Everyday atrocities: does internal (domestic) sex trafficking of British children satisfy the expectations of opportunity theories of crime?

Abstract

In this study we examine the internal (domestic) sex trafficking of British children using unique data from six major police investigations. This particular type of internal sex trafficking (sometimes known as ‘street grooming’) has been popularly conceptualised as a highly sophisticated, skilled and well-organised phenomenon. This study shows that this characterisation does not withstand empirical scrutiny. Instead, the routine activities and everyday associations of both offenders and victims are shown to play key roles in facilitating, sustaining and spreading the abuse. While the criminal acts associated with internal child sex trafficking can be atrocious, the people, places and processes involved are shown to be far from exceptional. In this respect, the results may be unsettling: they undermine explanations of an emotive crime that rest on reassuring but ultimately naïve errors of attribution. We argue that it is important, however, that preventative strategies are underpinned not by sensationalised narrative and untested assumptions but by sober and robust assessments of appropriate empirical data. The paper contributes to the theoretical and empirical literature on opportunity theories of crime, on human trafficking and on child sexual abuse/exploitation. While the sample size is not especially large (55 offenders and 43 victims), this study helps to expand a sorely limited knowledge base on a topical threat. It is also distinguished by its hard-to-access data and novel analytical approach. The work is likely to interest a broad and international audience of academics, practitioners and policy makers concerned with crime prevention and child protection.

Background

In 2004, the distinct criminal offence of ‘trafficking within the UK for the purposes of sexual exploitation’ came into force in England and Wales (Section 58 of the Sexual Offences Act 2003, amended in 2013 to Section 59a). One key function of the new legislation was to criminalise sex trafficking within the boundaries of a single country, otherwise known as internal (or domestic) sex trafficking. The new law was designed to cover both adult and child victims. In recent years, however, it is the internal sex trafficking of British children that has garnered particular attention and concern. For brevity’s sake we will refer to this issue hereafter simply as ‘internal child sex trafficking’ (ICST), although we recognise that children of other nationalities and/or in other countries can be internally sex trafficked too.

From 2008, a series of high-profile ICST cases such as those in Derby, Oxford and Rochdale have attracted concerted media, public and political attention (Brayley and Cockbain 2014; Cockbain 2013a). Growing concern around a possible systemic failure of prevention prompted various official enquires and reports (e.g., Association of Chief Police Officers 2012; Child Exploitation and Online Protection Centre 2011; Home Office 2011, 2012; House of Commons 2013). While the specifics of the recent high-profile cases varied, many shared certain common features. A ‘typical’ case that attracted extensive media attention: occurred in a town or city in the North and Midlands areas of England; involved groups of male perpetrators and female victims; and, particularly controversially, mostly involved South Asian (primarily Pakistani) offenders and white victims. Whether and if so to what extent such characteristics are representative of all ICST remains unclear, for reasons including major data gaps, confusion around definitions and potential biases in the news production process (see also Cockbain 2013a).

Like internal trafficking in general (Laczko and Gozdziak 2005; Winterdyk and Reichel 2010), ICST has been subject to little empirical research and its dynamics remain poorly understood. While ICST is clearly a form of child sexual abuse, it has some distinctive features. According to Brayley and Cockbain’s (2014) definition, which we use here, ICST involves multiple perpetrators and involves movement integral to the abuse (the ‘trafficking’ element).

In contrast, the involvement of multiple perpetrators is not characteristic of sexual offending against either children or adults (see, e.g., Horvath and Woodhams 2013; Smallbone and Wortley 2000). Compared to other offence types, sexual offences have repeatedly and consistently been found to have one of the lowest co-offending rates of any crime studied (Andreson and Felson 2012; Carrington 2009; Smallbone and Wortley 2000; van Mastrigt and Farrington 2009).

With the notable exception of a small-scale exploratory study on ICST victim and offender networking (Cockbain et al. 2011), there has been little research into the interpersonal associations and activities underpinning ICST. Nonetheless, the group-based nature of ICST has been characterised in the media and elsewhere as indicative of high levels of sophistication and organisation (e.g. Barnardo’s 2012).

