Sexual Abuse: A Journal

of Research and
Treatment
http://sax.sagepub.com/

Sex Offender Registration and Notification Policy Increases Juvenile Plea

Bargains
Elizabeth J. Letourneau, Kevin S. Armstrong, Dipankar Bandyopadhyay and Debajyoti
Sinha
Sex Abuse published online 22 August 2012
DOI: 10.1177/1079063212455667
The online version of this article can be found at:
http://sax.sagepub.com/content/early/2012/08/20/1079063212455667

Published by:
http://www.sagepublications.com

On behalf of:

Association for the Treatment of Sexual Abusers

Additional services and information for Sexual Abuse: A Journal of Research and Treatment can be
found at:
Email Alerts: http://sax.sagepub.com/cgi/alerts
Subscriptions: http://sax.sagepub.com/subscriptions
Reprints: http://www.sagepub.com/journalsReprints.nav
Permissions: http://www.sagepub.com/journalsPermissions.nav

Downloaded from sax.sagepub.com at JOHNS HOPKINS UNIV on September 12, 2012

>> OnlineFirst Version of Record - Aug 22, 2012

What is This?

Downloaded from sax.sagepub.com at JOHNS HOPKINS UNIV on September 12, 2012

455667
12455667Sexual AbuseLetourneau et al.

SAXXXX10.1177/10790632

Sex Offender Registration and

Notification Policy Increases
Juvenile Plea Bargains

Sexual Abuse: A Journal of

Research and Treatment
XX(X) 119
The Author(s) 2012
Reprints and permission:
sagepub.com/journalsPermissions.nav
DOI: 10.1177/1079063212455667
http://sax.sagepub.com

Elizabeth J. Letourneau1, Kevin S. Armstrong1,

Dipankar Bandyopadhyay2, and Debajyoti Sinha3

Abstract
The aim of this study was to test the hypothesis that South Carolinas sex offender
registration and notification policy influenced juvenile sex offense case plea bargains.
Two types of plea bargains were examined: initial sex offense charges amended to
nonsex offense charges and amended to lower severity charges. Comparison analyses
were conducted with juvenile assault and robbery offense cases. Archival data on
cases involving 19,215 male youth charged with sex, assault, and/or robbery offenses
between 1990 and 2004 informed analyses. Of these youth, 2,991 were charged with
one or more sex offense, 16,091 were charged with one or more assault offense, and
2,036 were charged with at one or more robbery offense. Generalized estimating
equations (GEE) were used to model changes in the probabilities of plea bargain
outcomes across three time intervals: before policy implementation (1990 to 1994),
after initial policy implementation (1995 to 1998), and after implementation of a
revised policy that included online registration requirements (1999 to 2004). Results
indicate significant increases in the probability of plea bargains for sex offense cases
across subsequent time periods, supporting the hypothesis that South Carolinas
initial and revised registration and notification policies were associated with significant
increases the likelihood of plea bargains to different types of charges and to lower
severity charges. Results were either nonsignificant or of much lower magnitude for
the comparison assault and robbery analyses. Suggestions for revising South Carolina
and national registration and notification policies are discussed.

Medical University of South Carolina, Charleston, SC, USA

University of Minnesota, Minneapolis, MN, USA
3
Florida State University, Tallahassee, FL, USA
2

Corresponding Author:
Elizabeth J. Letourneau, Medical University of South Carolina, Johns Hopkins Bloomberg School of Public
Health, Family Services Research Center, 67 President St., STE MC406, Charleston, SC 29425, USA
Email: eletourn@jhsph.edu

Downloaded from sax.sagepub.com at JOHNS HOPKINS UNIV on September 12, 2012

Sexual Abuse XX(X)

Keywords
adolescent sexual abusers, juvenile sex offender, Megans Law, notification, registration,
sex offender policy
Sex offender registration and notification policies aim to deter sexual offending,
reduce sexual recidivism, provide law enforcement personnel additional tools for
investigating alleged offenses and provide community members with information to
increase their and their childrens safety. Despite these aims, a growing body of
research generally fails to support the efficacy of these policies. For example, of 14
studies examining policy effects on recidivism, 10 reported no significant effects on
violent and/or sexual recidivism (Adkins, Huff, & Stageberg, 2000; Letourneau &
Armstrong, 2008; Letourneau, Bandyopadhyay, Sinha, & Armstrong, 2009b;
Letourneau, Levenson, Bandyopadhyay, Sinha, & Armstrong, 2010; Sandler, Freeman, & Socia, 2008; Schram & Milloy, 1995; Vsquez, Maddan, & Walker, 2008;
Veysey, Zgoba, & Dalessandro, 2009; Zevitz, 2006; Zgoba, Witt, Dalessandro, & Veysey, 2009). By comparison, just two studies reported reduced sexual and violent recidivism that might be attributable to policy effects (Duwe & Donnay, 2008; Washington
State Institute for Public Policy, 2005) and one study suggested increased sexual recidivism that might be attributable to policy effects (Freeman, in press). Prescott and
Rockoff (2008) separately examined the effects of registration versus notification on
recidivism. Their models suggested that registration laws might reduce sexual recidivism whereas notification laws might increase sexual recidivism.
Research also has examined whether registration and notification policies exert
unintended effects. Of particular concern is that registration and notification policies
perceived as unduly harsh might influence judicial case procedures (e.g., see Logan,
2009). Two studies have examined whether state registration and notification policies
influenced adult plea bargains. One study found that adults were significantly more
likely to plead guilty to amended (nonsex) offenses following South Carolinas initial
policy enactment and again following its modification to require online notification
(Letourneau, Levenson, Bandyopadhyay, Armstrong, & Sinha, 2010). A second study
found no effects on adult plea bargains following enactment of New Yorks registration and notification policy (Freeman, Sandler, & Socia, 2009). Letourneau and colleagues hypothesized that these discrepant outcomes were due to differences between
the two states policies. In particular, South Carolinas harsher policy was hypothesized as driving the increase in plea bargains for that states adult defendants.
Another study examined the influence of South Carolinas policy on judicial case
processing of juvenile sex offense cases (Letourneau, Bandyopadhyay, Sinha, &
Armstrong, 2009a). In that study, we reported that South Carolinas registration and
notification policy was associated with a significant decline in the likelihood of solicitors1 prosecuting youth charged with sex offenses. For cases that were prosecuted, the
policy was associated with a significant increase in the likelihood of guilty

