From Known to Stranger Crossover: A Retrospective Study of Child Sex Offenders Released Into the Community

Mei Wah M. Williams¹, Kirsty Blackwood², Jim van Rensburg³, David T. Jones², Susan W. Calvert⁴

¹ School of Psychology, Massey University, Auckland, New Zealand
² Department of Corrections, Auckland, New Zealand
³ Regional Forensic Psychiatric Services, Auckland, New Zealand
⁴ Clinical Psychologist

[Sexual Offender Treatment, Volume 11 (2016), Issue 1]

Abstract

Aim/Background: Strict enforcement of residential restrictions has been the norm although there is a paucity of empirical evidence as to whether recidivist child sex offenders will reoffend against an unknown child having had no previous history of offending against children unknown to them. This research investigated Known to Stranger crossover of 110 men who sexually reoffended against children after release from prison.
Methods: The offenders were released from prison between 1996 and 2003, and followed up until 2009 (M = 117.85 months, SD = 26.99). The mean age of the participants when first arrested for a sexual offence was 27.82 years (SD = 9.35: Range 14-68 years).
Results and Conclusions: The incidence and factors associated with the risk of crossover from Known to Stranger child victims were examined, with results showing that sexual reoffending was low at 6.8%. Crossover from Known to Stranger victims was lower still; at less than 1% of the men released into the community. Despite the low base rate of the Known to Stranger group, analysis showed that offenders whose first victim was from the Social relationship domain (i.e. the offender met the victim through someone known to the offender as a work colleague, friend, or even a family member) were more likely to reoffend against a stranger child than offenders from the other Known relationship domains. Being younger in age at the time of the first sexual offense was also associated with a risk of relationship crossover. The recidivist group was most likely to be rated as medium-high to high risk of re-offending at the time of their release from prison.

Key words: crossover of sexual offending, residential placement, child sex offenders, recidivism, correctional rehabilitation

Introduction

Sexual attacks on children by unknown assailants can be extremely terrifying, creating a climate of fear about the safety of the most vulnerable within the community. What may contribute to this fear is the uncertainty as to whether a released child sex offender will target an unknown victim, having had no prior record of sexual offences against a stranger victim. The research into crossover patterns of men who sexually offend against children has been relatively recent. Up until the early 90s the specialization of men who sexually abuse children was commonly held, in that it was believed that men who offended against family members did so only within the family and did not do so outside the family (Hanson & Bussière, 1998, Lussier, 2005).

Abel and his colleagues (Abel, Becker, Cunningham-Rathner, Mittelman, & Rouleau, 1988; Abel, Mittelman, & Becker, 1985; Abel & Osborn, 1992) first challenged the notion of specialization in men who committed sexual offences. The coexistence of multiple paraphilias was found to be much more common than previously believed, with little consistency in the victim type. Sexual offences were committed against both family and extrafamilial members, and across different age groups and gender; a phenomenon they termed "crossover". The incidents of crossover were not rare, with the prevalence of crossover in the age domain greatest at 43%, and for gender and incest/extrafamilial offences at 23% and 20% respectively. The findings, however, need to be treated with caution. Abel et al.'s research participants were a community sample of men voluntarily seeking treatment from a University-based center. Two-thirds had no connections with the criminal justice or forensic services, and the sexual offences included nontouching sexual behaviors or were against adults and/or children. Some of the paraphilias identified at the time were illegal but are no longer so, such as sadism, masochism, transvestism, and homosexuality. Therefore, the findings are not necessarily relevant to men who have been convicted of a sexual offence against children.

Subsequent to Abel et al.'s studies, Weinrott and Saylor (1991) found similar levels of versatility in men who committed these offences engaging in sexual as well as nonsexual criminal activity, although their sample included a number of men with sexual assaults against adults. Although the study did not specifically focus on crossover, offenders attending the treatment program who had sexual offences against family members also disclosed undetected sexual abuse of children outside the home. The extent of sexual offences reported also appeared to be underrepresented in official arrest records (Abel & Osborn, 1992). When assurance of confidentiality was given (Bradford, Boulet, & Pawlak, 1992) or under the pressure of polygraph testing (English, Jones, Patrick, & Pasini-Hill, 2003; Heil, Ahlmeyer, & Simons, 2003), participants disclosed a greater diversity and crossover of sexual offences compared to their official arrest records.