In the absence of evidence to the contrary, we begin with the assumption that ICST can be explained using models applied to child sexual abuse generally. Our study was informed by prior research that has examined the situational dynamics of child sexual abuse. Drawing on opportunity theories such as rational choice perspective (Clarke and Cornish 1985) and routine activity theory (Cohen and Felson 1979), situational analyses involve a shift in analytic focus. Like all human behaviour (Mischel 1968), crime can be understood as the product of an interaction between disposition and situation (Wortley 2012). Rather than focus on the distant factors leading to supposed ‘criminality’, situational researchers focus on the immediate context of ‘criminal events’.

Applied to the problem of child sex offending, the situational perspective would suggest that many offenders are flexible in their sexual preferences and take opportunities to sexually abuse children as and when the situational conditions are favourable (Wortley and Smallbone 2006a). Taking into account individual differences in criminal propensity, child sex offenders can be arguably be categorised as opportunity takers, seekers and creators (Wortley and Smallbone 2006b).

Research into other forms of child sexual abuse has demonstrated the importance of opportunity and offenders’ and victims’ routine activities. Offenders tend, for example, to abuse children to whom they have ready access and the offence itself often occurs at convenient locations where guardianship is absent and in the context of everyday care-giving or recreational activities (Deslauriers-Varin and Beauregard 2010; Leclerc and Felson 2014; Wortley and Smallbone 2006a). Many offenders do not display the specialised interest in sexually abusing children that might be expected of those psychologically driven to offend (Simon 1997, 2000; Smallbone and Wortley 2000, 2004). Smallbone and Wortley (2000), for example, examined the criminal records of 323 adult males in Australia who had been convicted of sexual offences against children. Although 62 % had previous conviction histories, only 21 % had a prior sexual offence conviction. Their findings indicate that the average child sex offender is more likely to be an ‘opportunity taker’ than a committed and specialised predator (see also Smallbone and Wortley 2004).

Marcus Felson and colleagues (Andreson and Felson 2010; 2012; Felson 2003; Felson and Boba 2010) have written from a routine activities perspective about the role of co-offending in crime generally. Felson’s work provides a basis for us to speculate about how the involvement of multiple offenders intrinsic to ICST might impact on the individual offending behaviour of those involved. Felson challenges the popular view of group-based crime as highly organised. Instead he characterises offender groups as loose structures with little cohesion or formal hierarchy, a view shared with other researchers whose work addresses co-offending and/or criminal networks (e.g. McGuire 2012; McGloin and Nguyen 2011; McGloin et al. 2008). From Felson’s perspective, the group may have a relatively stable core of members who account for most of the criminal activity, but most members are peripheral and drift in and out of the group. Members may co-offend with one or more other members and/or with others outside of the group and/or alone; the group rarely offends as a co-ordinated unit. Nevertheless co-offending can and does change the offending dynamic. Compared to solo-offenders, those involved in co-offending tend to have higher rates of offending and to commit more serious offences. If the patterns described above apply to ICST then, portrayals of offenders as highly sophisticated and organised are likely to be inaccurate. We might expect, however, to observe persistent and serious levels of offending among core group members in particular.

Our study examines the extent to which ICST conforms to the expectations of opportunity theories of crime, with a particular focus on the routine activity approach. We examine five propositions:

• 1. Many offenders are criminally versatile and have convictions for non-sexual offences;

• 2. Offenders’ routine activities facilitate access to suitable co-offenders, potential victims and appropriate crime facilitators;

• 3. Victims’ routine activities help explain their availability, attractiveness and vulnerability to offenders;

• 4. Better-connected offenders within the group typically offend at a higher rate; and

• 5. Abuse occurs at locations lacking supervision and familiar to offenders from their everyday lives.

Method

Data

This study took place within a broader programme of research into ICST, conducted by the first author for her doctoral studies (Cockbain 2013b). Gaining access to restricted police data was a lengthy process, involving negotiations with each investigative team to build trust and convey the value of the proposed research as well as the formulation of ethical approvals, security vetting procedures and data sharing contracts (see Cockbain 2015).

