Immaterial Boys? A Large-Scale Exploration of Gender-Based Differences in Child Sexual Exploitation Service Users

Abstract

Child sexual exploitation is increasingly recognized nationally and internationally as a pressing child protection, crime prevention, and public health issue. In the United Kingdom, for example, a recent series of high-profile cases has fueled pressure on policy makers and practitioners to improve responses. Yet, prevailing discourse, research, and interventions around child sexual exploitation have focused overwhelmingly on female victims. This study was designed to help redress fundamental knowledge gaps around boys affected by sexual exploitation. This was achieved through rigorous quantitative analysis of individual-level data for 9,042 users of child sexual exploitation services in the United Kingdom. One third of the sample were boys, and gender was associated with statistically significant differences on many variables. The results of this exploratory study highlight the need for further targeted research and more nuanced and inclusive counter-strategies.

Keywords

child sexual abuse child welfare victim risk assessment sex offenses

Introduction

Once routinely dismissed as consensual child prostitution, the issue of child sexual exploitation (CSE) is increasingly recognized as a serious child protection and crime prevention concern (e.g., Barnardo’s, 2012; Barrett & Melrose, 2003; Chase & Statham, 2005). Demands are growing nationally and internationally for improved responses to CSE (see, for example, Appleton, 2014; Home Affairs Select Committee, 2013). In the United Kingdom, a particular catalyst for change has been intense media scrutiny of recent cases involving large groups of offenders and of revelations of historic incidents featuring celebrities and politicians. Police and local authorities have been heavily criticized for failing to identify and respond adequately to CSE (e.g., Harvey, Hornsby, & Sattar, 2015; Jago et al., 2011; Jay, 2014) and demand is high for evidence to improve approaches. CSE’s ascent up the political agenda has been accompanied by the publication of numerous national and local reports, guidance, and action plans (e.g., Association of Chief Police Officers, 2012; Child Exploitation and Online Protection Centre, 2011; Crown Prosecution Service, 2013; Department for Education, 2011; Home Affairs Select Committee, 2013; Home Office, 2011, 2015; Office of the Children’s Commissioner for England, 2012). Yet, robust empirical research on CSE remains in short supply.

In accordance with the United Nations Convention on the Rights of the Child (United Nations, 1989), “children” refers here to individuals aged 17 years or younger. In the United Kingdom, however, the age of consent is 16 years; whether sexual activity involving 16- or 17-year-olds constitutes abuse depends on additional factors such as consent, the nature of the act, and the relationship between the parties involved. CSE is often characterized as a distinct subset of child sexual abuse (e.g., Jago et al., 2011; Jago & Pearce, 2008). Yet, there is no internationally agreed definition of CSE, and accepted national definitions may inadequately delineate exploitation from abuse more broadly (Brayley & Cockbain, 2014; Cockbain, 2013). In the United Kingdom, CSE is not a distinct offense defined in criminal law, and professionals typically instead work to the definition found in government safeguarding guidance (e.g., Department for Children, Schools and Families, 2009; Welsh Assembly Government, 2010). Central to this definition is the concept of exchange, in that CSE is said to involve

. . . exploitative situations, contexts and relationships where young people (or a third person or persons) receive “something” (e.g. food, accommodation, drugs, alcohol, cigarettes, affection, gifts, money) as a result of them performing, and/or another or others performing on them, sexual activities. (Department for Children, Schools and Families, 2009, p. 9)

This definition’s inclusivity may help to promote awareness and identification of victims, but its breadth and vague parameters can become problematic in scientific enquiry. In practice, our experience in the United Kingdom has been that CSE is often implicitly associated with the extra-familial abuse of teenagers. In policy, practice, and even research, however, the above definition is often applied uncritically, without mention or consideration of its inherent potential for inconsistencies in interpretation and application. The empirical and theoretical literature on CSE is limited and fragmented, quantitative studies are particularly rare, and fundamental considerations such as incidence and prevalence rates have yet to be established. The underdevelopment of the research base probably reflects, at least in part, the relatively recent growth of interest in CSE, issues in data collection, collation, and access, and the lack of definitional clarity and consensus (see Cockbain, 2013). A particular knowledge gap around sexual exploitation involving boys has been highlighted, usually accompanied by calls for targeted research (e.g., Child Exploitation and Online Protection Centre, 2011; Department for Education, 2011; Public Petitions Committee, 2014). In light of the limited research specific to CSE, we will contextualize our study by briefly reviewing some key findings from the better developed literature on child sexual abuse more broadly.

It is well established that boys constitute a minority of child sexual abuse victims (e.g., Cashmore & Shackel, 2014; Finkelhor, 1994; Priebe & Svedin, 2009; Stoltenborgh, van Ijzendoorn, Euser, & Bakermans-Kranenburg, 2011). Diverse epidemiological studies into prevalence and studies based on pre-identified groups of victims have respectively concluded that boys are at less risk of sexual abuse than girls and that there are far fewer boys than girls in most¹ victim samples (for a review, see, for example, Cashmore & Shackel, 2014; Stoltenborgh et al., 2011). Despite these broad points of consensus, individual studies have differed markedly in terms of the relative risk to and/or representation of boys and girls documented. Such variation is often attributed to conceptual, definitional, and methodological differences and/or to disparities in geographical and socio-cultural contexts (Cashmore & Shackel, 2014; Paolucci, Genuis, & Violato, 2001; Stoltenborgh et al., 2011). In a seminal review of the child sexual abuse literature in 21 countries, Finkelhor (1994) reported prevalence rates ranging from 7% to 36% for girls and 3% to 29% for boys. In most of the studies included, girl-to-boy ratios fell between 1.5:1 and 3:1. In a more recent meta-analysis of child sexual abuse prevalence globally, Stoltenborgh et al. (2011) reported European prevalence rates of 13.5% for girls and 5.6% for boys. Although specific prevalence rates varied by continent (from 11.3% to 21.5% for girls and 4.1% to 19.3% for boys), rates were higher (generally markedly so) for girls than boys in every continent but South America. In a recent interview-based study with a U.K.-wide random probability sample of 2,275 eleven- to 17-year-olds, Radford et al. (2011) found that self-reported lifetime contact child sexual abuse victimization rates were 2.7 times higher among female respondents (7.0%) than males (2.6%).

Given the well-established predominance of female victims and male offenders in child sexual abuse (see, for example, Smallbone, Marshall, & Wortley, 2008), it is largely unsurprising that much of the literature is filtered “through the prism of victim as female and perpetrator as male” (Cashmore & Shackel, 2014, p. 75). Although recent years have seen a gradual expansion of the literature on male victims (see, for example, Alaggia, 2005; Cashmore & Shackel, 2014; Edelson, 2012; Homma, Wang, Saewyc, & Kishor, 2012; Priebe & Svedin, 2009; Scrandis & Watt, 2014), it remains the norm to focus on female victims. With some notable exceptions (e.g., Priebe & Svedin, 2009; Smallbone & Wortley, 2000), where large analyses of child sexual abuse have factored in gender as a variable, it has typically been in relationship to risk of victimization or to the consequences of abuse (e.g., Paolucci et al., 2001; Stoltenborgh et al., 2011). Nonetheless, there is a small but important body of research indicating a relationship between victim gender and the characteristics and context of abuse incidents. Unlike girls, boys have been found to be more likely to be sexually abused in extra-familial than intra-familial contexts (e.g., Feiring, Taska, & Lewis, 1999; Moody, 1999; Smallbone & Wortley, 2000). A higher (albeit still relatively low) proportion of boys than girls are sexually abused by female offenders (e.g., Finkelhor, 1986; Nelson & Oliver, 1998). Overall, however, “there is limited research that specifically compares the experiences of male and female victims” (Cashmore & Shackel, 2014, p. 76). As a result, considerable knowledge gaps persist in respect to the interplay between victim gender and their individual characteristics, the abuse process, and official responses to them.

