Exploring the victim offender overlap among people with an intellectual disability.

BACKGROUND: While there is growing international evidence pointing to the increased risks of crime perpetration and victimisation for some people with an intellectual disability, the overlap between offending and victimisation (the victim-offender overlap) remains unclear.
METHOD: This study utilised a data linkage methodology of 2600 people with an intellectual disability, exploring their contacts with public mental health services and the police in Victoria, Australia.
RESULTS: Victim-offenders accounted for a small proportion of the sample (n = 148, 5.7%). The victim-offender overlap was evident for both violent and nonviolent nonsexual crimes, particularly for theft, burglary, and threat-related crimes. Key differences were also noted between males and females.
CONCLUSIONS: People with an intellectual disability who are both victim and offenders comprise a small but particularly complex justice-involved population. Future research should explore the victim offender overlap for males and females separately, as well as any additional risks and vulnerabilities associated with specific mental health diagnoses.

INTRODUCTION

Correlates of victimisation and offending have commonly been explored using general population samples. However, it has been strongly asserted that people with an intellectual disability are a group who may be additionally vulnerable to being justice‐involved (Gulati et al., 2018). Prior research has established that certain characteristics of intellectual disability, such as lowered self‐control, difficulty perceiving risk, being overly trustful, and acting compliantly can increase people's susceptibility to justice involvement more broadly (McClure et al., 2009; Modell et al., 2008). Some prior research indicates that this population are statistically over‐represented as victims and as offenders (Baldry et al., 2013; Hayes et al., 2007), although the available prevalence estimates do not account for individuals with mild and borderline intellectual disability who may go undiagnosed and undetected (Herrington, 2009). Focussing on offending, a systematic review (Fazel et al., 2008), pooling almost 12,000 prisoners, concluded that rates of clinically diagnosed intellectual disability were ‘at least as common as in the general population of similar age’ (p. 372). More recent research conducted in Ireland and Australia suggests an over‐representation in police custody (Gulati et al., 2018) and prison (Trofimovs et al., 2021).

It is well‐established that initial contacts with police can affect future justice involvement for people, including those with an intellectual disability (Modell et al., 2008; Spivak & Thomas, 2013); with Australian and international research finding that people with an intellectual disability are younger at age of initial victimisation, offending, and imprisonment compared to the general population (Baldry et al., 2013; Reppermund et al., 2019). Expanding on this, Cockram (2005) cautioned that younger age at first offence was linked with a greater likelihood of subsequent episodes of offending.

Despite recent Australian findings indicating the general population have higher risk of crime victimisation and equivalent rates of offending (operationalised as criminal charges) compared to people with intellectual disability (Fogden et al., 2016; Nixon et al., 2017), other research has found that the latter have been found to have a statistically increased likelihood of sexual and violent victimisation and offending (Jones et al., 2012; Krnjacki et al., 2016). For instance, Fogden et al. (2016) reported sexual victimisation and offending were 11 and eight times higher, respectively, among males with intellectual disability compared to males in a community sample of people without an intellectual disability. Nixon et al. (2017) reported people with intellectual disability were two and three times more likely to experience violent victimisation and offending, respectively, compared to a general community sample. In addition, Van der Put et al. (2014) reported sexual offending was more strongly related to sexual rather than physical victimisation, whereas earlier research by Hayes (2009) found that past physical victimisation could predict violence during subsequent offending. Some other prior research has investigated the prevalence of specific crimes among this population. Simpson and Hogg (2001) found that criminal damage and burglary were disproportionately high, while Faccini and Saide (2012) reported that written threats made by people with an intellectual disability were often targeted at people known to them. Of note, Jones (2007) concluded that arson, once believed to be highly prevalent due to biased sampling, was actually thought to be infrequent. Findings from a more recent meta‐analysis (Fisher et al., 2016) also reported a high prevalence of theft and assault‐related victimisation however, as noted elsewhere (Fogden et al., 2016; Wilson & Brewer, 1992), the prevalence of nonviolent nonsexual crime among those with an intellectual disability remains unclear.

People with an intellectual disability have been found to be more susceptible to experiencing mental illness than the general population (Mazza et al., 2020; Moss et al., 1997); estimates suggest up to half are believed to experience mental illness, known as dual disability (Emerson & Hatton, 2007). However, this prevalence may be underestimated as behaviours related to intellectual disability can overshadow symptoms of mental illness, making it difficult to diagnose (Hounsome et al., 2018). There remains only limited understanding of the statistical associations between dual disability and justice involvement; although some research suggests victimisation can trigger the onset of mental illness among people with an intellectual disability (Clark et al., 2016), while lowered self‐control could enhance externalising aggressive behaviour, increasing the risks of both offending (Barron et al., 2002; Dudley et al., 1999) and victimisation (Fisher et al., 2012). People with a dual disability seem especially vulnerable to justice involvement. For example, Thomas et al. (2019) reported that those with dual disability were almost three times and over three times as likely to be a victim or an offender, respectively, than those with intellectual disability alone.

