Production and Active Trading of Child Sexual Exploitation Images Depicting Identified Victims

See discussions, stats, and author profiles for this publication at: https://www.researchgate.
net/publication/325668273
Production and Active Trading of Child Sexual Exploitation Images Depicting

Identiﬁed Victims
Technical Report · January 2018
CITATIONS READS
6 519
4 authors, including:
Michael C Seto Ethel Quayle

University of Ottawa The University of Edinburgh
261 PUBLICATIONS 9,365 CITATIONS 97 PUBLICATIONS 2,003 CITATIONS
SEE PROFILE SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Incest aversion, Inbreeding avoidance, and Intra-familial sexual abuse View project
Child Pornography Offender Risk Tool (CPORT) View project
All content following this page was uploaded by Michael C Seto on 22 June 2018.
The user has requested enhancement of the downloaded file.

MARCH 2018
Production and Active Trading of Child Sexual

Exploitation Images Depicting Identified Victims
RESEARCH TEAM IN COOPERATION WITH
Michael C. Seto, Royal Ottawa Health National Center for Missing

Care Group & Exploited Children®
Cierra Buckman, Johns Hopkins University

FUNDED BY
R. Gregg Dwyer, Medical University
of South Carolina Thorn
Ethel Quayle, University of Edinburgh

Table of Contents
3 Executive Summary 15 Results | Historic Dataset
5 Project Aim 21 Results | Modern Dataset
6 Background 42 Conclusions
8 Data Collection Process 45 References
14 NCMEC Datasets 47 Appendix
NCMEC & Thorn Research Report

Executive PAGE
Summary 3
The National Center for Missing & Exploited Children (NCMEC) has
DATASETS
access to unique data about child exploitation images, particularly
1) HISTORICAL
those involving identified victims and offenders and reported by
• Actively traded cases
involving identified multiple law enforcement agencies.
victims: July 1, 2002
— June 30, 2014
• 518 cases involving
Through the cooperation of NCMEC, and with actively traded cases involving identified victims
933 identified victims the financial support of Thorn, we were able to from July 1, 2002, to June 30, 2014 (518 cases
extract data from NCMEC databases to address involving 933 victims); (2) a modern dataset
2) MODERN multiple research questions. encompassing all cases involving identified
• All cases involving victims from July 1, 2011, to June 30, 2014 (1,965
identified victims: The primary objective in this project was to
cases involving one offender and one victim, and
July 1, 2011 — June develop knowledge to assist law enforcement
30, 2014 633 cases involving multiple offenders and/or
in identifying victims of child sexual abuse
• 1,965 cases: one victims; only a small minority of these cases were
material and intervening to prevent child sexual
offender and one victim actively traded). The historical set allowed us to
exploitation and abuse.
• 633 cases: multiple examine trends over time, whereas the modern
offenders and/or victims dataset had more information due to a more
In this study, the first of its kind, we were able
to analyze data from two different datasets: comprehensive law enforcement submission
(1) a historical dataset that encompassed all form implemented in 2011.

Executive PAGE
Summary 4
Report Highlights
1 The most notable historical finding was a trend toward more egregious sexual content over
time, with more cases involving explicit sexual conduct in later years. In contrast, there were
no obvious trends in terms of child victim age or gender.
2 In cases involving a single victim and single offender, actively traded cases were associated
with having prepubescent victims. Actively traded cases were also associated with more
egregious content in terms of sexual activity, and more likely to involve familial offenders,
particularly nuclear family members.
3 While most cases involved male offenders who were unrelated to the child, cases involving
female offenders, younger children, or more egregious content were more likely to involve
familial offenders.
These results have implications for law enforcement investigations regarding adult-created child
pornography cases through a better understanding of the relationships between child, offender,
and offense characteristics.

Project PAGE
Aim 5
The primary objective in this project was to develop knowledge to assist

law enforcement in identifying victims of child sexual abuse material
and intervening to prevent child sexual exploitation and abuse.
WE WERE ABLE TO EXAMINE THE FOLLOWING RESEARCH QUESTIONS:
1. From data about all identified, actively traded single victim, and then including cases involving
cases (involving 5 or more reports to NCMEC) multiple offenders and/or victims:
from 2002 to 2013, are there longer-term
2. Was actively traded status (5 or more
trends in the nature of the content analyzed
reports to NCMEC) associated with victim
by NCMEC, in terms of the age, gender, or
age or gender, offender age or gender,
sexual activity involving depicted children?
sexual activity level, or relationship between
offender and victim?
rom data involving identified child victims
F
3. Given some offenders were family members
between July 1, 2011, and June 30, 2014, whether
of victims, was familial relationship
actively traded or not, we examined the following
associated with victim age or gender,
research questions, first by focusing on data
offender gender, or sexual activity level?
from cases involving a single offender and a

Background PAGE
Online child sexual exploitation is an international sociocultural contexts such as ethnicity or

