Content uploaded by Jill S Levenson
Author content
All content in this area was uploaded by Jill S Levenson on Jul 22, 2015
Content may be subject to copyright.
Content uploaded by Jill S Levenson
Author content
All content in this area was uploaded by Jill S Levenson on Jun 30, 2015
Content may be subject to copyright.
http://sax.sagepub.com
Research and Treatment
Sexual Abuse: A Journal of
DOI: 10.1177/1079063209340143
2009; 21; 363 originally published online Jul 6, 2009; SEX ABUSE
Mark Chaffin, Jill Levenson, Elizabeth Letourneau and Paul Stern
on Halloween
How Safe Are Trick-or-Treaters?: An Analysis of Child Sex Crime Rates
http://sax.sagepub.com/cgi/content/abstract/21/3/363
The online version of this article can be found at:
Published by:
http://www.sagepublications.com
On behalf of:
Association for the Treatment of Sexual Abusers
found at:
can beSexual Abuse: A Journal of Research and Treatment Additional services and information for
http://sax.sagepub.com/cgi/alerts Email Alerts:
http://sax.sagepub.com/subscriptions Subscriptions:
http://www.sagepub.com/journalsReprints.navReprints:
http://www.sagepub.com/journalsPermissions.navPermissions:
http://sax.sagepub.com/cgi/content/refs/21/3/363 Citations
at ATSA on August 13, 2009 http://sax.sagepub.comDownloaded from
363
Sexual Abuse: A Journal of
Research and Treatment
Volume 21 Number 3
September 2009 363-374
© 2009 Association for the
Treatment of Sexual Abusers
10.1177/1079063209340143
http://sajrt.sagepub.com
hosted at
http://online.sagepub.com
How Safe Are
Trick-or-Treaters?
An Analysis of Child Sex Crime
Rates on Halloween
Mark Chaffin
University of Oklahoma Health Sciences Center, Oklahoma City, OK
Jill Levenson
Lynn University, Boca Raton, Florida
Elizabeth Letourneau
Medical University of South Carolina Family Services Research Center,
Charleston, SC
Paul Stern
Snohomish County Prosecutors Office, Everett, WA
States, municipalities, and parole departments have adopted policies banning known
sex offenders from Halloween activities, based on the worry that there is unusual risk
on these days. The existence of this risk has not been empirically established. National
Incident-Base Reporting System crime report data from 1997 through 2005 were used
to examine daily population adjusted rates from 67,045 nonfamilial sex crimes against
children aged 12 years and less. Halloween rates were compared with expectations
based on time, seasonality, and weekday periodicity. Rates did not differ from expecta-
tion, no increased rate on or just before Halloween was found, and Halloween incidents
did not evidence unusual case characteristics. Findings were invariant across years,
both prior to and after these policies became popular. These findings raise questions
about the wisdom of diverting law enforcement resources to attend to a problem that
does not appear to exist.
Keywords: child sexual abuse; sex offenses; Halloween; National Incident-Base
Reporting System; NIBRS; policy and law
D
uring the past two decades, various sex crime policies, including registration,
community notification, and residence restrictions, have been implemented to
prevent and deter sexual violence against children. The newest trend in this effort
has been to prohibit known sex offenders from engaging in holiday festivities, par-
ticularly on or around Halloween. In previous generations, urban myths about
Authors’ Note: Please address correspondence to Jill Levenson, Lynn University, 3601 N. Military Trail,
Boca Raton, FL 33431; e-mail: jsljwm@bellsouth.net.
at ATSA on August 13, 2009 http://sax.sagepub.comDownloaded from
364 Sexual Abuse: A Journal of Research and Treatment
Halloween warned of tainted or poisoned candy, and parents were advised to check
for apples with razor blades tucked inside, despite no verification that such incidents
ever occurred (Glassner, 1999). Now parents are urged to be wary of potential sexual
predators handing out candy and to check their local sex offender registries to find
the addresses of homes to be avoided.