An accurate prevalence estimate of crime among people with intellectual disability remains difficult to ascertain, due to under‐reporting of victimisation and of offending. There are also recognised challenges around determining thresholds for challenging behaviour from that which is potentially criminal behaviour. For example, prior research has found that incidents in disability support facilities are infrequently reported to police, with staff having both a high tolerance towards offending behaviour and a reluctance to report such incidents or involve the police (Addicott et al., 2018; Lyall et al., 1995). However, negating to report incidents of crime possibly suggests that people with intellectual disability experience insufficient guardianship from experiencing victimisation as well as offending. Other barriers that have been reported that impede official reporting for people with an intellectual disability include fear of being sent to a more restrictive setting as a consequence of reporting the perpetrator (Petersilia, 2001), as well as difficulties communicating with police (Henshaw & Thomas, 2012).

Rationale and aim

The victim‐offender overlap has been widely investigated among the general population but less so among people with an intellectual disability. Some early work by Lindsay et al. (2001) noted that around a third of their sample of 46 sexual and 48 nonsexual offenders had histories of sexual abuse and physical abuse respectively. They argued that while these sexual abuse and physical abuse histories were significant, the so called ‘cycle of abuse’ was not deemed to be an inevitable consequence or adequate explanation for their offending. A more recent paper describing a treatment programme for men with an intellectual disability who were at risk of sexual offending (Murphy et al., 2010) also reported a higher overlap, with over half of their treatment sample having histories of sexual victimisation, mostly commonly in their childhoods. Against this background, the aim of the current study was to explore associations between offending and victimisation among people with intellectual disability, clarifying prevalence and characteristics of victim‐offenders overall and by sex. Victim‐offenders were compared to three other justice contact groups (nonvictim nonoffenders, pure victims, and pure offenders).

Three hypotheses were explored: Based on Cockram (2005), who demonstrated age of first offence was associated with future offending and Modell et al. (2008) who opined that initial police contact affected future justice involvement among people with intellectual disability, it was hypothesised that age of first police contact would be associated with the occurrence of victimisation and offending. Second, based on Thomas et al. (2019), who identified that people with dual disability were disproportionately represented as victims and offenders, it was hypothesised that there would be an association between dual disability and the victim‐offender classification. Finally, due to the lack of published evidence available, the third hypothesis was exploratory and considered both general and specific victim‐offender associations and differences between males and females.

METHOD

Participants

The sample were derived from the Restrictive Intervention Data System (RIDS), which is used to record instances of restrictive interventions for people with disabilities across the state of Victoria, Australia. Participants were eligible if they were listed on RIDS as having received one or more restrictive interventions (routine, emergency or otherwise) between the 1st July 2007 until 31st December 2012, and if they had an intellectual disability, as defined by the Department of Human Services (2007), assessed using a standardised measure of intellectual functioning and a standardised measure of adaptive behaviour (see Fogden et al., 2016). No exclusion criteria were applied. Eligible participants comprised 1684 males and 916 females (N = 2600), with an average age of 36.09 years (SD = 16.64, Mdn = 35.29).

Materials

Contact‐level data were extracted from three contacts‐based databases.

Restrictive intervention data system

The Victorian Office of the Senior Practitioner operate RIDS, which is used by over 150 Victorian government disability agencies to record restrictive intervention usage: This can include chemical restraints, physical restraints, or seclusion. Unique client identifiers are assigned to each individual on the database, with information recorded such as type of disability and date of birth.

Law Enforcement Assistance Program

Since its inception in 1993, Victoria Police have used the Law Enforcement Assistance Program (LEAP) database to record all police contact with individuals, whether they are victims, witnesses, suspects, offenders or otherwise. Offence data comprised of arrest date, charge, and outcome (conviction, diversion, or dismissed), whilst victimisation data are recorded using standard offence codes. Information on intervention orders (IVOs), such as who the complainant and defendant are in the matter, are also contained in LEAP.

Victorian Psychiatric Case Register

Operational since 1961, the Victorian Psychiatric Case Register (VPCR) database records public mental health service contact in Victoria. Data includes reason for contact, date of admission and discharge, as well as the World Health Organisation's (2018) ICD‐10 codes for mental disorder diagnosis. The register has a broad coverage of service contacts but does not include mental health service access outside the public health system (e.g., private practice), or interstate public mental health services contacts (Short et al., 2010).