problem, because these production and poverty. Existing research has provided some
distribution technologies transcend national insight about cohorts of identified children, or of
borders. There is also wide variation in child images seized from the computers of offenders,
pornography laws around the globe, according but there are many unaddressed questions. More
to a recent review by the International Centre for is known about perpetrators than victims at this
Missing & Exploited Children (2016). There are time (e.g., Long et al., 2016; Quayle & Jones, 2011;
growing concerns about the sexual exploitation Seto, 2013; Seto & Eke, 2015, in press; Seto, Wood,
and abuse of children as new technologies Babchishin & Flynn, 2012; Taylor & Quayle, 2003).
create more opportunities for perpetrators (Seto,
Knowledge gained from this research can lead
2013). Although the total number of children
to a better understanding of online victimization
who have been sexually exploited or abused and
and offending, thereby supporting more effective
photographed is unknown, the number of arrests
and efficient prevention and law enforcement
for adult-produced child pornography in the U.S.
initiatives to protect children. The results of
practically doubled between 2000 and 2009,
this research could have international impact
resulting in more than 1/3 of arrested producers
given the professional networks the researchers
of child pornography in 2009 being adults who
have in their respective countries and the
created the images of the children themselves
relationships NCMEC has with the international
(37%) (Wolak, Finkelhor, & Mitchell, 2012).
law enforcement community. For example, there
There is limited scientific understanding of are many questions about longer-term trends
the characteristics of these children who are in the production of child exploitation content,
victimized in child pornography images/videos and whether more actively traded content differs
and their relationship with those who have from non-traded content in victim or offender
sexually abused them. It is not known how these characteristics.
factors may change over time or vary across
Background | 7
Given younger children appear to be at greater age, are related to distribution or other offending
risk of sexual abuse by family members than from characteristics would support the development
non-relatives, reflecting access and opportunity of computer algorithms to categorize large
(Snyder, 2000), is it also the case that exploitation collections of child exploitation images. Seto and
content depicting younger children are more Eke’s (2015) predictive research has shown the
likely to involve familial offenders? Finding this ratio of boy to girl content is associated with the
association could help guide law enforcement likelihood of future sexual offending, so finding
investigations. As another example, evidence associations between child gender and other
that child characteristics, such as gender and study variables would extend this research.
THE TEAM
Principal Investigator: Michael Seto, PhD, Moore Center for the Prevention of Child Sexual
forensic research director at the Royal Ottawa Abuse, Bloomberg School of Public Health,
Health Care Group and an Associate Professor Johns Hopkins University (USA).
in Psychiatry at the University of Toronto, with
The work was completed in collaboration with the
cross-appointments to Ryerson University,
National Center for Missing & Exploited Children
Carleton University, and the University of
(NCMEC) staff.
Ottawa (Canada).
This project was funded by Thorn, a nonprofit
Co-Investigators: R. Gregg Dwyer, MD, EdD,
organization dedicated to driving technology
Associate Professor and Director of Community
innovation to combat child sexual exploitation.
and Public Safety Psychiatry, Director of the
Thorn partners with nonprofits and academic
Sexual Behaviors Clinic and Lab, Department
institutions to gather new insights into the role
of Psychiatry and Behavioral Sciences, Medical
technology plays in child sex trafficking, the
University of South Carolina (USA).
creation and proliferation of child pornography,
Ethel Quayle, PhD, Reader, School of Health and the normalization of child sexual exploitation.
and Social Science, University of Edinburgh Thorn then goes beyond insight to action to
(Scotland). develop the tools, systems, and approaches to
help address these issues (learn more at
Research Coordinator: Cierra Buckman, MHS,
www.wearethorn.org).
Senior Research Program Coordinator at the

Data Collection PAGE
Process 8
Database
NCMEC is a private, nonprofit organization When NCMEC introduced CVIP in 2002, the
established in 1984 (learn more at www. record-keeping for identified children was
missingkids.org). It was created to help find effective, but basic. As CVIP’s reputation and
missing children, reduce child sexual exploitation, recognition grew, so did their program. Law
and prevent child victimization. NCMEC serves as enforcement began seeking their assistance on
the national clearinghouse for families, victims, hundreds and thousands of cases. At this point,
private industry, law enforcement, and other CVIP moved their records to a case management
professionals on issues related to missing and system and formalized submissions with a law
sexually exploited children. NCMEC’s Exploited enforcement submission guideline form. The
Children Division operates the CyberTipline® and form’s first edition was still rather simple and
Child Victim Identification Program® (CVIP®). asked for standard information, such as data
CVIP primarily helps to verify whether or not about the victims involved in the case, the
child exploitation images appear to depict jurisdiction, a point of contact and such. [Please
children who have been previously identified note, NCMEC does not request nor record the
by law enforcement agencies as actual (rather names of victims in any of its systems.] However,
than virtual or computer-generated) children the process quickly became more sophisticated
and helps law enforcement identify new child and they added new variables and categories
pornography victims. CVIP maintains a database to each iteration of the guideline form. By 2014,
of information related to child sexual exploitation the submission guideline contained multiple
images, containing both identified and pages and additional case details were requested
unidentified children. and captured in the system. Data from each
submission was now entered into its respective
field and images were coded by analysts based
on their content.
Data Collection Process | 9
The CVIP database provides a unique opportunity Case/Series — A series is a group of images
for research related to online sexual exploitation focusing on a specific child(ren) and, when traded
and abuse of children because it is a central or collected, is most often done so as a set. “Series”
repository for data that crosses geographical, and “case” are synonymous within this report.
jurisdictional, and operational lines. CVIP
Actively Traded — A term designated by NCMEC
works in conjunction with several national
referring to a series that has been seen in 5 or
and international organizations to move cases
more CyberTipline Reports and/or CVIP case
along as quickly as possible. They also work
reviews.
closely with law enforcement agencies, internet
service providers, victim attorneys, and child Victim — Any child visible in the material who
welfare organizations throughout the U.S. They is younger than 18. Please note, only identified
have processed millions of cases and helped victim data have been included in this analysis.
to identify thousands of child sexual abuse
victims. As a result, their database has access Offender — The person who is proven or believed
to very large, broad, and unedited datasets, to have produced and/or enticed or coerced the
including information about child pornography images to be produced by a child.
collections, victim characteristics, and offender Age Category — This is coded based on physical
characteristics. development of the victim (limited to three
categories: infant/toddler, prepubescent, and
Vocabulary pubescent) based on the youngest appearance of
Given the evolution of the NCMEC database, the a child in a series of images or videos.
information in the CVIP records was not originally The database also includes other high-interest
intended or designed for research purposes. As variables, some which are coded on a series-level
such, historically they used in-house language and some which are coded on an individual-
for variables. In an effort to fully capture the level. Those coded on a series-level reflect data
nuances of their database, we have adopted their for the series overall regardless of number of
language for certain variables and categories victims and/or offenders. Those coded on an
as well. Below are a few key definitions to help individual-level reflect data for that specific child
readers understand the data as we describe our victim or offender. Below is a list of some of these
collection process and analytical methods. variables, and how they were coded.