However, these findings have been disputed by other studies. Guay, Proulx, Cusson, and Ouimet (2001) found a high level of stability in victim choice in terms of age, gender, and relationship to the victim in subsequent offences. That is, most men who sexually offended against children reoffended against victims with similar characteristics to their previous victims. If men whose sexual offence was against a familial member crossover, the victim was more likely to be someone known to them rather than someone who was unfamiliar. Sjöstedt, Långström, Sturidsson, and Grann (2004) followed up men with sexual offences for three to eight years after they were released in the community. They found 6% were reconvicted of a sexual offence during this period. Using retrospective official information of the repeat offenders, Sjöstedt et al. found offender-victim relationship to be highly stable in that those who were convicted of family related sexual offences in a previous offence were 27 times more likely to repeat their offence with similar victims. Repeat offences against stranger victims were nine times higher for those with prior offences with stranger victims. They concluded that repeat offending with similar victim-offender relationship was much more the norm than crossover between relationships.

Cann, Friendship, and Gozna (2007) examined the extent of crossover on victim characteristics such as age, gender, and relationship of men who sexually offended against children and were subsequently reimprisoned for sexual offences. Using official arrest records, Cann et al. found more than three quarters of the offenders did not demonstrate any boundary crossover. The prevalence of relationship crossover was found to be low at 14%; which upon review by Sim and Proeve (2010) deemed to be evidence of stability in victim choice. Cann et al. identified the factors associated with an increased likelihood of crossover, and these were offenders with a high risk score on the Static 99 (Hanson & Thornton, 1999) and offenders with a longer offending history.

Similar results to that of Cann et al. (2007) were reported by Sim and Proeve (2010). Unlike other studies, their participants consisted only of men who had sexually offended against children and who were attending a community based treatment program. The researchers found that although preference for age was less stable, the preference for the same gender and offender-victim relationship was highly stable with 70 to 80% of participants choosing victims on similar characteristics. The relationship between risk level and crossover found in Cann et al.'s (2007) were not replicated, although offenders who demonstrated crossover had more victims than those who did not. Offenders who were less than 25 years of age at the time of their first sexual offence were more likely to subsequently sexually assault stranger victims.

Levenson, Becker, and Morin (2008) investigated the gender crossover characteristics and its relationship to the victim's age. They proposed that gender crossover was much more likely to occur when the victim was at preschool age rather than when the victim was older. The findings from their study supported their hypothesis that the gender of the victim would remain stable as the age of the victim increased. However, the incidence of crossover of gender was negatively associated with the age of the child, in that the younger the child the greater the likelihood of crossover in sexual assaults against both male and female victims. Offenders who sexually offended against both genders had a greater likelihood of perpetrating sexual offences against a very young child, compared to offenders whose offences were against one gender only. The crossover of gender was particularly so when the child was a preschooler. Unlike Cann et al.'s (2007) study, the Static-99 risk score was not predictive of risk of gender crossover but was strongly associated with a diagnosis of pedophilia. The findings by Levenson et al. (2008) study that gender crossover was prevalent amongst offenders with sexual offences with very young victims seems to be inconsistent with the results of other studies. However, Levenson et al.'s investigations examined different age groups that included very young victims of sexual assaults from the age of 0-6 years, unlike other studies.

Most studies have investigated offenders taking part in a treatment program. Duwe, Donnay, and Tewksbury (2008) conducted research that followed 3166 men released in the community who had sexually offended against children. Although the researcher's intention was to examine the effectiveness of the residential zoning policy in reducing sexual recidivism, they found only a small proportion of men (7%) were reimprisoned for a new sexual offence during the 15 year period of the study. This finding is similar to Snyder's (2000) research in that the majority of sexual offences committed by the men occurred either through familial relationships or through social acquaintances. Only 21% of men crossed over to reoffend against victims unknown to them, although most of the victims were adults and not children. The sexual reoffending by the men appeared to support a pattern of stability in the victim relationship domain.

In summary, the findings from the crossover literature are mixed. The more recent studies (viz. Cann et al., 2007; Duwe et al., 2008; Sim & Proeve, 2010; Sjöstedt et al., 2004) revealed that crossover in victim types was not as common as suggested by Abel and his colleagues and earlier studies (such as English et al., 2003; Weinrott & Saylor, 1991). Other differences between the studies are also evident with Sim and Proeve (2010) reporting extrafamilial offenders to be relatively stable in their relatedness to the second victim compared to intrafamilial offenders, whereas Guay et al. (2001) found the reverse to be true. Cann et al. reported that the likelihood of men who crossover was positively associated with the risk of sexual recidivism, but this was not supported by Sim and Proeve.