Procedure

Research design

The study employed a data‐linkage research design, whereby name and date of birth from individuals in the RIDS cohort were corresponded against LEAP and VPCR databases to establish whether participants had any recorded contact with Victorian police or mental health services. Deidentified contact‐based data were then provided to the researchers, which were then linked, checked, and combined into a singular, fully deidentified dataset. Ethical approval was received from the host institution (Ref: 22870).

Data coding

Reason for and date of first police contact were recorded, as were initial and final dates of victimisation and offending. Offending and victimisation were initially broadly coded into three categories of crime (violent, sexual, and nonviolent nonsexual), as per Nixon et al. (2017). Offences were defined as any charges laid against the individual, as people with intellectual disability are often diverted from criminal proceedings (Herrington, 2009). Participants were classified as a nonvictim nonoffenders, pure victims, pure offenders, or as victim‐offenders. The number of IVOs and type of IVO contact (defendant, complainant, or both) were also coded.

Episodes of specific criminal charges, including fire‐based crime, theft, threat, criminal damage, stalking, assault, rape, and burglary were also coded. Assault comprised several offences: Common law assault, indecent assault, unlawful assault, intentionally/recklessly causing injury, and discharging a missile to injure. Fire‐based crime included criminal damage by fire (arson), giving false fire alarms, and lighting fires without authority.

Data analysis

Data were initially explored using descriptive statistics and testing assumptions of homogeneity of variance and normality. The first hypothesis utilised t tests, Spearman's rho correlation, and analysis of variance. Linear regressions were performed to predict total offending and victimisation episodes from a number of variables for individuals with any police contact, as well as victim‐offenders. The second hypothesis was tested using Chi‐squared tests of association and multinomial logistic regression to examine whether victim‐offenders had an increased likelihood of dual disability. The third hypothesis was tested using Chi‐squared tests of association, with only victim‐offenders selected for analyses. Sex‐based differences were also explored.

RESULTS

Sample characteristics

The sample comprised 2600 participants, with 1684 males (64.77%) and 916 females (35.23%). Female's age during the study (M = 38.98, SD = 16.64) was significantly older than males (M = 34.53, SD = 16.43), t(2598) = −6.57, p < .001, d = .27. Nineteen males and five females identified as being Indigenous or Torres Strait Islander. Culturally and linguistically diverse participants comprised three (2.03%) victim‐offenders, 11 (3.55%) pure victims, five (6.10%) pure offenders, and 99 (4.81%) nonvictims nonoffenders, with no significant differences between groups (p = .294, Φ = .04). Several significant differences were apparent when victim‐offenders (n = 148) were compared to pure victims (n = 310), pure offenders (n = 82) and nonvictim nonoffenders (n = 2060), regarding age of first police contact, age of first victimisation, age of first criminal charge, total number of victimisations, total number of criminal charges, and number of IVOs (see Table 1).

TABLE 1

Characteristics of sample by justice contact group

Characteristic variable	Victim‐offenders (n = 148)			Pure victims (n = 310)			Pure offenders (n = 82)			Nonvictim nonoffenders (n = 2,060)			Test statistic
	M	SD	Mdn	M	SD	Mdn	M	SD	Mdn	M	SD	Mdn
Age at time of study	34.26	11.72	34.31	40.25	16.14	40.53	38.94	13.95	39.06	35.49	17.01	34.36	F(3, 2596) = 8.86, p < .001, η 2 = 0.01
Age first police contact	18.86	9.69	17.32	28.36	14.56	27.47	25.95	12.56	24.65	22.65	14.31	18.54	F(3, 874) = 19.18, p < .001, η 2 = 0.06
Age first victimised	22.96	11.23	20.53	30.06	14.20	29.20	‐			‐			t(357.99) = 5.79, p <.001, d = .55
Male a	22.64	10.98	20.19	29.95	15.01	28.51	‐			‐			t(279.02) = 4.74, p <.001, d = .55
Female	24.01	12.09	23.96	30.18	13.20	29.96	‐			‐			t(172) = 2.51, p = .013, d = .49
No. of victimisations	3.20	3.66	2.00	1.53	1.08	1.00	‐			‐			t(159.31) = ‐5.46, p <.001, d = .62
Male	2.94	2.81	2.00	1.39	.81	1.00	‐			‐			t(124.53) = ‐5.72, p <.001, d = .75
Female	4.06	5.56	2.00	1.70	1.32	1.00	‐			‐			t(34.97) = ‐2.49, p = .018, d = .58
Age first charge	23.66	10.02	20.99	‐			28.59	12.17	26.22	‐			t(142.15) = 3.13, p = .002, d = .27
Male	22.81	9.73	20.01	‐			27.11	11.88	24.76	‐			t(178) = 2.64, p = .009, d = .40
Female	26.44	10.56	25.09	‐			35.20	11.59	35.46	‐			t(48) = 2.61, p = .012, d = .79
No. of charges	11.69	18.12	4.00	‐			5.66	13.46	2.00	‐			t(209.17) = ‐2.87, p = .005, d = .38
Male	13.97	20.06	5.00	‐			6.10	14.68	2.00	‐			t(170.17) = ‐3.02, p = .003, d = .45
Female	4.31	4.50	2.00	‐			3.67	5.18	1.00	‐			t(48) = ‐.45, p = .658, d = .13
No. of total contacts	14.89	19.00	7.00	1.53	1.08	1.00	5.66	13.46	2.00	‐			F(3, 874) = 106.71, p <.001, η2 = .27
Min and max	2‐129			1–11			1–107			‐
Number of IVOs	1.78	3.35	0	0.23	1.16	0	0.55	2.31	0	0.08	0.49	0	F(3, 161) = 5.81, p = .001, η 2 = 0.10
Min and max	1–17			1–16			1–19			1–6