The data used here derived from six of the earliest and largest ICST police investigations in the UK, all of which were concentrated in the North West and Midlands areas of England and the period 2008–2012. Table 1 provides an overview of the core characteristics of the six cases. Our data included thousands of pages of detailed police operational files and court records, offering a valuable opportunity for an in-depth exploration of the characteristics of the crime commission process and those involved. Our sample contained 55 offenders (54 of whom were male) and 43 victims (all female)¹.

Table 1 Key information on cases in the sample

While our dataset was both novel and extensive, we acknowledge that the sample size is not large in a statistical sense. Additionally, readers should be mindful that cases that are identified and investigated may differ from those that are not. This issue is common to much criminological research, however, and should not detract from the value of this novel exploration of a little-understood crime.

Procedure

Data were interrogated using content analysis (Berelson and Lazarsfeld 1948), selected due to its broad applicability, well-established nature and particular utility in structuring and systematising documentary analysis (Bryman 2001; Reason and Garcia 2007). In incorporating both qualitative and quantitative aspects to the analysis, we ensured a systematic and well-rounded exploration of ICST that did not reduce this complex social issue to numerical assessment alone.

There is no standardised procedural logic for content analysis and there is invariably some subjectivity in its design and conduct (Krippendorff 2004). To maximise reliability and transparency, we designed a clear and explicit coding framework (shown in Table 2). Due to the sheer volume of data to be reviewed, coding was a slow and iterative process. The first author sifted carefully through the available documentation, identifying and extracting any relevant information and entering it into the codesheet as she went (Excel spreadsheets). During the data extraction process, she continuously reviewed the information in the codesheet, updating it as necessary whenever new information emerged (e.g. information on an earlier offence that affected the code ‘age at first offence’). Unfortunately, we were unable to double-code the data due to their sensitive nature and the consequent constraints of our data-sharing agreements.

Table 2 Coding framework

For concision’s sake we present statistical results for the combined sample for the six investigations. Due to possible bias introduced by the clustering of individuals by case, we also ran case-by-case analysis and relevant discrepancies are highlighted in the results.

Results and discussion

We now examine each of the five propositions in turn, reviewing whether and to what extent they were supported. We present percentages to aid interpretation and facilitate comparison with other work, but readers should be mindful of our sample size when considering whether observed differences are likely to be meaningful.

Offender criminal versatility

Many offenders are criminally versatile and have convictions for non-sexual offences

With the notable exception of Operation Span, the majority of offenders on every case had previous convictions. The prior conviction rate (51 %, n = 22) was markedly higher than the comparable national baseline of 28 % of 18–52 year old males (Ministry of Justice 2010)². Additionally, prolific offenders (defined as having three or more convictions) comprised 29 % (n = 16) of our sample, compared with 11 % of males nationally (Ministry of Justice 2010). The modal number of convictions among those with records in our sample was 3.5 and the mean was 6.4 (SD = 7.3, range 1–28).

The results indicate a generalist pattern of offending among the ICST offenders with previous offences. Only two had prior sexual offence convictions; for the other 20 offenders, convictions came from a total of ten other offence types, as categorised by Blackstones (2013). Fifteen of the 22 offenders with convictions (including the two with sexual offending records) had records spanning multiple offence categories. Both in terms of prevalence and incidence, prior convictions were most commonly offences against the administration of justice, road traffic offences and theft offences.

The findings support the proposition that many ICST offenders are involved in other criminal activity beyond ICST, which in turn suggests a generalised lack of self-control and a readiness to seize criminal opportunities. Our findings are broadly comparable with previous research on sex offender versatility (Simon 1997, 2000; Smallbone and Wortley, 2000, 2004), although the proportion with prior sexual convictions (4 %, n = 2) in our sample was markedly lower than the 21 % documented by Smallbone and Wortley (2000). This discrepancy might suggest ICST could involve even more opportunism and less specialism than other forms of child sexual abuse.

Offenders’ routine activities

Offenders’ routine activities facilitate access to suitable co-offenders, potential victims and appropriate crime facilitators

Offenders were aged 17–56 years at the time of the first ICST offence recorded in our data: a proxy for age of ICST onset. The only minor was also the only female offender. The mean age was 29 years (SD = 9 years) but there was actually considerable variation in the age distribution by case (see Fig. 1). Overall, there was a general skew towards the younger age groups: 73 % (n = 40) of offenders were 31 years or younger at their first ICST offence (see Fig. 2).