There are clear empirical and theoretical reasons why victim gender could be an important factor to consider in understanding and responding to CSE. Yet, little has changed since Lillywhite and Skidmore (2006) highlighted the “persistent invisibility of boys” in public, political, practitioner, and academic discourse (p. 352). Amid the rapidly evolving landscape around CSE in the United Kingdom, a constant is the tendency to conceptualize the phenomenon as something perpetrated by men (or boys) against girls (see, for example, Cockbain, 2013). On one hand, the lack of research into the association between children’s gender and sexual exploitation might simply reflect the generalized underdevelopment of the CSE research literature. On the other hand, recent national studies into CSE in the United Kingdom that involved mixed-gender samples have missed the opportunity to investigate gendered differences (Child Exploitation and Online Protection Centre, 2011; Office of the Children’s Commissioner for England, 2012). Instead, descriptive analyses have been based on the samples as a whole, meaning the aggregate findings may inaccurately represent the characteristics of the male minorities of 13% (n = 271; Child Exploitation and Online Protection Centre, 2011) and 11% (n = 265; Office of the Children’s Commissioner for England, 2012), respectively.

It has been argued that CSE detection and disclosure rates are lower for boys than girls (Lillywhite and Skidmore, 2006). Among the commonly posited arguments for this are inadequate service provision, additional stigma associated with male sexual victimization, fear of homophobia, female-centric risk assessment tools, and limited awareness among professionals that boys can be sexually exploited (Forrest, 2007; Lillywhite & Skidmore, 2006; McNaughton-Nicholls, Harvey, & Paskell, 2014; Palmer, 2001). Although such claims have yet to be empirically investigated for CSE specifically, these barriers for boys are well documented in the child sexual abuse literature more broadly (see, for example, Homma et al., 2012). Consequently, it seems plausible that the limited official data on CSE particularly fails to capture the degree to which boys are affected. Nevertheless, the majority of current service provisions in the United Kingdom are “targeted towards young women rather than young men” (Lowe & Pearce, 2006, p. 289) and boys’ needs may be overlooked and under-served.

Amid this context, we designed this study as a large-scale empirical assessment of differences and commonalities between boys and girls accessing CSE support services in the United Kingdom. CSE is a sensitive and emotionally charged issue, and those affected by it are a vulnerable and hard-to-access research population (Palmer, 2001). Taking an unobtrusive approach to data collection, we were able to capitalize on an unusually extensive and detailed dataset spanning many thousands of CSE service users. As well as being among the first in-depth comparative analyses of gender and CSE, our study is unusual in its large sample size. Consequently, we expect the findings to have wider international relevance beyond the United Kingdom alone in informing future CSE research, policy, and practice.

Data

Our dataset derived from the national database of service users held by Barnardo’s, a non-governmental organization and the United Kingdom’s largest provider of CSE services. In the absence of a centralized national recording system for CSE, the Barnardo’s database (a centralized electronic case management system) is the single largest source of individual-level data about CSE in the United Kingdom. It is common in the United Kingdom for specialist support services for CSE and other such issues to be provided by non-governmental organizations, sometimes via directed governmental funding. Of course, access to such specialist services does not preclude more general involvement of statutory agencies in children’s welfare (e.g., via social workers or safeguarding teams).

We received raw data extracted from the Barnardo’s anonymized central database on November 1, 2013. This database includes information on service users supported for numerous different issues. To identify relevant service users for inclusion, Barnardo’s applied as a filter the tag “CSE”—a standardized tag used across all their services to highlight when sexual exploitation is involved.

We cleaned the dataset to remove duplicates and then went on to exclude a further 2,125 unique individuals who failed to meet our inclusion criteria (Table 1). Our final sample consisted of 9,042 unique CSE service users (also referred to here as cases): 6,056 females and 2,986 males from across 28 different services in England, Scotland, and Northern Ireland. Services from Wales, the United Kingdom’s fourth nation, did not feature in the study as they operate on a different case management system. The constraints of the original data limited us to examining differences between boys and girls, although we recognize that gender is a broader construct than the traditional male/female dichotomy alone.

Table 1.

Cases Excluded to Ensure Data Represented Only Sexually Exploited Children.

Inclusion criteria (applied sequentially in the order listed here)	Reason	Number of cases excluded
No apparent data entry mistake	To remove cases where values indicated a human error in entering the data	190
Age recorded	To allow filtering based on age	586
Individuals aged 17 years or younger when Barnardo’s began supporting them	To exclude sexually exploited adults erroneously recorded as CSE cases	960
Individuals aged 8 years or more when Barnardo’s began supporting them	To exclude very young children recorded in the database because they are the children of exploited adults	351
Gender recorded	To allow comparison of boys and girls	38
Total excluded cases		2,125

Note. CSE = child sexual exploitation.

Except for the excluded cases previously mentioned, our sample represents all children supported by Barnardo’s because of CSE between April 1, 2004, and November 1, 2013. We deliberately refer to this group as “service users” or children “affected by CSE” rather than as “victims.” The reason is that our sample included not only sexually exploited children but also children implicated in sexually exploiting other children (peer-on-peer CSE) and those at risk of either (or both) of these conditions. Our focus group participants described the threshold for being deemed “at risk” as quite high. They also stressed that referrals are based on a holistic assessment of individual children’s circumstances, rather than the presence of any given demographic factor associated with risk. On one hand, the breadth of the experiences covered in the sample is an asset because we were able to assess CSE in its entirety. On the other, we were limited by the fact that the original data did not distinguish into which category or categories each service user fell. Although Barnardo’s services all work to the same definition of CSE (Department for Children, Schools and Families, 2009), we cannot rule out the possibility of inconsistencies between and within services in how this definition is operationalized. In addition, services may vary in their capacity and criteria for accepting a referral, mainly due to different funding arrangements. The services do not cover the whole of the United Kingdom but rather certain geographical areas only.

In shaping our analysis, we were constrained by the design of the original data collection system. For confidentiality reasons, we are unable to share the data entry templates but instead will briefly explain their nature, provenance, and contents. Caseworkers enter into the database information about each service user at the point of starting work with him or her. They derive this information from sources including their own assessment of a child; discussions with the child, his or her parents, carers, or various practitioners (police, teachers, social workers, etc.); and written notes from formal records (e.g., safeguarding meetings). Caseworkers can revisit and revise these individual-level records should new information emerge.

In fact, two distinct types of records exist for CSE cases. We were able to match individuals across them as the same unique identifier is used in both. The first type of record (referred to here as the core record) is used across the board and covers various fundamental aspects about the service user and the support provided (e.g., age, gender, date of referral). The second (here, the additional record) comes from an extra (CSE specific) form that is used only by certain services. It contains additional fields—selected based on theoretical and experiential knowledge as especially relevant to CSE cases—such as a service users’ youth offending history and his or her peers’ involvement in exploitation.

Data from the additional records were supplied for 2,951 cases (33% of the overall study sample): 36.2% (n = 2,198) of the females and 25.2% (n = 753) of the males in the sample. This difference was significant, χ²(1, N = 9,042) = 111.61, p < .001, V = 0.11, but not meaningful as it simply reflects variation between services in the gender composition of those they support. The use of the additional record is not a function of systematic differences between services or individual cases but simply stems from local variation in data-recording practices—some services use this extra form, others do not. We cannot be certain that the subsample for which the extra records were available is representative of the Barnardo’s CSE service-user population at large. Nevertheless, we decided to include the variables from the additional records in our analyses due to their valuable information about factors rarely accessible to researchers at this scale and the potential to stimulate further targeted research.