Note: No. of total contacts is victim and offender contacts only, not including other contacts with police such as missing person, and so on.

Abbreviation: IVO, information on intervention order.

Males accounted for 113 (76.35%) victim‐offenders, 171 (55.16%) pure victims, 67 (81.71%) pure offenders and 1333 (64.71%) nonvictims nonoffenders, with significant differences between the groups, χ 2(3) = 31.55, p < .001, Φ = 0.11.

Victimisation data were recorded for 458 (17.62%) participants, of which 284 (62%) were male. One hundred and thirty‐one (131, 28.6%) of those with victimisation histories had their first victimisation episode recorded during their childhood (i.e., under the age of 18 years old). Age of first victimisation was younger for males (M = 27.04 years, SD = 14.00) compared to females (M = 28.94 years, SD = 13.19), but not significantly so, t(456) = −1.44, p = .151, d = 0.14.

Criminal charges were recorded for 230 participants (8.85%), of which 180 (78.26%) were male. Seventy‐four (74, 32.2%) of those with criminal charge histories had their first charge recorded when they were under the age of 18 years old. Age at first criminal charge was significantly younger for males (M = 24.41, SD = 10.75) than females (M = 29.07, SD = 11.50), t(228) = −2.67, p = .008, d = 0.42.

Age of first police contact

One or more instance of official police contact was recorded by 878 of the participants (33.77%), of which 595 (67.77%) were male, with significant differences between males and females (χ 2 = 5.22, p = .022, Φ = −.05, OR = 0.82, 95% CI = 0.67–0.97). The average age of first police contact was 24.33 years (SD = 13.99, Mdn = 21.40), with males being significantly younger at first police contact (M = 23.33 years, SD = 13.93) than females (M = 26.42 years, SD = 13.92), t(876) = −3.07, p = .002, d = 0.22. Among participants with any police contact, there was no correlation between age of first police contact and the total number of recorded victim and offender contacts, r s = .00, p = .897. However, a number of significant correlations between age at first police contact and other justice outcomes were observed for the victim‐offenders (see Table 2).

TABLE 2

Future justice outcomes correlated with age of first police contact

Justice outcome variables correlated with age of first police contact	Victim‐offenders (n = 148)		Pure victims (n = 310)		Pure offenders (n = 82)		Any police contact (n = 878)
	r s	p	r s	p	r s	p	r s	p
Total no. victim and offender contacts	−0.245	.003	−0.065	.254	−0.223	.044	−0.004	.897
Male	−0.261	.005	−0.037	.636	−0.145	.241	−0.005	.898
Female	−0.181	.298	−0.107	.212	−0.441	.100	−0.014	.810
Total no. of victimisations	−0.287	<.001	−0.065	.254	‐	‐	0.016	.629
Male	−0.299	.001	−0.037	.636	‐	‐	0.006	.884
Female	−0.247	.153	−0.107	.212	‐	‐	−0.016	.792
Total no. criminal charges	−0.209	.011	‐	‐	−0.223	.044	−0.122	<.001
Male	−0.209	.026	‐	‐	−0.145	.241	−0.108	.009
Female	−0.164	.347	‐	‐	−0.441	.100	−0.088	.141
Total no. IVOs	−0.317	<.001	−0.218	<.001	−0.303	.006	−0.375	<.001
Male	−0.337	<.001	−0.324	<.001	−0.332	.006	−0.402	<.001
Female	−0.273	.113	−0.078	.363	0.000	1.0	−0.301.	<.001

Note: Any police contact refers to all participants with any police contact of any capacity (criminal or non‐criminal).

Abbreviation: IVO, information on intervention order.