Fig. 1

Age distribution of offenders by case

Fig. 2

Age distribution of the full offender sample

All or most offenders on every case and 96 % (n = 53) overall were of Asian heritage—defined according to the UK census category as being from South Asia rather than the Far East. The remaining two offenders, including the female offender, were white. At 80 % (n = 44), Pakistani heritage offenders were clearly overrepresented relative to the demographics of the general English population (2 % Asian Pakistani) and of the relevant local authority areas (1–12 % Asian Pakistani) (Nomis 2013). Nationality data were available for the two white offenders, who were both British, and for 43 of the 53 Asian offenders. Contrary to media stereotypes that ICST is a Pakistani import (Cockbain 2013a), most of these ethnically Asian offenders (n = 34, 79 %) were British nationals. Typically they were born and raised in the UK, rather than emigrating later in life.

Almost half the offenders (45 %, n = 25) were in an adult relationship, although the precise configuration varied by case. Operation Central was anomalous here as it was only case in which most offenders were single. The majority of offenders in most cases and overall (65 %, n = 36) lived with their families (parents, wives or both).

The overall unemployment rate of 35 % (n = 9) was notably higher than comparable national rates: for Pakistani-heritage males these stand at 10 % for 16–24 year-olds and 8 % for 25–49 year-olds (Nomis 2013). The 60 % (n = 33) who were employed were typically in low-skill and low-pay jobs, most commonly working as taxi drivers, shop assistants, counter-staff in takeaways or food delivery drivers.

Offender connectivity was a major feature of all six cases: 96 % (n = 53) of the offenders were directly linked to at least one other offender. These connections were generally bonds typical of everyday life: many were relatives, friends, neighbours, flatmates and/or colleagues. Qualitative analysis of the data indicated that at least 62 % (n = 60) of the linked pairs were connected by social bonds extrinsic to and likely predating co-involvement in ICST. This finding undermines the stereotype of members of sex offender groups as deviant ‘others’ who converged around a shared interest in child abuse alone.

Not only were offenders embedded in networks of association but 89 % (n = 49) of them co-offended in ICST activity with one or more of their co-defendants and/or with other suspects beyond the immediate offender network. Interestingly, over half of these co-offenders (n = 27) also offended alone. Most offenders in each case and overall (84 %, n = 41) contributed tangible commodities to co-offences, including by providing alcohol, drugs, transportation, abuse locations and victims.

The above results support the proposition that offenders’ routine activities facilitate access to suitable co-offenders, potential victims and appropriate crime facilitators. Particularly important were the high levels of connectivity (offender networking) and co-offending. The level of group cohesion was, for some but not all cases, higher than that described by Felson (2003) for the typical offender group. This may be explained by the pre-existing nature of most of the social bonds linking offenders. The shared ethnicity of most of the offenders in each case corresponds with extensive research highlighting ethnic homogeneity as a common characteristic of co-offending groups (Bijleveld et al. 2007; Horvath and Kelly 2009; Reiss 1988; Warr 1996) and indeed of social groups in general (e.g. McPherson et al. 2001). From a routine activity perspective, it seems self-evident that the groups to which we belong would influence our associations and activities. As predicted by Felson, the ICST offenders engaged in both co- and solo- offending. This result suggests pattern of individual opportunism coupled with the spread of offending opportunities through the offender networks.

Higher-than-average unemployment levels are potentially informative since unemployment might result in less formal surveillance, reduced commitment to conventional action and more free time to offend. Further support for a routine activity explanation of ICST comes from employed offenders’ concentration in jobs that provided ready, opportunistic access to potential victims and enticements to offer them (e.g. free food and lifts).