VARIABLE CODING LEVEL DESCRIPTION
Gender Individual Gender of child victim/offender
Ethnicity Individual Ethnicity of child victim/offender
Relationship of Abuse Categorization of the relationship between each child

Individual
to Child victim and each offender
Specific U.S. state or “International” designation of
Jurisdiction Series
where the files were produced
Date NCMEC Received Date NCMEC received the case submission by

Series
the Submission law enforcement
Approximate Time Date provided by law enforcement indicating time

Series
Frame of Production frame of production of the series
Number of Images Series Approximate number of images and videos in the series
Categorization of the sexual activity depicted in the

Sexual Activity Checklist Series
images and videos
Additional fields on both individual- and series- videos is restricted, for the content variables, we
levels were also reviewed during the study. examined descriptors of series, counts of media,
However, the definitions of these variables and checklists of the sexual activity depicted.
are much more straightforward and align with Given the amount of missing information on
existing definitions in the research literature. early-identified victims, we chose to have
separate datasets from their database. Since we
Inclusion and Exclusion Criteria wanted to include some of the more historical
cases, in our first dataset, we chose to only
Data for our investigation was limited to cases
include actively traded cases with an identified
with an identified victim and one or more adult
victim. In our second dataset, we chose to
offenders. Since access to the images and
exclude cases before 2011.

Data Collection Method

The team’s research coordinator began working where there is only one identified victim-offender
onsite at NCMEC at the end of December 2014. relationship and cases with multiple identified
After a brief orientation, the lead analysts at victim-offender relationships. Using this layout,
NCMEC presented Ms. Buckman with several a closer examination of the relationship between
spreadsheets containing the raw variables for the offender and victims is possible, while still
first set of data. This first dataset, the historical answering questions surrounding case-level
dataset, contains all actively traded cases — information.
defined by NCMEC as having been reported
While CVIP’s services are extremely helpful to
on five or more times — from July 1, 2002, to
most law enforcement, many officers fail to
June 30, 2014. The second dataset, the modern
update cases with important information or
dataset, contains all cases involving identified
fully complete the law enforcement submission
child victims from July 1, 2011, to June 30, 2014.
form. To compensate for the missing data, Ms.
As discussed earlier, NCMEC receives a wide Buckman coded case documents for each series
range of information about a given case from to fill in as much missing information as possible
law enforcement, however, information is often and to capture additional variables that NCMEC
separated into different management systems. could not query from their data management
For instance, one system tracks information systems. The co-investigators determined that
related to images and videos, another tracks variables needed to be at least 80% complete
jurisdictional information and technical aspects to ensure reliable statistical calculations and
of a case, and a third stores information validity. At NCMEC’s suggestion, Ms. Buckman
concerning the series submission to NCMEC. developed a follow up survey for law enforcement
Given the method by which the variables were to collect the variables that still did not meet the
queried, compiling the data into one database 80% threshold.
would have greatly limited the analysis. Therefore,
A summary of the variables selected, scales for
the team opted for a more comprehensive
sexual activity, and scale for offender-victim
approach and asked Ms. Buckman to synthesize
relationship, is provided in the Appendix.
the raw data into two different databases: cases

Sample Selection
It is important to note, while this dataset is Furthermore, to ensure the goal of 80%
highly unique and offers a valuable insight into a completion per variable was achieved as well as
combination of perspectives (it combines offender, enough information for each perspective to work
victim, and content data), it also has limitations. with, NCMEC and the research team decided to
focus their research efforts on cases involving at
SAMPLE SELECTION OF LAW ENFORCEMENT least one identified child. This filter was applied
SUBMISSIONS
to both the historical dataset and the modern
dataset. As such, victim information is only
reflective of the victims identified in the case, not
Cases received necessarily all victims present in a case. Likewise,
by NCMEC CVIP information related to offenders is limited to
those who have been reported to NCMEC by
law enforcement. Additionally, self-produced
cases were excluded in the analyses since their
relationships represented a unique type of case.
Cases identified
by NCMEC CVIP Moreover, since some cases involve multiple
victims and some cases involve multiple
offenders or both, the data was looked at from
two perspectives: cases involving one victim-
Actively traded offender relationship and cases with multiple
cases reported victim-offender relationships. These different
back to NCMEC 5 perspectives allowed the use of variables that had
or more times been coded at either the case level or individual
level. It is important to make the distinction as
to what different variables say in relation to one
First, because NCMEC is entirely reliant on law another. For instance, in the historical dataset, a
enforcement for their information, the dataset refined sample (one victim-offender relationship
excludes cases law enforcement did not pursue cases) was used to fully explain things such as
as well as information remaining unknown to sexual activity and jurisdiction, both reported
law enforcement. at a case level. However, in order to maximize
the number of cases to work with, the multiple information concerning how a case was initiated.
relationship data was coded (i.e., if a case had all To make analyses more digestible, many of the
girl victims the case was considered girl victim or variables supplied to the team were grouped
if it had all infant/toddler victims it was counted and scaled to aid in dissemination and assist in
as infant/toddler victim). For cases with mixed mapping findings onto existing literature. These
samples (e.g., victims were both boys and girls), scales and groupings are listed in the Appendix.
a third “mixed” variable was created. This coding
The final sample criterion, only applied to the
strategy was applied to victim age and gender,
historical dataset, was actively traded status
relationships of victims and offenders, and
as defined by NCMEC as having been seen in
offenders’ gender.
5 or more CyberTipline reports and/or case
The raw NCMEC data was extremely rich with submissions by law enforcement to CVIP. Due
detail. NCMEC precisely codes the relationship to changes in the law enforcement submission
between offenders and victims, marks checklists guidelines it was impossible for early, non-
for sexual activity present, and keeps a variety actively traded cases to match the level of detail
of case information from distribution method to of modern non-actively traded cases.