Downloaded from sax.sagepub.com at JOHNS HOPKINS UNIV on September 12, 2012

Letourneau et al.

determinations. We hypothesized that the reduced likelihood of solicitors prosecuting

juvenile sex offense cases was due to their reluctance to subject some youth to South
Carolinas harsh registration requirements. We further hypothesized that the increased
rate of guilty determinations might be due, in part, to solicitors amending initial
charges to lesser (nonsex) offense charges at adjudication as part of plea bargain
negotiations.
In the present study, we test the hypothesis that South Carolinas registration and
notification policy was associated with increased probabilities of plea bargains. We
also examined plea bargain outcomes for juvenile assault and robbery offense cases to
address the possibility that any observed changes in sex offense plea bargains might
apply to violent offense cases in general. All models controlled for extra-legal factors
shown to influence juvenile case processing, including youth age, race, and prior
offenses (e.g., McCarthy & Lindquist, 1985; Minor, Hartman, & Terry, 1997) and
previously shown to predict the likelihood of solicitor decisions to prosecute juvenile
sex offense cases (Letourneau et al., 2009a). We were unable to control for extra-legal
variables specific to decisions makers (e.g., solicitors) that might influence juvenile
case dispositions, such as their beliefs, personalities, and demographic characteristics
have been less often studied (e.g., Davis, Severy, Kraus, & Whitaker, 1993). However,
research findings suggest that defendant variables are stronger predictors than decision maker variables (Davis et al., 1993). Thus, the present study seeks to determine
whether juvenile sex offender registration and/or public notification influenced case
outcomes after controlling the effects of youth age, race and prior offense history.

Method
South Carolinas Registration and Notification Policy
South Carolinas sex offender registration and notification policy was initially implemented on January 1, 1995. Charges that trigger automatic registration and notification requirements include most sex offenses for which juveniles can be adjudicated,
and range from noncontact offenses (voyeurism, indecent exposure) to serious sex
offenses against adults (e.g., criminal sexual conduct) and minors (e.g., criminal
sexual conduct with a minor). Youths adjudicated delinquent of sex offense charges
are required to register their personal information, including home and school
addresses, with the local sheriffs office twice per year for life (i.e., registration). For
all registered youth ages 12 and older, authorities inform their schools, their victims,
and nearby child-oriented businesses of the youths registration status (i.e., targeted
notification). Other individuals or businesses may request registration status information on specific youth. South Carolinas initial registration and notification policy was
amended in 1999 to include lifetime online notification. Online notification requirements apply to all registered adult offenders and to some registered juvenile offenders,
including those with prior sex offenses and those for whom the dispositional judge
recommends online notification.

Downloaded from sax.sagepub.com at JOHNS HOPKINS UNIV on September 12, 2012

Sexual Abuse XX(X)

Sample
A total of 19,215 male youth ages 6 to 21 years when charged with sex, assault, or
robbery offenses between 1990 and 2004 contributed data to the analyses. The majority of youth (90%) had been charged with just one offense type. Youth had assault
offense charges (73%), sex offense charges (11%), robbery offense charges (6%),
assault and sex offense charges (5%), or assault and robbery offense charges (5%). No
youth had all three types of charges. The sample was African American (66.3%) or
White (32.5%) and the mean age at first charge of any type was 14.7 years (SD = 1.5
years). Of youth charged with one or more sex offenses (n = 2,991), 59% were African
American, 40% were White, and the mean age at initial charge was 14.4 years (SD =
1.6 years). Approximately 6% had accrued more than one sex offense charge over
time (range of 2 to 5 charges) and approximately 5% had any assault or robbery
offense adjudications prior to their first sex offense charge. Of the youth charged with
at least one assault offense (n = 16,091), 67% were African American, 32% were
White, and the mean age at initial charge was 14.7 years (SD = 1.5 years). Twenty
percent of these youth had accrued more than one assault offense charge over time
(range of 2 to 9) and less than 1% had any sex or robbery offense adjudications prior
to their first assault charge. Of youth charged with one or more robbery offenses (n =
2,036), 88% were African American, 11% were White, and the mean age at initial
charge was 15.2 years (SD = 1.3 years). Approximately 8% had accrued more than
one robbery offense charge over time (range of 2 to 4) and 10% had any sex or assault
offense adjudications prior to their first robbery charge.

Procedures
All data were extracted from the South Carolina Department of Juvenile Justice (DJJ)
Management Information System in collaboration with the South Carolina Budget
and Control Board Office of Research and Statistics. Only cases initially charged
between January 1, 1990 and December 31, 2004 were included in the present study.
This timeframe permitted examination of cases processed during Time 1 (T1), which
included the five years prior to registration (January 1, 1990 through December 31,
1994), cases processed during Time 2 (T2), which included the four years following
initial implementation of the registration policy (January 1, 1995 through December
31, 1998), and Time 3 (T3), which included the five years following implementation
of online notification (January 1, 1999 through December 31, 2004).
All DJJ cases involving sex, assault, or robbery offense charges and that were
associated with final adjudication decisions were included in the analyses, with
minor exceptions. Exceptions included cases that involved female defendants
(which were rare, especially for sex offense cases) and cases that appeared to
involve very young (< 6 years) or old (> 21 years) juvenile defendants. In addition,
cases with multiple charges within an offense class (e.g., two or more sex offense

Downloaded from sax.sagepub.com at JOHNS HOPKINS UNIV on September 12, 2012

Letourneau et al.

charges filed on the same date), were removed from consideration due to difficulty
in linking initial to final charges. Just 8% of sex offense cases were removed due to
multiple charges, resulting in a final count of 3,176 sex offense cases that informed
analyses (the number of cases exceeds the number of youth because some youth
were charged with more than one sex offense over time, as noted earlier). Likewise,
just 6% of assault cases were removed due to multiple charges, for a final count of
20,424 cases informing the assault analyses and just 4% of robbery charges were
removed, for a final count of 2,212 cases informing the robbery analyses. Charges
filed against minors in general sessions (adult) court were not included in the analyses for this study. All data were de-identified prior to being made available for
analysis. Study procedures were approved by the Medical University of South
Carolinas Institutional Review Board and by the South Carolina Department of
Juvenile Justice.