The age profile of the offenders in our study differed in some respects from what has previously been documented in studies on child sexual abuse. On the one hand, the mean age of first documented ICST offence was—at 29 years—consistent with that found for extrafamilial child sexual offenders in an Australian study (Smallbone and Wortley 2000). The same study found differences in age profile between intrafamilial (n = 79) and extrafamilial (n = 60) offenders, with respective means of 33.1 and 29.4 years. The between-group differences were not statistically significant but this may be a reflection of the limited sample size. Offenders’ age, it was suggested, may influence the types of opportunities they encounter (or create) to abuse children inside or outside the family. On the other hand, the skew in our sample towards the younger age groups indicates ICST offenders may be different from that for extrafamilial child sexual abusers as a general group. The age distribution for our sample was more closely associated with the age-crime curve for crime more broadly (Hirschi and Gottfredson, 1983)—including sexual assault against adults (Gannon et al. 2008; Miller 2013)—than for child sexual abuse. The age distribution for child sexual abuse tends to be bimodal with twin peaks in adolescence/early adulthood and mid to late thirties (Smallbone et al. 2008).

Victims’ routine activities

Victims’ routine activities help explain their availability, attractiveness and vulnerability to offenders

Since we selected cases based on the inclusion criterion that the case had involved child trafficking, we were surprised to find four adults in the victim sample (all from Operation Retriever). After deliberation, we included them in the analysis since all were directly connected to one or more of the underage victims and were abused by the same offenders as the children were. Overall, victims ranged in age from 11 to 20 years with a mean of 15 (SD = 2 years). Although the precise age distribution varied by case (see Fig. 3), the overall distribution was normal with a modal age of 14 years (see Fig. 4).

Fig. 3

Age distribution of victims by case

Fig. 4

Age distribution of full victim sample

As shown in Table 3, most or all victims in each case and 79 % (n = 34) overall were white British. This figure was broadly proportionate with an English population that is 80 % white British (the individual local authority areas in question are 67–92 % white British) (Nomis 2013).

Table 3 Victims’ ethnicity

In each case and overall most victims (86 %, n = 37) were at school or college when the abuse began and most (81 %, n = 35) lived with their families. Only four girls were in local authority care initially; another eight went into care after the ICST began due to resultant behavioural deterioration. Although the available data did not support a comprehensive assessment of victims’ backgrounds, the limited information we could find suggested a clustering of trauma. At least 70 % of victims (n = 30) had a pre-ICST history involving one or more of the following: social services involvement (n = 10); going missing (n = 7); disruptive or aggressive behaviour (n = 9); drug abuse (n = 3); mental illness (n = 2); and sexual abuse (n = 16). Thirty per cent of the victims (n = 13), including some with no other prior traumas documented, had criminal records. A minority of victims had stereotypically ‘normal’ backgrounds, which highlights the dangers of assuming ICST only affects children with problematic lives.

There were multiple victims in all cases except Operation Wheat; 91 % (n = 39) of them were connected to at least one other victim. Common links included friendship, kinship and being schoolmates. Close analysis showed 75 % (n = 39) of the links were strong social bonds predating and extrinsic to the abuse.

As shown in Table 4, victims came into contact with their subsequent abuser(s) in three ways: a direct meeting; an introduction by another girl, usually a victim herself; or an introduction by another offender. Some victims met multiple offenders through multiple different pathways.

Table 4 Modes by which victims came into contact with their abusers

Introductions through offenders or other girls ranged from the casual to the calculated. Indicative of worryingly risky behaviour, nine victims described meeting up with strangers who had obtained their telephone numbers. Direct meetings were largely opportunistic and typically occurred on the street but also in parks, shopping centres, transport hubs, takeaways etc. There was evidence to suggest that many offenders happened across targets in the course of their everyday activities, such as driving taxis or simply ‘cruising’ around town.

Overall 91 % (n = 39) of the victims received tangible commodities from offenders. These were typically low-value consumables, especially free lifts, alcohol and drugs. These ‘gifts’ functioned variously as inducements to attract victims or means of lowering their inhibitions and facilitating abuse or securing continued compliance with offenders’ demands. The following quotation from a victim’s police interview captures the way in which the promise of such goods motivated some victims to respond to offenders’ initial advances:

You’re a young girl, Friday night, your mum and dad don’t give you much money… if someone pulls over and says they’re going to buy you a bottle of vodka obviously you’re going to get in and want to go for a drink.