The discrepancies in findings may be due to a number of methodological differences in the studies. As noted by Lussier (2005), methodological differences may depend upon the source of the data, the definitions used, and the characteristics of the offender in the studies. For example, when data were collected under conditions of anonymity and assurance of confidentiality (Weinrott & Saylor, 1991) or from polygraph testing (e.g. English et al., 2003; Heil et al., 2003, Hindman & Peters, 2001), greater levels of crossover and sexual and nonsexual crimes were disclosed. Much lower rates of crossovers were revealed in data from official records (Cann et al., 2007) or offenders attending a treatment facility (Abel et al., 1985; Sim & Proeve, 2010). Although official records may underrepresent the true rate of offending, the reliability and validity of polygraph testing and self-report under conditions of confidentiality assurances may also be questionable. Offenders may respond in a socially desirable manner and overinflate their level of offending, their memory recall may be subject to error, and the validity of the reports may be difficult to verify.

The inclusion of men who commit sexual offences against adults in the studies poses a number of difficulties in applying the findings to sexual offences against children. Except for Sim and Proeve's (2010) study, men who sexually assault adults have been included in all the studies and the composition of the groups have not always been reported. Men who sexually assault adults exhibit not only greater levels of antisociality (Abel et al., 1985; Nadesu, 2011) but are more likely to offend against both known and stranger victim (Beauregard, Leclerc, & Lussier, 2012), to use violence in the commission of their crime, and to be reconvicted of another crime within five years of their release (Nadesu, 2011). On the other hand, Miethe, Olson, and Mitchell (2006) reported men who sexually offend against children had a more restricted repertoire of criminal activity than men who sexually offend against adults. By including men who sexually offend against an adult in crossover studies it may inflate the nature and extent of the crossover, and generalizing such findings to men who offend against children appear inappropriate.

An additional methodological problem is the lack of consistency in the operational definition of offender-victim relationships, which limits comparisons across studies. Offender-victim relationships have been generally dichotomized into two broad categories. These may include "extrafamilial" and "incestuous relationships" (e.g. Abel et al., 1988; Doren, 1998; Guay et al., 2001); "within" and "outside the family" where "outside" includes victims known and unknown to the offender (Cann et al., 2007); and "family/related" and "stranger" (Sjöstedt et al., 2004) although social acquaintances were not included in either of the categories. Guay et al. (2001) differentiated relationships as "incest", "familiar" and "unfamiliar" in which neighbors were included in the "unfamiliar" category. Even within the terminology used, there is little consensus as to what the relationship categories represent. For example, incest may encompass biological relationships only (e.g. Abel & Osborn, 1992), or may include broader caregiver relationships such as step or foster relationships (e.g. Guay et al., 2001; Sim & Proeve, 2010; Sjöstedt et al., 2004), or "someone related to the victim" (e.g. English et al., 2003). Even less consistency is found in the extrafamilial category as it encompasses a diversity of relationships ranging from victims known to the offender to complete strangers, or is not defined at all.

The study's aims

Given the increased sensitivity as to the risk men who sexually offend against children pose to the community, particularly those who sexually offend against an unfamiliar victim having had no previous history of having done so, this study investigated the crossover patterns of New Zealand men reconvicted of sexual offenses against children after release from prison. Currently there is a gap in the literature in understanding the nature of crossover patterns of men who sexually reoffend against young children. To some extent this is exacerbated by a lack of consensus in the definition of the offender-victim relationship categories. Given the confusion around how offender-victim relationships are defined, the study aimed to develop a classification system that examined more closely the various offender/victim relatedness. Of particular interest to the study is the group of men who sexually assault children known to them and who subsequently sexually offend against an unknown victim. A second aim is to investigate the factors associated with the crossover of this group of men.

Method

Participants

Of the men incarcerated for sexual offences against children and released from prison between 1996 and 2003 and followed up until 2009, 153 were reconvicted of a sexual offence. The reoffending rate represented 6.8% of the 2,237 men released over a period of six to thirteen years. Of the 153 participants, 32 were excluded from the study because their reoffending was against an adult only (n = 28) or their reoffending was of a noncontact nature (n= 4), leaving 121 men who sexually reoffended against children. Information on the victim(s) was unable to be obtained for 11 in this group, leaving 110 cases for analysis. Of this group, the offenders were predominantly of New Zealand European descent (61.47%), with 31% being Māori. The remaining offenders were either of Pacific Island descent or of other ethnic origins. The mean age of the participants when first arrested for a sexual offence was 27.82 years (SD = 9.35: Range 14-68 years). Compared to the general New Zealand prison population, men who sexually offend against children were older and more likely to be of New Zealand European descent (Dept of Corrections, nd).