The two sets of records together covered a diverse range of variables, including characteristics relating to service users’ personal and demographic traits, the exploitation process, and official responses. We excluded certain variables for reasons such as low completion rates, duplication/overlap with other variables, and lack of relevance to an academic study. Others we combined to generate information more relevant to our study. Our final set of variables and the particular source (core or additional records) from which they originated are shown in Table 2.

Table 2.

Independent Variables.

Variable related to . . .	Core records	Additional records
Child	Age at referral Ethnicity Disability Looked-after child status	Youth offending history Involvement in gun or knife crime
Exploitation process		Peer involvement in exploitation
Official responses	Service providing careRegion/nation where service locatedSource of referralWaiting time	Reason for referral

Research Questions

In line with our exploratory approach, we deliberately framed our research question in broad and inclusive terms:

Research Question 1: What are the differences and commonalities between the recorded characteristics of males and females affected by CSE and supported by Barnardo’s services in the United Kingdom?

In doing so, we capitalized on an extensive and previously untapped dataset and the opportunity to expand the limited knowledge base on CSE. Although exploratory approaches can rarely provide conclusive answers, the insights they can generate are clearly superior to those gained (or lost) by research inactivity or reliance on untested popular wisdom. Such is especially the case when dealing with a notoriously hard-to-access research population such as children affected by CSE, an issue around which the demand for research evidence to inform policy and practice far outstrips what little is available.

There were several reasons why we felt it inappropriate to frame the study around a strict hypothesis, or set of hypotheses. First, there was a limited body of prior research on which to build. Second, using secondary data inherently limits what questions can be set and tested empirically. Third, there was substantial variation within our data in terms of the volume, relevance, and quality of evidence available for each variable to inform any hypotheses about their possible relationship to gender. A hypothesis-driven approach might have been feasible for some of the comparatively better documented variables but might have excluded other interesting and potentially informative factors that had little precedent in the research literature.

Despite our broad research question, we had some expectations about the relationships we might find between gender and the different independent variables. These are set out in turn below.

Age at Referral

From the child sexual abuse literature, there was reason to believe gender might be significantly associated with differences in age. More specifically, there was some evidence to suggest that the male CSE service users might be younger on average than their female counterparts. An important (albeit now somewhat dated) epidemiological study in Philadelphia demonstrated marked differences in the age distribution of male and female victims of child sexual abuse (De Jong, Hervada, & Emmett, 1983). The distribution for girls (n = 463) was bimodal with peaks at 6 and 15 years, whereas for boys (n = 103), the distribution had a single peak at 7 years.

Ethnicity

It has been suggested that sexual exploitation of Black and minority ethnicity (BME) children is particularly underrepresented in official records on CSE (Home Affairs Select Committee, 2013; Office of the Children’s Commissioner for England, 2012; Smeaton, 2013). Particular difficulties are said to exist in identifying children in BME communities who are affected by CSE (see, for example, Gohir, 2013; Ward & Patel, 2006). Such contentions could be expected to translate into a general under-representation of BME children in our sample, but we found no theoretical or empirical grounds to expect an association between service users’ gender and their ethnicity.

Disability

We had grounds to believe that the rates of disability (physical, cognitive, and/or emotional/behavioral) might be higher for the boys than the girls in our sample—as well as generally exceeding the national prevalence rate. The past two decades have seen increased interest in the possible association between disability and childhood maltreatment and abuse, including but not limited to child sexual abuse. In a major population-based epidemiological study (n = 50,278) in Omaha, Nebraska, Sullivan and Knutson (2000) reported maltreatment of 31% (n = 1,012) of disabled children compared with just 9% (n = 3,491) of non-disabled children. No disaggregated results were presented for males and females. In a study of children examined for suspected child sexual abuse in Norwegian hospitals in the mid-1990s (n = 1,293), Kvam (2000) found that 10% (n = 29) of boys examined had disabilities compared with 5% (n = 54) of girls. Moreover, the proportion of boys in the disabled group (35%) was higher than in the non-disabled group (22%), indicating a particular association may exist between gender, disability, and child sexual abuse.

Looked-After Child Status

Although looked-after children have been highlighted as a high-risk group for involvement in CSE (e.g., Child Exploitation and Online Protection Centre, 2011; Office of the Children’s Commissioner for England, 2012), we could find no prior research examining empirically an association with children’s gender. Looking simply at the closest available national baselines, around 55% of the looked-after population in England has been male for the past 5 years (Department for Education, 2014). We thought this pre-existing gender disparity in the looked-after population at large might be reflected in a small but significant relationship between service users’ gender and being looked after in our study.

Youth Offending History (Including Involvement in Gun or Knife Crime)

We expected to see significant differences between male and female service users in terms of youth offending rates, with males more likely to have a history of youth offending. Our reasons were twofold. First, among the general U.K. population, the majority of convicted young offenders are male (Ministry of Justice, 2014). Second, in a prior localized study that used a sample of CSE service users in one English city, males were 1.6 times more likely than females (55% vs. 35%) to have a criminal record (Cockbain & Brayley, 2012).

Peer Involvement in Exploitation

The importance of victim peer networks as a vector for CSE has only recently begun to attract concerted research attention (e.g., Cockbain, Brayley, & Laycock, 2011; Firmin, 2013). To date, such research has focused on cases involving exploited girls, and equivalent work on the role of boys’ peer structures in facilitating and/or spreading abuse is lacking. We saw, however, little reason to believe that gender should be associated with significant differences in peer involvement in CSE.

Service Providing Care (and Region/Nation Where It Was Located)

Jago et al. (2011) documented the extent to which areas vary in terms of the nature and quality of CSE service provisions. Detecting CSE is commonly characterized as requiring proper attention, awareness, and active investment of resources (Child Exploitation and Online Protection Centre, 2011; Office of the Children’s Commissioner for England, 2012). It would, therefore, seem reasonable to expect that the less well-understood phenomenon of CSE involving boys might be even more sensitive to local variations.

Source of Referral

Following on from our anticipation that youth offending rates would be higher among boys than girls in our sample, we expected a higher proportion of boys to be referred by criminal-justice agencies. We could find no pre-existing literature or theoretical grounds to inform any further predictions around referral source.

Reason for Referral

McNaughton-Nicholls et al. (2014) documented a strongly held belief among professionals involved in CSE cases that other practitioners reacted differently to the same CSE indicators or risk factors depending on whether the child in question was a boy or girl. Such factors, it was argued, were more likely to be overlooked, dismissed as normal, or denigrated as criminal (rather than signs of potential victimization) when seen in boys. Should, as the authors concluded, gendered norms and stereotypes influence responses to CSE, then we should expect to find differences in referral reason for boys and girls.

Method

Our main method was exploratory data analysis, an approach pioneered by Tukey (1977). In addition, we held a focus group with frontline service managers during the analysis phase of our research. We used information gathered from this focus group both to inform the analytical questions we considered and to help contextualize and interpret our results.

The main challenge we encountered in our analysis was missing data: At least one field was missing from all but 2% of core records and all but 5% of additional records. Out of the 10 core-data fields, six fields were missing for boys in the median case and five for girls, a significant difference, U = 6.4 × 10⁶, z = −23.0, p < .001. For the additional data, five fields (out of a possible nine) were missing for boys in the median case and three for girls, again a significant difference, U = 0.6 × 10⁶, z = −10.2, p < .001. A linear regression, R² = .06, p < .001, F(3) = 190, showed that core records for boys were significantly more likely than those for girls to have more missing fields, β = .23, p < .001. The service responsible for the user was also a significant predictor of the number of missing fields (β = .07, p < .001), as was the date on which the person began to be supported by Barnardo’s (β = −.05, p < .001, that is, later records had fewer missing fields). The small co-efficient of determination (R²) suggests that the apparently higher number of missing fields for boys was not solely an artifact of different recording practices, either over time or between services. Barnardo’s practitioners stated that caseworkers knew less about boys than girls because referring agencies often provide less information because the service users themselves tended to be less forthcoming in their disclosures.