Other identifiable push- and pull-factors that led victims to engage and continue engaging with offenders included: their friends’ involvement; boredom; thrill-seeking; lack of economic power; loneliness; conflict at home; and attraction or emotional attachment to offenders. While a detailed discussion of the interpersonal dynamics and grooming processes underpinning ICST is beyond this paper’s remit, it should be noted here that some victims described their entire lives becoming entrenched in the abuse. The routinisation of ICST is illustrated well in this quotation from a victim’s police interview:

It was like a lifestyle. Like get up, go out and then do whatever—like you’d get up and go to work.

Our findings support the proposition that victims’ everyday activities help explain their availability, attractiveness and vulnerability to offenders. We stress that acknowledging ways in which victims’ activities may translate into risk should not be confused with victim blaming. Nor is it an attempt at exculpating the offenders, with whom the ultimate moral and legal responsibility rests. Ignoring victims’ activities and their agency only serves to promote idealised stereotypes of victims as passive objects. As a basis for crime prevention, stylised and inaccurate conceptions of victims are at best useless and at worst counter-productive.

The victims were characterised by connectivity to one another and (even more so than for offenders) links between them were typically extrinsic to shared involvement in ICST. There has previously been very little exploration into victim networking and how it may promote, spread or sustain patterns of victimisation (an exception is Cockbain et al. 2011). Our results indicate that just as offenders’ social networks can create offending opportunities, so too victims’ social networks can create victimisation opportunities.

The limited information available on victims’ backgrounds indicated that many had chaotic or problematic histories. This finding resonates with prior research into child sexual exploitation as a broader category (Beckett 2011; Cockbain et al. forthcoming; Cockbain and Brayley 2012; Jago et al. 2011). Such findings are sometimes interpreted as implying offenders are well attuned to vulnerability and select victims accordingly. We would caution against over-estimating offenders’ skills by confusing correlation with causality. Offenders may deliberately target vulnerable individuals but equally victims’ vulnerabilities may translate into lifestyle factors that increase their exposure to offenders and/or make them more susceptible if targeted.

The overlap between internal sex trafficking of children and young adults suggests an opportunistic approach to victim selection, also supportive of routine activity theory. The concentration of victims in their teens is important, as this is an age at which children become more independent and spend more time unsupervised outside the home. Such entirely normal changes might explain the modal age of the victims, which was similar to that found in the Child Exploitation and Online Protection Centre’s (2011) study on localised grooming: another form of extrafamilial child sexual abuse. In general, academic reviews have shown victims of extrafamilial child sexual abuse to have a higher age at onset than those of intrafamilial abuse (Fischer and McDonald 1998; Smallbone et al. 2008). The ICST victims’ age may also help explain how openly many offenders acted; sexual activity with teens might reasonable be expected to be less taboo than that involving younger children.

While primarily driven by Operation Retriever (also the case with the most victims), black and minority ethnicity victims were not underrepresented relative to national demographics. This finding challenges some media and right-wing political contentions that so-called ‘Asian sex gangs’ prey exclusively and deliberately on white girls (Cockbain 2013a; Meleagrou-Hitchens and Brun 2013). The predominance of white victims may simply be a reflection of the demographics of the available ‘pool’ of potential victims locally. Other possible influences on the target pool include socio-cultural and socio-economic factors affecting both teenagers’ everyday activities and the norms and expectations that constrain them.

Offender connectivity and offending rates

Better-connected offenders typically offend at a higher rate

The majority of offenders in every case and overall (89 %, n = 49) were implicated in contact sexual offences; 48 of these were accused of penetrative abuse. The other contact offender was convicted on DNA evidence alone and no further material was available about his offending behaviour. The remaining six offenders, including the only female, offended purely in a facilitation capacity, for example by procuring victims. In this section, we focus on the 48 contact offenders just described.

We were able to identify 414 distinct incidents³ of contact sexual offences between these 48 offenders and the victims in our sample. All victims bar one were subject to contact sexual offences; for most (n = 32, 76 %) the abuse included at least one incident of anal, vaginal or oral penetration. The figure of 414 incidents is in fact conservative as some of the most heavily-abused victims struggled to recall every incident. Additionally, around half the contact offenders (49 %, n = 24) were implicated in further offences against victims outside the core sample that were not included in this figure.