The 110 offenders who reoffended against children under the age of 16 had a total of 412 victims (M = 3.75, SD = 2.07, median = 3, range 2-13). Nearly two thirds of the offenders (n = 72) committed sexual offences against females only, 11% (n = 12) reoffended against males only, and 24% (n = 26) reoffended against both. Nearly half were aged 25 years of age or less at the time of their first conviction (45.5%), with 38% aged between 26-35 years, and 16% aged 36 years and over.

Procedure

Ethical approval for this research was obtained from the New Zealand Department of Corrections. The study was a retrospective investigation of the official records of 2,237 men who sexually offended against children and who were released from prison between 1996 and 2003. Their sexual offences against children prior to their incarceration were reviewed, as well as the details of their reconvictions for sexual offences after release from prison. The followup period was until 2009; 6 to 13 years after release from prison (M = 117.85 months, SD = 26.99). The reconviction data were obtained from the New Zealand Department of Corrections Criminal History database, which is a nationwide electronic database on the entire offender population in New Zealand.

Data on the offender's age at first sexual offence, ethnicity, relationship to their victims, and gender of the victims were collected from official records including the Police Summary of Facts, Judge's Sentencing Notes, and Victim Impact Reports. If there was insufficient detail in the official records, information was sought from psychological reports on file. Information gathered on the victim was for gender only, as the database does not give the exact age of the victims. However, all victims were under the age of 16 years as it is illegal to have sexual contact with a person under the age of 16 in New Zealand.

The offenders' actuarial risk ratings at the date of release were obtained. In New Zealand all men who sexually offended against children receive an automated risk score that is generated from their official criminal data file; known as the Automated Sexual Recidivism Scale (ASRS; Skelton, Riley, Wales, & Vess, 2006). The seven items in the scale are derived from the Static 99 (Hanson & Thornton, 1999); items that are routinely captured in an offender's official criminal history record and can be extracted for the ASRS measure. This information includes prior sexual offences, convictions for noncontact sex offences, nonsexual violent offences, male sex victims, and age at time of index offence. According to Skelton et al. (2006) the ASRS receiver operating characteristic (ROC) area under the curve (AUC) for the measure was 0.78 at 5 years and 0.70 at 15 years followup period, which is comparable to the findings of Hanson and Thornton (1999) and Doren (2004) for the Static-99. The measure can reliably differentiate between four risk categories (low, medium low, medium high, and high risk offenders) (Skelton et al., 2006).

Development of Relationship Categories

Discrepancies in the way offender-victim relationships have been operationally defined and the ambiguity as to the exact nature of the construct have made it difficult to compare research findings. For example, incest is sometimes used to describe offending against a biological child (e.g. Quay et al., 2001), or distal biological relationships such as a niece or a grandchild (Hanson, Morton, & Harris, 2003), or stepchildren (Studer, Clelland, Aylwin, Reddon, & Monro, 2000). In spite of this, the literature has generally adopted two main types of victim relationships: related (familial/incest) and nonrelated (extrafamilial). Stranger victim (victim unknown to the offender at the time of the offending) was generally included as part of the extrafamilial category or not mentioned at all (except for Sjöstedt et al., 2004).

As the aim of this study was to investigate relationship crossover, particularly from known to stranger victims, it was important that the various victim relationships be explicitly defined. Firstly, the relationship of the offender to the victim was examined from case files. The relationships ranged from close biological relationships, brief to long term social contact with the victim or through collateral associations (such as work colleagues, friends of friends), and victims who were strangers to the offender. Each of the offender-victim relationships was recorded and through a series of iterative processes five main categories were eventually identified; Related, Parental Role, Formal Authority, Social, and Stranger (see Figure 1 for the definitions of these categories). The first four categories (Related, Parental, Formal, and Social) were then reduced into one category "Known". The "Stranger" category was defined as victims who were known to the offender less than 24 hours of contacting or meeting them, similar to that used by Sjöstedt et al. (2004). Where it was unclear from the data records how the offender knew the victim, but indicated that the offender was unknown to the victim at the time of the offending, this was included in the "Stranger" category. The processes of identifying the five categories are shown in the Appendix.

Figure 1: Categories of offender-victim relationships Note. Bolded variables represent aggregated categories. See Appendix for more detail on the development of the categories.

Two groups (Known and Stranger; refer Figure 1) were ultimately used to investigate the crossover patterns of men who sexually offended against children in this study; between victims known to the offender ("Known") and those unknown ("Stranger"). It should be noted that this distinction between "known" and "stranger" victims is different from that often found in the literature between "intrafamilial" and "extrafamilial" offenders. In this study the "Known" category encapsulated a much broader range of relationships that included not only relatives but children victimized by the offender through collateral social relationships.