Variables Included in the Models

Data extracted from DJJ records included prosecution and adjudication dates, labels
for initial (prosecution) and final (adjudication) charges and final disposition outcomes. All initial charges were classified as sex, assault, or robbery based on their
charge labels (see Letourneau et al., 2009a for more details on classification of
charges). Adjudication charges included these and other types of charges, as
detailed subsequently. Severity ratings for charges also were included in the data;
these ratings represent the number of years an adult would likely be sentenced to
serve in jail if convicted of a similar charge (Barrett, Katsiyannis, & Zhang, 2006;
Letourneau et al., 2009a). Charge severity can assume one of seven unequally
spaced values between 2 and 25. Specifically, sex offense severity ranges from 2 to
21, assault offense severity ranges from 2 to 25, and robbery offense severity ranges
from 5 to 21.
Two types of plea bargains were coded, including cases in which the initial charge
was amended to a lower severity charge at adjudication and cases in which the initial
charge was amended to a different type of charge at adjudication (e.g., if an initial sex
offense charge was amended to an assault offense charge). We included changes in
charge type because most sex offense charges, including many lower severity charges,
trigger lifetime registration and notification requirements in South Carolina. Thus, for
defendants pleading guilty, the primary way to avoid registration and notification is to
plead guilty to nonsex offense charges.
Three covariates included in all analyses were the youths age at initial charge, a
dichotomous indicator for the youths race (Minority or Nonminority), and a dichotomous indicator for prior violent (i.e., sex, assault, or robbery) offense adjudications
(any or none). Although potentially a continuous variable, the priors variable effectively functioned as a dichotomous variable (e.g., 94% of youth with sex offenses had
no prior violent charges and 5% had just one prior). Similar results occurred for the

Downloaded from sax.sagepub.com at JOHNS HOPKINS UNIV on September 12, 2012

Sexual Abuse XX(X)

youth with assault and robbery offenses. Charge severity was dichotomized to control
for low (2, 3, 5, 8) versus high (15, 21, 25) severity charge offenses and included as a
covariate in models estimating the likelihood of pleading to another type of offense
(charge severity is not included in models estimating the likelihood of pleading to
lower severity charges).

Data Analytic Procedures

Generalized estimating equations (GEE) methods (Liang & Zeger, 1986) were used
for modeling the effects of registration and notification on plea bargains. GEE provides robust and consistent marginal/population-averaged parameter estimates and
standard errors (Stokes, Davis, & Koch, 2000). Moreover, this analytic method is
characterized by two strengths that lend themselves to the needs of the present study.
First, unlike other regression methods, GEE is able to accommodate repeated, correlated binary responses. In the present study, some juveniles had multiple offenses, all
of which contributed to the analyses. Second, the GEE method uses an estimating
equations approach (versus, for example, a maximum likelihood method) for model
fitting and can accommodate large databases such as the one informing the present
studys results via readily available software like PROC GENMOD in SAS. Readers
are referred to our earlier publication for a more thorough discussion of the rationale
for using GEE versus alternative analytic approaches (Letourneau et al., 2009a).
Two primary outcomes were examined with GEE: (a) whether an initial charge was
amended to a charge of a different offense type and (b) whether an initial charge was
amended to a lower severity charge. The main objectives of the analyses were to evaluate the magnitude of the changes, if any, in the probabilities of these outcomes across
the three time intervals of relevance: T1 (pre-registration period from 1990 to 1994),
T2 (initial registration period from 1995 to 1998), and T3 (online registration period,
from 1999 to 2004). Models estimated the comparisons of outcomes at T2 versus T1
and T3 versus T2. A linear contrast then estimated the comparison of outcomes at T3
versus T1. All covariates (youth age, race, prior violent offenses, and charge severity)
were centered at their means. To control for a possible linear increase in the log likelihood of pleading over time we ran a series of models with integer charge year added
to the set of already reported predictors. The addition of linear charge year did not alter
results of models and was a significant predictor in just one of the six models (pleading
from a higher severity assault offense to a lower severity offense2). We therefore report
only the results of the original models that did not include linear charge year.
The adequacy of each GEE model was assessed using an adapted Hosmer and
Lemeshow (HL) goodness-of-fit test. This adapted test is an extension of the typical
HL goodness-of-fit test used with ordinary logistic regression and is appropriate for
models examining repeated binary response data as in the present study (Horton et al.,
1999). The adapted HL goodness-of-fit test statistic has an approximate chi-squared
distribution when models have been specified correctly.

Downloaded from sax.sagepub.com at JOHNS HOPKINS UNIV on September 12, 2012

Letourneau et al.

Results
Descriptive Results
Before reviewing the GEE results, descriptive information is provided regarding plea
bargains. Of all cases with initial sex offense charges, 73% were not amended (i.e.,
these cases retained a sex offense charge at adjudication); 24% pleaded out of type
(i.e., to a nonsex offense charge, 92% of which were also lower severity charges); and
3% pleaded to a lower severity sex offense charge. The percentage of cases that were
not amended declined over time, from 84% at T1 to 73% at T2 and to 65% at T3.
Of all cases with initial assault offense charges, 90% were not amended; 1% pleaded
out of type, 8% pleaded to a lower severity charge, and 1% pleaded to a lower severity
nonassault offense. The percentage of assault cases that were not amended remained
essentially unchanged over time (91%, 90%, and 89% at T1, T2, and T3, respectively).
Of all cases with initial robbery offense charges, 82% were not amended; 3%
pleaded out of type, 4% pleaded to a lower severity charge, and 11% pleaded to a
lower severity nonrobbery charge. The percentage of robbery cases that were not
amended declined slightly following the first time period (84%, 81%, and 82% at T1,
T2, and T,3 respectively).
Information on the type and severity of nonsex offense charges to which sex
offense charges were amended is presented in Table 1. In the great majority of cases,
charges were amended to assault offenses, of which the most common was aggravated assault and battery. Noncontact sex offenses were often amended to other types
of offenses, including public disorderly conduct, disturbing schools, simple larceny,
and trespassing.
Information on the type and severity of nonassault charges to which assault charges
were amended is presented in Table 2. The small subgroup of cases amended to nonassault offenses was characterized by significant variability with respect to adjudication
charges. Of the small subgroup of robbery cases amended to nonrobbery charges, most
amended charges related to theft (see Table 3).