Data were stratified by sex for the victim‐offender group, revealing males were younger at age of first police contact (n = 113, M = 18.38, SD = 9.01) than females (n = 35, M = 20.41, SD = 11.63), t(146) = −1.08, p = .280, d = 0.20. Males were younger at age of first victimisation (M = 22.64, SD = 10.98) than females (M = 24.01, SD = 12.09), t(146) = −0.63, p = .530, d = .12. Males were also younger at age of first criminal charge (M = 22.83, SD = 9.73) than females (M = 26.44, SD = 10.56), t(146) = −1.89, p = .061, d = 0.36. Females recorded more victimisations (M = 4.06, SD = 5.56) than males (M = 2.94, SD = 2.81), t(39.50) = −1.15, p = .259, d = 0.25. Males recorded a significantly higher number of criminal charges (M = 13.97, SD = 20.06) than females (M = 4.31, SD = 4.50), t(139.21) = 4.75, p < .001, d = 0.66.

Linear regression coefficient table results are displayed in Table 3. Among participants with any police contact, the model predicting the total number of charges was significant, F(7, 222) = 9.27, R = .476, R 2 adjusted = 0.202, p < .001, as was the model predicting total number of victimisations, F(7, 449) = 27.03, R = .544, R 2 adjusted = 0.286, p < .001. Among victim‐offenders, the model for predicting total number of charges was significant, F(6, 141) = 6.16, R = 0.456, R 2 adjusted = 0.174, p < .001, as was the model for predicting total number of victimisations, F(6, 141) = 7.05, R = .480, R 2 adjusted = 0.198, p < .001.

TABLE 3

Linear regression predicting total number of offences and victimisations

Variables	Victim‐offenders (n = 148)			Any police contact (n = 878)
	β	t	p	β	t	p
Total no. of criminal charges
Justice contact category	‐			0.09	1.35	.178
Sex	−0.22	−2.88	0.005	−0.17	−2.86	.005
Age first contact	0.19	1.53	0.129	0.19	1.53	.127
Age first charge	−0.40	−3.14	0.002	−0.34	−0.28	.006
Dual disability	0.16	2.05	0.042	0.17	0.28	.006
IVO history	−0.09	−0.83	0.408	−0.02	−0.24	.811
Number of IVOs	0.26	−2.88	0.005	0.29	3.61	<.001
Total no. of victimisations
Justice contact category	−			0.24	5.20	<.001
Sex	0.07	0.96	0.337	0.07	1.81	.072
Age first contact	0.31	2.12	0.035	0.27	2.17	.031
Age first victimisation	−0.37	−2.53	0.013	−0.29	−2.39	.017
Dual disability	0.03	0.41	0.684	0.03	0.65	.516
IVO history	−0.20	−1.92	0.057	−0.17	−3.00	.003
Number of IVOs	0.53	5.28	<0.001	0.52	9.40	<.001

Note: Any police contact refers to all participants with any police contact of any capacity (criminal or non‐criminal).

Abbreviation: IVO, information on intervention order.

Dual disability among victim‐offenders

Dual disability was identified for 709 (27.27%) participants, of which 477 (67.28%) were male, with no significant differences by sex, t(1936.14) = 1.66, p = .098, d = 0.07. There was a significant association between justice contact category and having dual disability (χ 2 = 135.99, p < .001, φ c = 0.23), which remained statistically significant for males (χ 2 = 81.87, p < .001, φ c = 0.22) and females when considered separately (χ 2 = 56.42, p < .001, φ c = 0.25). There was a significant association between being a victim‐offender and the presence of dual disability (χ 2 = 111.84, p < .001, Φ = 0.21, OR = 5.54, 95% CI 3.90–7.86), which remained significant when considered separately for males (χ 2 = 67.44, p < .001, Φ = 0.20, OR = 4.66, 95% CI 3.13–6.92) and females (χ 2 = 46.12, p < .001, Φ = .22, OR = 9.47, 95% CI 4.37–20.52). The likelihood of victim‐offenders having dual disability compared to the other three justice contact groups are displayed in Table 4.

TABLE 4

Victim‐offender likelihood of dual disability compared to other justice contact groups

Justice contact group		Male			Female			p
	n (%) with dual disability	OR	95%CI OR LL–UL	p	n (%) with dual disability	OR	95%CI OR LL–UL
Pure victim (n = 310)	51 (61.45%)	3.83	2.32–6.33	<.001	40 (12.90%)	7.15	3.08–16.60	<.001
Pure offender (n = 82)	30 (36.59%)	2.00	1.09–3.71	.026	8 (9.76%)	2.53	0.71–8.97	.151
Nonvictim nonoffender (n = 2,060)	326 (15.83%)	5.03	3.37–7.50	<.001	158 (7.67%)	10.40	4.78–22.66	<.001

Multinomial regression revealed that victim‐offenders were significantly more likely than pure victims, pure offenders, and nonvictim nonoffenders to have dual diagnosis. Victim‐offenders were more likely to have initial police contact at a younger age, be male, and have a history of IVOs than pure victims and nonvictim nonoffenders (see Table 5 for further comparisons).