Results

Independent-samples t-test was used to compare men who sexually reoffended against children with those who had not reoffended during the period of 1996 to 2003. The level of crossover from Known to Stranger was examined using descriptive statistics. The variables associated with the likelihood of Known to Stranger crossover were investigated using point-biserial correlation for dichotomous variables and Pearson correlation for continuous variables.

Comparisons on the level of risk between those who reoffended and those who did not are shown in Table 1. The recidivist group was at higher risk of reoffending with a greater proportion rated as medium-high to high on the ASRS (M = 2.54, SD = .98), compared to those who did not reoffend (M = 1.72, SD = .71). The difference in risk scores for the two groups was statistically significant (t(2164) = 9.99, p = .001 (one tailed), d = .97).

Crossover analyses

The relationship of the victim to the offender was not always recorded in the file information thus resulting in differing sample sizes in the crossover analyses. Although the majority of men who sexually reoffended against children showed some degree of crossover (see Table 2), 85% of the crossover was within the same relationship category (i.e. Known to Known). Fourteen offenders crossed from Known to Stranger victims; being 13.7% of the recidivist offenders and 0.63% of the cohort of men who sexually offended against children released from prison during the period of 1993 and 2006.

More detailed analysis of the crossover patterns is presented in Table 3. The rows represent the first known relationship type and the columns subsequent categories of victim relatedness. The figures in the table represent the cross-over pattern for each relationship type. Thus, for offenders who had a Parental relationship with their first victim, 35.7% subsequently reoffended against the same type of victim, 7.1% against Formal or Stranger victims, and 50% were against a Social victim whom they knew through their social activities or contacts. Within the categories of Known relationships (Parental, Formal, Related, and Social), those categorized as Social offenders committed nearly 80% of the Known to Stranger crossover. Furthermore, this group of offenders showed greater diversity in their crossover of victims in targeting victims from both the Known categories and the Stranger groups. These offenders also had a higher proportion of crossover (34.7%) compared to the other groups of offenders (refer to Table 2).

Given the disproportionate number of offenders from the Social group who subsequently sexually abused a Stranger victim, the characteristics of this group were compared with the other sexual offenders from the Known category. As shown in Table 4, 73% of offenders in the Social group who crossed over from Known to Stranger victims were aged 25 years or less when they first offended. In comparison, offenders from the other Known categories who crossed over from Known to Stranger were from an older cohort group. Based on the risk ratio, the risk of reoffending was 4.1 times greater for the Social group than the other three Known groups combined.

Point-biserial correlation analyses were conducted to investigate the relationship between age at first offence and number of victims with crossover from Known to Stranger victim based on the first victim. The number of victims and crossover from Known to Stranger victims was statistically significant (r_pb = .28, p < .05). This suggested that crossover was positively related to the number of victims. When the base rate of the phenomenon of interest in the total sample population is small, McGrath and Meyer (2006) recommend that .24 is regarded as a moderate effect size; therefore the magnitude of the correlation coefficient is deemed to be moderate. The relationship between age and stranger crossover (r_pb = -.18) was not statistically significant. Pearson correlation coefficient showed a statistically significant inverse relationship between age and number of victims (r = -.25, p < 0.5), although this relationship was small (Cohen, 1988). Offenders who were older at their first sexual offence had fewer victims than offenders who began offending at a younger age.

Discussion

The aim of the study was to examine the offender-victim relationship patterns of men who reoffended and were convicted of child sexual offences, with specific reference to the prevalence of crossover from known to stranger victims. Five broad categories of offender-victim relationship were identified, and from these the Known and Stranger categories were derived.

Of the sample of men who were released from prison between 1996 and 2003 for sexual offences against children, nearly 7% of men were reconvicted of sexual offences against children. This recidivism rate is lower than the 11% found by Vess and Skelton (2010), although their followup period was from 13-18 years whereas this study had a followup period of 6-13 years. This confirms the findings by other studies (e.g. Hanson & Bussiere, 1998; Hanson & Morton-Bourgon, 2005; Harris & Hanson, 2004; Sample & Bray, 2003; 2006) that the base rate of reoffending is much lower than for nonsexual types of offences (Miethe, Olson, & Mitchell, 2006; Sample & Bray, 2003; 2006).

Of the group of released offenders who reoffended, the phenomenon of crossover was common with nearly 70% of offenders demonstrating relationship crossover. However, most of these occurred within the Known relationship groups. Of the group of men whose reoffence victims were available, the crossover of sexual offences against children from Known to Stranger was 14.9%. This group made up less than 1% of all the men released from prison between 1996 and 2003 sexually reoffending against a Stranger victim, and having had no previous history of such type of offending.