GEE Results
Models estimating the probability of cases being amended to a different charge type
or a lower charge severity are presented first for cases with sex offense charges, then
for assault and robbery offense charges. Results for all models are summarized in
Table 4 and depicted in Figure 1. Goodness-of-fit tests for the GEE models are presented in the text. All tests except one were characterized by nonsignificant results,
supporting adequate model fit.
Sex offense cases, charge type. There was statistically significant evidence of
increases in the likelihood of plea bargains to nonsex offense charges from each time
period to the next (Table 4, column 3). Specifically, there was a 128% increase in plea
bargains to nonsex offense charges from the T1 to T2 and an additional 52% increase
from T2 to T3. These results are depicted in Figure 1. Model results also indicated that

Downloaded from sax.sagepub.com at JOHNS HOPKINS UNIV on September 12, 2012

Sexual Abuse XX(X)

Table 1. Type, Frequency, and Severity of Nonsex Offense Charges to Which Initial Sex
Offense Charges Were Amended.
Amended charge: Most
frequent
Initial charge

Amended charge: Second

most frequent

Count Severity Offense Percent Severity Offense Percent Severity

Criminal sexual conduct w. minor

1st degree
253
21
AGASBA
.89
(victim < 11
years)
2nd degree
15
AGASBA
.72
78
(victim 11
14 years)
2nd degree
15
AGASBA
.84
32
(victim < 16
years)
Lewd act on
15
AGASBA
.71
85
child under 16
Criminal sexual conduct
21
AGASBA
.81
1st degree
84
15
AGASBA
.79
2nd degree
53
3rd degree
25
8
AGASBA
.60
Assault with intent to commit criminal sexual conduct
1st degree
67
21
AGASBA
.64
2nd degree
7
15
AGASBA
.71
3rd degree
26
8
AGASBA
.58
Indecent
25
5
PUBDIS
.48
exposure
Exhibiting
9
2
DISTSL
.22
indecent/
obscene
pictures
Peeping Tom
8
3
TRESPS
.75
(Eavesdropping)
6
2
AGASBA
.50
Other sex
offenses
Attempted criminal sexual conduct
2nd degree
2
15
AGASBA
.50
3rd degree
1
15
SIMASL
1.00
Buggery
ATMPII
.50
2
5
Engaging child
2
15
AGASBA 1.00
under 18
for sexual
performance

SIMA&B

.06

SIMA&B

.18

SIMA&B

.16

SIMA&B

.15

8
8
8

SIMA&B
SIMA&B
AGASLT

.10
.17
.20

2
2
8

8
8
8
2

SIMA&B
SIMA&B
SIMA&B
DISTSL

.25
.14
.42
.32

2
2
2
2

PETLAR

.22

PETLAR

.13

SIMA&B

.33

SIMA&B

ATMPIII

.50

.50

8
8

(continued)

Downloaded from sax.sagepub.com at JOHNS HOPKINS UNIV on September 12, 2012

Letourneau et al.
Table 1. (continued)
Amended charge: Most
frequent
Initial charge

Amended charge: Second

most frequent

Count Severity Offense Percent Severity Offense Percent Severity

Sexual
exploitation
of a minor, 3rd
degree

AGASLT

.50

PUBDIS

.50

Produce,
promote,
direct, sexual
performance by
child

AGASBA

.50

SIMA&B

.50

Notes: AGASBA = assault and battery of a high and aggravated nature. AGASLT = assault of high and
aggravated nature. ATMPII = attempt to commit a category II offense. ATMPIII = attempt to commit a
category III offense. DISTSL = disturbing schools. PETLAR = simple larceny. PUBDIS = public disorderly
conduct. SIMA&B = simple assault and battery. SIMASL = simple assault. TRESPS = trespassing.

two covariates were significantly associated with plea bargains to nonsex offense
charges: youths age at charge and the dichotomized charge severity indicator. Specifically, each 1-year increase in age at charge accounted for an estimated 7% reduction in
the odds of plea bargains to nonsex offense charges. There was a 74% increase in the
odds of plea bargains to nonsex offense charges for youth with higher versus. lower
severity charges. The goodness-of-fit test yielded a Chi 2 value of 7.00 and a nonsignificant p-value of 0.67.
Sex offense cases, charge severity. There was statistically significant evidence of
increases in the likelihood of plea bargains to lower severity charges at each time period
(Table 4, column 4). Specifically, there was a 96% increase in plea bargains to lower
severity offense charges from T1 to T2 and an additional 44% increase from T2 to T3
(see Figure 1). Model results also indicated that youths age at charge and Minority status
were significant predictors. Each 1-year increase in age at charge accounted for a 5%
reduction in the odds of plea bargains to lower severity charges. There was a 19%
decrease in the odds of plea bargains to lower severity charges for Minority versus Nonminority defendants. The goodness-of-fit test yielded a Chi 2 value of 13.04 and a nonsignificant p-value of 0.16.
Assault offense cases, charge type. Model results (Table 4, column 5) indicated statistically significant evidence of some increase in the odds of plea bargains from assault
to nonassault offense charges from T2 to T3. However, as depicted in Figure 1, the
magnitude of this change was trivial (i.e., recall that the percent of cases associated
with plea bargains from assault to nonassault charges was 2%, 2%, and 3% for T1-T3,
respectively). The indicator for prior violent offense adjudications and charge severity

Downloaded from sax.sagepub.com at JOHNS HOPKINS UNIV on September 12, 2012

10

Sexual Abuse XX(X)

Table 2. Type, Frequency, and Severity of Nonassault Offense Charges to Which Initial Assault
Offense Charges Were Amended.
Amended charge: Most
frequent
Offense

Count Severity

Simple assault & 136

battery
Simple assault
111
Criminal domestic violence
1st/2nd
81
offense
3rd or
1
subsequent
offense
High and
2
aggravated
nature
Aggravated
43
assault &
battery
Assault &
32
battery w.
intent to kill
Assault w. intent 21
to kill or w.
concealed
weapon
Murder
19
Assault &
12
battery by
a student
on a school
employee
Lynching
2nd degree
8
Aggravated
10
assault
Kidnapping
6