TABLE 5

Multinomial logistic regression comparing justice contact groups to victim‐offenders

Variables	b (SE)	Wald χ 2	p	OR	95%CI OR LL–UL
Nonvictim nonoffender versus victim‐offender
Intercept	−0.30 (0.30)	1.01	.314	‐	‐
Age of first police contact	0.01 (0.01)	1.06	.303	1.01	0.99–1.03
No IVO history	0.69 (0.24)	8.21	.004	1.99	1.24–3.18
Male	−0.22 (0.24)	0.84	.359	0.80	0.50–1.28
Intellectual disability‐only	1.20 (0.21)	32.68	<.001	3.33	2.20–5.03
Pure victim versus victim‐offender
Intercept	−1.21 (0.34)	12.81	<.001	‐	‐
Age of first police contact	0.03 (0.01)	12.43	<.001	1.03	1.01–1.05
No IVO history	1.43 (0.29)	24.33	<.001	4.18	2.37–7.39
Male	−0.94 (0.24)	15.08	<.001	0.39	0.24–0.63
Intellectual disability‐only	1.29 (0.22)	33.13	<.001	3.63	2.34–5.63
Pure offender versus victim‐offender
Intercept	−2.39 (0.46)	26.56	<.001	‐	‐
Age of first police contact	0.03 (0.01)	6.65	.010	1.03	1.01–1.05
No IVO history	0.81 (0.37)	4.80	.028	2.25	1.09–4.65
Male	0.37 (0.35)	1.13	.288	1.45	0.73–2.89
Intellectual disability‐only	0.59 (0.29)	4.17	.041	1.80	1.02–3.15

Note: R 2 = 0.171 (Cox‐Snell), 0.186 (Nagelkerke). Model χ 2 (12) = 164.50, p < .001.

Abbreviation: IVO, information on intervention order.

General victim‐offender overlap

Crosstabulations exploring the general victim‐offender overlap among violent, sexual, and nonviolent nonsexual crime are displayed in Table 6. When explored by sex, associations between violent victimisation and violent offending remained significant for males (χ 2 = 12.28, p < .001, Φ = .33, OR = 8.61, 95% CI 2.19–33.89) but not for females (p = 1.0, Φ = −0.09, OR = 0.53, 95% CI 0.05–5.22), despite 21 (60.00%) female victim‐offenders recording overlap. Sexual victimisation and sexual offending associations were nonsignificant for victim‐offenders and remained nonsignificant when considering males (χ 2 = 0.78, p = .377, Φ = 0.08, OR = 1.46, 95% CI 0.65–3.07) and females (p = .457, Φ = −0.19, OR = 0.94, 95% CI = 0.83–1.06) separately. Nonviolent nonsexual victimisation and nonviolent nonsexual offending were significantly associated, but these associations became nonsignificant when considering males (χ 2 = 3.47, p = .063, Φ = 0.18, OR = 2.06, 95% CI 0.96–4.41) and females (p = .264, Φ = 0.24, OR = 2.97, 95% CI = 0.62–14.22) separately.

TABLE 6

Victim‐offender overlap for general and specific types of crimes

Types of offence	n	χ 2	p	Φ	OR	95% CI OR LL–UL
General overlap
Violent crime	114	5.38	.020	0.19	3.49	1.16–10.55
Sexual crime	21	0.034	.854	−0.02	0.94	0.47–1.86
Nonviolent nonsexual crime	36	5.17	.023	0.19	2.19	1.11–4.31
Specific overlap
Theft	26	7.35	.007	0.22	2.63	1.30–5.34
Assault	93	1.15	.284	0.09	1.63	0.66–4.02
Burglary	6	4.46	.035	0.17	3.17	1.04–9.67
Criminal damage	5	0.016	.899	−0.01	0.92	0.27–3.17
Threat	5	‐	.003 a	0.29	12.19	2.23–66.56
Rape	4	‐	1.0 a	−0.03	0.80	0.25–2.55
Stalking	0	‐	1.0 a	−0.02	0.95	0.91–0.98
Arson	0	‐	1.0 a	‐	‐	‐

Note: Crosstabulations were not computed for arson; while fifteen victim‐offenders had recorded arson offences, none recorded arson victimisation.

Fisher's exact test (FET) was used in this table and in‐text when at least one cell in Chi‐squared analysis had less than five counts.