We found that the 414 incidents were not equally distributed across the 48 contact offenders. Individual offenders were implicated in one to 46 incidents each with a mean of eight (SD = 10) and a mode of three. The high standard deviation relative to the means and the large range indicates wide variance both between and within cases in levels of offending. Nonetheless, levels of repeat offending (88 %, n = 42) were high across all cases.

To test for a relationship between offender connectivity with their criminal counterparts and abuse rates we used inferential statistical tests. Here we had to take some additional precautions due to the clustering of offenders by case, which risked violating the assumption of independence underpinning most inferential statistical tests (Kreft and de Leeuw 1998). This problem is common to most co-offending research as the datasets involved often feature clustering or nesting. With a few notable exceptions (e.g. McGloin and Nguyen 2012), the issue has often simply been overlooked when running statistical analyses. As Kreft and de Leeuw (1998) note, inferential statistics may still be applied if it can first be shown that the independent variable is not correlated with the case⁴. We first calculated the intraclass correlation coefficient to check for the effects of clustering. The low⁵ score (0.09), together with the non-significant f test result [f(51) = 1.2, p = ns] provided confidence in the legitimacy of proceeding with inferential tests.

We then proceeded to test for a correlation between offender connectivity and ICST offending rates. Connectivity was measured using individuals’ degree scores that were calculated previously using social network analysis (Cockbain 2013b). Degree is a common metric used in social network analysis that is calculated based on an individual’s direct links to others in the network relative to total network size. It measures the extent to which an individual entity is directly connected to others or ‘in the thick of things’ (Morselli 2009: 390).

Due to the non-normal distribution of the offending rates we used non-parametric tests. The results were significant at the p < 0.001 level with a Spearman’s correlation co-efficient of 0.56. The associated R² value was 0.31, indicating that offender connectivity had a medium sized effect on offending rates.

Our results support the proposition that better-connected offenders tend to commit more offences. It is generally well established in the crime literature that a disproportionately large amount of offences are committed by a small number of prolific offenders (Everson 2003). The positive correlation we found between offender connectivity and offending rates supports Felson’s arguments that better-connected offenders commit more offences. One possible explanation for this is that the more connected a child sex offender is to other child sex offenders the more opportunities (s)he has to offend. Another explanation, which is complementary rather than competing, is that the group dynamics of sexual offender networks serve to normalise and neutralise the abuse. Better-connected members may well have fewer inhibitions and thus be more willing to offend. In routine activity terms, being embedded in an offender network may help create ‘likely’ offenders.

Locations for abuse

Abuse occurs at locations lacking supervision and familiar to offenders from their everyday lives

Here we used offenders as our unit of analysis to permit comparison with prior research. Table 5 shows the locations where offenders abused victims, either directly or as facilitators. Just over half (54 %, n = 29) of them abused victims at multiple location types.

Table 5 Offenders’ abuse locations (sample excludes offender prosecuted on DNA evidence alone)

Qualitative analysis of the data showed variations in abuse locations between and within cases that seem likely to be connected to offenders’ age and routine activities. Parks and playing fields, for example, were more common among younger offenders, whereas abuse in takeaways and restaurants typically involved employees of these premises or their friends. The three most prevalent location types overall were cars or taxis (43 %), offenders’ homes (35 %) and other private properties to which offenders had access (46 %) (typically co-offenders’ homes). The sheer volume of abuse locations and variable level of information about exact times, dates and addresses precluded more sophisticated analysis such as hotspot mapping. Nonetheless, the results provided partial support for the proposition, in that offenders appear to have tended towards locations with which they were already familiar.

There was some evidence to suggest ICST may differ from other child sexual abuse in terms of the offence locations. It was previously found that offenders’ or their victims’ homes were the most common location to abuse children sexually (Smallbone and Wortley 2000). In our study, such places were common but so too were semi-public and public locations.