Amended charge: Second

most frequent

Offense Percent Severity Offense Percent Severity

DISTSL

0.45

PUBDIS

.25

DISTSL

0.49

PUBDIS

.15

INCORR

0.72

MALRPT

.06

ATMPIV

MLIAPP

INCORR

.50

DISTSL

0.21

TRESPS

.12

15

ACBAX

0.16

15

ACBAI

.16

15

MLIAPP

0.19

PTFIRA

.14

25
8

IMANSL
DISTSL

0.53
0.58

8
2

ACBAX
TEACHR

.21
.42

15
5

15
8

PUBDIS
LEWDAT

0.50
0.20

2
15

DISTSL
PUBDIS

.25
.20

2
2

21

ATMPX

0.33

21

INTCS1

.17

21

Notes: ACBAI = accessory before/after the fact to a category I offense. ACBAX = accessory before/
after the fact to a category X offense. ATMPIV = attempt to commit a category IV offense. ATMPX =
attempt to commit a category X offense. DISTSL = disturbing schools. IMANSL = involuntary manslaughter. INCORR = incorrigible. INTCS1 = assault w. intent to commit criminal sexual conduct, 1st degree.
LEWDAT = lewd act on child under 16. MALRPT = malicious injury to real property. MLIAPP = malicious
injury to personal property. PTFIRA = pointing or presenting fire arms at a person. PUBDIS = public
disorderly conduct. TEACHR = threaten life, person, family of public official/school teacher. TRESPS =
trespassing.

Downloaded from sax.sagepub.com at JOHNS HOPKINS UNIV on September 12, 2012

11

Letourneau et al.
Table 3. Type, Frequency, and Severity of Nonrobbery Offense Charges to Which Initial
Robbery Offense Charges Were Amended.
Amended charge: Most
frequent
Offense
Strong armed
robbery
Armed robbery
Purse snatching
Attempted
armed robbery

Amended charge: Second

most frequent

Count Severity Offense Percent Severity Offense

Percent Severity

165

SIMA&B

0.26

PETLAR

.25

114
17
1

21
5
15

ACBAX
PETLAR
ATMPII

0.28
0.24
1

15
2
8

ATMPX
RECSTG

.28
.24

21
1

Notes: ACBAX = accessory before/after the fact to a category X offense. ATMPII = attempt to commit
a category II offense. ATMPX = attempt to commit a category X offense. BREACH = breach of trust w.
fraudulent intent, US$1,000 or less. PETLAR = simple larceny. RECSTG = receiving stolen goods less than
US$200. SIMA&B = simple assault and battery.

were also was significant predictors. Specifically, there was a 76% reduction in the
odds of plea bargains to nonassault offense charges for youth with prior violent
offenses compared with youth with no such priors. There was a 55% increase in the
odds of plea bargains to nonassault offense charges for youth with higher versus lower
severity charges. The goodness-of-fit test yielded a Chi 2 value of 6.95 and a nonsignificant p-value of 0.64.
Assault offense cases, charge severity. Model results (Table 4, column 6) indicated
statistically significant evidence of an increase in the likelihood of plea bargains to
lower severity charges from T1 to T2. Again, as depicted in Figure 1, the magnitude of
change was trivial (i.e., recall that the percent of assault cases associated with plea
bargains to lower severity charges was 8%, 9%, and 9% for T1-T3, respectively).
Youth race and prior violent offenses were significant predictors in this model. There
was an estimated 42% increase in the odds of plea bargains to lower severity charges
for Minority versus Nonminority youth, and an estimated 27% increase in the odds of
plea bargains to lower severity charges for youth with prior violent offense adjudications versus those without such priors. The goodness-of-fit test yielded a Chi 2 value
of 15.06 and a significant p-value of .005, suggesting that the number and/or type of
covariates in this model are inadequate to explain the increase in charge severity from
T1 to T2.
Robbery offense cases, charge type. Model results (Table 4, column 7) indicated no
statistically significant differences in the odds of plea bargains from robbery to nonrobbery offense charges between time periods. Age of youth, prior violent offenses,
and charge severity did significantly influence these plea bargains. Specifically, each
1-year increase in age was associated with an estimated 12% reduction in the odds of
plea bargains from robbery to nonrobbery offense charges. There was an estimated

Downloaded from sax.sagepub.com at JOHNS HOPKINS UNIV on September 12, 2012

12

Sexual Abuse XX(X)

Table 4. GEE Parameter Estimates For Plea Bargain Analyses in Juvenile Sex, Assault, and
Robbery Offense Cases. Revised results table20120504.
Sex (N = 3,176)1
Parameter
Intercept
(T1)

T2 vs. T1

T3 vs. T2

T3 vs. T1

Age

Race

Priors

Severity

Type

Severity

Assault
(N = 20,424)2
Type

Severity

Robbery
(N = 2,212)3
Type

Severity

1.986

1.758

4.093

2.476

1.966

1.973

p
OR
95% CIOR
b
p
OR
95% CIOR
b
p
OR
95% CIOR
b
p
OR
95% CIOR
b
p
OR
95% CIOR
b
p
OR
95% CIOR
b
p
OR
95% CIOR
b
p
OR
95% CIOR

<.000
0.137
0.11, 0.17
0.824
<.000
2.280
1.77, 2.94
0.419
<.000
1.520
1.24, 1.86
1.243
<.000
3.467
2.77, 4.34
.071
0.007
0.931
0.88, 0.98
0.155
0.075
0.856
0.72, 1.02
0.017
0.924
0.984
0.70, 1.38
0.553
<.000
1.739
1.36, 2.22

<.000
0.172
0.15, 0.21
0.672
<.000
1.958
1.54, 2.49
0.363
<.000
1.438
1.18, 1.76
1.035
<.000
2.815
2.28, 3.47
0.056
0.032
0.946
0.99, 1.00
0.206
0.016
0.814
0.69, 0.96
0.019
0.910
1.019
0.73, 1.42
___
___
___
___

<.000
0.017
0.01, 0.02
0.107
0.436
1.078
0.82, 1.42
0.392
<.000
1.480
1.17, 1.62
0.50
<.000
1.648
1.31,2.08
0.018
0.550
1.018
0.96, 1.08
0.099
0.333
0.906
0.74, 1.11
1.422
<.000
0.241
0.15, 0.40
0.436
<.000
1.546
1.22, 1.96

<.000
0.084
0.08, 0.09
0.149
0.035
1.161
1.01, 1.33
0.014
0.818
0.986
0.88, 1.11
0.135
0.037
1.145
1.01, 1.30
0.013
0.393
0.987
0.96, 1.02
0.351
<.000
1.420
1.26, 1.60
0.242
0.001
1.274
1.10, 1.48
___
___
___
___