NCMEC PAGE
Datasets 14
The first dataset was a historical, cross- some of the older cases (including their case
DATASETS sectional slice encompassing all available actively documents) were missing information that is
traded cases involving identified victims from now included in the NCMEC database. Thus, the
1) HISTORICAL
July 1, 2002, to June 30, 2014. This provided a modern dataset had the benefit of being more
• Actively traded cases
involving identified
look at longer-term trends over time in the nature complete as well as larger than the historical
victims: July 1, 2002 of the production cases submitted to NCMEC. dataset. There were 1,965 cases involving one
— June 30, 2014 For the analyses presented in this report, we victim and one offender, and approximately
• 518 cases involving looked at 518 actively traded cases, which 7% (N=143) of those cases were actively traded.
933 identified victims
involved 933 victims. There were 633 cases that involved multiple
relationships between victims and offenders,
2) MODERN The second dataset was modern and
and approximately 12% (N=75) were actively
• All cases involving encompassed all cases involving identified child
identified victims: traded. The larger amount of cases gave
victims from July 1, 2011, to June 30, 2014.
July 1, 2011 — June increased confidence in statistically significant
30, 2014 The dates for this cross-sectional slice were
differences between subgroups and comparisons,
• 1,965 cases: one chosen based on the implementation of a more
such as male versus female victims, male
offender and one victim comprehensive law enforcement submission
versus female offenders, and familial versus
• 633 cases: multiple form in 2011. While the historical dataset shows
offenders and/or victims non-familial relationships.
interesting trends among actively traded cases,

Results PAGE
15
Historic
Dataset
Results | Historic Dataset 16
Results Historic Dataset

The NCMEC historical and modern datasets are In this report, we have focused our analyses on
unique and rich and could be used to address the modern dataset since it is larger and speaks
a number of different questions. In this report, to current trends in child sexual abuse material,
we explore the interaction between victim but have presented a few timewise trends from
characteristics, offender characteristics, actively the historical dataset in the tables below.
traded status, and the content of the material,
Table 1 shows the number of actively traded
with the ultimate goal of developing knowledge
cases involving at least one identified victim by
to aid NCMEC and law enforcement in their work.
year, using the year the case was first recorded
An overview of the project was presented at NCMEC.
in August 2015 at the Dallas Crimes Against
Children Conference (http://www.cacconference.
org). Preliminary results were presented at
the annual meeting of the Association for the
Treatment of Sexual Abusers (www.atsa.com),
which took place in Montreal, Quebec, in
October 2015.
Table 1 | Historic Dataset Distribution by Year
YEAR COUNT OF CASES
2002 and 2003 71
2004 and 2005 75
2006 and 2007 82
2008 and 2009 72
2010 and 2011 110
2012 and 2013 108

1
Figure 1 | Year Distribution by Sexual Activity
Figure 1 shows the sexual activity of each case across years. Of note, sexual activity is graded for the highest activity
depicted in a series of images. The association is explained by a greater prevalence of images at levels 3 or 4 in
later years.
1
Percentages don’t always sum to 100% due to rounding. The sexual activities scale is shown on the following page and in the Appendix as Table 4.

Table 2 | 4 Point Sexual Activity Scale
Nudity or erotic posing with no sexual activity. (Level 1 on SAP Scale)

1
• Fully clothed erotica
• Erotica present
• Exposed genitals or anus
• Exposed breasts or chest
• Other sexual explicit content (i.e. fetishes)
• Full nudity
Non-penetrative sexual activity between children, adults and children, or masturbation.

2 (Level 2+3 on SAP Scale)
• Licking
• Kissing
• Manual stimulation
• Oral copulation
Penetrative sexual activity between adults and children. (Level 4 on SAP Scale)
3
• Anal or vaginal penetration
• Ejaculation seen
Sadism or Bestiality (Level 5 on SAP Scale)

4
• Drugged / Sleeping
• Bestiality
• Bondage
• Defecation
• Urination

Figure 2 | Year by Victim Age

KEY INSIGHT
There is a
relative shift
for 2008-2009
with similar
proportions in
other years.
Figure 2 shows the age category of all identified number of actively traded cases. The association
victims across the year distribution. Of note, is explained by a shift to relatively more
some cases involve multiple victims, hence pubescent victims in 2008 - 2009, with more
why the number of victims is larger than the similar proportions in other years.

Figure 3 | Year by Victim Gender
KEY
Figure 3 shows the gender of all identified victims There was significant variation in the proportion
across the year distribution. Of note, some cases of male victims over time, but there was no
involve multiple victims, hence why the number obvious trend because the percentage increased
of victims is larger than the number of actively from the first period and then substantially
traded cases. decreased in the last period.