Several states have passed laws or enacted local regulations restricting known sex
offenders’ activities on Halloween (O’Connor, 2005). In some New York counties,
for example, sex offenders on probation are required to attend a 4-hour education
program on Halloween night. In New Jersey, New York, Virginia, Wisconsin,
California, South Carolina, and North Carolina, curfew policies prohibit registered
sex offenders from going out or opening their doors on Halloween (Wood, 2006).
New Jersey sex offenders caught giving out candy are considered in violation of
probation and face up to 3 years in jail. In Tennessee, sex offenders on probation are
banned from Halloween costume parties and cannot put up decorations, and in Ohio,
Illinois, Virginia, and North Carolina, offenders are ordered to attend meetings with
law enforcement or probation officials during the evening hours of Halloween
(Associated Press, 2006; Walberg, 2006). In Idaho, Maryland, Florida, and Texas,
sex offenders on probation are warned not to decorate their homes and are told to
keep their houses dark on Halloween night. In Kentucky, sex offenders receive a
letter from police telling them not to give out candy or have unauthorized contact
with children. Michigan police told a Lansing television station that there are no
laws in the state prohibiting a sex offender from handing out candy but recom-
mended that parents check the sex offender registry before trick-or-treating.
These policies are premised on the theory that Halloween provides an opportunity
for sex offenders to make contact with youngsters for improper purposes or to use
costumes to conceal their identities and avoid detection (O’Connor, 2005). In par-
ticular, Halloween policies are aimed at protecting children from sexual assault by
nonfamily members and individuals with whom a child has little routine contact
outside of the Halloween context. A second premise is the perception that sex
offenders have alarmingly high recidivism rates, generating understandable concern
and mobilizing efforts toward community protection policies (Levenson, Brannon,
Fortney, & Baker, 2007). In fact, it has been found that a minority of convicted sex
offenders go on to be rearrested for new sex crimes (Bureau of Justice Statistics,
2003; Hall, 1995; Hanson & Bussiere, 1998; Hanson & Morton-Bourgon, 2004;
Harris & Hanson, 2004). Furthermore, apprehensions about “stranger danger” can
be misleading, as more than 90% of sexually abused children are victimized by
someone known to them, often a relative or close acquaintance (Bureau of Justice
Statistics, 2000, 2004). A study in Jacksonville, Florida, found that 15 out of 230 sex
offenders (about 7%) were neighbors of their victims (Bartkowiak & Daniels, 2007),
and in Minnesota, about 3% to 4% of repeat sex offenders victimized neighbors
(Duwe, Donnay, & Tewksbury, 2008).
at ATSA on August 13, 2009 http://sax.sagepub.comDownloaded from
Chaffin et al. / Sex Offenders on Halloween 365
Law enforcement officials note that Halloween laws were not developed in
response to actual attacks reported to have occurred on 31 October or in a trick-or-
treat context (O’Connor, 2005). Police have generally not expressed alarm about an
observed increase in incidents where children are snatched off porches or victim-
ized on Halloween. Notably, the chair of the parole board in New Jersey stated that
there were no known assaults by registered sex offenders committed on Halloween
(Wood, 2006). Kristen Anderson, director of the case analysis division for sex
offender tracking at the National Center for Missing and Exploited Children, was
quoted as saying that only nine nonfamily child abductions were reported in the
United States between 29 October and 1 November over a 5-year period, none of
which appeared to have any connection to trick-or-treating (Kriel, 2008).
Although the types of offenses targeted by Halloween sex offender policies may
be rare, a rationale for these policies might exist nonetheless. On one hand, it could
be the case that sex crimes show an unusual spike on or around Halloween, justify-
ing a corresponding intensification in vigilance and restrictions. On the other hand,
the holiday could be a completely typical and routine day as far as these types of
offenses are concerned, contradicting a major part of the rationale on which the
policies are based. Whether or not there is actually a measured elevation in child sex
crime risk on Halloween is unknown. This is a question that has received little or
no empirical study. The purpose of this study was to empirically examine whether
there is a Halloween effect on nonfamilial sex crimes against children—that is,
whether the rate of these crimes on Halloween differs from what we would expect
if it were just another day.