The problem of missing data meant that we compared most variables to gender using bivariate tests only. We used non-parametric tests: the chi-square test for comparisons of categorical variables and the Mann–Whitney U test for continuous variables. To estimate effect size, we calculated the Cramér’s (1946) V statistic for categorical variables and the absolute value of r (Rosenthal, 1991) for continuous variables. To aid interpretation of the results, we report the statistical tests along with the set of records from which each variable came (core or additional) and its completion rate in our original dataset.

Finally, we ran a logistic regression to determine the associations between gender (as the dependent variable) and each of the other variables in conjunction with one another. To maximize the number of cases included in this regression, we limited it to those variables for which there were the fewest missing values²: geographical location, source of referral, reason for referral, ethnicity, and looked-after status.

Results

One third of the overall sample was male (n = 2,986, 33%). The proportion of boys varied by year—although no linear temporal trend existed—from a high of 39% in 2010 to 24% in 2013 (Figure 1). Perhaps more noteworthy was the large change over time in the number of referrals to Barnardo’s: More than 4 times as many cases were dealt with in the first 10 months of 2013 than in the whole of 2008. There are many potential reasons for such variations, including changes in the capacity and funding of projects over time, the increased profile of CSE in the media, and more awareness around the issue, which might have led in turn to more referrals.

Figure 1.

Number of children with whom Barnardo’s began work, by year.

Geographical Variation

The proportion of boys supported by each of the Barnardo’s services in the sample varied substantially. For services that supported at least 100 children in our sample, the proportion of male service users varied between 5% and 57%, with a median of 36%. Geographical variation remained pronounced when service-level data were aggregated to the level of individual regions or nations (due to the small number of Scottish and Northern Irish services, we aggregated their statistics to national level to protect services’ confidentiality). We found the proportion of male service users to range from a low of 6% in Northern Ireland to a high of 47% in the South East of England (with an average of 28% across the individual services in the sample). Such variation should be interpreted cautiously, however, as there are many potential explanations other than underlying differences in the actual proportion of boys among those affected by CSE in a given area. Such possible explanations include differences in staff training and awareness, the availability of male staff, and the services offered in specific locations. The observed variation might also reflect variations between areas in the specific characteristics of the CSE problem, for example, higher proportions of male victims might be seen in an area where the exploitation of boys tended to occur in public places (which might lead to more reporting) or where there was a particularly prolific group of offenders targeting boys.

Source of Referral

Service users were referred to Barnardo’s by a variety of other agencies responsible for the care and supervision of children, as shown in Table 3. Overall, social services were the most frequent source of referrals, followed by criminal-justice agencies such as the police, courts, or probation service. There was a significant difference between boys and girls in the frequency with which they were referred by the different services, χ²(6, N = 8,149: 90% of core records) = 516.8, p < .001, V = 0.25. Boys were 1.7 times more likely than girls to be referred by criminal-justice agencies and only more than half as likely as girls to be referred by social services. These differences may reflect variation in the extent to which boys and girls come into contact with these agencies and/or discrepancies between sectors in the level of training and awareness around CSE and its different forms. It is perhaps noteworthy that very few referrals for either gender came from education or health agencies, even though almost all children would have had contact with professionals in those sectors. This may indicate a lack of awareness of CSE among those working in health or education, or procedural reasons that mean CSE concerns rarely lead to direct referrals to Barnardo’s.

Table 3.

Frequency of Referrals From Different Agencies.

	Girls (%)	Boys (%)	Both (%)
Social services	41	24	35
Criminal justice	32	55	40
Education	6	2	5
Health	2	1	1
Other	19	19	19

To determine whether the gender-based disparity in referral agency was associated with other variables in the data, we ran two additional tests. Contrary to our expectations, children with a criminal record were not referred significantly more often by criminal-justice agencies than those without criminal records, χ²(1, N = 1,482: 50% of additional records) = 1.8, p = .17, V = −0.04. This finding indicated that the more frequent referral of boys by criminal-justice agencies is not solely a result of more boys than girls being known to them due to higher rates of youth offending among boys. In contrast, “looked-after” children (i.e., those under the care of the local authority) were more likely than other children to be referred by social services, χ²(1, N = 8,149: 90% of core records) = 226.2, p < .001, V = 0.17.

Reason for Referral

Children were referred to Barnardo’s CSE services for a variety of reasons, as shown in Table 4. There were significant differences between the genders, χ²(4, N = 2,790: 95% of additional records) = 309.1, p < .001, V = 0.33. Boys were almost twice as likely as girls to be referred because of having gone missing. Going missing accounted for more than half of all referrals overall. It was not possible to explore the relationship between going missing and CSE in any more detail (because data were not available on how many children who go missing are not involved in CSE), but several other studies have highlighted a link between the two issues and further targeted research is needed (Beckett, 2011; Jago et al., 2011; Scott & Skidmore, 2006; Sharp, 2012; Smeaton, 2013).

Table 4.

Frequency of Referrals for Different Reasons.

	Girls (%)	Boys (%)	Both (%)
Going missing	42	80	51
Suspected exploitation	25	7	21
Concerns about relationship with adult	14	4	12
Disclosure of exploitation	5	1	4
Other	15	9	13

Boys were generally referred after going missing. Although going missing was the most common referral reason for girls too, girls were more than 3 times as likely as boys to be referred because of specific concerns about suspected exploitation. This result resonates with previous studies (McNaughton-Nicholls et al., 2014; Smeaton, 2013) in which it has been suggested that practitioners are more likely to identify signs of CSE in girls than in boys. Girls were 4 times more likely than boys to be referred following direct disclosure of CSE, although such direct disclosure was uncommon across the sample as a whole (4%). The rarity with which children in the sample came forward independently to say they had been exploited emphasizes the importance of professionals being able to identify signs of CSE in the absence of a self-disclosure.

Age

The median age of referral for boys (14 years and 5 months) was significantly lower than the median age for girls (14 years and 9 months), U = 6.2 × 10⁶, z = −9.2, p < .001, r = .18. Figure 2 shows that, prior to the age of 11, more boys than girls were referred to Barnardo’s. After the age of 12, however, girls substantially outnumbered boys. Once again, there are various possible explanations for this finding. It may reflect differences in the preferences of those who sexually exploit boys and girls or that boys spend time away from guardians at a younger age than girls do. Alternatively, older boys may be less susceptible than older girls to offenders’ advances, or referring agencies may be less aware of the signs of CSE in teenage boys than in pre-teen boys.

Figure 2.

Age of child at referral to Barnardo’s.

The time from the date of a child’s referral to Barnardo’s to when he or she was first seen by a caseworker was shorter for boys (a median wait of 97 days) than for girls (168 days). This difference was significant, U = 5.8 × 10⁶, z = −13.3, p < .001, r = .17. Barnardo’s practitioners stated that this reflected their experiences that CSE-related concerns for boys tended to be reported at a later stage than for girls, meaning boys’ situations were typically more severe and required more immediate intervention. They said that referring agencies were more aware of vulnerability and early warning signs in girls than in boys, another finding that resonated with work by McNaughton-Nicholls et al. (2014). There are also administrative reasons why there might be a delay in beginning work with a particular child (e.g., resourcing or capacity constraints or difficulty establishing contact), which mean alternative explanations for differences in waiting time cannot be excluded.

Ethnicity

The distribution of service users’ ethnicities (as recorded by their caseworkers) was broadly in line with that of the general youth population, with no obvious under- or over-representation of any particular ethnic group (Table 5). This result runs counter to the commonly heard argument that sexual exploitation involving BME children has been overlooked relative to that involving White children (Home Affairs Select Committee, 2013; Office of the Children’s Commissioner for England, 2012; Smeaton, 2013).

Table 5.

Ethnicity of Service Users.