<.000
0.140
0.11, 0.17
0.097
0.528
1.102
0.82, 1.49
0.093
0.574
0.911
0.66, 1.26
0.003
0.983
1.003
0.74, 1.37
0.123
0.008
0.884
0.81, 0.97
0.096
0.632
0.909
0.61, 1.35
0.931
0.001
0.394
0.23, 0.68
0.379
0.004
0.685
0.53, 0.88

<.000
0.139
0.11, 0.17
0.334
0.025
1.397
1.04, 1.87
0.092
0.552
0.912
0.67, 1.24
0.242
0.115
1.274
0.94, 1.72
0.093
0.030
0.911
0.84, 0.99
0.174
0.360
0.840
0.58, 1.22
0.247
0.234
0.781
0.52, 1.17
___
___
___
___

Notes. T1 refers to the pre-registration time period (1990-1994), T2 refers to the registration time period (19951998), and T3 refers to the online notification time period (1999-2004). OR = odds ratio. 95% CI OR refers to the 95%
confidence interval around the estimated OR. The analyses comparing T3 vs. T1 were linear contrasts. Age refers to age
at charge. Race is coded 1 for Minority race. Priors refers to prior violent offense adjudications and is coded 1 for one
or more prior violent offense adjudications. The severity predictor variable was not included in the analyses in which
severity was an outcome.
Multiple arrests resulted in 13,176 cases for 2,991 offenders, 220,424 cases for 16,091 offenders, and 32,212 cases for
2,036 offenders.

Downloaded from sax.sagepub.com at JOHNS HOPKINS UNIV on September 12, 2012

13

Letourneau et al.

60% reduction in the odds of plea bargains for youth with versus without prior violent
offense adjudications. There was an estimated 31% reduction in the odds of plea bargains from robbery to nonrobbery offense charges for youth with higher versus lower
severity charges. The goodness-of-fit test yielded a Chi 2 value of 14.79 and a nonsignificant p-value of 0.10.
Robbery offense cases, charge severity. Model results (Table 4, column 8) indicated
statistically significant evidence of an increase in the likelihood of plea bargains to
lower severity charges from T1 to T2. As depicted in Figure 1, this change was
trivial and there was no significant change from T2 to T3. Youth age was also a
significant predictor. Specifically, each 1-year increase in age at initial charge corresponded with an estimated 9% reduction in the odds of plea bargains to lower
severity charges. The goodness-of-fit test yielded a Chi 2 value of 11.28 and a nonsignificant p-value of 0.26.

Discussion
Results from the present study indicated marked increases in plea bargains for juvenile sex offense cases following initial enactment of South Carolinas registration and
notification policy in 1995 and following the 1999 revision for online notification.
Previously, we reported that the probability of prosecuting these types of cases
declined across these same time periods, whereas probability of guilty determinations
increased (Letourneau et al., 2009a). We hypothesized that these earlier findings were

100%
90%
Probability of Plea Bargains

80%
70%

Sex Type

60%

Sex Severity

50%

Assault Type
Assault Severity

40%

Robbery Type

30%

Robbery Severity

20%
10%
0%
T1

T2

T3

Figure 1. Estimated likelihood of plea bargains to charges amended per type or severity for
juvenile sex, assault, and robbery offense cases.

Downloaded from sax.sagepub.com at JOHNS HOPKINS UNIV on September 12, 2012

14

Sexual Abuse XX(X)

attributable to the imposition of harsh registration and notification policies on juveniles. Specifically, we suggested that solicitor reluctance to prosecute juvenile sex
offense cases reflected a protective concern that not all youth who committed sex
offenses deserved the lifetime registration and notification requirements imposed by
South Carolinas policy. We also suggested that the increased rate of guilty dispositions was attributable to increased use of plea bargains, which also permit youth to
avoid registration and notification requirements. The present study supports this
hypothesis. Specifically, results indicated large and statistically significant increases
in the probability of plea bargains following initial enactment of South Carolinas
policy and subsequent enactment of online notification.
Results from the comparison assault and robbery analyses further bolster this
hypothesis. First, unlike sex offense cases, assault and robbery offense cases were
rarely associated with pleading guilty to other charge types, suggesting that amending
charges out of type is a sex offense-specific legal maneuver, presumably aimed at
avoiding sex offense-specific legal consequences such as registration and notification.
Second, whereas there were dramatic increases in the probability of sex offense cases
plea bargaining to lower severity charges over time, changes assault and robbery
offense cases were comparatively trivial, even when statistically significant (see
Figure 1). Thus, although plea bargaining to lower severity charges is clearly not a sex
offense-specific defensive strategy, findings nevertheless support the hypothesis that
sex offense-specific factors were influencing this process.
Adding to our confidence that findings in this study are attributable to South
Carolinas registration and notification policy, several South Carolina solicitors have
indicated that juvenile sex offenders are hidden to avoid the penalties imposed by
this policy. Notably, one attorney told the first author that the family court judge in her
jurisdiction had forbidden solicitors from bringing juvenile sex offense cases before
his court. Although anecdotal, such comments in combination with our results suggest
that South Carolinas policy is exerting powerful yet entirely unintended effects on
juvenile sex offense adjudication procedures.

Study Strengths and Limitations

Results from this study should be weighed in light of its strengths and weaknesses.
Study strengths include inclusion of nearly all South Carolina juvenile sex offense
cases associated with adjudications across a 15-year period of time, inclusion of several covariates that might influence plea bargain outcomes, and comparison analyses
with assault and robbery offense cases to address the potential influence of alternative
(nonspecific) factors on plea bargain outcomes. The most significant limitations of
this study are the nonexperimental study design and the risk of model misspecification. In particular, models could have been misspecified due to the absence of data on
other relevant predictors such as decision maker characteristics (e.g., prosecutor or
judge beliefs or demographic characteristics) at the individual case level and of any
predictors at higher levels (e.g., jurisdiction). We partly addressed these concerns by

Downloaded from sax.sagepub.com at JOHNS HOPKINS UNIV on September 12, 2012

15

Letourneau et al.

including goodness-of-fit tests, which supported model adequacy in all but one model,
and by including the comparison assault and robbery analyses which we believe supported the argument that South Carolinas registration and notification policy exerted
sex offense-specific effects.