Method
Data for the study were drawn from publicly downloadable National Incident-
Based Reporting System (NIBRS) data sets. NIBRS was instituted in the United
States as an updated and more detailed replacement for the uniform crime reporting
system that has been in effect since the 1930s. Similar to the uniform crime report-
ing, NIBRS draws data from local law enforcement agencies in individual reporting
jurisdictions. Some or all of a state’s jurisdictions may participate in NIBRS.
Participating jurisdictions are required to demonstrate that their local data categories
can be translated reliably and accurately into NIBRS coding categories. One of the
main advantages of NIBRS is that it provides detailed data on individual, not aggre-
gated, incidents known to law enforcement, including information about the incident
date, the victim(s) and offender(s) involved, the type of crime(s) involved in the
incident, victim(s) characteristics, offender(s) characteristics, and expanded victim–
offender relationship information. NIBRS participation is limited but accelerating.
In 1995, jurisdictions in 9 states were contributing data; by 1997 the number was 12
at ATSA on August 13, 2009 http://sax.sagepub.comDownloaded from
366 Sexual Abuse: A Journal of Research and Treatment
states, and by 2004, jurisdictions in 29 states submitted reports. By 2004, NIBRS
covered approximately 20% of the nation’s population and recorded data on more
than 4 million individual crime incidents annually, which constitutes approximately
16% of crimes in the United States. Although not a representative sample, NIBRS
does provide a rich database suited to examining trends and patterns in more finely
grained crime categories.
NIBRS is particularly well suited to the purposes of this study because it provides
detailed individual-level information about crime incidents, including the occur-
rence date, the type of crime, victim age, and the type of preexisting victim–offender
relationship. It also covers a time frame beginning prior to the widespread popularity
of Halloween sex offender policies up until the recent past. NIBRS data allow us to
narrow the examination to sex crimes that might reasonably be addressed by
Halloween sex offender policies. For example, sex crimes against children commit-
ted by family members, close friends, babysitters, teachers, or others are not directly
addressed by Halloween-related sex offender policies. Rather, these policies are
most directly concerned with preventing sex crimes against children in trick-or-treat
and similar contexts that might bring the child into contact with sex offenders who
are strangers, acquaintances, neighbors, or individuals with no preexisting close
relationship to the child. Similarly, these policies are most directly concerned with
preventing sex offenses against young children rather than sex offenses against all
victim age groups. Because these policies are often directed toward offenders on sex
offender registries, they can apply to both adult and juvenile offenders depending on
local registration practices and policies. Individual incidents were selectively
included from NIBRS data with these considerations in mind.
Victim data from the 1997 through 2005 NIBRS reporting years were used for the
study. These years captured data from jurisdictions in 10 states in 1997, expanding
to 30 states by 2005. Victimization incidents were included if they met the following
criteria: (a) the victim was age 12 years or less on the occurrence date (not the
reporting date) of the incident, (b) the primary offender in the incident had no
reported preexisting family or role relationship to the victim, and (c) the offense(s)
coded included at least one sex crime other than a prostitution-related offense. The
offenses included were forcible sodomy, forcible rape, statutory rape, sexual assault
with an object, child pornography–related crimes, and forcible fondling. The result-
ing combined sample included 67,045 victims.
Results
Because Halloween-related activities sometimes occur on the days immediately
preceding Halloween, as well as on Halloween itself, a span of 3 days (Halloween
and 2 days prior) were designated as Halloween days for all statistical comparisons.
Victims were 73% female and 27% male with a mean and median age of 8 years.
at ATSA on August 13, 2009 http://sax.sagepub.comDownloaded from
Chaffin et al. / Sex Offenders on Halloween 367
Victim age was bimodally distributed, and the most common ages of victims were
12 and 11, followed by 4 and 5 years of age (see Figure 1). Victims were 75% White,
19% Black, 5% unknown, and less than 1% Native American/Alaska Native and
Asian/Pacific Islander. Five percent were reported as ethnically Hispanic. Victim
gender, age distribution by year, median age, race, and ethnicity did not significantly
differ between Halloween and other days. Primary offenders were 94% male with a
mean age of 24 and a median age of 18, which did not differ between Halloween and
other days.