Victim‐offender overlap for specific offences

Results for victim‐offender associations overall for the specific crimes of theft, threat, criminal damage, arson, burglary, stalking, assault, and rape are displayed in Table 6. When stratified by sex, victim‐offender associations for theft remained significant for males (χ 2 = 4.26, p = .039, Φ = 0.19, OR = 2.29, 95% CI 1.04–5.05) but not females (p = .103, Φ = 0.31, OR = 4.17, 95% CI 0.84–20.64). Burglary victim‐offender associations did not remain significant for males (χ 2 = 2.61 p = .107, Φ = 0.15, OR = 2.66, 95% CI 0.78–9.06) or females (p = .242, Φ = 0.27, OR = 7.50, 95% CI = 0.46–122.70). Threat victim‐offender associations remained significant for males (p = .011, Φ = 0.28, OR = 14.78, 95% CI 1.58–138.75) but not females (p = .113, Φ = 0.47, OR = 32.00, 95% CI 1.06–970.81). Assault victim‐offender associations remained nonsignificant for males (χ 2 = 2.56, p = .110, Φ = 0.15, OR = 2.46, 95% CI 0.80–7.62) and females (p = .709, Φ = −0.10, OR = 0.63, 95% CI 0.13–2.98). Criminal damage victim‐offender associations were overall nonsignificant and remained so for males (χ 2 = 0.01, p = .930, Φ = 0.01, OR = 1.06, 95% CI 0.29–3.88). Females recorded zero overlap between criminal damage victimisation and offending (p = 1.0, Φ = −0.15). Victim‐offender associations for rape were overall nonsignificant and remained so for males (p = .474, Φ = 0.08, OR = 1.73, 95% CI 0.49–6.10). Females recorded zero rape offences but 18 (51.40%) recorded rape victimisation. No stalking or arson victimisation and offending associations were recorded.

DISCUSSION

This study utilised data‐linkage methodology to explore the presence, prevalence, and characteristics of victim‐offenders, among a cohort of people with an intellectual disability who had experienced one or more restrictive interventions.

The first hypothesis extended prior findings by Cockram (2005), Baldry et al. (2013) and Reppermund et al. (2019), finding that victim‐offenders with intellectual disability had significantly earlier initial police contacts, and that this was associated with the extent of their subsequent justice involvement. Interestingly, earlier age of police contact appeared to be more influential on future victimisation outcomes than on offending‐related outcomes, reflected by the variability in the effect sizes reported here, which varied from small to moderate, the latter being the case for the association between age of first police contact and victimisation. Moderate effect sizes were observed when female victim‐offenders and female pure offenders were compared on age of first offence too. Previous application of routine activities theory among the general population suggests that females have lower risk of victimisation due to engaging in less risky lifestyles (e.g., Lauritsen et al., 1991). Although, given the nonsignificant differences between males and females regarding total number and age of first victimisation, it may be that those with intellectual disability, regardless of sex, have inherently risky lifestyles (e.g., living in residential facilities with others) which may create an environment where they could be more readily victimised (Nixon et al., 2017). The present study found females were more likely to be victim‐offenders than pure offenders, therefore consistent with research from Engström (2018) who proposed that females' risky lifestyles primarily result in victimisation. These findings also resonate with other prior research opining that males are more likely than females to respond to victimisation with aggression or offending, which has been linked to low self‐control (Menard & Covey, 2016).

Whereas prior research has investigated the link between dual disability and victimisation and offending separately, this study found dual disability was most common for the victim‐offender group. Interpreting these results through the lens of routine activities theory, it is possible that the presence of comorbid mental illness may further enhance the likelihood of a person with intellectual disability being exposed to risky environments and situations (Fogden et al., 2016). Taking this argument forward, challenging or aggressive behaviours may be more common among people with dual disability than people with intellectual disability alone (Dudley et al., 1999), which may serve to further enhance risk of precipitating victimisation and retaliatory behaviours (Clark et al., 2016; Fisher et al., 2012). Supporting this stance, Fogden et al. (2016) contended that treating symptoms of mental illness alongside intellectual disability deficits could have beneficial impacts on reducing the levels of justice involvement among those with a dual disability. Further, strengthening the types and range of supports available to support people with intellectual disability contacting police should occur, given the continued need for police legitimacy with respect to its community encounters and the known associations between initial police contact and future justice involvement, demonstrated in this study as well as previous work (Baldry et al., 2013; Modell et al., 2008; Spivak & Thomas, 2013).