The prevalence of non-private spaces was unexpected given the taboo nature of child sexual abuse. This apparent failure of guardianship could be linked various factors, including: the physical absence of guardians: inattentiveness of those guardians present; reluctance of potential guardians to intervene (e.g. due to fear of repercussions, lack of moral censorship of the offending itself or perhaps ambiguity over the victims’ apparent age); and/or physical characteristics of the spaces that limited natural surveillance (e.g. offences in parks might be hidden due to dense vegetation or limited lighting). The fact that many offenders co-offended with their friends, family members, colleagues and so forth (see section on offender connectivity) is also worth highlighting here. From a guardianship perspective, such contacts would normally be expected to act as intimate handlers rather than actively participating in offences.

Most offenders took few if any steps to conceal these locations from their victims in order to frustrate any eventual investigation. They employed few if any other safeguards against detection or investigation. The following behaviours, for example, were common: not masking faces; limited or erratic condom use⁶; using their normal telephones⁷; and using their own names. Such a general lack of precautions undermines the media portrayal of the offenders as calculating predators. It also indicates that ICST activity was not perceived as particularly risky. Given the long periods over which much of the offending had gone unchallenged, such a belief might have been well-founded—rational even.

Conclusion

When faced with an issue as emotive and widely stigmatised as child sexual abuse there is an understandable tendency to assume that abhorrent crimes must involve exceptional people and processes. There are many myths and untested assumptions around internal trafficking of British children, or ‘on street grooming’ as it is sometimes known. In this paper, we closely examined the characteristics of victims, offenders and crime events across six major cases. We found that in many respects ICST satisfies the expectations of opportunity theories of crime.

Our results provided strong support for four of the five propositions tested: many of the offenders were criminal generalists; offenders’ everyday activities facilitated access to co-offenders, victims and crime facilitators; victims’ everyday activities helped explain their availability, attractiveness and vulnerability; and better-connected offenders typically committed more offences. There was limited evidence supporting our final proposition: offenders abused victims at a wide range of locations familiar from their everyday lives, but it was not clear whether and how supervision was lacking. Our study makes an important empirical contribution towards expanding the limited knowledge base on ICST and internal trafficking more broadly. It also makes a contribution to the literature on opportunity theories by highlighting their explanatory utility in the context of a serious and complex crime.

In light of on-going efforts to improve responses to ICST, our results also have implications for practice. The findings offer a counterbalance to popular myths and assumptions about ICST. When examined carefully and dispassionately, the evidence about ICST and the people, places and processes involved recalls Arendt’s (1963) famous notion of the ‘banality of evil’. While treating ICST offenders as uniquely deviant criminal masterminds may be comforting, it is unlikely to assist in designing and delivering effective counter-measures. In contrast, recognising the importance of opportunity and routine activities opens up new avenues for interventions. As Wortley and Smallbone (2006b) have suggested, offenders who are ‘opportunity takers’ may well be deterred through situational crime prevention (Clarke 1980).

It is a fundamental principle of situational prevention that interventions need to be tailored to the specific dynamics of the crime problem in question. For this reason we are loath to suggest a cook-book of generic interventions for ICST based on our findings. However, by way of illustration, in the cases we examined mapping the social connections of existing victims (e.g. via social network analysis) could help identify others at particular risk of involvement in ICST. Preventative and protective interventions—such as helping young people look out for one another—might be then be targeted at high-risk children (e.g. friends or classmates of existing victims) rather than delivered through universal education efforts. Given the public and semi-public nature of much of the ICST we examined, efforts might also be directed at improving surveillance at known pick-up and offending locations (hotspots), be that through enhancing CCTV coverage, educating place managers to look out for suspicious behaviour, or increasing targeted police patrols.

There is growing evidence on the effectiveness of situational interventions in tackling diverse crime problems (e.g. Bowers et al. 2011; Clarke 1997). Yet, the majority of such work to date has focused on so-called ‘volume’ crimes, such as burglary. It is only more recently that researchers have begun to explore situational crime prevention’s application to more serious or complex crimes (Bullock et al. 2010; Wortley and Smallbone 2006c). Our study contributes to a small but growing literature that suggests even the most reviled of crimes may be responsive to situational interventions.