Results PAGE
21
Modern
Dataset
Results | Modern Dataset 22
Modern Dataset
In the modern dataset, we subdivided the data activity (40%), whereas those series with multiple
into cases that involved only one relationship (one relationships involved more level 3 content (30%).
victim and one offender) and cases that involved See Table 3 on the following pages.
multiple relationships (either multiple victims or
We have not included offender ethnicity in our
offenders or both).
general descriptive table or our analysis because
In the one relationship subgroup, victims were of the extent of the missing data; this variable
predominantly white (85%), pubescent (61%) did not meet our 80% threshold. This is most
females (76%) with non-familial relationships likely due to the fact that it is hard to judge the
(74%) to white (86%) male (98%) offenders. In ethnicity of depicted persons as a result of poor
our second perspective, which looked at cases photo quality or when only partial figures (i.e.,
with multiple victim-offender relationships, arms, legs) were visible. NCMEC coding was
victims were also predominantly pubescent (42%) conservative and so entries were only made if
female (62%) with non-familial relationships analysts were confident. The law enforcement
(59%) to male (82%) offenders. Most content in survey also had a low yield for this variable
the one-to-one series involved level 1 sexual as well.

Table 3 | Modern Dataset Characteristics
ONE ON ONE (N=1,965) MULTIPLE RELATIONSHIPS (N=633)
RELATIONSHIP BREAKDOWN
Not Family (Closer in Proximity) 37% (728) 59% (374)
Not Family (Unknown To Child)2 37% (723) -- (--)
Family (Extended Family) 16% (305) 21% (134)
Family (Nuclear Family) 11% (209) -- (--)
Mixed -- (--) 20% (125)
SEXUAL ACTIVITY
1 40% (764) 28% (173)
2 20% (376) 21% (129)
3 26% (490) 30% (183)
4 14% (271) 20% (125)
64 cases were coded as unclear 23 cases were coded as unclear
VICTIM AGE
Infant/Toddler 6% (112) 3% (22)
Prepubescent 33% (644) 31% (196)
Pubescent 61% (1,209) 42% (264)
Mixed -- 24% (151)
2
Unknown to child includes those some refer to as “strangers,” sex traffickers, and missing data (see Appendix - Table 5).
Table 3 | Modern Dataset Characteristics (continued)
ONE ON ONE (N=1,965) MULTIPLE RELATIONSHIPS (N=633)
VICTIM GENDER
Female 76% (1,486) 62% (393)
Male 24% (479) 22% (141)
Mixed -- (--) 16% (99)
OFFENDER GENDER
Female 2% (41) 3% (17)
Male 98% (1,726) 82% (494)
Mixed -- (--) 15% (87)
198 observations missing 35 observations missing
ACTIVELY TRADED STATUS
Not Traded 93% (1,822) 88% (558)

Traded 7% (143) 12% (75)

Modern Dataset: One Relationship

As noted in the sample selection section, generalizations about the case. The tables below
NCMEC records information at both the case and show the distribution of the data (N=1,965). If any
individual level. Since key variables are coded at data is missing from the comparison, it is also
the case level, such as sexual activity, we chose noted in the table. Statistically significant odds
to analyze the subset of cases with only one ratios are reported beneath the tables. If the odds
victim-offender relationship separately. These ratio is not presented, it can be assumed the
cases allow us to speak directly to the variables difference was not statistically significant.
involved rather than providing grouped values or
Figure 4 | One Relationship - Victim Gender by Actively Traded Status
KEY
There was no significant difference in actively

traded status based on victim gender.

Figure 5 | One Relationship - Victim Age by Actively Traded Status
KEY
Cases with a prepubescent victim were much more likely to be traded than either
infant/toddler or pubescent victims, suggesting this was the preferred age category.
Figure 6 | One Relationship - Offender Gender by Actively Traded Status
KEY
There was no statistically significant difference

between male and female offenders.

Figure 7 | One Relationship - Sexual Activity Scale by Actively Traded Status
KEY
Each one point increase in the sexual activity scale

was associated with a greater likelihood of being
actively traded.

Figure 8 | One Relationship - Relationship by Actively Traded Status
KEY (NON-FAMILIAL)
KEY (FAMILIAL)
Cases with a familial relationship were more most likely to be actively traded, and cases
likely to be actively traded than cases with a involving a person who is unknown to the victim
non-familial relationship. Additional analysis or with whom the victim is unacquainted to be
revealed the difference was explained by cases the least likely. Relationship coding is listed in
involving nuclear family members being the Table 5 (Appendix).
3
Unknown to child includes those some refer to as “strangers,” sex traffickers, and missing data (see Appendix - Table 5). NCMEC & Thorn Research Report
Figure 9 | One Relationship – Offender Gender by Relationship
KEY
Though male offenders far outnumbered

female offenders, cases involving female
offenders were much more likely to
involve a familial relationship.
Figure 10 | One Relationship – Victim Gender by Relationship
KEY
Cases involving female children

were more likely to involve a familial
relationship with the offender.

Figure 11 | One Relationship – Victim Age by Relationship
KEY
Cases involving pubescent victims were much

less likely to involve a familial relationship
(14% compared to 59% for infants/toddlers and
43% for prepubescent victims).

Figure 12 | One Relationship – Sexual Activity Scale by Relationship
KEY
Cases involving the most egregious content

(level 4) were much more likely to involve a
familial relationship between offender
and victim.

Cases Involving Multiple Offender-Victim Relationships
The findings presented below included cases “male only,” and if the case included both male
involving multiple offenders and/or multiple and female victims it would be marked “mixed.”
victims, unlike the one-to-one cases just This has also been applied to victim age, offender
described. To capture this, we added a “mixed” gender, and relationships. The figures below show
category: For example, if a case involved only the distribution of the data (N=633). If any data
female victims it would be marked “female only,” are missing from the comparison, it is also noted
if it involved only male victims it would be marked in the figure.
Figure 13 | Multiple Relationships – Victim Gender by Actively Traded Status
KEY
Cases with both male and female victims

(“mixed”) were more likely to be actively traded.

