
Granville 8 59916 0 32.8
Greene 1 21362 1 37.3
Guilford 110 488406 11 32.5
Halifax 15 54691 5 53.2
Harnett 26 114678 6 20.9
Haywood 7 59036 2 1.1
Henderson 11 106740 1 3.0
Hertford 12 24669 2 60.5
Hoke 7 46952 0 33.5
Hyde 0 5810 0 31.6
Iredell 10 159437 4 11.9
Jackson 6 40271 0 1.8
Johnston 27 168878 11 15.1
Jones 1 10153 1 32.4
Lee 16 57866 0 20.0
Lenoir 17 59495 2 40.5
Lincoln 3 78265 0 5.5
Macon 2 33922 0 3.8
Madison 1 20765 0 1.3
Martin 10 24505 1 1.2
McDowell 12 44996 1 43.5
Mecklenburg 102 919628 15 30.8
Mitchell 1 15579 0 0.4
Montgomery 1 27798 1 18.8
Moore 18 88247 3 13.4
Nash 22 95840 3 37.2
New Hanover 38 202667 8 14.8
Northampton 8 22099 0 58.4

36
 Onslow 32 177772 7 15.6
Orange 11 133801 0 11.9
Pamlico 0 13144 0 20.0
Pasquotank 10 40661 0 37.8
Pender 9 52217 1 17.8
Perquimans 2 13453 0 24.9
Person 5 39464 0 27.0
Pitt 18 168148 7 34.1
Polk 1 20510 2 4.5
Randolph 19 141752 10 5.8
Richmond 8 46639 6 30.6
Robeson 53 134168 10 24.3
Rockingham 14 93643 5 18.9
Rowan 33 138428 7 16.2
Rutherford 15 67810 4 10.1
Sampson 17 63431 4 27.0
Scotland 9 36157 2 38.6
Stanly 7 60585 4 10.9
Stokes 6 47401 2 4.0
Surry 6 73673 3 3.7
Swain 7 13981 0 0.5
Transylvania 4 33090 0 3.9
Tyrrell 0 4407 0 38.2
Union 13 201292 4 11.7
Vance 6 45422 0 49.9
Wake 131 900993 19 20.7
Warren 0 20972 1 52.3
Washington 0 13228 1 49.8
Watauga 8 51079 0 1.7
Wayne 28 122623 8 31.4
Wilkes 18 69340 3 4.1

37
 Wilson 25 81234 3 39.0
Yadkin 6 38406 1 3.1
Yancey 2 17818 0 0.8

38