A total of 67,307 sex offenses were coded for the 67,045 victims. The most com-
mon offense category was forcible fondling (63%), followed by forcible rape (15%),
forcible sodomy (12%), sexual assault with an object (6%), and statutory rape (4%).
The distribution of offense categories did not differ significantly between Halloween
days and other days. The victim–offender relationship for the sample was coded as
acquaintances (43%), followed by an unknown person (28%), neighbor (8%),
stranger (5%), or a person otherwise known to the victim but not an unspecified
family member (15%), and the distribution did not significantly differ between
Halloween days and all other days.
Figure 1
Age of Victims
at ATSA on August 13, 2009 http://sax.sagepub.comDownloaded from
368 Sexual Abuse: A Journal of Research and Treatment
The overall analytic strategy for examining offense rates involved three steps.
First, the time series of daily rates over 9 years was examined for trend, seasonality,
and periodicity. Second, a predictive model was specified to account for seasonal
and periodic cycles. Finally, the Halloween effect was tested for deviation from
prediction. Daily report numbers over the 9 years were calculated by aggregating the
number of victims for each day in the time series.
A distinct and invariant day-of-month effect was observed in the raw data, reflect-
ing very high reporting on the first day of every month, likely because of interval
censoring for incidents where the incident month may be known but not the exact
date. Additionally, elevated annual (on 1 January of each year) and midannual (on 1
June of each year) spikes are present also likely because of censoring. The first day
of each month was removed from the overall time series to manage this irregularity.
Daily report counts were transformed into rates per 100,000 of population. Summed
annual population figures associated with each individual NIBRS reporting jurisdic-
tion for each year were used for estimating rates. Linear interyear interpolations
were used to create a series of estimated daily population numbers, which served as
the denominator in daily time series rate calculations. The time series of rates was
decomposed into trend and seasonal components using a Loess decomposition pro-
cedure in R with optimized time span settings (Cleveland, Cleveland, McRae, &
Terpenning, 1990). This is shown in Figure 2. Regular, sinusoidal seasonal patterns
were observed, along with an overall declining trend over the 1997 to 2005 time
period. Finally, weekday periodicity was observed, with weekend incidents being
somewhat less prevalent than Monday to Friday incidents. Trend and seasonality
components seen in Figure 2 were modeled as a Fourier series
where Y
t
is the incident count at time t, l is the period length, and e is residual error.
Trend components are reflected in the linear and quadratic terms involving t, and
seasonality components are reflected in the sine wave (sin and cos) terms. The
Fourier model fits the data well, accounted for 50% of observed variability, and all
parameters were significant. The fit of the Fourier prediction model with observed
counts, aggregated across years to show fit for 1 mean year, is shown in Figure 3.
Note the substantial rate increases during the summer months. Predicted counts from
the Fourier model for each day in the series were saved and converted to rates per
100,000 of population for use as covariates in the final model of Halloween
effects.
Daily rates were modeled as the outcome using a mixed model procedure with
restricted maximum likelihood estimation of model parameters. Days of the year
were treated as subjects and years as a repeated factor with a diagonal covariance
Y
t
= a
0
+ a
1
t + a
2
t
2
+ b cos
2p
l
+ g sin
2p
l
+ e
t
,
at ATSA on August 13, 2009 http://sax.sagepub.comDownloaded from
Chaffin et al. / Sex Offenders on Halloween 369
structure. Explanatory variables included the predicted rate estimated by the Fourier
model to account for linear + quadratic trend and sinusoidal seasonality, an indicator
variable coded for day of the week to account for weekday periodicity, and an indi-
cator variable for Halloween (including the two prior days) as the effect of interest.
No significant effect attributed to Halloween days was found (see Table 1).