Figure 14 | Multiple Relationships — Victim Age by Actively Traded Status

KEY INSIGHT
Infant/toddler
only content was
the most likely
to be actively
traded.
Unlike the one-to-one analysis, where cases One possible explanation is that multiple
involving prepubescent victims were the most relationship cases in which images were actively
likely to be actively traded, infant/toddler traded were more likely to involve a family
content was the most likely to be actively member (see figure 17) than one-to-one cases
traded in cases involving multiple offenders and family members have more access to
and/or victims. infants or toddlers.

Figure 15 | Multiple Relationships – Sexual Activity Scale by Actively

Traded Status
KEY
Again, the likelihood of being actively traded increased with sexual activity level.

Figure 16 | Multiple Relationships – Offender Gender by Actively

KEY INSIGHT
Traded Status
Cases more
likely to be
actively traded:
• male and
female
offenders
• familial only
relationships Cases involving both male and female offenders were more likely to involve actively traded content.
Figure 17 | Multiple Relationships – Relationships by Actively Traded Status
Again, cases involving family offenders were more likely to be actively traded.

Figure 18 | Multiple Relationships – Offender Gender by Relationship
KEY
Figure 19 | Multiple Relationships – Victim Gender by Relationship
KEY

Figure 20 | Multiple Relationships – Victim Age by Relationship

KEY INSIGHT
Data shows
there is a relative
association
between
pubescent
victims and
non-familial
offenders.
The most notable result here is the relative

association between pubescent victims and
non-familial offenders, paralleling the finding
reported in Figure 11 (one relationship).

Figure 21 | Multiple Relationships – Sexual Activity by Relationship
KEY
There was no significant difference in each one point increase in the sexual activity scale when
comparing familial only to non-familial only, familial only to mixed, or non-familial only to mixed.

Figure 22 | Multiple Relationships – Victim Age by Offender Gender
KEY
Most victims in all age groups were victimized by male offenders.

However, cases involving infant/toddler or prepubescent victims
differed from other age groups by being more likely to involve
both male and female offenders.

Figure 23 | Multiple Relationships – Victim Gender by Offender Gender
KEY
There was no obvious trend in the relationship between offender and victim
gender. It was still the case that most offenders in these cases were male
(including mixed) and the majority of victims were female.

Figure 24 | Multiple Relationships – Sexual Activity by Offender Gender
KEY
There was no obvious trend in terms of sexual activity level relative to offender gender.

Conclusions PAGE
42
In the following, we first discuss the results from the historical dataset
KEY INSIGHT
and then from the modern dataset, because the data were organized
Most notable and examined differently. We then conclude with implications for policy
historical finding: and practice regarding law enforcement.
trend toward
more egregious Historical Dataset
sexual content The historical dataset suggests there has over the years, with two-thirds (63%) involving
over time. indeed been a shift toward more egregious girls. The only exception was in 2013, when boys
content over time, with more content rated were in the majority (Quayle, E., Jonsson, L.,
at levels 3 or 4 on the sexual activity scale in Cooper, K., Trayner, J. and Svedin, C-G., 2018).
later years. This is different from other analyses The Internet Watch Foundation (2013) data
of child images, such as those reported by Wolak indicated 26% of identified images were boys
et al. (2011, 2012), but results are not directly in 2011, going down to 11% over the following
comparable because we focused our analysis two years.
on cases involving adult producers and actively
traded content. Modern Dataset
The NCMEC data indicated variation but no Actively traded cases were associated with
obvious trend in the proportion of boy victims having prepubescent victims. Actively traded
over time. Wolak, Finkelhor, and Mitchell (2011) cases were also associated with more
reported a small decrease in boy victims between egregious content in terms of sexual activity,
2000 to 2006 (20% to 13%). With the INTERPOL and more likely to involve familial offenders,
database analysis, gender of the children varied particularly nuclear family members. A familial

Conclusions | 43
KEY INSIGHT offender-victim relationship was relevant in a and girl victims.

number of other ways besides being more likely
Though we are aware of cases where unrelated
to be actively traded. Cases involving familial
Cases involving male offenders contact women with custody of
relationships were more likely to involve female
female offenders offenders, female victims, more egregious
children in order to produce child pornography
were more likely content, and younger victims.
content that may then be distributed to others,
this is largely an unexplored phenomenon in the
to also involve:
Male offenders were much more common than scientific literature. These cases suggest female
• male offenders, female offenders in this dataset, similar to offenders are not explained by current models
• related victims, previous research on online offending samples of online offending developed using research
• younger victims, (see Babchishin, Hanson, & VanZuylen, 2015). with male offenders. Case studies of female
• and to be actively Nonetheless, the involvement of female producers of child pornography suggest their
traded offenders was important when examining criminal conduct is not related to pedophilia (Prat,
cases involving multiple offender-victim Bertsch, Chudzik & Réveillère, 2014). Prat et al.
relationships. Cases involving female offenders hypothesized that producing images allowed
were more likely to also involve male offenders, women perpetrators to meet the desires or needs
related victims, younger victims, and to be of their romantic partners, which was important
actively traded. enough to the women to overcome any inhibitions
against exploiting children in their care.
Integrating these different findings, the
pattern of associations we observed indicates This pattern of results can be interpreted in light
male offenders are the most common, as of how we understand access and opportunity
demonstrated in multiple prior studies. These play a role in child sexual exploitation and abuse.
male offenders are more likely to target girls who Through social media and other online channels,
are unrelated to them, especially pubescent girls. men can have contact with pubescent children,
However, there were also cases involving both particularly girls (Whittle, Hamilton-Giachritsis,
male and female offenders. Different associations Beech & Collings, 2013; Quayle, Allegro, Hutton,
suggest these cases involve mostly unrelated Sheath & Lööf, 2014; Winters, Kaylor & Jeglic,
male offenders co-offending with female 2017). Some men who are interested in younger
offenders who were more likely to be related children, however, may contact and eventually
to victims, especially younger victims. These conspire with women who have contact with
co-offending cases, denoted by mixed gender younger children.
offenders, are more likely to have both boy
Conclusions | 44
Implications for Law Enforcement