The victim‐offender overlap was identified for violent crime, thus adding to the consistent literature base identifying violent victim‐offender overlap (Hiday et al., 2001; Sampson & Lauritsen, 1990; Silver et al., 2011), and building on past research indicating a statistically increased risk of violent crime among people with an intellectual disability (Krnjacki et al., 2016; Nixon et al., 2017). Although evidence for nonviolent nonsexual victim‐offender overlap was observed in this study, the veracity of this finding needs to be established, due to both the exploratory nature of this research and the mixed findings reported previously (Fogden et al., 2016; Nixon et al., 2017). Interestingly, sexual victim‐offender overlap was not found to be statistically significant, despite previous research reporting an increased statistical likelihood of sexual victimisation and offending among people with an intellectual disability (Fogden et al., 2016; Van der Put et al., 2014). Given the nature of the RIDS sample, it is likely the participants more commonly lived in residential services; prior research has indicated several barriers impede reporting in these services (e.g., Lyall et al., 1995) as well as challenges reported by police regarding formalising criminal charges (Addicott et al., 2018; Petersilia, 2001). Further, the low base rates of sexual offending and sexual victimisation reported for this sample, especially when the data were stratified by sex, may have compromised the statistical comparisons able to be made.

While acknowledging the small numbers involved here, there was some preliminary evidence of the specificity of the victim‐offender overlap for theft, burglary, and threat offences; the acquisitive crime findings being consistent with prior findings across time and jurisdictions (Fisher et al., 2016; Simpson & Hogg, 2001). These findings potentially suggest an ongoing lack of guardianship from these more opportunistic crimes, although future research could investigate what motivates or otherwise enables the opportunity for these crimes among people with an intellectual disability. Differences were also apparent for the males and females in the sample, with specific overlap regarding charges for theft and threat offences remaining significant only for males, building upon past similar research by Menard and Covey (2016) which demonstrated males (although, among a sample with no known intellectual disability) had more significant victim‐offender specificity than females regarding property crime.

Limitations and future directions

Whilst this study utilised a large sample of people with an intellectual disability and a powerful data‐linkage methodology using contacts‐based databases, the results should be interpreted in light of potential methodological limitations. Firstly, the results may not be broadly generalisable for people with intellectual disability, as this sample may include individuals with more complex presentations of intellectual disability (as evident by receiving at least one restrictive intervention and hence being recorded on the RIDS database used here). As a result, and as previously noted by Fogden et al. (2016), these findings may over‐ or under‐estimate the true rates of offending and victimisation. Replicating this study with a sample of people with an intellectual disability without restrictive intervention episode histories could potentially provide more broadly generalisable results regarding the generality and specificity of the victim‐offender overlap.

Secondly, rates of criminal offending were potentially over‐inflated due to operationalising offending as any criminal charges filed instead of convictions recorded. This was done because those with intellectual disabilities are regularly diverted from prosecution, thereby not recording a conviction (Herrington, 2009). Additionally, using formally recorded data may limit the present study's ability to accurately gauge the true prevalence of crime among the sample; further, Addicott et al. (2018) reported that staff in residential facilities may not report crime, or even where they do that police may choose to not formalise or record the incident. Future study should therefore seek to explore the circumstances and range of factors that influence staff and family members decisions to report and also formalise episodes of victimisation and offending with the police.

Thirdly, this study did not consider the degree to which victimisation and offending trajectories were inter‐dispersed and may have changed over time and place. Future research could explore whether victim‐offending among people with an intellectual disability is confined to a particular age range and whether the nature of this association and overlap changes over time. While just over a quarter of those who had histories of victimisation had their first recorded incident in their childhood, the rates of childhood victimisation are likely to be under‐represented here, due to the nature of the methodology not being able to overcome deficits with both timely reporting and investigation. Much work therefore remains to be done before a complete understanding of victim‐offender overlap among people with intellectual disability is established. Of note, gaps remain most prominent with our knowledge with respect to the specificity of the victim‐offender overlap of certain offences among males and females with an intellectual disability, both with and without comorbid mental illness.

CONCLUSION

The present study furthered the endeavour to understand more about justice involvement for people with an intellectual disability. Elucidating information about what justice contact looks like for people with intellectual disability who are victim‐offenders has important implications for needs‐based treatments, supports, and risk management strategies (Taylor & Lindsay, 2018). Of note, the present findings have potential implications for suggesting more effective intervention strategies may be required upon first contact with police, preventing early contact with the justice system turning into cyclic victimisation and offending. Drawing from established general population research on victim‐offenders (Bucerius et al., 2020; Schreck, 1999), and noting that people with an intellectual disability require different approaches to offender rehabilitation and treatment (Herrington, 2009; Rose et al., 2012), it is clear that victim‐offenders with an intellectual disability require a different approach that adequately takes into account their learning styles and living situations. Despite representing a small proportion of people with an intellectual disability, the available evidence suggests that those who are justice involved as both victims and offenders are not being sufficiently protected from harm to themselves and to others.

CONFLICT OF INTEREST

The authors declare no conflict of interest.