Examining Figure 2, greater residual variability was noted during earlier periods of
the time series. Residual weights were applied to the model to correct for this, using
the reciprocal of the time predicted squared model residual, but this did not alter
Figure 2
Loess Decomposition of Trend and Seasonality
0.00
data
−0.02
seasonal
0.052
trend
−0.04
1998
remainder
time
0.00 0.04 0.08
0.00 0.02
0.0600.048
0.056
0.150.10
0.05
2000 2002 2004 2006
at ATSA on August 13, 2009 http://sax.sagepub.comDownloaded from
370 Sexual Abuse: A Journal of Research and Treatment
results. Data on Halloween sex offender policies in the individual jurisdictions were
unavailable, but it is likely that several began to implement sex offender policies at
some point given that the policies have seen widespread uptake over the study time
frame. If these policies produced a net effect over time on overall Halloween victim-
ization, we would predict that the rates of offenses on Halloween would show
greater decline over time relative to the rates for other days. If true, this might
explain the absence of a simple Halloween effect. To test whether there may have
been greater reductions in offense rates on Halloween relative to other days over the
9-year span, a year-by-Halloween interaction term was added to the model. The
interaction did not approach significance. The observed rates and predicted rates
from a restricted model (excluding the Halloween effect) for Halloween day and the
surrounding ±14 days are presented in Figure 4.
To contextualize sex crimes against children by nonfamily members, we examined
only the most recent year included in the study (2005), which included information on
Figure 3
Fit of Fourier Model With Observed Counts, Aggregated
Across Years, With Reference Line at Halloween
at ATSA on August 13, 2009 http://sax.sagepub.comDownloaded from
Chaffin et al. / Sex Offenders on Halloween 371
more than 5 million victimizations from 30 states with reporting jurisdictions. The
most common types of crime from among the incidents reported on Halloween and
adjacent days were theft (32%), destruction or vandalism of property (21%), assault
(19%), and burglary (9%). Vandalism and property destruction accounted for a sig-
nificantly greater proportion of crime around Halloween compared with all other days
(21% vs. 14% of all reports). Sex crimes of all types accounted for slightly more than
1% of all Halloween crime. Sex crimes against children of the kind examined in this
study accounted for less than .2% of all Halloween crime incidents.
Discussion
This study found no significant increase in risk for nonfamilial child sexual abuse
on or just prior to Halloween. Although sex offenders may use seemingly innocent
Figure 4
Observed Rates Versus Restricted Model Predicted Rates
(Excluding Halloween Variable From the Prediction Model) for
the Month Surrounding Halloween. Reference line at Halloween
at ATSA on August 13, 2009 http://sax.sagepub.comDownloaded from
372 Sexual Abuse: A Journal of Research and Treatment
opportunities to engage children and sexually abuse (Conte, Wolf, & Smith, 1989;
Salter, 1995) and therefore might be hypothesized to use trick-or-treat for ulterior
purposes, this logic does not appear to translate into any actual unusual rate of sex
offenses on Halloween. The absence of a Halloween effect remained constant over
the 9-year period, beginning well before the current interest in Halloween sex
offender policies and extending to recent years. Any Halloween policies that have
been adopted by reporting jurisdictions during that period appear not to have
affected the overall sex offense rate.
Halloween was also typical in terms of victim and offender characteristics, the
types of child sex offenses reported, and the categories of victim–offender relation-
ships involved. As with all other days of the year, young children are sexually vic-
timized on Halloween. We do not suggest that there is no risk on Halloween or that
anecdotal accounts of Halloween molestations should be dismissed. Nor do we sug-
gest that parents should abandon caution and reasonable supervision of their chil-
dren. But there does not appear to be need for alarm concerning sexual abuse on
these particular days. In short, Halloween appears to be just another autumn day
where rates of sex crimes against children are concerned. If anything, increased
vigilance concerning risk should be directed to the summer months in general, where
regular seasonal increases in rates are readily seen (see Figure 3).