KEY INSIGHT
Given limited law enforcement resources relative A significant finding is that prepubescent children
to the number of tips, reported contacts, and are sought more than other age groups from
Images depicting responses to personas, decisions have to be this study’s sample, and the finding that victim
younger children made about how to allocate time and effort. cases involving young children are more likely
and/or more Decision-making algorithms would be helpful to involve at least one perpetrator related to the
egregious in this regard, and was the rationale for the child and both male and female perpetrators.
development of the Child Pornography Offender Our findings also suggest images depicting
content are
Risk Tool (CPORT: Seto & Eke, 2015). However, younger children and/or more egregious
more likely to producers of child exploitation content were a content are more likely to involve familial
involve familial minority of the sample in the development of the offenders. This information could be useful to law
offenders and CPORT, and there may be important differences enforcement during victim identification efforts
were more between producers and possessors of this content. and investigations, perhaps keeping in mind that
while looking for very young victims, the offender
likely to be Knowing more about those engaging in
is relatively likely a family member.
actively traded. production of child exploitation images can serve
to improve the efficiency of law enforcement Another finding relevant for law enforcement is
operations. Knowing what type of images are most not all images are traded with equal frequency.
likely to be produced can help law enforcement Those with the most egregious content and
plan their investigations accordingly. Using a those with a familial relationship between
large dataset, this study has replicated and victim and perpetrator were more likely to
extended previous research. be actively traded.

Conclusions | 45
References
Babchishin, K. M., Hanson, R. K., & VanZuylen, H. International Center for Missing & Exploited
(2015). Online child pornography offenders are Children. (2016). Child Pornography: Model
different: A meta-analysis of the characteristics Legislation & Global Review, 8th Edition.
of online and offline sex offenders against Available at: http://www.icmec.org/wp-content/
children. Archives of Sexual Behavior, 44, 45-66. uploads/2016/02/Child-Pornography-Model-
Law-8th-Ed-Final-linked.pdf
Burgess, A. W., Carretta, C. M., & Burgess, A. G.
(2012). Patterns of federal Internet offenders: A Internet Watch Foundation (2013). Annual Report.
pilot study. Journal of Forensic Nursing, 8, 112-121. Available online from https://www.iwf.org.uk/
sites/default/files/reports/2016-03/ar_final_
Burgess, A. W., Mahoney, M., Visk, J., &
web_low%20res.pdf.
Morgenbesser, L. I. (2008). Cyber child sexual
exploitation. Journal of Psychosocial Nursing & Internet Watch Foundation (2016). Annual
Mental Health Services, 46, 38-45. Report. Available online from https://www.iwf.
org.uk/sites/default/files/reports/2017-04/iwf_
DeHart, D., Dwyer, G., Seto, M. C., Moran, R.,
report_2016.pdf.
Letourneau, E., & Schwarz-Watts, D. (2017).
Internet sexual solicitation of children: A Long, M., Alison, L., Tejeiro, R., Hendricks, E., &
proposed typology of offenders based on their Giles, S. (2016). KIRAT. Psychology, Public Policy,
chats, E-mails, and social network posts. Journal and Law, 22, 12-21.
of Sexual Aggression, 23, 77-89.
Prat, S., Bertsch, I., Chudzik, L., & Réveillère, C.
DeLong, R. L. (2012). Developing a typology for (2014). Women convicted of a sexual offence,
understanding internet sexual offenders. Doctoral including child pornography production: Two case
dissertation, Walden University, Minneapolis, MN. reports. Journal of Forensic and Legal Medicine,
23, 22.
Dwyer, R. G., Seto, M. C., DeHart, D., Letourneau,
E., McKee, T., & Moran R. (2016). Protecting Quayle, Allegro, Hutton, Sheath, & Lööf. (2014).
children online: Using research-based Rapid skill acquisition and online sexual grooming
algorithms to prioritize law enforcement Internet of children. Computers in Human Behavior, 39,
investigations (Research Report). Washington, 368-375.
DC: U.S. Department of Justice.

Conclusions | 46
Quayle, E., & Jones, T. (2011). Sexualized images Shelton, J., Eakin, J., Hoffer, T., Muirhead,
of children on the Internet. Sexual Abuse: A Y., & Owens, J. (2016). Online child sexual
Journal of Research and Treatment, 23, 7-21. exploitation: An investigative analysis of
offender characteristics and offending behavior.
Quayle, E., Jonsson, L., Cooper, K., Trayner, J.
Aggression and Violent Behavior, 30, 15-23.
and Svedin, C-G. (2018). Children identified in
sexual images - who are they? Self and non-self Snyder, H. N. (2000, July). Sexual assault of
taken images in the International Child Sexual young children as reported to law enforcement:
Exploitation Image Database 2006-2015. Child Victim, incident, and offender characteristics
Abuse Review.DOI:10.1002/car.2507. (Report NCJ 182990). Bureau of Justice
Statistics, U. S. Department of Justice, Office of
Seto, M. C. (2013). Internet sex offenders.
Justice Programs. Retrieved from https://www.
Washington, DC: American Psychological
bjs.gov/content/pub/pdf/saycrle.pdf.
Association.
Taylor, M., & Quayle, E. (2003). Child pornography:
Seto, M.C., & Eke, A.W. (2015). Predicting
An Internet crime. New York: Brunner-Routledge.
recidivism among adult male child pornography
offenders: Development of the Child Pornography Whittle, Hamilton-Giachritsis, Beech, & Collings.
Offender Risk Tool (CPORT). Law and Human (2013). A review of young people’s vulnerabilities
Behavior, 39, 416-429. to online grooming. Aggression and Violent
Behavior, 18, 135-146.
Seto, M. C., & Eke, A. W. (in press). Correlates of
admitted sexual interest in children among child Winters, G., Kaylor, L., & Jeglic, E. (2017). Sexual
pornography offenders. Law and Human Behavior. offenders contacting children online: An
examination of transcripts of sexual grooming.
Seto, M. C., Hanson, R. K., & Babchishin, K. M.
Journal of Sexual Aggression, 23, 62-76.
(2011). Contact sexual offending by men arrested
for child pornography offenses. Sexual Abuse: A Wolak, J., Finkelhor, D., & Mitchell, K. J. (2012).
Journal of Research and Treatment, 23, 124-145. Trends in arrests for child pornography
production: The Third National Juvenile Online
Seto, M. C., Wood, J. M., Babchishin, K. M., &
Victimization Study (NJOV‐3). Retrieved from
Flynn, S. (2012). Online solicitation offenders are
https://scholars.unh.edu/ccrc/47/
different from child pornography offenders and
lower risk contact sexual offenders. Law and
Human Behavior, 36, 320-330.
Conclusions | 47
Appendix
Table 4 | 4 Point Sexual Activity Scale
Nudity or erotic posing with no sexual activity. (Level 1 on SAP Scale)