	Barnardo’s service users			Youth population^a
	Girls (%)	Boys (%)	Both (%)	Youth population^a
White	80	83	81	82
Black	7	6	6	4
Asian	6	5	5	9
Mixed race	5	3	5	4
Other	2	3	5	1

Due to differences in publication of census data, these proportions are based on figures for people aged 8 to 17 years in England and 10 to 17 years in Scotland.

There was an apparent difference between the ethnicities of male and female service users, χ²(4, N = 5,952: 66% of core records) = 12.4, p = .02, V = 0.05. It should be noted that this difference was significant only at the p < .05 level and that the effect size was very small; as shown in Table 5, in no case was the gender disparity greater than 3% of all cases. Given that the present study made multiple comparisons between variables, this result may be spurious, and we have therefore not treated it as a significant finding in our subsequent discussion.

Disability

Male service users were significantly more likely than females to have a disability recorded on their files, χ²(2, N = 3,639: 40% of core records) = 177.1, p < .001, V = 0.27. Overall, 35% of these boys had a recorded disability, compared with 13% of these girls. The gender disparity found in our sample far exceeded our tentative expectation that boys would be over-represented among those with a disability, based on Kvam’s (2000) Norwegian research. Gender considerations aside, the overall over-representation of children with recorded disabilities in our sample corresponds to and builds upon a growing literature documenting the association between disability and increased risk of maltreatment and various forms of abuse (e.g., Sullivan & Knutson, 2000).

Figure 3 shows the disabilities recorded for service users. At least two of the three most common disabilities for both sexes—behavioral disabilities, learning disabilities, and autism spectrum disorder (ASD)—were more common among the study sample than the general youth population. Emerson (2003) found that 1.4% of girls and 3.7% of boys in Great Britain had learning difficulties, notably lower than the figure of 6.7% of girls and 8.7% of boys in the present study. Service users also had a higher prevalence of ASD (1.7% of girls and 8.1% of boys) than that of 1.2% estimated for the general youth population (Baird et al., 2006). The ratio of boys to girls with ASD on record in the current sample (4.8:1) was slightly greater than the mean ratio of 4.3:1 found by Fombonne (2005) in a review of 37 studies of children with ASD. Nonetheless, some caution should be exercized in considering these differences. The previous studies discussed here were typically based on diagnostic tests, whereas our data may include disabilities that had not been clinically diagnosed. In addition, Barnardo’s data entry system only allowed for one disability to be recorded per service user.

Figure 3.

Disabilities recorded for service users.

There are several potential reasons why recorded disabilities may be more common in service users than in the youth population. For example, some offenders may preferentially target disabled children because of perceived weaknesses or reduced credibility as victims. Alternatively, or additionally, certain disabilities (e.g., cognitive or behavioral ones) may impair children’s abilities to evaluate risk, which could in turn lead to greater exposure to would-be offenders and/or susceptibility to their advances. Another possible, but by no means mutually exclusive, explanation for the high rates of disabilities in our data is that disabled children may simply be subject to better safeguarding than non-disabled children. According to this explanation, vulnerability to CSE would be more likely to lead to a referral for support. Further targeted work is required to investigate the relationship between CSE and disability more thoroughly.

“Looked-After” Children

Eighteen percent of the service users were “looked-after” children, a definition that includes those living in children’s homes, secure units, and foster care as well as those living at home with their parents but under the supervision of social services. There was no significant difference between the proportion of boys and girls who were looked after, χ²(1, N = 9,042: 100% of core records) = 0.02, p = .90, V = 0.001.

Although equivalent U.K.-wide baseline figures are not available, around 0.6% of children in England are looked after (Department for Education, 2014). Compared with this figure, our results indicate a substantial over-representation of looked-after children in the service-user sample relative to the youth population at large. In this respect, our findings correspond with previous studies into CSE that have found high rates of looked-after children in their samples (Child Exploitation and Online Protection Centre, 2011; Office of the Children’s Commissioner for England, 2012). Nationally, looked-after children are known to suffer disproportionately to the general youth population from abuse, neglect, and family dysfunction (Department for Education, 2014), which may increase their level of risk in relation to CSE. These findings suggest that those caring for looked-after children have a particularly important role to play in spotting the signs of exploitation. No study has yet looked in detail on a large scale at how being looked after and experiences of CSE interact. Consequently, the reasons behind the over-representation of looked-after children among CSE samples remain unclear. As with the high rate of children with disabilities, the preponderance of looked-after children could be explained via mechanisms such as preferential targeting by offenders, above-average vulnerability to offenders’ approaches, greater exposure to offenders in the first place, and/or closer monitoring by practitioners—whose statutory responsibilities to report concerns could then lead to higher levels of identification.

Youth Offending

Significantly more male (48%) than female service users (28%) were known to have a criminal record, χ²(1, N = 1,567: 53% of additional records) = 45.6, p < .001, V = 0.17. Furthermore, 10% of male and 4% of female service users were suspected of being involved in knife or gun crime (among the 46% of cases of which this information was known). These results corresponded fully with our expectations, having close parallels with findings from Cockbain and Brayley’s (2012) study into links between CSE and youth offending in the English city of Derby. They found that 48% (n = 53) of male and 35% (n = 158) of female CSE service users had criminal records. The proportions of service users with a criminal record in their and our samples exceed that in the general youth population: In 2013, 1.6% of boys and 0.4% of girls aged 10 to 17 years in England and Wales had a criminal record.³

There are several reasons why such a high proportion of CSE service users might have a criminal record: Children may commit offenses as a result of involvement in CSE, children may come into contact with sexual offenders through their own criminal activity, there may be more opportunities to detect CSE among children who are already under the supervision of a youth offending team, and/or CSE and youth offending may be the product of shared environmental risk factors. We had no information about the specific offenses committed by those in our sample but it would be wrong to assume particular links to sexually harmful behavior. To this point, Cockbain and Brayley (2012) found that just 1% of all recorded offenses committed by CSE service users in their sample were sexual in nature, in line with the proportion for recorded youth crime in England and Wales at large.

The gender disparity in offending rates appears to be less for CSE service users than in the general youth population. After controlling for gender imbalances in the samples at large, the ratio of males to females with a criminal record was 1.7:1 in the current sample, broadly analogous to Cockbain and Brayley’s (2012) figure of 1.4:1. Among the general population in 2013, in contrast, for every girl with a criminal record, there were 3.7 boys. Youth offending therefore appears to be more closely associated with CSE for girls than for boys. One possible explanation for this might lie in the fact that boys and girls vary in terms of what offenses they tend to commit (Youth Justice Board, 2009).

Peer Involvement in CSE

Peer involvement in CSE was known or suspected in the cases of 31% of male and 56% of female service users, χ²(1, N = 1,323: 45% of additional records) = 47.3, p < .001, V = 0.19. This disparity may be because certain types of CSE (namely, those involving solo victimization) may be more common for boys than girls (Brayley, Cockbain, & Gibson, 2014) or that boys discuss their involvement in CSE less with their peers due to the additional stigma associated with male-on-male sex (meaning that male service users might simply be less aware of their male friends’ involvement in CSE). Barnardo’s practitioners suggested the difference might arise due to boys’ comparative unwillingness to discuss either their own exploitation or that of their peers. Previous studies into CSE involving groups of victims have focused on female victims (Brayley, Cockbain, & Laycock, 2011; Cockbain et al., 2011; Firmin, 2013) and little is currently known about equivalent CSE affecting boys.

Multivariate Analysis

The problem of missing data meant that we carried out our most detailed analyses using bivariate tests. We also ran a binary logistic regression to determine whether the results of those tests changed when controlling for the other variables and potential interactions between them. To maximize the sample size available for this analysis, we only included those variables present in more than 60% of cases. Of the original sample, 23% of cases (n = 2,080) had complete data for all five variables in question and could therefore be included in the multivariate analysis.