Although the sample available through NIBRS did not include all states, the
similarities in overall time trends to those found by other researchers using 50-state
data sources are encouraging. During the study time frame, decreases in violent
crime have been observed in general (Bureau of Justice Assistance, 2008). Official
crime data and victim reports both indicate that rape arrest rates peaked in 1990 and
have decreased steadily since 1991 (Maguire & Pastore, 2003). Examining child
welfare data, a 51% decline in child sexual abuse rates have been observed between
1990 and 2005 (Finkelhor & Jones, 2004; Jones & Finkelhor, 2003). Ironically, these
welcome decreases appear to coincide with escalating anxiety, fueling fear in parents
and prompting legislators to propose laws to assuage the concerns of their constitu-
ents (Sample & Kadleck, 2008). In this case, worries and good intentions might have
inspired advocates and lawmakers to propose legislation that combats a nonexistent
problem. The findings suggest that Halloween policies may in fact be targeting a
new urban myth similar to past myths warning of tainted treats. The results are
Table 1
Model Results Testing the Effect of Halloween Versus Other Days, Controlling
for Fourier Trend and Seasonality, Plus Weekday Effects
Source Numerator df Denominator df F Significance
Intercept 1 400.068 26.691 .000
Halloween 1 331.727 .258 .612
Fourier prediction 1 550.670 497.325 .000
Weekday 6 2670.719 114.329 .000
Year 8 1952.419 3.880 .000
at ATSA on August 13, 2009 http://sax.sagepub.comDownloaded from
Chaffin et al. / Sex Offenders on Halloween 373
consistent with observations offered by law enforcement officials who do not
describe any epidemic of trick-or-treaters being assaulted by known sex offenders
and who have observed no unusual rate of child sexual assault events on Halloween
(O’Connor, 2005).
It might be argued that Halloween sex offender policies are worthwhile even if
they prevent only a single child from being victimized. However, this line of reason-
ing fails to consider the cost side of the cost–benefit equation. The wide net cast by
Halloween laws places some degree of burden on law enforcement officers whose
time would otherwise be allocated to addressing more probable dangerous events.
For example, a particularly salient threat to children on Halloween comes from
motor vehicle accidents. Children aged 5 to 14 years are four times more likely to
be killed in a pedestrian–motor vehicle accident on Halloween than on any other day
of the year (Centers for Disease Control, 1997). Regarding criminal activity on
Halloween, alcohol-related offenses and vandalism are particularly common (Siverts,
2002). Although we do not know the precise amount of law enforcement resources
consumed by Halloween sex offender policies, it will be important for policy makers
to estimate and consider allocation of resources in light of the actual increased risks
that exist in other areas, such as pedestrian–vehicle fatalities. Our findings indicated
that sex crimes against children by nonfamily members account for 2 out of every
1,000 Halloween crimes, calling into question the justification for diverting law
enforcement resources away from more prevalent public safety concerns.
Study results should be considered in light of the strengths and limitations. The
data drawn from NIBRS captured more than 67,000 sex offenses reported to law
enforcement agencies in 30 states across 9 years and did allow exclusion of familial
or similar sex crime cases that are not targeted by these policies. As with all second-
ary crime data sets, data entry error or jurisdictional idiosyncrasies might affect the
precision of the results. Also, the data set did not contain variables that would have
been valuable for even more precisely focusing the study incidents, such as informa-
tion about whether the perpetrator was a registered sex offender or whether the
incident specifically involved a trick-or-treat context. The data set did not provide
information about any situations where Halloween activities may have contributed
to grooming or set up an incident that took place on a later date. This study does,
however, provide the first empirical analysis of child sex crimes occurring on
Halloween and, therefore, can inform policy development, law enforcement resource
distribution, and case management practices.
References
Associated Press. (2006). States ban sex offenders from Halloween festivities. Retrieved November 21,
2006, from http://www.cnn.com
Bartkowiak, C., & Daniels, L. L. (2007). Sex offenders who live among us in northeast Florida. Retrieved
February 5, 2007 from http://sexoffenderstatistics.net/abstract.pdf
at ATSA on August 13, 2009 http://sax.sagepub.comDownloaded from
374 Sexual Abuse: A Journal of Research and Treatment
Bureau of Justice Assistance. (2008). Criminal victimization 2007. Washington DC: U.S. Department of
Justice.