1
• Fully clothed erotica
• Erotica present
• Exposed genitals or anus
• Exposed breasts or chest
• Other sexual explicit content (i.e. fetishes)
• Full nudity
Non-penetrative sexual activity between children, adults and children, or masturbation.

2 (Level 2+3 on SAP Scale)
• Licking
• Kissing
• Manual stimulation
• Oral copulation
Penetrative sexual activity between adults and children. (Level 4 on SAP Scale)
3
• Anal or vaginal penetration
• Ejaculation seen
Sadism or Bestiality (Level 5 on SAP Scale)

4
• Drugged / Sleeping
• Bestiality
• Bondage
• Defecation
• Urination

Conclusions | 48
Table 5 | Relationship Scale
FAMILIAL Nuclear Family • Mother

• Father
• Brother
• Sister
• Half Sibling
Extended Family • Step-Father

• Step-Mother
• Aunt
• Uncle
• Grandfather
• Step-Grandparent
• Brother-in-Law
• Cousin
• Legal Guardian
• Other Relative
NON-FAMILIAL Close Proximity • Babysitter/mentor/coach/teacher

• Boyfriend
• Guardian’s Partner
• Neighbor/Family Friend
Unknown to Victim • No Relationship

• Online Enticement/Self & Perp Produced
• Photographer
• Sex Trafficker
• Stranger
• Unknown

MARCH 2018
Production and Active Trading of Child Sexual

Exploitation Images Depicting Identified Victims
View publication stats

Production and Active Trading of Child Sexual Exploitation Images Depicting Identified Victims

Uploaded by

Copyright:

Available Formats

Production and Active Trading of Child Sexual Exploitation Images Depicting Identified Victims

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Production and Active Trading of Child Sexual Exploitation Images Depicting Identified Victims

Uploaded by

Copyright:

Available Formats

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

Production and Active Trading of Child Sexual Exploitation Images Depicting

Technical Report · January 2018

Michael C Seto Ethel Quayle

SEE PROFILE SEE PROFILE

Child Pornography Offender Risk Tool (CPORT) View project

The user has requested enhancement of the downloaded file.

Production and Active Trading of Child Sexual

Michael C. Seto, Royal Ottawa Health National Center for Missing

Cierra Buckman, Johns Hopkins University

Ethel Quayle, University of Edinburgh

NCMEC & Thorn Research Report

NCMEC & Thorn Research Report

NCMEC & Thorn Research Report

The primary objective in this project was to develop knowledge to assist

WE WERE ABLE TO EXAMINE THE FOLLOWING RESEARCH QUESTIONS:

NCMEC & Thorn Research Report

Online child sexual exploitation is an international sociocultural contexts such as ethnicity or

NCMEC & Thorn Research Report

NCMEC & Thorn Research Report

VARIABLE CODING LEVEL DESCRIPTION

Gender Individual Gender of child victim/offender

Ethnicity Individual Ethnicity of child victim/offender

Relationship of Abuse Categorization of the relationship between each child

Date NCMEC Received Date NCMEC received the case submission by

Approximate Time Date provided by law enforcement indicating time

Categorization of the sexual activity depicted in the

NCMEC & Thorn Research Report

Data Collection Method

NCMEC & Thorn Research Report

NCMEC & Thorn Research Report

NCMEC & Thorn Research Report

Results Historic Dataset

Table 1 | Historic Dataset Distribution by Year

YEAR COUNT OF CASES

2002 and 2003 71

2004 and 2005 75

2006 and 2007 82

2008 and 2009 72

2010 and 2011 110

2012 and 2013 108

NCMEC & Thorn Research Report

Table 2 | 4 Point Sexual Activity Scale

Nudity or erotic posing with no sexual activity. (Level 1 on SAP Scale)

Non-penetrative sexual activity between children, adults and children, or masturbation.

Sadism or Bestiality (Level 5 on SAP Scale)

NCMEC & Thorn Research Report

Figure 2 | Year by Victim Age

NCMEC & Thorn Research Report

Figure 3 | Year by Victim Gender

NCMEC & Thorn Research Report

NCMEC & Thorn Research Report

Table 3 | Modern Dataset Characteristics

ONE ON ONE (N=1,965) MULTIPLE RELATIONSHIPS (N=633)

Table 3 | Modern Dataset Characteristics (continued)

ONE ON ONE (N=1,965) MULTIPLE RELATIONSHIPS (N=633)

ACTIVELY TRADED STATUS

Not Traded 93% (1,822) 88% (558)

NCMEC & Thorn Research Report