Overall, the model containing the independent variables shown in Table 6 was a significant predictor of whether a service user would be male or female, χ²(21, N = 2,080) = 278.2, p < .001. The pseudo-R² value of .20, calculated as described by Cragg and Uhler (1970), indicated that there was unexplained variation in the model. This was to be expected, because there are likely to be differences between male and female service users that could not be studied using the present data. The results of Wald tests for individual predictors generally accorded with those found in bivariate tests. The region of the United Kingdom in which a service user lived, the agency referring them, and the reason for the referral were all significant predictors of gender; bivariate tests had already shown significant differences by gender for each of these variables individually. Ethnicity and being a looked-after child were not significant predictors of gender: neither of these were factors for which a significant gender disparity (at p < .01 level) had been found in the bivariate tests. For those predictors that were significant, the direction of the odds of a service user being male (e^β, Table 6) was the same as would be predicted by the bivariate tests (e^β is not shown for the region variable to protect confidentiality). The results of the regression model give increased confidence in the findings of the bivariate tests, speaking against the results being artifacts caused by differences in other variables.

Table 6.

Logistic Regression Results.

	Wald χ²	df	p	e ^β
Region of the United Kingdom	16.5	8	.036
Source of referral	19.5	4	.001
Social services				0.69
Criminal justice				1.20
Education				0.46
Health				0.74
Referral reason	71.8	4	<.001
Going missing				1.70
Suspected exploitation				0.47
Concerns about relationship with older person				0.34
Disclosure of exploitation				0.36
Ethnicity	7.32	4	.120
Looked-after child	3.5	1	.060

Discussion

This study was designed to explore empirically and systematically the relationship between CSE and the gender of the children it affects. Having already considered the interpretation and implications of individual results in the preceding session, here we focus on the study and its findings in their entirety. One of our most striking findings is also one of the simplest: One third of our sample was boys. The proportion of boys was markedly higher than in prior national studies into various forms of CSE in the United Kingdom or its constituent nations (e.g., Child Exploitation and Online Protection Centre, 2011; Office of the Children’s Commissioner for England, 2012). This discrepancy may be due in part to our broad definition of CSE, whereas these prior studies focused on sub-types such as “group and gang associated CSE.” It might seem a valid challenge to suggest that the high proportion of boys might stem from our inclusion of children suspected or at risk of perpetrating CSE. When we raised this in the focus group with frontline Barnardo’s service providers, they felt this to be inaccurate. They expressed the strong conviction that the majority of boys (even more so than girls) they support are CSE victims. Whatever the reason for it, the very fact that nearly 3,000 boys were supported by Barnardo’s for CSE emphasizes the importance of better incorporating their needs into research, policy, and practice.

Taken as a whole, our other findings indicate that the relationship between CSE and service users’ gender is both complex and nuanced. No significant differences (at p < .01 level) were observed between the boys and girls in terms of ethnicity and being looked after. In addition, we found no clear trend in the proportion of male service users per year over the core study period from 2008 to 2013 inclusive, the years from which >99% (n = 9,020) of our sample derived. The substantial (and significant) variations between individual projects and regions or nations in terms of the proportion of boys, although interesting, should be treated cautiously for reasons previously explained (e.g., differences in specialist service provision and other agencies’ awareness that CSE affects boys). In contrast, such external factors alone would inadequately explain the observed differences for the other statistically significant variations between the genders. To recap, the boys in the sample were, in comparison with the girls, more likely to be referred by certain agencies (in particular criminal-justice agencies) and less likely to be referred by others (in particular social services); more likely to be referred because of concerns related to going missing, several months younger on average; more likely to have recorded disabilities; more likely to have criminal records; and less likely to have peers also thought to be affected by CSE.

Many of our findings resonate with previous research in which associations have been documented qualitatively—and to a lesser extent quantitatively—between CSE and youth offending, going missing, and being in care (e.g., Beckett, 2011; Cockbain & Brayley, 2012; Child Exploitation and Online Protection Centre, 2011; Clutton & Coles, 2007; Jago et al., 2011; Office of the Children’s Commissioner for England, 2012; Pearce, Williams, & Galvin, 2003; Scott & Skidmore, 2006). In what appears to be an isolated example of prior research covering CSE referral pathways in the United Kingdom, McNaughton-Nicholls et al. (2014) reported that practitioners dealing with CSE perceived boys to be seen as less vulnerable than girls. Such differences in perception of gender and risk might help account for the differences in referral agencies and referral reasons documented in the current study, although this is clearly an area in need of greater attention. Even more poorly documented in the existing research literature is the relationship between disability, gender, and CSE. Future investigation into CSE and disabilities could usefully incorporate gender into the analysis, given the substantial and significant differences in disability rates for boys and girls in our sample.

In considering the strengths and weaknesses of our study, several factors distinguish it from most prior enquiries into CSE in the United Kingdom or elsewhere. First is our use of the quantitative approach. Although fairly common in the better developed literature on child sexual abuse, quantitative methods have rarely been applied to the study of CSE. Instead the literature has been dominated by qualitative studies, with notable exceptions including evaluations of policy and its implementation (e.g., Jago et al., 2011) and small-scale studies into certain types of CSE (e.g., Cockbain et al., 2011). Quantitative studies such as ours can develop the evidence base by helping identify more general—and possibly generalizable—patterns. The second distinguishing factor is the large size of the sample: 9,042 unique cases. The prevailing tendency toward small samples even in quantitative studies of CSE may be linked to difficulties in accessing data on this sensitive and hidden (or simply overlooked) issue. In our study, the large sample size reduces the risk of sampling error and promotes confidence in the results. The third key strength is that the data were individual level, which permitted types of analysis not used even in previous large-scale national scoping studies into CSE (e.g., Child Exploitation and Online Protection Centre, 2011; Office of the Children’s Commissioner for England, 2012). Individual-level data were a critical prerequisite for us to be able to compare boys and girls systematically across a wide range of independent variables and to examine possible interactions between the independent variables. Using inferential statistics to test for the significance of the observed relationships and for possible confounding variables promotes greater confidence in the validity of our results. Other strengths include the current nature of the data (the vast majority of cases came from 2008-2013 inclusive), the coverage of three countries (England, Northern Ireland, and Scotland), the broad range of study variables, and the unobtrusive approach to data collection.

Of course, the study also has limitations that affect the interpretation and application of our findings. The ethical benefits of relying on a pre-existing sample to research this sensitive issue are counterbalanced by the downsides to using data collected primarily for administrative rather than research purposes. The way variables had been categorized meant it was not always easy to find appropriate and directly comparable baseline data against which to assess the findings in a meaningful fashion. We could not address certain issues of clear theoretical and practical interest (such as exploiters’ age and gender, the nature and context of any abuse suffered, and factors promoting resilience) simply because the necessary data had not been collected. As previously stated, there was no way of distinguishing whether a given child had been exploited, exploited others, and/or was at risk of experiencing or perpetrating CSE. Instead, all these permutations were captured under the broad tag “affected by CSE.” Missing data was another problem, limiting the number of variables and proportion of cases that we could include in the multivariate analysis. More sophisticated analyses were not possible, and the denominator for the bivariate analyses varied depending on completion rate. Although it is understandable that frontline service providers might not see data entry as a key concern, studies such as this can help demonstrate the value of detailed and comprehensive record keeping. Perhaps the most obvious limitation of the study is the issue of external validity. Our data derived from one organization only, albeit one that is the United Kingdom’s biggest provider of CSE services. Barnardo’s has a broad geographical reach in its services but the cases covered here are clustered around service locations rather than being evenly distributed across the United Kingdom’s nations. Linked to the issue of external validity is the perennial question of much social research of whether—and to what extent—identified cases (the study sample) differ systematically from all those unidentified cases that make up the “dark figure” of crime.