Of specific interest to the study were the reoffenders who crossed from sexually assaulting Known to Stranger children. Analysis of the Known to Stranger crossover indicated that offenders whose first victim was in the Social category were statistically at significantly greater risk of crossover than for offenders in the other Known categories. Over 80% of this group committed Known to Stranger crossover. Not only did this group show greater diversity in targeting victims across all the relationship categories, their risk of reoffending against a Stranger was four times greater than the other three Known groups combined. This suggests that men who sexually abuse a child met through collateral relationships, such as a work colleague or children of friends, are more likely to cross over to offend against a child unknown to them.

The question is why would Social offenders target a Stranger victim in subsequent offences? This is perhaps not unexpected, in that it suggests that men who initially offend against children outside the family environment are more likely to make the additional step towards offending against complete strangers. Although speculative, there may be similar characteristics between the Social offender and the Stranger offender in that both are opportunists taking advantage of their social network or activities to offend. The level of familiarity with the victim may be minimal at the time of the offence compared to the Related and other Known categories, and the risk of detection would be higher. These factors may appeal more to the predatory type of offenders who have a preference for risky antisocial behaviors, such as young offenders. This type of offender may represent the group identified by Ward and Siegart (2002) as the offenders who sexually abuse children as part of a general pattern of antisocial propensity. Offenders in this group will tend to manipulate opportunities for their own sexual gratification and disregard societal norms around sexual activities with a child. Although it is beyond the scope of the study, particular characteristics of men who sexually offended against children may promote crossover of relationship boundaries to offend against an unknown child (e.g. Mandeville-Norden & Beech, 2009; Robertiello & Terry, 2007; Ward & Beech, 2005).

Furthermore, offenders who were younger at age of first offence (25 years or less) were also more likely to crossover from Known to Stranger victims than older offenders from the other Known categories, although the difference was not statistically significant. The findings are similar to research that indicate age at first offence is associated with greater deviance and recidivism (Craig, 2011; Hanson & Bussière, 1998; Hanson et al., 2003) and that young offenders are more likely to sexually abuse unknown victims (Sim & Proeve, 2010).

In conclusion, crossover was not an uncommon event although it typically occurred within other Known relationships rather than across the Known to Stranger category. Crossover from Known to Stranger was rare, supporting Guay et al.'s (2001) findings that men who sexually offended against children tend to reoffend against victims who are familiar rather than targeting someone unknown to them. Although the risk of Known to Stranger crossover is small, even one incident of crossover from known to stranger relationships would be unacceptable to the community. The factors most related to risk of Known to Stranger crossover seemed to be men whose first sexual offences were against children in the Social relationship category and whose first sexual offence was committed at a young age. Furthermore, offenders who were rated as medium-high to high risk of recidivism at the time of their release from prison were significantly more likely to be reconvicted of a sexual offence against children.

This study is the first to examine crossover for men who sexually offended against children using several relationship categories. It is noted that this study was reliant on reconviction data and any undetected offending could not be taken into account. The true rate of reoffending may well be higher and the diversity of victim selection greater. The detection and conviction of a sexual offense, however, is highly dependent on a range of factors, such as the victim being willing to report the offence, the Police being able to locate the assailant who might be unknown to the victim, and the outcome of a lengthy and arduous court process. Therefore the true rate of sexual offences against an unknown child will be difficult to estimate. It is also noted that only a small sample of reoffenders was available for analysis. The low rate of recidivism contributed to this, although this was exacerbated by the unavailability of data for 9% of the participants; thus further reducing the sample size. The findings of the study may, therefore, not be generalizable given the small sample size.

Future directions for the research would be to replicate the study to determine if the findings can be generalized with a larger sample. This study also did not investigate patterns as to when and why the crossover in relationships occurred. This information could be of interest in terms of determining differences between offenders who diversify early from offenders who crossover in later offending, and the factors that promote or facilitate crossover. Such qualitative information could assist in determining the factors that promote risk of cross-over, and could potentially be useful in treatment programs for men who sexually offend against children.