Policy Recommendations
South Carolinas registration and notifications policy makes few distinctions between
juveniles and adults. Nearly all sex offense adjudication (juveniles) and convictions
(adults) trigger an automatic requirement for lifetime registration and notification. All
juvenile registrants are subjected to targeted community notification whereas all adult
registrants and some juvenile registrants are subjected to lifetime online notification.
Neither individual recidivism risk nor case-specific circumstances are permitted to
influence registration and notification requirements and there is no mechanism for
registrants to reduce the duration of requirements from life. Against this harsh reality,
many South Carolina solicitors appear to have altered the manner in which they do
business. Now, more than ever, juvenile sex offense cases are dismissed outright or
when formal adjudication is desired, initial sex offense charges are pled to nonsex
offense charges. These findings are consistent with the broader criminal justice literature, in which it has been reported that policies perceived by judicial decision makers
as unnecessarily harsh are under-enforced (e.g., see research on uneven enforcement
of mandatory minimum sentences with adult offenders; McCoy, 1984; Zimring,
Hawkins, & Kamin, 2001).
In summary, our program of research indicates that South Carolinas registration
and notification policy fails to deter new juvenile sex crimes (Letourneau,
Bandyopadhyay, Armstrong, & Sinha, 2010), fails to deter juvenile sexual or violent
recidivism (Letourneau & Armstrong, 2008; Letourneau et al., 2009b), and increases
the risk that youth will incur new nonsex, nonviolent misdemeanor charges (Letourneau
et al., 2009b). Other researchers have demonstrated that federal standards for juvenile
sex offender registration fail to distinguish between youth who will reoffend or not
(Batastini, Hunt, Present-Koller, & DeMatteo, 2011; Caldwell, Ziemke, & Vittacco,
2008) as do state-specific methods in New Jersey, Texas, and Wisconsin (Caldwell &
Dickinson, 2009; Caldwell et al., 2008). The basis for these federal and state policy
failures might lie, in part, with the low sexual recidivism rate of youth adjudicated for
sex offenses (Caldwell, 2002; Vandiver, 2006). Indeed, results from one study indicated that the risk of sexual recidivism was statistically equal for youth treated in a
residential facility for either sexual or nonsexual delinquent offenses (Caldwell, 2007).
Thus, distinguishing between youth likely to sexually reoffend or not involves more
than simply knowing that a youth has a history of such offending.
Given these policy failures, the avoidance of registration and notification requirements by some youth via dismissed or amended charges seems unlikely to harm community safety per se. However, there are additional potential consequences of these
outcomes. For example, juveniles who have actually committed sexual offenses but

Downloaded from sax.sagepub.com at JOHNS HOPKINS UNIV on September 12, 2012

16

Sexual Abuse XX(X)

whose charges were dismissed or amended to nonsex offense charges might not
receive appropriate clinical services or supervision. Moreover, youth age and race
influenced solicitor decisions to permit plea bargains, introducing the possibility of
inequity. Rather than relying upon solicitors to protect youth from ineffective and possibly harmful legal policies, a more just and equitable solution would be to alter the
onerous policies themselves. The available evidence from four state policies and the
federal registration policy fails to support any community safety effect of juvenile
registration or notification. More research evaluating other state policies would helpful. However, given the consistency of findings to date, we submit that youth adjudicated delinquent as minors should be exempt from sex offender registration and
notification requirements until such time as empirically rigorous evidence emerges
indicating that these policies can be crafted in such a way as to improve community
safety.
Acknowledgments
The authors wish to thank Ms. Trudie Trotti, director of research and statistics, South Carolina
Department of Juvenile Justice and Mr. Charles Bradberry, formerly senior statistician, South
Carolina State Budget and Control Board, Office of Research and Statistics.

Declaration of Conflicting Interests

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

Funding
The author(s) disclosed receipt of the following financial support for the research, authorship,
and/or publication of this article: This study was supported in part by the Centers for Disease
Control and Prevention (R49 CE00567) and the National Science Foundation (SES 0455124).
The conclusions expressed in this article are those of the authors and do not necessarily represent those of the funding agencies or South Carolina state agencies.

Notes
1. In South Carolina prosecutors are referred to as solicitors.
2. The addition of charge year to this model results in a statistically significant increase of
0.04% in the odds of pleading to a lower severity offense annually, p = .016, with a 20%
overall reduction between the periods commencing 1995 and 1999, p = .038.

References
Adkins, G., Huff, D., & Stageberg, P. (2000). The Iowa sex offender registry and recidivism. Des
Moines: Iowa Department of Human Rights.
Barrett, D. E., Katsiyannis, A., & Zhang, D. (2006). Predictors of offense severity, prosecution,
incarceration and repeat violations for adolescent male and female offenders. Journal of
Child and Family Studies, 15, 709-719.

Downloaded from sax.sagepub.com at JOHNS HOPKINS UNIV on September 12, 2012

17

Letourneau et al.