Given the complex set of processes likely to be at play in mediating children’s exposure, vulnerability, and resilience to sexual exploitation, it would be useful in future to consider population-based studies into CSE. The inclusion of both victims and non-victims in such studies might help disentangle potentially interrelated variables (e.g., youth offending and going missing) and enable the identification of those variables with the maximum predictive utility. The identification of empirically substantiated risk factors (and resilience factors) could support more effective and targeted deployment of ever dwindling resources (e.g., via the creation of predictive risk maps of the type used by epidemiologists). The careful and deliberate collection of temporally ordered data (including but not limited to the type used in longitudinal studies) could be beneficial in moving beyond identifying significant correlations toward investigating processes of cause and effect. Many of the factors routinely referred to as “risk factors” for or “indicators” of CSE are currently only actually known to be correlates. Although optimistic about the benefits of further empirical investigations of this nature, we would caution that CSE’s complexity may mean even well-designed and well-executed research may not uncover clear, linear, and predictable causal relationships. To illustrate this point, going missing may be a CSE risk factor for some children, a consequence for others, and a simple correlate for others still.

Conclusion

Although CSE is increasingly recognized as an important child protection and crime prevention priority, the demand for robust research evidence to inform responses outpaces its supply. Among the most pronounced and frequently cited gaps in the literature is the sexual exploitation of boys. Our study’s systematic and large-scale exploration of the relationship between gender and CSE represents a novel and timely addition to the limited research base. We drew upon a rich and previously untapped resource, and our study sample (9,042 cases) was unusually large for the field. Following the approach of exploratory data analysis, we subjected individual-level data for diverse variables to descriptive and inferential statistical analyses. Our results demonstrate statistically significant differences by gender across many variables and crosscutting commonalities across others. Clear overlaps between CSE and other serious social concerns such as youth offending and missing children were also noteworthy. The results of this exploratory study should not be overstated, and more work is clearly needed to disentangle the complex relationship between CSE and gender. Gender, we would argue, is an important factor to consider but still just one piece of the proverbial puzzle. Nonetheless, our findings highlight the danger of continuing to ignore or overlook boys when it comes to tackling CSE. The observed differences between the male and female service users in our sample suggest that current female-centric approaches to policy making, victim identification, and service provision may not be serving boys adequately. In future, gender might usefully be factored more effectively into the design and delivery of research, policy, and practice.

Footnotes

Acknowledgements

In addition to the funding agency, the authors thank Barnardo’s (in particular, Carron Fox and Cassandra Harrison) for the invaluable input, support, and data provision. We are also grateful to Professor Richard Wortley, the journal editors, and anonymous peer reviewers for their insightful comments on earlier drafts of this article.

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 by the Nuffield Foundation (grant reference CPF/41512).

Notes

References

Alaggia

(2005). Disclosing the trauma of child sexual abuse: A gender analysis. Journal of Loss & Trauma, 10, 453-470. doi:10.1080/15325020500193895

Appleton

(2014). Child sexual exploitation, victimisation and vulnerability. Child Abuse Review, 23, 155-158. doi:10.1002/car.2336

Association of Chief Police Officers. (2012). ACPO child sexual exploitation action plan. London, England: ACPO Child Protection and Abuse Investigation Working Group. Retrieved from http://www.ceop.police.uk/Documents/ACPOCSEPlanNPM.pdf

Baird

Simonoff

Pickles

Chandler

Loucas

Medlrum

Charman

(2006). Prevalence of disorders of the autism spectrum in a population cohort of children in South Thames: The Special Needs Autism Project (SNAP). The Lancet, 378, 210-215. doi:10.1016/S0140-6736(06)69041-7

Barnardo’s. (2012). Cutting them free: How is the UK progressing in protecting its children from sexual exploitation? London, England: Author.

Barrett

Melrose

(2003). Courting controversy—Children sexually abused through prostitution—Are they everybody’s distant relative but nobody’s children? Child and Family Law Quarterly, 15, 371-382.

Beckett

(2011). Not a world away: The sexual exploitation of children and young people in Northern Ireland. Belfast, Ireland: Barnardo’s.

Brayley

Cockbain

(2014). British children can be trafficked too: Towards an inclusive definition of internal child sex trafficking. Child Abuse Review, 23, 171-184. doi:10.1002/car.2307

Brayley

Cockbain

Gibson

(2014). Rapid evidence assessment: The sexual exploitation of boys and young men. London, England: Barnardo’s.

10.

Brayley

Cockbain

Laycock

(2011). The value of crime scripting: Deconstructing internal child sex trafficking. Policing: A Journal of Policy and Practice, 5, 132-143. doi:10.1093/police/par024

11.

Cashmore

Shackel

(2014). Gender differences in the context and consequences of child sexual abuse. Current Issues in Criminal Justice, 24, 75-104.

12.

Chase

Statham

(2005). Commercial and sexual exploitation of children and young people in the UK—A review. Child Abuse Review, 14, 4-25. doi:10.1002/car.881

13.

Child Exploitation and Online Protection Centre. (2011). Out of mind, out of sight: Breaking down the barriers to understanding child sexual exploitation. London, England: Author.

14.

Clutton

Coles

(2007). Sexual exploitation risk assessment framework: A pilot study. Cardiff, UK: Barnardo’s Cymru.

15.

Cockbain

(2013). Grooming and the “Asian sex gang predator”: The construction of a racial crime threat. Race & Class, 54, 22-32. doi:10.1177/0306396813475983

16.

Cockbain

Brayley

(2012). Child sexual exploitation and youth offending: A research note. European Journal of Criminology, 9, 689-700. doi:10.1177/1477370812453401

17.

Cockbain

Brayley

Laycock

(2011). Exploring internal child sex trafficking using social network analysis. Policing: A Journal of Policy and Practice, 5, 144-157. doi:10.1093/police/par025

18.

Cragg

J. G. G.

Uhler

R. S.

(1970). The demand for automobiles. Canadian Journal of Economics, 3, 386-406. doi:10.2307/133656

19.

Cramér

(1946). Mathematical methods of statistics. Princeton, NJ: Princeton University Press.

20.

Crown Prosecution Service. (2013). Interim guidance on prosecuting cases of child sexual abuse. London, England: Author.

21.

De Jong

A. R.

Hervada

A. R.

Emmett

G. A

. (1983). Epidemiological variations in childhood sexual abuse. Child Abuse & Neglect, 7, 155-162. doi:10.1016/0145-2134(83)90067-4

22.

Department for Children, Schools and Families. (2009). Safeguarding children and young people from sexual exploitation (Supplementary Guidance to Working Together to Safeguard Children). London, England: Home Office.

23.

Department for Education. (2011). Tackling child sexual exploitation: Action plan. London, England: Author.

24.

Department for Education. (2014). Children looked after in England (including adoption and care leavers) year ending 31 March 2014. London, England: Office for National Statistics.

25.

Disch

Avery

(2001). Sex in the consulting room, the examining room, and the sacristy: Survivors of sexual abuse by professionals. American Journal of Orthopsychiatry, 71, 214-217. doi:10.1037/0002-9432.71.2.204

26.

Edelson

(2012). Why have all the boys gone? Gender differences in prosecution acceptance of child sexual abuse cases. Sexual Abuse: A Journal of Research and Treatment, 25, 461-481. doi:10.1177/1079063212465900

27.

Emerson

(2003). Prevalence of psychiatric disorders in children and adolescents with and without intellectual disability. Journal of Intellectual Disability Research, 47, 51-58. doi:10.1046/j.1365-2788.2003.00464.x

28.

Feiring

Taska

Lewis

(1999). Age and gender differences in children’s and adolescents’ adaptation to sexual abuse. Child Abuse & Neglect, 23, 115-128. doi:10.1016/S0145-2134(98)00116-1

29.