Bureau of Justice Statistics. (2000). Sexual assault of young children as reported to law enforcement:
Victim, incident, and offender characteristics (No. NCJ 182990). Washington, DC: U.S. Department
of Justice.
Bureau of Justice Statistics. (2003). Recidivism of sex offenders released from prison in 1994 (No. NCJ
198281). Washington, DC: U.S. Department of Justice.
Bureau of Justice Statistics. (2004). National Crime Victimization Survey (No. NCJ205455). Washington,
DC: U.S. Department of Justice.
Centers for Disease Control. (1997). Childhood pedestrian deaths during Halloween—United States,
1975-1996. Morbidity and Mortality Weekly Report, 46, 987-990.
Cleveland, R. B., Cleveland, W. S., McRae, J. E., & Terpenning, I. (1990). STL: A Seasonal-Trend
Decomposition Procedure Based on Loess. Journal of Official Statistics, 6, 3-73.
Conte, J. R., Wolf, S., & Smith, T. (1989). What sexual offenders tell us about prevention strategies. Child
Abuse & Neglect, 13, 293-301.
Duwe, G., Donnay, W., & Tewksbury, R. (2008). Does residential proximity matter? A geographic analy-
sis of sex offense recidivism. Criminal Justice and Behavior, 35, 484-504.
Finkelhor, D., & Jones, L. M. (2004). Explanations for the decline in child sexual abuse cases.
Washington, DC: Office of Juvenile Justice and Delinquency Prevention.
Glassner, B. (1999). The culture of fear. New York: Basic Books.
Hall, G. C. N. (1995). Sexual offender recidivism revisited: A meta-analysis of recent treatment studies.
Journal of Consulting and Clinical Psychology, 63, 802-809.
Hanson, R. K., & Bussiere, M. T. (1998). Predicting relapse: A meta-analysis of sexual offender recidi-
vism studies. Journal of Consulting and Clinical Psychology, 66, 348-362.
Hanson, R. K., & Morton-Bourgon, K. (2004). Predictors of sexual recidivism: An updated meta-analysis.
Ottawa, Ontario, Canada: Public Works and Government Services.
Harris, A. J. R., & Hanson, R. K. (2004). Sex offender recidivism: A simple question (No. 2004-03).
Ottawa, Ontario, Canada: Public Safety and Emergency Preparedness Canada.
Jones, K. D., & Finkelhor, D. (2003). Putting together evidence on declining trends in sexual abuse:
A complex puzzle. Child Abuse & Neglect, 27, 133-135.
Kriel, L. (2008). Roundup targets sex offenders. Retrieved July 12, 2008, from http://www.mysanantonio
.com/news/local_news/Roundup_targets_sex_offenders.html
Levenson, J. S., Brannon, Y., Fortney, T., & Baker, J. (2007). Public perceptions about sex offenders and
community protection policies. Analyses of Social Issues and Public Policy, 7, 1-25.
Maguire, K., & Pastore, A. L. (2003). Sourcebook of criminal justice statistics 2003. Washington, DC:
U. S. Department of Justice, Bureau of Justice Statistics.
O’Connor, A. (2005, October 26). Sex offenders see new limits for Halloween. New York Times.
Salter, A. (1995). Transforming trauma. Thousand Oaks, CA: Sage.
Sample, L. L., & Kadleck, C. (2008). Sex offender laws: Legislators’ accounts of the need for policy.
Criminal Justice Policy Review, 19, 40-62.
Siverts, J. B. (2002). IVFP reports mild 2002 Halloween crime rates. Daily Nexus, Vol. 33, p. 1.
Walberg, M. (2006, November 1). No tricks, treats at Halloween “parties”. Chicago Tribune.
Wood, S. (2006, October 10). Halloween curfew revived for sex offenders. Philadelphia Inquirer.
at ATSA on August 13, 2009 http://sax.sagepub.comDownloaded from