Batastini, A. B., Hunt, E., Present-Koller, J., & DeMatteo, D. (2011). Federal standards for community registration of juvenile sex offenders: An evaluation of risk prediction and future
implications. Psychology, Public Policy and Law, 17, 451-474.
Caldwell, M. F. (2002). What we do not know about juvenile sexual reoffense risk. Child Maltreatment, 7, 291-302.
Caldwell, M. F. (2007). Sexual offense adjudication and sexual recidivism among juvenile
offenders. Sexual Abuse: A Journal of Research and Treatment, 19, 107-113.
Caldwell, M. F., & Dickinson, C. (2009). Sex offender registration and recidivism risk in juvenile sexual offenders. Behavioral Sciences and the Law, 27, 941-956.
Caldwell, M. F., Ziemke, M. H., & Vittacco, M. J. (2008). An examination of the sex offender
registration and notification act as applied to juveniles: Evaluating the ability to predict
sexual recidivism. Psychology, Public Policy and Law, 14, 89-114.
Davis, T. L., Severy, L. J., Kraus, S. J., & Whitaker, J. M. (1993). Predictors of sentencing
decisions: The beliefs, personality variables, and demographic factors of juvenile justice
personnel. Journal of Applied Social Psychology, 23, 451-477.
Duwe, G., & Donnay, W. (2008). The impact of Megans Law on sex offender recidivism: The
Minnesota experience. Criminology, 46, 411-446.
Freeman, N. J. (2012). The public safety impact of community notification laws: Rearrest of
convicted sex offenders. Crime and Delinquency, 58(4), 539-564.
Freeman, N. J., Sandler, J. C., & Socia, K. M. (2009). A time-series analysis on the impact of sex
offender registration and community notification laws on plea bargaining rates. Criminal
Justice Studies, 22, 153-165.
Horton N. J., Bebchuk, J. D., Jones, C. L., Lipsitz, S. R., Catalano, P. J., Zahner, G. E., &
Fitzmaurice, G. M. (1999). Goodness-of-fit for GEE: An example with mental health service utilization. Statistics in Medicine, 18, 213-222.
Letourneau, E. J., & Armstrong, K. S. (2008). Recidivism rates for registered and nonregistered
juvenile sexual offenders. Sexual Abuse: A Journal of Research and Treatment, 20, 393-408.
Letourneau, E. J., Bandyopadhyay, D., Sinha, D., & Armstrong, K. S. (2009a). The effects of
sex offender registration policies on juvenile justice decision making. Sexual Abuse: A Journal of Research and Treatment, 21, 149-165.
Letourneau, E. J., Bandyopadhyay, D., Sinha, D., & Armstrong, K. S. (2009b). The influence
of sex offender registration on juvenile sexual recidivism. Criminal Justice Policy Review,
20, 136-153.
Letourneau, E. J., Bandyopadhyay, D., Armstrong, K. S., & Sinha, D. (2010). Do sex offender
registration and notification requirements deter juvenile sex crimes? Criminal Justice and
Behavior, 37, 553-569.
Letourneau, E. J., Levenson, J. S., Bandyopadhyay, D., Armstrong, K., S., & Sinha, D. (2010).
The effects of sex offender registration and notification on judicial decisions. Criminal Justice Review, 35, 295-317.
Letourneau, E. J., Levenson, J. S., Bandyopadhyay, D., Sinha, D., & Armstrong, K. S. (2010).
Effects of South Carolinas sex offender registration and notification policy on adult recidivism. Criminal Justice Policy Review, 21, 435-458.

Downloaded from sax.sagepub.com at JOHNS HOPKINS UNIV on September 12, 2012

18

Sexual Abuse XX(X)

Liang, K.-Y., & Zeger, S. L. (1986). Longitudinal data analysis using generalized linear models.
Biometrika, 73, 13-22.
Logan, W. A. (2009). Knowledge as power: Criminal registration and community notification
laws in America. Stanford, CA: Stanford Law Press.
McCarthy, B. R., & Lindquist, C. A. (1985). Ambiguity and conflict in sentencing research:
Partial resolution through crime-specific analysis. Journal of Criminal Justice, 13, 155-163.
McCoy, C. (1984). Determinate sentencing, plea bargaining bans, and hydraulic discretion in
California. Justice System Journal, 9, 256-275.
Minor, K. I., Hartmann, D. J., & Terry, S. (1997). Predictors of juvenile court actions and recidivism. Crime & Delinquency, 43, 328-344.
Prescott, J. J., & Rockoff, J. E. (2008). Do sex offender registration and notification laws
affect criminal behavior? Unpublished manuscript. Retrieved from http://ssrn.com/abstract
=1100663 (cited with author permission).
Sandler, J. C., Freeman, N. J., & Socia, K. M. (2008). Does a watched pot boil? A time-series
analysis of New York States sex offender registration and notification law. Psychology,
Public Policy, and Law, 14, 284-302.
Schram, D., & Milloy, C. D. (1995). Community notification: A study of offender characteristics
and recidivism. Olympia, WA: Washington Institute for Public Policy.
Stokes, M. E., Davis, C. S., & Koch, G. G. (2000). Categorical data analysis using the SAS
system. New York, NY: Wiley.
Vandiver, D. (2006). A prospective analysis of juvenile male sex offenders: Characteristics and
recidivism rates as adults. Journal of Interpersonal Violence, 21, 673-688.
Vsquez, B. E., Maddan, S., & Walker, J. T. (2008). The influence of sex offender registration
and notification laws in the United States: A time-series analysis. Crime and Delinquency,
54, 175-192.
Veysey, B., Zgoba, K., & Dalessandro, M. (2009). A preliminary step towards evaluating the
impact of Megans Law: A trend analysis of sexual offenses in New Jersey from 1985 to
2005. Justice Research and Policy, 10(2), 1-18.
Washington State Institute for Public Policy (2005). Sex offender sentencing in Washington
State: Has community notification reduced recidivism? (Document No. 05-12-1202). Olympia Author. Retrieved from www.wsipp.wa.gov
Zevitz, R. G. (2006). Sex offender community notification: Its role in recidivism and offender
reintegration. Criminal Justice Studies, 19, 193-208.
Zgoba, K., Witt, P., Dalessandro, M., & Veysey, B. (2009). Megans Law: Assessing the practical and monetary efficacy. Retrieved from http://www.ncjrs.gov/pdffiles1/nij/grants/
225370.pdf
Zimring, F. E., Hawkins, G., & Kamin, S. (2001). Punishment and democracy: Three strikes
and youre out in California. New York, NY: Oxford University Press.

Bios
Elizabeth J. Letourneau, Ph.D. is Associate Professor, Mental Health Department, Bloomberg
School of Public Health, Johns Hopkins University. She has conducted sexual offense-related

Downloaded from sax.sagepub.com at JOHNS HOPKINS UNIV on September 12, 2012

19

Letourneau et al.

research for more than 20 years, serving as principal investigator for four federally funded
evaluations of sex offender registration and notification policies.
Kevin S. Armstrong is Statistical Analyst and Data Manager at Family Services Research
Center, Department of Psychiatry, Medical University of South Carolina. Since 2003 Mr.
Armstrong has served in numerous capacities on eight funded research projects and is co-author
on nine publications, the majority of which evaluated the effects of sex offender registration and
notification policies.
Dipankar Bandyopadhyay, Ph.D. is Associate Professor, Division of Biostatistics, School of
Public Health, University of Minnesota. His primary research interests are in Bayesian biostatistics, spatial data analysis, survival analysis and other parametric and nonparametric methods
and their applications to problems in epidemiology and behavioral studies.
Debahyoti Sinha, Ph.D. is the Ron & Carolyn Hobbs Endowed Chair in Statistics, Florida
State University. His main research interests are in survival analysis, Bayesian biostatistics,
modeling cancer prevention data, cure rate survival and relapse data. He co-authored Bayesian
Survival Analysis (2001; Springer-Veralg: New York).

Downloaded from sax.sagepub.com at JOHNS HOPKINS UNIV on September 12, 2012