Finkelhor

(1986). A source book on child sexual abuse. Newbury Park, CA: Sage.

30.

Finkelhor

(1994). The international epidemiology of child sexual abuse. Child Abuse & Neglect, 18, 409-417. doi:10.1016/0145-2134(94)90026-4

31.

Firmin

(2013). Busting the “gang-rape” myth: Girls’ victimisation and agency in gang-associated sexual violence. In Horvath

Woodhams

(Eds.), Handbook on the study of multiple perpetrator rape (pp. 97-115). Abingdon, UK: Routledge.

32.

Fombonne

(2005). The changing epidemiology of autism. Journal of Applied Research in Intellectual Disabilities, 18, 281-294. doi:10.1111/j.1468-3148.2005.00266.x

33.

Forrest

(2007). Boys, young men and sexual health services: A summary of a review of the academic literature. London, England: Brook.

34.

Gohir

(2013). Unheard voices: The sexual exploitation of Asian girls and young women. Birmingham: Muslim Women’s Network UK.

35.

Harvey

Hornsby

Sattar

(2015). Disjointed service: An English case study of multi-agency provision in tackling child trafficking. British Journal of Criminology, 55, 494-513. doi:10.1093/bjc/azu115

36.

Home Affairs Select Committee. (2013). Child sexual exploitation and the response to localised grooming: Second report of session 2013-14. London, England: The Stationery Office.

37.

Home Office. (2011). UK action plan on tackling child sexual exploitation. London, England: The Stationery Office.

38.

Home Office. (2015). Tackling child sexual exploitation. London, England: The Stationery Office.

39.

Homma

Wang

Saewyc

Kishor

(2012). The relationship between sexual abuse and risky sexual behavior among adolescent boys: A meta-analysis. Journal of Adolescent Health, 51, 18-24. doi:10.1016/j.jadohealth.2011.12.032

40.

Jago

Arocha

Brodie

Melrose

Pearce

Warrington

(2011). What’s going on to safeguard children and young people from sexual exploitation? How local partnerships respond to child sexual exploitation. Luton, UK: University of Bedfordshire.

41.

Jago

Pearce

(2008). Gathering evidence of sexual exploitation of children and young people: A scoping exercise. Luton, UK: University of Bedfordshire.

42.

Jay

(2014). Independent inquiry into child sexual exploitation in Rotherham 1997-2013. Retrieved from www.rotherham.gov.uk/downloads/file/1407/independent_inquiry_cse_in_rotherham

43.

John Jay College of Criminal Justice. (2006). The nature and scope of sexual abuse of minors by Catholic Priests and Deacons in the United States, 1950-2002. Washington, DC: United States Conference of Catholic Bishops.

44.

Kvam

(2000). Is sexual abuse of children with disabilities disclosed? A retrospective analysis of child disability and the likelihood of sexual abuse among those attending Norwegian hospitals. Child Abuse & Neglect, 24, 1073-1084. doi:10.1016/S0145-2134(00)00159-9

45.

Lillywhite

Skidmore

(2006). Boys are not sexually exploited? A challenge to practitioners. Child Abuse Review, 15, 351-361. doi:10.1002/car.952

46.

Lowe

Pearce

(2006). Young people and sexual exploitation. Child Abuse Review, 15, 289-293. doi:10.1002/car.958

47.

McNaughton-Nicholls

Harvey

Paskell

(2014). Sexual exploitation of boys and young men: Interviews with professionals. London, England: Barnardo’s.

48.

Ministry of Justice. (2014). Youth justice annual statistics: 2012 to 2013. London, England: Author.

49.

Moody

(1999). Male child sexual abuse. Journal of Pediatric Health Care, 13, 112-119. doi:10.1016/S0891-5245(99)90072-X

50.

National Statistics. (2014). Population estimates for UK, England and Wales, Scotland and Northern Ireland, Mid-2013. London, England: Office for National Statistics.

51.

Nelson

Oliver

(1998). Gender and the construction of consent in child-adult sexual contact beyond gender neutrality and male monopoly. Gender & Society, 12, 554-577. doi:10.1177/089124398012005004

52.

Office of the Children’s Commissioner for England. (2012). “I thought I was the only one. The only one in the world.” The Office of the Children’s Commissioner’s inquiry into child sexual exploitation in gangs and groups, interim report. London, England: Author.

53.

Palmer

(2001). No son of mine! Children abused through prostitution. London, England: Barnardo’s.

54.

Paolucci

Genuis

Violato

(2001). A meta-analysis of the published research on the effects of child sexual abuse. The Journal of Psychology: Interdisciplinary and Applied, 135, 17-36. doi:10.1080/00223980109603677

55.

Parkinson

Oates

Jayakody

(2010). Breaking the long silence: Reports of child sexual abuse in the Anglican Church of Australia. Ecclesiology, 6, 183-200. doi:10.1163/174413610X493791

56.

Pearce

Williams

Galvin

(2003). It’s someone taking a part of you: A study of young women and sexual exploitation. London, England: National Children’s Bureau.

57.

Priebe

Svedin

(2009). Prevalence, characteristics and associations of sexual abuse with sociodemographics and consensual sex in a population-based sample of Swedish adolescents. Journal of Child Sexual Abuse, 18, 19-39. doi:10.1080/10538710802584635

58.

Public Petitions Committee. (2014). Inquiry into tackling child sexual exploitation in Scotland. Edinburgh, Scotland: Author.

59.

Radford

Corral

Bradley

Fisher

Bassett

Howat

Collishaw

(2011). Child abuse and neglect in the UK today. London, England: NSPCC.

60.

Rosenthal

(1991). Meta-analytic procedures for social research. Newbury Park, CA: Sage.

61.

Scott

Skidmore

(2006). Reducing the risk: Barnardo’s support for sexually exploited young people: A two-year evaluation. London, England: Barnardo’s.

62.

Scrandis

Watt

(2014). Child sexual abuse in boys: Implications for primary care. The Journal for Nurse Practitioners, 10, 706-713. doi:10.1016/j.nurpra.2014.07.021

63.

Sharp

(2012). Still hidden? Going missing as an indicator of child sexual exploitation. London, England: Missing People.

64.

Smallbone

Marshall

Wortley

(2008). Preventing child sexual abuse: Evidence, policy and practice. Cullompton, UK: Willan.

65.

Smallbone

Wortley

(2000). Child sexual abuse in Queensland: Offender characteristics and modus operandi. Brisbane, Australia: Queensland Crime Commission.

66.

Smeaton

(2013). Running from hate to what you think is love: The relationship between running away and child sexual exploitation. London, England: Barnardo’s.

67.

Stoltenborgh

van Ijzendoorn

Euser

Bakermans-Kranenburg

(2011). A global perspective on child sexual abuse: Meta-analysis of prevalence around the world. Child Maltreatment, 16, 79-101. doi:10.1177/1077559511403920

68.

Sullivan

Knutson

(2000). Maltreatment and disabilities: A population-based epidemiological study. Child Abuse & Neglect, 24, 1257-1273. doi:10.1016/S0145-2134(00)00190-3

69.

Tukey

(1977). Exploratory data analysis. New York, NY: Pearson.

70.

United Nations. (1989). United Nations convention on the rights of the child (UNCRC). Geneva, Switzerland: United Nations.

71.

Ward

Patel

(2006). Broadening the discussion on “sexual exploitation”: Ethnicity, sexual exploitation and young people. Child Abuse Review, 15, 341-350. doi:10.1002/car.953

72.

Welsh Assembly Government. (2010). Safeguarding children and young people from sexual exploitation: Supplementary guidance to safeguarding children: Working Together Under the Children Act 2004. Cardiff, UK: Author.

73.

Youth Justice Board. (2009). Girls and offending: Patterns, perceptions and interventions. London, England: Author.