References

1. Abel, G. G., Becker, J. V., Cunningham-Rathner, J., Mittelman, M., & Rouleau, J-L. (1988). Multiple paraphiliac diagnoses among sex offenders. Bulletin of the American Academy of Psychiatry and the Law, 16, 153-168.
2. Abel, G.G., Mittelman, M.S., & Becker, J.V. (1985). Sexual offenders: Results of assessment and recommendations for treatment. In M.H. Ben-Aron, S.J. Hucker, & C.D. Webstr (Eds.). Clinical Criminology: The Assessment and Treatment of Criminal Behaviour. (pp. 191-205). Clarke Institute of Psychiatry: University of Toronto, Canada.
3. Abel, G. C. & Osborn, C. (1992). The paraphilias: The extent and nature of sexually deviant and criminal behavior. Psychiatric Clinics of North America, 15, 675-687.
4. Bagley, A.L. (2008). An era of human zoning: Banishing sex offenders from communities through residence and work restrictions. Emory Law Journal, 57, 1347-1392.
5. Beauregard, E., Leclerc, B., & Lussier, P. (2012). Decision making in the crime commission process: Comparing rapists, child molesters, and victim cross-over sex offenders. Criminal Justice and Behavior, 39, 1275-1295.
6. Bradford, J. M. W., Boulet, J., & Pawlak, A. (1992). The paraphilias: A multiplicity of deviant behaviours. Canadian Journal of Psychiatry, 37, 104-108.
7. Bulman, P. (2010). Electronic monitoring reduces recidivism. Corrections Today, December 2010, 72-73.
8. Cann, J., Friendship, C., & Gozna, L. (2007). Assessing crossover in a sample of sexual offenders with multiple victims. Legal and Criminological Psychology, 12, 149-163.
9. Comartin, E.B., Kernsmith, P.D., & Kernsmith, R.M. (2009). Sanctions for sex offenders: Fear and public policy. Journal of Offender Rehabilitation, 48, 605-619.
10. Craig, L.A. (2011). The effect of age on sexual and violent reconviction. International Journal of Offender Therapy and Comparative Criminology, 55, 75-97.
11. Department of Corrections (nd). Census of prison inmates and home detainees 2003. Retrieved from http://www.corrections.govt.nz/research/census-of-prison-inmates-and-home-detainees/census-of-prison-inmates-and-home-detainees-2003/introduction.html
12. Department of Corrections (2012). Child sex offenders, determining accommodation suitability. Retrieved from http://www.corrections.govt.nz/policy-and-legislation/cpps-operations
13. Doren, D.M. (1998). Recidivism base rates, predictions of sex offender recidivism, and the "sexual predator" commitment laws. Behavioral Sciences and the Law, 16, 97-114.
14. Doren, D.M. (2004). Stability of the interpretative risk percentages of the RRASOR and Static-99. Sexual Abuse: A Journal of Research and Treatment, 16, 25-36.
15. Duwe, G., Donnay, W., & Tewksbury, R. (2008). Does residential proximity matter? A geographic analysis of sex offender recidivism. Criminal Justice and Behavior, 35, 484-504.
16. English, K., Jones, L., Patrick, D., & Pasini-Hill, D. (2003). Sexual offender containment: Use of the postconviction polygraph. Annals New York Academy of Sciences, 989, 411-427.
17. Guay, J-P., Proulx, J., Cusson, M., & Ouimet, M. (2001). Victim-choice polymorphia among serious sex offenders. Archives of Sexual Behavior, 30, 521-533.
18. Hanson, R. K. & Bussière, M. T. (1998). Predicting relapse: A meta-analysis of sexual offender recidivism studies. Journal of Consulting and Clinical Psychology, 66, 348-362.
19. Hanson, R. K. & Morton, K. E., & Harris, A. J. (2003). Sexual offender recidivism risk: What we know and what we need to know. Annals of New York Academy of Science, 989, 154-166.
20. Hanson, R. K. & Morton-Bourgon, K. E. (2005). The characteristics of persistent Sexual offenders: A meta-analysis of recidivism studies. Journal of Consulting and Clinical Pyschology, 73, 1154-1163.
21. Hanson, R.K., & Thornton, D. (1999). Static-99: Improving Actuarial Risk Assessments for Sex Offenders (User report no. 1999-02). Ottawa: Department of the Solicitor General of Canada.
22. Harris, A.J.R., & Hanson, R. K. (2004). Sex offender recidivism: A simple question. Ottawa: Public Safety and Emergency Preparedness Canada.
23. Heil, P., Ahlmeyer, S., & Simons, D. (2003). Crossover sexual offences. Sexual Abuse: A Journal of Research and Treatment, 15, 221-236.
24. Hindman, J., & Peters, J.M. (2001). Polygraph testing leads to better understanding adult and juvenile sex offenders. Federal Probation, 65, 8-14.
25. Levenson, J. S., Becker, J., & Morin, J. W. (2008). The relationship between victim age and gender crossover among sex offenders. Sexual Abuse, 20, 43-
26. Levenson, J. S., & Cotter, L. P. (2005). The effect of Megan's law on sex offender reintegration. Journal of Contemporary Criminal Justice, 21, 49-66.
27. Lussier, P. (2005). The criminal activity of sexual offenders in adulthood: Revisiting the specialization debate. Sexual Abuse: A Journal of Research and Treatment, 17, 269-292.
28. Mandeville-Norden, R., & Beech, A. R. (2009). Development of a psychometric typology of child molesters: Implications for treatment. Journal of Interpersonal Violence, 24, 307-325.
29. McGrath, R. E., & Meyer, G. J. (2006). When effect sizes disagree: The case of r and d. Psychological Methods, 11, 386-401.
30. Miethe, T. D., Olson, J., & Mitchell, O. (2006). Specialization and persistence in the arrest histories of sex offenders. A comparative analysis of alternative measures and offense types. Journal of Research in Crime and Delinquency, 43, 204-229.
31. Nadesu, A. (2011). Reconviction Rates of Sex Offenders: Five Year Follow-Up Study. Department of Corrections; Wellington, New Zealand.
32. Petrunik, M. (2005). Dangerousness and its discontent: A discourse on the socio-politics of dangerousness. Sociology of Crime, Law and Deviance, 6, 49-74.
33. Quay, J-P, Proulx, J-P, Cusson, M., & Ouimet, M. (2001). Victim-choice polymorphia among serious sex offenders. Archives of Sexual Behavior, 30, 521-533.
34. Queensland Corrective Services, (2007). Supervising DPSOA offenders in the community: Electronic monitoring and other requirements. Queensland Government: Australia.
35. Robertiello, G., & Terry, K.J. (2007). Can we profile sex offenders? A review of sex offender typologies. Aggression and Violent Behavior, 12, 508-518.
36. Sample, L. L., & Bray, T. M. (2003). Are sex offenders dangerous? Criminology and Public Policy, 3, 59-82.
37. Sample, L. L., & Bray, T. M. (2006). Are sex offenders different? An examination of rearrest patterns. Criminology and Public Policy,17(1), 83-102.
38. Sim, D. J., & Proeve, M. (2010). Crossover and stability of victim type in child molesters. Legal and Criminological Psychology, 15, 401-413.
39. Sjöstedt, G., Långström, N., Sturidsson, K., & Grann, M. (2004). Stability of modus operandi in sexual offending. Criminal Justice and Behavior, 31, 609-623.
40. Skelton, A., Riley, D., Wales, D., & Vess, J. (2006). Assessing risk for sexual offenders in New Zealand: Development and validation of a computer-scored risk measure. Journal of Sexual Aggression, 12, 277-286.
41. Snyder, H. N. (2000). Sexual Assault of Young Children as Reported to Law Enforcement: Victim, Incident, and Offender Characteristics. Bureau of Justice Statistics; US Department of Justice.
42. Socia, K. M., & Stamatel, J. P. (2012). Neighborhood characteristics and the social control of registered sex offenders. Crime & Delinquency, 58, 565-587.
43. Soothill, K., Francis, B., Sanderson, B., & Ackerley, E. (2000). Sex offenders: Specialists, generalists - or both? The British Journal of Criminology, 40, 56-67.
44. Stanlas, L. J. (2004). Adult sex offenders on community supervision: A review of recent assessment strategies and treatment. Criminal Justice and Behavior, 31, 564-608.
45. Studer, L. H., Clelland, S. R., Aylwin, A. S., Reddon, J. R., & Monro, A. (2000). Rethinking risk assessment for incest offenders. International Journal of Law and Psychiatry, 23, 15-22.
46. Tewksbury, R., & Levenson, J. (2007). When evidence is ignored: Residential restrictions for sex offenders. Corrections Today, 69, 54-57.
47. Ward, T., & Beech, A. (2005). An integrated theory of sexual offending. Aggression and Violent Behavior, 11, 44-63.
48. Ward, T., & Siegert, R. J. (2002). Toward a comprehensive theory of child sexual abuse: A theory knitting perspective. Psychology, Crime, & Law, 8, 319-351.
49. Weinrott, M. R., & Saylor, M. (1991). Self-report of crimes committed by sex offenders. Journal of Interpersonal Violence, 6, 286-300.
50. Willis, G. M., Levenson, J., & Ward, T. (2010). Desistance and attitudes towards sex offenders: Facilitation or hindrance. Journal of Family Violence, 25, 545-556.

Author address

Dr Mei Wah Williams
School of Psychology
Massey University
Private Bag 102 904, North Shore City
Auckland 0745, New Zealand
m.w.williams@massey.ac.nz